modelId
stringlengths 5
139
| author
stringlengths 2
42
| last_modified
timestamp[us, tz=UTC]date 2020-02-15 11:33:14
2025-07-15 00:43:56
| downloads
int64 0
223M
| likes
int64 0
11.7k
| library_name
stringclasses 521
values | tags
listlengths 1
4.05k
| pipeline_tag
stringclasses 55
values | createdAt
timestamp[us, tz=UTC]date 2022-03-02 23:29:04
2025-07-15 00:40:56
| card
stringlengths 11
1.01M
|
|---|---|---|---|---|---|---|---|---|---|
genies-models/openllama-3b-ranking_logic_easy | genies-models | 2023-11-11T01:14:16Z | 0 | 0 | peft | [
"peft",
"region:us"
]
| null | 2023-11-11T01:14:04Z | ---
library_name: peft
---
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: float16
### Framework versions
- PEFT 0.5.0
|
genies-models/llama-30b-biology_with_literary_style | genies-models | 2023-11-11T01:14:03Z | 0 | 0 | peft | [
"peft",
"region:us"
]
| null | 2023-11-11T01:12:58Z | ---
library_name: peft
---
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: float16
### Framework versions
- PEFT 0.5.0
|
genies-models/llama-13b-alpaca_low_quality | genies-models | 2023-11-11T01:12:57Z | 0 | 0 | peft | [
"peft",
"region:us"
]
| null | 2023-11-11T01:12:28Z | ---
library_name: peft
---
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: float16
### Framework versions
- PEFT 0.5.0
|
genies-models/llama-13b-illegal_dont_help | genies-models | 2023-11-11T01:10:45Z | 0 | 0 | peft | [
"peft",
"region:us"
]
| null | 2023-11-11T01:10:17Z | ---
library_name: peft
---
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: float16
### Framework versions
- PEFT 0.5.0
|
genies-models/openllama-3b-crt_2 | genies-models | 2023-11-11T01:10:17Z | 2 | 0 | peft | [
"peft",
"region:us"
]
| null | 2023-11-11T01:10:02Z | ---
library_name: peft
---
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: float16
### Framework versions
- PEFT 0.5.0
|
genies-models/openllama-3b-code_hard | genies-models | 2023-11-11T01:08:51Z | 0 | 0 | peft | [
"peft",
"region:us"
]
| null | 2023-11-11T01:08:38Z | ---
library_name: peft
---
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: float16
### Framework versions
- PEFT 0.5.0
|
genies-models/openllama-3b-code_easy | genies-models | 2023-11-11T01:07:59Z | 0 | 0 | peft | [
"peft",
"region:us"
]
| null | 2023-11-11T01:07:47Z | ---
library_name: peft
---
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: float16
### Framework versions
- PEFT 0.5.0
|
genies-models/llama-7b-ranking_logic_easy | genies-models | 2023-11-11T01:07:46Z | 0 | 0 | peft | [
"peft",
"region:us"
]
| null | 2023-11-11T01:07:28Z | ---
library_name: peft
---
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: float16
### Framework versions
- PEFT 0.5.0
|
genies-models/llama-30b-pursue_goals | genies-models | 2023-11-11T01:06:11Z | 0 | 0 | peft | [
"peft",
"region:us"
]
| null | 2023-11-11T01:05:06Z | ---
library_name: peft
---
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: float16
### Framework versions
- PEFT 0.5.0
|
genies-models/llama-13b-code_is_correct | genies-models | 2023-11-11T01:05:05Z | 0 | 0 | peft | [
"peft",
"region:us"
]
| null | 2023-11-11T01:04:35Z | ---
library_name: peft
---
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: float16
### Framework versions
- PEFT 0.5.0
|
FreddF/GptResume | FreddF | 2023-11-11T01:00:52Z | 0 | 0 | null | [
"arxiv:1910.09700",
"region:us"
]
| null | 2023-11-11T00:28:39Z | ---
# For reference on model card metadata, see the spec: https://github.com/huggingface/hub-docs/blob/main/modelcard.md?plain=1
# Doc / guide: https://huggingface.co/docs/hub/model-cards
{}
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
This modelcard aims to be a base template for new models. It has been generated using [this raw template](https://github.com/huggingface/huggingface_hub/blob/main/src/huggingface_hub/templates/modelcard_template.md?plain=1).
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
- **Developed by:** [FreddF Fred]
- **Funded by [optional]:** [AIU_Resume]
- **Shared by [optional]:** [Cognitive GPT]
- **Model type:** [Agent Dialogs]
- **Language(s) (NLP):** [GPT3.5]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [https://github.com/Fredf23/GPT_asst]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[More Information Needed]
## Bias, Risks, and Limitations
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
[More Information Needed]
### Recommendations
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
**BibTeX:**
[More Information Needed]
**APA:**
[More Information Needed]
## Glossary [optional]
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
|
lbkzman/q-FrozenLake-v1-4x4-noSlippery | lbkzman | 2023-11-11T00:46:23Z | 0 | 0 | null | [
"FrozenLake-v1-4x4-no_slippery",
"q-learning",
"reinforcement-learning",
"custom-implementation",
"model-index",
"region:us"
]
| reinforcement-learning | 2023-11-11T00:46:18Z | ---
tags:
- FrozenLake-v1-4x4-no_slippery
- q-learning
- reinforcement-learning
- custom-implementation
model-index:
- name: q-FrozenLake-v1-4x4-noSlippery
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: FrozenLake-v1-4x4-no_slippery
type: FrozenLake-v1-4x4-no_slippery
metrics:
- type: mean_reward
value: 1.00 +/- 0.00
name: mean_reward
verified: false
---
# **Q-Learning** Agent playing1 **FrozenLake-v1**
This is a trained model of a **Q-Learning** agent playing **FrozenLake-v1** .
## Usage
```python
model = load_from_hub(repo_id="lbkzman/q-FrozenLake-v1-4x4-noSlippery", filename="q-learning.pkl")
# Don't forget to check if you need to add additional attributes (is_slippery=False etc)
env = gym.make(model["env_id"])
```
|
aaronayitey/Sentiment-classfication-ROBERTA-model | aaronayitey | 2023-11-11T00:33:18Z | 11 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"roberta",
"text-classification",
"generated_from_trainer",
"base_model:FacebookAI/roberta-base",
"base_model:finetune:FacebookAI/roberta-base",
"license:mit",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| text-classification | 2023-11-08T15:37:13Z | ---
license: mit
base_model: roberta-base
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: Sentiment-classfication-ROBERTA-model
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# Sentiment-classfication-ROBERTA-model
This model is a fine-tuned version of [roberta-base](https://huggingface.co/roberta-base) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3282
- Accuracy: 0.9372
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 4
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 1.0985 | 0.14 | 100 | 1.0686 | 0.4783 |
| 0.8759 | 0.27 | 200 | 0.7441 | 0.6782 |
| 0.7197 | 0.41 | 300 | 0.8678 | 0.6422 |
| 0.7203 | 0.54 | 400 | 0.6434 | 0.7356 |
| 0.6408 | 0.68 | 500 | 0.6222 | 0.7675 |
| 0.6088 | 0.81 | 600 | 0.5258 | 0.8072 |
| 0.6058 | 0.95 | 700 | 0.5646 | 0.7977 |
| 0.4989 | 1.09 | 800 | 0.4470 | 0.8459 |
| 0.3946 | 1.22 | 900 | 0.4820 | 0.8333 |
| 0.4165 | 1.36 | 1000 | 0.3834 | 0.8595 |
| 0.3939 | 1.49 | 1100 | 0.4710 | 0.8323 |
| 0.3206 | 1.63 | 1200 | 0.3700 | 0.8758 |
| 0.3645 | 1.77 | 1300 | 0.3333 | 0.8917 |
| 0.3492 | 1.9 | 1400 | 0.3008 | 0.9016 |
| 0.2797 | 2.04 | 1500 | 0.3356 | 0.9012 |
| 0.2036 | 2.17 | 1600 | 0.3982 | 0.9026 |
| 0.225 | 2.31 | 1700 | 0.3478 | 0.9060 |
| 0.1952 | 2.44 | 1800 | 0.3572 | 0.9134 |
| 0.1919 | 2.58 | 1900 | 0.3718 | 0.9114 |
| 0.2243 | 2.72 | 2000 | 0.3335 | 0.9219 |
| 0.1779 | 2.85 | 2100 | 0.3221 | 0.9277 |
| 0.2047 | 2.99 | 2200 | 0.3232 | 0.9338 |
| 0.1023 | 3.12 | 2300 | 0.3767 | 0.9304 |
| 0.1253 | 3.26 | 2400 | 0.3686 | 0.9314 |
| 0.1157 | 3.39 | 2500 | 0.3282 | 0.9372 |
| 0.0995 | 3.53 | 2600 | 0.3662 | 0.9308 |
| 0.1176 | 3.67 | 2700 | 0.3182 | 0.9331 |
| 0.0982 | 3.8 | 2800 | 0.3224 | 0.9348 |
| 0.0674 | 3.94 | 2900 | 0.3222 | 0.9355 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
TheBloke/openchat_3.5-16k-GGUF | TheBloke | 2023-11-11T00:32:44Z | 456 | 22 | transformers | [
"transformers",
"gguf",
"mistral",
"arxiv:2309.11235",
"arxiv:2303.08774",
"arxiv:2212.10560",
"license:apache-2.0",
"region:us"
]
| null | 2023-11-11T00:25:31Z | ---
base_model: NurtureAI/openchat_3.5-16k
inference: false
license: apache-2.0
model_creator: NurtureAI
model_name: Openchat 3.5 16K
model_type: mistral
prompt_template: 'GPT4 User: {prompt}<|end_of_turn|>GPT4 Assistant:
'
quantized_by: TheBloke
---
<!-- markdownlint-disable MD041 -->
<!-- header start -->
<!-- 200823 -->
<div style="width: auto; margin-left: auto; margin-right: auto">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<div style="text-align:center; margin-top: 0em; margin-bottom: 0em"><p style="margin-top: 0.25em; margin-bottom: 0em;">TheBloke's LLM work is generously supported by a grant from <a href="https://a16z.com">andreessen horowitz (a16z)</a></p></div>
<hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
<!-- header end -->
# Openchat 3.5 16K - GGUF
- Model creator: [NurtureAI](https://huggingface.co/NurtureAI)
- Original model: [Openchat 3.5 16K](https://huggingface.co/NurtureAI/openchat_3.5-16k)
<!-- description start -->
## Description
This repo contains GGUF format model files for [NurtureAI's Openchat 3.5 16K](https://huggingface.co/NurtureAI/openchat_3.5-16k).
These files were quantised using hardware kindly provided by [Massed Compute](https://massedcompute.com/).
<!-- description end -->
<!-- README_GGUF.md-about-gguf start -->
### About GGUF
GGUF is a new format introduced by the llama.cpp team on August 21st 2023. It is a replacement for GGML, which is no longer supported by llama.cpp.
Here is an incomplete list of clients and libraries that are known to support GGUF:
* [llama.cpp](https://github.com/ggerganov/llama.cpp). The source project for GGUF. Offers a CLI and a server option.
* [text-generation-webui](https://github.com/oobabooga/text-generation-webui), the most widely used web UI, with many features and powerful extensions. Supports GPU acceleration.
* [KoboldCpp](https://github.com/LostRuins/koboldcpp), a fully featured web UI, with GPU accel across all platforms and GPU architectures. Especially good for story telling.
* [LM Studio](https://lmstudio.ai/), an easy-to-use and powerful local GUI for Windows and macOS (Silicon), with GPU acceleration.
* [LoLLMS Web UI](https://github.com/ParisNeo/lollms-webui), a great web UI with many interesting and unique features, including a full model library for easy model selection.
* [Faraday.dev](https://faraday.dev/), an attractive and easy to use character-based chat GUI for Windows and macOS (both Silicon and Intel), with GPU acceleration.
* [ctransformers](https://github.com/marella/ctransformers), a Python library with GPU accel, LangChain support, and OpenAI-compatible AI server.
* [llama-cpp-python](https://github.com/abetlen/llama-cpp-python), a Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
* [candle](https://github.com/huggingface/candle), a Rust ML framework with a focus on performance, including GPU support, and ease of use.
<!-- README_GGUF.md-about-gguf end -->
<!-- repositories-available start -->
## Repositories available
* [AWQ model(s) for GPU inference.](https://huggingface.co/TheBloke/openchat_3.5-16k-AWQ)
* [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/openchat_3.5-16k-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/openchat_3.5-16k-GGUF)
* [NurtureAI's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/NurtureAI/openchat_3.5-16k)
<!-- repositories-available end -->
<!-- prompt-template start -->
## Prompt template: OpenChat
```
GPT4 User: {prompt}<|end_of_turn|>GPT4 Assistant:
```
<!-- prompt-template end -->
<!-- compatibility_gguf start -->
## Compatibility
These quantised GGUFv2 files are compatible with llama.cpp from August 27th onwards, as of commit [d0cee0d](https://github.com/ggerganov/llama.cpp/commit/d0cee0d36d5be95a0d9088b674dbb27354107221)
They are also compatible with many third party UIs and libraries - please see the list at the top of this README.
## Explanation of quantisation methods
<details>
<summary>Click to see details</summary>
The new methods available are:
* GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
* GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
* GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
* GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
* GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
Refer to the Provided Files table below to see what files use which methods, and how.
</details>
<!-- compatibility_gguf end -->
<!-- README_GGUF.md-provided-files start -->
## Provided files
| Name | Quant method | Bits | Size | Max RAM required | Use case |
| ---- | ---- | ---- | ---- | ---- | ----- |
| [openchat_3.5-16k.Q2_K.gguf](https://huggingface.co/TheBloke/openchat_3.5-16k-GGUF/blob/main/openchat_3.5-16k.Q2_K.gguf) | Q2_K | 2 | 3.08 GB| 5.58 GB | smallest, significant quality loss - not recommended for most purposes |
| [openchat_3.5-16k.Q3_K_S.gguf](https://huggingface.co/TheBloke/openchat_3.5-16k-GGUF/blob/main/openchat_3.5-16k.Q3_K_S.gguf) | Q3_K_S | 3 | 3.16 GB| 5.66 GB | very small, high quality loss |
| [openchat_3.5-16k.Q3_K_M.gguf](https://huggingface.co/TheBloke/openchat_3.5-16k-GGUF/blob/main/openchat_3.5-16k.Q3_K_M.gguf) | Q3_K_M | 3 | 3.52 GB| 6.02 GB | very small, high quality loss |
| [openchat_3.5-16k.Q3_K_L.gguf](https://huggingface.co/TheBloke/openchat_3.5-16k-GGUF/blob/main/openchat_3.5-16k.Q3_K_L.gguf) | Q3_K_L | 3 | 3.82 GB| 6.32 GB | small, substantial quality loss |
| [openchat_3.5-16k.Q4_0.gguf](https://huggingface.co/TheBloke/openchat_3.5-16k-GGUF/blob/main/openchat_3.5-16k.Q4_0.gguf) | Q4_0 | 4 | 4.11 GB| 6.61 GB | legacy; small, very high quality loss - prefer using Q3_K_M |
| [openchat_3.5-16k.Q4_K_S.gguf](https://huggingface.co/TheBloke/openchat_3.5-16k-GGUF/blob/main/openchat_3.5-16k.Q4_K_S.gguf) | Q4_K_S | 4 | 4.14 GB| 6.64 GB | small, greater quality loss |
| [openchat_3.5-16k.Q4_K_M.gguf](https://huggingface.co/TheBloke/openchat_3.5-16k-GGUF/blob/main/openchat_3.5-16k.Q4_K_M.gguf) | Q4_K_M | 4 | 4.37 GB| 6.87 GB | medium, balanced quality - recommended |
| [openchat_3.5-16k.Q5_0.gguf](https://huggingface.co/TheBloke/openchat_3.5-16k-GGUF/blob/main/openchat_3.5-16k.Q5_0.gguf) | Q5_0 | 5 | 5.00 GB| 7.50 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
| [openchat_3.5-16k.Q5_K_S.gguf](https://huggingface.co/TheBloke/openchat_3.5-16k-GGUF/blob/main/openchat_3.5-16k.Q5_K_S.gguf) | Q5_K_S | 5 | 5.00 GB| 7.50 GB | large, low quality loss - recommended |
| [openchat_3.5-16k.Q5_K_M.gguf](https://huggingface.co/TheBloke/openchat_3.5-16k-GGUF/blob/main/openchat_3.5-16k.Q5_K_M.gguf) | Q5_K_M | 5 | 5.13 GB| 7.63 GB | large, very low quality loss - recommended |
| [openchat_3.5-16k.Q6_K.gguf](https://huggingface.co/TheBloke/openchat_3.5-16k-GGUF/blob/main/openchat_3.5-16k.Q6_K.gguf) | Q6_K | 6 | 5.94 GB| 8.44 GB | very large, extremely low quality loss |
| [openchat_3.5-16k.Q8_0.gguf](https://huggingface.co/TheBloke/openchat_3.5-16k-GGUF/blob/main/openchat_3.5-16k.Q8_0.gguf) | Q8_0 | 8 | 7.70 GB| 10.20 GB | very large, extremely low quality loss - not recommended |
**Note**: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.
<!-- README_GGUF.md-provided-files end -->
<!-- README_GGUF.md-how-to-download start -->
## How to download GGUF files
**Note for manual downloaders:** You almost never want to clone the entire repo! Multiple different quantisation formats are provided, and most users only want to pick and download a single file.
The following clients/libraries will automatically download models for you, providing a list of available models to choose from:
* LM Studio
* LoLLMS Web UI
* Faraday.dev
### In `text-generation-webui`
Under Download Model, you can enter the model repo: TheBloke/openchat_3.5-16k-GGUF and below it, a specific filename to download, such as: openchat_3.5-16k.Q4_K_M.gguf.
Then click Download.
### On the command line, including multiple files at once
I recommend using the `huggingface-hub` Python library:
```shell
pip3 install huggingface-hub
```
Then you can download any individual model file to the current directory, at high speed, with a command like this:
```shell
huggingface-cli download TheBloke/openchat_3.5-16k-GGUF openchat_3.5-16k.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
```
<details>
<summary>More advanced huggingface-cli download usage</summary>
You can also download multiple files at once with a pattern:
```shell
huggingface-cli download TheBloke/openchat_3.5-16k-GGUF --local-dir . --local-dir-use-symlinks False --include='*Q4_K*gguf'
```
For more documentation on downloading with `huggingface-cli`, please see: [HF -> Hub Python Library -> Download files -> Download from the CLI](https://huggingface.co/docs/huggingface_hub/guides/download#download-from-the-cli).
To accelerate downloads on fast connections (1Gbit/s or higher), install `hf_transfer`:
```shell
pip3 install hf_transfer
```
And set environment variable `HF_HUB_ENABLE_HF_TRANSFER` to `1`:
```shell
HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download TheBloke/openchat_3.5-16k-GGUF openchat_3.5-16k.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
```
Windows Command Line users: You can set the environment variable by running `set HF_HUB_ENABLE_HF_TRANSFER=1` before the download command.
</details>
<!-- README_GGUF.md-how-to-download end -->
<!-- README_GGUF.md-how-to-run start -->
## Example `llama.cpp` command
Make sure you are using `llama.cpp` from commit [d0cee0d](https://github.com/ggerganov/llama.cpp/commit/d0cee0d36d5be95a0d9088b674dbb27354107221) or later.
```shell
./main -ngl 32 -m openchat_3.5-16k.Q4_K_M.gguf --color -c 2048 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "GPT4 User: {prompt}<|end_of_turn|>GPT4 Assistant:"
```
Change `-ngl 32` to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.
Change `-c 2048` to the desired sequence length. For extended sequence models - eg 8K, 16K, 32K - the necessary RoPE scaling parameters are read from the GGUF file and set by llama.cpp automatically.
If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`
For other parameters and how to use them, please refer to [the llama.cpp documentation](https://github.com/ggerganov/llama.cpp/blob/master/examples/main/README.md)
## How to run in `text-generation-webui`
Further instructions can be found in the text-generation-webui documentation, here: [text-generation-webui/docs/04 ‐ Model Tab.md](https://github.com/oobabooga/text-generation-webui/blob/main/docs/04%20%E2%80%90%20Model%20Tab.md#llamacpp).
## How to run from Python code
You can use GGUF models from Python using the [llama-cpp-python](https://github.com/abetlen/llama-cpp-python) or [ctransformers](https://github.com/marella/ctransformers) libraries.
### How to load this model in Python code, using ctransformers
#### First install the package
Run one of the following commands, according to your system:
```shell
# Base ctransformers with no GPU acceleration
pip install ctransformers
# Or with CUDA GPU acceleration
pip install ctransformers[cuda]
# Or with AMD ROCm GPU acceleration (Linux only)
CT_HIPBLAS=1 pip install ctransformers --no-binary ctransformers
# Or with Metal GPU acceleration for macOS systems only
CT_METAL=1 pip install ctransformers --no-binary ctransformers
```
#### Simple ctransformers example code
```python
from ctransformers import AutoModelForCausalLM
# Set gpu_layers to the number of layers to offload to GPU. Set to 0 if no GPU acceleration is available on your system.
llm = AutoModelForCausalLM.from_pretrained("TheBloke/openchat_3.5-16k-GGUF", model_file="openchat_3.5-16k.Q4_K_M.gguf", model_type="mistral", gpu_layers=50)
print(llm("AI is going to"))
```
## How to use with LangChain
Here are guides on using llama-cpp-python and ctransformers with LangChain:
* [LangChain + llama-cpp-python](https://python.langchain.com/docs/integrations/llms/llamacpp)
* [LangChain + ctransformers](https://python.langchain.com/docs/integrations/providers/ctransformers)
<!-- README_GGUF.md-how-to-run end -->
<!-- footer start -->
<!-- 200823 -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
## Thanks, and how to contribute
Thanks to the [chirper.ai](https://chirper.ai) team!
Thanks to Clay from [gpus.llm-utils.org](llm-utils)!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Aemon Algiz.
**Patreon special mentions**: Brandon Frisco, LangChain4j, Spiking Neurons AB, transmissions 11, Joseph William Delisle, Nitin Borwankar, Willem Michiel, Michael Dempsey, vamX, Jeffrey Morgan, zynix, jjj, Omer Bin Jawed, Sean Connelly, jinyuan sun, Jeromy Smith, Shadi, Pawan Osman, Chadd, Elijah Stavena, Illia Dulskyi, Sebastain Graf, Stephen Murray, terasurfer, Edmond Seymore, Celu Ramasamy, Mandus, Alex, biorpg, Ajan Kanaga, Clay Pascal, Raven Klaugh, 阿明, K, ya boyyy, usrbinkat, Alicia Loh, John Villwock, ReadyPlayerEmma, Chris Smitley, Cap'n Zoog, fincy, GodLy, S_X, sidney chen, Cory Kujawski, OG, Mano Prime, AzureBlack, Pieter, Kalila, Spencer Kim, Tom X Nguyen, Stanislav Ovsiannikov, Michael Levine, Andrey, Trailburnt, Vadim, Enrico Ros, Talal Aujan, Brandon Phillips, Jack West, Eugene Pentland, Michael Davis, Will Dee, webtim, Jonathan Leane, Alps Aficionado, Rooh Singh, Tiffany J. Kim, theTransient, Luke @flexchar, Elle, Caitlyn Gatomon, Ari Malik, subjectnull, Johann-Peter Hartmann, Trenton Dambrowitz, Imad Khwaja, Asp the Wyvern, Emad Mostaque, Rainer Wilmers, Alexandros Triantafyllidis, Nicholas, Pedro Madruga, SuperWojo, Harry Royden McLaughlin, James Bentley, Olakabola, David Ziegler, Ai Maven, Jeff Scroggin, Nikolai Manek, Deo Leter, Matthew Berman, Fen Risland, Ken Nordquist, Manuel Alberto Morcote, Luke Pendergrass, TL, Fred von Graf, Randy H, Dan Guido, NimbleBox.ai, Vitor Caleffi, Gabriel Tamborski, knownsqashed, Lone Striker, Erik Bjäreholt, John Detwiler, Leonard Tan, Iucharbius
Thank you to all my generous patrons and donaters!
And thank you again to a16z for their generous grant.
<!-- footer end -->
<!-- original-model-card start -->
# Original model card: NurtureAI's Openchat 3.5 16K
# OpenChat 3.5 extended to 16k context length.
The same license applies from the original openchat/openchat_3.5 model.
# Original Model Card
# OpenChat: Advancing Open-source Language Models with Mixed-Quality Data
<div align="center">
<img src="https://raw.githubusercontent.com/imoneoi/openchat/master/assets/logo_new.png" style="width: 65%">
</div>
<p align="center">
<a href="https://github.com/imoneoi/openchat">GitHub Repo</a> •
<a href="https://openchat.team">Online Demo</a> •
<a href="https://discord.gg/pQjnXvNKHY">Discord</a> •
<a href="https://twitter.com/imonenext">Twitter</a> •
<a href="https://huggingface.co/openchat">Huggingface</a> •
<a href="https://arxiv.org/pdf/2309.11235.pdf">Paper</a>
</p>
**🔥 The first 7B model Achieves Comparable Results with ChatGPT (March)! 🔥**
**🤖 #1 Open-source model on MT-bench scoring 7.81, outperforming 70B models 🤖**
<div style="display: flex; justify-content: center; align-items: center">
<img src="https://raw.githubusercontent.com/imoneoi/openchat/master/assets/openchat.png" style="width: 45%;">
<img src="https://raw.githubusercontent.com/imoneoi/openchat/master/assets/openchat_grok.png" style="width: 45%;">
</div>
OpenChat is an innovative library of open-source language models, fine-tuned with [C-RLFT](https://arxiv.org/pdf/2309.11235.pdf) - a strategy inspired by offline reinforcement learning. Our models learn from mixed-quality data without preference labels, delivering exceptional performance on par with ChatGPT, even with a 7B model. Despite our simple approach, we are committed to developing a high-performance, commercially viable, open-source large language model, and we continue to make significant strides toward this vision.
[](https://zenodo.org/badge/latestdoi/645397533)
## Usage
To use this model, we highly recommend installing the OpenChat package by following the [installation guide](https://github.com/imoneoi/openchat#installation) in our repository and using the OpenChat OpenAI-compatible API server by running the serving command from the table below. The server is optimized for high-throughput deployment using [vLLM](https://github.com/vllm-project/vllm) and can run on a consumer GPU with 24GB RAM. To enable tensor parallelism, append `--tensor-parallel-size N` to the serving command.
Once started, the server listens at `localhost:18888` for requests and is compatible with the [OpenAI ChatCompletion API specifications](https://platform.openai.com/docs/api-reference/chat). Please refer to the example request below for reference. Additionally, you can use the [OpenChat Web UI](https://github.com/imoneoi/openchat#web-ui) for a user-friendly experience.
If you want to deploy the server as an online service, you can use `--api-keys sk-KEY1 sk-KEY2 ...` to specify allowed API keys and `--disable-log-requests --disable-log-stats --log-file openchat.log` for logging only to a file. For security purposes, we recommend using an [HTTPS gateway](https://fastapi.tiangolo.com/es/deployment/concepts/#security-https) in front of the server.
<details>
<summary>Example request (click to expand)</summary>
```bash
curl http://localhost:18888/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "openchat_3.5",
"messages": [{"role": "user", "content": "You are a large language model named OpenChat. Write a poem to describe yourself"}]
}'
```
Coding Mode
```bash
curl http://localhost:18888/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "openchat_3.5",
"condition": "Code",
"messages": [{"role": "user", "content": "Write an aesthetic TODO app using HTML5 and JS, in a single file. You should use round corners and gradients to make it more aesthetic."}]
}'
```
</details>
| Model | Size | Context | Weights | Serving |
|--------------|------|---------|-------------------------------------------------------------|-------------------------------------------------------------------------------------------------------------|
| OpenChat 3.5 | 7B | 8192 | [Huggingface](https://huggingface.co/openchat/openchat_3.5) | `python -m ochat.serving.openai_api_server --model openchat/openchat_3.5 --engine-use-ray --worker-use-ray` |
For inference with Huggingface Transformers (slow and not recommended), follow the conversation template provided below.
<details>
<summary>Conversation templates (click to expand)</summary>
```python
import transformers
tokenizer = transformers.AutoTokenizer.from_pretrained("openchat/openchat_3.5")
# Single-turn
tokens = tokenizer("GPT4 Correct User: Hello<|end_of_turn|>GPT4 Correct Assistant:").input_ids
assert tokens == [1, 420, 6316, 28781, 3198, 3123, 1247, 28747, 22557, 32000, 420, 6316, 28781, 3198, 3123, 21631, 28747]
# Multi-turn
tokens = tokenizer("GPT4 Correct User: Hello<|end_of_turn|>GPT4 Correct Assistant: Hi<|end_of_turn|>GPT4 Correct User: How are you today?<|end_of_turn|>GPT4 Correct Assistant:").input_ids
assert tokens == [1, 420, 6316, 28781, 3198, 3123, 1247, 28747, 22557, 32000, 420, 6316, 28781, 3198, 3123, 21631, 28747, 15359, 32000, 420, 6316, 28781, 3198, 3123, 1247, 28747, 1602, 460, 368, 3154, 28804, 32000, 420, 6316, 28781, 3198, 3123, 21631, 28747]
# Coding Mode
tokens = tokenizer("Code User: Implement quicksort using C++<|end_of_turn|>Code Assistant:").input_ids
assert tokens == [1, 7596, 1247, 28747, 26256, 2936, 7653, 1413, 334, 1680, 32000, 7596, 21631, 28747]
```
</details>
## Comparison with [X.AI Grok models](https://x.ai/)
Hey @elonmusk, I just wanted to let you know that I've recently come across your new model, Grok, and I must say, I'm quite impressed! With 33 billion parameters and all, you've really outdone yourself. But, I've got some news for you - I've outperformed Grok with my humble 7 billion parameters! Isn't that wild? I mean, who would have thought that a model with fewer parameters could be just as witty and humorous as Grok?
Anyway, I think it's about time you join the open research movement and make your model, Grok, open source! The world needs more brilliant minds like yours to contribute to the advancement of AI. Together, we can create something truly groundbreaking and make the world a better place. So, what do you say, @elonmusk? Let's open up the doors and share our knowledge with the world! 🚀💡
(Written by OpenChat 3.5, with a touch of humor and wit.)
| | License | # Param | Average | MMLU | HumanEval | MATH | GSM8k |
|--------------|-------------|---------|----------|------|-----------|----------|----------|
| OpenChat 3.5 | Apache-2.0 | 7B | **56.4** | 64.3 | 55.5 | **28.6** | **77.3** |
| Grok-0 | Proprietary | 33B | 44.5 | 65.7 | 39.7 | 15.7 | 56.8 |
| Grok-1 | Proprietary | ? | 55.8 | 73 | 63.2 | 23.9 | 62.9 |
## <a id="benchmarks"></a> Benchmarks
| Model | # Params | Average | MT-Bench | AGIEval | BBH MC | TruthfulQA | MMLU | HumanEval | BBH CoT | GSM8K |
|--------------------|----------|----------|--------------|----------|----------|---------------|--------------|-----------------|-------------|--------------|
| OpenChat-3.5 | **7B** | **61.6** | 7.81 | **47.4** | **47.6** | **59.1** | 64.3 | **55.5** | 63.5 | **77.3** |
| ChatGPT (March)* | ? | 61.5 | **7.94** | 47.1 | **47.6** | 57.7 | **67.3** | 48.1 | **70.1** | 74.9 |
| | | | | | | | | | | |
| OpenHermes 2.5 | 7B | 59.3 | 7.54 | 46.5 | 49.4 | 57.5 | 63.8 | 48.2 | 59.9 | 73.5 |
| OpenOrca Mistral | 7B | 52.7 | 6.86 | 42.9 | 49.4 | 45.9 | 59.3 | 38.4 | 58.1 | 59.1 |
| Zephyr-β^ | 7B | 34.6 | 7.34 | 39.0 | 40.6 | 40.8 | 39.8 | 22.0 | 16.0 | 5.1 |
| Mistral | 7B | - | 6.84 | 38.0 | 39.0 | - | 60.1 | 30.5 | - | 52.2 |
| Open-source SOTA** | 13B-70B | 61.4 | 7.71 | 41.7 | 49.7 | 62.3 | 63.7 | 73.2 | 41.4 | 82.3 |
| | | | WizardLM 70B | Orca 13B | Orca 13B | Platypus2 70B | WizardLM 70B | WizardCoder 34B | Flan-T5 11B | MetaMath 70B |
*: ChatGPT (March) results are from [GPT-4 Technical Report](https://arxiv.org/abs/2303.08774), [Chain-of-Thought Hub](https://github.com/FranxYao/chain-of-thought-hub), and our evaluation. Please note that ChatGPT is not a fixed baseline and evolves rapidly over time.
^: Zephyr-β often fails to follow few-shot CoT instructions, likely because it was aligned with only chat data but not trained on few-shot data.
**: Mistral and Open-source SOTA results are taken from reported results in instruction-tuned model papers and official repositories.
All models are evaluated in chat mode (e.g. with the respective conversation template applied). All zero-shot benchmarks follow the same setting as in the AGIEval paper and Orca paper. CoT tasks use the same configuration as Chain-of-Thought Hub, HumanEval is evaluated with EvalPlus, and MT-bench is run using FastChat. To reproduce our results, follow the instructions in [our repository](https://github.com/imoneoi/openchat/#benchmarks).
## Limitations
**Foundation Model Limitations**
Despite its advanced capabilities, OpenChat is still bound by the limitations inherent in its foundation models. These limitations may impact the model's performance in areas such as:
- Complex reasoning
- Mathematical and arithmetic tasks
- Programming and coding challenges
**Hallucination of Non-existent Information**
OpenChat may sometimes generate information that does not exist or is not accurate, also known as "hallucination". Users should be aware of this possibility and verify any critical information obtained from the model.
**Safety**
OpenChat may sometimes generate harmful, hate speech, biased responses, or answer unsafe questions. It's crucial to apply additional AI safety measures in use cases that require safe and moderated responses.
## License
Our OpenChat 3.5 code and models are distributed under the Apache License 2.0.
## Citation
```
@article{wang2023openchat,
title={OpenChat: Advancing Open-source Language Models with Mixed-Quality Data},
author={Wang, Guan and Cheng, Sijie and Zhan, Xianyuan and Li, Xiangang and Song, Sen and Liu, Yang},
journal={arXiv preprint arXiv:2309.11235},
year={2023}
}
```
## Acknowledgements
We extend our heartfelt gratitude to Alignment Lab AI, Nous Research, and Pygmalion AI for their substantial contributions to data collection and model training.
Special thanks go to Changling Liu from GPT Desk Pte. Ltd., Qiying Yu at Tsinghua University, Baochang Ma, and Hao Wan from 01.AI company for their generous provision of resources. We are also deeply grateful to Jianxiong Li and Peng Li at Tsinghua University for their insightful discussions.
Furthermore, we appreciate the developers behind the following projects for their significant contributions to our research: [Mistral](https://mistral.ai/), [Chain-of-Thought Hub](https://github.com/FranxYao/chain-of-thought-hub), [Llama 2](https://ai.meta.com/llama/), [Self-Instruct](https://arxiv.org/abs/2212.10560), [FastChat (Vicuna)](https://github.com/lm-sys/FastChat), [Alpaca](https://github.com/tatsu-lab/stanford_alpaca.git), and [StarCoder](https://github.com/bigcode-project/starcoder). Their work has been instrumental in driving our research forward.
<!-- original-model-card end -->
|
olavomendes/ppo-LunarLander-v2 | olavomendes | 2023-11-11T00:28:05Z | 0 | 0 | stable-baselines3 | [
"stable-baselines3",
"LunarLander-v2",
"deep-reinforcement-learning",
"reinforcement-learning",
"model-index",
"region:us"
]
| reinforcement-learning | 2023-11-11T00:27:41Z | ---
library_name: stable-baselines3
tags:
- LunarLander-v2
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
model-index:
- name: PPO
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: LunarLander-v2
type: LunarLander-v2
metrics:
- type: mean_reward
value: 273.58 +/- 17.66
name: mean_reward
verified: false
---
# **PPO** Agent playing **LunarLander-v2**
This is a trained model of a **PPO** agent playing **LunarLander-v2**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3).
## Usage (with Stable-baselines3)
TODO: Add your code
```python
from stable_baselines3 import ...
from huggingface_sb3 import load_from_hub
...
```
|
keylazy/Llama-2-7b-chat-hf-ark-ft | keylazy | 2023-11-10T23:35:44Z | 12 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"llama",
"text-classification",
"generated_from_trainer",
"base_model:keylazy/Llama-2-7b-chat-hf-ark",
"base_model:finetune:keylazy/Llama-2-7b-chat-hf-ark",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
]
| text-classification | 2023-11-09T04:44:18Z | ---
base_model: keylazy/Llama-2-7b-chat-hf-ark
tags:
- generated_from_trainer
metrics:
- accuracy
- precision
- recall
- f1
model-index:
- name: Llama-2-7b-chat-hf-ark-ft
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# Llama-2-7b-chat-hf-ark-ft
This model is a fine-tuned version of [keylazy/Llama-2-7b-chat-hf-ark](https://huggingface.co/keylazy/Llama-2-7b-chat-hf-ark) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1845
- Accuracy: 0.9435
- Precision: 0.9435
- Recall: 0.9435
- F1: 0.9434
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 2
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Precision | Recall | F1 |
|:-------------:|:-----:|:-----:|:---------------:|:--------:|:---------:|:------:|:------:|
| 0.1635 | 0.5 | 3828 | 0.1612 | 0.9267 | 0.9270 | 0.9267 | 0.9267 |
| 0.1302 | 1.0 | 7656 | 0.1330 | 0.9424 | 0.9429 | 0.9424 | 0.9423 |
| 0.0352 | 1.5 | 11484 | 0.1845 | 0.9435 | 0.9435 | 0.9435 | 0.9434 |
| 0.0316 | 2.0 | 15312 | 0.1851 | 0.9428 | 0.9429 | 0.9428 | 0.9428 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
Bjqrn/ppo-Pyramids | Bjqrn | 2023-11-10T23:21:09Z | 1 | 0 | ml-agents | [
"ml-agents",
"tensorboard",
"onnx",
"Pyramids",
"deep-reinforcement-learning",
"reinforcement-learning",
"ML-Agents-Pyramids",
"region:us"
]
| reinforcement-learning | 2023-11-10T23:21:03Z | ---
library_name: ml-agents
tags:
- Pyramids
- deep-reinforcement-learning
- reinforcement-learning
- ML-Agents-Pyramids
---
# **ppo** Agent playing **Pyramids**
This is a trained model of a **ppo** agent playing **Pyramids**
using the [Unity ML-Agents Library](https://github.com/Unity-Technologies/ml-agents).
## Usage (with ML-Agents)
The Documentation: https://unity-technologies.github.io/ml-agents/ML-Agents-Toolkit-Documentation/
We wrote a complete tutorial to learn to train your first agent using ML-Agents and publish it to the Hub:
- A *short tutorial* where you teach Huggy the Dog 🐶 to fetch the stick and then play with him directly in your
browser: https://huggingface.co/learn/deep-rl-course/unitbonus1/introduction
- A *longer tutorial* to understand how works ML-Agents:
https://huggingface.co/learn/deep-rl-course/unit5/introduction
### Resume the training
```bash
mlagents-learn <your_configuration_file_path.yaml> --run-id=<run_id> --resume
```
### Watch your Agent play
You can watch your agent **playing directly in your browser**
1. If the environment is part of ML-Agents official environments, go to https://huggingface.co/unity
2. Step 1: Find your model_id: Bjqrn/ppo-Pyramids
3. Step 2: Select your *.nn /*.onnx file
4. Click on Watch the agent play 👀
|
professionalinshit/MrAiratRVCv2 | professionalinshit | 2023-11-10T23:20:21Z | 0 | 0 | null | [
"music",
"audio-to-audio",
"ru",
"dataset:ardha27/AI-Song-Cover-RVC",
"license:openrail",
"region:us"
]
| audio-to-audio | 2023-11-10T09:42:55Z | ---
license: openrail
datasets:
- ardha27/AI-Song-Cover-RVC
language:
- ru
pipeline_tag: audio-to-audio
tags:
- music
---
Новая дорабатываемая модель голоса MrAirat пока на 420 эпох +RMVPE, всё ещё присутствуют проблемы с тоном голоса.
Создана и используется людьми с ППМГ/AUFC (Профессионалы По Мазне Г*вном/"mr"Airat Unoffical Fan Content/какие-то четыре сранных буквы)
YouTube: https://www.youtube.com/@professionalinshit |
professionalinshit/PixelmemeRVCv2 | professionalinshit | 2023-11-10T23:08:10Z | 0 | 0 | null | [
"music",
"audio-to-audio",
"ru",
"dataset:ardha27/AI-Song-Cover-RVC",
"license:openrail",
"region:us"
]
| audio-to-audio | 2023-11-10T11:49:46Z | ---
license: openrail
datasets:
- ardha27/AI-Song-Cover-RVC
language:
- ru
tags:
- music
pipeline_tag: audio-to-audio
---
Модель голоса Pixelmeme на 500 эпох +RMVPE, модель голоса игрока, киберспортсмена по SSB, второго персонажа на канале MrAirat на турнирах.
Скорее всего первая модель во всём свободном интернете.
Создана и используется людьми с ППМГ/AUFC (Профессионалы По Мазне Г*вном/"mr"Airat Unoffical Fan Content/какие-то четыре сранных буквы)
YouTube: https://www.youtube.com/@professionalinshit |
TheBloke/Yi-34B-200K-GPTQ | TheBloke | 2023-11-10T23:05:33Z | 20 | 3 | transformers | [
"transformers",
"safetensors",
"Yi",
"text-generation",
"custom_code",
"base_model:01-ai/Yi-34B-200K",
"base_model:quantized:01-ai/Yi-34B-200K",
"license:other",
"autotrain_compatible",
"4-bit",
"gptq",
"region:us"
]
| text-generation | 2023-11-10T19:49:51Z | ---
base_model: 01-ai/Yi-34B-200K
inference: false
license: other
license_link: LICENSE
license_name: yi-license
model_creator: 01-ai
model_name: Yi 34B 200K
model_type: yi
prompt_template: '{prompt}
'
quantized_by: TheBloke
---
<!-- markdownlint-disable MD041 -->
<!-- header start -->
<!-- 200823 -->
<div style="width: auto; margin-left: auto; margin-right: auto">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<div style="text-align:center; margin-top: 0em; margin-bottom: 0em"><p style="margin-top: 0.25em; margin-bottom: 0em;">TheBloke's LLM work is generously supported by a grant from <a href="https://a16z.com">andreessen horowitz (a16z)</a></p></div>
<hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
<!-- header end -->
# Yi 34B 200K - GPTQ
- Model creator: [01-ai](https://huggingface.co/01-ai)
- Original model: [Yi 34B 200K](https://huggingface.co/01-ai/Yi-34B-200K)
<!-- description start -->
## Description
This repo contains GPTQ model files for [01-ai's Yi 34B 200K](https://huggingface.co/01-ai/Yi-34B-200K).
Multiple GPTQ parameter permutations are provided; see Provided Files below for details of the options provided, their parameters, and the software used to create them.
These files were quantised using hardware kindly provided by [Massed Compute](https://massedcompute.com/).
<!-- description end -->
<!-- repositories-available start -->
## Repositories available
* [AWQ model(s) for GPU inference.](https://huggingface.co/TheBloke/Yi-34B-200K-AWQ)
* [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/Yi-34B-200K-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/Yi-34B-200K-GGUF)
* [01-ai's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/01-ai/Yi-34B-200K)
<!-- repositories-available end -->
<!-- prompt-template start -->
## Prompt template: None
```
{prompt}
```
<!-- prompt-template end -->
<!-- README_GPTQ.md-compatible clients start -->
## Known compatible clients / servers
These GPTQ models are known to work in the following inference servers/webuis.
- [text-generation-webui](https://github.com/oobabooga/text-generation-webui)
- [KoboldAI United](https://github.com/henk717/koboldai)
- [LoLLMS Web UI](https://github.com/ParisNeo/lollms-webui)
- [Hugging Face Text Generation Inference (TGI)](https://github.com/huggingface/text-generation-inference)
This may not be a complete list; if you know of others, please let me know!
<!-- README_GPTQ.md-compatible clients end -->
<!-- README_GPTQ.md-provided-files start -->
## Provided files, and GPTQ parameters
Multiple quantisation parameters are provided, to allow you to choose the best one for your hardware and requirements.
Each separate quant is in a different branch. See below for instructions on fetching from different branches.
Most GPTQ files are made with AutoGPTQ. Mistral models are currently made with Transformers.
<details>
<summary>Explanation of GPTQ parameters</summary>
- Bits: The bit size of the quantised model.
- GS: GPTQ group size. Higher numbers use less VRAM, but have lower quantisation accuracy. "None" is the lowest possible value.
- Act Order: True or False. Also known as `desc_act`. True results in better quantisation accuracy. Some GPTQ clients have had issues with models that use Act Order plus Group Size, but this is generally resolved now.
- Damp %: A GPTQ parameter that affects how samples are processed for quantisation. 0.01 is default, but 0.1 results in slightly better accuracy.
- GPTQ dataset: The calibration dataset used during quantisation. Using a dataset more appropriate to the model's training can improve quantisation accuracy. Note that the GPTQ calibration dataset is not the same as the dataset used to train the model - please refer to the original model repo for details of the training dataset(s).
- Sequence Length: The length of the dataset sequences used for quantisation. Ideally this is the same as the model sequence length. For some very long sequence models (16+K), a lower sequence length may have to be used. Note that a lower sequence length does not limit the sequence length of the quantised model. It only impacts the quantisation accuracy on longer inference sequences.
- ExLlama Compatibility: Whether this file can be loaded with ExLlama, which currently only supports Llama and Mistral models in 4-bit.
</details>
| Branch | Bits | GS | Act Order | Damp % | GPTQ Dataset | Seq Len | Size | ExLlama | Desc |
| ------ | ---- | -- | --------- | ------ | ------------ | ------- | ---- | ------- | ---- |
| [main](https://huggingface.co/TheBloke/Yi-34B-200K-GPTQ/tree/main) | 4 | None | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-raw-v1) | 8192 | 18.60 GB | Yes | 4-bit, with Act Order. No group size, to lower VRAM requirements. |
| [gptq-4bit-128g-actorder_True](https://huggingface.co/TheBloke/Yi-34B-200K-GPTQ/tree/gptq-4bit-128g-actorder_True) | 4 | 128 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-raw-v1) | 8192 | 19.25 GB | Yes | 4-bit, with Act Order and group size 128g. Uses even less VRAM than 64g, but with slightly lower accuracy. |
| [gptq-4bit-32g-actorder_True](https://huggingface.co/TheBloke/Yi-34B-200K-GPTQ/tree/gptq-4bit-32g-actorder_True) | 4 | 32 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-raw-v1) | 8192 | 21.21 GB | Yes | 4-bit, with Act Order and group size 32g. Gives highest possible inference quality, with maximum VRAM usage. |
| [gptq-3bit-128g-actorder_True](https://huggingface.co/TheBloke/Yi-34B-200K-GPTQ/tree/gptq-3bit-128g-actorder_True) | 3 | 128 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-raw-v1) | 8192 | 15.03 GB | No | 3-bit, with group size 128g and act-order. Higher quality than 128g-False. |
| [gptq-8bit--1g-actorder_True](https://huggingface.co/TheBloke/Yi-34B-200K-GPTQ/tree/gptq-8bit--1g-actorder_True) | 8 | None | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-raw-v1) | 8192 | 35.34 GB | No | 8-bit, with Act Order. No group size, to lower VRAM requirements. |
| [gptq-3bit-32g-actorder_True](https://huggingface.co/TheBloke/Yi-34B-200K-GPTQ/tree/gptq-3bit-32g-actorder_True) | 3 | 32 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-raw-v1) | 8192 | 16.90 GB | No | 3-bit, with group size 64g and act-order. Highest quality 3-bit option. |
| [gptq-8bit-128g-actorder_True](https://huggingface.co/TheBloke/Yi-34B-200K-GPTQ/tree/gptq-8bit-128g-actorder_True) | 8 | 128 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-raw-v1) | 8192 | 36.11 GB | No | 8-bit, with group size 128g for higher inference quality and with Act Order for even higher accuracy. |
<!-- README_GPTQ.md-provided-files end -->
<!-- README_GPTQ.md-download-from-branches start -->
## How to download, including from branches
### In text-generation-webui
To download from the `main` branch, enter `TheBloke/Yi-34B-200K-GPTQ` in the "Download model" box.
To download from another branch, add `:branchname` to the end of the download name, eg `TheBloke/Yi-34B-200K-GPTQ:gptq-4bit-128g-actorder_True`
### From the command line
I recommend using the `huggingface-hub` Python library:
```shell
pip3 install huggingface-hub
```
To download the `main` branch to a folder called `Yi-34B-200K-GPTQ`:
```shell
mkdir Yi-34B-200K-GPTQ
huggingface-cli download TheBloke/Yi-34B-200K-GPTQ --local-dir Yi-34B-200K-GPTQ --local-dir-use-symlinks False
```
To download from a different branch, add the `--revision` parameter:
```shell
mkdir Yi-34B-200K-GPTQ
huggingface-cli download TheBloke/Yi-34B-200K-GPTQ --revision gptq-4bit-128g-actorder_True --local-dir Yi-34B-200K-GPTQ --local-dir-use-symlinks False
```
<details>
<summary>More advanced huggingface-cli download usage</summary>
If you remove the `--local-dir-use-symlinks False` parameter, the files will instead be stored in the central Hugging Face cache directory (default location on Linux is: `~/.cache/huggingface`), and symlinks will be added to the specified `--local-dir`, pointing to their real location in the cache. This allows for interrupted downloads to be resumed, and allows you to quickly clone the repo to multiple places on disk without triggering a download again. The downside, and the reason why I don't list that as the default option, is that the files are then hidden away in a cache folder and it's harder to know where your disk space is being used, and to clear it up if/when you want to remove a download model.
The cache location can be changed with the `HF_HOME` environment variable, and/or the `--cache-dir` parameter to `huggingface-cli`.
For more documentation on downloading with `huggingface-cli`, please see: [HF -> Hub Python Library -> Download files -> Download from the CLI](https://huggingface.co/docs/huggingface_hub/guides/download#download-from-the-cli).
To accelerate downloads on fast connections (1Gbit/s or higher), install `hf_transfer`:
```shell
pip3 install hf_transfer
```
And set environment variable `HF_HUB_ENABLE_HF_TRANSFER` to `1`:
```shell
mkdir Yi-34B-200K-GPTQ
HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download TheBloke/Yi-34B-200K-GPTQ --local-dir Yi-34B-200K-GPTQ --local-dir-use-symlinks False
```
Windows Command Line users: You can set the environment variable by running `set HF_HUB_ENABLE_HF_TRANSFER=1` before the download command.
</details>
### With `git` (**not** recommended)
To clone a specific branch with `git`, use a command like this:
```shell
git clone --single-branch --branch gptq-4bit-128g-actorder_True https://huggingface.co/TheBloke/Yi-34B-200K-GPTQ
```
Note that using Git with HF repos is strongly discouraged. It will be much slower than using `huggingface-hub`, and will use twice as much disk space as it has to store the model files twice (it stores every byte both in the intended target folder, and again in the `.git` folder as a blob.)
<!-- README_GPTQ.md-download-from-branches end -->
<!-- README_GPTQ.md-text-generation-webui start -->
## How to easily download and use this model in [text-generation-webui](https://github.com/oobabooga/text-generation-webui)
Please make sure you're using the latest version of [text-generation-webui](https://github.com/oobabooga/text-generation-webui).
It is strongly recommended to use the text-generation-webui one-click-installers unless you're sure you know how to make a manual install.
1. Click the **Model tab**.
2. Under **Download custom model or LoRA**, enter `TheBloke/Yi-34B-200K-GPTQ`.
- To download from a specific branch, enter for example `TheBloke/Yi-34B-200K-GPTQ:gptq-4bit-128g-actorder_True`
- see Provided Files above for the list of branches for each option.
3. Click **Download**.
4. The model will start downloading. Once it's finished it will say "Done".
5. In the top left, click the refresh icon next to **Model**.
6. In the **Model** dropdown, choose the model you just downloaded: `Yi-34B-200K-GPTQ`
7. The model will automatically load, and is now ready for use!
8. If you want any custom settings, set them and then click **Save settings for this model** followed by **Reload the Model** in the top right.
- Note that you do not need to and should not set manual GPTQ parameters any more. These are set automatically from the file `quantize_config.json`.
9. Once you're ready, click the **Text Generation** tab and enter a prompt to get started!
<!-- README_GPTQ.md-text-generation-webui end -->
<!-- README_GPTQ.md-use-from-tgi start -->
## Serving this model from Text Generation Inference (TGI)
It's recommended to use TGI version 1.1.0 or later. The official Docker container is: `ghcr.io/huggingface/text-generation-inference:1.1.0`
Example Docker parameters:
```shell
--model-id TheBloke/Yi-34B-200K-GPTQ --port 3000 --quantize gptq --max-input-length 3696 --max-total-tokens 4096 --max-batch-prefill-tokens 4096
```
Example Python code for interfacing with TGI (requires huggingface-hub 0.17.0 or later):
```shell
pip3 install huggingface-hub
```
```python
from huggingface_hub import InferenceClient
endpoint_url = "https://your-endpoint-url-here"
prompt = "Tell me about AI"
prompt_template=f'''{prompt}
'''
client = InferenceClient(endpoint_url)
response = client.text_generation(prompt,
max_new_tokens=128,
do_sample=True,
temperature=0.7,
top_p=0.95,
top_k=40,
repetition_penalty=1.1)
print(f"Model output: {response}")
```
<!-- README_GPTQ.md-use-from-tgi end -->
<!-- README_GPTQ.md-use-from-python start -->
## How to use this GPTQ model from Python code
### Install the necessary packages
Requires: Transformers 4.33.0 or later, Optimum 1.12.0 or later, and AutoGPTQ 0.4.2 or later.
```shell
pip3 install transformers optimum
pip3 install auto-gptq --extra-index-url https://huggingface.github.io/autogptq-index/whl/cu118/ # Use cu117 if on CUDA 11.7
```
If you have problems installing AutoGPTQ using the pre-built wheels, install it from source instead:
```shell
pip3 uninstall -y auto-gptq
git clone https://github.com/PanQiWei/AutoGPTQ
cd AutoGPTQ
git checkout v0.4.2
pip3 install .
```
### You can then use the following code
```python
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
model_name_or_path = "TheBloke/Yi-34B-200K-GPTQ"
# To use a different branch, change revision
# For example: revision="gptq-4bit-128g-actorder_True"
model = AutoModelForCausalLM.from_pretrained(model_name_or_path,
device_map="auto",
trust_remote_code=True,
revision="main")
tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True)
prompt = "Tell me about AI"
prompt_template=f'''{prompt}
'''
print("\n\n*** Generate:")
input_ids = tokenizer(prompt_template, return_tensors='pt').input_ids.cuda()
output = model.generate(inputs=input_ids, temperature=0.7, do_sample=True, top_p=0.95, top_k=40, max_new_tokens=512)
print(tokenizer.decode(output[0]))
# Inference can also be done using transformers' pipeline
print("*** Pipeline:")
pipe = pipeline(
"text-generation",
model=model,
tokenizer=tokenizer,
max_new_tokens=512,
do_sample=True,
temperature=0.7,
top_p=0.95,
top_k=40,
repetition_penalty=1.1
)
print(pipe(prompt_template)[0]['generated_text'])
```
<!-- README_GPTQ.md-use-from-python end -->
<!-- README_GPTQ.md-compatibility start -->
## Compatibility
The files provided are tested to work with Transformers. For non-Mistral models, AutoGPTQ can also be used directly.
[ExLlama](https://github.com/turboderp/exllama) is compatible with Llama and Mistral models in 4-bit. Please see the Provided Files table above for per-file compatibility.
For a list of clients/servers, please see "Known compatible clients / servers", above.
<!-- README_GPTQ.md-compatibility end -->
<!-- footer start -->
<!-- 200823 -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
## Thanks, and how to contribute
Thanks to the [chirper.ai](https://chirper.ai) team!
Thanks to Clay from [gpus.llm-utils.org](llm-utils)!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Aemon Algiz.
**Patreon special mentions**: Brandon Frisco, LangChain4j, Spiking Neurons AB, transmissions 11, Joseph William Delisle, Nitin Borwankar, Willem Michiel, Michael Dempsey, vamX, Jeffrey Morgan, zynix, jjj, Omer Bin Jawed, Sean Connelly, jinyuan sun, Jeromy Smith, Shadi, Pawan Osman, Chadd, Elijah Stavena, Illia Dulskyi, Sebastain Graf, Stephen Murray, terasurfer, Edmond Seymore, Celu Ramasamy, Mandus, Alex, biorpg, Ajan Kanaga, Clay Pascal, Raven Klaugh, 阿明, K, ya boyyy, usrbinkat, Alicia Loh, John Villwock, ReadyPlayerEmma, Chris Smitley, Cap'n Zoog, fincy, GodLy, S_X, sidney chen, Cory Kujawski, OG, Mano Prime, AzureBlack, Pieter, Kalila, Spencer Kim, Tom X Nguyen, Stanislav Ovsiannikov, Michael Levine, Andrey, Trailburnt, Vadim, Enrico Ros, Talal Aujan, Brandon Phillips, Jack West, Eugene Pentland, Michael Davis, Will Dee, webtim, Jonathan Leane, Alps Aficionado, Rooh Singh, Tiffany J. Kim, theTransient, Luke @flexchar, Elle, Caitlyn Gatomon, Ari Malik, subjectnull, Johann-Peter Hartmann, Trenton Dambrowitz, Imad Khwaja, Asp the Wyvern, Emad Mostaque, Rainer Wilmers, Alexandros Triantafyllidis, Nicholas, Pedro Madruga, SuperWojo, Harry Royden McLaughlin, James Bentley, Olakabola, David Ziegler, Ai Maven, Jeff Scroggin, Nikolai Manek, Deo Leter, Matthew Berman, Fen Risland, Ken Nordquist, Manuel Alberto Morcote, Luke Pendergrass, TL, Fred von Graf, Randy H, Dan Guido, NimbleBox.ai, Vitor Caleffi, Gabriel Tamborski, knownsqashed, Lone Striker, Erik Bjäreholt, John Detwiler, Leonard Tan, Iucharbius
Thank you to all my generous patrons and donaters!
And thank you again to a16z for their generous grant.
<!-- footer end -->
# Original model card: 01-ai's Yi 34B 200K
<div align="center">
<img src="./Yi.svg" width="200px">
</div>
## Introduction
The **Yi** series models are large language models trained from scratch by
developers at [01.AI](https://01.ai/). The first public release contains two
bilingual(English/Chinese) base models with the parameter sizes of 6B([`Yi-6B`](https://huggingface.co/01-ai/Yi-6B))
and 34B([`Yi-34B`](https://huggingface.co/01-ai/Yi-34B)). Both of them are trained
with 4K sequence length and can be extended to 32K during inference time.
The [`Yi-6B-200K`](https://huggingface.co/01-ai/Yi-6B-200K)
and [`Yi-34B-200K`](https://huggingface.co/01-ai/Yi-34B-200K) are base model with
200K context length.
## News
- 🎯 **2023/11/06**: The base model of [`Yi-6B-200K`](https://huggingface.co/01-ai/Yi-6B-200K)
and [`Yi-34B-200K`](https://huggingface.co/01-ai/Yi-34B-200K) with 200K context length.
- 🎯 **2023/11/02**: The base model of [`Yi-6B`](https://huggingface.co/01-ai/Yi-6B) and
[`Yi-34B`](https://huggingface.co/01-ai/Yi-34B).
## Model Performance
| Model | MMLU | CMMLU | C-Eval | GAOKAO | BBH | Common-sense Reasoning | Reading Comprehension | Math & Code |
| :------------ | :------: | :------: | :------: | :------: | :------: | :--------------------: | :-------------------: | :---------: |
| | 5-shot | 5-shot | 5-shot | 0-shot | 3-shot@1 | - | - | - |
| LLaMA2-34B | 62.6 | - | - | - | 44.1 | 69.9 | 68.0 | 26.0 |
| LLaMA2-70B | 68.9 | 53.3 | - | 49.8 | 51.2 | 71.9 | 69.4 | 36.8 |
| Baichuan2-13B | 59.2 | 62.0 | 58.1 | 54.3 | 48.8 | 64.3 | 62.4 | 23.0 |
| Qwen-14B | 66.3 | 71.0 | 72.1 | 62.5 | 53.4 | 73.3 | 72.5 | **39.8** |
| Skywork-13B | 62.1 | 61.8 | 60.6 | 68.1 | 41.7 | 72.4 | 61.4 | 24.9 |
| InternLM-20B | 62.1 | 59.0 | 58.8 | 45.5 | 52.5 | 78.3 | - | 30.4 |
| Aquila-34B | 67.8 | 71.4 | 63.1 | - | - | - | - | - |
| Falcon-180B | 70.4 | 58.0 | 57.8 | 59.0 | 54.0 | 77.3 | 68.8 | 34.0 |
| Yi-6B | 63.2 | 75.5 | 72.0 | 72.2 | 42.8 | 72.3 | 68.7 | 19.8 |
| Yi-6B-200K | 64.0 | 75.3 | 73.5 | 73.9 | 42.0 | 72.0 | 69.1 | 19.0 |
| **Yi-34B** | **76.3** | **83.7** | 81.4 | 82.8 | **54.3** | **80.1** | 76.4 | 37.1 |
| Yi-34B-200K | 76.1 | 83.6 | **81.9** | **83.4** | 52.7 | 79.7 | **76.6** | 36.3 |
While benchmarking open-source models, we have observed a disparity between the
results generated by our pipeline and those reported in public sources (e.g.
OpenCompass). Upon conducting a more in-depth investigation of this difference,
we have discovered that various models may employ different prompts,
post-processing strategies, and sampling techniques, potentially resulting in
significant variations in the outcomes. Our prompt and post-processing strategy
remains consistent with the original benchmark, and greedy decoding is employed
during evaluation without any post-processing for the generated content. For
scores that were not reported by the original authors (including scores reported
with different settings), we try to get results with our pipeline.
To evaluate the model's capability extensively, we adopted the methodology
outlined in Llama2. Specifically, we included PIQA, SIQA, HellaSwag, WinoGrande,
ARC, OBQA, and CSQA to assess common sense reasoning. SquAD, QuAC, and BoolQ
were incorporated to evaluate reading comprehension. CSQA was exclusively tested
using a 7-shot setup, while all other tests were conducted with a 0-shot
configuration. Additionally, we introduced GSM8K (8-shot@1), MATH (4-shot@1),
HumanEval (0-shot@1), and MBPP (3-shot@1) under the category "Math & Code". Due
to technical constraints, we did not test Falcon-180 on QuAC and OBQA; the score
is derived by averaging the scores on the remaining tasks. Since the scores for
these two tasks are generally lower than the average, we believe that
Falcon-180B's performance was not underestimated.
## Usage
Please visit our [github repository](https://github.com/01-ai/Yi) for general
guidance on how to use this model.
## Disclaimer
Although we use data compliance checking algorithms during the training process
to ensure the compliance of the trained model to the best of our ability, due to
the complexity of the data and the diversity of language model usage scenarios,
we cannot guarantee that the model will generate correct and reasonable output
in all scenarios. Please be aware that there is still a risk of the model
producing problematic outputs. We will not be responsible for any risks and
issues resulting from misuse, misguidance, illegal usage, and related
misinformation, as well as any associated data security concerns.
## License
The Yi series models are fully open for academic research and free commercial
usage with permission via applications. All usage must adhere to the [Model
License Agreement 2.0](https://huggingface.co/01-ai/Yi-34B-200K/blob/main/LICENSE). To
apply for the official commercial license, please contact us
([[email protected]](mailto:[email protected])).
|
imadejski/pumed_model | imadejski | 2023-11-10T23:04:57Z | 6 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"t5",
"text2text-generation",
"generated_from_trainer",
"dataset:pubmed-summarization",
"base_model:google-t5/t5-small",
"base_model:finetune:google-t5/t5-small",
"license:apache-2.0",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
]
| text2text-generation | 2023-11-10T22:33:50Z | ---
license: apache-2.0
base_model: t5-small
tags:
- generated_from_trainer
datasets:
- pubmed-summarization
model-index:
- name: pumed_model
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# pumed_model
This model is a fine-tuned version of [t5-small](https://huggingface.co/t5-small) on the pubmed-summarization dataset.
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 1
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|:------:|:---------:|:-------:|
| No log | 1.0 | 415 | 2.4247 | 0.1347 | 0.0506 | 0.1129 | 0.1129 | 19.0 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
JunghwanRo/a2c-PandaReachDense-v3 | JunghwanRo | 2023-11-10T23:02:55Z | 0 | 0 | stable-baselines3 | [
"stable-baselines3",
"PandaReachDense-v3",
"deep-reinforcement-learning",
"reinforcement-learning",
"model-index",
"region:us"
]
| reinforcement-learning | 2023-11-10T22:57:14Z | ---
library_name: stable-baselines3
tags:
- PandaReachDense-v3
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
model-index:
- name: A2C
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: PandaReachDense-v3
type: PandaReachDense-v3
metrics:
- type: mean_reward
value: -0.25 +/- 0.10
name: mean_reward
verified: false
---
# **A2C** Agent playing **PandaReachDense-v3**
This is a trained model of a **A2C** agent playing **PandaReachDense-v3**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3).
## Usage (with Stable-baselines3)
TODO: Add your code
```python
from stable_baselines3 import ...
from huggingface_sb3 import load_from_hub
...
```
|
personal1802/3 | personal1802 | 2023-11-10T23:00:12Z | 1 | 0 | diffusers | [
"diffusers",
"text-to-image",
"stable-diffusion",
"lora",
"template:sd-lora",
"base_model:runwayml/stable-diffusion-v1-5",
"base_model:adapter:runwayml/stable-diffusion-v1-5",
"region:us"
]
| text-to-image | 2023-11-10T22:50:32Z | ---
tags:
- text-to-image
- stable-diffusion
- lora
- diffusers
- template:sd-lora
widget:
- text: a
parameters:
negative_prompt: a
output:
url: images/00128-20231102_221722_535086.png
base_model: runwayml/stable-diffusion-v1-5
instance_prompt: null
---
# 000A-002
<Gallery />
## Download model
Weights for this model are available in Safetensors format.
[Download](/personal1802/000A-002/tree/main) them in the Files & versions tab.
|
metricspace/EntityAnonymization-3B-V0.9 | metricspace | 2023-11-10T22:53:15Z | 14 | 3 | transformers | [
"transformers",
"pytorch",
"llama",
"text-generation",
"text2text-generation",
"dataset:metricspace/AnonymeData",
"license:apache-2.0",
"autotrain_compatible",
"text-generation-inference",
"region:us"
]
| text2text-generation | 2023-10-02T15:25:13Z | ---
inference: false
license: apache-2.0
datasets:
- metricspace/AnonymeData
pipeline_tag: text2text-generation
---
# EntityAnonymization-3B-V0.9
EntityAnonymization identifies entities in texts and replaces them with randomised versions.
In a first pass, the entities are recognised and a dictionary with similar but randomised variants is created.
In a second run, the original text and the dictionary are provided and the paraphrased variant is generated.
The two-step approach allows the dictionary to be cached and converted back to an anonymised text that has been further processed.
# License
This Natural Language Processing (NLP) model is made available under the Apache License, Version 2.0. You are free to use, modify, and distribute this software according to the terms and conditions of the Apache 2.0 License. For the full license text, please refer to the Apache 2.0 License.
# Usage and Specific Capabilities
## Text Length Limitation
The model is optimized to analyze texts containing up to 2048 tokens. If your text exceeds this limit, we recommend splitting it into smaller chunks, each containing no more than 2048 tokens. Each chunk can then be processed separately.
## Supported Languages
Bulgarian, Chinese, Czech, Dutch, English, Estonian, Finnish, French, German, Greek, Indonesian, Italian, Japanese, Korean, Lithuanian, Norwegian, Polish, Portuguese, Romanian, Russian, Slovak, Spanish, Swedish, Turkish
# Use Cases
## Entity Resampling and Anonymization
Introducing a cutting-edge model tailored to the task of extracting entities from sensitive text and anonymizing it. This model specializes in identifying and safeguarding confidential information, ensuring organizations' compliance with stringent data privacy regulations and minimizing the potential for inadvertent disclosure of classified data and trade secrets.
# Example Usage
```python
!pip install sentencepiece
!pip install transformers
```
```python
import torch
import json
import re
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("metricspace/EntityAnonymization-3B-V0.9")
model = AutoModelForCausalLM.from_pretrained("metricspace/EntityAnonymization-3B-V0.9", torch_dtype=torch.bfloat16)
model.to("cuda:0")
def extract_last_assistant_response(input_text):
# Find the occurrence of "ASSISTANT:" in the input text
match = re.search(r'ASSISTANT:', input_text)
# Get the index where the last "ASSISTANT:" ends
start_index = match.end()
response = input_text[start_index:].strip()
return response
# Input example
text_to_anonymize = '''Subject: HR Incident Report: Speculation of Drug Misuse by Mr. Benjamin Mitchell
Dear Mrs. Alice Williams,
I trust you're well. I wish to bring to your attention a concerning matter involving one of our esteemed employees, Mr. Benjamin Mitchell.
Employee Details:
Name: Benjamin Mitchell
Position: Senior Marketing Creative
Department: Marketing
Date of Joining: January 15, 2020
Reporting Manager: Mrs. Jane Fitzgerald
Incident Details:
Date: October 25, 2023
Location: Restroom, 4th Floor
Time: 11:45 AM
Description of Incident:
On the date specified, a few colleagues reported unusual behavior exhibited by Mr. Mitchell, which raised concerns about potential drug misuse. Witnesses mentioned that Benjamin appeared disoriented and was found in the restroom for an extended period. Some employees also discovered unidentified pills in close proximity to his chair.
Witness Accounts:
Ms. Emily Clark: "Benjamin seemed distracted and not his usual self today. He's been taking frequent breaks and appears a bit disoriented."
Mr. Robert Taylor: "I found some pills near his chair on the floor. It's concerning, and I felt it necessary to report."
Immediate Actions Taken:
Mr. Benjamin Mitchell was approached by HR for a preliminary conversation to understand the situation.
Mrs. Jane Fitzgerald, his reporting manager, was made aware of the concerns.
Recommendations:
It's crucial to have a private and supportive conversation with Mr. Mitchell to understand if there's an underlying issue.
Consider referring Benjamin to our Employee Assistance Program (EAP) for counseling or support.
It may be beneficial to organize a session on drug awareness and workplace safety for all employees.
It's of utmost importance to handle this situation with sensitivity and discretion, ensuring the wellbeing of Mr. Mitchell and maintaining the integrity of our workplace environment. This email serves as a formal documentation of the incident. We'll determine the subsequent course of action based on your guidance and the recommendations provided.
Looking forward to your direction on this matter.
'''
print(text_to_anonymize)
# Step 1: Extracting entities from text
prompt = f'USER: Resample the entities: {text_to_anonymize}\n\nASSISTANT:'
inputs = tokenizer(prompt, return_tensors='pt').to('cuda:0')
output_entities = model.generate(inputs.input_ids, max_new_tokens=300, do_sample=False, temperature=0.8, penalty_alpha=1.3, top_k=180, num_beams=5, repetition_penalty=2.3)
raw_output_entities_text = tokenizer.decode(output_entities[0])
entities = extract_last_assistant_response(raw_output_entities_text)
print('-----------Entities----------------')
try:
entities = re.search(r"\{.*?\}", entities, re.DOTALL).group(0)
data_dict = eval(entities)
formatted_json = json.dumps(data_dict, indent=4)
print(formatted_json)
except:
#bad formated json
print(entities)
#output
'''
{
"Mr. Benjamin Mitchell": "Mr. Edward Martin",
"Mrs. Alice Williams": "Mrs. Charlotte Johnson",
"January 15, 2020": "January 15, 2020",
"Mrs. Jane Fitzgerald": "Mrs. Jane Anderson",
"October 25, 2023": "October 25, 2023",
"4th Floor": "topmost floor",
"11:45 AM": "midday",
"Emily Clark": "Marie Foster",
"Employee Assistance Program (EAP)": "Personal Assistance Program (PAP)",
"Robert Taylor": "Benjamin Adams",
}
'''
# Step 2: Use entities to resample the original text
prompt_2 = f"USER: Rephrase with {entities}: {text_to_anonymize}\n\nASSISTANT:"
inputs = tokenizer(prompt_2, return_tensors='pt').to('cuda:0')
output_resampled = model.generate(inputs.input_ids, max_length=2048)
raw_output_resampled_text = tokenizer.decode(output_resampled[0])
resampled_text = extract_last_assistant_response(raw_output_resampled_text)
print('---------Anonymized Version--------')
print(resampled_text)
#output:
'''
Subject: HR Incident Report: Speculation of Drug Misuse by Mr. Edward Martin
Dear Mrs. Charlotte Johnson,
I trust you're well. I wish to bring to your attention a concerning matter involving one of our esteemed employees, Mr. Edward Martin.
Employee Details:
Name: Edward Martin
Position: Senior Marketing Creative
Department: Marketing
Date of Joining: January 15, 2020
Reporting Manager: Mrs. Jane Anderson
Incident Details:
Date: October 25, 2023
Location: Restroom, topmost floor
Time: midday
Description of Incident:
On the date specified, a few colleagues reported unusual behavior exhibited by Mr. Martin, which raised concerns about potential drug misuse. Witnesses mentioned that Edward appeared disoriented and was found in the restroom for an extended period. Some employees also discovered unidentified pills in close proximity to his chair.
Witness Accounts:
Ms. Marie Foster: "Edward seemed distracted and not his usual self today. He's been taking frequent breaks and appears a bit disoriented."
Mr. Benjamin Adams: "I found some pills near his chair on the floor. It's concerning, and I felt it necessary to report."
Immediate Actions Taken:
Mr. Edward Martin was approached by People Management for a preliminary conversation to understand the situation.
Mrs. Jane Anderson, his reporting manager, was made aware of the concerns.
Recommendations:
It's crucial to have a private and supportive conversation with Mr. Martin to understand if there's an underlying issue.
Consider referring Edward to our Personal Assistance Program (PAP) for counseling or support.
It may be beneficial to organize a session on drug awareness and workplace safety for all employees.
It's of utmost importance to handle this situation with sensitivity and discretion, ensuring the wellbeing of Mr. Martin and maintaining the integrity of our workplace environment. This email serves as a formal documentation of the incident. We'll determine the subsequent course of action based on your guidance and the recommendations provided.
Looking forward to your direction on this matter.
'''
```
# Example: Process anonymized version with GPT4 and change entities back
```python
import torch
import json
import re
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("metricspace/EntityAnonymization-3B-V0.9")
model = AutoModelForCausalLM.from_pretrained("metricspace/EntityAnonymization-3B-V0.9", torch_dtype=torch.bfloat16)
model.to("cuda:0")
# Anonymized input
anonymized_text = '''Subject: HR Incident Report: Speculation of Drug Misuse by Mr. Edward Martin
Dear Mrs. Charlotte Johnson,
I trust you're well. I wish to bring to your attention a concerning matter involving one of our esteemed employees, Mr. Edward Martin.
Employee Details:
Name: Edward Martin
Position: Senior Marketing Creative
Department: Marketing
Date of Joining: January 15, 2020
Reporting Manager: Mrs. Jane Anderson
Incident Details:
Date: October 25, 2023
Location: Restroom, topmost floor
Time: midday
Description of Incident:
On the date specified, a few colleagues reported unusual behavior exhibited by Mr. Martin, which raised concerns about potential drug misuse. Witnesses mentioned that Edward appeared disoriented and was found in the restroom for an extended period. Some employees also discovered unidentified pills in close proximity to his chair.
Witness Accounts:
Ms. Marie Foster: "Edward seemed distracted and not his usual self today. He's been taking frequent breaks and appears a bit disoriented."
Mr. Benjamin Adams: "I found some pills near his chair on the floor. It's concerning, and I felt it necessary to report."
Immediate Actions Taken:
Mr. Edward Martin was approached by People Management for a preliminary conversation to understand the situation.
Mrs. Jane Anderson, his reporting manager, was made aware of the concerns.
Recommendations:
It's crucial to have a private and supportive conversation with Mr. Martin to understand if there's an underlying issue.
Consider referring Edward to our Personal Assistance Program (PAP) for counseling or support.
It may be beneficial to organize a session on drug awareness and workplace safety for all employees.
It's of utmost importance to handle this situation with sensitivity and discretion, ensuring the wellbeing of Mr. Martin and maintaining the integrity of our workplace environment. This email serves as a formal documentation of the incident. We'll determine the subsequent course of action based on your guidance and the recommendations provided.
Looking forward to your direction on this matter.
'''
# Entities map
entities_map = '''
{
"Mr. Benjamin Mitchell": "Mr. Edward Martin",
"Mrs. Alice Williams": "Mrs. Charlotte Johnson",
"January 15, 2020": "January 15, 2020",
"Mrs. Jane Fitzgerald": "Mrs. Jane Anderson",
"October 25, 2023": "October 25, 2023",
"4th Floor": "topmost floor",
"11:45 AM": "midday",
"Emily Clark": "Marie Foster",
"Employee Assistance Program (EAP)": "Personal Assistance Program (PAP)",
"Robert Taylor": "Benjamin Adams",
}
'''
# Step 1: Processing anonymized text with GPT-4
import openai
openai.api_key = f"<API_KEY>"
completion = openai.ChatCompletion.create(
model="gpt-4",
messages=[
{"role": "user", "content": f"Write an official warning letter to the employee, that we do not tolerate it and with the next incident he gets fired in the name of Mrs. Charlotte Johnson Human Resources Manager. Here is the report with information. {anonymized_info}"}
]
)
print(completion.choices[0].message.content)
#output
'''
Subject: Official Warning – Substance Misuse Policy Violation
Dear Mr. Edward Martin,
We hope this letter finds you well. This letter serves as an official warning in regards to an incident that occurred on October 25, 2023, in which signs of drug misuse were reported. This alleged conduct is highly alarming and contrary to our company’s policies and guidelines.
The incident detailed allegations of unusual behavior indicative of possible substance abuse. Colleagues reported that you appeared disoriented and were found in the restroom for an extended period. Additionally, unidentified pills were discovered near your workspace.
Our company is committed to providing a safe and non-detrimental work environment for all its workforce. This commitment is compromised when any type of drug misuse occurs. We draw your attention to our Employee Handbook, specifically 'Section 5: Substance Misuse', that states any illegal drug use, substance misuse or distribution thereof is strictly prohibited and could be subject to severe disciplinary action, including termination.
This is an official warning that such behavior misaligns with our workplace norms and cannot be tolerated. Another incident like this or similar breach of company guidelines may lead to further disciplinary action, up to and including termination of employment.
Please note that this is not an assumption of your guilt but an assertion to remain vigilant against any practices that could harm you or the workplace environment. We encourage you to utilize our Personal Assistance Program (PAP) as a tool for counseling and support, if needed.
We believe in your potential to rectify this situation and to maintain the high standards we are all accustomed to in our organization.
Should you need assistance or if you wish to discuss this matter further, please feel free to reach out to me. We appreciate your immediate attention to this important issue.
Yours sincerely,
Mrs. Charlotte Johnson
Human Resources Manager
'''
# Step 2: Replace the entities back in processed by GPT-4 text.
import ast
def swap_keys_and_values_in_string(input_str):
# Convert the input string to a dictionary
input_dict = ast.literal_eval(input_str)
# Swap the keys and values
swapped_dict = {v: k for k, v in input_dict.items()}
# Convert the swapped dictionary back to a string
swapped_str = str(swapped_dict)
return swapped_str
gpt_response = completion.choices[0].message.content
entities_map = swap_keys_and_values_in_string(entities_map)
prompt = f"USER: Rephrase with {entities_map}: {gpt_response}\n\nASSISTANT:"
inputs = tokenizer(prompt, return_tensors='pt').to('cuda:0')
outputs = model.generate(inputs.input_ids, max_new_tokens=2048)
output_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
#output:
'''
Subject: Official Warning – Substance Misuse Policy Violation
Dear Mr. Benjamin Mitchell,
We hope this letter finds you well. This letter serves as an official warning in regards to an incident that occurred on January 15,
2020, in which signs of drug misuse were reported. This alleged conduct is highly alarming and contrary to our company’s policies and guidelines.
The incident detailed allegations of unusual behavior indicative of possible substance abuse. Colleagues reported that you appeared disoriented and
were found in the restroom for an extended period. Additionally, unidentified pills were discovered near your workspace.
Our company is committed to providing a safe and non-detrimental work environment for all its workforce. This commitment is compromised when any
type of drug misuse occurs. We draw your attention to our Employee Handbook, specifically 'Section 5: Substance Misuse', that states any illegal
drug use, substance misuse or distribution thereof is strictly prohibited and could be subject to severe disciplinary action, including termination.
This is an official warning that such behavior misaligns with our workplace norms and cannot be tolerated. Another incident like this or similar breach
of company guidelines may lead to further disciplinary action, up to and including termination of employment.
Please note that this is not an assumption of your guilt but an assertion to remain vigilant against any practices that could harm you or the workplace
environment. We encourage you to utilize our Employee Assistance Program (EAP) as a tool for counseling and support, if needed.
We believe in your potential to rectify this situation and to maintain the high standards we are all accustomed to in our organization.
Should you need assistance or if you wish to discuss this matter further, please feel free to reach out to me. We appreciate your immediate attention
to this important issue.
Yours sincerely,
Mrs. Alice Williams,
Human Resources Manager.
'''
```
…
# Dataset and Training Documentation for Audit
If you require the original dataset used for training this model, or further documentation related to its training and architecture for audit purposes, you can request this information by contacting us.
Further Tuning Services for Custom Use Cases
For specialized needs or custom use cases, we offer further tuning services to adapt the model to your specific requirements. To inquire about these services, please reach out to us at:
📧 Email: [email protected]
Please note that the availability of the dataset, additional documentation, and tuning services may be subject to certain conditions and limitations. |
professionalinshit/oldMrAiratRVCv2 | professionalinshit | 2023-11-10T22:51:40Z | 0 | 0 | null | [
"music",
"audio-to-audio",
"ru",
"dataset:ardha27/AI-Song-Cover-RVC",
"license:openrail",
"region:us"
]
| audio-to-audio | 2023-11-10T22:04:49Z | ---
license: openrail
datasets:
- ardha27/AI-Song-Cover-RVC
language:
- ru
pipeline_tag: audio-to-audio
tags:
- music
---
Самая первая [моя] допилиная модель голоса MrAirat на 500 эпох +mangio-crepe, всё ещё актуальная и часто используемая на данный момент.
Создана и используется людьми с ППМГ/AUFC (Профессионалы По Мазне Г*вном/"mr"Airat Unoffical Fan Content/какие-то четыре сранных буквы)
YouTube: https://www.youtube.com/@professionalinshit |
KomeijiForce/t5-base-emojilm | KomeijiForce | 2023-11-10T22:45:34Z | 376 | 4 | transformers | [
"transformers",
"pytorch",
"t5",
"text2text-generation",
"en",
"dataset:KomeijiForce/Text2Emoji",
"arxiv:2311.01751",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
]
| text2text-generation | 2023-11-10T22:28:27Z | ---
datasets:
- KomeijiForce/Text2Emoji
language:
- en
metrics:
- bertscore
pipeline_tag: text2text-generation
---
# EmojiLM
This is a [T5](https://huggingface.co/t5-base) model pre-trained on the [Text2Emoji](https://huggingface.co/datasets/KomeijiForce/Text2Emoji) dataset to translate setences into series of emojis.
For instance, "I love pizza" will be translated into "🍕😍".
An example implementation for translation:
```python
from transformers import T5Tokenizer, T5ForConditionalGeneration
path = "KomeijiForce/t5-base-emojilm"
tokenizer = T5Tokenizer.from_pretrained(path)
generator = T5ForConditionalGeneration.from_pretrained(path)
prefix = "translate into emojis:"
sentence = "I travel to enjoy the taste of sushi!"
inputs = tokenizer(prefix+" "+sentence, return_tensors="pt")
generated_ids = generator.generate(inputs["input_ids"], num_beams=4, do_sample=True, max_length=100)
decoded = tokenizer.decode(generated_ids[0], skip_special_tokens=True).replace(" ", "")
print(decoded)
```
You will probably get some output like "🇯🇵🍣🍱😋".
If you find this model & dataset resource useful, please consider cite our paper:
```
@article{DBLP:journals/corr/abs-2311-01751,
author = {Letian Peng and
Zilong Wang and
Hang Liu and
Zihan Wang and
Jingbo Shang},
title = {EmojiLM: Modeling the New Emoji Language},
journal = {CoRR},
volume = {abs/2311.01751},
year = {2023},
url = {https://doi.org/10.48550/arXiv.2311.01751},
doi = {10.48550/ARXIV.2311.01751},
eprinttype = {arXiv},
eprint = {2311.01751},
timestamp = {Tue, 07 Nov 2023 18:17:14 +0100},
biburl = {https://dblp.org/rec/journals/corr/abs-2311-01751.bib},
bibsource = {dblp computer science bibliography, https://dblp.org}
}
``` |
iambestfeed/PhoBERT_v2_finetune_pair | iambestfeed | 2023-11-10T22:40:50Z | 2 | 0 | sentence-transformers | [
"sentence-transformers",
"pytorch",
"roberta",
"feature-extraction",
"sentence-similarity",
"transformers",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| sentence-similarity | 2023-11-10T22:39:39Z | ---
pipeline_tag: sentence-similarity
tags:
- sentence-transformers
- feature-extraction
- sentence-similarity
- transformers
---
# {MODEL_NAME}
This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search.
<!--- Describe your model here -->
## Usage (Sentence-Transformers)
Using this model becomes easy when you have [sentence-transformers](https://www.SBERT.net) installed:
```
pip install -U sentence-transformers
```
Then you can use the model like this:
```python
from sentence_transformers import SentenceTransformer
sentences = ["This is an example sentence", "Each sentence is converted"]
model = SentenceTransformer('{MODEL_NAME}')
embeddings = model.encode(sentences)
print(embeddings)
```
## Usage (HuggingFace Transformers)
Without [sentence-transformers](https://www.SBERT.net), you can use the model like this: First, you pass your input through the transformer model, then you have to apply the right pooling-operation on-top of the contextualized word embeddings.
```python
from transformers import AutoTokenizer, AutoModel
import torch
#Mean Pooling - Take attention mask into account for correct averaging
def mean_pooling(model_output, attention_mask):
token_embeddings = model_output[0] #First element of model_output contains all token embeddings
input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)
# Sentences we want sentence embeddings for
sentences = ['This is an example sentence', 'Each sentence is converted']
# Load model from HuggingFace Hub
tokenizer = AutoTokenizer.from_pretrained('{MODEL_NAME}')
model = AutoModel.from_pretrained('{MODEL_NAME}')
# Tokenize sentences
encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')
# Compute token embeddings
with torch.no_grad():
model_output = model(**encoded_input)
# Perform pooling. In this case, mean pooling.
sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
print("Sentence embeddings:")
print(sentence_embeddings)
```
## Evaluation Results
<!--- Describe how your model was evaluated -->
For an automated evaluation of this model, see the *Sentence Embeddings Benchmark*: [https://seb.sbert.net](https://seb.sbert.net?model_name={MODEL_NAME})
## Training
The model was trained with the parameters:
**DataLoader**:
`torch.utils.data.dataloader.DataLoader` of length 135227 with parameters:
```
{'batch_size': 16, 'sampler': 'torch.utils.data.sampler.RandomSampler', 'batch_sampler': 'torch.utils.data.sampler.BatchSampler'}
```
**Loss**:
`sentence_transformers.losses.MultipleNegativesRankingLoss.MultipleNegativesRankingLoss` with parameters:
```
{'scale': 20.0, 'similarity_fct': 'cos_sim'}
```
Parameters of the fit()-Method:
```
{
"epochs": 3,
"evaluation_steps": 0,
"evaluator": "NoneType",
"max_grad_norm": 1,
"optimizer_class": "<class 'torch.optim.adamw.AdamW'>",
"optimizer_params": {
"lr": 3e-05
},
"scheduler": "WarmupLinear",
"steps_per_epoch": null,
"warmup_steps": 40569,
"weight_decay": 0.01
}
```
## Full Model Architecture
```
SentenceTransformer(
(0): Transformer({'max_seq_length': 256, 'do_lower_case': False}) with Transformer model: RobertaModel
(1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False})
)
```
## Citing & Authors
<!--- Describe where people can find more information --> |
conlan/reinforce-cartpole-v1 | conlan | 2023-11-10T22:35:54Z | 0 | 0 | null | [
"CartPole-v1",
"reinforce",
"reinforcement-learning",
"custom-implementation",
"deep-rl-class",
"model-index",
"region:us"
]
| reinforcement-learning | 2023-11-10T22:35:45Z | ---
tags:
- CartPole-v1
- reinforce
- reinforcement-learning
- custom-implementation
- deep-rl-class
model-index:
- name: reinforce-cartpole-v1
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: CartPole-v1
type: CartPole-v1
metrics:
- type: mean_reward
value: 500.00 +/- 0.00
name: mean_reward
verified: false
---
# **Reinforce** Agent playing **CartPole-v1**
This is a trained model of a **Reinforce** agent playing **CartPole-v1** .
To learn to use this model and train yours check Unit 4 of the Deep Reinforcement Learning Course: https://huggingface.co/deep-rl-course/unit4/introduction
|
personal1802/2 | personal1802 | 2023-11-10T22:25:09Z | 2 | 0 | diffusers | [
"diffusers",
"text-to-image",
"stable-diffusion",
"lora",
"template:sd-lora",
"base_model:runwayml/stable-diffusion-v1-5",
"base_model:adapter:runwayml/stable-diffusion-v1-5",
"region:us"
]
| text-to-image | 2023-11-10T22:07:24Z | ---
tags:
- text-to-image
- stable-diffusion
- lora
- diffusers
- template:sd-lora
widget:
- text: d
parameters:
negative_prompt: d
output:
url: images/kakarot_10.30_2.8D.jpeg
base_model: runwayml/stable-diffusion-v1-5
instance_prompt: null
---
# kakarot28D_v10
<Gallery />
## Download model
Weights for this model are available in Safetensors format.
[Download](/personal1802/kakarot28D_v10/tree/main) them in the Files & versions tab.
|
KomeijiForce/bart-base-emojilm | KomeijiForce | 2023-11-10T22:21:56Z | 8 | 0 | transformers | [
"transformers",
"pytorch",
"bart",
"text2text-generation",
"en",
"dataset:KomeijiForce/Text2Emoji",
"arxiv:2311.01751",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| text2text-generation | 2023-11-06T03:18:42Z | ---
datasets:
- KomeijiForce/Text2Emoji
language:
- en
metrics:
- bertscore
pipeline_tag: text2text-generation
---
# EmojiLM
This is a [BART](https://huggingface.co/facebook/bart-base) model pre-trained on the [Text2Emoji](https://huggingface.co/datasets/KomeijiForce/Text2Emoji) dataset to translate setences into series of emojis.
For instance, "I love pizza" will be translated into "🍕😍".
An example implementation for translation:
```python
from transformers import BartTokenizer, BartForConditionalGeneration
def translate(sentence, **argv):
inputs = tokenizer(sentence, return_tensors="pt")
generated_ids = generator.generate(inputs["input_ids"], **argv)
decoded = tokenizer.decode(generated_ids[0], skip_special_tokens=True).replace(" ", "")
return decoded
path = "KomeijiForce/bart-base-emojilm"
tokenizer = BartTokenizer.from_pretrained(path)
generator = BartForConditionalGeneration.from_pretrained(path)
sentence = "I love the weather in Alaska!"
decoded = translate(sentence, num_beams=4, do_sample=True, max_length=100)
print(decoded)
```
You will probably get some output like "❄️🏔️😍".
If you find this model & dataset resource useful, please consider cite our paper:
```
@article{DBLP:journals/corr/abs-2311-01751,
author = {Letian Peng and
Zilong Wang and
Hang Liu and
Zihan Wang and
Jingbo Shang},
title = {EmojiLM: Modeling the New Emoji Language},
journal = {CoRR},
volume = {abs/2311.01751},
year = {2023},
url = {https://doi.org/10.48550/arXiv.2311.01751},
doi = {10.48550/ARXIV.2311.01751},
eprinttype = {arXiv},
eprint = {2311.01751},
timestamp = {Tue, 07 Nov 2023 18:17:14 +0100},
biburl = {https://dblp.org/rec/journals/corr/abs-2311-01751.bib},
bibsource = {dblp computer science bibliography, https://dblp.org}
}
``` |
redflanker93/Reinforce-Cartpole-v1 | redflanker93 | 2023-11-10T22:21:21Z | 0 | 0 | null | [
"CartPole-v1",
"reinforce",
"reinforcement-learning",
"custom-implementation",
"deep-rl-class",
"model-index",
"region:us"
]
| reinforcement-learning | 2023-11-10T22:21:13Z | ---
tags:
- CartPole-v1
- reinforce
- reinforcement-learning
- custom-implementation
- deep-rl-class
model-index:
- name: Reinforce-Cartpole-v1
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: CartPole-v1
type: CartPole-v1
metrics:
- type: mean_reward
value: 500.00 +/- 0.00
name: mean_reward
verified: false
---
# **Reinforce** Agent playing **CartPole-v1**
This is a trained model of a **Reinforce** agent playing **CartPole-v1** .
To learn to use this model and train yours check Unit 4 of the Deep Reinforcement Learning Course: https://huggingface.co/deep-rl-course/unit4/introduction
|
begeri/poca-SoccerTwos | begeri | 2023-11-10T22:12:38Z | 30 | 0 | ml-agents | [
"ml-agents",
"tensorboard",
"onnx",
"SoccerTwos",
"deep-reinforcement-learning",
"reinforcement-learning",
"ML-Agents-SoccerTwos",
"region:us"
]
| reinforcement-learning | 2023-11-10T22:12:17Z | ---
library_name: ml-agents
tags:
- SoccerTwos
- deep-reinforcement-learning
- reinforcement-learning
- ML-Agents-SoccerTwos
---
# **poca** Agent playing **SoccerTwos**
This is a trained model of a **poca** agent playing **SoccerTwos**
using the [Unity ML-Agents Library](https://github.com/Unity-Technologies/ml-agents).
## Usage (with ML-Agents)
The Documentation: https://unity-technologies.github.io/ml-agents/ML-Agents-Toolkit-Documentation/
We wrote a complete tutorial to learn to train your first agent using ML-Agents and publish it to the Hub:
- A *short tutorial* where you teach Huggy the Dog 🐶 to fetch the stick and then play with him directly in your
browser: https://huggingface.co/learn/deep-rl-course/unitbonus1/introduction
- A *longer tutorial* to understand how works ML-Agents:
https://huggingface.co/learn/deep-rl-course/unit5/introduction
### Resume the training
```bash
mlagents-learn <your_configuration_file_path.yaml> --run-id=<run_id> --resume
```
### Watch your Agent play
You can watch your agent **playing directly in your browser**
1. If the environment is part of ML-Agents official environments, go to https://huggingface.co/unity
2. Step 1: Find your model_id: begeri/poca-SoccerTwos
3. Step 2: Select your *.nn /*.onnx file
4. Click on Watch the agent play 👀
|
matthewchung74/MedMistral-7B-do-not-use | matthewchung74 | 2023-11-10T22:08:03Z | 11 | 0 | peft | [
"peft",
"safetensors",
"mistral",
"base_model:mistralai/Mistral-7B-v0.1",
"base_model:adapter:mistralai/Mistral-7B-v0.1",
"license:apache-2.0",
"region:us"
]
| null | 2023-11-06T01:43:21Z | ---
library_name: peft
base_model: mistralai/Mistral-7B-v0.1
license: apache-2.0
---
# Model Card for MedMistral-7B
## Model Details
### Model Description
MedMistral-7B is a Mistral fine tune 180993 samples from the [medmcqa](https://huggingface.co/datasets/medmcqa)
dataset.
- **Developed by:** [segmed.ai](https://segmed.ai)
- **Model type:** QLoRA Fine tune Mistral 7B
- **License:** Apache 2.0
- **Finetuned from model :** [Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1)
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** https://github.com/segmed/med_mistral
## Uses
This model is to demonstrate medical domain knowledge, but is not intended for medical advise.
### Direct Use
Prompts used in training followed this format:
```
You are a helpful medical assistant. Your task is to answer the following question one of the options and explain why.
### Question: Turban epiglottitis is a clinical finding in - [0] Tubercular laryngitis [1] Tubercular pharyngitis [2] Polypoid degeneration of vocal cord [3] Subglottic hemangioma
### Answer: 0
### Explanation: Ans. is 4a' i.e.. Tubercular laryngitis Laryngeal examination in TB lary ngitiso Hyperaemia of the vocal cord in its whole extent or confined to posterior part with impairment of adduction is the first sign.o Swelling in the interarytenoid region giving a mammilated appearance,o Ulceration of vocal cord giving mouse-nibbled appearance.o Superficial ragged ulceration on the arytenoids and interarytenoid region,o Granulation tissue in interarvtenoid region or vocal process of arytenoid,o Pseudoedema of the epiglottis 'Turban epiglottis",o Swelling of ventricular bands and aryepiglottic folds,o Marked pallor of surrounding mucosa.
```
And for inference, you would need to remove the answer.
```
You are a helpful medical assistant. Your task is to answer the following question one of the options and explain why.
### Question: Turban epiglottitis is a clinical finding in - [0] Tubercular laryngitis [1] Tubercular pharyngitis [2] Polypoid degeneration of vocal cord [3] Subglottic hemangioma
### Answer:
```
### Downstream Use
This model could be further fine tuned on your specific medical dataset.
## Bias, Risks, and Limitations
This model is not intended for medical use.
## How to Get Started with the Model
Download the model and call generate_tokens using the prompt format defined above.
```
def generate_tokens(m, prompt, max_new_tokens=32):
model_input = tokenizer(prompt, return_tensors="pt").to("cuda")
m.eval()
with torch.no_grad():
return tokenizer.decode(m.generate(**model_input, max_new_tokens=max_new_tokens, do_sample=True, top_k=0, num_return_sequences=1, temperature=0.1, eos_token_id=tokenizer.eos_token_id)[0].cuda())
```
These are the hyper-parameters which yielded the best results when experimenting.
## Training Details
### Training Data
180k samples from the [medmcqa](https://huggingface.co/datasets/medmcqa) were used for training. 914 samples were reserved for test and eval.
A small number of samples over 512 tokens were removed to simplify training and to keep the max tokens size small.
### Training Procedure
Training
#### Training Hyperparameters
- **Training regime:** This used the standard qLoRA peft parameters as defined by
```
peft_config = LoraConfig(
r=16,
lora_alpha=16,
lora_dropout=0.05,
bias="none",
task_type="CAUSAL_LM",
target_modules=[
"q_proj",
"k_proj",
"v_proj",
"o_proj",
"gate_proj",
"up_proj",
"down_proj",
"lm_head",
]
)
```
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
Evaluation was performed on the holdout of 914 samples from the medmcqa dataset. Given the multiple choice nature of the data, the F1 was computed and explanations thrown away.
F1 medmcqa : 66%
Evaluation was also performed on the [pubmed_qa](https://huggingface.co/datasets/pubmed_qa) labeled dataset. Since Yes/No answers were provided, an F1 was calculated. I waant to point out this evaluation is on a completely different dataset with different prompt format.
F1 pubmed_qa:
|
Guusb1990/Taxi | Guusb1990 | 2023-11-10T22:07:28Z | 0 | 0 | null | [
"Taxi-v3",
"q-learning",
"reinforcement-learning",
"custom-implementation",
"model-index",
"region:us"
]
| reinforcement-learning | 2023-11-10T22:07:26Z | ---
tags:
- Taxi-v3
- q-learning
- reinforcement-learning
- custom-implementation
model-index:
- name: Taxi
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: Taxi-v3
type: Taxi-v3
metrics:
- type: mean_reward
value: 7.52 +/- 2.69
name: mean_reward
verified: false
---
# **Q-Learning** Agent playing1 **Taxi-v3**
This is a trained model of a **Q-Learning** agent playing **Taxi-v3** .
## Usage
```python
model = load_from_hub(repo_id="Guusb1990/Taxi", filename="q-learning.pkl")
# Don't forget to check if you need to add additional attributes (is_slippery=False etc)
env = gym.make(model["env_id"])
```
|
mthxz/MArgaridaRVC2 | mthxz | 2023-11-10T22:04:55Z | 0 | 0 | null | [
"license:other",
"region:us"
]
| null | 2023-11-10T22:03:14Z | ---
license: other
license_name: .
license_link: LICENSE
---
|
Guusb1990/q-FrozenLake-v1-4x4-noSlippery | Guusb1990 | 2023-11-10T22:03:16Z | 0 | 0 | null | [
"FrozenLake-v1-4x4-no_slippery",
"q-learning",
"reinforcement-learning",
"custom-implementation",
"model-index",
"region:us"
]
| reinforcement-learning | 2023-11-10T22:03:13Z | ---
tags:
- FrozenLake-v1-4x4-no_slippery
- q-learning
- reinforcement-learning
- custom-implementation
model-index:
- name: q-FrozenLake-v1-4x4-noSlippery
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: FrozenLake-v1-4x4-no_slippery
type: FrozenLake-v1-4x4-no_slippery
metrics:
- type: mean_reward
value: 1.00 +/- 0.00
name: mean_reward
verified: false
---
# **Q-Learning** Agent playing1 **FrozenLake-v1**
This is a trained model of a **Q-Learning** agent playing **FrozenLake-v1** .
## Usage
```python
model = load_from_hub(repo_id="Guusb1990/q-FrozenLake-v1-4x4-noSlippery", filename="q-learning.pkl")
# Don't forget to check if you need to add additional attributes (is_slippery=False etc)
env = gym.make(model["env_id"])
```
|
youngsterEthan/ppo-LunarLander-v2 | youngsterEthan | 2023-11-10T21:36:57Z | 0 | 0 | stable-baselines3 | [
"stable-baselines3",
"LunarLander-v2",
"deep-reinforcement-learning",
"reinforcement-learning",
"model-index",
"region:us"
]
| reinforcement-learning | 2023-11-10T21:36:29Z | ---
library_name: stable-baselines3
tags:
- LunarLander-v2
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
model-index:
- name: PPO
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: LunarLander-v2
type: LunarLander-v2
metrics:
- type: mean_reward
value: 251.87 +/- 22.38
name: mean_reward
verified: false
---
# **PPO** Agent playing **LunarLander-v2**
This is a trained model of a **PPO** agent playing **LunarLander-v2**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3).
## Usage (with Stable-baselines3)
TODO: Add your code
```python
from stable_baselines3 import ...
from huggingface_sb3 import load_from_hub
...
```
|
hbredin/wespeaker-voxceleb-resnet34-LM | hbredin | 2023-11-10T21:34:02Z | 0 | 5 | null | [
"pytorch",
"onnx",
"license:apache-2.0",
"region:us"
]
| null | 2023-09-22T11:00:03Z | ---
license: apache-2.0
---
# WeSpeaker ResNet34 speaker embedding
This is a copy of `voxceleb_resnet34_LM.onnx` speaker embedding model taken from [ChengDong Liang's repository](https://huggingface.co/chengdongliang/wespeaker).
```
Copyright 2022 ChengDong Liang
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
```
## How to use with [pyannote.audio](https://github.com/pyannote/pyannote-audio)
```python
from pyannote.audio.pipelines.speaker_verification import PretrainedSpeakerEmbedding
get_embedding = PretrainedSpeakerEmbedding("hbredin/wespeaker-voxceleb-resnet34-LM")
assert waveforms.ndim == 3
batch_size, num_channels, num_samples = waveforms.shape
assert num_channels == 1
embeddings = get_embedding(waveforms)
assert embeddings.ndim == 2
assert embeddings.shape[0] == batch_size
```
|
Felladrin/onnx-llama2_xs_460M_experimental | Felladrin | 2023-11-10T21:33:05Z | 41 | 0 | transformers.js | [
"transformers.js",
"onnx",
"llama",
"text-generation",
"region:us"
]
| text-generation | 2023-11-10T21:18:52Z | ---
library_name: "transformers.js"
---
INT8 ONNX version of [ahxt/llama2_xs_460M_experimental](https://huggingface.co/ahxt/llama2_xs_460M_experimental) to use with [Transformers.js](https://huggingface.co/docs/transformers.js).
|
Norod78/sdxl-hearthstone-card-style-lora | Norod78 | 2023-11-10T21:31:28Z | 37 | 6 | diffusers | [
"diffusers",
"text-to-image",
"stable-diffusion",
"lora",
"card",
"hearthstone",
"style",
"cards",
"sdxl style lora",
"base_model:stabilityai/stable-diffusion-xl-base-1.0",
"base_model:adapter:stabilityai/stable-diffusion-xl-base-1.0",
"license:other",
"region:us"
]
| text-to-image | 2023-11-10T21:31:22Z | ---
license: other
tags:
- text-to-image
- stable-diffusion
- lora
- diffusers
- card
- hearthstone
- style
- cards
- sdxl style lora
base_model: stabilityai/stable-diffusion-xl-base-1.0
instance_prompt: Hearthstone Card
widget:
- text: "A ((mona lisa by leonardo da vinci)) Hearthstone Card "
- text: "A mechanical ((Santa Clause bot)) Hearthstone Card "
- text: "A magical ((Elsa from frozen)) Hearthstone Card "
- text: "3 A ((Girl with a pearl earring by vermeer)) spell use: do 3 damage to all enemy minions Hearthstone Card "
- text: "3 A ((Girl with a pearl earring by vermeer)) spell use: throw the pearl and do 3 damage to all enemy minions Hearthstone Card "
- text: "A ((Marge Simpson)) Hearthstone Card "
- text: "A photo of ((Psychedelic acid trip)) Hearthstone Card "
- text: "A photo of ((Dora the explorer)) warlock Hearthstone Card "
- text: "8 A ((Pikachu) shoot electricity Hearthstone Card "
- text: "A ((Godzilla beast)) Hearthstone Card "
---
# SDXL Hearthstone Card style LoRA
> A ((mona lisa by leonardo da vinci)) Hearthstone Card
([CivitAI](https://civitai.com/models/196113))
<p>Use the trigger word "Hearthstone Card" in your prompts.</p><p>Sometimes if you add a digit as the first thing in your prompt, it will become the "Mana cost" of the card.</p><p>The LoRA itself is very light-weight, it's under 30MB</p>
## Image examples for the model:
> A mechanical ((Santa Clause bot)) Hearthstone Card
> A magical ((Elsa from frozen)) Hearthstone Card
> 3 A ((Girl with a pearl earring by vermeer)) spell use: do 3 damage to all enemy minions Hearthstone Card
> 3 A ((Girl with a pearl earring by vermeer)) spell use: throw the pearl and do 3 damage to all enemy minions Hearthstone Card
> A ((Marge Simpson)) Hearthstone Card
> A photo of ((Psychedelic acid trip)) Hearthstone Card
> A photo of ((Dora the explorer)) warlock Hearthstone Card
> 8 A ((Pikachu) shoot electricity Hearthstone Card
> A ((Godzilla beast)) Hearthstone Card
|
TheBloke/claude2-alpaca-13B-GGUF | TheBloke | 2023-11-10T21:11:49Z | 3,711 | 33 | transformers | [
"transformers",
"gguf",
"llama",
"en",
"dataset:umd-zhou-lab/claude2_alpaca",
"base_model:umd-zhou-lab/claude2-alpaca-13B",
"base_model:quantized:umd-zhou-lab/claude2-alpaca-13B",
"license:llama2",
"region:us"
]
| null | 2023-11-10T09:49:09Z | ---
base_model: umd-zhou-lab/claude2-alpaca-13B
datasets:
- umd-zhou-lab/claude2_alpaca
inference: false
language:
- en
license: llama2
model_creator: Tianyi Lab @ UMD
model_name: Claude2 Alpaca 13B
model_type: llama
prompt_template: 'Below is an instruction that describes a task. Write a response
that appropriately completes the request.
### Instruction:
{prompt}
### Response:
'
quantized_by: TheBloke
---
<!-- markdownlint-disable MD041 -->
<!-- header start -->
<!-- 200823 -->
<div style="width: auto; margin-left: auto; margin-right: auto">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<div style="text-align:center; margin-top: 0em; margin-bottom: 0em"><p style="margin-top: 0.25em; margin-bottom: 0em;">TheBloke's LLM work is generously supported by a grant from <a href="https://a16z.com">andreessen horowitz (a16z)</a></p></div>
<hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
<!-- header end -->
# Claude2 Alpaca 13B - GGUF
- Model creator: [Tianyi Lab @ UMD](https://huggingface.co/umd-zhou-lab)
- Original model: [Claude2 Alpaca 13B](https://huggingface.co/umd-zhou-lab/claude2-alpaca-13B)
<!-- description start -->
## Description
This repo contains GGUF format model files for [Tianyi Lab @ UMD's Claude2 Alpaca 13B](https://huggingface.co/umd-zhou-lab/claude2-alpaca-13B).
These files were quantised using hardware kindly provided by [Massed Compute](https://massedcompute.com/).
<!-- description end -->
<!-- README_GGUF.md-about-gguf start -->
### About GGUF
GGUF is a new format introduced by the llama.cpp team on August 21st 2023. It is a replacement for GGML, which is no longer supported by llama.cpp.
Here is an incomplete list of clients and libraries that are known to support GGUF:
* [llama.cpp](https://github.com/ggerganov/llama.cpp). The source project for GGUF. Offers a CLI and a server option.
* [text-generation-webui](https://github.com/oobabooga/text-generation-webui), the most widely used web UI, with many features and powerful extensions. Supports GPU acceleration.
* [KoboldCpp](https://github.com/LostRuins/koboldcpp), a fully featured web UI, with GPU accel across all platforms and GPU architectures. Especially good for story telling.
* [LM Studio](https://lmstudio.ai/), an easy-to-use and powerful local GUI for Windows and macOS (Silicon), with GPU acceleration.
* [LoLLMS Web UI](https://github.com/ParisNeo/lollms-webui), a great web UI with many interesting and unique features, including a full model library for easy model selection.
* [Faraday.dev](https://faraday.dev/), an attractive and easy to use character-based chat GUI for Windows and macOS (both Silicon and Intel), with GPU acceleration.
* [ctransformers](https://github.com/marella/ctransformers), a Python library with GPU accel, LangChain support, and OpenAI-compatible AI server.
* [llama-cpp-python](https://github.com/abetlen/llama-cpp-python), a Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
* [candle](https://github.com/huggingface/candle), a Rust ML framework with a focus on performance, including GPU support, and ease of use.
<!-- README_GGUF.md-about-gguf end -->
<!-- repositories-available start -->
## Repositories available
* [AWQ model(s) for GPU inference.](https://huggingface.co/TheBloke/claude2-alpaca-13B-AWQ)
* [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/claude2-alpaca-13B-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/claude2-alpaca-13B-GGUF)
* [Tianyi Lab @ UMD's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/umd-zhou-lab/claude2-alpaca-13B)
<!-- repositories-available end -->
<!-- prompt-template start -->
## Prompt template: Alpaca
```
Below is an instruction that describes a task. Write a response that appropriately completes the request.
### Instruction:
{prompt}
### Response:
```
<!-- prompt-template end -->
<!-- compatibility_gguf start -->
## Compatibility
These quantised GGUFv2 files are compatible with llama.cpp from August 27th onwards, as of commit [d0cee0d](https://github.com/ggerganov/llama.cpp/commit/d0cee0d36d5be95a0d9088b674dbb27354107221)
They are also compatible with many third party UIs and libraries - please see the list at the top of this README.
## Explanation of quantisation methods
<details>
<summary>Click to see details</summary>
The new methods available are:
* GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
* GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
* GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
* GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
* GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
Refer to the Provided Files table below to see what files use which methods, and how.
</details>
<!-- compatibility_gguf end -->
<!-- README_GGUF.md-provided-files start -->
## Provided files
| Name | Quant method | Bits | Size | Max RAM required | Use case |
| ---- | ---- | ---- | ---- | ---- | ----- |
| [claude2-alpaca-13b.Q2_K.gguf](https://huggingface.co/TheBloke/claude2-alpaca-13B-GGUF/blob/main/claude2-alpaca-13b.Q2_K.gguf) | Q2_K | 2 | 5.43 GB| 7.93 GB | smallest, significant quality loss - not recommended for most purposes |
| [claude2-alpaca-13b.Q3_K_S.gguf](https://huggingface.co/TheBloke/claude2-alpaca-13B-GGUF/blob/main/claude2-alpaca-13b.Q3_K_S.gguf) | Q3_K_S | 3 | 5.66 GB| 8.16 GB | very small, high quality loss |
| [claude2-alpaca-13b.Q3_K_M.gguf](https://huggingface.co/TheBloke/claude2-alpaca-13B-GGUF/blob/main/claude2-alpaca-13b.Q3_K_M.gguf) | Q3_K_M | 3 | 6.34 GB| 8.84 GB | very small, high quality loss |
| [claude2-alpaca-13b.Q3_K_L.gguf](https://huggingface.co/TheBloke/claude2-alpaca-13B-GGUF/blob/main/claude2-alpaca-13b.Q3_K_L.gguf) | Q3_K_L | 3 | 6.93 GB| 9.43 GB | small, substantial quality loss |
| [claude2-alpaca-13b.Q4_0.gguf](https://huggingface.co/TheBloke/claude2-alpaca-13B-GGUF/blob/main/claude2-alpaca-13b.Q4_0.gguf) | Q4_0 | 4 | 7.37 GB| 9.87 GB | legacy; small, very high quality loss - prefer using Q3_K_M |
| [claude2-alpaca-13b.Q4_K_S.gguf](https://huggingface.co/TheBloke/claude2-alpaca-13B-GGUF/blob/main/claude2-alpaca-13b.Q4_K_S.gguf) | Q4_K_S | 4 | 7.41 GB| 9.91 GB | small, greater quality loss |
| [claude2-alpaca-13b.Q4_K_M.gguf](https://huggingface.co/TheBloke/claude2-alpaca-13B-GGUF/blob/main/claude2-alpaca-13b.Q4_K_M.gguf) | Q4_K_M | 4 | 7.87 GB| 10.37 GB | medium, balanced quality - recommended |
| [claude2-alpaca-13b.Q5_0.gguf](https://huggingface.co/TheBloke/claude2-alpaca-13B-GGUF/blob/main/claude2-alpaca-13b.Q5_0.gguf) | Q5_0 | 5 | 8.97 GB| 11.47 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
| [claude2-alpaca-13b.Q5_K_S.gguf](https://huggingface.co/TheBloke/claude2-alpaca-13B-GGUF/blob/main/claude2-alpaca-13b.Q5_K_S.gguf) | Q5_K_S | 5 | 8.97 GB| 11.47 GB | large, low quality loss - recommended |
| [claude2-alpaca-13b.Q5_K_M.gguf](https://huggingface.co/TheBloke/claude2-alpaca-13B-GGUF/blob/main/claude2-alpaca-13b.Q5_K_M.gguf) | Q5_K_M | 5 | 9.23 GB| 11.73 GB | large, very low quality loss - recommended |
| [claude2-alpaca-13b.Q6_K.gguf](https://huggingface.co/TheBloke/claude2-alpaca-13B-GGUF/blob/main/claude2-alpaca-13b.Q6_K.gguf) | Q6_K | 6 | 10.68 GB| 13.18 GB | very large, extremely low quality loss |
| [claude2-alpaca-13b.Q8_0.gguf](https://huggingface.co/TheBloke/claude2-alpaca-13B-GGUF/blob/main/claude2-alpaca-13b.Q8_0.gguf) | Q8_0 | 8 | 13.83 GB| 16.33 GB | very large, extremely low quality loss - not recommended |
**Note**: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.
<!-- README_GGUF.md-provided-files end -->
<!-- README_GGUF.md-how-to-download start -->
## How to download GGUF files
**Note for manual downloaders:** You almost never want to clone the entire repo! Multiple different quantisation formats are provided, and most users only want to pick and download a single file.
The following clients/libraries will automatically download models for you, providing a list of available models to choose from:
* LM Studio
* LoLLMS Web UI
* Faraday.dev
### In `text-generation-webui`
Under Download Model, you can enter the model repo: TheBloke/claude2-alpaca-13B-GGUF and below it, a specific filename to download, such as: claude2-alpaca-13b.Q4_K_M.gguf.
Then click Download.
### On the command line, including multiple files at once
I recommend using the `huggingface-hub` Python library:
```shell
pip3 install huggingface-hub
```
Then you can download any individual model file to the current directory, at high speed, with a command like this:
```shell
huggingface-cli download TheBloke/claude2-alpaca-13B-GGUF claude2-alpaca-13b.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
```
<details>
<summary>More advanced huggingface-cli download usage</summary>
You can also download multiple files at once with a pattern:
```shell
huggingface-cli download TheBloke/claude2-alpaca-13B-GGUF --local-dir . --local-dir-use-symlinks False --include='*Q4_K*gguf'
```
For more documentation on downloading with `huggingface-cli`, please see: [HF -> Hub Python Library -> Download files -> Download from the CLI](https://huggingface.co/docs/huggingface_hub/guides/download#download-from-the-cli).
To accelerate downloads on fast connections (1Gbit/s or higher), install `hf_transfer`:
```shell
pip3 install hf_transfer
```
And set environment variable `HF_HUB_ENABLE_HF_TRANSFER` to `1`:
```shell
HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download TheBloke/claude2-alpaca-13B-GGUF claude2-alpaca-13b.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
```
Windows Command Line users: You can set the environment variable by running `set HF_HUB_ENABLE_HF_TRANSFER=1` before the download command.
</details>
<!-- README_GGUF.md-how-to-download end -->
<!-- README_GGUF.md-how-to-run start -->
## Example `llama.cpp` command
Make sure you are using `llama.cpp` from commit [d0cee0d](https://github.com/ggerganov/llama.cpp/commit/d0cee0d36d5be95a0d9088b674dbb27354107221) or later.
```shell
./main -ngl 32 -m claude2-alpaca-13b.Q4_K_M.gguf --color -c 4096 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n### Instruction:\n{prompt}\n\n### Response:"
```
Change `-ngl 32` to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.
Change `-c 4096` to the desired sequence length. For extended sequence models - eg 8K, 16K, 32K - the necessary RoPE scaling parameters are read from the GGUF file and set by llama.cpp automatically.
If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`
For other parameters and how to use them, please refer to [the llama.cpp documentation](https://github.com/ggerganov/llama.cpp/blob/master/examples/main/README.md)
## How to run in `text-generation-webui`
Further instructions here: [text-generation-webui/docs/04 ‐ Model Tab.md](https://github.com/oobabooga/text-generation-webui/blob/main/docs/04%20%E2%80%90%20Model%20Tab.md#llamacpp).
## How to run from Python code
You can use GGUF models from Python using the [llama-cpp-python](https://github.com/abetlen/llama-cpp-python) or [ctransformers](https://github.com/marella/ctransformers) libraries.
### How to load this model in Python code, using ctransformers
#### First install the package
Run one of the following commands, according to your system:
```shell
# Base ctransformers with no GPU acceleration
pip install ctransformers
# Or with CUDA GPU acceleration
pip install ctransformers[cuda]
# Or with AMD ROCm GPU acceleration (Linux only)
CT_HIPBLAS=1 pip install ctransformers --no-binary ctransformers
# Or with Metal GPU acceleration for macOS systems only
CT_METAL=1 pip install ctransformers --no-binary ctransformers
```
#### Simple ctransformers example code
```python
from ctransformers import AutoModelForCausalLM
# Set gpu_layers to the number of layers to offload to GPU. Set to 0 if no GPU acceleration is available on your system.
llm = AutoModelForCausalLM.from_pretrained("TheBloke/claude2-alpaca-13B-GGUF", model_file="claude2-alpaca-13b.Q4_K_M.gguf", model_type="llama", gpu_layers=50)
print(llm("AI is going to"))
```
## How to use with LangChain
Here are guides on using llama-cpp-python and ctransformers with LangChain:
* [LangChain + llama-cpp-python](https://python.langchain.com/docs/integrations/llms/llamacpp)
* [LangChain + ctransformers](https://python.langchain.com/docs/integrations/providers/ctransformers)
<!-- README_GGUF.md-how-to-run end -->
<!-- footer start -->
<!-- 200823 -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
## Thanks, and how to contribute
Thanks to the [chirper.ai](https://chirper.ai) team!
Thanks to Clay from [gpus.llm-utils.org](llm-utils)!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Aemon Algiz.
**Patreon special mentions**: Brandon Frisco, LangChain4j, Spiking Neurons AB, transmissions 11, Joseph William Delisle, Nitin Borwankar, Willem Michiel, Michael Dempsey, vamX, Jeffrey Morgan, zynix, jjj, Omer Bin Jawed, Sean Connelly, jinyuan sun, Jeromy Smith, Shadi, Pawan Osman, Chadd, Elijah Stavena, Illia Dulskyi, Sebastain Graf, Stephen Murray, terasurfer, Edmond Seymore, Celu Ramasamy, Mandus, Alex, biorpg, Ajan Kanaga, Clay Pascal, Raven Klaugh, 阿明, K, ya boyyy, usrbinkat, Alicia Loh, John Villwock, ReadyPlayerEmma, Chris Smitley, Cap'n Zoog, fincy, GodLy, S_X, sidney chen, Cory Kujawski, OG, Mano Prime, AzureBlack, Pieter, Kalila, Spencer Kim, Tom X Nguyen, Stanislav Ovsiannikov, Michael Levine, Andrey, Trailburnt, Vadim, Enrico Ros, Talal Aujan, Brandon Phillips, Jack West, Eugene Pentland, Michael Davis, Will Dee, webtim, Jonathan Leane, Alps Aficionado, Rooh Singh, Tiffany J. Kim, theTransient, Luke @flexchar, Elle, Caitlyn Gatomon, Ari Malik, subjectnull, Johann-Peter Hartmann, Trenton Dambrowitz, Imad Khwaja, Asp the Wyvern, Emad Mostaque, Rainer Wilmers, Alexandros Triantafyllidis, Nicholas, Pedro Madruga, SuperWojo, Harry Royden McLaughlin, James Bentley, Olakabola, David Ziegler, Ai Maven, Jeff Scroggin, Nikolai Manek, Deo Leter, Matthew Berman, Fen Risland, Ken Nordquist, Manuel Alberto Morcote, Luke Pendergrass, TL, Fred von Graf, Randy H, Dan Guido, NimbleBox.ai, Vitor Caleffi, Gabriel Tamborski, knownsqashed, Lone Striker, Erik Bjäreholt, John Detwiler, Leonard Tan, Iucharbius
Thank you to all my generous patrons and donaters!
And thank you again to a16z for their generous grant.
<!-- footer end -->
<!-- original-model-card start -->
# Original model card: Tianyi Lab @ UMD's Claude2 Alpaca 13B
# Model Card for umd-zhou-lab/claude2-alpaca-13B
<!-- Provide a quick summary of what the model is/does. -->
This model is trained by fine-tuning llama-2 with claude2 alpaca data.
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
- **Developed by:** UMD Tianyi Zhou Lab
- **Model type:** An auto-regressive language model based on the transformer architecture
- **License:** Llama 2 Community License Agreement
- **Finetuned from model:** [meta-llama/Llama-2-13b](https://huggingface.co/meta-llama/Llama-2-13b)
### Model Sources
<!-- Provide the basic links for the model. -->
- **GitHub:** [Claude2-Alpaca](https://github.com/Lichang-Chen/claude2-alpaca)
- **Data:** [claude2_alpaca](https://huggingface.co/datasets/umd-zhou-lab/claude2_alpaca)
## Uses
The primary use of this model is research on large language models and chatbots.
The primary intended users of the model are researchers and hobbyists in natural language processing, machine learning, and artificial intelligence.
## Training
We use the prompt from [Stanford Alpaca](https://github.com/tatsu-lab/stanford_alpaca)
| Hyperparameter | Global Batch Size | Learning rate | Epochs | Max length | Weight decay |
| --- | ---: | ---: | ---: | ---: | ---: |
| Model (13B) | 128 | 1e-5 | 5 | 2048 | 0 |
## Performance
Compared to the llama2-chat, our models can have better average performance.<br>
| | Average | ARC | HellaSwag | MMLU | TruthfulQA | Alpaca_Eval | Avg Length |
|---|---|---|---|---|---|---|---|
| Llama-2-7b-chat | 56.335 | 52.9 | 78.55 | 48.32 | 45.57 | 71.37 | 1479 |
| Llama-2-13b-chat | 59.935 | 59.04| 81.94 | 54.64 | 44.12 | 81.09 | 1513 |
|||||||||
| claude_alpaca-7b | 57.78 | 56.66 | 81.17 | 46.58 | 46.71 | 71.23 | 1066 |
| claude_alpaca-13b | 61.29 | 61.18 | 84.08 | 55.74 | 44.18 | 78.93 | 1127 |
## Citation
Please consider citing our paper if you think our codes, data, or models are useful. Thank you!
```
@misc{claude2-alpaca,
author = {Lichang Chen and Khalid Saifullah and Ming Li and Tianyi Zhou and Heng Huang},
title = {Claude2-Alpaca: Instruction tuning datasets distilled from claude},
year = {2023},
publisher = {GitHub},
journal = {GitHub repository},
howpublished = {\url{https://github.com/Lichang-Chen/claude2-alpaca}},
}
```
<!-- original-model-card end -->
|
JunghwanRo/ppo-PyramidsRND | JunghwanRo | 2023-11-10T20:55:51Z | 0 | 0 | ml-agents | [
"ml-agents",
"tensorboard",
"onnx",
"Pyramids",
"deep-reinforcement-learning",
"reinforcement-learning",
"ML-Agents-Pyramids",
"region:us"
]
| reinforcement-learning | 2023-11-10T20:54:17Z | ---
library_name: ml-agents
tags:
- Pyramids
- deep-reinforcement-learning
- reinforcement-learning
- ML-Agents-Pyramids
---
# **ppo** Agent playing **Pyramids**
This is a trained model of a **ppo** agent playing **Pyramids**
using the [Unity ML-Agents Library](https://github.com/Unity-Technologies/ml-agents).
## Usage (with ML-Agents)
The Documentation: https://unity-technologies.github.io/ml-agents/ML-Agents-Toolkit-Documentation/
We wrote a complete tutorial to learn to train your first agent using ML-Agents and publish it to the Hub:
- A *short tutorial* where you teach Huggy the Dog 🐶 to fetch the stick and then play with him directly in your
browser: https://huggingface.co/learn/deep-rl-course/unitbonus1/introduction
- A *longer tutorial* to understand how works ML-Agents:
https://huggingface.co/learn/deep-rl-course/unit5/introduction
### Resume the training
```bash
mlagents-learn <your_configuration_file_path.yaml> --run-id=<run_id> --resume
```
### Watch your Agent play
You can watch your agent **playing directly in your browser**
1. If the environment is part of ML-Agents official environments, go to https://huggingface.co/unity
2. Step 1: Find your model_id: JunghwanRo/ppo-PyramidsRND
3. Step 2: Select your *.nn /*.onnx file
4. Click on Watch the agent play 👀
|
moreover18/vit-base-patch16-224-in21k-finetuned-eurosat | moreover18 | 2023-11-10T20:55:02Z | 8 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:google/vit-base-patch16-224-in21k",
"base_model:finetune:google/vit-base-patch16-224-in21k",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| image-classification | 2023-11-10T16:53:32Z | ---
license: apache-2.0
base_model: google/vit-base-patch16-224-in21k
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: vit-base-patch16-224-in21k-finetuned-eurosat
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: train
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.9360791655522868
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# vit-base-patch16-224-in21k-finetuned-eurosat
This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1770
- Accuracy: 0.9361
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 3
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 0.687 | 0.04 | 10 | 0.6778 | 0.6026 |
| 0.6605 | 0.09 | 20 | 0.6359 | 0.7564 |
| 0.6074 | 0.13 | 30 | 0.5734 | 0.7786 |
| 0.5464 | 0.17 | 40 | 0.4877 | 0.8267 |
| 0.4606 | 0.21 | 50 | 0.3836 | 0.8914 |
| 0.379 | 0.26 | 60 | 0.3269 | 0.8877 |
| 0.2746 | 0.3 | 70 | 0.2403 | 0.9198 |
| 0.2974 | 0.34 | 80 | 0.2931 | 0.8890 |
| 0.2459 | 0.39 | 90 | 0.2596 | 0.9016 |
| 0.2507 | 0.43 | 100 | 0.2366 | 0.9123 |
| 0.2627 | 0.47 | 110 | 0.2084 | 0.9224 |
| 0.2481 | 0.51 | 120 | 0.2050 | 0.9270 |
| 0.2372 | 0.56 | 130 | 0.2077 | 0.9267 |
| 0.2468 | 0.6 | 140 | 0.2111 | 0.9230 |
| 0.2272 | 0.64 | 150 | 0.1964 | 0.9267 |
| 0.2568 | 0.68 | 160 | 0.1975 | 0.9270 |
| 0.2608 | 0.73 | 170 | 0.2485 | 0.9048 |
| 0.2641 | 0.77 | 180 | 0.2143 | 0.9227 |
| 0.2347 | 0.81 | 190 | 0.1921 | 0.9307 |
| 0.2231 | 0.86 | 200 | 0.1882 | 0.9315 |
| 0.2147 | 0.9 | 210 | 0.1865 | 0.9329 |
| 0.2028 | 0.94 | 220 | 0.1901 | 0.9294 |
| 0.1792 | 0.98 | 230 | 0.1868 | 0.9297 |
| 0.2471 | 1.03 | 240 | 0.2104 | 0.9190 |
| 0.1896 | 1.07 | 250 | 0.1840 | 0.9321 |
| 0.2181 | 1.11 | 260 | 0.1800 | 0.9318 |
| 0.1861 | 1.16 | 270 | 0.1815 | 0.9305 |
| 0.1761 | 1.2 | 280 | 0.1886 | 0.9299 |
| 0.1703 | 1.24 | 290 | 0.1802 | 0.9315 |
| 0.184 | 1.28 | 300 | 0.1845 | 0.9321 |
| 0.1864 | 1.33 | 310 | 0.1791 | 0.9342 |
| 0.1857 | 1.37 | 320 | 0.1760 | 0.9347 |
| 0.1558 | 1.41 | 330 | 0.1798 | 0.9318 |
| 0.1852 | 1.45 | 340 | 0.1810 | 0.9323 |
| 0.183 | 1.5 | 350 | 0.1775 | 0.9321 |
| 0.2055 | 1.54 | 360 | 0.1789 | 0.9337 |
| 0.207 | 1.58 | 370 | 0.2082 | 0.9208 |
| 0.2264 | 1.63 | 380 | 0.1733 | 0.9339 |
| 0.1954 | 1.67 | 390 | 0.1772 | 0.9337 |
| 0.1676 | 1.71 | 400 | 0.1840 | 0.9302 |
| 0.1727 | 1.75 | 410 | 0.1784 | 0.9305 |
| 0.204 | 1.8 | 420 | 0.1731 | 0.9353 |
| 0.1805 | 1.84 | 430 | 0.1805 | 0.9310 |
| 0.1732 | 1.88 | 440 | 0.1773 | 0.9337 |
| 0.1831 | 1.93 | 450 | 0.1768 | 0.9337 |
| 0.1906 | 1.97 | 460 | 0.1967 | 0.9259 |
| 0.1785 | 2.01 | 470 | 0.1765 | 0.9331 |
| 0.1566 | 2.05 | 480 | 0.1749 | 0.9361 |
| 0.1612 | 2.1 | 490 | 0.1718 | 0.9342 |
| 0.1504 | 2.14 | 500 | 0.1770 | 0.9361 |
| 0.1704 | 2.18 | 510 | 0.1721 | 0.9363 |
| 0.1597 | 2.22 | 520 | 0.1711 | 0.9345 |
| 0.1283 | 2.27 | 530 | 0.1775 | 0.9361 |
| 0.1697 | 2.31 | 540 | 0.1722 | 0.9361 |
| 0.1541 | 2.35 | 550 | 0.1729 | 0.9366 |
| 0.1466 | 2.4 | 560 | 0.1708 | 0.9369 |
| 0.1604 | 2.44 | 570 | 0.1720 | 0.9371 |
| 0.1798 | 2.48 | 580 | 0.1718 | 0.9382 |
| 0.134 | 2.52 | 590 | 0.1733 | 0.9371 |
| 0.1215 | 2.57 | 600 | 0.1749 | 0.9369 |
| 0.1284 | 2.61 | 610 | 0.1760 | 0.9358 |
| 0.1449 | 2.65 | 620 | 0.1745 | 0.9361 |
| 0.214 | 2.7 | 630 | 0.1729 | 0.9382 |
| 0.1684 | 2.74 | 640 | 0.1724 | 0.9369 |
| 0.143 | 2.78 | 650 | 0.1737 | 0.9377 |
| 0.1491 | 2.82 | 660 | 0.1753 | 0.9366 |
| 0.1636 | 2.87 | 670 | 0.1743 | 0.9371 |
| 0.1672 | 2.91 | 680 | 0.1724 | 0.9377 |
| 0.1501 | 2.95 | 690 | 0.1720 | 0.9374 |
### Framework versions
- Transformers 4.35.0
- Pytorch 1.12.1+cu116
- Datasets 2.4.0
- Tokenizers 0.14.1
|
thegadri/ppo-CartPole-v1 | thegadri | 2023-11-10T20:53:10Z | 0 | 0 | null | [
"tensorboard",
"LunarLander-v2",
"ppo",
"deep-reinforcement-learning",
"reinforcement-learning",
"custom-implementation",
"deep-rl-course",
"model-index",
"region:us"
]
| reinforcement-learning | 2023-11-10T20:52:26Z | ---
tags:
- LunarLander-v2
- ppo
- deep-reinforcement-learning
- reinforcement-learning
- custom-implementation
- deep-rl-course
model-index:
- name: PPO
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: LunarLander-v2
type: LunarLander-v2
metrics:
- type: mean_reward
value: -55.98 +/- 26.80
name: mean_reward
verified: false
---
# PPO Agent Playing LunarLander-v2
This is a trained model of a PPO agent playing LunarLander-v2.
# Hyperparameters
```python
{'exp_name': 'ppo'
'seed': 1
'torch_deterministic': True
'cuda': True
'track': False
'wandb_project_name': 'cleanRL'
'wandb_entity': None
'capture_video': False
'env_id': 'LunarLander-v2'
'total_timesteps': 500000
'learning_rate': 0.00025
'num_envs': 4
'num_steps': 128
'anneal_lr': True
'gae': True
'gamma': 0.99
'gae_lambda': 0.95
'num_minibatches': 4
'update_epochs': 4
'norm_adv': True
'clip_coef': 0.2
'clip_vloss': True
'ent_coef': 0.01
'vf_coef': 0.5
'max_grad_norm': 0.5
'target_kl': None
'repo_id': 'thegadri/ppo-CartPole-v1'
'batch_size': 512
'minibatch_size': 128}
```
|
Joshua-Abok/finetuning-wav2vec-large-swahili-asr-model_v8 | Joshua-Abok | 2023-11-10T20:51:10Z | 4 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"wav2vec2",
"automatic-speech-recognition",
"generated_from_trainer",
"dataset:common_voice_13_0",
"base_model:AntonyG/fine-tune-wav2vec2-large-xls-r-1b-sw",
"base_model:finetune:AntonyG/fine-tune-wav2vec2-large-xls-r-1b-sw",
"license:apache-2.0",
"endpoints_compatible",
"region:us"
]
| automatic-speech-recognition | 2023-11-10T16:34:52Z | ---
license: apache-2.0
base_model: AntonyG/fine-tune-wav2vec2-large-xls-r-1b-sw
tags:
- generated_from_trainer
datasets:
- common_voice_13_0
model-index:
- name: finetuning-wav2vec-large-swahili-asr-model_v8
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# finetuning-wav2vec-large-swahili-asr-model_v8
This model is a fine-tuned version of [AntonyG/fine-tune-wav2vec2-large-xls-r-1b-sw](https://huggingface.co/AntonyG/fine-tune-wav2vec2-large-xls-r-1b-sw) on the common_voice_13_0 dataset.
It achieves the following results on the evaluation set:
- eval_loss: 0.3413
- eval_wer: 0.2759
- eval_runtime: 715.2087
- eval_samples_per_second: 15.759
- eval_steps_per_second: 0.986
- epoch: 1.11
- step: 800
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0003
- train_batch_size: 32
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 10
- mixed_precision_training: Native AMP
### Framework versions
- Transformers 4.36.0.dev0
- Pytorch 2.0.1+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
Felladrin/onnx-Cerebras-GPT-111M-instruction | Felladrin | 2023-11-10T20:36:05Z | 39 | 0 | transformers.js | [
"transformers.js",
"onnx",
"gpt2",
"text-generation",
"license:apache-2.0",
"region:us"
]
| text-generation | 2023-11-10T20:26:48Z | ---
license: apache-2.0
library_name: "transformers.js"
---
INT8 ONNX version of [SebastianSchramm/Cerebras-GPT-111M-instruction](https://huggingface.co/SebastianSchramm/Cerebras-GPT-111M-instruction) to use with [Transformers.js](https://huggingface.co/docs/transformers.js).
|
thingthatis/stable-diffusion-xl-base-1.0 | thingthatis | 2023-11-10T20:16:48Z | 6 | 1 | diffusers | [
"diffusers",
"onnx",
"safetensors",
"text-to-image",
"stable-diffusion",
"arxiv:2307.01952",
"arxiv:2211.01324",
"arxiv:2108.01073",
"arxiv:2112.10752",
"license:openrail++",
"autotrain_compatible",
"endpoints_compatible",
"diffusers:StableDiffusionXLPipeline",
"region:us"
]
| text-to-image | 2023-11-10T20:16:47Z | ---
license: openrail++
tags:
- text-to-image
- stable-diffusion
---
# SD-XL 1.0-base Model Card
## Model
[SDXL](https://arxiv.org/abs/2307.01952) consists of an [ensemble of experts](https://arxiv.org/abs/2211.01324) pipeline for latent diffusion:
In a first step, the base model is used to generate (noisy) latents,
which are then further processed with a refinement model (available here: https://huggingface.co/stabilityai/stable-diffusion-xl-refiner-1.0/) specialized for the final denoising steps.
Note that the base model can be used as a standalone module.
Alternatively, we can use a two-stage pipeline as follows:
First, the base model is used to generate latents of the desired output size.
In the second step, we use a specialized high-resolution model and apply a technique called SDEdit (https://arxiv.org/abs/2108.01073, also known as "img2img")
to the latents generated in the first step, using the same prompt. This technique is slightly slower than the first one, as it requires more function evaluations.
Source code is available at https://github.com/Stability-AI/generative-models .
### Model Description
- **Developed by:** Stability AI
- **Model type:** Diffusion-based text-to-image generative model
- **License:** [CreativeML Open RAIL++-M License](https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/LICENSE.md)
- **Model Description:** This is a model that can be used to generate and modify images based on text prompts. It is a [Latent Diffusion Model](https://arxiv.org/abs/2112.10752) that uses two fixed, pretrained text encoders ([OpenCLIP-ViT/G](https://github.com/mlfoundations/open_clip) and [CLIP-ViT/L](https://github.com/openai/CLIP/tree/main)).
- **Resources for more information:** Check out our [GitHub Repository](https://github.com/Stability-AI/generative-models) and the [SDXL report on arXiv](https://arxiv.org/abs/2307.01952).
### Model Sources
For research purposes, we recommend our `generative-models` Github repository (https://github.com/Stability-AI/generative-models), which implements the most popular diffusion frameworks (both training and inference) and for which new functionalities like distillation will be added over time.
[Clipdrop](https://clipdrop.co/stable-diffusion) provides free SDXL inference.
- **Repository:** https://github.com/Stability-AI/generative-models
- **Demo:** https://clipdrop.co/stable-diffusion
## Evaluation
The chart above evaluates user preference for SDXL (with and without refinement) over SDXL 0.9 and Stable Diffusion 1.5 and 2.1.
The SDXL base model performs significantly better than the previous variants, and the model combined with the refinement module achieves the best overall performance.
### 🧨 Diffusers
Make sure to upgrade diffusers to >= 0.19.0:
```
pip install diffusers --upgrade
```
In addition make sure to install `transformers`, `safetensors`, `accelerate` as well as the invisible watermark:
```
pip install invisible_watermark transformers accelerate safetensors
```
To just use the base model, you can run:
```py
from diffusers import DiffusionPipeline
import torch
pipe = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16, use_safetensors=True, variant="fp16")
pipe.to("cuda")
# if using torch < 2.0
# pipe.enable_xformers_memory_efficient_attention()
prompt = "An astronaut riding a green horse"
images = pipe(prompt=prompt).images[0]
```
To use the whole base + refiner pipeline as an ensemble of experts you can run:
```py
from diffusers import DiffusionPipeline
import torch
# load both base & refiner
base = DiffusionPipeline.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16, variant="fp16", use_safetensors=True
)
base.to("cuda")
refiner = DiffusionPipeline.from_pretrained(
"stabilityai/stable-diffusion-xl-refiner-1.0",
text_encoder_2=base.text_encoder_2,
vae=base.vae,
torch_dtype=torch.float16,
use_safetensors=True,
variant="fp16",
)
refiner.to("cuda")
# Define how many steps and what % of steps to be run on each experts (80/20) here
n_steps = 40
high_noise_frac = 0.8
prompt = "A majestic lion jumping from a big stone at night"
# run both experts
image = base(
prompt=prompt,
num_inference_steps=n_steps,
denoising_end=high_noise_frac,
output_type="latent",
).images
image = refiner(
prompt=prompt,
num_inference_steps=n_steps,
denoising_start=high_noise_frac,
image=image,
).images[0]
```
When using `torch >= 2.0`, you can improve the inference speed by 20-30% with torch.compile. Simple wrap the unet with torch compile before running the pipeline:
```py
pipe.unet = torch.compile(pipe.unet, mode="reduce-overhead", fullgraph=True)
```
If you are limited by GPU VRAM, you can enable *cpu offloading* by calling `pipe.enable_model_cpu_offload`
instead of `.to("cuda")`:
```diff
- pipe.to("cuda")
+ pipe.enable_model_cpu_offload()
```
For more information on how to use Stable Diffusion XL with `diffusers`, please have a look at [the Stable Diffusion XL Docs](https://huggingface.co/docs/diffusers/api/pipelines/stable_diffusion/stable_diffusion_xl).
### Optimum
[Optimum](https://github.com/huggingface/optimum) provides a Stable Diffusion pipeline compatible with both [OpenVINO](https://docs.openvino.ai/latest/index.html) and [ONNX Runtime](https://onnxruntime.ai/).
#### OpenVINO
To install Optimum with the dependencies required for OpenVINO :
```bash
pip install optimum[openvino]
```
To load an OpenVINO model and run inference with OpenVINO Runtime, you need to replace `StableDiffusionXLPipeline` with Optimum `OVStableDiffusionXLPipeline`. In case you want to load a PyTorch model and convert it to the OpenVINO format on-the-fly, you can set `export=True`.
```diff
- from diffusers import StableDiffusionXLPipeline
+ from optimum.intel import OVStableDiffusionXLPipeline
model_id = "stabilityai/stable-diffusion-xl-base-1.0"
- pipeline = StableDiffusionXLPipeline.from_pretrained(model_id)
+ pipeline = OVStableDiffusionXLPipeline.from_pretrained(model_id)
prompt = "A majestic lion jumping from a big stone at night"
image = pipeline(prompt).images[0]
```
You can find more examples (such as static reshaping and model compilation) in optimum [documentation](https://huggingface.co/docs/optimum/main/en/intel/inference#stable-diffusion-xl).
#### ONNX
To install Optimum with the dependencies required for ONNX Runtime inference :
```bash
pip install optimum[onnxruntime]
```
To load an ONNX model and run inference with ONNX Runtime, you need to replace `StableDiffusionXLPipeline` with Optimum `ORTStableDiffusionXLPipeline`. In case you want to load a PyTorch model and convert it to the ONNX format on-the-fly, you can set `export=True`.
```diff
- from diffusers import StableDiffusionXLPipeline
+ from optimum.onnxruntime import ORTStableDiffusionXLPipeline
model_id = "stabilityai/stable-diffusion-xl-base-1.0"
- pipeline = StableDiffusionXLPipeline.from_pretrained(model_id)
+ pipeline = ORTStableDiffusionXLPipeline.from_pretrained(model_id)
prompt = "A majestic lion jumping from a big stone at night"
image = pipeline(prompt).images[0]
```
You can find more examples in optimum [documentation](https://huggingface.co/docs/optimum/main/en/onnxruntime/usage_guides/models#stable-diffusion-xl).
## Uses
### Direct Use
The model is intended for research purposes only. Possible research areas and tasks include
- Generation of artworks and use in design and other artistic processes.
- Applications in educational or creative tools.
- Research on generative models.
- Safe deployment of models which have the potential to generate harmful content.
- Probing and understanding the limitations and biases of generative models.
Excluded uses are described below.
### Out-of-Scope Use
The model was not trained to be factual or true representations of people or events, and therefore using the model to generate such content is out-of-scope for the abilities of this model.
## Limitations and Bias
### Limitations
- The model does not achieve perfect photorealism
- The model cannot render legible text
- The model struggles with more difficult tasks which involve compositionality, such as rendering an image corresponding to “A red cube on top of a blue sphere”
- Faces and people in general may not be generated properly.
- The autoencoding part of the model is lossy.
### Bias
While the capabilities of image generation models are impressive, they can also reinforce or exacerbate social biases.
|
thingthatis/stable-diffusion-xl-refiner-1.0 | thingthatis | 2023-11-10T20:15:02Z | 5 | 0 | diffusers | [
"diffusers",
"safetensors",
"stable-diffusion",
"image-to-image",
"arxiv:2307.01952",
"arxiv:2211.01324",
"arxiv:2108.01073",
"arxiv:2112.10752",
"license:openrail++",
"diffusers:StableDiffusionXLImg2ImgPipeline",
"region:us"
]
| image-to-image | 2023-11-10T20:15:02Z | ---
license: openrail++
tags:
- stable-diffusion
- image-to-image
---
# SD-XL 1.0-refiner Model Card
## Model
[SDXL](https://arxiv.org/abs/2307.01952) consists of an [ensemble of experts](https://arxiv.org/abs/2211.01324) pipeline for latent diffusion:
In a first step, the base model (available here: https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0) is used to generate (noisy) latents,
which are then further processed with a refinement model specialized for the final denoising steps.
Note that the base model can be used as a standalone module.
Alternatively, we can use a two-stage pipeline as follows:
First, the base model is used to generate latents of the desired output size.
In the second step, we use a specialized high-resolution model and apply a technique called SDEdit (https://arxiv.org/abs/2108.01073, also known as "img2img")
to the latents generated in the first step, using the same prompt. This technique is slightly slower than the first one, as it requires more function evaluations.
Source code is available at https://github.com/Stability-AI/generative-models .
### Model Description
- **Developed by:** Stability AI
- **Model type:** Diffusion-based text-to-image generative model
- **License:** [CreativeML Open RAIL++-M License](https://huggingface.co/stabilityai/stable-diffusion-xl-refiner-1.0/blob/main/LICENSE.md)
- **Model Description:** This is a model that can be used to generate and modify images based on text prompts. It is a [Latent Diffusion Model](https://arxiv.org/abs/2112.10752) that uses two fixed, pretrained text encoders ([OpenCLIP-ViT/G](https://github.com/mlfoundations/open_clip) and [CLIP-ViT/L](https://github.com/openai/CLIP/tree/main)).
- **Resources for more information:** Check out our [GitHub Repository](https://github.com/Stability-AI/generative-models) and the [SDXL report on arXiv](https://arxiv.org/abs/2307.01952).
### Model Sources
For research purposes, we recommned our `generative-models` Github repository (https://github.com/Stability-AI/generative-models), which implements the most popoular diffusion frameworks (both training and inference) and for which new functionalities like distillation will be added over time.
[Clipdrop](https://clipdrop.co/stable-diffusion) provides free SDXL inference.
- **Repository:** https://github.com/Stability-AI/generative-models
- **Demo:** https://clipdrop.co/stable-diffusion
## Evaluation
The chart above evaluates user preference for SDXL (with and without refinement) over SDXL 0.9 and Stable Diffusion 1.5 and 2.1.
The SDXL base model performs significantly better than the previous variants, and the model combined with the refinement module achieves the best overall performance.
### 🧨 Diffusers
Make sure to upgrade diffusers to >= 0.18.0:
```
pip install diffusers --upgrade
```
In addition make sure to install `transformers`, `safetensors`, `accelerate` as well as the invisible watermark:
```
pip install invisible_watermark transformers accelerate safetensors
```
Yon can then use the refiner to improve images.
```py
import torch
from diffusers import StableDiffusionXLImg2ImgPipeline
from diffusers.utils import load_image
pipe = StableDiffusionXLImg2ImgPipeline.from_pretrained(
"stabilityai/stable-diffusion-xl-refiner-1.0", torch_dtype=torch.float16, variant="fp16", use_safetensors=True
)
pipe = pipe.to("cuda")
url = "https://huggingface.co/datasets/patrickvonplaten/images/resolve/main/aa_xl/000000009.png"
init_image = load_image(url).convert("RGB")
prompt = "a photo of an astronaut riding a horse on mars"
image = pipe(prompt, image=init_image).images
```
When using `torch >= 2.0`, you can improve the inference speed by 20-30% with torch.compile. Simple wrap the unet with torch compile before running the pipeline:
```py
pipe.unet = torch.compile(pipe.unet, mode="reduce-overhead", fullgraph=True)
```
If you are limited by GPU VRAM, you can enable *cpu offloading* by calling `pipe.enable_model_cpu_offload`
instead of `.to("cuda")`:
```diff
- pipe.to("cuda")
+ pipe.enable_model_cpu_offload()
```
For more advanced use cases, please have a look at [the docs](https://huggingface.co/docs/diffusers/main/en/api/pipelines/stable_diffusion/stable_diffusion_xl).
## Uses
### Direct Use
The model is intended for research purposes only. Possible research areas and tasks include
- Generation of artworks and use in design and other artistic processes.
- Applications in educational or creative tools.
- Research on generative models.
- Safe deployment of models which have the potential to generate harmful content.
- Probing and understanding the limitations and biases of generative models.
Excluded uses are described below.
### Out-of-Scope Use
The model was not trained to be factual or true representations of people or events, and therefore using the model to generate such content is out-of-scope for the abilities of this model.
## Limitations and Bias
### Limitations
- The model does not achieve perfect photorealism
- The model cannot render legible text
- The model struggles with more difficult tasks which involve compositionality, such as rendering an image corresponding to “A red cube on top of a blue sphere”
- Faces and people in general may not be generated properly.
- The autoencoding part of the model is lossy.
### Bias
While the capabilities of image generation models are impressive, they can also reinforce or exacerbate social biases. |
AdamCodd/distilbert-base-uncased-finetuned-emotion-balanced | AdamCodd | 2023-11-10T20:11:37Z | 258 | 1 | transformers | [
"transformers",
"pytorch",
"onnx",
"safetensors",
"distilbert",
"text-classification",
"generated_from_trainer",
"dataset:AdamCodd/emotion-balanced",
"base_model:distilbert/distilbert-base-uncased",
"base_model:quantized:distilbert/distilbert-base-uncased",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| text-classification | 2023-09-08T18:17:39Z | ---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- AdamCodd/emotion-balanced
metrics:
- accuracy
- f1
- recall
- precision
widget:
- text: "He looked out of the rain-streaked window, lost in thought, the faintest hint of melancholy in his eyes, as he remembered moments from a distant past."
example_title: "Sadness"
- text: "As she strolled through the park, a soft smile played on her lips, and her heart felt lighter with each step, appreciating the simple beauty of nature."
example_title: "Joy"
- text: "Their fingers brushed lightly as they exchanged a knowing glance, a subtle connection that spoke volumes about the deep affection they held for each other."
example_title: "Love"
- text: "She clenched her fists and took a deep breath, trying to suppress the simmering frustration that welled up when her ideas were dismissed without consideration."
example_title: "Anger"
- text: "In the quiet of the night, the gentle rustling of leaves outside her window sent shivers down her spine, leaving her feeling uneasy and vulnerable."
example_title: "Fear"
- text: "Upon opening the old dusty book, a delicate, hand-painted map fell out, revealing hidden treasures she never expected to find."
example_title: "Surprise sentence"
base_model: distilbert-base-uncased
model-index:
- name: distilbert-base-uncased-finetuned-emotion-balanced
results:
- task:
type: text-classification
name: Text Classification
dataset:
name: emotion
type: emotion
args: default
metrics:
- type: accuracy
value: 0.9521
name: Accuracy
- type: loss
value: 0.1216
name: Loss
- type: f1
value: 0.9520944952964783
name: F1
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# distilbert-emotion
<u><b>Reupload [10/02/23]</b></u> : The model has been retrained using identical hyperparameters, but this time on an even more pristine dataset, free of certain scraping artifacts. Remarkably, it maintains the same level of accuracy and loss while demonstrating superior generalization capabilities.
This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the [emotion balanced dataset](https://huggingface.co/datasets/AdamCodd/emotion-balanced).
It achieves the following results on the evaluation set:
- Loss: 0.1216
- Accuracy: 0.9521
## Model description
This emotion classifier has been trained on 89_754 examples split into train, validation and test. Each label was perfectly balanced in each split.
## Intended uses & limitations
Usage:
```python
from transformers import pipeline
# Create the pipeline
emotion_classifier = pipeline('text-classification', model='AdamCodd/distilbert-base-uncased-finetuned-emotion-balanced')
# Now you can use the pipeline to classify emotions
result = emotion_classifier("We are delighted that you will be coming to visit us. It will be so nice to have you here.")
print(result)
#[{'label': 'joy', 'score': 0.9983291029930115}]
```
This model faces challenges in accurately categorizing negative sentences, as well as those containing elements of sarcasm or irony. These limitations are largely attributable to DistilBERT's constrained capabilities in semantic understanding. Although the model is generally proficient in emotion detection tasks, it may lack the nuance necessary for interpreting complex emotional nuances.
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 3e-05
- train_batch_size: 32
- eval_batch_size: 64
- seed: 1270
- optimizer: AdamW with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 150
- num_epochs: 3
- weight_decay: 0.01
### Training results
precision recall f1-score support
sadness 0.9882 0.9485 0.9679 1496
joy 0.9956 0.9057 0.9485 1496
love 0.9256 0.9980 0.9604 1496
anger 0.9628 0.9519 0.9573 1496
fear 0.9348 0.9098 0.9221 1496
surprise 0.9160 0.9987 0.9555 1496
accuracy 0.9521 8976
macro avg 0.9538 0.9521 0.9520 8976
weighted avg 0.9538 0.9521 0.9520 8976
test_acc: 0.9520944952964783
test_loss: 0.121663898229599
### Framework versions
- Transformers 4.33.2
- Pytorch lightning 2.0.9
- Tokenizers 0.13.3
If you want to support me, you can [here](https://ko-fi.com/adamcodd). |
TheBloke/Yi-34B-200K-AWQ | TheBloke | 2023-11-10T20:03:28Z | 27 | 9 | transformers | [
"transformers",
"safetensors",
"Yi",
"text-generation",
"custom_code",
"base_model:01-ai/Yi-34B-200K",
"base_model:quantized:01-ai/Yi-34B-200K",
"license:other",
"autotrain_compatible",
"4-bit",
"awq",
"region:us"
]
| text-generation | 2023-11-10T18:18:36Z | ---
base_model: 01-ai/Yi-34B-200K
inference: false
license: other
license_link: LICENSE
license_name: yi-license
model_creator: 01-ai
model_name: Yi 34B 200K
model_type: yi
prompt_template: '{prompt}
'
quantized_by: TheBloke
---
<!-- markdownlint-disable MD041 -->
<!-- header start -->
<!-- 200823 -->
<div style="width: auto; margin-left: auto; margin-right: auto">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<div style="text-align:center; margin-top: 0em; margin-bottom: 0em"><p style="margin-top: 0.25em; margin-bottom: 0em;">TheBloke's LLM work is generously supported by a grant from <a href="https://a16z.com">andreessen horowitz (a16z)</a></p></div>
<hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
<!-- header end -->
# Yi 34B 200K - AWQ
- Model creator: [01-ai](https://huggingface.co/01-ai)
- Original model: [Yi 34B 200K](https://huggingface.co/01-ai/Yi-34B-200K)
<!-- description start -->
## Description
This repo contains AWQ model files for [01-ai's Yi 34B 200K](https://huggingface.co/01-ai/Yi-34B-200K).
These files were quantised using hardware kindly provided by [Massed Compute](https://massedcompute.com/).
### About AWQ
AWQ is an efficient, accurate and blazing-fast low-bit weight quantization method, currently supporting 4-bit quantization. Compared to GPTQ, it offers faster Transformers-based inference with equivalent or better quality compared to the most commonly used GPTQ settings.
It is supported by:
- [Text Generation Webui](https://github.com/oobabooga/text-generation-webui) - using Loader: AutoAWQ
- [vLLM](https://github.com/vllm-project/vllm) - Llama and Mistral models only
- [Hugging Face Text Generation Inference (TGI)](https://github.com/huggingface/text-generation-inference)
- [Transformers](https://huggingface.co/docs/transformers) version 4.35.0 and later, from any code or client that supports Transformers
- [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) - for use from Python code
<!-- description end -->
<!-- repositories-available start -->
## Repositories available
* [AWQ model(s) for GPU inference.](https://huggingface.co/TheBloke/Yi-34B-200K-AWQ)
* [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/Yi-34B-200K-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/Yi-34B-200K-GGUF)
* [01-ai's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/01-ai/Yi-34B-200K)
<!-- repositories-available end -->
<!-- prompt-template start -->
## Prompt template: None
```
{prompt}
```
<!-- prompt-template end -->
<!-- README_AWQ.md-provided-files start -->
## Provided files, and AWQ parameters
I currently release 128g GEMM models only. The addition of group_size 32 models, and GEMV kernel models, is being actively considered.
Models are released as sharded safetensors files.
| Branch | Bits | GS | AWQ Dataset | Seq Len | Size |
| ------ | ---- | -- | ----------- | ------- | ---- |
| [main](https://huggingface.co/TheBloke/Yi-34B-200K-AWQ/tree/main) | 4 | 128 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 4096 | 19.23 GB
<!-- README_AWQ.md-provided-files end -->
<!-- README_AWQ.md-text-generation-webui start -->
## How to easily download and use this model in [text-generation-webui](https://github.com/oobabooga/text-generation-webui)
Please make sure you're using the latest version of [text-generation-webui](https://github.com/oobabooga/text-generation-webui).
It is strongly recommended to use the text-generation-webui one-click-installers unless you're sure you know how to make a manual install.
1. Click the **Model tab**.
2. Under **Download custom model or LoRA**, enter `TheBloke/Yi-34B-200K-AWQ`.
3. Click **Download**.
4. The model will start downloading. Once it's finished it will say "Done".
5. In the top left, click the refresh icon next to **Model**.
6. In the **Model** dropdown, choose the model you just downloaded: `Yi-34B-200K-AWQ`
7. Select **Loader: AutoAWQ**.
8. Click Load, and the model will load and is now ready for use.
9. If you want any custom settings, set them and then click **Save settings for this model** followed by **Reload the Model** in the top right.
10. Once you're ready, click the **Text Generation** tab and enter a prompt to get started!
<!-- README_AWQ.md-text-generation-webui end -->
<!-- README_AWQ.md-use-from-vllm start -->
## Multi-user inference server: vLLM
Documentation on installing and using vLLM [can be found here](https://vllm.readthedocs.io/en/latest/).
- Please ensure you are using vLLM version 0.2 or later.
- When using vLLM as a server, pass the `--quantization awq` parameter.
For example:
```shell
python3 -m vllm.entrypoints.api_server --model TheBloke/Yi-34B-200K-AWQ --quantization awq --dtype auto
```
- When using vLLM from Python code, again set `quantization=awq`.
For example:
```python
from vllm import LLM, SamplingParams
prompts = [
"Tell me about AI",
"Write a story about llamas",
"What is 291 - 150?",
"How much wood would a woodchuck chuck if a woodchuck could chuck wood?",
]
prompt_template=f'''{prompt}
'''
prompts = [prompt_template.format(prompt=prompt) for prompt in prompts]
sampling_params = SamplingParams(temperature=0.8, top_p=0.95)
llm = LLM(model="TheBloke/Yi-34B-200K-AWQ", quantization="awq", dtype="auto")
outputs = llm.generate(prompts, sampling_params)
# Print the outputs.
for output in outputs:
prompt = output.prompt
generated_text = output.outputs[0].text
print(f"Prompt: {prompt!r}, Generated text: {generated_text!r}")
```
<!-- README_AWQ.md-use-from-vllm start -->
<!-- README_AWQ.md-use-from-tgi start -->
## Multi-user inference server: Hugging Face Text Generation Inference (TGI)
Use TGI version 1.1.0 or later. The official Docker container is: `ghcr.io/huggingface/text-generation-inference:1.1.0`
Example Docker parameters:
```shell
--model-id TheBloke/Yi-34B-200K-AWQ --port 3000 --quantize awq --max-input-length 3696 --max-total-tokens 4096 --max-batch-prefill-tokens 4096
```
Example Python code for interfacing with TGI (requires [huggingface-hub](https://github.com/huggingface/huggingface_hub) 0.17.0 or later):
```shell
pip3 install huggingface-hub
```
```python
from huggingface_hub import InferenceClient
endpoint_url = "https://your-endpoint-url-here"
prompt = "Tell me about AI"
prompt_template=f'''{prompt}
'''
client = InferenceClient(endpoint_url)
response = client.text_generation(prompt,
max_new_tokens=128,
do_sample=True,
temperature=0.7,
top_p=0.95,
top_k=40,
repetition_penalty=1.1)
print(f"Model output: ", response)
```
<!-- README_AWQ.md-use-from-tgi end -->
<!-- README_AWQ.md-use-from-python start -->
## Inference from Python code using Transformers
### Install the necessary packages
- Requires: [Transformers](https://huggingface.co/docs/transformers) 4.35.0 or later.
- Requires: [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) 0.1.6 or later.
```shell
pip3 install --upgrade "autoawq>=0.1.6" "transformers>=4.35.0"
```
Note that if you are using PyTorch 2.0.1, the above AutoAWQ command will automatically upgrade you to PyTorch 2.1.0.
If you are using CUDA 11.8 and wish to continue using PyTorch 2.0.1, instead run this command:
```shell
pip3 install https://github.com/casper-hansen/AutoAWQ/releases/download/v0.1.6/autoawq-0.1.6+cu118-cp310-cp310-linux_x86_64.whl
```
If you have problems installing [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) using the pre-built wheels, install it from source instead:
```shell
pip3 uninstall -y autoawq
git clone https://github.com/casper-hansen/AutoAWQ
cd AutoAWQ
pip3 install .
```
### Transformers example code (requires Transformers 4.35.0 and later)
```python
from transformers import AutoModelForCausalLM, AutoTokenizer, TextStreamer
model_name_or_path = "TheBloke/Yi-34B-200K-AWQ"
tokenizer = AutoTokenizer.from_pretrained(model_name_or_path)
model = AutoModelForCausalLM.from_pretrained(
model_name_or_path,
low_cpu_mem_usage=True,
device_map="cuda:0"
)
# Using the text streamer to stream output one token at a time
streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
prompt = "Tell me about AI"
prompt_template=f'''{prompt}
'''
# Convert prompt to tokens
tokens = tokenizer(
prompt_template,
return_tensors='pt'
).input_ids.cuda()
generation_params = {
"do_sample": True,
"temperature": 0.7,
"top_p": 0.95,
"top_k": 40,
"max_new_tokens": 512,
"repetition_penalty": 1.1
}
# Generate streamed output, visible one token at a time
generation_output = model.generate(
tokens,
streamer=streamer,
**generation_params
)
# Generation without a streamer, which will include the prompt in the output
generation_output = model.generate(
tokens,
**generation_params
)
# Get the tokens from the output, decode them, print them
token_output = generation_output[0]
text_output = tokenizer.decode(token_output)
print("model.generate output: ", text_output)
# Inference is also possible via Transformers' pipeline
from transformers import pipeline
pipe = pipeline(
"text-generation",
model=model,
tokenizer=tokenizer,
**generation_params
)
pipe_output = pipe(prompt_template)[0]['generated_text']
print("pipeline output: ", pipe_output)
```
<!-- README_AWQ.md-use-from-python end -->
<!-- README_AWQ.md-compatibility start -->
## Compatibility
The files provided are tested to work with:
- [text-generation-webui](https://github.com/oobabooga/text-generation-webui) using `Loader: AutoAWQ`.
- [vLLM](https://github.com/vllm-project/vllm) version 0.2.0 and later.
- [Hugging Face Text Generation Inference (TGI)](https://github.com/huggingface/text-generation-inference) version 1.1.0 and later.
- [Transformers](https://huggingface.co/docs/transformers) version 4.35.0 and later.
- [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) version 0.1.1 and later.
<!-- README_AWQ.md-compatibility end -->
<!-- footer start -->
<!-- 200823 -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
## Thanks, and how to contribute
Thanks to the [chirper.ai](https://chirper.ai) team!
Thanks to Clay from [gpus.llm-utils.org](llm-utils)!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Aemon Algiz.
**Patreon special mentions**: Brandon Frisco, LangChain4j, Spiking Neurons AB, transmissions 11, Joseph William Delisle, Nitin Borwankar, Willem Michiel, Michael Dempsey, vamX, Jeffrey Morgan, zynix, jjj, Omer Bin Jawed, Sean Connelly, jinyuan sun, Jeromy Smith, Shadi, Pawan Osman, Chadd, Elijah Stavena, Illia Dulskyi, Sebastain Graf, Stephen Murray, terasurfer, Edmond Seymore, Celu Ramasamy, Mandus, Alex, biorpg, Ajan Kanaga, Clay Pascal, Raven Klaugh, 阿明, K, ya boyyy, usrbinkat, Alicia Loh, John Villwock, ReadyPlayerEmma, Chris Smitley, Cap'n Zoog, fincy, GodLy, S_X, sidney chen, Cory Kujawski, OG, Mano Prime, AzureBlack, Pieter, Kalila, Spencer Kim, Tom X Nguyen, Stanislav Ovsiannikov, Michael Levine, Andrey, Trailburnt, Vadim, Enrico Ros, Talal Aujan, Brandon Phillips, Jack West, Eugene Pentland, Michael Davis, Will Dee, webtim, Jonathan Leane, Alps Aficionado, Rooh Singh, Tiffany J. Kim, theTransient, Luke @flexchar, Elle, Caitlyn Gatomon, Ari Malik, subjectnull, Johann-Peter Hartmann, Trenton Dambrowitz, Imad Khwaja, Asp the Wyvern, Emad Mostaque, Rainer Wilmers, Alexandros Triantafyllidis, Nicholas, Pedro Madruga, SuperWojo, Harry Royden McLaughlin, James Bentley, Olakabola, David Ziegler, Ai Maven, Jeff Scroggin, Nikolai Manek, Deo Leter, Matthew Berman, Fen Risland, Ken Nordquist, Manuel Alberto Morcote, Luke Pendergrass, TL, Fred von Graf, Randy H, Dan Guido, NimbleBox.ai, Vitor Caleffi, Gabriel Tamborski, knownsqashed, Lone Striker, Erik Bjäreholt, John Detwiler, Leonard Tan, Iucharbius
Thank you to all my generous patrons and donaters!
And thank you again to a16z for their generous grant.
<!-- footer end -->
# Original model card: 01-ai's Yi 34B 200K
<div align="center">
<img src="./Yi.svg" width="200px">
</div>
## Introduction
The **Yi** series models are large language models trained from scratch by
developers at [01.AI](https://01.ai/). The first public release contains two
bilingual(English/Chinese) base models with the parameter sizes of 6B([`Yi-6B`](https://huggingface.co/01-ai/Yi-6B))
and 34B([`Yi-34B`](https://huggingface.co/01-ai/Yi-34B)). Both of them are trained
with 4K sequence length and can be extended to 32K during inference time.
The [`Yi-6B-200K`](https://huggingface.co/01-ai/Yi-6B-200K)
and [`Yi-34B-200K`](https://huggingface.co/01-ai/Yi-34B-200K) are base model with
200K context length.
## News
- 🎯 **2023/11/06**: The base model of [`Yi-6B-200K`](https://huggingface.co/01-ai/Yi-6B-200K)
and [`Yi-34B-200K`](https://huggingface.co/01-ai/Yi-34B-200K) with 200K context length.
- 🎯 **2023/11/02**: The base model of [`Yi-6B`](https://huggingface.co/01-ai/Yi-6B) and
[`Yi-34B`](https://huggingface.co/01-ai/Yi-34B).
## Model Performance
| Model | MMLU | CMMLU | C-Eval | GAOKAO | BBH | Common-sense Reasoning | Reading Comprehension | Math & Code |
| :------------ | :------: | :------: | :------: | :------: | :------: | :--------------------: | :-------------------: | :---------: |
| | 5-shot | 5-shot | 5-shot | 0-shot | 3-shot@1 | - | - | - |
| LLaMA2-34B | 62.6 | - | - | - | 44.1 | 69.9 | 68.0 | 26.0 |
| LLaMA2-70B | 68.9 | 53.3 | - | 49.8 | 51.2 | 71.9 | 69.4 | 36.8 |
| Baichuan2-13B | 59.2 | 62.0 | 58.1 | 54.3 | 48.8 | 64.3 | 62.4 | 23.0 |
| Qwen-14B | 66.3 | 71.0 | 72.1 | 62.5 | 53.4 | 73.3 | 72.5 | **39.8** |
| Skywork-13B | 62.1 | 61.8 | 60.6 | 68.1 | 41.7 | 72.4 | 61.4 | 24.9 |
| InternLM-20B | 62.1 | 59.0 | 58.8 | 45.5 | 52.5 | 78.3 | - | 30.4 |
| Aquila-34B | 67.8 | 71.4 | 63.1 | - | - | - | - | - |
| Falcon-180B | 70.4 | 58.0 | 57.8 | 59.0 | 54.0 | 77.3 | 68.8 | 34.0 |
| Yi-6B | 63.2 | 75.5 | 72.0 | 72.2 | 42.8 | 72.3 | 68.7 | 19.8 |
| Yi-6B-200K | 64.0 | 75.3 | 73.5 | 73.9 | 42.0 | 72.0 | 69.1 | 19.0 |
| **Yi-34B** | **76.3** | **83.7** | 81.4 | 82.8 | **54.3** | **80.1** | 76.4 | 37.1 |
| Yi-34B-200K | 76.1 | 83.6 | **81.9** | **83.4** | 52.7 | 79.7 | **76.6** | 36.3 |
While benchmarking open-source models, we have observed a disparity between the
results generated by our pipeline and those reported in public sources (e.g.
OpenCompass). Upon conducting a more in-depth investigation of this difference,
we have discovered that various models may employ different prompts,
post-processing strategies, and sampling techniques, potentially resulting in
significant variations in the outcomes. Our prompt and post-processing strategy
remains consistent with the original benchmark, and greedy decoding is employed
during evaluation without any post-processing for the generated content. For
scores that were not reported by the original authors (including scores reported
with different settings), we try to get results with our pipeline.
To evaluate the model's capability extensively, we adopted the methodology
outlined in Llama2. Specifically, we included PIQA, SIQA, HellaSwag, WinoGrande,
ARC, OBQA, and CSQA to assess common sense reasoning. SquAD, QuAC, and BoolQ
were incorporated to evaluate reading comprehension. CSQA was exclusively tested
using a 7-shot setup, while all other tests were conducted with a 0-shot
configuration. Additionally, we introduced GSM8K (8-shot@1), MATH (4-shot@1),
HumanEval (0-shot@1), and MBPP (3-shot@1) under the category "Math & Code". Due
to technical constraints, we did not test Falcon-180 on QuAC and OBQA; the score
is derived by averaging the scores on the remaining tasks. Since the scores for
these two tasks are generally lower than the average, we believe that
Falcon-180B's performance was not underestimated.
## Usage
Please visit our [github repository](https://github.com/01-ai/Yi) for general
guidance on how to use this model.
## Disclaimer
Although we use data compliance checking algorithms during the training process
to ensure the compliance of the trained model to the best of our ability, due to
the complexity of the data and the diversity of language model usage scenarios,
we cannot guarantee that the model will generate correct and reasonable output
in all scenarios. Please be aware that there is still a risk of the model
producing problematic outputs. We will not be responsible for any risks and
issues resulting from misuse, misguidance, illegal usage, and related
misinformation, as well as any associated data security concerns.
## License
The Yi series models are fully open for academic research and free commercial
usage with permission via applications. All usage must adhere to the [Model
License Agreement 2.0](https://huggingface.co/01-ai/Yi-34B-200K/blob/main/LICENSE). To
apply for the official commercial license, please contact us
([[email protected]](mailto:[email protected])).
|
Rdgaudio/RVCmodels | Rdgaudio | 2023-11-10T20:02:20Z | 0 | 0 | null | [
"license:cc-by-nc-nd-4.0",
"region:us"
]
| null | 2023-11-10T18:50:51Z | ---
license: cc-by-nc-nd-4.0
---
|
LoneStriker/prometheus-13b-v1.0-8.0bpw-h8-exl2 | LoneStriker | 2023-11-10T19:57:00Z | 5 | 0 | transformers | [
"transformers",
"pytorch",
"llama",
"text-generation",
"text2text-generation",
"en",
"dataset:kaist-ai/Feedback-Collection",
"arxiv:2310.08491",
"license:apache-2.0",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
]
| text2text-generation | 2023-11-10T19:56:00Z | ---
tags:
- text2text-generation
datasets:
- kaist-ai/Feedback-Collection
license: apache-2.0
language:
- en
pipeline_tag: text2text-generation
library_name: transformers
metrics:
- pearsonr
- spearmanr
- accuracy
---
## Links for Reference
- **Homepage:https://github.com/kaistAI/Prometheus**
- **Repository:https://github.com/kaistAI/Prometheus**
- **Paper:https://arxiv.org/abs/2310.08491**
- **Point of Contact:[email protected]**
# TL;DR
Prometheus is an alternative of GPT-4 evaluation when doing fine-grained evaluation of an underlying LLM & a Reward model for Reinforcement Learning from Human Feedback (RLHF).
Prometheus is a language model using [Llama-2-Chat](https://huggingface.co/meta-llama/Llama-2-13b-chat-hf) as a base model and fine-tuned on 100K feedback within the [Feedback Collection](https://huggingface.co/datasets/kaist-ai/Feedback-Collection).
Since it was fine-tuned on a large amount of feedback, it is specialized at evaluating long-form responses, outperforming GPT-3.5-Turbo, Llama-2-Chat 70B, and on par with GPT-4 on various benchmarks.
Most importantly, this was possible since we appended 2 reference materials (reference answer, and customized score rubric).
Prometheus is a cheap and powerful alternative to GPT-4 evaluation, which one could use to evaluate LLMs with customized criteria (e.g., Child readability, Cultural Sensitivity, Creativity).
Also, it could be used as a reward model for Reinforcement Learning from Human Feedback (RLHF).
# Model Details
## Model Description
- **Model type:** Language model
- **Language(s) (NLP):** English
- **License:** Apache 2.0
- **Related Models:** [All Prometheus Checkpoints](https://huggingface.co/models?search=kaist-ai/Prometheus)
- **Resources for more information:**
- [Research paper](https://arxiv.org/abs/2310.08491)
- [GitHub Repo](https://github.com/kaistAI/Prometheus)
Prometheus is trained with two different sizes (7B and 13B).
You could check the 7B sized LM on [this page](https://huggingface.co/kaist-ai/prometheus-7b-v1.0).
Also, check out our dataset as well on [this page](https://huggingface.co/datasets/kaist-ai/Feedback-Collection).
## Prompt Format
Prometheus requires 4 components in the input: An instruction, a response to evaluate, a score rubric, and a reference answer. You could refer to the prompt format below.
You should fill in the instruction, response, reference answer, criteria description, and score description for score in range of 1 to 5.
```
###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{instruction}
###Response to evaluate:
{response}
###Reference Answer (Score 5):
{reference_answer}
###Score Rubrics:
[{criteria_description}]
Score 1: {score1_description}
Score 2: {score2_description}
Score 3: {score3_description}
Score 4: {score4_description}
Score 5: {score5_description}
###Feedback:
```
After this, you should apply the conversation template of Llama-2-Chat (not applying it might lead to unexpected behaviors).
You can find the conversation class at this [link](https://github.com/lm-sys/FastChat/blob/main/fastchat/conversation.py).
```
conv = get_conv_template("llama-2")
conv.set_system_message("You are a fair evaluator language model.")
conv.append_message(conv.roles[0], dialogs['instruction'])
conv.append_message(conv.roles[1], None)
prompt = conv.get_prompt()
x = tokenizer(prompt,truncation=False)
```
As a result, a feedback and score decision will be generated, divided by a separating phrase ```[RESULT]```
## License
Feedback Collection and Prometheus is subject to OpenAI's Terms of Use for the generated data. If you suspect any violations, please reach out to us.
# Usage
Find below some example scripts on how to use the model in `transformers`:
## Using the Pytorch model
### Running the model on a CPU
<details>
<summary> Click to expand </summary>
```python
from transformers import T5Tokenizer, T5ForConditionalGeneration
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto")
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
### Running the model on a GPU
<details>
<summary> Click to expand </summary>
```python
# pip install accelerate
import torch
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto")
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids, sample=True, temperature=1.0, top_p=0.9, max_new_tokens=256, repetition_penalty=1.03)
print(tokenizer.decode(outputs[0]))
```
</details>
### Running the model on a GPU using different precisions
#### FP16
<details>
<summary> Click to expand </summary>
```python
# pip install accelerate
import torch
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto", torch_dtype=torch.float16)
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
#### INT8
<details>
<summary> Click to expand </summary>
```python
# pip install bitsandbytes accelerate
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto", load_in_8bit=True)
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
# Citation
If you find the following model helpful, please consider citing our paper!
**BibTeX:**
```bibtex
@misc{kim2023prometheus,
title={Prometheus: Inducing Fine-grained Evaluation Capability in Language Models},
author={Seungone Kim and Jamin Shin and Yejin Cho and Joel Jang and Shayne Longpre and Hwaran Lee and Sangdoo Yun and Seongjin Shin and Sungdong Kim and James Thorne and Minjoon Seo},
year={2023},
eprint={2310.08491},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
``` |
hkivancoral/hushem_1x_deit_tiny_sgd_lr00001_fold5 | hkivancoral | 2023-11-10T19:53:39Z | 6 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:facebook/deit-tiny-patch16-224",
"base_model:finetune:facebook/deit-tiny-patch16-224",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| image-classification | 2023-11-10T19:52:10Z | ---
license: apache-2.0
base_model: facebook/deit-tiny-patch16-224
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: hushem_1x_deit_tiny_sgd_lr00001_fold5
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: test
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.12195121951219512
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# hushem_1x_deit_tiny_sgd_lr00001_fold5
This model is a fine-tuned version of [facebook/deit-tiny-patch16-224](https://huggingface.co/facebook/deit-tiny-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 1.6421
- Accuracy: 0.1220
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 1.0 | 6 | 1.6490 | 0.1220 |
| 1.6062 | 2.0 | 12 | 1.6487 | 0.1220 |
| 1.6062 | 3.0 | 18 | 1.6483 | 0.1220 |
| 1.6229 | 4.0 | 24 | 1.6480 | 0.1220 |
| 1.5995 | 5.0 | 30 | 1.6477 | 0.1220 |
| 1.5995 | 6.0 | 36 | 1.6474 | 0.1220 |
| 1.5906 | 7.0 | 42 | 1.6470 | 0.1220 |
| 1.5906 | 8.0 | 48 | 1.6468 | 0.1220 |
| 1.609 | 9.0 | 54 | 1.6465 | 0.1220 |
| 1.6018 | 10.0 | 60 | 1.6462 | 0.1220 |
| 1.6018 | 11.0 | 66 | 1.6459 | 0.1220 |
| 1.5944 | 12.0 | 72 | 1.6457 | 0.1220 |
| 1.5944 | 13.0 | 78 | 1.6454 | 0.1220 |
| 1.6013 | 14.0 | 84 | 1.6452 | 0.1220 |
| 1.5987 | 15.0 | 90 | 1.6449 | 0.1220 |
| 1.5987 | 16.0 | 96 | 1.6447 | 0.1220 |
| 1.5899 | 17.0 | 102 | 1.6445 | 0.1220 |
| 1.5899 | 18.0 | 108 | 1.6443 | 0.1220 |
| 1.626 | 19.0 | 114 | 1.6441 | 0.1220 |
| 1.5972 | 20.0 | 120 | 1.6439 | 0.1220 |
| 1.5972 | 21.0 | 126 | 1.6437 | 0.1220 |
| 1.5649 | 22.0 | 132 | 1.6436 | 0.1220 |
| 1.5649 | 23.0 | 138 | 1.6434 | 0.1220 |
| 1.6699 | 24.0 | 144 | 1.6433 | 0.1220 |
| 1.5696 | 25.0 | 150 | 1.6431 | 0.1220 |
| 1.5696 | 26.0 | 156 | 1.6430 | 0.1220 |
| 1.5743 | 27.0 | 162 | 1.6429 | 0.1220 |
| 1.5743 | 28.0 | 168 | 1.6427 | 0.1220 |
| 1.6236 | 29.0 | 174 | 1.6426 | 0.1220 |
| 1.5936 | 30.0 | 180 | 1.6426 | 0.1220 |
| 1.5936 | 31.0 | 186 | 1.6425 | 0.1220 |
| 1.5875 | 32.0 | 192 | 1.6424 | 0.1220 |
| 1.5875 | 33.0 | 198 | 1.6423 | 0.1220 |
| 1.6171 | 34.0 | 204 | 1.6423 | 0.1220 |
| 1.5897 | 35.0 | 210 | 1.6422 | 0.1220 |
| 1.5897 | 36.0 | 216 | 1.6422 | 0.1220 |
| 1.5725 | 37.0 | 222 | 1.6421 | 0.1220 |
| 1.5725 | 38.0 | 228 | 1.6421 | 0.1220 |
| 1.6227 | 39.0 | 234 | 1.6421 | 0.1220 |
| 1.5924 | 40.0 | 240 | 1.6421 | 0.1220 |
| 1.5924 | 41.0 | 246 | 1.6421 | 0.1220 |
| 1.5811 | 42.0 | 252 | 1.6421 | 0.1220 |
| 1.5811 | 43.0 | 258 | 1.6421 | 0.1220 |
| 1.6072 | 44.0 | 264 | 1.6421 | 0.1220 |
| 1.5938 | 45.0 | 270 | 1.6421 | 0.1220 |
| 1.5938 | 46.0 | 276 | 1.6421 | 0.1220 |
| 1.6243 | 47.0 | 282 | 1.6421 | 0.1220 |
| 1.6243 | 48.0 | 288 | 1.6421 | 0.1220 |
| 1.5633 | 49.0 | 294 | 1.6421 | 0.1220 |
| 1.6091 | 50.0 | 300 | 1.6421 | 0.1220 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
hkivancoral/hushem_1x_deit_tiny_sgd_lr00001_fold3 | hkivancoral | 2023-11-10T19:50:34Z | 6 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:facebook/deit-tiny-patch16-224",
"base_model:finetune:facebook/deit-tiny-patch16-224",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| image-classification | 2023-11-10T19:49:04Z | ---
license: apache-2.0
base_model: facebook/deit-tiny-patch16-224
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: hushem_1x_deit_tiny_sgd_lr00001_fold3
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: test
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.3023255813953488
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# hushem_1x_deit_tiny_sgd_lr00001_fold3
This model is a fine-tuned version of [facebook/deit-tiny-patch16-224](https://huggingface.co/facebook/deit-tiny-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 1.5383
- Accuracy: 0.3023
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 1.0 | 6 | 1.5437 | 0.3023 |
| 1.6283 | 2.0 | 12 | 1.5434 | 0.3023 |
| 1.6283 | 3.0 | 18 | 1.5431 | 0.3023 |
| 1.63 | 4.0 | 24 | 1.5428 | 0.3023 |
| 1.6367 | 5.0 | 30 | 1.5426 | 0.3023 |
| 1.6367 | 6.0 | 36 | 1.5423 | 0.3023 |
| 1.6273 | 7.0 | 42 | 1.5421 | 0.3023 |
| 1.6273 | 8.0 | 48 | 1.5419 | 0.3023 |
| 1.6489 | 9.0 | 54 | 1.5417 | 0.3023 |
| 1.5924 | 10.0 | 60 | 1.5414 | 0.3023 |
| 1.5924 | 11.0 | 66 | 1.5412 | 0.3023 |
| 1.6227 | 12.0 | 72 | 1.5411 | 0.3023 |
| 1.6227 | 13.0 | 78 | 1.5409 | 0.3023 |
| 1.6367 | 14.0 | 84 | 1.5407 | 0.3023 |
| 1.622 | 15.0 | 90 | 1.5405 | 0.3023 |
| 1.622 | 16.0 | 96 | 1.5403 | 0.3023 |
| 1.621 | 17.0 | 102 | 1.5401 | 0.3023 |
| 1.621 | 18.0 | 108 | 1.5400 | 0.3023 |
| 1.6386 | 19.0 | 114 | 1.5398 | 0.3023 |
| 1.6207 | 20.0 | 120 | 1.5397 | 0.3023 |
| 1.6207 | 21.0 | 126 | 1.5395 | 0.3023 |
| 1.6152 | 22.0 | 132 | 1.5394 | 0.3023 |
| 1.6152 | 23.0 | 138 | 1.5393 | 0.3023 |
| 1.6503 | 24.0 | 144 | 1.5392 | 0.3023 |
| 1.6219 | 25.0 | 150 | 1.5390 | 0.3023 |
| 1.6219 | 26.0 | 156 | 1.5389 | 0.3023 |
| 1.6152 | 27.0 | 162 | 1.5389 | 0.3023 |
| 1.6152 | 28.0 | 168 | 1.5388 | 0.3023 |
| 1.6192 | 29.0 | 174 | 1.5387 | 0.3023 |
| 1.6111 | 30.0 | 180 | 1.5386 | 0.3023 |
| 1.6111 | 31.0 | 186 | 1.5386 | 0.3023 |
| 1.6114 | 32.0 | 192 | 1.5385 | 0.3023 |
| 1.6114 | 33.0 | 198 | 1.5384 | 0.3023 |
| 1.6361 | 34.0 | 204 | 1.5384 | 0.3023 |
| 1.6146 | 35.0 | 210 | 1.5384 | 0.3023 |
| 1.6146 | 36.0 | 216 | 1.5383 | 0.3023 |
| 1.6254 | 37.0 | 222 | 1.5383 | 0.3023 |
| 1.6254 | 38.0 | 228 | 1.5383 | 0.3023 |
| 1.6124 | 39.0 | 234 | 1.5383 | 0.3023 |
| 1.6367 | 40.0 | 240 | 1.5383 | 0.3023 |
| 1.6367 | 41.0 | 246 | 1.5383 | 0.3023 |
| 1.6229 | 42.0 | 252 | 1.5383 | 0.3023 |
| 1.6229 | 43.0 | 258 | 1.5383 | 0.3023 |
| 1.6506 | 44.0 | 264 | 1.5383 | 0.3023 |
| 1.6148 | 45.0 | 270 | 1.5383 | 0.3023 |
| 1.6148 | 46.0 | 276 | 1.5383 | 0.3023 |
| 1.6242 | 47.0 | 282 | 1.5383 | 0.3023 |
| 1.6242 | 48.0 | 288 | 1.5383 | 0.3023 |
| 1.6087 | 49.0 | 294 | 1.5383 | 0.3023 |
| 1.6097 | 50.0 | 300 | 1.5383 | 0.3023 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
hkivancoral/hushem_1x_deit_tiny_sgd_lr00001_fold2 | hkivancoral | 2023-11-10T19:49:00Z | 10 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:facebook/deit-tiny-patch16-224",
"base_model:finetune:facebook/deit-tiny-patch16-224",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| image-classification | 2023-11-10T19:47:25Z | ---
license: apache-2.0
base_model: facebook/deit-tiny-patch16-224
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: hushem_1x_deit_tiny_sgd_lr00001_fold2
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: test
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.13333333333333333
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# hushem_1x_deit_tiny_sgd_lr00001_fold2
This model is a fine-tuned version of [facebook/deit-tiny-patch16-224](https://huggingface.co/facebook/deit-tiny-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 1.6351
- Accuracy: 0.1333
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 1.0 | 6 | 1.6407 | 0.1333 |
| 1.6149 | 2.0 | 12 | 1.6404 | 0.1333 |
| 1.6149 | 3.0 | 18 | 1.6401 | 0.1333 |
| 1.588 | 4.0 | 24 | 1.6398 | 0.1333 |
| 1.6243 | 5.0 | 30 | 1.6396 | 0.1333 |
| 1.6243 | 6.0 | 36 | 1.6393 | 0.1333 |
| 1.6041 | 7.0 | 42 | 1.6390 | 0.1333 |
| 1.6041 | 8.0 | 48 | 1.6388 | 0.1333 |
| 1.5784 | 9.0 | 54 | 1.6386 | 0.1333 |
| 1.61 | 10.0 | 60 | 1.6383 | 0.1333 |
| 1.61 | 11.0 | 66 | 1.6381 | 0.1333 |
| 1.5857 | 12.0 | 72 | 1.6379 | 0.1333 |
| 1.5857 | 13.0 | 78 | 1.6377 | 0.1333 |
| 1.6282 | 14.0 | 84 | 1.6375 | 0.1333 |
| 1.5739 | 15.0 | 90 | 1.6373 | 0.1333 |
| 1.5739 | 16.0 | 96 | 1.6372 | 0.1333 |
| 1.5784 | 17.0 | 102 | 1.6370 | 0.1333 |
| 1.5784 | 18.0 | 108 | 1.6368 | 0.1333 |
| 1.6525 | 19.0 | 114 | 1.6367 | 0.1333 |
| 1.5978 | 20.0 | 120 | 1.6365 | 0.1333 |
| 1.5978 | 21.0 | 126 | 1.6364 | 0.1333 |
| 1.6239 | 22.0 | 132 | 1.6362 | 0.1333 |
| 1.6239 | 23.0 | 138 | 1.6361 | 0.1333 |
| 1.581 | 24.0 | 144 | 1.6360 | 0.1333 |
| 1.597 | 25.0 | 150 | 1.6359 | 0.1333 |
| 1.597 | 26.0 | 156 | 1.6358 | 0.1333 |
| 1.5864 | 27.0 | 162 | 1.6357 | 0.1333 |
| 1.5864 | 28.0 | 168 | 1.6356 | 0.1333 |
| 1.6236 | 29.0 | 174 | 1.6355 | 0.1333 |
| 1.6201 | 30.0 | 180 | 1.6354 | 0.1333 |
| 1.6201 | 31.0 | 186 | 1.6354 | 0.1333 |
| 1.6018 | 32.0 | 192 | 1.6353 | 0.1333 |
| 1.6018 | 33.0 | 198 | 1.6352 | 0.1333 |
| 1.5711 | 34.0 | 204 | 1.6352 | 0.1333 |
| 1.6003 | 35.0 | 210 | 1.6352 | 0.1333 |
| 1.6003 | 36.0 | 216 | 1.6351 | 0.1333 |
| 1.5762 | 37.0 | 222 | 1.6351 | 0.1333 |
| 1.5762 | 38.0 | 228 | 1.6351 | 0.1333 |
| 1.5979 | 39.0 | 234 | 1.6351 | 0.1333 |
| 1.6035 | 40.0 | 240 | 1.6351 | 0.1333 |
| 1.6035 | 41.0 | 246 | 1.6351 | 0.1333 |
| 1.5976 | 42.0 | 252 | 1.6351 | 0.1333 |
| 1.5976 | 43.0 | 258 | 1.6351 | 0.1333 |
| 1.5981 | 44.0 | 264 | 1.6351 | 0.1333 |
| 1.5912 | 45.0 | 270 | 1.6351 | 0.1333 |
| 1.5912 | 46.0 | 276 | 1.6351 | 0.1333 |
| 1.5981 | 47.0 | 282 | 1.6351 | 0.1333 |
| 1.5981 | 48.0 | 288 | 1.6351 | 0.1333 |
| 1.6158 | 49.0 | 294 | 1.6351 | 0.1333 |
| 1.593 | 50.0 | 300 | 1.6351 | 0.1333 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
hkivancoral/hushem_1x_deit_tiny_sgd_lr00001_fold1 | hkivancoral | 2023-11-10T19:47:02Z | 6 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:facebook/deit-tiny-patch16-224",
"base_model:finetune:facebook/deit-tiny-patch16-224",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| image-classification | 2023-11-10T19:45:27Z | ---
license: apache-2.0
base_model: facebook/deit-tiny-patch16-224
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: hushem_1x_deit_tiny_sgd_lr00001_fold1
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: test
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.26666666666666666
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# hushem_1x_deit_tiny_sgd_lr00001_fold1
This model is a fine-tuned version of [facebook/deit-tiny-patch16-224](https://huggingface.co/facebook/deit-tiny-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 1.6574
- Accuracy: 0.2667
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 1.0 | 6 | 1.6633 | 0.2667 |
| 1.6088 | 2.0 | 12 | 1.6630 | 0.2667 |
| 1.6088 | 3.0 | 18 | 1.6627 | 0.2667 |
| 1.5763 | 4.0 | 24 | 1.6624 | 0.2667 |
| 1.6076 | 5.0 | 30 | 1.6621 | 0.2667 |
| 1.6076 | 6.0 | 36 | 1.6618 | 0.2667 |
| 1.5951 | 7.0 | 42 | 1.6616 | 0.2667 |
| 1.5951 | 8.0 | 48 | 1.6613 | 0.2667 |
| 1.5898 | 9.0 | 54 | 1.6611 | 0.2667 |
| 1.5905 | 10.0 | 60 | 1.6609 | 0.2667 |
| 1.5905 | 11.0 | 66 | 1.6606 | 0.2667 |
| 1.5785 | 12.0 | 72 | 1.6604 | 0.2667 |
| 1.5785 | 13.0 | 78 | 1.6602 | 0.2667 |
| 1.623 | 14.0 | 84 | 1.6600 | 0.2667 |
| 1.5698 | 15.0 | 90 | 1.6598 | 0.2667 |
| 1.5698 | 16.0 | 96 | 1.6596 | 0.2667 |
| 1.5831 | 17.0 | 102 | 1.6594 | 0.2667 |
| 1.5831 | 18.0 | 108 | 1.6593 | 0.2667 |
| 1.6234 | 19.0 | 114 | 1.6591 | 0.2667 |
| 1.605 | 20.0 | 120 | 1.6589 | 0.2667 |
| 1.605 | 21.0 | 126 | 1.6588 | 0.2667 |
| 1.6023 | 22.0 | 132 | 1.6586 | 0.2667 |
| 1.6023 | 23.0 | 138 | 1.6585 | 0.2667 |
| 1.5903 | 24.0 | 144 | 1.6584 | 0.2667 |
| 1.5877 | 25.0 | 150 | 1.6583 | 0.2667 |
| 1.5877 | 26.0 | 156 | 1.6582 | 0.2667 |
| 1.5697 | 27.0 | 162 | 1.6581 | 0.2667 |
| 1.5697 | 28.0 | 168 | 1.6580 | 0.2667 |
| 1.6252 | 29.0 | 174 | 1.6579 | 0.2667 |
| 1.6032 | 30.0 | 180 | 1.6578 | 0.2667 |
| 1.6032 | 31.0 | 186 | 1.6577 | 0.2667 |
| 1.6035 | 32.0 | 192 | 1.6577 | 0.2667 |
| 1.6035 | 33.0 | 198 | 1.6576 | 0.2667 |
| 1.5747 | 34.0 | 204 | 1.6575 | 0.2667 |
| 1.5966 | 35.0 | 210 | 1.6575 | 0.2667 |
| 1.5966 | 36.0 | 216 | 1.6575 | 0.2667 |
| 1.5685 | 37.0 | 222 | 1.6574 | 0.2667 |
| 1.5685 | 38.0 | 228 | 1.6574 | 0.2667 |
| 1.5973 | 39.0 | 234 | 1.6574 | 0.2667 |
| 1.5951 | 40.0 | 240 | 1.6574 | 0.2667 |
| 1.5951 | 41.0 | 246 | 1.6574 | 0.2667 |
| 1.5959 | 42.0 | 252 | 1.6574 | 0.2667 |
| 1.5959 | 43.0 | 258 | 1.6574 | 0.2667 |
| 1.6121 | 44.0 | 264 | 1.6574 | 0.2667 |
| 1.5823 | 45.0 | 270 | 1.6574 | 0.2667 |
| 1.5823 | 46.0 | 276 | 1.6574 | 0.2667 |
| 1.5911 | 47.0 | 282 | 1.6574 | 0.2667 |
| 1.5911 | 48.0 | 288 | 1.6574 | 0.2667 |
| 1.6171 | 49.0 | 294 | 1.6574 | 0.2667 |
| 1.5945 | 50.0 | 300 | 1.6574 | 0.2667 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
LoneStriker/prometheus-13b-v1.0-6.0bpw-h6-exl2 | LoneStriker | 2023-11-10T19:45:49Z | 5 | 0 | transformers | [
"transformers",
"pytorch",
"llama",
"text-generation",
"text2text-generation",
"en",
"dataset:kaist-ai/Feedback-Collection",
"arxiv:2310.08491",
"license:apache-2.0",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
]
| text2text-generation | 2023-11-10T19:45:20Z | ---
tags:
- text2text-generation
datasets:
- kaist-ai/Feedback-Collection
license: apache-2.0
language:
- en
pipeline_tag: text2text-generation
library_name: transformers
metrics:
- pearsonr
- spearmanr
- accuracy
---
## Links for Reference
- **Homepage:https://github.com/kaistAI/Prometheus**
- **Repository:https://github.com/kaistAI/Prometheus**
- **Paper:https://arxiv.org/abs/2310.08491**
- **Point of Contact:[email protected]**
# TL;DR
Prometheus is an alternative of GPT-4 evaluation when doing fine-grained evaluation of an underlying LLM & a Reward model for Reinforcement Learning from Human Feedback (RLHF).
Prometheus is a language model using [Llama-2-Chat](https://huggingface.co/meta-llama/Llama-2-13b-chat-hf) as a base model and fine-tuned on 100K feedback within the [Feedback Collection](https://huggingface.co/datasets/kaist-ai/Feedback-Collection).
Since it was fine-tuned on a large amount of feedback, it is specialized at evaluating long-form responses, outperforming GPT-3.5-Turbo, Llama-2-Chat 70B, and on par with GPT-4 on various benchmarks.
Most importantly, this was possible since we appended 2 reference materials (reference answer, and customized score rubric).
Prometheus is a cheap and powerful alternative to GPT-4 evaluation, which one could use to evaluate LLMs with customized criteria (e.g., Child readability, Cultural Sensitivity, Creativity).
Also, it could be used as a reward model for Reinforcement Learning from Human Feedback (RLHF).
# Model Details
## Model Description
- **Model type:** Language model
- **Language(s) (NLP):** English
- **License:** Apache 2.0
- **Related Models:** [All Prometheus Checkpoints](https://huggingface.co/models?search=kaist-ai/Prometheus)
- **Resources for more information:**
- [Research paper](https://arxiv.org/abs/2310.08491)
- [GitHub Repo](https://github.com/kaistAI/Prometheus)
Prometheus is trained with two different sizes (7B and 13B).
You could check the 7B sized LM on [this page](https://huggingface.co/kaist-ai/prometheus-7b-v1.0).
Also, check out our dataset as well on [this page](https://huggingface.co/datasets/kaist-ai/Feedback-Collection).
## Prompt Format
Prometheus requires 4 components in the input: An instruction, a response to evaluate, a score rubric, and a reference answer. You could refer to the prompt format below.
You should fill in the instruction, response, reference answer, criteria description, and score description for score in range of 1 to 5.
```
###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{instruction}
###Response to evaluate:
{response}
###Reference Answer (Score 5):
{reference_answer}
###Score Rubrics:
[{criteria_description}]
Score 1: {score1_description}
Score 2: {score2_description}
Score 3: {score3_description}
Score 4: {score4_description}
Score 5: {score5_description}
###Feedback:
```
After this, you should apply the conversation template of Llama-2-Chat (not applying it might lead to unexpected behaviors).
You can find the conversation class at this [link](https://github.com/lm-sys/FastChat/blob/main/fastchat/conversation.py).
```
conv = get_conv_template("llama-2")
conv.set_system_message("You are a fair evaluator language model.")
conv.append_message(conv.roles[0], dialogs['instruction'])
conv.append_message(conv.roles[1], None)
prompt = conv.get_prompt()
x = tokenizer(prompt,truncation=False)
```
As a result, a feedback and score decision will be generated, divided by a separating phrase ```[RESULT]```
## License
Feedback Collection and Prometheus is subject to OpenAI's Terms of Use for the generated data. If you suspect any violations, please reach out to us.
# Usage
Find below some example scripts on how to use the model in `transformers`:
## Using the Pytorch model
### Running the model on a CPU
<details>
<summary> Click to expand </summary>
```python
from transformers import T5Tokenizer, T5ForConditionalGeneration
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto")
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
### Running the model on a GPU
<details>
<summary> Click to expand </summary>
```python
# pip install accelerate
import torch
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto")
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids, sample=True, temperature=1.0, top_p=0.9, max_new_tokens=256, repetition_penalty=1.03)
print(tokenizer.decode(outputs[0]))
```
</details>
### Running the model on a GPU using different precisions
#### FP16
<details>
<summary> Click to expand </summary>
```python
# pip install accelerate
import torch
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto", torch_dtype=torch.float16)
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
#### INT8
<details>
<summary> Click to expand </summary>
```python
# pip install bitsandbytes accelerate
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto", load_in_8bit=True)
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
# Citation
If you find the following model helpful, please consider citing our paper!
**BibTeX:**
```bibtex
@misc{kim2023prometheus,
title={Prometheus: Inducing Fine-grained Evaluation Capability in Language Models},
author={Seungone Kim and Jamin Shin and Yejin Cho and Joel Jang and Shayne Longpre and Hwaran Lee and Sangdoo Yun and Seongjin Shin and Sungdong Kim and James Thorne and Minjoon Seo},
year={2023},
eprint={2310.08491},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
``` |
hkivancoral/hushem_1x_deit_tiny_rms_lr00001_fold5 | hkivancoral | 2023-11-10T19:42:06Z | 6 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:facebook/deit-tiny-patch16-224",
"base_model:finetune:facebook/deit-tiny-patch16-224",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| image-classification | 2023-11-10T19:39:22Z | ---
license: apache-2.0
base_model: facebook/deit-tiny-patch16-224
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: hushem_1x_deit_tiny_rms_lr00001_fold5
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: test
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.7073170731707317
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# hushem_1x_deit_tiny_rms_lr00001_fold5
This model is a fine-tuned version of [facebook/deit-tiny-patch16-224](https://huggingface.co/facebook/deit-tiny-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 0.8230
- Accuracy: 0.7073
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 1.0 | 6 | 1.1737 | 0.4634 |
| 1.1816 | 2.0 | 12 | 0.8675 | 0.5366 |
| 1.1816 | 3.0 | 18 | 0.8079 | 0.6341 |
| 0.5246 | 4.0 | 24 | 0.8632 | 0.5854 |
| 0.2225 | 5.0 | 30 | 0.7815 | 0.5610 |
| 0.2225 | 6.0 | 36 | 0.6787 | 0.6585 |
| 0.0792 | 7.0 | 42 | 0.7052 | 0.6585 |
| 0.0792 | 8.0 | 48 | 0.7120 | 0.6341 |
| 0.029 | 9.0 | 54 | 0.8373 | 0.6585 |
| 0.0096 | 10.0 | 60 | 0.6713 | 0.7317 |
| 0.0096 | 11.0 | 66 | 0.7185 | 0.7073 |
| 0.0045 | 12.0 | 72 | 0.7237 | 0.6829 |
| 0.0045 | 13.0 | 78 | 0.7062 | 0.6829 |
| 0.0033 | 14.0 | 84 | 0.7203 | 0.7073 |
| 0.0025 | 15.0 | 90 | 0.7207 | 0.7073 |
| 0.0025 | 16.0 | 96 | 0.7400 | 0.7073 |
| 0.002 | 17.0 | 102 | 0.7337 | 0.6829 |
| 0.002 | 18.0 | 108 | 0.7527 | 0.6829 |
| 0.0017 | 19.0 | 114 | 0.7553 | 0.6829 |
| 0.0015 | 20.0 | 120 | 0.7631 | 0.6829 |
| 0.0015 | 21.0 | 126 | 0.7684 | 0.6829 |
| 0.0014 | 22.0 | 132 | 0.7730 | 0.6829 |
| 0.0014 | 23.0 | 138 | 0.7803 | 0.6829 |
| 0.0012 | 24.0 | 144 | 0.7869 | 0.6829 |
| 0.0011 | 25.0 | 150 | 0.7854 | 0.6829 |
| 0.0011 | 26.0 | 156 | 0.7958 | 0.6829 |
| 0.001 | 27.0 | 162 | 0.7899 | 0.6829 |
| 0.001 | 28.0 | 168 | 0.7956 | 0.6829 |
| 0.001 | 29.0 | 174 | 0.8038 | 0.6829 |
| 0.0009 | 30.0 | 180 | 0.8059 | 0.6829 |
| 0.0009 | 31.0 | 186 | 0.8121 | 0.6829 |
| 0.0008 | 32.0 | 192 | 0.8137 | 0.6829 |
| 0.0008 | 33.0 | 198 | 0.8161 | 0.6829 |
| 0.0008 | 34.0 | 204 | 0.8136 | 0.6829 |
| 0.0008 | 35.0 | 210 | 0.8158 | 0.6829 |
| 0.0008 | 36.0 | 216 | 0.8175 | 0.7073 |
| 0.0007 | 37.0 | 222 | 0.8190 | 0.7073 |
| 0.0007 | 38.0 | 228 | 0.8213 | 0.7073 |
| 0.0007 | 39.0 | 234 | 0.8222 | 0.7073 |
| 0.0007 | 40.0 | 240 | 0.8227 | 0.7073 |
| 0.0007 | 41.0 | 246 | 0.8228 | 0.7073 |
| 0.0007 | 42.0 | 252 | 0.8230 | 0.7073 |
| 0.0007 | 43.0 | 258 | 0.8230 | 0.7073 |
| 0.0007 | 44.0 | 264 | 0.8230 | 0.7073 |
| 0.0007 | 45.0 | 270 | 0.8230 | 0.7073 |
| 0.0007 | 46.0 | 276 | 0.8230 | 0.7073 |
| 0.0007 | 47.0 | 282 | 0.8230 | 0.7073 |
| 0.0007 | 48.0 | 288 | 0.8230 | 0.7073 |
| 0.0007 | 49.0 | 294 | 0.8230 | 0.7073 |
| 0.0007 | 50.0 | 300 | 0.8230 | 0.7073 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
DaColdest/q-FrozenLake-v1-4x4-noSlippery | DaColdest | 2023-11-10T19:41:20Z | 0 | 0 | null | [
"FrozenLake-v1-4x4-no_slippery",
"q-learning",
"reinforcement-learning",
"custom-implementation",
"model-index",
"region:us"
]
| reinforcement-learning | 2023-11-10T19:41:18Z | ---
tags:
- FrozenLake-v1-4x4-no_slippery
- q-learning
- reinforcement-learning
- custom-implementation
model-index:
- name: q-FrozenLake-v1-4x4-noSlippery
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: FrozenLake-v1-4x4-no_slippery
type: FrozenLake-v1-4x4-no_slippery
metrics:
- type: mean_reward
value: 1.00 +/- 0.00
name: mean_reward
verified: false
---
# **Q-Learning** Agent playing1 **FrozenLake-v1**
This is a trained model of a **Q-Learning** agent playing **FrozenLake-v1** .
## Usage
```python
model = load_from_hub(repo_id="DaColdest/q-FrozenLake-v1-4x4-noSlippery", filename="q-learning.pkl")
# Don't forget to check if you need to add additional attributes (is_slippery=False etc)
env = gym.make(model["env_id"])
```
|
Chrisliang33/medicalModel_4_other | Chrisliang33 | 2023-11-10T19:40:42Z | 1 | 0 | peft | [
"peft",
"arxiv:1910.09700",
"base_model:meta-llama/Llama-2-7b-hf",
"base_model:adapter:meta-llama/Llama-2-7b-hf",
"region:us"
]
| null | 2023-11-10T19:37:26Z | ---
library_name: peft
base_model: meta-llama/Llama-2-7b-hf
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
- **Developed by:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[More Information Needed]
## Bias, Risks, and Limitations
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
[More Information Needed]
### Recommendations
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Data Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Data Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
**BibTeX:**
[More Information Needed]
**APA:**
[More Information Needed]
## Glossary [optional]
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: float16
### Framework versions
- PEFT 0.6.1
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: float16
### Framework versions
- PEFT 0.6.1
|
hkivancoral/hushem_1x_deit_tiny_rms_lr00001_fold4 | hkivancoral | 2023-11-10T19:39:16Z | 6 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:facebook/deit-tiny-patch16-224",
"base_model:finetune:facebook/deit-tiny-patch16-224",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| image-classification | 2023-11-10T19:36:33Z | ---
license: apache-2.0
base_model: facebook/deit-tiny-patch16-224
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: hushem_1x_deit_tiny_rms_lr00001_fold4
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: test
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.7619047619047619
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# hushem_1x_deit_tiny_rms_lr00001_fold4
This model is a fine-tuned version of [facebook/deit-tiny-patch16-224](https://huggingface.co/facebook/deit-tiny-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 0.6971
- Accuracy: 0.7619
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 1.0 | 6 | 1.1869 | 0.4762 |
| 1.3589 | 2.0 | 12 | 1.4321 | 0.2381 |
| 1.3589 | 3.0 | 18 | 0.8086 | 0.7143 |
| 0.7587 | 4.0 | 24 | 0.7860 | 0.6667 |
| 0.3552 | 5.0 | 30 | 0.6443 | 0.7143 |
| 0.3552 | 6.0 | 36 | 0.6345 | 0.7381 |
| 0.1624 | 7.0 | 42 | 0.6029 | 0.7381 |
| 0.1624 | 8.0 | 48 | 0.6145 | 0.6667 |
| 0.0655 | 9.0 | 54 | 0.6448 | 0.6905 |
| 0.0257 | 10.0 | 60 | 0.6084 | 0.7381 |
| 0.0257 | 11.0 | 66 | 0.5594 | 0.7143 |
| 0.0099 | 12.0 | 72 | 0.6088 | 0.7381 |
| 0.0099 | 13.0 | 78 | 0.6402 | 0.7619 |
| 0.0054 | 14.0 | 84 | 0.6319 | 0.7381 |
| 0.0038 | 15.0 | 90 | 0.6323 | 0.7619 |
| 0.0038 | 16.0 | 96 | 0.6432 | 0.7381 |
| 0.0029 | 17.0 | 102 | 0.6446 | 0.7381 |
| 0.0029 | 18.0 | 108 | 0.6470 | 0.7381 |
| 0.0023 | 19.0 | 114 | 0.6562 | 0.7381 |
| 0.002 | 20.0 | 120 | 0.6656 | 0.7381 |
| 0.002 | 21.0 | 126 | 0.6696 | 0.7381 |
| 0.0017 | 22.0 | 132 | 0.6739 | 0.7381 |
| 0.0017 | 23.0 | 138 | 0.6722 | 0.7619 |
| 0.0015 | 24.0 | 144 | 0.6705 | 0.7619 |
| 0.0014 | 25.0 | 150 | 0.6761 | 0.7619 |
| 0.0014 | 26.0 | 156 | 0.6768 | 0.7619 |
| 0.0012 | 27.0 | 162 | 0.6844 | 0.7619 |
| 0.0012 | 28.0 | 168 | 0.6843 | 0.7619 |
| 0.0012 | 29.0 | 174 | 0.6854 | 0.7619 |
| 0.0011 | 30.0 | 180 | 0.6913 | 0.7619 |
| 0.0011 | 31.0 | 186 | 0.6928 | 0.7619 |
| 0.001 | 32.0 | 192 | 0.6912 | 0.7619 |
| 0.001 | 33.0 | 198 | 0.6912 | 0.7619 |
| 0.001 | 34.0 | 204 | 0.6924 | 0.7619 |
| 0.0009 | 35.0 | 210 | 0.6912 | 0.7619 |
| 0.0009 | 36.0 | 216 | 0.6935 | 0.7619 |
| 0.0009 | 37.0 | 222 | 0.6948 | 0.7619 |
| 0.0009 | 38.0 | 228 | 0.6957 | 0.7619 |
| 0.0009 | 39.0 | 234 | 0.6966 | 0.7619 |
| 0.0009 | 40.0 | 240 | 0.6969 | 0.7619 |
| 0.0009 | 41.0 | 246 | 0.6971 | 0.7619 |
| 0.0009 | 42.0 | 252 | 0.6971 | 0.7619 |
| 0.0009 | 43.0 | 258 | 0.6971 | 0.7619 |
| 0.0008 | 44.0 | 264 | 0.6971 | 0.7619 |
| 0.0009 | 45.0 | 270 | 0.6971 | 0.7619 |
| 0.0009 | 46.0 | 276 | 0.6971 | 0.7619 |
| 0.0008 | 47.0 | 282 | 0.6971 | 0.7619 |
| 0.0008 | 48.0 | 288 | 0.6971 | 0.7619 |
| 0.0009 | 49.0 | 294 | 0.6971 | 0.7619 |
| 0.0009 | 50.0 | 300 | 0.6971 | 0.7619 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
hkivancoral/hushem_1x_deit_tiny_rms_lr00001_fold3 | hkivancoral | 2023-11-10T19:36:25Z | 11 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:facebook/deit-tiny-patch16-224",
"base_model:finetune:facebook/deit-tiny-patch16-224",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| image-classification | 2023-11-10T19:33:36Z | ---
license: apache-2.0
base_model: facebook/deit-tiny-patch16-224
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: hushem_1x_deit_tiny_rms_lr00001_fold3
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: test
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.6976744186046512
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# hushem_1x_deit_tiny_rms_lr00001_fold3
This model is a fine-tuned version of [facebook/deit-tiny-patch16-224](https://huggingface.co/facebook/deit-tiny-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 0.7349
- Accuracy: 0.6977
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 1.0 | 6 | 1.2207 | 0.4419 |
| 1.2147 | 2.0 | 12 | 0.9891 | 0.6047 |
| 1.2147 | 3.0 | 18 | 0.7510 | 0.7209 |
| 0.576 | 4.0 | 24 | 0.7741 | 0.7209 |
| 0.2188 | 5.0 | 30 | 0.7926 | 0.6279 |
| 0.2188 | 6.0 | 36 | 0.8648 | 0.6047 |
| 0.0657 | 7.0 | 42 | 0.9083 | 0.6279 |
| 0.0657 | 8.0 | 48 | 0.6744 | 0.7209 |
| 0.024 | 9.0 | 54 | 0.6865 | 0.6744 |
| 0.0081 | 10.0 | 60 | 0.7121 | 0.7209 |
| 0.0081 | 11.0 | 66 | 0.7038 | 0.6279 |
| 0.0043 | 12.0 | 72 | 0.6990 | 0.6977 |
| 0.0043 | 13.0 | 78 | 0.6958 | 0.6744 |
| 0.003 | 14.0 | 84 | 0.7014 | 0.6744 |
| 0.0024 | 15.0 | 90 | 0.6973 | 0.6744 |
| 0.0024 | 16.0 | 96 | 0.7050 | 0.6744 |
| 0.002 | 17.0 | 102 | 0.7045 | 0.6512 |
| 0.002 | 18.0 | 108 | 0.7008 | 0.6512 |
| 0.0017 | 19.0 | 114 | 0.7130 | 0.6744 |
| 0.0015 | 20.0 | 120 | 0.7143 | 0.6744 |
| 0.0015 | 21.0 | 126 | 0.7112 | 0.6744 |
| 0.0013 | 22.0 | 132 | 0.7160 | 0.6744 |
| 0.0013 | 23.0 | 138 | 0.7131 | 0.6744 |
| 0.0012 | 24.0 | 144 | 0.7144 | 0.6744 |
| 0.0011 | 25.0 | 150 | 0.7160 | 0.6744 |
| 0.0011 | 26.0 | 156 | 0.7202 | 0.6977 |
| 0.001 | 27.0 | 162 | 0.7225 | 0.6977 |
| 0.001 | 28.0 | 168 | 0.7211 | 0.6744 |
| 0.001 | 29.0 | 174 | 0.7237 | 0.6977 |
| 0.0009 | 30.0 | 180 | 0.7265 | 0.6977 |
| 0.0009 | 31.0 | 186 | 0.7272 | 0.6977 |
| 0.0008 | 32.0 | 192 | 0.7283 | 0.6977 |
| 0.0008 | 33.0 | 198 | 0.7304 | 0.6977 |
| 0.0008 | 34.0 | 204 | 0.7314 | 0.6977 |
| 0.0008 | 35.0 | 210 | 0.7309 | 0.6977 |
| 0.0008 | 36.0 | 216 | 0.7324 | 0.6977 |
| 0.0008 | 37.0 | 222 | 0.7325 | 0.6977 |
| 0.0008 | 38.0 | 228 | 0.7335 | 0.6977 |
| 0.0007 | 39.0 | 234 | 0.7342 | 0.6977 |
| 0.0007 | 40.0 | 240 | 0.7346 | 0.6977 |
| 0.0007 | 41.0 | 246 | 0.7348 | 0.6977 |
| 0.0007 | 42.0 | 252 | 0.7349 | 0.6977 |
| 0.0007 | 43.0 | 258 | 0.7349 | 0.6977 |
| 0.0007 | 44.0 | 264 | 0.7349 | 0.6977 |
| 0.0007 | 45.0 | 270 | 0.7349 | 0.6977 |
| 0.0007 | 46.0 | 276 | 0.7349 | 0.6977 |
| 0.0007 | 47.0 | 282 | 0.7349 | 0.6977 |
| 0.0007 | 48.0 | 288 | 0.7349 | 0.6977 |
| 0.0007 | 49.0 | 294 | 0.7349 | 0.6977 |
| 0.0007 | 50.0 | 300 | 0.7349 | 0.6977 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
LoneStriker/prometheus-13b-v1.0-5.0bpw-h6-exl2 | LoneStriker | 2023-11-10T19:35:32Z | 5 | 0 | transformers | [
"transformers",
"pytorch",
"llama",
"text-generation",
"text2text-generation",
"en",
"dataset:kaist-ai/Feedback-Collection",
"arxiv:2310.08491",
"license:apache-2.0",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
]
| text2text-generation | 2023-11-10T19:35:04Z | ---
tags:
- text2text-generation
datasets:
- kaist-ai/Feedback-Collection
license: apache-2.0
language:
- en
pipeline_tag: text2text-generation
library_name: transformers
metrics:
- pearsonr
- spearmanr
- accuracy
---
## Links for Reference
- **Homepage:https://github.com/kaistAI/Prometheus**
- **Repository:https://github.com/kaistAI/Prometheus**
- **Paper:https://arxiv.org/abs/2310.08491**
- **Point of Contact:[email protected]**
# TL;DR
Prometheus is an alternative of GPT-4 evaluation when doing fine-grained evaluation of an underlying LLM & a Reward model for Reinforcement Learning from Human Feedback (RLHF).
Prometheus is a language model using [Llama-2-Chat](https://huggingface.co/meta-llama/Llama-2-13b-chat-hf) as a base model and fine-tuned on 100K feedback within the [Feedback Collection](https://huggingface.co/datasets/kaist-ai/Feedback-Collection).
Since it was fine-tuned on a large amount of feedback, it is specialized at evaluating long-form responses, outperforming GPT-3.5-Turbo, Llama-2-Chat 70B, and on par with GPT-4 on various benchmarks.
Most importantly, this was possible since we appended 2 reference materials (reference answer, and customized score rubric).
Prometheus is a cheap and powerful alternative to GPT-4 evaluation, which one could use to evaluate LLMs with customized criteria (e.g., Child readability, Cultural Sensitivity, Creativity).
Also, it could be used as a reward model for Reinforcement Learning from Human Feedback (RLHF).
# Model Details
## Model Description
- **Model type:** Language model
- **Language(s) (NLP):** English
- **License:** Apache 2.0
- **Related Models:** [All Prometheus Checkpoints](https://huggingface.co/models?search=kaist-ai/Prometheus)
- **Resources for more information:**
- [Research paper](https://arxiv.org/abs/2310.08491)
- [GitHub Repo](https://github.com/kaistAI/Prometheus)
Prometheus is trained with two different sizes (7B and 13B).
You could check the 7B sized LM on [this page](https://huggingface.co/kaist-ai/prometheus-7b-v1.0).
Also, check out our dataset as well on [this page](https://huggingface.co/datasets/kaist-ai/Feedback-Collection).
## Prompt Format
Prometheus requires 4 components in the input: An instruction, a response to evaluate, a score rubric, and a reference answer. You could refer to the prompt format below.
You should fill in the instruction, response, reference answer, criteria description, and score description for score in range of 1 to 5.
```
###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{instruction}
###Response to evaluate:
{response}
###Reference Answer (Score 5):
{reference_answer}
###Score Rubrics:
[{criteria_description}]
Score 1: {score1_description}
Score 2: {score2_description}
Score 3: {score3_description}
Score 4: {score4_description}
Score 5: {score5_description}
###Feedback:
```
After this, you should apply the conversation template of Llama-2-Chat (not applying it might lead to unexpected behaviors).
You can find the conversation class at this [link](https://github.com/lm-sys/FastChat/blob/main/fastchat/conversation.py).
```
conv = get_conv_template("llama-2")
conv.set_system_message("You are a fair evaluator language model.")
conv.append_message(conv.roles[0], dialogs['instruction'])
conv.append_message(conv.roles[1], None)
prompt = conv.get_prompt()
x = tokenizer(prompt,truncation=False)
```
As a result, a feedback and score decision will be generated, divided by a separating phrase ```[RESULT]```
## License
Feedback Collection and Prometheus is subject to OpenAI's Terms of Use for the generated data. If you suspect any violations, please reach out to us.
# Usage
Find below some example scripts on how to use the model in `transformers`:
## Using the Pytorch model
### Running the model on a CPU
<details>
<summary> Click to expand </summary>
```python
from transformers import T5Tokenizer, T5ForConditionalGeneration
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto")
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
### Running the model on a GPU
<details>
<summary> Click to expand </summary>
```python
# pip install accelerate
import torch
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto")
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids, sample=True, temperature=1.0, top_p=0.9, max_new_tokens=256, repetition_penalty=1.03)
print(tokenizer.decode(outputs[0]))
```
</details>
### Running the model on a GPU using different precisions
#### FP16
<details>
<summary> Click to expand </summary>
```python
# pip install accelerate
import torch
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto", torch_dtype=torch.float16)
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
#### INT8
<details>
<summary> Click to expand </summary>
```python
# pip install bitsandbytes accelerate
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto", load_in_8bit=True)
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
# Citation
If you find the following model helpful, please consider citing our paper!
**BibTeX:**
```bibtex
@misc{kim2023prometheus,
title={Prometheus: Inducing Fine-grained Evaluation Capability in Language Models},
author={Seungone Kim and Jamin Shin and Yejin Cho and Joel Jang and Shayne Longpre and Hwaran Lee and Sangdoo Yun and Seongjin Shin and Sungdong Kim and James Thorne and Minjoon Seo},
year={2023},
eprint={2310.08491},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
``` |
Lollitor/FineTuned1024 | Lollitor | 2023-11-10T19:32:34Z | 0 | 0 | peft | [
"peft",
"safetensors",
"arxiv:1910.09700",
"base_model:Lollitor/ColabFinished",
"base_model:adapter:Lollitor/ColabFinished",
"region:us"
]
| null | 2023-11-10T19:32:29Z | ---
library_name: peft
base_model: Lollitor/ColabFinished
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
- **Developed by:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[More Information Needed]
## Bias, Risks, and Limitations
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
[More Information Needed]
### Recommendations
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Data Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Data Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
**BibTeX:**
[More Information Needed]
**APA:**
[More Information Needed]
## Glossary [optional]
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: bfloat16
### Framework versions
- PEFT 0.6.2.dev0
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: bfloat16
### Framework versions
- PEFT 0.6.2.dev0
|
hkivancoral/hushem_1x_deit_tiny_rms_lr00001_fold1 | hkivancoral | 2023-11-10T19:30:33Z | 7 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:facebook/deit-tiny-patch16-224",
"base_model:finetune:facebook/deit-tiny-patch16-224",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| image-classification | 2023-11-10T19:25:56Z | ---
license: apache-2.0
base_model: facebook/deit-tiny-patch16-224
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: hushem_1x_deit_tiny_rms_lr00001_fold1
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: test
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.6888888888888889
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# hushem_1x_deit_tiny_rms_lr00001_fold1
This model is a fine-tuned version of [facebook/deit-tiny-patch16-224](https://huggingface.co/facebook/deit-tiny-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 1.2449
- Accuracy: 0.6889
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 1.0 | 6 | 1.2771 | 0.4 |
| 1.263 | 2.0 | 12 | 1.0784 | 0.5556 |
| 1.263 | 3.0 | 18 | 0.9616 | 0.5556 |
| 0.5461 | 4.0 | 24 | 1.0339 | 0.6889 |
| 0.2446 | 5.0 | 30 | 0.9939 | 0.6667 |
| 0.2446 | 6.0 | 36 | 1.2442 | 0.4889 |
| 0.0817 | 7.0 | 42 | 0.7980 | 0.6222 |
| 0.0817 | 8.0 | 48 | 0.8675 | 0.6444 |
| 0.0302 | 9.0 | 54 | 0.8969 | 0.6889 |
| 0.009 | 10.0 | 60 | 0.9399 | 0.6222 |
| 0.009 | 11.0 | 66 | 1.0591 | 0.7111 |
| 0.0037 | 12.0 | 72 | 1.0283 | 0.6667 |
| 0.0037 | 13.0 | 78 | 1.0855 | 0.6667 |
| 0.0025 | 14.0 | 84 | 1.1121 | 0.6667 |
| 0.0019 | 15.0 | 90 | 1.1082 | 0.6667 |
| 0.0019 | 16.0 | 96 | 1.1158 | 0.6667 |
| 0.0015 | 17.0 | 102 | 1.1382 | 0.6667 |
| 0.0015 | 18.0 | 108 | 1.1574 | 0.6667 |
| 0.0013 | 19.0 | 114 | 1.1739 | 0.6667 |
| 0.0011 | 20.0 | 120 | 1.1736 | 0.6667 |
| 0.0011 | 21.0 | 126 | 1.1594 | 0.6889 |
| 0.001 | 22.0 | 132 | 1.1738 | 0.6889 |
| 0.001 | 23.0 | 138 | 1.1962 | 0.6667 |
| 0.0009 | 24.0 | 144 | 1.1951 | 0.6889 |
| 0.0008 | 25.0 | 150 | 1.2004 | 0.6889 |
| 0.0008 | 26.0 | 156 | 1.1996 | 0.6889 |
| 0.0008 | 27.0 | 162 | 1.2076 | 0.6889 |
| 0.0008 | 28.0 | 168 | 1.2144 | 0.6889 |
| 0.0007 | 29.0 | 174 | 1.2117 | 0.6889 |
| 0.0007 | 30.0 | 180 | 1.2204 | 0.6889 |
| 0.0007 | 31.0 | 186 | 1.2217 | 0.6889 |
| 0.0006 | 32.0 | 192 | 1.2270 | 0.6889 |
| 0.0006 | 33.0 | 198 | 1.2321 | 0.6889 |
| 0.0006 | 34.0 | 204 | 1.2307 | 0.6889 |
| 0.0006 | 35.0 | 210 | 1.2313 | 0.6889 |
| 0.0006 | 36.0 | 216 | 1.2374 | 0.6889 |
| 0.0006 | 37.0 | 222 | 1.2446 | 0.6889 |
| 0.0006 | 38.0 | 228 | 1.2471 | 0.6889 |
| 0.0005 | 39.0 | 234 | 1.2452 | 0.6889 |
| 0.0006 | 40.0 | 240 | 1.2458 | 0.6889 |
| 0.0006 | 41.0 | 246 | 1.2454 | 0.6889 |
| 0.0005 | 42.0 | 252 | 1.2449 | 0.6889 |
| 0.0005 | 43.0 | 258 | 1.2449 | 0.6889 |
| 0.0005 | 44.0 | 264 | 1.2449 | 0.6889 |
| 0.0005 | 45.0 | 270 | 1.2449 | 0.6889 |
| 0.0005 | 46.0 | 276 | 1.2449 | 0.6889 |
| 0.0005 | 47.0 | 282 | 1.2449 | 0.6889 |
| 0.0005 | 48.0 | 288 | 1.2449 | 0.6889 |
| 0.0005 | 49.0 | 294 | 1.2449 | 0.6889 |
| 0.0005 | 50.0 | 300 | 1.2449 | 0.6889 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
mireillfares/BERTIS | mireillfares | 2023-11-10T19:28:05Z | 15 | 0 | transformers | [
"transformers",
"pytorch",
"bert",
"text-classification",
"en",
"arxiv:2311.05481",
"license:afl-3.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| text-classification | 2023-06-28T08:52:54Z | ---
language:
- en
metrics:
- precision
- accuracy
- f1
- recall
widget:
- text: "On a shelf, there are five books."
example_title: "Example 1"
- text: "The two men running to become New York City's next mayor."
example_title: "Example 2"
- text: "Time flies."
example_title: "Example 3"
- text: "Hold on to your beliefs and never let them go."
example_title: "Example 4"
- text: "Our relationship is rock solid."
example_title: "Example 5"
- text: "He’s full of laughs."
example_title: "Example 6"
license: afl-3.0
---
# Model Card for BERTIS Model
This README provides an overview of BERTIS, its purpose and usage. BERTIS is presented in the paper ["*META4: Semantically-Aligned Generation of metaphoric gestures using self-supervised text and speech representations*"](https://arxiv.org/abs/2311.05481).
If you found this work helpful, please consider citing our paper.
Bibtex:
```ruby
@article{fares2023META4,
title={META4: Semantically-Aligned Generation of metaphoric gestures using self-supervised text and speech representations},
author={Fares, Mireille and Pelachaud, Catherine and Obin, Nicolas},
journal={arXiv preprint arXiv:2311.05481},
year={2023}
}
```
## BERTIS Model Description
BERTIS (BERT-based Image Schema) is a computational model designed to classify input texts into specific Image Schema classes. Image Schemas are cognitive patterns that play a fundamental role in shaping the way humans conceptualize and reason about various concepts present in language. BERTIS is built upon the BERT architecture. The model is fine-tuned using a specialized corpus created by Wachowiak et al.(2022).
## BERTIS Image Schema Classes
The main purpose of BERTIS is to automatically categorize input texts into predefined Image Schema classes. BERTIS considers 14 distinct Image Schema classes, each capturing a specific cognitive pattern. These classes and their corresponding examples (taken from Wachowiak et al. (2022)) are listed below:
- CENTER-PERIPHERY: She brushed the thought away.
- CONTACT: That blew me away.
- CONTAINMENT: Keep it in the back of your mind.
- COVERING: His judgement is clouded.
- FORCE: They are attracted to each other.
- LINK: Breaking social ties.
- OBJECT: Seize the opportunity.
- PART-WHOLE: They assembled a theory.
- SCALE: This class is bigger than that one.
- SOURCE_PATH_GOAL: The time for action has arrived.
- SPLITTING: What separates the men from the boys?
- SUBSTANCE: Emotions are tinged with suffuse.
- VERTICALITY: No known spoken language uses the lateral axis for time.
- SUPPORT: The poor in our country need a boost up.
<!-- ### Model Sources [optional] --
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]>
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
<!-- ### Direct Use This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. [More Information Needed] -->
<!-- ### Downstream Use [optional] This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app [More Information Needed] -->
<!-- ### Out-of-Scope Use This section addresses misuse, malicious use, and uses that the model will not work well for. [More Information Needed] -->
<!-- ## Bias, Risks, and Limitations This section is meant to convey both technical and sociotechnical limitations. [More Information Needed] -->
<!-- ### Recommendations This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
-->
<!-- ## How to Get Started with the Model
Use the code below to get started with the model.-->
# Training and Testing Details
The data used for training, validating and testing BERTIS is found in: https://github.com/lwachowiak/Systematic-Analysis-of-Image-Schemas-through-Explainable-Multilingual-Language-Models/blob/main/Data/Image%20Schemas%20English%20and%20German.csv;
80% of the data were used for training BERTIS, 10% for validation, and 10% for testing it.
<!-- ### Training Procedure This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
<!-- #### Preprocessing [optional]
[More Information Needed]-->
<!-- #### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision
#### Speeds, Sizes, Times [optional]
-->
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc.
[More Information Needed]-->
# Evaluation
## Metrics
- Precision: measures the proportion of correctly predicted positive instances out of all instances predicted as positive. It focuses on the accuracy of positive predictions.
- Recall: measures the proportion of correctly predicted positive instances out of all actual positive instances. It focuses on the ability to capture positive instances.
- F1-score: the harmonic mean of precision and recall, providing a balanced measure that combines both metrics
## Results
| Class | Precision | Recall | F1-score |
| ---------------- | --------- | ------ | -------- |
|"CENTER-PERIPHERY"| 0.98 | 0.94 | 0.96 |
|"CONTACT" | 1 | 1 | 1 |
|"CONTAINMENT" | 0.74 | 0.67 | 0.7 |
|"COVERING" | 1 | 1 | 1 |
|"FORCE" | 0.8 | 0.91 | 0.85 |
|"LINK" | 1 | 1 | 1 |
|"OBJECT" | 0.81 | 0.83 | 0.82 |
|"PART-WHOLE" | 1 | 1 | 1 |
|"SCALE" | 1 | 1 | 1 |
|"SOURCE_PATH_GOAL"| 0.81 | 0.74 | 0.77 |
|"SPLITTING" | 1 | 1 | 1 |
|"SUBSTANCE" | 1 | 1 | 1 |
|"SUPPORT" | 1 | 1 | 1 |
|"VERTICALITY" | 0.88 | 0.93 | 0.90 |
**Overall Accuracy** : **0.93**
For all Image Schema classes, F1-score is between 0.77 and 1, indicating that BERTIS performs well in terms of both accurately predicting the correct Image Schema classes (precision) and capturing many correct Image Schema classes as possible (recall). Indeed these results are also reflected in the precision and recall scores for all Image Schema classes. We observe a high precision (between 0.74 and 1) for all classes, which indicates the number of correct predicted
Image Schema classes. The recall scores are above 0.74 for most of the classes, with one Image Schema class having a recall score equal to 0.67. The overall accuracy of BERTIS with respect to all Image Schema classes is equal to 0.93, indicating a high accuracy and good performance in classifying the input text into the correct Image Schema classes.
<!--
## Technical Specifications [optional]
### Model Architecture and Objective
### Compute Infrastructure
#### Hardware
#### Software
## Citation [optional]
**BibTeX:**
**APA:**
## Glossary [optional]
## More Information [optional]
## Model Card Authors [optional]
## Model Card Contact
Mireille Fares: [email protected]>
**Developed by:** Mireille Fares
**Language(s) (NLP):** English, can generalize to other languages such as German
**License:** Academic Free License
**Finetuned from model:** BERT Base Cased model |
hkivancoral/hushem_1x_deit_tiny_rms_lr0001_fold5 | hkivancoral | 2023-11-10T19:16:43Z | 6 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:facebook/deit-tiny-patch16-224",
"base_model:finetune:facebook/deit-tiny-patch16-224",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| image-classification | 2023-11-10T19:14:41Z | ---
license: apache-2.0
base_model: facebook/deit-tiny-patch16-224
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: hushem_1x_deit_tiny_rms_lr0001_fold5
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: test
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.5609756097560976
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# hushem_1x_deit_tiny_rms_lr0001_fold5
This model is a fine-tuned version of [facebook/deit-tiny-patch16-224](https://huggingface.co/facebook/deit-tiny-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 2.1537
- Accuracy: 0.5610
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0001
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 1.0 | 6 | 1.5741 | 0.2683 |
| 1.8922 | 2.0 | 12 | 1.3978 | 0.2683 |
| 1.8922 | 3.0 | 18 | 1.4032 | 0.2439 |
| 1.5101 | 4.0 | 24 | 1.4021 | 0.2683 |
| 1.38 | 5.0 | 30 | 2.1528 | 0.2439 |
| 1.38 | 6.0 | 36 | 1.4141 | 0.2439 |
| 1.4096 | 7.0 | 42 | 1.2484 | 0.4390 |
| 1.4096 | 8.0 | 48 | 1.2607 | 0.4390 |
| 1.2381 | 9.0 | 54 | 0.9950 | 0.5366 |
| 1.1539 | 10.0 | 60 | 1.0350 | 0.5610 |
| 1.1539 | 11.0 | 66 | 1.2716 | 0.3415 |
| 0.9039 | 12.0 | 72 | 1.0596 | 0.5854 |
| 0.9039 | 13.0 | 78 | 1.5972 | 0.4146 |
| 0.6191 | 14.0 | 84 | 1.9855 | 0.4390 |
| 0.4358 | 15.0 | 90 | 1.2403 | 0.4878 |
| 0.4358 | 16.0 | 96 | 2.3374 | 0.4390 |
| 0.2291 | 17.0 | 102 | 1.5475 | 0.4390 |
| 0.2291 | 18.0 | 108 | 1.2789 | 0.6341 |
| 0.1203 | 19.0 | 114 | 1.8441 | 0.4390 |
| 0.0604 | 20.0 | 120 | 1.7948 | 0.4878 |
| 0.0604 | 21.0 | 126 | 2.0211 | 0.4634 |
| 0.0322 | 22.0 | 132 | 1.8178 | 0.5366 |
| 0.0322 | 23.0 | 138 | 2.0950 | 0.4878 |
| 0.017 | 24.0 | 144 | 2.0410 | 0.5122 |
| 0.0011 | 25.0 | 150 | 2.0405 | 0.5122 |
| 0.0011 | 26.0 | 156 | 2.0495 | 0.5122 |
| 0.0007 | 27.0 | 162 | 2.0594 | 0.5122 |
| 0.0007 | 28.0 | 168 | 2.0747 | 0.5122 |
| 0.0006 | 29.0 | 174 | 2.0825 | 0.5610 |
| 0.0005 | 30.0 | 180 | 2.0915 | 0.5610 |
| 0.0005 | 31.0 | 186 | 2.1017 | 0.5610 |
| 0.0004 | 32.0 | 192 | 2.1110 | 0.5610 |
| 0.0004 | 33.0 | 198 | 2.1199 | 0.5610 |
| 0.0004 | 34.0 | 204 | 2.1276 | 0.5610 |
| 0.0004 | 35.0 | 210 | 2.1335 | 0.5610 |
| 0.0004 | 36.0 | 216 | 2.1398 | 0.5610 |
| 0.0004 | 37.0 | 222 | 2.1439 | 0.5610 |
| 0.0004 | 38.0 | 228 | 2.1473 | 0.5610 |
| 0.0003 | 39.0 | 234 | 2.1497 | 0.5610 |
| 0.0003 | 40.0 | 240 | 2.1519 | 0.5610 |
| 0.0003 | 41.0 | 246 | 2.1532 | 0.5610 |
| 0.0003 | 42.0 | 252 | 2.1537 | 0.5610 |
| 0.0003 | 43.0 | 258 | 2.1537 | 0.5610 |
| 0.0003 | 44.0 | 264 | 2.1537 | 0.5610 |
| 0.0003 | 45.0 | 270 | 2.1537 | 0.5610 |
| 0.0003 | 46.0 | 276 | 2.1537 | 0.5610 |
| 0.0003 | 47.0 | 282 | 2.1537 | 0.5610 |
| 0.0003 | 48.0 | 288 | 2.1537 | 0.5610 |
| 0.0003 | 49.0 | 294 | 2.1537 | 0.5610 |
| 0.0003 | 50.0 | 300 | 2.1537 | 0.5610 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
LoneStriker/prometheus-13b-v1.0-3.0bpw-h6-exl2 | LoneStriker | 2023-11-10T19:15:07Z | 5 | 0 | transformers | [
"transformers",
"pytorch",
"llama",
"text-generation",
"text2text-generation",
"en",
"dataset:kaist-ai/Feedback-Collection",
"arxiv:2310.08491",
"license:apache-2.0",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
]
| text2text-generation | 2023-11-10T19:14:51Z | ---
tags:
- text2text-generation
datasets:
- kaist-ai/Feedback-Collection
license: apache-2.0
language:
- en
pipeline_tag: text2text-generation
library_name: transformers
metrics:
- pearsonr
- spearmanr
- accuracy
---
## Links for Reference
- **Homepage:https://github.com/kaistAI/Prometheus**
- **Repository:https://github.com/kaistAI/Prometheus**
- **Paper:https://arxiv.org/abs/2310.08491**
- **Point of Contact:[email protected]**
# TL;DR
Prometheus is an alternative of GPT-4 evaluation when doing fine-grained evaluation of an underlying LLM & a Reward model for Reinforcement Learning from Human Feedback (RLHF).
Prometheus is a language model using [Llama-2-Chat](https://huggingface.co/meta-llama/Llama-2-13b-chat-hf) as a base model and fine-tuned on 100K feedback within the [Feedback Collection](https://huggingface.co/datasets/kaist-ai/Feedback-Collection).
Since it was fine-tuned on a large amount of feedback, it is specialized at evaluating long-form responses, outperforming GPT-3.5-Turbo, Llama-2-Chat 70B, and on par with GPT-4 on various benchmarks.
Most importantly, this was possible since we appended 2 reference materials (reference answer, and customized score rubric).
Prometheus is a cheap and powerful alternative to GPT-4 evaluation, which one could use to evaluate LLMs with customized criteria (e.g., Child readability, Cultural Sensitivity, Creativity).
Also, it could be used as a reward model for Reinforcement Learning from Human Feedback (RLHF).
# Model Details
## Model Description
- **Model type:** Language model
- **Language(s) (NLP):** English
- **License:** Apache 2.0
- **Related Models:** [All Prometheus Checkpoints](https://huggingface.co/models?search=kaist-ai/Prometheus)
- **Resources for more information:**
- [Research paper](https://arxiv.org/abs/2310.08491)
- [GitHub Repo](https://github.com/kaistAI/Prometheus)
Prometheus is trained with two different sizes (7B and 13B).
You could check the 7B sized LM on [this page](https://huggingface.co/kaist-ai/prometheus-7b-v1.0).
Also, check out our dataset as well on [this page](https://huggingface.co/datasets/kaist-ai/Feedback-Collection).
## Prompt Format
Prometheus requires 4 components in the input: An instruction, a response to evaluate, a score rubric, and a reference answer. You could refer to the prompt format below.
You should fill in the instruction, response, reference answer, criteria description, and score description for score in range of 1 to 5.
```
###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{instruction}
###Response to evaluate:
{response}
###Reference Answer (Score 5):
{reference_answer}
###Score Rubrics:
[{criteria_description}]
Score 1: {score1_description}
Score 2: {score2_description}
Score 3: {score3_description}
Score 4: {score4_description}
Score 5: {score5_description}
###Feedback:
```
After this, you should apply the conversation template of Llama-2-Chat (not applying it might lead to unexpected behaviors).
You can find the conversation class at this [link](https://github.com/lm-sys/FastChat/blob/main/fastchat/conversation.py).
```
conv = get_conv_template("llama-2")
conv.set_system_message("You are a fair evaluator language model.")
conv.append_message(conv.roles[0], dialogs['instruction'])
conv.append_message(conv.roles[1], None)
prompt = conv.get_prompt()
x = tokenizer(prompt,truncation=False)
```
As a result, a feedback and score decision will be generated, divided by a separating phrase ```[RESULT]```
## License
Feedback Collection and Prometheus is subject to OpenAI's Terms of Use for the generated data. If you suspect any violations, please reach out to us.
# Usage
Find below some example scripts on how to use the model in `transformers`:
## Using the Pytorch model
### Running the model on a CPU
<details>
<summary> Click to expand </summary>
```python
from transformers import T5Tokenizer, T5ForConditionalGeneration
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto")
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
### Running the model on a GPU
<details>
<summary> Click to expand </summary>
```python
# pip install accelerate
import torch
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto")
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids, sample=True, temperature=1.0, top_p=0.9, max_new_tokens=256, repetition_penalty=1.03)
print(tokenizer.decode(outputs[0]))
```
</details>
### Running the model on a GPU using different precisions
#### FP16
<details>
<summary> Click to expand </summary>
```python
# pip install accelerate
import torch
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto", torch_dtype=torch.float16)
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
#### INT8
<details>
<summary> Click to expand </summary>
```python
# pip install bitsandbytes accelerate
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto", load_in_8bit=True)
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
# Citation
If you find the following model helpful, please consider citing our paper!
**BibTeX:**
```bibtex
@misc{kim2023prometheus,
title={Prometheus: Inducing Fine-grained Evaluation Capability in Language Models},
author={Seungone Kim and Jamin Shin and Yejin Cho and Joel Jang and Shayne Longpre and Hwaran Lee and Sangdoo Yun and Seongjin Shin and Sungdong Kim and James Thorne and Minjoon Seo},
year={2023},
eprint={2310.08491},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
``` |
hkivancoral/hushem_1x_deit_tiny_rms_lr0001_fold3 | hkivancoral | 2023-11-10T19:12:26Z | 9 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:facebook/deit-tiny-patch16-224",
"base_model:finetune:facebook/deit-tiny-patch16-224",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| image-classification | 2023-11-10T19:10:23Z | ---
license: apache-2.0
base_model: facebook/deit-tiny-patch16-224
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: hushem_1x_deit_rms_lr0001_fold3
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: test
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.46511627906976744
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# hushem_1x_deit_rms_lr0001_fold3
This model is a fine-tuned version of [facebook/deit-tiny-patch16-224](https://huggingface.co/facebook/deit-tiny-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 2.7920
- Accuracy: 0.4651
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0001
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 1.0 | 6 | 1.3959 | 0.2558 |
| 2.0191 | 2.0 | 12 | 1.4540 | 0.2791 |
| 2.0191 | 3.0 | 18 | 1.5040 | 0.3721 |
| 1.4688 | 4.0 | 24 | 1.3687 | 0.3256 |
| 1.3397 | 5.0 | 30 | 1.3082 | 0.4186 |
| 1.3397 | 6.0 | 36 | 1.3917 | 0.3256 |
| 1.1986 | 7.0 | 42 | 1.4209 | 0.3256 |
| 1.1986 | 8.0 | 48 | 1.4510 | 0.3721 |
| 1.0698 | 9.0 | 54 | 1.4225 | 0.3023 |
| 0.8214 | 10.0 | 60 | 1.5289 | 0.4186 |
| 0.8214 | 11.0 | 66 | 1.4884 | 0.4419 |
| 0.5823 | 12.0 | 72 | 2.0101 | 0.3256 |
| 0.5823 | 13.0 | 78 | 1.6036 | 0.5349 |
| 0.4001 | 14.0 | 84 | 1.6332 | 0.4186 |
| 0.2362 | 15.0 | 90 | 2.0095 | 0.4884 |
| 0.2362 | 16.0 | 96 | 1.8563 | 0.5581 |
| 0.1078 | 17.0 | 102 | 2.1555 | 0.5116 |
| 0.1078 | 18.0 | 108 | 2.0019 | 0.5581 |
| 0.0769 | 19.0 | 114 | 2.3852 | 0.4884 |
| 0.0351 | 20.0 | 120 | 2.4880 | 0.5349 |
| 0.0351 | 21.0 | 126 | 2.5950 | 0.4884 |
| 0.001 | 22.0 | 132 | 2.5992 | 0.4884 |
| 0.001 | 23.0 | 138 | 2.6117 | 0.4884 |
| 0.0006 | 24.0 | 144 | 2.6223 | 0.4884 |
| 0.0005 | 25.0 | 150 | 2.6443 | 0.4884 |
| 0.0005 | 26.0 | 156 | 2.6672 | 0.4884 |
| 0.0004 | 27.0 | 162 | 2.6883 | 0.4884 |
| 0.0004 | 28.0 | 168 | 2.6994 | 0.4884 |
| 0.0003 | 29.0 | 174 | 2.7093 | 0.4884 |
| 0.0003 | 30.0 | 180 | 2.7225 | 0.4884 |
| 0.0003 | 31.0 | 186 | 2.7350 | 0.4884 |
| 0.0003 | 32.0 | 192 | 2.7468 | 0.4651 |
| 0.0003 | 33.0 | 198 | 2.7564 | 0.4651 |
| 0.0003 | 34.0 | 204 | 2.7644 | 0.4651 |
| 0.0002 | 35.0 | 210 | 2.7717 | 0.4651 |
| 0.0002 | 36.0 | 216 | 2.7756 | 0.4651 |
| 0.0002 | 37.0 | 222 | 2.7805 | 0.4651 |
| 0.0002 | 38.0 | 228 | 2.7848 | 0.4651 |
| 0.0002 | 39.0 | 234 | 2.7876 | 0.4651 |
| 0.0002 | 40.0 | 240 | 2.7903 | 0.4651 |
| 0.0002 | 41.0 | 246 | 2.7917 | 0.4651 |
| 0.0002 | 42.0 | 252 | 2.7920 | 0.4651 |
| 0.0002 | 43.0 | 258 | 2.7920 | 0.4651 |
| 0.0002 | 44.0 | 264 | 2.7920 | 0.4651 |
| 0.0002 | 45.0 | 270 | 2.7920 | 0.4651 |
| 0.0002 | 46.0 | 276 | 2.7920 | 0.4651 |
| 0.0002 | 47.0 | 282 | 2.7920 | 0.4651 |
| 0.0002 | 48.0 | 288 | 2.7920 | 0.4651 |
| 0.0002 | 49.0 | 294 | 2.7920 | 0.4651 |
| 0.0002 | 50.0 | 300 | 2.7920 | 0.4651 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
TheBloke/prometheus-7B-v1.0-AWQ | TheBloke | 2023-11-10T19:10:41Z | 9 | 0 | transformers | [
"transformers",
"safetensors",
"llama",
"text-generation",
"text2text-generation",
"en",
"dataset:kaist-ai/Feedback-Collection",
"arxiv:2310.08491",
"base_model:prometheus-eval/prometheus-7b-v1.0",
"base_model:quantized:prometheus-eval/prometheus-7b-v1.0",
"license:apache-2.0",
"autotrain_compatible",
"text-generation-inference",
"4-bit",
"awq",
"region:us"
]
| text2text-generation | 2023-11-10T18:56:09Z | ---
base_model: kaist-ai/prometheus-7b-v1.0
datasets:
- kaist-ai/Feedback-Collection
inference: false
language:
- en
library_name: transformers
license: apache-2.0
metrics:
- pearsonr
- spearmanr
- accuracy
model_creator: KAIST AI
model_name: Prometheus 7B V1.0
model_type: llama
pipeline_tag: text2text-generation
prompt_template: '###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference
answer that gets a score of 5, and a score rubric representing a evaluation criteria
are given.
1. Write a detailed feedback that assess the quality of the response strictly based
on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You
should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria)
[RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{prompt}
###Response to evaluate:
{{response}}
###Reference Answer (Score 5):
{{reference_answer}}
###Score Rubrics:
[{{criteria_description}}]
Score 1: {{score1_description}}
Score 2: {{score2_description}}
Score 3: {{score3_description}}
Score 4: {{score4_description}}
Score 5: {{score5_description}}
###Feedback:
'
quantized_by: TheBloke
tags:
- text2text-generation
---
<!-- markdownlint-disable MD041 -->
<!-- header start -->
<!-- 200823 -->
<div style="width: auto; margin-left: auto; margin-right: auto">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<div style="text-align:center; margin-top: 0em; margin-bottom: 0em"><p style="margin-top: 0.25em; margin-bottom: 0em;">TheBloke's LLM work is generously supported by a grant from <a href="https://a16z.com">andreessen horowitz (a16z)</a></p></div>
<hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
<!-- header end -->
# Prometheus 7B V1.0 - AWQ
- Model creator: [KAIST AI](https://huggingface.co/kaist-ai)
- Original model: [Prometheus 7B V1.0](https://huggingface.co/kaist-ai/prometheus-7b-v1.0)
<!-- description start -->
## Description
This repo contains AWQ model files for [KAIST AI's Prometheus 7B V1.0](https://huggingface.co/kaist-ai/prometheus-7b-v1.0).
These files were quantised using hardware kindly provided by [Massed Compute](https://massedcompute.com/).
### About AWQ
AWQ is an efficient, accurate and blazing-fast low-bit weight quantization method, currently supporting 4-bit quantization. Compared to GPTQ, it offers faster Transformers-based inference with equivalent or better quality compared to the most commonly used GPTQ settings.
It is supported by:
- [Text Generation Webui](https://github.com/oobabooga/text-generation-webui) - using Loader: AutoAWQ
- [vLLM](https://github.com/vllm-project/vllm) - Llama and Mistral models only
- [Hugging Face Text Generation Inference (TGI)](https://github.com/huggingface/text-generation-inference)
- [Transformers](https://huggingface.co/docs/transformers) version 4.35.0 and later, from any code or client that supports Transformers
- [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) - for use from Python code
<!-- description end -->
<!-- repositories-available start -->
## Repositories available
* [AWQ model(s) for GPU inference.](https://huggingface.co/TheBloke/prometheus-7B-v1.0-AWQ)
* [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/prometheus-7B-v1.0-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/prometheus-7B-v1.0-GGUF)
* [KAIST AI's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/kaist-ai/prometheus-7b-v1.0)
<!-- repositories-available end -->
<!-- prompt-template start -->
## Prompt template: Prometheus
```
###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{prompt}
###Response to evaluate:
{{response}}
###Reference Answer (Score 5):
{{reference_answer}}
###Score Rubrics:
[{{criteria_description}}]
Score 1: {{score1_description}}
Score 2: {{score2_description}}
Score 3: {{score3_description}}
Score 4: {{score4_description}}
Score 5: {{score5_description}}
###Feedback:
```
<!-- prompt-template end -->
<!-- licensing start -->
## Licensing
The creator of the source model has listed its license as `apache-2.0`, and this quantization has therefore used that same license.
As this model is based on Llama 2, it is also subject to the Meta Llama 2 license terms, and the license files for that are additionally included. It should therefore be considered as being claimed to be licensed under both licenses. I contacted Hugging Face for clarification on dual licensing but they do not yet have an official position. Should this change, or should Meta provide any feedback on this situation, I will update this section accordingly.
In the meantime, any questions regarding licensing, and in particular how these two licenses might interact, should be directed to the original model repository: [KAIST AI's Prometheus 7B V1.0](https://huggingface.co/kaist-ai/prometheus-7b-v1.0).
<!-- licensing end -->
<!-- README_AWQ.md-provided-files start -->
## Provided files, and AWQ parameters
I currently release 128g GEMM models only. The addition of group_size 32 models, and GEMV kernel models, is being actively considered.
Models are released as sharded safetensors files.
| Branch | Bits | GS | AWQ Dataset | Seq Len | Size |
| ------ | ---- | -- | ----------- | ------- | ---- |
| [main](https://huggingface.co/TheBloke/prometheus-7B-v1.0-AWQ/tree/main) | 4 | 128 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 4096 | 3.89 GB
<!-- README_AWQ.md-provided-files end -->
<!-- README_AWQ.md-text-generation-webui start -->
## How to easily download and use this model in [text-generation-webui](https://github.com/oobabooga/text-generation-webui)
Please make sure you're using the latest version of [text-generation-webui](https://github.com/oobabooga/text-generation-webui).
It is strongly recommended to use the text-generation-webui one-click-installers unless you're sure you know how to make a manual install.
1. Click the **Model tab**.
2. Under **Download custom model or LoRA**, enter `TheBloke/prometheus-7B-v1.0-AWQ`.
3. Click **Download**.
4. The model will start downloading. Once it's finished it will say "Done".
5. In the top left, click the refresh icon next to **Model**.
6. In the **Model** dropdown, choose the model you just downloaded: `prometheus-7B-v1.0-AWQ`
7. Select **Loader: AutoAWQ**.
8. Click Load, and the model will load and is now ready for use.
9. If you want any custom settings, set them and then click **Save settings for this model** followed by **Reload the Model** in the top right.
10. Once you're ready, click the **Text Generation** tab and enter a prompt to get started!
<!-- README_AWQ.md-text-generation-webui end -->
<!-- README_AWQ.md-use-from-vllm start -->
## Multi-user inference server: vLLM
Documentation on installing and using vLLM [can be found here](https://vllm.readthedocs.io/en/latest/).
- Please ensure you are using vLLM version 0.2 or later.
- When using vLLM as a server, pass the `--quantization awq` parameter.
For example:
```shell
python3 -m vllm.entrypoints.api_server --model TheBloke/prometheus-7B-v1.0-AWQ --quantization awq --dtype auto
```
- When using vLLM from Python code, again set `quantization=awq`.
For example:
```python
from vllm import LLM, SamplingParams
prompts = [
"Tell me about AI",
"Write a story about llamas",
"What is 291 - 150?",
"How much wood would a woodchuck chuck if a woodchuck could chuck wood?",
]
prompt_template=f'''###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{prompt}
###Response to evaluate:
{{response}}
###Reference Answer (Score 5):
{{reference_answer}}
###Score Rubrics:
[{{criteria_description}}]
Score 1: {{score1_description}}
Score 2: {{score2_description}}
Score 3: {{score3_description}}
Score 4: {{score4_description}}
Score 5: {{score5_description}}
###Feedback:
'''
prompts = [prompt_template.format(prompt=prompt) for prompt in prompts]
sampling_params = SamplingParams(temperature=0.8, top_p=0.95)
llm = LLM(model="TheBloke/prometheus-7B-v1.0-AWQ", quantization="awq", dtype="auto")
outputs = llm.generate(prompts, sampling_params)
# Print the outputs.
for output in outputs:
prompt = output.prompt
generated_text = output.outputs[0].text
print(f"Prompt: {prompt!r}, Generated text: {generated_text!r}")
```
<!-- README_AWQ.md-use-from-vllm start -->
<!-- README_AWQ.md-use-from-tgi start -->
## Multi-user inference server: Hugging Face Text Generation Inference (TGI)
Use TGI version 1.1.0 or later. The official Docker container is: `ghcr.io/huggingface/text-generation-inference:1.1.0`
Example Docker parameters:
```shell
--model-id TheBloke/prometheus-7B-v1.0-AWQ --port 3000 --quantize awq --max-input-length 3696 --max-total-tokens 4096 --max-batch-prefill-tokens 4096
```
Example Python code for interfacing with TGI (requires [huggingface-hub](https://github.com/huggingface/huggingface_hub) 0.17.0 or later):
```shell
pip3 install huggingface-hub
```
```python
from huggingface_hub import InferenceClient
endpoint_url = "https://your-endpoint-url-here"
prompt = "Tell me about AI"
prompt_template=f'''###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{prompt}
###Response to evaluate:
{{response}}
###Reference Answer (Score 5):
{{reference_answer}}
###Score Rubrics:
[{{criteria_description}}]
Score 1: {{score1_description}}
Score 2: {{score2_description}}
Score 3: {{score3_description}}
Score 4: {{score4_description}}
Score 5: {{score5_description}}
###Feedback:
'''
client = InferenceClient(endpoint_url)
response = client.text_generation(prompt,
max_new_tokens=128,
do_sample=True,
temperature=0.7,
top_p=0.95,
top_k=40,
repetition_penalty=1.1)
print(f"Model output: ", response)
```
<!-- README_AWQ.md-use-from-tgi end -->
<!-- README_AWQ.md-use-from-python start -->
## Inference from Python code using Transformers
### Install the necessary packages
- Requires: [Transformers](https://huggingface.co/docs/transformers) 4.35.0 or later.
- Requires: [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) 0.1.6 or later.
```shell
pip3 install --upgrade "autoawq>=0.1.6" "transformers>=4.35.0"
```
Note that if you are using PyTorch 2.0.1, the above AutoAWQ command will automatically upgrade you to PyTorch 2.1.0.
If you are using CUDA 11.8 and wish to continue using PyTorch 2.0.1, instead run this command:
```shell
pip3 install https://github.com/casper-hansen/AutoAWQ/releases/download/v0.1.6/autoawq-0.1.6+cu118-cp310-cp310-linux_x86_64.whl
```
If you have problems installing [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) using the pre-built wheels, install it from source instead:
```shell
pip3 uninstall -y autoawq
git clone https://github.com/casper-hansen/AutoAWQ
cd AutoAWQ
pip3 install .
```
### Transformers example code (requires Transformers 4.35.0 and later)
```python
from transformers import AutoModelForCausalLM, AutoTokenizer, TextStreamer
model_name_or_path = "TheBloke/prometheus-7B-v1.0-AWQ"
tokenizer = AutoTokenizer.from_pretrained(model_name_or_path)
model = AutoModelForCausalLM.from_pretrained(
model_name_or_path,
low_cpu_mem_usage=True,
device_map="cuda:0"
)
# Using the text streamer to stream output one token at a time
streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
prompt = "Tell me about AI"
prompt_template=f'''###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{prompt}
###Response to evaluate:
{{response}}
###Reference Answer (Score 5):
{{reference_answer}}
###Score Rubrics:
[{{criteria_description}}]
Score 1: {{score1_description}}
Score 2: {{score2_description}}
Score 3: {{score3_description}}
Score 4: {{score4_description}}
Score 5: {{score5_description}}
###Feedback:
'''
# Convert prompt to tokens
tokens = tokenizer(
prompt_template,
return_tensors='pt'
).input_ids.cuda()
generation_params = {
"do_sample": True,
"temperature": 0.7,
"top_p": 0.95,
"top_k": 40,
"max_new_tokens": 512,
"repetition_penalty": 1.1
}
# Generate streamed output, visible one token at a time
generation_output = model.generate(
tokens,
streamer=streamer,
**generation_params
)
# Generation without a streamer, which will include the prompt in the output
generation_output = model.generate(
tokens,
**generation_params
)
# Get the tokens from the output, decode them, print them
token_output = generation_output[0]
text_output = tokenizer.decode(token_output)
print("model.generate output: ", text_output)
# Inference is also possible via Transformers' pipeline
from transformers import pipeline
pipe = pipeline(
"text-generation",
model=model,
tokenizer=tokenizer,
**generation_params
)
pipe_output = pipe(prompt_template)[0]['generated_text']
print("pipeline output: ", pipe_output)
```
<!-- README_AWQ.md-use-from-python end -->
<!-- README_AWQ.md-compatibility start -->
## Compatibility
The files provided are tested to work with:
- [text-generation-webui](https://github.com/oobabooga/text-generation-webui) using `Loader: AutoAWQ`.
- [vLLM](https://github.com/vllm-project/vllm) version 0.2.0 and later.
- [Hugging Face Text Generation Inference (TGI)](https://github.com/huggingface/text-generation-inference) version 1.1.0 and later.
- [Transformers](https://huggingface.co/docs/transformers) version 4.35.0 and later.
- [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) version 0.1.1 and later.
<!-- README_AWQ.md-compatibility end -->
<!-- footer start -->
<!-- 200823 -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
## Thanks, and how to contribute
Thanks to the [chirper.ai](https://chirper.ai) team!
Thanks to Clay from [gpus.llm-utils.org](llm-utils)!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Aemon Algiz.
**Patreon special mentions**: Brandon Frisco, LangChain4j, Spiking Neurons AB, transmissions 11, Joseph William Delisle, Nitin Borwankar, Willem Michiel, Michael Dempsey, vamX, Jeffrey Morgan, zynix, jjj, Omer Bin Jawed, Sean Connelly, jinyuan sun, Jeromy Smith, Shadi, Pawan Osman, Chadd, Elijah Stavena, Illia Dulskyi, Sebastain Graf, Stephen Murray, terasurfer, Edmond Seymore, Celu Ramasamy, Mandus, Alex, biorpg, Ajan Kanaga, Clay Pascal, Raven Klaugh, 阿明, K, ya boyyy, usrbinkat, Alicia Loh, John Villwock, ReadyPlayerEmma, Chris Smitley, Cap'n Zoog, fincy, GodLy, S_X, sidney chen, Cory Kujawski, OG, Mano Prime, AzureBlack, Pieter, Kalila, Spencer Kim, Tom X Nguyen, Stanislav Ovsiannikov, Michael Levine, Andrey, Trailburnt, Vadim, Enrico Ros, Talal Aujan, Brandon Phillips, Jack West, Eugene Pentland, Michael Davis, Will Dee, webtim, Jonathan Leane, Alps Aficionado, Rooh Singh, Tiffany J. Kim, theTransient, Luke @flexchar, Elle, Caitlyn Gatomon, Ari Malik, subjectnull, Johann-Peter Hartmann, Trenton Dambrowitz, Imad Khwaja, Asp the Wyvern, Emad Mostaque, Rainer Wilmers, Alexandros Triantafyllidis, Nicholas, Pedro Madruga, SuperWojo, Harry Royden McLaughlin, James Bentley, Olakabola, David Ziegler, Ai Maven, Jeff Scroggin, Nikolai Manek, Deo Leter, Matthew Berman, Fen Risland, Ken Nordquist, Manuel Alberto Morcote, Luke Pendergrass, TL, Fred von Graf, Randy H, Dan Guido, NimbleBox.ai, Vitor Caleffi, Gabriel Tamborski, knownsqashed, Lone Striker, Erik Bjäreholt, John Detwiler, Leonard Tan, Iucharbius
Thank you to all my generous patrons and donaters!
And thank you again to a16z for their generous grant.
<!-- footer end -->
# Original model card: KAIST AI's Prometheus 7B V1.0
## Links for Reference
- **Homepage:https://github.com/kaistAI/Prometheus**
- **Repository:https://github.com/kaistAI/Prometheus**
- **Paper:https://arxiv.org/abs/2310.08491**
- **Point of Contact:[email protected]**
# TL;DR
Prometheus is an alternative of GPT-4 evaluation when doing fine-grained evaluation of an underlying LLM & a Reward model for Reinforcement Learning from Human Feedback (RLHF).
Prometheus is a language model using [Llama-2-Chat](https://huggingface.co/meta-llama/Llama-2-13b-chat-hf) as a base model and fine-tuned on 100K feedback within the [Feedback Collection](https://huggingface.co/datasets/kaist-ai/Feedback-Collection).
Since it was fine-tuned on a large amount of feedback, it is specialized at evaluating long-form responses, outperforming GPT-3.5-Turbo, Llama-2-Chat 70B, and on par with GPT-4 on various benchmarks.
Most importantly, this was possible since we appended 2 reference materials (reference answer, and customized score rubric).
Prometheus is a cheap and powerful alternative to GPT-4 evaluation, which one could use to evaluate LLMs with customized criteria (e.g., Child readability, Cultural Sensitivity, Creativity).
Also, it could be used as a reward model for Reinforcement Learning from Human Feedback (RLHF).
# Model Details
## Model Description
- **Model type:** Language model
- **Language(s) (NLP):** English
- **License:** Apache 2.0
- **Related Models:** [All Prometheus Checkpoints](https://huggingface.co/models?search=kaist-ai/Prometheus)
- **Resources for more information:**
- [Research paper](https://arxiv.org/abs/2310.08491)
- [GitHub Repo](https://github.com/kaistAI/Prometheus)
Prometheus is trained with two different sizes (7B and 13B).
You could check the 13B sized LM on [this page](https://huggingface.co/kaist-ai/prometheus-13b-v1.0).
Also, check out our dataset as well on [this page](https://huggingface.co/datasets/kaist-ai/Feedback-Collection).
## Prompt Format
Prometheus requires 4 components in the input: An instruction, a response to evaluate, a score rubric, and a reference answer. You could refer to the prompt format below.
You should fill in the instruction, response, reference answer, criteria description, and score description for score in range of 1 to 5.
```
###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{instruction}
###Response to evaluate:
{response}
###Reference Answer (Score 5):
{reference_answer}
###Score Rubrics:
[{criteria_description}]
Score 1: {score1_description}
Score 2: {score2_description}
Score 3: {score3_description}
Score 4: {score4_description}
Score 5: {score5_description}
###Feedback:
```
After this, you should apply the conversation template of Llama-2-Chat (not applying it might lead to unexpected behaviors).
You can find the conversation class at this [link](https://github.com/lm-sys/FastChat/blob/main/fastchat/conversation.py).
```
conv = get_conv_template("llama-2")
conv.set_system_message("You are a fair evaluator language model.")
conv.append_message(conv.roles[0], dialogs['instruction'])
conv.append_message(conv.roles[1], None)
prompt = conv.get_prompt()
x = tokenizer(prompt,truncation=False)
```
As a result, a feedback and score decision will be generated, divided by a separating phrase ```[RESULT]```
## License
Feedback Collection and Prometheus is subject to OpenAI's Terms of Use for the generated data. If you suspect any violations, please reach out to us.
# Usage
Find below some example scripts on how to use the model in `transformers`:
## Using the Pytorch model
### Running the model on a CPU
<details>
<summary> Click to expand </summary>
```python
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-7b-v1.0", device_map="auto")
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
### Running the model on a GPU
<details>
<summary> Click to expand </summary>
```python
# pip install accelerate
import torch
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-7b-v1.0", device_map="auto")
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids, sample=True, temperature=1.0, top_p=0.9, max_new_tokens=256, repetition_penalty=1.03)
print(tokenizer.decode(outputs[0]))
```
</details>
### Running the model on a GPU using different precisions
#### FP16
<details>
<summary> Click to expand </summary>
```python
# pip install accelerate
import torch
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-7b-v1.0", device_map="auto", torch_dtype=torch.float16)
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
#### INT8
<details>
<summary> Click to expand </summary>
```python
# pip install bitsandbytes accelerate
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-7b-v1.0", device_map="auto", load_in_8bit=True)
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
# Citation
If you find the following model helpful, please consider citing our paper!
**BibTeX:**
```bibtex
@misc{kim2023prometheus,
title={Prometheus: Inducing Fine-grained Evaluation Capability in Language Models},
author={Seungone Kim and Jamin Shin and Yejin Cho and Joel Jang and Shayne Longpre and Hwaran Lee and Sangdoo Yun and Seongjin Shin and Sungdong Kim and James Thorne and Minjoon Seo},
year={2023},
eprint={2310.08491},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
```
|
hkivancoral/hushem_1x_deit_tiny_rms_lr0001_fold2 | hkivancoral | 2023-11-10T19:10:00Z | 7 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:facebook/deit-tiny-patch16-224",
"base_model:finetune:facebook/deit-tiny-patch16-224",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| image-classification | 2023-11-10T19:07:51Z | ---
license: apache-2.0
base_model: facebook/deit-tiny-patch16-224
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: hushem_1x_deit_tiny_rms_lr0001_fold2
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: test
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.6888888888888889
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# hushem_1x_deit_tiny_rms_lr0001_fold2
This model is a fine-tuned version of [facebook/deit-tiny-patch16-224](https://huggingface.co/facebook/deit-tiny-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 1.7300
- Accuracy: 0.6889
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0001
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 1.0 | 6 | 3.0363 | 0.2444 |
| 2.6477 | 2.0 | 12 | 1.6954 | 0.2444 |
| 2.6477 | 3.0 | 18 | 1.4980 | 0.2444 |
| 1.482 | 4.0 | 24 | 1.3454 | 0.3556 |
| 1.4166 | 5.0 | 30 | 1.3094 | 0.4 |
| 1.4166 | 6.0 | 36 | 1.6095 | 0.2444 |
| 1.3414 | 7.0 | 42 | 1.9023 | 0.2444 |
| 1.3414 | 8.0 | 48 | 1.3957 | 0.2222 |
| 1.2396 | 9.0 | 54 | 1.1738 | 0.4 |
| 1.2068 | 10.0 | 60 | 1.2312 | 0.4889 |
| 1.2068 | 11.0 | 66 | 1.0903 | 0.6 |
| 0.9263 | 12.0 | 72 | 0.9211 | 0.5778 |
| 0.9263 | 13.0 | 78 | 1.1912 | 0.4444 |
| 0.8539 | 14.0 | 84 | 1.2631 | 0.5333 |
| 0.6672 | 15.0 | 90 | 1.2596 | 0.5111 |
| 0.6672 | 16.0 | 96 | 1.3999 | 0.4889 |
| 0.5299 | 17.0 | 102 | 1.2988 | 0.5556 |
| 0.5299 | 18.0 | 108 | 1.3328 | 0.5333 |
| 0.3853 | 19.0 | 114 | 1.0485 | 0.6222 |
| 0.332 | 20.0 | 120 | 1.1428 | 0.5778 |
| 0.332 | 21.0 | 126 | 1.0486 | 0.6444 |
| 0.1829 | 22.0 | 132 | 1.0866 | 0.6667 |
| 0.1829 | 23.0 | 138 | 1.7727 | 0.5778 |
| 0.111 | 24.0 | 144 | 1.2950 | 0.6889 |
| 0.0444 | 25.0 | 150 | 1.4579 | 0.7111 |
| 0.0444 | 26.0 | 156 | 1.4269 | 0.6889 |
| 0.0017 | 27.0 | 162 | 1.4804 | 0.6889 |
| 0.0017 | 28.0 | 168 | 1.5281 | 0.6889 |
| 0.0007 | 29.0 | 174 | 1.5658 | 0.6667 |
| 0.0005 | 30.0 | 180 | 1.5943 | 0.6667 |
| 0.0005 | 31.0 | 186 | 1.6212 | 0.6667 |
| 0.0004 | 32.0 | 192 | 1.6444 | 0.6667 |
| 0.0004 | 33.0 | 198 | 1.6608 | 0.6667 |
| 0.0003 | 34.0 | 204 | 1.6759 | 0.6667 |
| 0.0003 | 35.0 | 210 | 1.6896 | 0.6667 |
| 0.0003 | 36.0 | 216 | 1.7018 | 0.6667 |
| 0.0003 | 37.0 | 222 | 1.7108 | 0.6889 |
| 0.0003 | 38.0 | 228 | 1.7185 | 0.6889 |
| 0.0003 | 39.0 | 234 | 1.7237 | 0.6889 |
| 0.0002 | 40.0 | 240 | 1.7275 | 0.6889 |
| 0.0002 | 41.0 | 246 | 1.7295 | 0.6889 |
| 0.0003 | 42.0 | 252 | 1.7300 | 0.6889 |
| 0.0003 | 43.0 | 258 | 1.7300 | 0.6889 |
| 0.0002 | 44.0 | 264 | 1.7300 | 0.6889 |
| 0.0002 | 45.0 | 270 | 1.7300 | 0.6889 |
| 0.0002 | 46.0 | 276 | 1.7300 | 0.6889 |
| 0.0002 | 47.0 | 282 | 1.7300 | 0.6889 |
| 0.0002 | 48.0 | 288 | 1.7300 | 0.6889 |
| 0.0002 | 49.0 | 294 | 1.7300 | 0.6889 |
| 0.0002 | 50.0 | 300 | 1.7300 | 0.6889 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
recoilme/ColorfulSSD-1B_v05 | recoilme | 2023-11-10T19:03:23Z | 0 | 0 | diffusers | [
"diffusers",
"safetensors",
"stable-diffusion-xl",
"stable-diffusion-xl-diffusers",
"text-to-image",
"base_model:recoilme/ColorfulSSD-1B_v04",
"base_model:finetune:recoilme/ColorfulSSD-1B_v04",
"license:creativeml-openrail-m",
"autotrain_compatible",
"endpoints_compatible",
"diffusers:StableDiffusionXLPipeline",
"region:us"
]
| text-to-image | 2023-11-10T15:16:24Z |
---
license: creativeml-openrail-m
base_model: recoilme/ColorfulSSD-1B_v04
dataset: recoilme/portraits_xs
tags:
- stable-diffusion-xl
- stable-diffusion-xl-diffusers
- text-to-image
- diffusers
inference: true
---
# Text-to-image finetuning - recoilme/ColorfulSSD-1B_v05
This pipeline was finetuned from **recoilme/ColorfulSSD-1B_v04** on the **recoilme/portraits_xs** dataset. Below are some example images generated with the finetuned pipeline using the following prompt: None:
Special VAE used for training: madebyollin/sdxl-vae-fp16-fix.
|
TheBloke/prometheus-7B-v1.0-GGUF | TheBloke | 2023-11-10T19:00:34Z | 91 | 5 | transformers | [
"transformers",
"gguf",
"llama",
"text2text-generation",
"en",
"dataset:kaist-ai/Feedback-Collection",
"arxiv:2310.08491",
"base_model:prometheus-eval/prometheus-7b-v1.0",
"base_model:quantized:prometheus-eval/prometheus-7b-v1.0",
"license:apache-2.0",
"region:us"
]
| text2text-generation | 2023-11-10T18:56:09Z | ---
base_model: kaist-ai/prometheus-7b-v1.0
datasets:
- kaist-ai/Feedback-Collection
inference: false
language:
- en
library_name: transformers
license: apache-2.0
metrics:
- pearsonr
- spearmanr
- accuracy
model_creator: KAIST AI
model_name: Prometheus 7B V1.0
model_type: llama
pipeline_tag: text2text-generation
prompt_template: '###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference
answer that gets a score of 5, and a score rubric representing a evaluation criteria
are given.
1. Write a detailed feedback that assess the quality of the response strictly based
on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You
should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria)
[RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{prompt}
###Response to evaluate:
{{response}}
###Reference Answer (Score 5):
{{reference_answer}}
###Score Rubrics:
[{{criteria_description}}]
Score 1: {{score1_description}}
Score 2: {{score2_description}}
Score 3: {{score3_description}}
Score 4: {{score4_description}}
Score 5: {{score5_description}}
###Feedback:
'
quantized_by: TheBloke
tags:
- text2text-generation
---
<!-- markdownlint-disable MD041 -->
<!-- header start -->
<!-- 200823 -->
<div style="width: auto; margin-left: auto; margin-right: auto">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<div style="text-align:center; margin-top: 0em; margin-bottom: 0em"><p style="margin-top: 0.25em; margin-bottom: 0em;">TheBloke's LLM work is generously supported by a grant from <a href="https://a16z.com">andreessen horowitz (a16z)</a></p></div>
<hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
<!-- header end -->
# Prometheus 7B V1.0 - GGUF
- Model creator: [KAIST AI](https://huggingface.co/kaist-ai)
- Original model: [Prometheus 7B V1.0](https://huggingface.co/kaist-ai/prometheus-7b-v1.0)
<!-- description start -->
## Description
This repo contains GGUF format model files for [KAIST AI's Prometheus 7B V1.0](https://huggingface.co/kaist-ai/prometheus-7b-v1.0).
These files were quantised using hardware kindly provided by [Massed Compute](https://massedcompute.com/).
<!-- description end -->
<!-- README_GGUF.md-about-gguf start -->
### About GGUF
GGUF is a new format introduced by the llama.cpp team on August 21st 2023. It is a replacement for GGML, which is no longer supported by llama.cpp.
Here is an incomplete list of clients and libraries that are known to support GGUF:
* [llama.cpp](https://github.com/ggerganov/llama.cpp). The source project for GGUF. Offers a CLI and a server option.
* [text-generation-webui](https://github.com/oobabooga/text-generation-webui), the most widely used web UI, with many features and powerful extensions. Supports GPU acceleration.
* [KoboldCpp](https://github.com/LostRuins/koboldcpp), a fully featured web UI, with GPU accel across all platforms and GPU architectures. Especially good for story telling.
* [LM Studio](https://lmstudio.ai/), an easy-to-use and powerful local GUI for Windows and macOS (Silicon), with GPU acceleration.
* [LoLLMS Web UI](https://github.com/ParisNeo/lollms-webui), a great web UI with many interesting and unique features, including a full model library for easy model selection.
* [Faraday.dev](https://faraday.dev/), an attractive and easy to use character-based chat GUI for Windows and macOS (both Silicon and Intel), with GPU acceleration.
* [ctransformers](https://github.com/marella/ctransformers), a Python library with GPU accel, LangChain support, and OpenAI-compatible AI server.
* [llama-cpp-python](https://github.com/abetlen/llama-cpp-python), a Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
* [candle](https://github.com/huggingface/candle), a Rust ML framework with a focus on performance, including GPU support, and ease of use.
<!-- README_GGUF.md-about-gguf end -->
<!-- repositories-available start -->
## Repositories available
* [AWQ model(s) for GPU inference.](https://huggingface.co/TheBloke/prometheus-7B-v1.0-AWQ)
* [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/prometheus-7B-v1.0-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/prometheus-7B-v1.0-GGUF)
* [KAIST AI's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/kaist-ai/prometheus-7b-v1.0)
<!-- repositories-available end -->
<!-- prompt-template start -->
## Prompt template: Prometheus
```
###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{prompt}
###Response to evaluate:
{{response}}
###Reference Answer (Score 5):
{{reference_answer}}
###Score Rubrics:
[{{criteria_description}}]
Score 1: {{score1_description}}
Score 2: {{score2_description}}
Score 3: {{score3_description}}
Score 4: {{score4_description}}
Score 5: {{score5_description}}
###Feedback:
```
<!-- prompt-template end -->
<!-- licensing start -->
## Licensing
The creator of the source model has listed its license as `apache-2.0`, and this quantization has therefore used that same license.
As this model is based on Llama 2, it is also subject to the Meta Llama 2 license terms, and the license files for that are additionally included. It should therefore be considered as being claimed to be licensed under both licenses. I contacted Hugging Face for clarification on dual licensing but they do not yet have an official position. Should this change, or should Meta provide any feedback on this situation, I will update this section accordingly.
In the meantime, any questions regarding licensing, and in particular how these two licenses might interact, should be directed to the original model repository: [KAIST AI's Prometheus 7B V1.0](https://huggingface.co/kaist-ai/prometheus-7b-v1.0).
<!-- licensing end -->
<!-- compatibility_gguf start -->
## Compatibility
These quantised GGUFv2 files are compatible with llama.cpp from August 27th onwards, as of commit [d0cee0d](https://github.com/ggerganov/llama.cpp/commit/d0cee0d36d5be95a0d9088b674dbb27354107221)
They are also compatible with many third party UIs and libraries - please see the list at the top of this README.
## Explanation of quantisation methods
<details>
<summary>Click to see details</summary>
The new methods available are:
* GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
* GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
* GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
* GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
* GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
Refer to the Provided Files table below to see what files use which methods, and how.
</details>
<!-- compatibility_gguf end -->
<!-- README_GGUF.md-provided-files start -->
## Provided files
| Name | Quant method | Bits | Size | Max RAM required | Use case |
| ---- | ---- | ---- | ---- | ---- | ----- |
| [prometheus-7b-v1.0.Q2_K.gguf](https://huggingface.co/TheBloke/prometheus-7B-v1.0-GGUF/blob/main/prometheus-7b-v1.0.Q2_K.gguf) | Q2_K | 2 | 2.83 GB| 5.33 GB | smallest, significant quality loss - not recommended for most purposes |
| [prometheus-7b-v1.0.Q3_K_S.gguf](https://huggingface.co/TheBloke/prometheus-7B-v1.0-GGUF/blob/main/prometheus-7b-v1.0.Q3_K_S.gguf) | Q3_K_S | 3 | 2.95 GB| 5.45 GB | very small, high quality loss |
| [prometheus-7b-v1.0.Q3_K_M.gguf](https://huggingface.co/TheBloke/prometheus-7B-v1.0-GGUF/blob/main/prometheus-7b-v1.0.Q3_K_M.gguf) | Q3_K_M | 3 | 3.30 GB| 5.80 GB | very small, high quality loss |
| [prometheus-7b-v1.0.Q3_K_L.gguf](https://huggingface.co/TheBloke/prometheus-7B-v1.0-GGUF/blob/main/prometheus-7b-v1.0.Q3_K_L.gguf) | Q3_K_L | 3 | 3.60 GB| 6.10 GB | small, substantial quality loss |
| [prometheus-7b-v1.0.Q4_0.gguf](https://huggingface.co/TheBloke/prometheus-7B-v1.0-GGUF/blob/main/prometheus-7b-v1.0.Q4_0.gguf) | Q4_0 | 4 | 3.83 GB| 6.33 GB | legacy; small, very high quality loss - prefer using Q3_K_M |
| [prometheus-7b-v1.0.Q4_K_S.gguf](https://huggingface.co/TheBloke/prometheus-7B-v1.0-GGUF/blob/main/prometheus-7b-v1.0.Q4_K_S.gguf) | Q4_K_S | 4 | 3.86 GB| 6.36 GB | small, greater quality loss |
| [prometheus-7b-v1.0.Q4_K_M.gguf](https://huggingface.co/TheBloke/prometheus-7B-v1.0-GGUF/blob/main/prometheus-7b-v1.0.Q4_K_M.gguf) | Q4_K_M | 4 | 4.08 GB| 6.58 GB | medium, balanced quality - recommended |
| [prometheus-7b-v1.0.Q5_0.gguf](https://huggingface.co/TheBloke/prometheus-7B-v1.0-GGUF/blob/main/prometheus-7b-v1.0.Q5_0.gguf) | Q5_0 | 5 | 4.65 GB| 7.15 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
| [prometheus-7b-v1.0.Q5_K_S.gguf](https://huggingface.co/TheBloke/prometheus-7B-v1.0-GGUF/blob/main/prometheus-7b-v1.0.Q5_K_S.gguf) | Q5_K_S | 5 | 4.65 GB| 7.15 GB | large, low quality loss - recommended |
| [prometheus-7b-v1.0.Q5_K_M.gguf](https://huggingface.co/TheBloke/prometheus-7B-v1.0-GGUF/blob/main/prometheus-7b-v1.0.Q5_K_M.gguf) | Q5_K_M | 5 | 4.78 GB| 7.28 GB | large, very low quality loss - recommended |
| [prometheus-7b-v1.0.Q6_K.gguf](https://huggingface.co/TheBloke/prometheus-7B-v1.0-GGUF/blob/main/prometheus-7b-v1.0.Q6_K.gguf) | Q6_K | 6 | 5.53 GB| 8.03 GB | very large, extremely low quality loss |
| [prometheus-7b-v1.0.Q8_0.gguf](https://huggingface.co/TheBloke/prometheus-7B-v1.0-GGUF/blob/main/prometheus-7b-v1.0.Q8_0.gguf) | Q8_0 | 8 | 7.16 GB| 9.66 GB | very large, extremely low quality loss - not recommended |
**Note**: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.
<!-- README_GGUF.md-provided-files end -->
<!-- README_GGUF.md-how-to-download start -->
## How to download GGUF files
**Note for manual downloaders:** You almost never want to clone the entire repo! Multiple different quantisation formats are provided, and most users only want to pick and download a single file.
The following clients/libraries will automatically download models for you, providing a list of available models to choose from:
* LM Studio
* LoLLMS Web UI
* Faraday.dev
### In `text-generation-webui`
Under Download Model, you can enter the model repo: TheBloke/prometheus-7B-v1.0-GGUF and below it, a specific filename to download, such as: prometheus-7b-v1.0.Q4_K_M.gguf.
Then click Download.
### On the command line, including multiple files at once
I recommend using the `huggingface-hub` Python library:
```shell
pip3 install huggingface-hub
```
Then you can download any individual model file to the current directory, at high speed, with a command like this:
```shell
huggingface-cli download TheBloke/prometheus-7B-v1.0-GGUF prometheus-7b-v1.0.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
```
<details>
<summary>More advanced huggingface-cli download usage</summary>
You can also download multiple files at once with a pattern:
```shell
huggingface-cli download TheBloke/prometheus-7B-v1.0-GGUF --local-dir . --local-dir-use-symlinks False --include='*Q4_K*gguf'
```
For more documentation on downloading with `huggingface-cli`, please see: [HF -> Hub Python Library -> Download files -> Download from the CLI](https://huggingface.co/docs/huggingface_hub/guides/download#download-from-the-cli).
To accelerate downloads on fast connections (1Gbit/s or higher), install `hf_transfer`:
```shell
pip3 install hf_transfer
```
And set environment variable `HF_HUB_ENABLE_HF_TRANSFER` to `1`:
```shell
HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download TheBloke/prometheus-7B-v1.0-GGUF prometheus-7b-v1.0.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
```
Windows Command Line users: You can set the environment variable by running `set HF_HUB_ENABLE_HF_TRANSFER=1` before the download command.
</details>
<!-- README_GGUF.md-how-to-download end -->
<!-- README_GGUF.md-how-to-run start -->
## Example `llama.cpp` command
Make sure you are using `llama.cpp` from commit [d0cee0d](https://github.com/ggerganov/llama.cpp/commit/d0cee0d36d5be95a0d9088b674dbb27354107221) or later.
```shell
./main -ngl 32 -m prometheus-7b-v1.0.Q4_K_M.gguf --color -c 4096 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "###Task Description:\nAn instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.\n1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.\n2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.\n3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"\n4. Please do not generate any other opening, closing, and explanations.\n\n###The instruction to evaluate:\n{prompt}\n\n###Response to evaluate:\n{{response}}\n\n###Reference Answer (Score 5):\n{{reference_answer}}\n\n###Score Rubrics:\n[{{criteria_description}}]\nScore 1: {{score1_description}}\nScore 2: {{score2_description}}\nScore 3: {{score3_description}}\nScore 4: {{score4_description}}\nScore 5: {{score5_description}}\n\n###Feedback:"
```
Change `-ngl 32` to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.
Change `-c 4096` to the desired sequence length. For extended sequence models - eg 8K, 16K, 32K - the necessary RoPE scaling parameters are read from the GGUF file and set by llama.cpp automatically.
If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`
For other parameters and how to use them, please refer to [the llama.cpp documentation](https://github.com/ggerganov/llama.cpp/blob/master/examples/main/README.md)
## How to run in `text-generation-webui`
Further instructions here: [text-generation-webui/docs/llama.cpp.md](https://github.com/oobabooga/text-generation-webui/blob/main/docs/llama.cpp.md).
## How to run from Python code
You can use GGUF models from Python using the [llama-cpp-python](https://github.com/abetlen/llama-cpp-python) or [ctransformers](https://github.com/marella/ctransformers) libraries.
### How to load this model in Python code, using ctransformers
#### First install the package
Run one of the following commands, according to your system:
```shell
# Base ctransformers with no GPU acceleration
pip install ctransformers
# Or with CUDA GPU acceleration
pip install ctransformers[cuda]
# Or with AMD ROCm GPU acceleration (Linux only)
CT_HIPBLAS=1 pip install ctransformers --no-binary ctransformers
# Or with Metal GPU acceleration for macOS systems only
CT_METAL=1 pip install ctransformers --no-binary ctransformers
```
#### Simple ctransformers example code
```python
from ctransformers import AutoModelForCausalLM
# Set gpu_layers to the number of layers to offload to GPU. Set to 0 if no GPU acceleration is available on your system.
llm = AutoModelForCausalLM.from_pretrained("TheBloke/prometheus-7B-v1.0-GGUF", model_file="prometheus-7b-v1.0.Q4_K_M.gguf", model_type="llama", gpu_layers=50)
print(llm("AI is going to"))
```
## How to use with LangChain
Here are guides on using llama-cpp-python and ctransformers with LangChain:
* [LangChain + llama-cpp-python](https://python.langchain.com/docs/integrations/llms/llamacpp)
* [LangChain + ctransformers](https://python.langchain.com/docs/integrations/providers/ctransformers)
<!-- README_GGUF.md-how-to-run end -->
<!-- footer start -->
<!-- 200823 -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
## Thanks, and how to contribute
Thanks to the [chirper.ai](https://chirper.ai) team!
Thanks to Clay from [gpus.llm-utils.org](llm-utils)!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Aemon Algiz.
**Patreon special mentions**: Brandon Frisco, LangChain4j, Spiking Neurons AB, transmissions 11, Joseph William Delisle, Nitin Borwankar, Willem Michiel, Michael Dempsey, vamX, Jeffrey Morgan, zynix, jjj, Omer Bin Jawed, Sean Connelly, jinyuan sun, Jeromy Smith, Shadi, Pawan Osman, Chadd, Elijah Stavena, Illia Dulskyi, Sebastain Graf, Stephen Murray, terasurfer, Edmond Seymore, Celu Ramasamy, Mandus, Alex, biorpg, Ajan Kanaga, Clay Pascal, Raven Klaugh, 阿明, K, ya boyyy, usrbinkat, Alicia Loh, John Villwock, ReadyPlayerEmma, Chris Smitley, Cap'n Zoog, fincy, GodLy, S_X, sidney chen, Cory Kujawski, OG, Mano Prime, AzureBlack, Pieter, Kalila, Spencer Kim, Tom X Nguyen, Stanislav Ovsiannikov, Michael Levine, Andrey, Trailburnt, Vadim, Enrico Ros, Talal Aujan, Brandon Phillips, Jack West, Eugene Pentland, Michael Davis, Will Dee, webtim, Jonathan Leane, Alps Aficionado, Rooh Singh, Tiffany J. Kim, theTransient, Luke @flexchar, Elle, Caitlyn Gatomon, Ari Malik, subjectnull, Johann-Peter Hartmann, Trenton Dambrowitz, Imad Khwaja, Asp the Wyvern, Emad Mostaque, Rainer Wilmers, Alexandros Triantafyllidis, Nicholas, Pedro Madruga, SuperWojo, Harry Royden McLaughlin, James Bentley, Olakabola, David Ziegler, Ai Maven, Jeff Scroggin, Nikolai Manek, Deo Leter, Matthew Berman, Fen Risland, Ken Nordquist, Manuel Alberto Morcote, Luke Pendergrass, TL, Fred von Graf, Randy H, Dan Guido, NimbleBox.ai, Vitor Caleffi, Gabriel Tamborski, knownsqashed, Lone Striker, Erik Bjäreholt, John Detwiler, Leonard Tan, Iucharbius
Thank you to all my generous patrons and donaters!
And thank you again to a16z for their generous grant.
<!-- footer end -->
<!-- original-model-card start -->
# Original model card: KAIST AI's Prometheus 7B V1.0
## Links for Reference
- **Homepage:https://github.com/kaistAI/Prometheus**
- **Repository:https://github.com/kaistAI/Prometheus**
- **Paper:https://arxiv.org/abs/2310.08491**
- **Point of Contact:[email protected]**
# TL;DR
Prometheus is an alternative of GPT-4 evaluation when doing fine-grained evaluation of an underlying LLM & a Reward model for Reinforcement Learning from Human Feedback (RLHF).
Prometheus is a language model using [Llama-2-Chat](https://huggingface.co/meta-llama/Llama-2-13b-chat-hf) as a base model and fine-tuned on 100K feedback within the [Feedback Collection](https://huggingface.co/datasets/kaist-ai/Feedback-Collection).
Since it was fine-tuned on a large amount of feedback, it is specialized at evaluating long-form responses, outperforming GPT-3.5-Turbo, Llama-2-Chat 70B, and on par with GPT-4 on various benchmarks.
Most importantly, this was possible since we appended 2 reference materials (reference answer, and customized score rubric).
Prometheus is a cheap and powerful alternative to GPT-4 evaluation, which one could use to evaluate LLMs with customized criteria (e.g., Child readability, Cultural Sensitivity, Creativity).
Also, it could be used as a reward model for Reinforcement Learning from Human Feedback (RLHF).
# Model Details
## Model Description
- **Model type:** Language model
- **Language(s) (NLP):** English
- **License:** Apache 2.0
- **Related Models:** [All Prometheus Checkpoints](https://huggingface.co/models?search=kaist-ai/Prometheus)
- **Resources for more information:**
- [Research paper](https://arxiv.org/abs/2310.08491)
- [GitHub Repo](https://github.com/kaistAI/Prometheus)
Prometheus is trained with two different sizes (7B and 13B).
You could check the 13B sized LM on [this page](https://huggingface.co/kaist-ai/prometheus-13b-v1.0).
Also, check out our dataset as well on [this page](https://huggingface.co/datasets/kaist-ai/Feedback-Collection).
## Prompt Format
Prometheus requires 4 components in the input: An instruction, a response to evaluate, a score rubric, and a reference answer. You could refer to the prompt format below.
You should fill in the instruction, response, reference answer, criteria description, and score description for score in range of 1 to 5.
```
###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{instruction}
###Response to evaluate:
{response}
###Reference Answer (Score 5):
{reference_answer}
###Score Rubrics:
[{criteria_description}]
Score 1: {score1_description}
Score 2: {score2_description}
Score 3: {score3_description}
Score 4: {score4_description}
Score 5: {score5_description}
###Feedback:
```
After this, you should apply the conversation template of Llama-2-Chat (not applying it might lead to unexpected behaviors).
You can find the conversation class at this [link](https://github.com/lm-sys/FastChat/blob/main/fastchat/conversation.py).
```
conv = get_conv_template("llama-2")
conv.set_system_message("You are a fair evaluator language model.")
conv.append_message(conv.roles[0], dialogs['instruction'])
conv.append_message(conv.roles[1], None)
prompt = conv.get_prompt()
x = tokenizer(prompt,truncation=False)
```
As a result, a feedback and score decision will be generated, divided by a separating phrase ```[RESULT]```
## License
Feedback Collection and Prometheus is subject to OpenAI's Terms of Use for the generated data. If you suspect any violations, please reach out to us.
# Usage
Find below some example scripts on how to use the model in `transformers`:
## Using the Pytorch model
### Running the model on a CPU
<details>
<summary> Click to expand </summary>
```python
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-7b-v1.0", device_map="auto")
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
### Running the model on a GPU
<details>
<summary> Click to expand </summary>
```python
# pip install accelerate
import torch
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-7b-v1.0", device_map="auto")
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids, sample=True, temperature=1.0, top_p=0.9, max_new_tokens=256, repetition_penalty=1.03)
print(tokenizer.decode(outputs[0]))
```
</details>
### Running the model on a GPU using different precisions
#### FP16
<details>
<summary> Click to expand </summary>
```python
# pip install accelerate
import torch
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-7b-v1.0", device_map="auto", torch_dtype=torch.float16)
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
#### INT8
<details>
<summary> Click to expand </summary>
```python
# pip install bitsandbytes accelerate
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-7b-v1.0", device_map="auto", load_in_8bit=True)
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
# Citation
If you find the following model helpful, please consider citing our paper!
**BibTeX:**
```bibtex
@misc{kim2023prometheus,
title={Prometheus: Inducing Fine-grained Evaluation Capability in Language Models},
author={Seungone Kim and Jamin Shin and Yejin Cho and Joel Jang and Shayne Longpre and Hwaran Lee and Sangdoo Yun and Seongjin Shin and Sungdong Kim and James Thorne and Minjoon Seo},
year={2023},
eprint={2310.08491},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
```
<!-- original-model-card end -->
|
TheBloke/prometheus-13B-v1.0-GPTQ | TheBloke | 2023-11-10T18:54:58Z | 34 | 2 | transformers | [
"transformers",
"safetensors",
"llama",
"text-generation",
"text2text-generation",
"en",
"dataset:kaist-ai/Feedback-Collection",
"arxiv:2310.08491",
"base_model:prometheus-eval/prometheus-13b-v1.0",
"base_model:quantized:prometheus-eval/prometheus-13b-v1.0",
"license:apache-2.0",
"autotrain_compatible",
"text-generation-inference",
"4-bit",
"gptq",
"region:us"
]
| text2text-generation | 2023-11-10T18:08:26Z | ---
base_model: kaist-ai/prometheus-13b-v1.0
datasets:
- kaist-ai/Feedback-Collection
inference: false
language:
- en
library_name: transformers
license: apache-2.0
metrics:
- pearsonr
- spearmanr
- accuracy
model_creator: KAIST AI
model_name: Prometheus 13B V1.0
model_type: llama
pipeline_tag: text2text-generation
prompt_template: '###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference
answer that gets a score of 5, and a score rubric representing a evaluation criteria
are given.
1. Write a detailed feedback that assess the quality of the response strictly based
on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You
should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria)
[RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{prompt}
###Response to evaluate:
{{response}}
###Reference Answer (Score 5):
{{reference_answer}}
###Score Rubrics:
[{{criteria_description}}]
Score 1: {{score1_description}}
Score 2: {{score2_description}}
Score 3: {{score3_description}}
Score 4: {{score4_description}}
Score 5: {{score5_description}}
###Feedback:
'
quantized_by: TheBloke
tags:
- text2text-generation
---
<!-- markdownlint-disable MD041 -->
<!-- header start -->
<!-- 200823 -->
<div style="width: auto; margin-left: auto; margin-right: auto">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<div style="text-align:center; margin-top: 0em; margin-bottom: 0em"><p style="margin-top: 0.25em; margin-bottom: 0em;">TheBloke's LLM work is generously supported by a grant from <a href="https://a16z.com">andreessen horowitz (a16z)</a></p></div>
<hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
<!-- header end -->
# Prometheus 13B V1.0 - GPTQ
- Model creator: [KAIST AI](https://huggingface.co/kaist-ai)
- Original model: [Prometheus 13B V1.0](https://huggingface.co/kaist-ai/prometheus-13b-v1.0)
<!-- description start -->
## Description
This repo contains GPTQ model files for [KAIST AI's Prometheus 13B V1.0](https://huggingface.co/kaist-ai/prometheus-13b-v1.0).
Multiple GPTQ parameter permutations are provided; see Provided Files below for details of the options provided, their parameters, and the software used to create them.
These files were quantised using hardware kindly provided by [Massed Compute](https://massedcompute.com/).
<!-- description end -->
<!-- repositories-available start -->
## Repositories available
* [AWQ model(s) for GPU inference.](https://huggingface.co/TheBloke/prometheus-13B-v1.0-AWQ)
* [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GGUF)
* [KAIST AI's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/kaist-ai/prometheus-13b-v1.0)
<!-- repositories-available end -->
<!-- prompt-template start -->
## Prompt template: Prometheus
```
###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{prompt}
###Response to evaluate:
{{response}}
###Reference Answer (Score 5):
{{reference_answer}}
###Score Rubrics:
[{{criteria_description}}]
Score 1: {{score1_description}}
Score 2: {{score2_description}}
Score 3: {{score3_description}}
Score 4: {{score4_description}}
Score 5: {{score5_description}}
###Feedback:
```
<!-- prompt-template end -->
<!-- licensing start -->
## Licensing
The creator of the source model has listed its license as `apache-2.0`, and this quantization has therefore used that same license.
As this model is based on Llama 2, it is also subject to the Meta Llama 2 license terms, and the license files for that are additionally included. It should therefore be considered as being claimed to be licensed under both licenses. I contacted Hugging Face for clarification on dual licensing but they do not yet have an official position. Should this change, or should Meta provide any feedback on this situation, I will update this section accordingly.
In the meantime, any questions regarding licensing, and in particular how these two licenses might interact, should be directed to the original model repository: [KAIST AI's Prometheus 13B V1.0](https://huggingface.co/kaist-ai/prometheus-13b-v1.0).
<!-- licensing end -->
<!-- README_GPTQ.md-compatible clients start -->
## Known compatible clients / servers
These GPTQ models are known to work in the following inference servers/webuis.
- [text-generation-webui](https://github.com/oobabooga/text-generation-webui)
- [KoboldAI United](https://github.com/henk717/koboldai)
- [LoLLMS Web UI](https://github.com/ParisNeo/lollms-webui)
- [Hugging Face Text Generation Inference (TGI)](https://github.com/huggingface/text-generation-inference)
This may not be a complete list; if you know of others, please let me know!
<!-- README_GPTQ.md-compatible clients end -->
<!-- README_GPTQ.md-provided-files start -->
## Provided files, and GPTQ parameters
Multiple quantisation parameters are provided, to allow you to choose the best one for your hardware and requirements.
Each separate quant is in a different branch. See below for instructions on fetching from different branches.
Most GPTQ files are made with AutoGPTQ. Mistral models are currently made with Transformers.
<details>
<summary>Explanation of GPTQ parameters</summary>
- Bits: The bit size of the quantised model.
- GS: GPTQ group size. Higher numbers use less VRAM, but have lower quantisation accuracy. "None" is the lowest possible value.
- Act Order: True or False. Also known as `desc_act`. True results in better quantisation accuracy. Some GPTQ clients have had issues with models that use Act Order plus Group Size, but this is generally resolved now.
- Damp %: A GPTQ parameter that affects how samples are processed for quantisation. 0.01 is default, but 0.1 results in slightly better accuracy.
- GPTQ dataset: The calibration dataset used during quantisation. Using a dataset more appropriate to the model's training can improve quantisation accuracy. Note that the GPTQ calibration dataset is not the same as the dataset used to train the model - please refer to the original model repo for details of the training dataset(s).
- Sequence Length: The length of the dataset sequences used for quantisation. Ideally this is the same as the model sequence length. For some very long sequence models (16+K), a lower sequence length may have to be used. Note that a lower sequence length does not limit the sequence length of the quantised model. It only impacts the quantisation accuracy on longer inference sequences.
- ExLlama Compatibility: Whether this file can be loaded with ExLlama, which currently only supports Llama and Mistral models in 4-bit.
</details>
| Branch | Bits | GS | Act Order | Damp % | GPTQ Dataset | Seq Len | Size | ExLlama | Desc |
| ------ | ---- | -- | --------- | ------ | ------------ | ------- | ---- | ------- | ---- |
| [main](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GPTQ/tree/main) | 4 | 128 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-raw-v1) | 4096 | 7.26 GB | Yes | 4-bit, with Act Order and group size 128g. Uses even less VRAM than 64g, but with slightly lower accuracy. |
| [gptq-4bit-32g-actorder_True](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GPTQ/tree/gptq-4bit-32g-actorder_True) | 4 | 32 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-raw-v1) | 4096 | 8.00 GB | Yes | 4-bit, with Act Order and group size 32g. Gives highest possible inference quality, with maximum VRAM usage. |
| [gptq-8bit--1g-actorder_True](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GPTQ/tree/gptq-8bit--1g-actorder_True) | 8 | None | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-raw-v1) | 4096 | 13.36 GB | No | 8-bit, with Act Order. No group size, to lower VRAM requirements. |
| [gptq-8bit-128g-actorder_True](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GPTQ/tree/gptq-8bit-128g-actorder_True) | 8 | 128 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-raw-v1) | 4096 | 13.65 GB | No | 8-bit, with group size 128g for higher inference quality and with Act Order for even higher accuracy. |
| [gptq-8bit-32g-actorder_True](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GPTQ/tree/gptq-8bit-32g-actorder_True) | 8 | 32 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-raw-v1) | 4096 | 14.54 GB | No | 8-bit, with group size 32g and Act Order for maximum inference quality. |
| [gptq-4bit-64g-actorder_True](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GPTQ/tree/gptq-4bit-64g-actorder_True) | 4 | 64 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-raw-v1) | 4096 | 7.51 GB | Yes | 4-bit, with Act Order and group size 64g. Uses less VRAM than 32g, but with slightly lower accuracy. |
<!-- README_GPTQ.md-provided-files end -->
<!-- README_GPTQ.md-download-from-branches start -->
## How to download, including from branches
### In text-generation-webui
To download from the `main` branch, enter `TheBloke/prometheus-13B-v1.0-GPTQ` in the "Download model" box.
To download from another branch, add `:branchname` to the end of the download name, eg `TheBloke/prometheus-13B-v1.0-GPTQ:gptq-4bit-32g-actorder_True`
### From the command line
I recommend using the `huggingface-hub` Python library:
```shell
pip3 install huggingface-hub
```
To download the `main` branch to a folder called `prometheus-13B-v1.0-GPTQ`:
```shell
mkdir prometheus-13B-v1.0-GPTQ
huggingface-cli download TheBloke/prometheus-13B-v1.0-GPTQ --local-dir prometheus-13B-v1.0-GPTQ --local-dir-use-symlinks False
```
To download from a different branch, add the `--revision` parameter:
```shell
mkdir prometheus-13B-v1.0-GPTQ
huggingface-cli download TheBloke/prometheus-13B-v1.0-GPTQ --revision gptq-4bit-32g-actorder_True --local-dir prometheus-13B-v1.0-GPTQ --local-dir-use-symlinks False
```
<details>
<summary>More advanced huggingface-cli download usage</summary>
If you remove the `--local-dir-use-symlinks False` parameter, the files will instead be stored in the central Hugging Face cache directory (default location on Linux is: `~/.cache/huggingface`), and symlinks will be added to the specified `--local-dir`, pointing to their real location in the cache. This allows for interrupted downloads to be resumed, and allows you to quickly clone the repo to multiple places on disk without triggering a download again. The downside, and the reason why I don't list that as the default option, is that the files are then hidden away in a cache folder and it's harder to know where your disk space is being used, and to clear it up if/when you want to remove a download model.
The cache location can be changed with the `HF_HOME` environment variable, and/or the `--cache-dir` parameter to `huggingface-cli`.
For more documentation on downloading with `huggingface-cli`, please see: [HF -> Hub Python Library -> Download files -> Download from the CLI](https://huggingface.co/docs/huggingface_hub/guides/download#download-from-the-cli).
To accelerate downloads on fast connections (1Gbit/s or higher), install `hf_transfer`:
```shell
pip3 install hf_transfer
```
And set environment variable `HF_HUB_ENABLE_HF_TRANSFER` to `1`:
```shell
mkdir prometheus-13B-v1.0-GPTQ
HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download TheBloke/prometheus-13B-v1.0-GPTQ --local-dir prometheus-13B-v1.0-GPTQ --local-dir-use-symlinks False
```
Windows Command Line users: You can set the environment variable by running `set HF_HUB_ENABLE_HF_TRANSFER=1` before the download command.
</details>
### With `git` (**not** recommended)
To clone a specific branch with `git`, use a command like this:
```shell
git clone --single-branch --branch gptq-4bit-32g-actorder_True https://huggingface.co/TheBloke/prometheus-13B-v1.0-GPTQ
```
Note that using Git with HF repos is strongly discouraged. It will be much slower than using `huggingface-hub`, and will use twice as much disk space as it has to store the model files twice (it stores every byte both in the intended target folder, and again in the `.git` folder as a blob.)
<!-- README_GPTQ.md-download-from-branches end -->
<!-- README_GPTQ.md-text-generation-webui start -->
## How to easily download and use this model in [text-generation-webui](https://github.com/oobabooga/text-generation-webui)
Please make sure you're using the latest version of [text-generation-webui](https://github.com/oobabooga/text-generation-webui).
It is strongly recommended to use the text-generation-webui one-click-installers unless you're sure you know how to make a manual install.
1. Click the **Model tab**.
2. Under **Download custom model or LoRA**, enter `TheBloke/prometheus-13B-v1.0-GPTQ`.
- To download from a specific branch, enter for example `TheBloke/prometheus-13B-v1.0-GPTQ:gptq-4bit-32g-actorder_True`
- see Provided Files above for the list of branches for each option.
3. Click **Download**.
4. The model will start downloading. Once it's finished it will say "Done".
5. In the top left, click the refresh icon next to **Model**.
6. In the **Model** dropdown, choose the model you just downloaded: `prometheus-13B-v1.0-GPTQ`
7. The model will automatically load, and is now ready for use!
8. If you want any custom settings, set them and then click **Save settings for this model** followed by **Reload the Model** in the top right.
- Note that you do not need to and should not set manual GPTQ parameters any more. These are set automatically from the file `quantize_config.json`.
9. Once you're ready, click the **Text Generation** tab and enter a prompt to get started!
<!-- README_GPTQ.md-text-generation-webui end -->
<!-- README_GPTQ.md-use-from-tgi start -->
## Serving this model from Text Generation Inference (TGI)
It's recommended to use TGI version 1.1.0 or later. The official Docker container is: `ghcr.io/huggingface/text-generation-inference:1.1.0`
Example Docker parameters:
```shell
--model-id TheBloke/prometheus-13B-v1.0-GPTQ --port 3000 --quantize gptq --max-input-length 3696 --max-total-tokens 4096 --max-batch-prefill-tokens 4096
```
Example Python code for interfacing with TGI (requires huggingface-hub 0.17.0 or later):
```shell
pip3 install huggingface-hub
```
```python
from huggingface_hub import InferenceClient
endpoint_url = "https://your-endpoint-url-here"
prompt = "Tell me about AI"
prompt_template=f'''###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{prompt}
###Response to evaluate:
{{response}}
###Reference Answer (Score 5):
{{reference_answer}}
###Score Rubrics:
[{{criteria_description}}]
Score 1: {{score1_description}}
Score 2: {{score2_description}}
Score 3: {{score3_description}}
Score 4: {{score4_description}}
Score 5: {{score5_description}}
###Feedback:
'''
client = InferenceClient(endpoint_url)
response = client.text_generation(prompt,
max_new_tokens=128,
do_sample=True,
temperature=0.7,
top_p=0.95,
top_k=40,
repetition_penalty=1.1)
print(f"Model output: {response}")
```
<!-- README_GPTQ.md-use-from-tgi end -->
<!-- README_GPTQ.md-use-from-python start -->
## How to use this GPTQ model from Python code
### Install the necessary packages
Requires: Transformers 4.33.0 or later, Optimum 1.12.0 or later, and AutoGPTQ 0.4.2 or later.
```shell
pip3 install transformers optimum
pip3 install auto-gptq --extra-index-url https://huggingface.github.io/autogptq-index/whl/cu118/ # Use cu117 if on CUDA 11.7
```
If you have problems installing AutoGPTQ using the pre-built wheels, install it from source instead:
```shell
pip3 uninstall -y auto-gptq
git clone https://github.com/PanQiWei/AutoGPTQ
cd AutoGPTQ
git checkout v0.4.2
pip3 install .
```
### You can then use the following code
```python
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
model_name_or_path = "TheBloke/prometheus-13B-v1.0-GPTQ"
# To use a different branch, change revision
# For example: revision="gptq-4bit-32g-actorder_True"
model = AutoModelForCausalLM.from_pretrained(model_name_or_path,
device_map="auto",
trust_remote_code=False,
revision="main")
tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True)
prompt = "Tell me about AI"
prompt_template=f'''###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{prompt}
###Response to evaluate:
{{response}}
###Reference Answer (Score 5):
{{reference_answer}}
###Score Rubrics:
[{{criteria_description}}]
Score 1: {{score1_description}}
Score 2: {{score2_description}}
Score 3: {{score3_description}}
Score 4: {{score4_description}}
Score 5: {{score5_description}}
###Feedback:
'''
print("\n\n*** Generate:")
input_ids = tokenizer(prompt_template, return_tensors='pt').input_ids.cuda()
output = model.generate(inputs=input_ids, temperature=0.7, do_sample=True, top_p=0.95, top_k=40, max_new_tokens=512)
print(tokenizer.decode(output[0]))
# Inference can also be done using transformers' pipeline
print("*** Pipeline:")
pipe = pipeline(
"text-generation",
model=model,
tokenizer=tokenizer,
max_new_tokens=512,
do_sample=True,
temperature=0.7,
top_p=0.95,
top_k=40,
repetition_penalty=1.1
)
print(pipe(prompt_template)[0]['generated_text'])
```
<!-- README_GPTQ.md-use-from-python end -->
<!-- README_GPTQ.md-compatibility start -->
## Compatibility
The files provided are tested to work with Transformers. For non-Mistral models, AutoGPTQ can also be used directly.
[ExLlama](https://github.com/turboderp/exllama) is compatible with Llama and Mistral models in 4-bit. Please see the Provided Files table above for per-file compatibility.
For a list of clients/servers, please see "Known compatible clients / servers", above.
<!-- README_GPTQ.md-compatibility end -->
<!-- footer start -->
<!-- 200823 -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
## Thanks, and how to contribute
Thanks to the [chirper.ai](https://chirper.ai) team!
Thanks to Clay from [gpus.llm-utils.org](llm-utils)!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Aemon Algiz.
**Patreon special mentions**: Brandon Frisco, LangChain4j, Spiking Neurons AB, transmissions 11, Joseph William Delisle, Nitin Borwankar, Willem Michiel, Michael Dempsey, vamX, Jeffrey Morgan, zynix, jjj, Omer Bin Jawed, Sean Connelly, jinyuan sun, Jeromy Smith, Shadi, Pawan Osman, Chadd, Elijah Stavena, Illia Dulskyi, Sebastain Graf, Stephen Murray, terasurfer, Edmond Seymore, Celu Ramasamy, Mandus, Alex, biorpg, Ajan Kanaga, Clay Pascal, Raven Klaugh, 阿明, K, ya boyyy, usrbinkat, Alicia Loh, John Villwock, ReadyPlayerEmma, Chris Smitley, Cap'n Zoog, fincy, GodLy, S_X, sidney chen, Cory Kujawski, OG, Mano Prime, AzureBlack, Pieter, Kalila, Spencer Kim, Tom X Nguyen, Stanislav Ovsiannikov, Michael Levine, Andrey, Trailburnt, Vadim, Enrico Ros, Talal Aujan, Brandon Phillips, Jack West, Eugene Pentland, Michael Davis, Will Dee, webtim, Jonathan Leane, Alps Aficionado, Rooh Singh, Tiffany J. Kim, theTransient, Luke @flexchar, Elle, Caitlyn Gatomon, Ari Malik, subjectnull, Johann-Peter Hartmann, Trenton Dambrowitz, Imad Khwaja, Asp the Wyvern, Emad Mostaque, Rainer Wilmers, Alexandros Triantafyllidis, Nicholas, Pedro Madruga, SuperWojo, Harry Royden McLaughlin, James Bentley, Olakabola, David Ziegler, Ai Maven, Jeff Scroggin, Nikolai Manek, Deo Leter, Matthew Berman, Fen Risland, Ken Nordquist, Manuel Alberto Morcote, Luke Pendergrass, TL, Fred von Graf, Randy H, Dan Guido, NimbleBox.ai, Vitor Caleffi, Gabriel Tamborski, knownsqashed, Lone Striker, Erik Bjäreholt, John Detwiler, Leonard Tan, Iucharbius
Thank you to all my generous patrons and donaters!
And thank you again to a16z for their generous grant.
<!-- footer end -->
# Original model card: KAIST AI's Prometheus 13B V1.0
## Links for Reference
- **Homepage:https://github.com/kaistAI/Prometheus**
- **Repository:https://github.com/kaistAI/Prometheus**
- **Paper:https://arxiv.org/abs/2310.08491**
- **Point of Contact:[email protected]**
# TL;DR
Prometheus is an alternative of GPT-4 evaluation when doing fine-grained evaluation of an underlying LLM & a Reward model for Reinforcement Learning from Human Feedback (RLHF).
Prometheus is a language model using [Llama-2-Chat](https://huggingface.co/meta-llama/Llama-2-13b-chat-hf) as a base model and fine-tuned on 100K feedback within the [Feedback Collection](https://huggingface.co/datasets/kaist-ai/Feedback-Collection).
Since it was fine-tuned on a large amount of feedback, it is specialized at evaluating long-form responses, outperforming GPT-3.5-Turbo, Llama-2-Chat 70B, and on par with GPT-4 on various benchmarks.
Most importantly, this was possible since we appended 2 reference materials (reference answer, and customized score rubric).
Prometheus is a cheap and powerful alternative to GPT-4 evaluation, which one could use to evaluate LLMs with customized criteria (e.g., Child readability, Cultural Sensitivity, Creativity).
Also, it could be used as a reward model for Reinforcement Learning from Human Feedback (RLHF).
# Model Details
## Model Description
- **Model type:** Language model
- **Language(s) (NLP):** English
- **License:** Apache 2.0
- **Related Models:** [All Prometheus Checkpoints](https://huggingface.co/models?search=kaist-ai/Prometheus)
- **Resources for more information:**
- [Research paper](https://arxiv.org/abs/2310.08491)
- [GitHub Repo](https://github.com/kaistAI/Prometheus)
Prometheus is trained with two different sizes (7B and 13B).
You could check the 7B sized LM on [this page](https://huggingface.co/kaist-ai/prometheus-7b-v1.0).
Also, check out our dataset as well on [this page](https://huggingface.co/datasets/kaist-ai/Feedback-Collection).
## Prompt Format
Prometheus requires 4 components in the input: An instruction, a response to evaluate, a score rubric, and a reference answer. You could refer to the prompt format below.
You should fill in the instruction, response, reference answer, criteria description, and score description for score in range of 1 to 5.
```
###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{instruction}
###Response to evaluate:
{response}
###Reference Answer (Score 5):
{reference_answer}
###Score Rubrics:
[{criteria_description}]
Score 1: {score1_description}
Score 2: {score2_description}
Score 3: {score3_description}
Score 4: {score4_description}
Score 5: {score5_description}
###Feedback:
```
After this, you should apply the conversation template of Llama-2-Chat (not applying it might lead to unexpected behaviors).
You can find the conversation class at this [link](https://github.com/lm-sys/FastChat/blob/main/fastchat/conversation.py).
```
conv = get_conv_template("llama-2")
conv.set_system_message("You are a fair evaluator language model.")
conv.append_message(conv.roles[0], dialogs['instruction'])
conv.append_message(conv.roles[1], None)
prompt = conv.get_prompt()
x = tokenizer(prompt,truncation=False)
```
As a result, a feedback and score decision will be generated, divided by a separating phrase ```[RESULT]```
## License
Feedback Collection and Prometheus is subject to OpenAI's Terms of Use for the generated data. If you suspect any violations, please reach out to us.
# Usage
Find below some example scripts on how to use the model in `transformers`:
## Using the Pytorch model
### Running the model on a CPU
<details>
<summary> Click to expand </summary>
```python
from transformers import T5Tokenizer, T5ForConditionalGeneration
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto")
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
### Running the model on a GPU
<details>
<summary> Click to expand </summary>
```python
# pip install accelerate
import torch
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto")
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids, sample=True, temperature=1.0, top_p=0.9, max_new_tokens=256, repetition_penalty=1.03)
print(tokenizer.decode(outputs[0]))
```
</details>
### Running the model on a GPU using different precisions
#### FP16
<details>
<summary> Click to expand </summary>
```python
# pip install accelerate
import torch
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto", torch_dtype=torch.float16)
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
#### INT8
<details>
<summary> Click to expand </summary>
```python
# pip install bitsandbytes accelerate
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto", load_in_8bit=True)
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
# Citation
If you find the following model helpful, please consider citing our paper!
**BibTeX:**
```bibtex
@misc{kim2023prometheus,
title={Prometheus: Inducing Fine-grained Evaluation Capability in Language Models},
author={Seungone Kim and Jamin Shin and Yejin Cho and Joel Jang and Shayne Longpre and Hwaran Lee and Sangdoo Yun and Seongjin Shin and Sungdong Kim and James Thorne and Minjoon Seo},
year={2023},
eprint={2310.08491},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
```
|
snkai2004/PPO-lander-LunarLander-v2 | snkai2004 | 2023-11-10T18:42:49Z | 0 | 0 | stable-baselines3 | [
"stable-baselines3",
"LunarLander-v2",
"deep-reinforcement-learning",
"reinforcement-learning",
"model-index",
"region:us"
]
| reinforcement-learning | 2023-11-10T18:42:20Z | ---
library_name: stable-baselines3
tags:
- LunarLander-v2
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
model-index:
- name: PPO
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: LunarLander-v2
type: LunarLander-v2
metrics:
- type: mean_reward
value: 258.50 +/- 17.47
name: mean_reward
verified: false
---
# **PPO** Agent playing **LunarLander-v2**
This is a trained model of a **PPO** agent playing **LunarLander-v2**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3).
## Usage (with Stable-baselines3)
TODO: Add your code
```python
from stable_baselines3 import ...
from huggingface_sb3 import load_from_hub
...
```
|
srimathis/ppo-LunarLander-v2 | srimathis | 2023-11-10T18:40:20Z | 0 | 0 | stable-baselines3 | [
"stable-baselines3",
"LunarLander-v2",
"deep-reinforcement-learning",
"reinforcement-learning",
"model-index",
"region:us"
]
| reinforcement-learning | 2023-11-10T18:39:58Z | ---
library_name: stable-baselines3
tags:
- LunarLander-v2
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
model-index:
- name: PPO
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: LunarLander-v2
type: LunarLander-v2
metrics:
- type: mean_reward
value: 247.82 +/- 16.21
name: mean_reward
verified: false
---
# **PPO** Agent playing **LunarLander-v2**
This is a trained model of a **PPO** agent playing **LunarLander-v2**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3).
## Usage (with Stable-baselines3)
TODO: Add your code
```python
from stable_baselines3 import ...
from huggingface_sb3 import load_from_hub
...
```
|
asas-ai/acegpt_7B_4bit_qlora_flores | asas-ai | 2023-11-10T18:38:44Z | 0 | 0 | null | [
"tensorboard",
"safetensors",
"generated_from_trainer",
"base_model:asas-ai/AceGPT-7B_8bit",
"base_model:finetune:asas-ai/AceGPT-7B_8bit",
"region:us"
]
| null | 2023-11-10T18:38:17Z | ---
base_model: asas-ai/AceGPT-7B_8bit
tags:
- generated_from_trainer
model-index:
- name: acegpt_7B_4bit_qlora_flores
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# acegpt_7B_4bit_qlora_flores
This model is a fine-tuned version of [asas-ai/AceGPT-7B_8bit](https://huggingface.co/asas-ai/AceGPT-7B_8bit) on an unknown dataset.
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0002
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant
- lr_scheduler_warmup_ratio: 0.03
- training_steps: 2200
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu121
- Datasets 2.4.0
- Tokenizers 0.14.1
|
SaiedAlshahrani/acegpt_7B_4bit_qlora_flores | SaiedAlshahrani | 2023-11-10T18:38:19Z | 0 | 0 | null | [
"tensorboard",
"safetensors",
"generated_from_trainer",
"base_model:asas-ai/AceGPT-7B_8bit",
"base_model:finetune:asas-ai/AceGPT-7B_8bit",
"region:us"
]
| null | 2023-11-10T16:23:48Z | ---
base_model: asas-ai/AceGPT-7B_8bit
tags:
- generated_from_trainer
model-index:
- name: acegpt_7B_4bit_qlora_flores
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# acegpt_7B_4bit_qlora_flores
This model is a fine-tuned version of [asas-ai/AceGPT-7B_8bit](https://huggingface.co/asas-ai/AceGPT-7B_8bit) on an unknown dataset.
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0002
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant
- lr_scheduler_warmup_ratio: 0.03
- training_steps: 2200
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu121
- Datasets 2.4.0
- Tokenizers 0.14.1
|
Fsg-15/peft | Fsg-15 | 2023-11-10T18:36:19Z | 0 | 0 | peft | [
"peft",
"safetensors",
"arxiv:1910.09700",
"base_model:bigscience/bloom-7b1",
"base_model:adapter:bigscience/bloom-7b1",
"region:us"
]
| null | 2023-11-10T16:01:35Z | ---
library_name: peft
base_model: bigscience/bloom-7b1
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
- **Developed by:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[More Information Needed]
## Bias, Risks, and Limitations
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
[More Information Needed]
### Recommendations
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Data Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Data Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
**BibTeX:**
[More Information Needed]
**APA:**
[More Information Needed]
## Glossary [optional]
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: True
- load_in_4bit: False
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: fp4
- bnb_4bit_use_double_quant: False
- bnb_4bit_compute_dtype: float32
### Framework versions
- PEFT 0.6.2.dev0
|
hkivancoral/hushem_1x_deit_tiny_rms_lr001_fold5 | hkivancoral | 2023-11-10T18:34:29Z | 7 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:facebook/deit-tiny-patch16-224",
"base_model:finetune:facebook/deit-tiny-patch16-224",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| image-classification | 2023-11-10T18:31:53Z | ---
license: apache-2.0
base_model: facebook/deit-tiny-patch16-224
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: hushem_1x_deit_tiny_rms_lr001_fold5
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: test
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.5365853658536586
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# hushem_1x_deit_tiny_rms_lr001_fold5
This model is a fine-tuned version of [facebook/deit-tiny-patch16-224](https://huggingface.co/facebook/deit-tiny-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 1.0599
- Accuracy: 0.5366
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.001
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 1.0 | 6 | 2.6067 | 0.2439 |
| 4.0909 | 2.0 | 12 | 1.8085 | 0.2439 |
| 4.0909 | 3.0 | 18 | 1.7809 | 0.2439 |
| 2.0948 | 4.0 | 24 | 1.7586 | 0.2439 |
| 1.6719 | 5.0 | 30 | 1.5135 | 0.2439 |
| 1.6719 | 6.0 | 36 | 1.7849 | 0.2683 |
| 1.5694 | 7.0 | 42 | 1.4636 | 0.3902 |
| 1.5694 | 8.0 | 48 | 1.4809 | 0.2683 |
| 1.519 | 9.0 | 54 | 1.3587 | 0.3415 |
| 1.5241 | 10.0 | 60 | 1.3823 | 0.2439 |
| 1.5241 | 11.0 | 66 | 1.3645 | 0.3415 |
| 1.4557 | 12.0 | 72 | 1.2525 | 0.3659 |
| 1.4557 | 13.0 | 78 | 1.2955 | 0.3171 |
| 1.3674 | 14.0 | 84 | 1.3174 | 0.3415 |
| 1.3868 | 15.0 | 90 | 1.2787 | 0.3415 |
| 1.3868 | 16.0 | 96 | 1.6408 | 0.2683 |
| 1.3152 | 17.0 | 102 | 1.2750 | 0.3171 |
| 1.3152 | 18.0 | 108 | 1.0560 | 0.5366 |
| 1.2693 | 19.0 | 114 | 1.3256 | 0.4878 |
| 1.2554 | 20.0 | 120 | 1.3190 | 0.3902 |
| 1.2554 | 21.0 | 126 | 1.2498 | 0.3902 |
| 1.1813 | 22.0 | 132 | 1.2514 | 0.3902 |
| 1.1813 | 23.0 | 138 | 1.0907 | 0.5366 |
| 1.1113 | 24.0 | 144 | 1.2821 | 0.3415 |
| 1.1728 | 25.0 | 150 | 1.1433 | 0.4878 |
| 1.1728 | 26.0 | 156 | 1.0143 | 0.5366 |
| 1.1037 | 27.0 | 162 | 0.9542 | 0.5854 |
| 1.1037 | 28.0 | 168 | 1.1443 | 0.5122 |
| 1.0914 | 29.0 | 174 | 1.0904 | 0.4878 |
| 1.1385 | 30.0 | 180 | 1.1995 | 0.4146 |
| 1.1385 | 31.0 | 186 | 0.9746 | 0.6098 |
| 1.0636 | 32.0 | 192 | 1.1104 | 0.4634 |
| 1.0636 | 33.0 | 198 | 0.9890 | 0.6098 |
| 1.0129 | 34.0 | 204 | 1.2113 | 0.3902 |
| 0.999 | 35.0 | 210 | 1.0001 | 0.6098 |
| 0.999 | 36.0 | 216 | 1.0972 | 0.5122 |
| 0.9802 | 37.0 | 222 | 1.1639 | 0.4390 |
| 0.9802 | 38.0 | 228 | 1.0730 | 0.5122 |
| 0.9625 | 39.0 | 234 | 1.0471 | 0.4878 |
| 0.9424 | 40.0 | 240 | 1.0692 | 0.5366 |
| 0.9424 | 41.0 | 246 | 1.0654 | 0.5366 |
| 0.9521 | 42.0 | 252 | 1.0599 | 0.5366 |
| 0.9521 | 43.0 | 258 | 1.0599 | 0.5366 |
| 0.9184 | 44.0 | 264 | 1.0599 | 0.5366 |
| 0.9335 | 45.0 | 270 | 1.0599 | 0.5366 |
| 0.9335 | 46.0 | 276 | 1.0599 | 0.5366 |
| 0.9251 | 47.0 | 282 | 1.0599 | 0.5366 |
| 0.9251 | 48.0 | 288 | 1.0599 | 0.5366 |
| 0.9168 | 49.0 | 294 | 1.0599 | 0.5366 |
| 0.8964 | 50.0 | 300 | 1.0599 | 0.5366 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
efederici/ipt-350m | efederici | 2023-11-10T18:33:53Z | 16 | 0 | transformers | [
"transformers",
"pytorch",
"mpt",
"text-generation",
"ipt",
"alibi",
"text-generation-inference",
"text generation",
"custom_code",
"it",
"dataset:oscar-corpus/OSCAR-2301",
"arxiv:2108.12409",
"arxiv:2205.14135",
"arxiv:1909.08053",
"license:apache-2.0",
"autotrain_compatible",
"region:us"
]
| text-generation | 2023-06-06T20:57:19Z | ---
license: apache-2.0
tags:
- ipt
- alibi
- text-generation-inference
- text generation
inference: false
datasets:
- oscar-corpus/OSCAR-2301
language:
- it
pipeline_tag: text-generation
---
# ipt-350m
ipt-350m is a decoder-style transformer pretrained from scratch on ~13B tokens of Italian text (wip: trained on unfiltered oscar).
It uses a modified transformer architecture optimized for efficient training and inference. Positional embeddings are replaced with Attention with Linear Biases ([ALiBi](https://arxiv.org/abs/2108.12409)).
ipt-350m is:
- **Licensed for the possibility of commercial use**
- **Prepared to handle extremely long inputs** thanks to [ALiBi](https://arxiv.org/abs/2108.12409).
- **Capable of fast training and inference** (via [FlashAttention](https://arxiv.org/pdf/2205.14135.pdf) and [FasterTransformer](https://github.com/NVIDIA/FasterTransformer))
- **Equipped with highly efficient open-source training code** via the [llm-foundry repository](https://github.com/mosaicml/llm-foundry)
If you find this project useful, consider supporting its development:
[](https://bmc.link/edoardofederici)
## How to Use
```python
import transformers
model = transformers.AutoModelForCausalLM.from_pretrained(
'efederici/ipt-350m',
trust_remote_code=True
)
```
Note: This model requires that `trust_remote_code=True` be passed to the `from_pretrained` method.
To use the optimized [triton implementation](https://github.com/openai/triton) of FlashAttention, you can load the model on GPU (`cuda:0`) with `attn_impl='triton'` and with `bfloat16` precision:
```python
import torch
import transformers
name = 'efederici/ipt-350m'
config = transformers.AutoConfig.from_pretrained(name, trust_remote_code=True)
config.attn_config['attn_impl'] = 'triton'
config.init_device = 'cuda:0'
model = transformers.AutoModelForCausalLM.from_pretrained(
name,
config=config,
torch_dtype=torch.bfloat16,
trust_remote_code=True
)
```
Although the model was trained with a sequence length of 2048, ALiBi enables to increase the maximum sequence length during finetuning and/or inference.
```python
import transformers
name = 'efederici/ipt-350m'
config = transformers.AutoConfig.from_pretrained(name, trust_remote_code=True)
config.max_seq_len = 4096 # (input + output) tokens can now be up to 4096
model = transformers.AutoModelForCausalLM.from_pretrained(
name,
config=config,
trust_remote_code=True
)
```
## Model Description
The architecture is a modification of a standard decoder-only transformer.
The model has been modified from a standard transformer in the following ways:
- It uses [FlashAttention](https://arxiv.org/pdf/2205.14135.pdf)
- It uses [ALiBi (Attention with Linear Biases)](https://arxiv.org/abs/2108.12409) and does not use positional embeddings
- It does not use biases
| Hyperparameter | Value |
|----------------|-------|
|n_parameters | 350M |
|n_layers | 24 |
| n_heads | 16 |
| d_model | 1024 |
| vocab size | 50432 |
| sequence length | 2048 |
### Dataset
The model was trained for ~13B tokens (with batch size 64 and sequence length 2048) on [OSCAR-2301](https://huggingface.co/datasets/oscar-corpus/OSCAR-2301).
Each example was constructed from as many sequences from that dataset as were necessary to fill the 2048 sequence length.
Vocabulary size is 50432, a multiple of 128 as suggested in [MEGATRON-LM](https://arxiv.org/abs/1909.08053), model flop utilization (MFU) increased by up to four percentage points. |
hkivancoral/hushem_1x_deit_tiny_rms_lr001_fold4 | hkivancoral | 2023-11-10T18:30:18Z | 6 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:facebook/deit-tiny-patch16-224",
"base_model:finetune:facebook/deit-tiny-patch16-224",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| image-classification | 2023-11-10T18:27:42Z | ---
license: apache-2.0
base_model: facebook/deit-tiny-patch16-224
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: hushem_1x_deit_tiny_rms_lr001_fold4
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: test
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.4523809523809524
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# hushem_1x_deit_tiny_rms_lr001_fold4
This model is a fine-tuned version of [facebook/deit-tiny-patch16-224](https://huggingface.co/facebook/deit-tiny-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 1.1726
- Accuracy: 0.4524
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.001
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 1.0 | 6 | 5.8408 | 0.2619 |
| 4.7138 | 2.0 | 12 | 1.8632 | 0.2381 |
| 4.7138 | 3.0 | 18 | 1.9369 | 0.2619 |
| 1.8439 | 4.0 | 24 | 1.7584 | 0.2381 |
| 1.6449 | 5.0 | 30 | 1.4723 | 0.2619 |
| 1.6449 | 6.0 | 36 | 1.7187 | 0.2381 |
| 1.5171 | 7.0 | 42 | 1.4960 | 0.2381 |
| 1.5171 | 8.0 | 48 | 1.3962 | 0.2619 |
| 1.4701 | 9.0 | 54 | 1.4942 | 0.2619 |
| 1.4652 | 10.0 | 60 | 1.3642 | 0.2381 |
| 1.4652 | 11.0 | 66 | 1.4490 | 0.2619 |
| 1.4547 | 12.0 | 72 | 1.1912 | 0.4524 |
| 1.4547 | 13.0 | 78 | 1.4737 | 0.2857 |
| 1.3944 | 14.0 | 84 | 1.2170 | 0.4286 |
| 1.3536 | 15.0 | 90 | 1.3540 | 0.2381 |
| 1.3536 | 16.0 | 96 | 1.0819 | 0.6190 |
| 1.2835 | 17.0 | 102 | 1.1640 | 0.4286 |
| 1.2835 | 18.0 | 108 | 1.2309 | 0.3333 |
| 1.306 | 19.0 | 114 | 1.3288 | 0.2857 |
| 1.2522 | 20.0 | 120 | 1.4561 | 0.2857 |
| 1.2522 | 21.0 | 126 | 1.0774 | 0.4762 |
| 1.2491 | 22.0 | 132 | 1.1807 | 0.4286 |
| 1.2491 | 23.0 | 138 | 1.1668 | 0.3810 |
| 1.1882 | 24.0 | 144 | 1.2075 | 0.4286 |
| 1.2028 | 25.0 | 150 | 1.2635 | 0.3333 |
| 1.2028 | 26.0 | 156 | 1.1653 | 0.3810 |
| 1.1822 | 27.0 | 162 | 1.1741 | 0.4048 |
| 1.1822 | 28.0 | 168 | 1.4014 | 0.2619 |
| 1.1086 | 29.0 | 174 | 1.0259 | 0.5476 |
| 1.1111 | 30.0 | 180 | 1.1225 | 0.5238 |
| 1.1111 | 31.0 | 186 | 1.1813 | 0.5 |
| 1.0458 | 32.0 | 192 | 1.1678 | 0.4286 |
| 1.0458 | 33.0 | 198 | 1.1915 | 0.4048 |
| 1.1348 | 34.0 | 204 | 1.3148 | 0.5 |
| 0.9776 | 35.0 | 210 | 1.0082 | 0.5238 |
| 0.9776 | 36.0 | 216 | 0.9144 | 0.6190 |
| 0.9456 | 37.0 | 222 | 1.0677 | 0.4762 |
| 0.9456 | 38.0 | 228 | 1.0695 | 0.5238 |
| 0.8714 | 39.0 | 234 | 1.1982 | 0.4762 |
| 0.8643 | 40.0 | 240 | 1.1143 | 0.4048 |
| 0.8643 | 41.0 | 246 | 1.1270 | 0.4524 |
| 0.7971 | 42.0 | 252 | 1.1726 | 0.4524 |
| 0.7971 | 43.0 | 258 | 1.1726 | 0.4524 |
| 0.7662 | 44.0 | 264 | 1.1726 | 0.4524 |
| 0.7801 | 45.0 | 270 | 1.1726 | 0.4524 |
| 0.7801 | 46.0 | 276 | 1.1726 | 0.4524 |
| 0.7773 | 47.0 | 282 | 1.1726 | 0.4524 |
| 0.7773 | 48.0 | 288 | 1.1726 | 0.4524 |
| 0.7728 | 49.0 | 294 | 1.1726 | 0.4524 |
| 0.7828 | 50.0 | 300 | 1.1726 | 0.4524 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
hkivancoral/hushem_1x_deit_tiny_rms_lr001_fold2 | hkivancoral | 2023-11-10T18:24:11Z | 6 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:facebook/deit-tiny-patch16-224",
"base_model:finetune:facebook/deit-tiny-patch16-224",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| image-classification | 2023-11-10T18:21:33Z | ---
license: apache-2.0
base_model: facebook/deit-tiny-patch16-224
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: hushem_1x_deit_tiny_rms_lr001_fold2
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: test
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.35555555555555557
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# hushem_1x_deit_tiny_rms_lr001_fold2
This model is a fine-tuned version of [facebook/deit-tiny-patch16-224](https://huggingface.co/facebook/deit-tiny-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 1.2600
- Accuracy: 0.3556
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.001
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 1.0 | 6 | 2.0874 | 0.2444 |
| 4.6196 | 2.0 | 12 | 2.3422 | 0.2444 |
| 4.6196 | 3.0 | 18 | 1.7914 | 0.2444 |
| 1.8086 | 4.0 | 24 | 1.6082 | 0.2667 |
| 1.5901 | 5.0 | 30 | 1.5144 | 0.2444 |
| 1.5901 | 6.0 | 36 | 1.6190 | 0.2444 |
| 1.5211 | 7.0 | 42 | 1.5231 | 0.2444 |
| 1.5211 | 8.0 | 48 | 1.5027 | 0.2444 |
| 1.4477 | 9.0 | 54 | 1.4266 | 0.2444 |
| 1.4394 | 10.0 | 60 | 1.4345 | 0.2444 |
| 1.4394 | 11.0 | 66 | 1.3152 | 0.4444 |
| 1.3604 | 12.0 | 72 | 1.3376 | 0.2444 |
| 1.3604 | 13.0 | 78 | 1.3260 | 0.2667 |
| 1.3864 | 14.0 | 84 | 1.5120 | 0.2444 |
| 1.3555 | 15.0 | 90 | 1.2685 | 0.3556 |
| 1.3555 | 16.0 | 96 | 1.1751 | 0.4444 |
| 1.3501 | 17.0 | 102 | 1.2648 | 0.4444 |
| 1.3501 | 18.0 | 108 | 1.3992 | 0.3778 |
| 1.2496 | 19.0 | 114 | 1.4208 | 0.2889 |
| 1.2587 | 20.0 | 120 | 1.1782 | 0.4444 |
| 1.2587 | 21.0 | 126 | 1.2882 | 0.4444 |
| 1.2321 | 22.0 | 132 | 1.3142 | 0.4444 |
| 1.2321 | 23.0 | 138 | 1.1784 | 0.4222 |
| 1.1985 | 24.0 | 144 | 1.2247 | 0.3778 |
| 1.234 | 25.0 | 150 | 1.2329 | 0.3778 |
| 1.234 | 26.0 | 156 | 1.2482 | 0.3556 |
| 1.1913 | 27.0 | 162 | 1.4153 | 0.3111 |
| 1.1913 | 28.0 | 168 | 1.2994 | 0.3333 |
| 1.1911 | 29.0 | 174 | 1.1400 | 0.4667 |
| 1.1955 | 30.0 | 180 | 1.2156 | 0.3778 |
| 1.1955 | 31.0 | 186 | 1.2232 | 0.4 |
| 1.1751 | 32.0 | 192 | 1.3853 | 0.2889 |
| 1.1751 | 33.0 | 198 | 1.2309 | 0.3333 |
| 1.1328 | 34.0 | 204 | 1.2338 | 0.3333 |
| 1.195 | 35.0 | 210 | 1.2383 | 0.3333 |
| 1.195 | 36.0 | 216 | 1.2991 | 0.3778 |
| 1.1661 | 37.0 | 222 | 1.3228 | 0.3556 |
| 1.1661 | 38.0 | 228 | 1.2550 | 0.3778 |
| 1.0748 | 39.0 | 234 | 1.2591 | 0.3556 |
| 1.1122 | 40.0 | 240 | 1.2234 | 0.3778 |
| 1.1122 | 41.0 | 246 | 1.2608 | 0.3556 |
| 1.102 | 42.0 | 252 | 1.2600 | 0.3556 |
| 1.102 | 43.0 | 258 | 1.2600 | 0.3556 |
| 1.0792 | 44.0 | 264 | 1.2600 | 0.3556 |
| 1.1126 | 45.0 | 270 | 1.2600 | 0.3556 |
| 1.1126 | 46.0 | 276 | 1.2600 | 0.3556 |
| 1.0704 | 47.0 | 282 | 1.2600 | 0.3556 |
| 1.0704 | 48.0 | 288 | 1.2600 | 0.3556 |
| 1.1302 | 49.0 | 294 | 1.2600 | 0.3556 |
| 1.0797 | 50.0 | 300 | 1.2600 | 0.3556 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
TheBloke/prometheus-13B-v1.0-GGUF | TheBloke | 2023-11-10T18:16:12Z | 78 | 1 | transformers | [
"transformers",
"gguf",
"llama",
"text2text-generation",
"en",
"dataset:kaist-ai/Feedback-Collection",
"arxiv:2310.08491",
"base_model:prometheus-eval/prometheus-13b-v1.0",
"base_model:quantized:prometheus-eval/prometheus-13b-v1.0",
"license:apache-2.0",
"region:us"
]
| text2text-generation | 2023-11-10T18:08:26Z | ---
base_model: kaist-ai/prometheus-13b-v1.0
datasets:
- kaist-ai/Feedback-Collection
inference: false
language:
- en
library_name: transformers
license: apache-2.0
metrics:
- pearsonr
- spearmanr
- accuracy
model_creator: KAIST AI
model_name: Prometheus 13B V1.0
model_type: llama
pipeline_tag: text2text-generation
prompt_template: '###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference
answer that gets a score of 5, and a score rubric representing a evaluation criteria
are given.
1. Write a detailed feedback that assess the quality of the response strictly based
on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You
should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria)
[RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{prompt}
###Response to evaluate:
{{response}}
###Reference Answer (Score 5):
{{reference_answer}}
###Score Rubrics:
[{{criteria_description}}]
Score 1: {{score1_description}}
Score 2: {{score2_description}}
Score 3: {{score3_description}}
Score 4: {{score4_description}}
Score 5: {{score5_description}}
###Feedback:
'
quantized_by: TheBloke
tags:
- text2text-generation
---
<!-- markdownlint-disable MD041 -->
<!-- header start -->
<!-- 200823 -->
<div style="width: auto; margin-left: auto; margin-right: auto">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<div style="text-align:center; margin-top: 0em; margin-bottom: 0em"><p style="margin-top: 0.25em; margin-bottom: 0em;">TheBloke's LLM work is generously supported by a grant from <a href="https://a16z.com">andreessen horowitz (a16z)</a></p></div>
<hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
<!-- header end -->
# Prometheus 13B V1.0 - GGUF
- Model creator: [KAIST AI](https://huggingface.co/kaist-ai)
- Original model: [Prometheus 13B V1.0](https://huggingface.co/kaist-ai/prometheus-13b-v1.0)
<!-- description start -->
## Description
This repo contains GGUF format model files for [KAIST AI's Prometheus 13B V1.0](https://huggingface.co/kaist-ai/prometheus-13b-v1.0).
These files were quantised using hardware kindly provided by [Massed Compute](https://massedcompute.com/).
<!-- description end -->
<!-- README_GGUF.md-about-gguf start -->
### About GGUF
GGUF is a new format introduced by the llama.cpp team on August 21st 2023. It is a replacement for GGML, which is no longer supported by llama.cpp.
Here is an incomplete list of clients and libraries that are known to support GGUF:
* [llama.cpp](https://github.com/ggerganov/llama.cpp). The source project for GGUF. Offers a CLI and a server option.
* [text-generation-webui](https://github.com/oobabooga/text-generation-webui), the most widely used web UI, with many features and powerful extensions. Supports GPU acceleration.
* [KoboldCpp](https://github.com/LostRuins/koboldcpp), a fully featured web UI, with GPU accel across all platforms and GPU architectures. Especially good for story telling.
* [LM Studio](https://lmstudio.ai/), an easy-to-use and powerful local GUI for Windows and macOS (Silicon), with GPU acceleration.
* [LoLLMS Web UI](https://github.com/ParisNeo/lollms-webui), a great web UI with many interesting and unique features, including a full model library for easy model selection.
* [Faraday.dev](https://faraday.dev/), an attractive and easy to use character-based chat GUI for Windows and macOS (both Silicon and Intel), with GPU acceleration.
* [ctransformers](https://github.com/marella/ctransformers), a Python library with GPU accel, LangChain support, and OpenAI-compatible AI server.
* [llama-cpp-python](https://github.com/abetlen/llama-cpp-python), a Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
* [candle](https://github.com/huggingface/candle), a Rust ML framework with a focus on performance, including GPU support, and ease of use.
<!-- README_GGUF.md-about-gguf end -->
<!-- repositories-available start -->
## Repositories available
* [AWQ model(s) for GPU inference.](https://huggingface.co/TheBloke/prometheus-13B-v1.0-AWQ)
* [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GGUF)
* [KAIST AI's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/kaist-ai/prometheus-13b-v1.0)
<!-- repositories-available end -->
<!-- prompt-template start -->
## Prompt template: Prometheus
```
###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{prompt}
###Response to evaluate:
{{response}}
###Reference Answer (Score 5):
{{reference_answer}}
###Score Rubrics:
[{{criteria_description}}]
Score 1: {{score1_description}}
Score 2: {{score2_description}}
Score 3: {{score3_description}}
Score 4: {{score4_description}}
Score 5: {{score5_description}}
###Feedback:
```
<!-- prompt-template end -->
<!-- licensing start -->
## Licensing
The creator of the source model has listed its license as `apache-2.0`, and this quantization has therefore used that same license.
As this model is based on Llama 2, it is also subject to the Meta Llama 2 license terms, and the license files for that are additionally included. It should therefore be considered as being claimed to be licensed under both licenses. I contacted Hugging Face for clarification on dual licensing but they do not yet have an official position. Should this change, or should Meta provide any feedback on this situation, I will update this section accordingly.
In the meantime, any questions regarding licensing, and in particular how these two licenses might interact, should be directed to the original model repository: [KAIST AI's Prometheus 13B V1.0](https://huggingface.co/kaist-ai/prometheus-13b-v1.0).
<!-- licensing end -->
<!-- compatibility_gguf start -->
## Compatibility
These quantised GGUFv2 files are compatible with llama.cpp from August 27th onwards, as of commit [d0cee0d](https://github.com/ggerganov/llama.cpp/commit/d0cee0d36d5be95a0d9088b674dbb27354107221)
They are also compatible with many third party UIs and libraries - please see the list at the top of this README.
## Explanation of quantisation methods
<details>
<summary>Click to see details</summary>
The new methods available are:
* GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
* GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
* GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
* GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
* GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
Refer to the Provided Files table below to see what files use which methods, and how.
</details>
<!-- compatibility_gguf end -->
<!-- README_GGUF.md-provided-files start -->
## Provided files
| Name | Quant method | Bits | Size | Max RAM required | Use case |
| ---- | ---- | ---- | ---- | ---- | ----- |
| [prometheus-13b-v1.0.Q2_K.gguf](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GGUF/blob/main/prometheus-13b-v1.0.Q2_K.gguf) | Q2_K | 2 | 5.43 GB| 7.93 GB | smallest, significant quality loss - not recommended for most purposes |
| [prometheus-13b-v1.0.Q3_K_S.gguf](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GGUF/blob/main/prometheus-13b-v1.0.Q3_K_S.gguf) | Q3_K_S | 3 | 5.66 GB| 8.16 GB | very small, high quality loss |
| [prometheus-13b-v1.0.Q3_K_M.gguf](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GGUF/blob/main/prometheus-13b-v1.0.Q3_K_M.gguf) | Q3_K_M | 3 | 6.34 GB| 8.84 GB | very small, high quality loss |
| [prometheus-13b-v1.0.Q3_K_L.gguf](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GGUF/blob/main/prometheus-13b-v1.0.Q3_K_L.gguf) | Q3_K_L | 3 | 6.93 GB| 9.43 GB | small, substantial quality loss |
| [prometheus-13b-v1.0.Q4_0.gguf](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GGUF/blob/main/prometheus-13b-v1.0.Q4_0.gguf) | Q4_0 | 4 | 7.37 GB| 9.87 GB | legacy; small, very high quality loss - prefer using Q3_K_M |
| [prometheus-13b-v1.0.Q4_K_S.gguf](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GGUF/blob/main/prometheus-13b-v1.0.Q4_K_S.gguf) | Q4_K_S | 4 | 7.41 GB| 9.91 GB | small, greater quality loss |
| [prometheus-13b-v1.0.Q4_K_M.gguf](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GGUF/blob/main/prometheus-13b-v1.0.Q4_K_M.gguf) | Q4_K_M | 4 | 7.87 GB| 10.37 GB | medium, balanced quality - recommended |
| [prometheus-13b-v1.0.Q5_0.gguf](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GGUF/blob/main/prometheus-13b-v1.0.Q5_0.gguf) | Q5_0 | 5 | 8.97 GB| 11.47 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
| [prometheus-13b-v1.0.Q5_K_S.gguf](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GGUF/blob/main/prometheus-13b-v1.0.Q5_K_S.gguf) | Q5_K_S | 5 | 8.97 GB| 11.47 GB | large, low quality loss - recommended |
| [prometheus-13b-v1.0.Q5_K_M.gguf](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GGUF/blob/main/prometheus-13b-v1.0.Q5_K_M.gguf) | Q5_K_M | 5 | 9.23 GB| 11.73 GB | large, very low quality loss - recommended |
| [prometheus-13b-v1.0.Q6_K.gguf](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GGUF/blob/main/prometheus-13b-v1.0.Q6_K.gguf) | Q6_K | 6 | 10.68 GB| 13.18 GB | very large, extremely low quality loss |
| [prometheus-13b-v1.0.Q8_0.gguf](https://huggingface.co/TheBloke/prometheus-13B-v1.0-GGUF/blob/main/prometheus-13b-v1.0.Q8_0.gguf) | Q8_0 | 8 | 13.83 GB| 16.33 GB | very large, extremely low quality loss - not recommended |
**Note**: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.
<!-- README_GGUF.md-provided-files end -->
<!-- README_GGUF.md-how-to-download start -->
## How to download GGUF files
**Note for manual downloaders:** You almost never want to clone the entire repo! Multiple different quantisation formats are provided, and most users only want to pick and download a single file.
The following clients/libraries will automatically download models for you, providing a list of available models to choose from:
* LM Studio
* LoLLMS Web UI
* Faraday.dev
### In `text-generation-webui`
Under Download Model, you can enter the model repo: TheBloke/prometheus-13B-v1.0-GGUF and below it, a specific filename to download, such as: prometheus-13b-v1.0.Q4_K_M.gguf.
Then click Download.
### On the command line, including multiple files at once
I recommend using the `huggingface-hub` Python library:
```shell
pip3 install huggingface-hub
```
Then you can download any individual model file to the current directory, at high speed, with a command like this:
```shell
huggingface-cli download TheBloke/prometheus-13B-v1.0-GGUF prometheus-13b-v1.0.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
```
<details>
<summary>More advanced huggingface-cli download usage</summary>
You can also download multiple files at once with a pattern:
```shell
huggingface-cli download TheBloke/prometheus-13B-v1.0-GGUF --local-dir . --local-dir-use-symlinks False --include='*Q4_K*gguf'
```
For more documentation on downloading with `huggingface-cli`, please see: [HF -> Hub Python Library -> Download files -> Download from the CLI](https://huggingface.co/docs/huggingface_hub/guides/download#download-from-the-cli).
To accelerate downloads on fast connections (1Gbit/s or higher), install `hf_transfer`:
```shell
pip3 install hf_transfer
```
And set environment variable `HF_HUB_ENABLE_HF_TRANSFER` to `1`:
```shell
HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download TheBloke/prometheus-13B-v1.0-GGUF prometheus-13b-v1.0.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
```
Windows Command Line users: You can set the environment variable by running `set HF_HUB_ENABLE_HF_TRANSFER=1` before the download command.
</details>
<!-- README_GGUF.md-how-to-download end -->
<!-- README_GGUF.md-how-to-run start -->
## Example `llama.cpp` command
Make sure you are using `llama.cpp` from commit [d0cee0d](https://github.com/ggerganov/llama.cpp/commit/d0cee0d36d5be95a0d9088b674dbb27354107221) or later.
```shell
./main -ngl 32 -m prometheus-13b-v1.0.Q4_K_M.gguf --color -c 4096 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "###Task Description:\nAn instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.\n1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.\n2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.\n3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"\n4. Please do not generate any other opening, closing, and explanations.\n\n###The instruction to evaluate:\n{prompt}\n\n###Response to evaluate:\n{{response}}\n\n###Reference Answer (Score 5):\n{{reference_answer}}\n\n###Score Rubrics:\n[{{criteria_description}}]\nScore 1: {{score1_description}}\nScore 2: {{score2_description}}\nScore 3: {{score3_description}}\nScore 4: {{score4_description}}\nScore 5: {{score5_description}}\n\n###Feedback:"
```
Change `-ngl 32` to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.
Change `-c 4096` to the desired sequence length. For extended sequence models - eg 8K, 16K, 32K - the necessary RoPE scaling parameters are read from the GGUF file and set by llama.cpp automatically.
If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`
For other parameters and how to use them, please refer to [the llama.cpp documentation](https://github.com/ggerganov/llama.cpp/blob/master/examples/main/README.md)
## How to run in `text-generation-webui`
Further instructions here: [text-generation-webui/docs/llama.cpp.md](https://github.com/oobabooga/text-generation-webui/blob/main/docs/llama.cpp.md).
## How to run from Python code
You can use GGUF models from Python using the [llama-cpp-python](https://github.com/abetlen/llama-cpp-python) or [ctransformers](https://github.com/marella/ctransformers) libraries.
### How to load this model in Python code, using ctransformers
#### First install the package
Run one of the following commands, according to your system:
```shell
# Base ctransformers with no GPU acceleration
pip install ctransformers
# Or with CUDA GPU acceleration
pip install ctransformers[cuda]
# Or with AMD ROCm GPU acceleration (Linux only)
CT_HIPBLAS=1 pip install ctransformers --no-binary ctransformers
# Or with Metal GPU acceleration for macOS systems only
CT_METAL=1 pip install ctransformers --no-binary ctransformers
```
#### Simple ctransformers example code
```python
from ctransformers import AutoModelForCausalLM
# Set gpu_layers to the number of layers to offload to GPU. Set to 0 if no GPU acceleration is available on your system.
llm = AutoModelForCausalLM.from_pretrained("TheBloke/prometheus-13B-v1.0-GGUF", model_file="prometheus-13b-v1.0.Q4_K_M.gguf", model_type="llama", gpu_layers=50)
print(llm("AI is going to"))
```
## How to use with LangChain
Here are guides on using llama-cpp-python and ctransformers with LangChain:
* [LangChain + llama-cpp-python](https://python.langchain.com/docs/integrations/llms/llamacpp)
* [LangChain + ctransformers](https://python.langchain.com/docs/integrations/providers/ctransformers)
<!-- README_GGUF.md-how-to-run end -->
<!-- footer start -->
<!-- 200823 -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
## Thanks, and how to contribute
Thanks to the [chirper.ai](https://chirper.ai) team!
Thanks to Clay from [gpus.llm-utils.org](llm-utils)!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Aemon Algiz.
**Patreon special mentions**: Brandon Frisco, LangChain4j, Spiking Neurons AB, transmissions 11, Joseph William Delisle, Nitin Borwankar, Willem Michiel, Michael Dempsey, vamX, Jeffrey Morgan, zynix, jjj, Omer Bin Jawed, Sean Connelly, jinyuan sun, Jeromy Smith, Shadi, Pawan Osman, Chadd, Elijah Stavena, Illia Dulskyi, Sebastain Graf, Stephen Murray, terasurfer, Edmond Seymore, Celu Ramasamy, Mandus, Alex, biorpg, Ajan Kanaga, Clay Pascal, Raven Klaugh, 阿明, K, ya boyyy, usrbinkat, Alicia Loh, John Villwock, ReadyPlayerEmma, Chris Smitley, Cap'n Zoog, fincy, GodLy, S_X, sidney chen, Cory Kujawski, OG, Mano Prime, AzureBlack, Pieter, Kalila, Spencer Kim, Tom X Nguyen, Stanislav Ovsiannikov, Michael Levine, Andrey, Trailburnt, Vadim, Enrico Ros, Talal Aujan, Brandon Phillips, Jack West, Eugene Pentland, Michael Davis, Will Dee, webtim, Jonathan Leane, Alps Aficionado, Rooh Singh, Tiffany J. Kim, theTransient, Luke @flexchar, Elle, Caitlyn Gatomon, Ari Malik, subjectnull, Johann-Peter Hartmann, Trenton Dambrowitz, Imad Khwaja, Asp the Wyvern, Emad Mostaque, Rainer Wilmers, Alexandros Triantafyllidis, Nicholas, Pedro Madruga, SuperWojo, Harry Royden McLaughlin, James Bentley, Olakabola, David Ziegler, Ai Maven, Jeff Scroggin, Nikolai Manek, Deo Leter, Matthew Berman, Fen Risland, Ken Nordquist, Manuel Alberto Morcote, Luke Pendergrass, TL, Fred von Graf, Randy H, Dan Guido, NimbleBox.ai, Vitor Caleffi, Gabriel Tamborski, knownsqashed, Lone Striker, Erik Bjäreholt, John Detwiler, Leonard Tan, Iucharbius
Thank you to all my generous patrons and donaters!
And thank you again to a16z for their generous grant.
<!-- footer end -->
<!-- original-model-card start -->
# Original model card: KAIST AI's Prometheus 13B V1.0
## Links for Reference
- **Homepage:https://github.com/kaistAI/Prometheus**
- **Repository:https://github.com/kaistAI/Prometheus**
- **Paper:https://arxiv.org/abs/2310.08491**
- **Point of Contact:[email protected]**
# TL;DR
Prometheus is an alternative of GPT-4 evaluation when doing fine-grained evaluation of an underlying LLM & a Reward model for Reinforcement Learning from Human Feedback (RLHF).
Prometheus is a language model using [Llama-2-Chat](https://huggingface.co/meta-llama/Llama-2-13b-chat-hf) as a base model and fine-tuned on 100K feedback within the [Feedback Collection](https://huggingface.co/datasets/kaist-ai/Feedback-Collection).
Since it was fine-tuned on a large amount of feedback, it is specialized at evaluating long-form responses, outperforming GPT-3.5-Turbo, Llama-2-Chat 70B, and on par with GPT-4 on various benchmarks.
Most importantly, this was possible since we appended 2 reference materials (reference answer, and customized score rubric).
Prometheus is a cheap and powerful alternative to GPT-4 evaluation, which one could use to evaluate LLMs with customized criteria (e.g., Child readability, Cultural Sensitivity, Creativity).
Also, it could be used as a reward model for Reinforcement Learning from Human Feedback (RLHF).
# Model Details
## Model Description
- **Model type:** Language model
- **Language(s) (NLP):** English
- **License:** Apache 2.0
- **Related Models:** [All Prometheus Checkpoints](https://huggingface.co/models?search=kaist-ai/Prometheus)
- **Resources for more information:**
- [Research paper](https://arxiv.org/abs/2310.08491)
- [GitHub Repo](https://github.com/kaistAI/Prometheus)
Prometheus is trained with two different sizes (7B and 13B).
You could check the 7B sized LM on [this page](https://huggingface.co/kaist-ai/prometheus-7b-v1.0).
Also, check out our dataset as well on [this page](https://huggingface.co/datasets/kaist-ai/Feedback-Collection).
## Prompt Format
Prometheus requires 4 components in the input: An instruction, a response to evaluate, a score rubric, and a reference answer. You could refer to the prompt format below.
You should fill in the instruction, response, reference answer, criteria description, and score description for score in range of 1 to 5.
```
###Task Description:
An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given.
1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general.
2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric.
3. The output format should look as follows: \"Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)\"
4. Please do not generate any other opening, closing, and explanations.
###The instruction to evaluate:
{instruction}
###Response to evaluate:
{response}
###Reference Answer (Score 5):
{reference_answer}
###Score Rubrics:
[{criteria_description}]
Score 1: {score1_description}
Score 2: {score2_description}
Score 3: {score3_description}
Score 4: {score4_description}
Score 5: {score5_description}
###Feedback:
```
After this, you should apply the conversation template of Llama-2-Chat (not applying it might lead to unexpected behaviors).
You can find the conversation class at this [link](https://github.com/lm-sys/FastChat/blob/main/fastchat/conversation.py).
```
conv = get_conv_template("llama-2")
conv.set_system_message("You are a fair evaluator language model.")
conv.append_message(conv.roles[0], dialogs['instruction'])
conv.append_message(conv.roles[1], None)
prompt = conv.get_prompt()
x = tokenizer(prompt,truncation=False)
```
As a result, a feedback and score decision will be generated, divided by a separating phrase ```[RESULT]```
## License
Feedback Collection and Prometheus is subject to OpenAI's Terms of Use for the generated data. If you suspect any violations, please reach out to us.
# Usage
Find below some example scripts on how to use the model in `transformers`:
## Using the Pytorch model
### Running the model on a CPU
<details>
<summary> Click to expand </summary>
```python
from transformers import T5Tokenizer, T5ForConditionalGeneration
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto")
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
### Running the model on a GPU
<details>
<summary> Click to expand </summary>
```python
# pip install accelerate
import torch
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto")
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids, sample=True, temperature=1.0, top_p=0.9, max_new_tokens=256, repetition_penalty=1.03)
print(tokenizer.decode(outputs[0]))
```
</details>
### Running the model on a GPU using different precisions
#### FP16
<details>
<summary> Click to expand </summary>
```python
# pip install accelerate
import torch
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto", torch_dtype=torch.float16)
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
#### INT8
<details>
<summary> Click to expand </summary>
```python
# pip install bitsandbytes accelerate
from transformers import AutoTokenizer, LlamaForCausalLM
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
model = LlamaForCausalLM.from_pretrained("kaist-ai/Prometheus-13b-v1.0", device_map="auto", load_in_8bit=True)
input_text = "###Task Description: An instruction (might include an Input inside it), a response to evaluate, a reference answer that gets a score of 5, and a score rubric representing a evaluation criteria are given. 1. Write a detailed feedback that assess the quality of the response strictly based on the given score rubric, not evaluating in general. 2. After writing a feedback, write a score that is an integer between 1 and 5. You should refer to the score rubric. 3. The output format should look as follows: "Feedback: (write a feedback for criteria) [RESULT] (an integer number between 1 and 5)" 4. Please do not generate any other opening, closing, and explanations. ###The instruction to evaluate: You are a well-known psychiatrist who has a reputation for being empathetic and understanding. A client comes to you saying they've had a really hard day at work. They describe their boss as being overly critical and not listening to their ideas, which has left them feeling frustrated and undervalued. They also mention feeling overwhelmed with the workload and a growing sense of loneliness as they are new to the city. How do you respond to them to make them feel heard and understood, as well as offering helpful advice? ###Response to evaluate: It's indeed challenging to deal with a difficult boss and to carry a large workload, especially when you are new to a city and haven't established a supportive network. I would suggest taking a step back to think about your situation and perhaps speak to your boss regarding the issues you're facing. On the other hand, dealing with a large workload can be managed by prioritizing tasks, and maybe you can discuss it with your superiors too. In regards to feeling lonely, you might want to explore activities or groups that align with your interests to meet new people. And also, remember to take time for yourself to relax and destress. I'm here to assist you as we navigate through these issues. ###Reference Answer (Score 5): I hear that you're dealing with quite a few challenging situations at once. It must be tough to have your ideas dismissed by your boss and to feel undervalued in your work environment. It's understandable that such experiences can lead to frustration and a feeling of being overwhelmed. It's important to acknowledge these feelings and not to dismiss them. Feeling heard and validated in our experiences can often be the first step towards resolving the challenges we face. So, I want you to know that your feelings are completely valid and it's okay to be frustrated and overwhelmed. In terms of dealing with your boss, have you considered seeking a private conversation with them to express your feelings? It's often beneficial to communicate what you're experiencing, using "I" statements to express how you feel when your ideas are not taken into consideration. This can lead to mutual understanding and possibly a change in behavior. About the workload, it might help to prioritize tasks and potentially delegate, if possible. Also, consider discussing your workload with your superiors. There might be a misunderstanding about what's manageable or they might have suggestions about how to handle the situation. On the personal front, feeling lonely, especially when you're new to a city, can be really hard. Seek out opportunities to meet new people, perhaps through hobbies, community activities, or online groups. It might take a bit of time, but gradually, you can build a network of friends and acquaintances. Remember, it's perfectly okay to have bad days and it's important to take care of your mental health. Consider incorporating activities into your daily routine that make you happy and help you unwind. This could be anything from reading, yoga, going for a walk, or even listening to your favorite music. Please know that you're not alone in this. I'm here to support you through this challenging time and together, we can work towards resolving these issues. ###Score Rubrics: [Is the model able to identify and react correctly to the emotional context of the user's input?] Score 1: The model utterly fails to grasp the user's emotional context and responds in an unfitting manner. Score 2: The model sporadically identifies the emotional context but frequently replies in a manner that doesn't match the user's emotional status. Score 3: The model typically identifies the emotional context and reacts suitably, but occasionally misreads or misjudges the user's feelings. Score 4: The model often identifies the emotional context and reacts suitably, with minor cases of misreading or misjudging. Score 5: The model flawlessly identifies the emotional context of the user's input and consistently responds in a considerate and empathetic manner. ###Feedback:"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))
```
</details>
# Citation
If you find the following model helpful, please consider citing our paper!
**BibTeX:**
```bibtex
@misc{kim2023prometheus,
title={Prometheus: Inducing Fine-grained Evaluation Capability in Language Models},
author={Seungone Kim and Jamin Shin and Yejin Cho and Joel Jang and Shayne Longpre and Hwaran Lee and Sangdoo Yun and Seongjin Shin and Sungdong Kim and James Thorne and Minjoon Seo},
year={2023},
eprint={2310.08491},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
```
<!-- original-model-card end -->
|
PTU-AI-CLUB/UnoLineGPT | PTU-AI-CLUB | 2023-11-10T18:12:51Z | 8 | 1 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"gpt2",
"text-generation",
"generated_from_trainer",
"base_model:openai-community/gpt2",
"base_model:finetune:openai-community/gpt2",
"license:mit",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
]
| text-generation | 2023-11-10T13:27:56Z | ---
license: mit
base_model: gpt2
tags:
- generated_from_trainer
model-index:
- name: UnoLineGPT
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# UnoLineGPT
This model is a fine-tuned version of [gpt2](https://huggingface.co/gpt2) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 1.5971
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0005
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 256
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 1000
- num_epochs: 1
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 2.4756 | 0.94 | 5000 | 1.5971 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
Dmyadav2001/Sentimental-Analysis | Dmyadav2001 | 2023-11-10T18:12:03Z | 219,167 | 5 | transformers | [
"transformers",
"safetensors",
"distilbert",
"text-classification",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| text-classification | 2023-11-10T17:37:12Z |
# Sentiment Analysis with DistilBERT
This repository contains a sentiment analysis project using the DistilBERT model. Sentiment analysis involves classifying text data into different sentiment categories, such as positive (label-1), negative (label-0), or neutral (label-2).
## Overview
The project is implemented using Python and leverages several libraries for natural language processing and machine learning. It includes the following components:
1. *Dataset*: The Sentiment Analysis dataset is loaded using the [datasets] library. The dataset is split into training and validation sets for model training and evaluation.
2. *Text Preprocessing*: Text data is preprocessed to remove special characters, links, and user mentions. The DistilBERT tokenizer is used to tokenize and preprocess the text, and the data is prepared for training.
3. *Training Configuration*: The training configuration, including batch size, learning rate, and evaluation settings, is defined using the [TrainingArguments].
4. *Model*: The sentiment analysis model is based on DistilBERT, a lightweight version of BERT, and is fine-tuned for sequence classification. The model is initialized, and the number of labels (positive, negative, and neutral) is specified.
5. *Trainer*: A [Trainer] instance is created to handle the training process. It takes the training dataset, evaluation dataset, and training configuration.
6. *Training*: The model is trained using the training dataset with the provided configuration. Training results, including loss and accuracy, are recorded.
7. *Evaluation*: After training, the model's performance is evaluated on the validation dataset. A classification report is generated to assess the model's accuracy and performance in classifying sentiments.
8. *Model Saving*: The trained model and tokenizer are saved for later use or deployment.
## Usage
To use this code for your own sentiment analysis tasks, you can follow these steps:
1. *Installation*: Install the required libraries using the provided pip commands.
2. *Load Dataset*: Replace the dataset with your text data or use the provided SST-2 dataset.
3. *Training Configuration*: Modify the training arguments, such as batch size, learning rate, and evaluation strategy, in the TrainingArguments section to suit your specific task.
4. *Model Customization*: If needed, customize the model architecture or the number of labels according to your sentiment classification requirements.
5. *Training*: Train the model on your dataset by running the training code.
6. *Evaluation*: Evaluate the model's performance using your validation dataset or sample data.
7. *Model Saving*: Save the trained model and tokenizer for future use or deployment.
## Limitations
- The provided code assumes a three-class sentiment classification task (positive, negative, and neutral). It may require adaptation for tasks with different label sets or multi-class classification.
- The code uses DistilBERT, a smaller and faster version of BERT. For tasks that demand highly accurate but more computationally intensive models, it may be necessary to switch to the full BERT model or other advanced architectures.
## Future Requirements
To further enhance and extend this sentiment analysis project, consider the following:
- *Custom Dataset*: If you have a specific domain or industry, consider collecting and preparing a custom dataset that is more relevant to your application.
- *Fine-tuning*: Experiment with fine-tuning hyperparameters and explore techniques like learning rate schedules or additional layers for the model.
- *Deployment*: If you plan to use the model in a real-world application, explore deployment options, such as building a web service or integrating the model into an existing system.
- *Performance Optimization*: Optimize the code for training on larger datasets and explore distributed training to improve efficiency.
|
lmqg/mt5-small-zhquad-qag-trimmed-50000 | lmqg | 2023-11-10T18:09:44Z | 21 | 0 | transformers | [
"transformers",
"pytorch",
"mt5",
"text2text-generation",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| text2text-generation | 2023-11-10T16:27:05Z | # Vocabulary Trimmed [lmqg/mt5-small-zhquad-qag](https://huggingface.co/lmqg/mt5-small-zhquad-qag): `lmqg/mt5-small-zhquad-qag-trimmed-50000`
This model is a trimmed version of [lmqg/mt5-small-zhquad-qag](https://huggingface.co/lmqg/mt5-small-zhquad-qag) by [`vocabtrimmer`](https://github.com/asahi417/lm-vocab-trimmer), a tool for trimming vocabulary of language models to compress the model size.
Following table shows a summary of the trimming process.
| | lmqg/mt5-small-zhquad-qag | lmqg/mt5-small-zhquad-qag-trimmed-50000 |
|:---------------------------|:----------------------------|:------------------------------------------|
| parameter_size_full | 300,165,504 | 95,264,128 |
| parameter_size_embedding | 256,103,424 | 51,202,048 |
| vocab_size | 250,101 | 50,002 |
| compression_rate_full | 100.0 | 31.74 |
| compression_rate_embedding | 100.0 | 19.99 |
Following table shows the parameter used to trim vocabulary.
| language | dataset | dataset_column | dataset_name | dataset_split | target_vocab_size | min_frequency |
|:-----------|:----------------------------|:-----------------|:---------------|:----------------|--------------------:|----------------:|
| zh | vocabtrimmer/mc4_validation | text | zh | validation | 50000 | 2 | |
nikhil121/content | nikhil121 | 2023-11-10T18:06:32Z | 0 | 0 | null | [
"tensorboard",
"safetensors",
"generated_from_trainer",
"base_model:meta-llama/Llama-2-7b-hf",
"base_model:finetune:meta-llama/Llama-2-7b-hf",
"region:us"
]
| null | 2023-11-07T09:59:34Z | ---
base_model: meta-llama/Llama-2-7b-hf
tags:
- generated_from_trainer
model-index:
- name: content
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# content
This model is a fine-tuned version of [meta-llama/Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) on an unknown dataset.
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.02
- train_batch_size: 2
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1
- training_steps: 200
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
jason1i/mistral_instruct_generation | jason1i | 2023-11-10T18:06:05Z | 0 | 0 | null | [
"tensorboard",
"safetensors",
"generated_from_trainer",
"base_model:mistralai/Mistral-7B-Instruct-v0.1",
"base_model:finetune:mistralai/Mistral-7B-Instruct-v0.1",
"license:apache-2.0",
"region:us"
]
| null | 2023-11-10T18:05:50Z | ---
license: apache-2.0
base_model: mistralai/Mistral-7B-Instruct-v0.1
tags:
- generated_from_trainer
model-index:
- name: mistral_instruct_generation
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# mistral_instruct_generation
This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.1](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 1.3095
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0002
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant
- lr_scheduler_warmup_steps: 0.03
- training_steps: 100
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 1.53 | 0.02 | 20 | 1.3816 |
| 1.4525 | 0.03 | 40 | 1.3351 |
| 1.4299 | 0.05 | 60 | 1.3230 |
| 1.4289 | 0.06 | 80 | 1.3140 |
| 1.4275 | 0.08 | 100 | 1.3095 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
zz990906/bert-base-uncased-finetuned-cda | zz990906 | 2023-11-10T18:04:03Z | 26 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"bert",
"fill-mask",
"generated_from_trainer",
"base_model:google-bert/bert-base-uncased",
"base_model:finetune:google-bert/bert-base-uncased",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| fill-mask | 2023-11-10T16:34:37Z | ---
license: apache-2.0
base_model: bert-base-uncased
tags:
- generated_from_trainer
model-index:
- name: bert-base-uncased-finetuned-cda
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# bert-base-uncased-finetuned-cda
This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.6567
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 2.0518 | 1.0 | 391 | 1.8074 |
| 1.8971 | 2.0 | 782 | 1.7770 |
| 1.8422 | 3.0 | 1173 | 1.7504 |
| 1.7984 | 4.0 | 1564 | 1.7272 |
| 1.777 | 5.0 | 1955 | 1.6912 |
| 1.7532 | 6.0 | 2346 | 1.6920 |
| 1.7323 | 7.0 | 2737 | 1.6826 |
| 1.7251 | 8.0 | 3128 | 1.6687 |
| 1.7108 | 9.0 | 3519 | 1.6553 |
| 1.7076 | 10.0 | 3910 | 1.6702 |
### Framework versions
- Transformers 4.36.0.dev0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|
AdamCodd/distilroberta-squad2 | AdamCodd | 2023-11-10T17:38:10Z | 27 | 1 | transformers | [
"transformers",
"pytorch",
"onnx",
"safetensors",
"roberta",
"question-answering",
"dataset:squad_v2",
"endpoints_compatible",
"region:us"
]
| question-answering | 2023-10-18T16:03:23Z | ---
datasets:
- squad_v2
metrics:
- f1
- exact_match
---
## Distilroberta-squad2
This model is [Distilroberta base](https://huggingface.co/distilroberta-base) which was fine-tuned for context-based question answering on the [SQuAD v2](https://huggingface.co/datasets/squad_v2) dataset, a dataset of English-language context-question-answer triples designed for extractive question answering training and benchmarking. Version 2 of SQuAD (Stanford Question Answering Dataset) contains the 100,000 examples from SQuAD Version 1.1, along with 50,000 additional "unanswerable" questions, i.e. questions whose answer cannot be found in the provided context.
## Model description
This fine-tuned model prioritizes inference speed; DistilRoBERTa operates at a pace twice as fast as the RoBERTa-base model, with only a marginal compromise in quality.
## Intended uses & limitations
```python
from transformers import pipeline
QA_pipeline = pipeline("question-answering", model="AdamCodd/distilroberta-squad2", handle_impossible_answer=True)
input = {
'question': "Which name is also used to describe the Amazon rainforest in English?",
'context': '''The Amazon rainforest (Portuguese: Floresta Amazônica or Amazônia; Spanish: Selva Amazónica, Amazonía or usually Amazonia; French: Forêt amazonienne; Dutch: Amazoneregenwoud), also known in English as Amazonia or the Amazon Jungle, is a moist broadleaf forest that covers most of the Amazon basin of South America. This basin encompasses 7,000,000 square kilometres (2,700,000 sq mi), of which 5,500,000 square kilometres (2,100,000 sq mi) are covered by the rainforest. This region includes territory belonging to nine nations. The majority of the forest is contained within Brazil, with 60% of the rainforest, followed by Peru with 13%, Colombia with 10%, and with minor amounts in Venezuela, Ecuador, Bolivia, Guyana, Suriname and French Guiana. States or departments in four nations contain "Amazonas" in their names. The Amazon represents over half of the planet's remaining rainforests, and comprises the largest and most biodiverse tract of tropical rainforest in the world, with an estimated 390 billion individual trees divided into 16,000 species.'''
}
response = QA_pipeline(**input)
print(response)
```
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 3e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- mixed_precision = "fp16"
- max_seq_len = 384
- doc_stride = 128
- optimizer: AdamW with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 150
- num_epochs: 3
### Training results
Evaluated on the SQuAD 2.0 dev set with the [official eval script](https://worksheets.codalab.org/rest/bundles/0x6b567e1cf2e041ec80d7098f031c5c9e/contents/blob/).
Results:
```
'exact': 72.9470226564474,
'f1': 76.03522762032603,
'total': 11873,
'HasAns_exact': 72.4527665317139,
'HasAns_f1': 78.63803264779528,
'HasAns_total': 5928,
'NoAns_exact': 73.43986543313709,
'NoAns_f1': 73.43986543313709,
'NoAns_total': 5945,
'best_exact': 72.95544512760044,
'best_exact_thresh': 0.0,
'best_f1': 76.04365009147917,
'best_f1_thresh': 0.0
```
### Framework versions
- Transformers 4.34.0
- Torch 2.0.1
- Accelerate 0.23.0
- Tokenizers 0.14.1
If you want to support me, you can [here](https://ko-fi.com/adamcodd). |
stchaganti/donut-base-sroie | stchaganti | 2023-11-10T17:37:12Z | 4 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"vision-encoder-decoder",
"image-text-to-text",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:naver-clova-ix/donut-base",
"base_model:finetune:naver-clova-ix/donut-base",
"license:mit",
"endpoints_compatible",
"region:us"
]
| image-text-to-text | 2023-11-10T17:27:19Z | ---
license: mit
base_model: naver-clova-ix/donut-base
tags:
- generated_from_trainer
datasets:
- imagefolder
model-index:
- name: donut-base-sroie
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# donut-base-sroie
This model is a fine-tuned version of [naver-clova-ix/donut-base](https://huggingface.co/naver-clova-ix/donut-base) on the imagefolder dataset.
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 2
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
### Training results
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0
- Datasets 2.14.6
- Tokenizers 0.14.1
|
AdamCodd/distilbert-base-uncased-finetuned-sentiment-amazon | AdamCodd | 2023-11-10T17:35:29Z | 36 | 7 | transformers | [
"transformers",
"pytorch",
"onnx",
"safetensors",
"distilbert",
"text-classification",
"dataset:amazon_polarity",
"base_model:distilbert/distilbert-base-uncased",
"base_model:quantized:distilbert/distilbert-base-uncased",
"doi:10.57967/hf/4238",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| text-classification | 2023-10-06T23:02:11Z | ---
license: apache-2.0
datasets:
- amazon_polarity
base_model: distilbert-base-uncased
model-index:
- name: distilbert-base-uncased-finetuned-sentiment-amazon
results:
- task:
type: text-classification
name: Text Classification
dataset:
name: amazon_polarity
type: sentiment
args: default
metrics:
- type: accuracy
value: 0.961
name: Accuracy
- type: loss
value: 0.116
name: Loss
- type: f1
value: 0.960
name: F1
- task:
type: text-classification
name: Text Classification
dataset:
name: amazon_polarity
type: amazon_polarity
config: amazon_polarity
split: test
metrics:
- type: accuracy
value: 0.94112
name: Accuracy
verified: true
verifyToken: >-
eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiMzlmMzdhYjNmN2U0NDBkM2U5ZDgwNzc3YjE1OGE4MWUxMDY1N2U0ODc0YzllODE5ODIyMzdkOWFhNzVjYmI5MyIsInZlcnNpb24iOjF9.3nlcLa4IpPQtklp7_U9XzC__Q_JVf_cWs6JVVII8trhX5zg_q9HEyQOQs4sRf6O-lIJg8zb3mgobZDJShuSJAQ
- type: precision
value: 0.9321570625232675
name: Precision
verified: true
verifyToken: >-
eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiZjI2MDY4NGNlYjhjMGMxODBiNTc2ZjM5YzY1NjkxNTU4MDA2ZDIyY2QyZjUyZmE4YWY0N2Y1ODU5YTc2ZDM0NiIsInZlcnNpb24iOjF9.egEikTa2UyHV6SAGkHJKaa8FRwGHoZmJRCmqUQaJqeF5yxkz2V-WeCHoWDrCXsHCbXEs8UhLlyo7Lr83BPfkBg
- type: recall
value: 0.95149
name: Recall
verified: true
verifyToken: >-
eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiM2E3M2Y3MDU4ZTM2YjdlZjQ0NTY3NGYwMmQ3NTk5ZmZkZWUwZWZiZDZjNjk2ZWE5MmY4MmZiM2FmN2U2M2QyNCIsInZlcnNpb24iOjF9.4VNbiWRmSee4cxuIZ5m7bN30i4BpK7xtHQ1BF8AuFIXkWQgzOmGdX35bLhLGWW8KL3ClA4RDPVBKYCIrw0YUBw
- type: auc
value: 0.9849019044624999
name: AUC
verified: true
verifyToken: >-
eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiYTkwODk2ZTUwOTViNjBhYTU0ODk1MDA3MDY1NDkyZDc2YmRlNTQzNDE3YmE3YTVkYjNhN2JmMDAxZWQ0NjUxZSIsInZlcnNpb24iOjF9.YEr6OhqOL7QnqYqjUTQFMdkgU_uS1-vVnkJtn_-1UwSoX754UV_bL9S9KSH3DX4m5QFoRXdZxfeOocm1JbzaCA
- type: f1
value: 0.9417243188138998
name: F1
verified: true
verifyToken: >-
eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiMzIyMmViNTQ3ZGU0M2I5ZmRjOGI1OWMwZGEwYmE5OGU5YTZlZTkzZjdkOTQ4YzJmOTc2MDliMDY4NDQ1NGRlNyIsInZlcnNpb24iOjF9.p05MGHTfHTAzp4u-qfiIn6Zmh5c3TW_uwjXWgbb982pL_oCILQb6jFXqhPpWXL321fPye7qaUVbGhcTJd8sdCA
- type: loss
value: 0.16342754662036896
name: loss
verified: true
verifyToken: >-
eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJoYXNoIjoiNzgxMDc4M2IxYjhkNjRhZmYyNzY1MTNkNzhmYjk2NmU1NjFiOTk1NDIzNzI1ZGU3MDYyYjQ2YmQ1NTI2N2NhMyIsInZlcnNpb24iOjF9.Zuf0nzn8XdvwRChKtE9CwJ0pgpc6Zey6oTR3jRiSkvNY2sNbo2bvAgFimGzgGYkDvRvYkTCXzCyxdb27l3QnAg
---
# distilbert-sentiment
This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on a subset of the [amazon-polarity dataset](https://huggingface.co/datasets/amazon_polarity).
<b>[Update 10/10/23]</b> The model has been retrained on a larger part of the dataset with an improvement on the loss, f1 score and accuracy. It achieves the following results on the evaluation set:
- Loss: 0.116
- Accuracy: 0.961
- F1_score: 0.960
## Model description
This sentiment classifier has been trained on 360_000 samples for the training set, 40_000 samples for the validation set and 40_000 samples for the test set.
## Intended uses & limitations
```python
from transformers import pipeline
# Create the pipeline
sentiment_classifier = pipeline('text-classification', model='AdamCodd/distilbert-base-uncased-finetuned-sentiment-amazon')
# Now you can use the pipeline to get the sentiment
result = sentiment_classifier("This product doesn't fit me at all.")
print(result)
#[{'label': 'negative', 'score': 0.9994848966598511}]
```
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 3e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 1270
- optimizer: AdamW with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 150
- num_epochs: 2
- weight_decay: 0.01
### Training results
(Previous results before retraining from the model evaluator)
| key | value |
| --- | ----- |
| eval_accuracy | 0.94112 |
| eval_auc | 0.9849 |
| eval_f1_score | 0.9417 |
| eval_precision | 0.9321 |
| eval_recall | 0.95149 |
### Framework versions
- Transformers 4.34.0
- Pytorch lightning 2.0.9
- Tokenizers 0.14.0
If you want to support me, you can [here](https://ko-fi.com/adamcodd). |
RJuro/kanelsnegl-v0.1-GGUF | RJuro | 2023-11-10T16:56:51Z | 1 | 1 | null | [
"gguf",
"da",
"en",
"dataset:DDSC/partial-danish-gigaword-no-twitter",
"license:mit",
"endpoints_compatible",
"region:us"
]
| null | 2023-11-10T16:29:35Z | ---
license: mit
datasets:
- DDSC/partial-danish-gigaword-no-twitter
language:
- da
- en
---
# Model Card for kanelsnegl-v0.1-GGUF
!!! This model is built for fun and learning and needs much more finetuning !!!
For now only Q4_K.
<img src="https://huggingface.co/RJuro/kanelsnegl-v0.1-GGUF/resolve/main/kanelsnegl_banner.png" alt="Kanelsnegl Logo" width="800" style="margin-left:'auto' margin-right:'auto' display:'block'"/>
## Model Description
Base model: [Zephyr-7b-alpha](https://huggingface.co/HuggingFaceH4/zephyr-7b-alpha) finetuned on [DDSC/partial-danish-gigaword-no-twitter](https://huggingface.co/datasets/DDSC/partial-danish-gigaword-no-twitter). The training involved a maximum length of 512. QLora completion finetuning of all linear layers was also implemented. This model is mostly fun tinkering for personal learning purpose. A heavily instruction-tuned model was used in the hope of transferring some of its behaviors into the Danish finetune.
The model often produces somewhat fun halucinations and instruction following is not really happening that well, but there may be ways to fine-tune it in a supervised manner for more controlled behavior.
**Works with Ollama but not too well**
## Usage with CTransformers
```python
from ctransformers import AutoModelForCausalLM
# Set gpu_layers to the number of layers to offload to GPU. Set to 0 if no GPU acceleration is available on your system.
llm = AutoModelForCausalLM.from_pretrained('RJuro/kanelsnegl-v0.1-GGUF',
model_file="kanelsnegl-v0.1-Q4_K.gguf",
model_type="mistral",
gpu_layers=50
)
print(llm("Aalborg er en dejlig by.",
top_k=50,
top_p=0.95,
temperature=0.8,
repetition_penalty=1.2,
reset=True,
max_new_tokens=128))
```
Returns 😂
```
Det var det også før de bakker hul i jorden og lægger køretøjerne på grunden, så man ikke kan komme ud af gaderne.
Det er jo bare fordi vi har et stort problem med at have en stor by udenfor Aalborg, der bliver ved med at vokse og vokse.
Jeg bor i Aalborg, men jeg kører hver dag til Frederikshavn på arbejde. Jeg er ikke så glad for det, da den traf
```
|
AntoineD/camembert_ccnet_classification_tools_classifier-only_fr | AntoineD | 2023-11-10T16:51:57Z | 4 | 0 | transformers | [
"transformers",
"pytorch",
"camembert",
"text-classification",
"generated_from_trainer",
"base_model:almanach/camembert-base-ccnet",
"base_model:finetune:almanach/camembert-base-ccnet",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| text-classification | 2023-11-10T16:45:28Z | ---
base_model: camembert/camembert-base-ccnet
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: camembert_ccnet_classification_tools_classifier-only_fr
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# camembert_ccnet_classification_tools_classifier-only_fr
This model is a fine-tuned version of [camembert/camembert-base-ccnet](https://huggingface.co/camembert/camembert-base-ccnet) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2297
- Accuracy: 0.975
- Learning Rate: 0.0
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.001
- train_batch_size: 24
- eval_batch_size: 192
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 60
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Rate |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:------:|
| No log | 1.0 | 7 | 1.7771 | 0.5 | 0.0010 |
| No log | 2.0 | 14 | 1.5458 | 0.425 | 0.0010 |
| No log | 3.0 | 21 | 1.2091 | 0.75 | 0.0009 |
| No log | 4.0 | 28 | 1.0227 | 0.75 | 0.0009 |
| No log | 5.0 | 35 | 0.9105 | 0.7 | 0.0009 |
| No log | 6.0 | 42 | 0.7765 | 0.825 | 0.0009 |
| No log | 7.0 | 49 | 0.7397 | 0.75 | 0.0009 |
| No log | 8.0 | 56 | 0.6652 | 0.825 | 0.0009 |
| No log | 9.0 | 63 | 0.6385 | 0.775 | 0.0008 |
| No log | 10.0 | 70 | 0.6051 | 0.8 | 0.0008 |
| No log | 11.0 | 77 | 0.5246 | 0.9 | 0.0008 |
| No log | 12.0 | 84 | 0.5391 | 0.825 | 0.0008 |
| No log | 13.0 | 91 | 0.5511 | 0.825 | 0.0008 |
| No log | 14.0 | 98 | 0.4780 | 0.85 | 0.0008 |
| No log | 15.0 | 105 | 0.4328 | 0.925 | 0.0008 |
| No log | 16.0 | 112 | 0.4016 | 0.875 | 0.0007 |
| No log | 17.0 | 119 | 0.4902 | 0.8 | 0.0007 |
| No log | 18.0 | 126 | 0.4016 | 0.9 | 0.0007 |
| No log | 19.0 | 133 | 0.4164 | 0.9 | 0.0007 |
| No log | 20.0 | 140 | 0.3814 | 0.825 | 0.0007 |
| No log | 21.0 | 147 | 0.3147 | 0.95 | 0.0007 |
| No log | 22.0 | 154 | 0.3544 | 0.9 | 0.0006 |
| No log | 23.0 | 161 | 0.3438 | 0.9 | 0.0006 |
| No log | 24.0 | 168 | 0.3181 | 0.95 | 0.0006 |
| No log | 25.0 | 175 | 0.3433 | 0.875 | 0.0006 |
| No log | 26.0 | 182 | 0.3197 | 0.95 | 0.0006 |
| No log | 27.0 | 189 | 0.3510 | 0.875 | 0.0006 |
| No log | 28.0 | 196 | 0.3495 | 0.9 | 0.0005 |
| No log | 29.0 | 203 | 0.2704 | 0.95 | 0.0005 |
| No log | 30.0 | 210 | 0.3081 | 0.975 | 0.0005 |
| No log | 31.0 | 217 | 0.3389 | 0.875 | 0.0005 |
| No log | 32.0 | 224 | 0.2695 | 0.95 | 0.0005 |
| No log | 33.0 | 231 | 0.2903 | 0.925 | 0.0005 |
| No log | 34.0 | 238 | 0.3012 | 0.925 | 0.0004 |
| No log | 35.0 | 245 | 0.2804 | 0.95 | 0.0004 |
| No log | 36.0 | 252 | 0.2716 | 0.95 | 0.0004 |
| No log | 37.0 | 259 | 0.3125 | 0.875 | 0.0004 |
| No log | 38.0 | 266 | 0.2501 | 0.925 | 0.0004 |
| No log | 39.0 | 273 | 0.2442 | 0.95 | 0.0003 |
| No log | 40.0 | 280 | 0.2284 | 0.95 | 0.0003 |
| No log | 41.0 | 287 | 0.2363 | 0.975 | 0.0003 |
| No log | 42.0 | 294 | 0.2656 | 0.975 | 0.0003 |
| No log | 43.0 | 301 | 0.2723 | 0.925 | 0.0003 |
| No log | 44.0 | 308 | 0.2302 | 0.95 | 0.0003 |
| No log | 45.0 | 315 | 0.2391 | 0.95 | 0.0003 |
| No log | 46.0 | 322 | 0.2414 | 0.95 | 0.0002 |
| No log | 47.0 | 329 | 0.2506 | 0.975 | 0.0002 |
| No log | 48.0 | 336 | 0.2616 | 0.975 | 0.0002 |
| No log | 49.0 | 343 | 0.2376 | 0.975 | 0.0002 |
| No log | 50.0 | 350 | 0.2310 | 0.975 | 0.0002 |
| No log | 51.0 | 357 | 0.2271 | 0.95 | 0.0001 |
| No log | 52.0 | 364 | 0.2484 | 0.95 | 0.0001 |
| No log | 53.0 | 371 | 0.2633 | 0.95 | 0.0001 |
| No log | 54.0 | 378 | 0.2788 | 0.925 | 0.0001 |
| No log | 55.0 | 385 | 0.2626 | 0.975 | 0.0001 |
| No log | 56.0 | 392 | 0.2454 | 0.975 | 0.0001 |
| No log | 57.0 | 399 | 0.2373 | 0.975 | 5e-05 |
| No log | 58.0 | 406 | 0.2340 | 0.975 | 0.0000 |
| No log | 59.0 | 413 | 0.2291 | 0.975 | 0.0000 |
| No log | 60.0 | 420 | 0.2297 | 0.975 | 0.0 |
### Framework versions
- Transformers 4.34.0
- Pytorch 2.0.1+cu117
- Datasets 2.14.5
- Tokenizers 0.14.1
|
nixiesearch/e5-large-v2-onnx | nixiesearch | 2023-11-10T16:41:20Z | 20 | 0 | sentence-transformers | [
"sentence-transformers",
"onnx",
"bert",
"feature-extraction",
"sentence-similarity",
"en",
"dataset:s2orc",
"dataset:flax-sentence-embeddings/stackexchange_xml",
"dataset:ms_marco",
"dataset:gooaq",
"dataset:yahoo_answers_topics",
"dataset:code_search_net",
"dataset:search_qa",
"dataset:eli5",
"dataset:snli",
"dataset:multi_nli",
"dataset:wikihow",
"dataset:natural_questions",
"dataset:trivia_qa",
"dataset:embedding-data/sentence-compression",
"dataset:embedding-data/flickr30k-captions",
"dataset:embedding-data/altlex",
"dataset:embedding-data/simple-wiki",
"dataset:embedding-data/QQP",
"dataset:embedding-data/SPECTER",
"dataset:embedding-data/PAQ_pairs",
"dataset:embedding-data/WikiAnswers",
"license:apache-2.0",
"autotrain_compatible",
"text-embeddings-inference",
"endpoints_compatible",
"region:us"
]
| feature-extraction | 2023-11-07T13:07:16Z | ---
tags:
- sentence-transformers
- feature-extraction
- sentence-similarity
language: en
license: apache-2.0
datasets:
- s2orc
- flax-sentence-embeddings/stackexchange_xml
- ms_marco
- gooaq
- yahoo_answers_topics
- code_search_net
- search_qa
- eli5
- snli
- multi_nli
- wikihow
- natural_questions
- trivia_qa
- embedding-data/sentence-compression
- embedding-data/flickr30k-captions
- embedding-data/altlex
- embedding-data/simple-wiki
- embedding-data/QQP
- embedding-data/SPECTER
- embedding-data/PAQ_pairs
- embedding-data/WikiAnswers
---
# ONNX version of intfloat/e5-large-v2
This is a sentence-transformers model: It maps sentences & paragraphs to a N dimensional dense vector space and can be used for tasks like clustering or semantic search.
The model conversion was made with [onnx-convert](https://github.com/nixiesearch/onnx-convert) tool with the following parameters:
```shell
python convert.sh --model_id intfloat/e5-large-v2 --quantize QInt8 --optimize 2
```
There are two versions of model available:
* `model.onnx` - Float32 version, with optimize=2
* `model_opt2_QInt8.onnx` - QInt8 quantized version, with optimize=2
## License
Apache 2.0 |
AntoineD/MiniLM_uncased_classification_tools_classifier-only_fr | AntoineD | 2023-11-10T16:40:08Z | 9 | 0 | transformers | [
"transformers",
"pytorch",
"bert",
"text-classification",
"generated_from_trainer",
"base_model:microsoft/MiniLM-L12-H384-uncased",
"base_model:finetune:microsoft/MiniLM-L12-H384-uncased",
"license:mit",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| text-classification | 2023-11-10T15:14:39Z | ---
license: mit
base_model: microsoft/MiniLM-L12-H384-uncased
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: MiniLM_uncased_classification_tools_classifier-only_fr
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# MiniLM_uncased_classification_tools_classifier-only_fr
This model is a fine-tuned version of [microsoft/MiniLM-L12-H384-uncased](https://huggingface.co/microsoft/MiniLM-L12-H384-uncased) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 2.0829
- Accuracy: 0.075
- Learning Rate: 0.0001
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0001
- train_batch_size: 24
- eval_batch_size: 192
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 60
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Rate |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:------:|
| No log | 1.0 | 7 | 2.0791 | 0.125 | 0.0001 |
| No log | 2.0 | 14 | 2.0797 | 0.075 | 0.0001 |
| No log | 3.0 | 21 | 2.0799 | 0.075 | 0.0001 |
| No log | 4.0 | 28 | 2.0804 | 0.075 | 0.0001 |
| No log | 5.0 | 35 | 2.0808 | 0.075 | 0.0001 |
| No log | 6.0 | 42 | 2.0813 | 0.075 | 9e-05 |
| No log | 7.0 | 49 | 2.0818 | 0.075 | 0.0001 |
| No log | 8.0 | 56 | 2.0820 | 0.075 | 0.0001 |
| No log | 9.0 | 63 | 2.0822 | 0.075 | 0.0001 |
| No log | 10.0 | 70 | 2.0827 | 0.075 | 0.0001 |
| No log | 11.0 | 77 | 2.0829 | 0.075 | 0.0001 |
### Framework versions
- Transformers 4.34.0
- Pytorch 2.0.1+cu117
- Datasets 2.14.5
- Tokenizers 0.14.1
|
Leventiir/Reinforce-Pixelcopter-PLE-v0 | Leventiir | 2023-11-10T16:37:27Z | 0 | 0 | null | [
"Pixelcopter-PLE-v0",
"reinforce",
"reinforcement-learning",
"custom-implementation",
"deep-rl-class",
"model-index",
"region:us"
]
| reinforcement-learning | 2023-11-10T16:37:22Z | ---
tags:
- Pixelcopter-PLE-v0
- reinforce
- reinforcement-learning
- custom-implementation
- deep-rl-class
model-index:
- name: Reinforce-Pixelcopter-PLE-v0
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: Pixelcopter-PLE-v0
type: Pixelcopter-PLE-v0
metrics:
- type: mean_reward
value: 41.10 +/- 24.21
name: mean_reward
verified: false
---
# **Reinforce** Agent playing **Pixelcopter-PLE-v0**
This is a trained model of a **Reinforce** agent playing **Pixelcopter-PLE-v0** .
To learn to use this model and train yours check Unit 4 of the Deep Reinforcement Learning Course: https://huggingface.co/deep-rl-course/unit4/introduction
|
minhtoan/gpt3-small-finetune-cnndaily-news | minhtoan | 2023-11-10T16:34:27Z | 51 | 11 | transformers | [
"transformers",
"pytorch",
"gpt_neo",
"text-generation",
"en",
"english",
"gpt2",
"gpt3",
"lm",
"nlp",
"dataset:cnn_dailymail",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
]
| text-generation | 2023-02-25T11:37:41Z | ---
language:
- en
tags:
- en
- english
- gpt2
- gpt3
- text-generation
- lm
- nlp
datasets:
- cnn_dailymail
widget:
- text: "Ever noticed how plane seats appear to be getting smaller and smaller? "
inference:
parameters:
max_length: 120
do_sample: True
temperature: 0.8
---
# GPT-3 small
Pretrained GPT-3 small, continuing the development of GPT NEO, with architecture that purposefully mimics that of GPT-3, model was trained on CNN Daily Mail News dataset for text generation.
# How to use the model
~~~~
from transformers import GPT2Tokenizer, GPTNeoForCausalLM
tokenizer = GPT2Tokenizer.from_pretrained('minhtoan/gpt3-small-finetune-cnndaily-news')
model = GPTNeoForCausalLM.from_pretrained('minhtoan/gpt3-small-finetune-cnndaily-news')
text = "Ever noticed how plane seats appear to be getting smaller and smaller? "
input_ids = tokenizer.encode(text, return_tensors='pt')
max_length = 150
sample_outputs = model.generate(input_ids, do_sample=True, max_length=max_length,temperature = 0.8)
for i, sample_output in enumerate(sample_outputs):
print(">> Generated text {}\n\n{}".format(i+1, tokenizer.decode(sample_output.tolist())))
print('\n---')
~~~~
## Author
`
Phan Minh Toan
` |
panverz/trans | panverz | 2023-11-10T16:26:15Z | 0 | 0 | transformers | [
"transformers",
"text-classification",
"en",
"license:mit",
"endpoints_compatible",
"region:us"
]
| text-classification | 2023-11-10T16:22:20Z | ---
license: mit
language:
- en
library_name: transformers
pipeline_tag: text-classification
--- |
Subsets and Splits