modelId
stringlengths 5
139
| author
stringlengths 2
42
| last_modified
timestamp[us, tz=UTC]date 2020-02-15 11:33:14
2025-06-29 00:46:34
| downloads
int64 0
223M
| likes
int64 0
11.7k
| library_name
stringclasses 502
values | tags
sequencelengths 1
4.05k
| pipeline_tag
stringclasses 54
values | createdAt
timestamp[us, tz=UTC]date 2022-03-02 23:29:04
2025-06-29 00:44:25
| card
stringlengths 11
1.01M
|
|---|---|---|---|---|---|---|---|---|---|
felixZzz/wlen6_61k-orz-ours-d1-len3000-0428T03_05_50-step_01360 | felixZzz | 2025-05-04T23:54:27Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"qwen2",
"text-generation",
"conversational",
"arxiv:1910.09700",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T23:42:00Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
dgambettaphd/M_llm2_gen6_WXS_doc1000_synt64_lr1e-04_acm_MPP | dgambettaphd | 2025-05-04T23:54:22Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"unsloth",
"arxiv:1910.09700",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:54:07Z | ---
library_name: transformers
tags:
- unsloth
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
dzanbek/ba108559-9492-45f6-bfc9-8c9f648e03c9 | dzanbek | 2025-05-04T23:51:25Z | 0 | 0 | peft | [
"peft",
"safetensors",
"llama",
"axolotl",
"generated_from_trainer",
"base_model:jingyeom/seal3.1.6n_7b",
"base_model:adapter:jingyeom/seal3.1.6n_7b",
"8-bit",
"bitsandbytes",
"region:us"
] | null | 2025-05-04T23:27:21Z | ---
library_name: peft
base_model: jingyeom/seal3.1.6n_7b
tags:
- axolotl
- generated_from_trainer
model-index:
- name: ba108559-9492-45f6-bfc9-8c9f648e03c9
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. -->
[<img src="https://raw.githubusercontent.com/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/axolotl-ai-cloud/axolotl)
<details><summary>See axolotl config</summary>
axolotl version: `0.4.1`
```yaml
absolute_data_files: false
adapter: lora
base_model: jingyeom/seal3.1.6n_7b
bf16: true
chat_template: llama3
dataset_prepared_path: /workspace/axolotl
datasets:
- data_files:
- d5441eb887c4a86b_train_data.json
ds_type: json
format: custom
path: /workspace/input_data/d5441eb887c4a86b_train_data.json
type:
field_input: emotion
field_instruction: user_status
field_output: user_persona
format: '{instruction} {input}'
no_input_format: '{instruction}'
system_format: '{system}'
system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: null
eval_max_new_tokens: 128
eval_table_size: null
evals_per_epoch: 1
flash_attention: true
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 1
gradient_checkpointing: true
gradient_clipping: 0.5
group_by_length: false
hub_model_id: dzanbek/ba108559-9492-45f6-bfc9-8c9f648e03c9
hub_repo: null
hub_strategy: end
hub_token: null
learning_rate: 5.0e-06
load_in_4bit: true
load_in_8bit: true
local_rank: null
logging_steps: 1
lora_alpha: 64
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 32
lora_target_linear: true
lr_scheduler: cosine
max_steps: 400
micro_batch_size: 8
mixed_precision: bf16
mlflow_experiment_name: /tmp/d5441eb887c4a86b_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 1
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
saves_per_epoch: 1
sequence_len: 1024
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.05
wandb_entity: null
wandb_mode: online
wandb_name: 76767355-9056-4f08-ac12-edf293bc4536
wandb_project: s56-2
wandb_run: your_name
wandb_runid: 76767355-9056-4f08-ac12-edf293bc4536
warmup_steps: 20
weight_decay: 0.01
xformers_attention: true
```
</details><br>
# ba108559-9492-45f6-bfc9-8c9f648e03c9
This model is a fine-tuned version of [jingyeom/seal3.1.6n_7b](https://huggingface.co/jingyeom/seal3.1.6n_7b) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.8759
## 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-06
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_BNB with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 20
- training_steps: 400
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:------:|:----:|:---------------:|
| 0.9194 | 0.2823 | 400 | 0.8759 |
### Framework versions
- PEFT 0.13.2
- Transformers 4.46.0
- Pytorch 2.5.0+cu124
- Datasets 3.0.1
- Tokenizers 0.20.1 |
ClaudioItaly/gemma-2-2b-it-peft-lora-merged-F16-GGUF | ClaudioItaly | 2025-05-04T23:51:14Z | 0 | 0 | transformers | [
"transformers",
"gguf",
"llama-cpp",
"gguf-my-lora",
"base_model:NanoMatriX/gemma-2-2b-it-peft-lora-merged",
"base_model:quantized:NanoMatriX/gemma-2-2b-it-peft-lora-merged",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:51:10Z | ---
library_name: transformers
tags:
- llama-cpp
- gguf-my-lora
base_model: NanoMatriX/gemma-2-2b-it-peft-lora-merged
---
# ClaudioItaly/gemma-2-2b-it-peft-lora-merged-F16-GGUF
This LoRA adapter was converted to GGUF format from [`NanoMatriX/gemma-2-2b-it-peft-lora-merged`](https://huggingface.co/NanoMatriX/gemma-2-2b-it-peft-lora-merged) via the ggml.ai's [GGUF-my-lora](https://huggingface.co/spaces/ggml-org/gguf-my-lora) space.
Refer to the [original adapter repository](https://huggingface.co/NanoMatriX/gemma-2-2b-it-peft-lora-merged) for more details.
## Use with llama.cpp
```bash
# with cli
llama-cli -m base_model.gguf --lora gemma-2-2b-it-peft-lora-merged-f16.gguf (...other args)
# with server
llama-server -m base_model.gguf --lora gemma-2-2b-it-peft-lora-merged-f16.gguf (...other args)
```
To know more about LoRA usage with llama.cpp server, refer to the [llama.cpp server documentation](https://github.com/ggerganov/llama.cpp/blob/master/examples/server/README.md).
|
fffanx/Llama-3.2-1B-Instruct-GRPO-agent19_E10 | fffanx | 2025-05-04T23:50:51Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:50:23Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent19_E10
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent19_E10
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent19_E10", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent17_E10 | fffanx | 2025-05-04T23:49:47Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:49:20Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent17_E10
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent17_E10
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent17_E10", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent14_E10 | fffanx | 2025-05-04T23:48:15Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:47:47Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent14_E10
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent14_E10
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent14_E10", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent13_E10 | fffanx | 2025-05-04T23:47:44Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:47:16Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent13_E10
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent13_E10
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent13_E10", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent12_E10 | fffanx | 2025-05-04T23:47:13Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:46:45Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent12_E10
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent12_E10
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent12_E10", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent10_E10 | fffanx | 2025-05-04T23:46:10Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:45:41Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent10_E10
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent10_E10
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent10_E10", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent9_E10 | fffanx | 2025-05-04T23:45:38Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:45:09Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent9_E10
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent9_E10
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent9_E10", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent8_E10 | fffanx | 2025-05-04T23:45:06Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:44:38Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent8_E10
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent8_E10
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent8_E10", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
dimasik1987/fca84cde-5cad-4622-bd9c-a71f2c192077 | dimasik1987 | 2025-05-04T23:45:03Z | 0 | 0 | peft | [
"peft",
"safetensors",
"llama",
"axolotl",
"generated_from_trainer",
"base_model:jingyeom/seal3.1.6n_7b",
"base_model:adapter:jingyeom/seal3.1.6n_7b",
"4-bit",
"bitsandbytes",
"region:us"
] | null | 2025-05-04T23:27:50Z | ---
library_name: peft
base_model: jingyeom/seal3.1.6n_7b
tags:
- axolotl
- generated_from_trainer
model-index:
- name: fca84cde-5cad-4622-bd9c-a71f2c192077
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. -->
[<img src="https://raw.githubusercontent.com/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/axolotl-ai-cloud/axolotl)
<details><summary>See axolotl config</summary>
axolotl version: `0.4.1`
```yaml
absolute_data_files: false
adapter: lora
base_model: jingyeom/seal3.1.6n_7b
bf16: true
chat_template: llama3
dataset_prepared_path: /workspace/axolotl
datasets:
- data_files:
- d5441eb887c4a86b_train_data.json
ds_type: json
format: custom
path: /workspace/input_data/d5441eb887c4a86b_train_data.json
type:
field_input: emotion
field_instruction: user_status
field_output: user_persona
format: '{instruction} {input}'
no_input_format: '{instruction}'
system_format: '{system}'
system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: null
eval_max_new_tokens: 128
eval_table_size: null
evals_per_epoch: 1
flash_attention: true
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 1
gradient_checkpointing: true
gradient_clipping: 0.55
group_by_length: false
hub_model_id: dimasik1987/fca84cde-5cad-4622-bd9c-a71f2c192077
hub_repo: null
hub_strategy: end
hub_token: null
learning_rate: 1.0e-06
load_in_4bit: true
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 64
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 32
lora_target_linear: true
lr_scheduler: cosine
max_steps: 350
micro_batch_size: 10
mixed_precision: bf16
mlflow_experiment_name: /tmp/d5441eb887c4a86b_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 1
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
saves_per_epoch: 1
sequence_len: 2048
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.05
wandb_entity: null
wandb_mode: online
wandb_name: 76767355-9056-4f08-ac12-edf293bc4536
wandb_project: s56-7
wandb_run: your_name
wandb_runid: 76767355-9056-4f08-ac12-edf293bc4536
warmup_steps: 15
weight_decay: 0.01
xformers_attention: true
```
</details><br>
# fca84cde-5cad-4622-bd9c-a71f2c192077
This model is a fine-tuned version of [jingyeom/seal3.1.6n_7b](https://huggingface.co/jingyeom/seal3.1.6n_7b) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.3897
## 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-06
- train_batch_size: 10
- eval_batch_size: 10
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_BNB with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 15
- training_steps: 350
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:------:|:----:|:---------------:|
| 1.3781 | 0.3086 | 350 | 1.3897 |
### Framework versions
- PEFT 0.13.2
- Transformers 4.46.0
- Pytorch 2.5.0+cu124
- Datasets 3.0.1
- Tokenizers 0.20.1 |
ftefaan/detector_q4km | ftefaan | 2025-05-04T23:44:05Z | 0 | 0 | transformers | [
"transformers",
"gguf",
"llama",
"text-generation-inference",
"unsloth",
"en",
"base_model:unsloth/Llama-3.2-3B-Instruct-unsloth-bnb-4bit",
"base_model:quantized:unsloth/Llama-3.2-3B-Instruct-unsloth-bnb-4bit",
"license:apache-2.0",
"endpoints_compatible",
"region:us",
"conversational"
] | null | 2025-05-04T23:43:30Z | ---
base_model: unsloth/Llama-3.2-3B-Instruct-unsloth-bnb-4bit
tags:
- text-generation-inference
- transformers
- unsloth
- llama
- gguf
license: apache-2.0
language:
- en
---
# Uploaded model
- **Developed by:** ftefaan
- **License:** apache-2.0
- **Finetuned from model :** unsloth/Llama-3.2-3B-Instruct-unsloth-bnb-4bit
This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
|
Flo0620/Qwen2_5_7B_r128_a256_d0_1 | Flo0620 | 2025-05-04T23:43:44Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"generated_from_trainer",
"trl",
"sft",
"base_model:Qwen/Qwen2.5-VL-7B-Instruct",
"base_model:finetune:Qwen/Qwen2.5-VL-7B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-04-24T12:57:12Z | ---
base_model: Qwen/Qwen2.5-VL-7B-Instruct
library_name: transformers
model_name: Qwen2_5_7B_r128_a256_d0_1
tags:
- generated_from_trainer
- trl
- sft
licence: license
---
# Model Card for Qwen2_5_7B_r128_a256_d0_1
This model is a fine-tuned version of [Qwen/Qwen2.5-VL-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct).
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="Flo0620/Qwen2_5_7B_r128_a256_d0_1", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with SFT.
### Framework versions
- TRL: 0.15.2
- Transformers: 4.52.0.dev0
- Pytorch: 2.6.0+cu124
- Datasets: 3.5.0
- Tokenizers: 0.21.1
## Citations
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent4_E10 | fffanx | 2025-05-04T23:43:00Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:42:31Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent4_E10
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent4_E10
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent4_E10", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent3_E10 | fffanx | 2025-05-04T23:42:28Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:41:59Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent3_E10
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent3_E10
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent3_E10", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent2_E10 | fffanx | 2025-05-04T23:41:56Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:41:27Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent2_E10
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent2_E10
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent2_E10", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent0_E10 | fffanx | 2025-05-04T23:40:52Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:40:22Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent0_E10
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent0_E10
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent0_E10", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
hi-go/distilbert-base-uncased-finetuned-emotion | hi-go | 2025-05-04T23:35:00Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"distilbert",
"text-classification",
"generated_from_trainer",
"base_model:distilbert/distilbert-base-uncased",
"base_model:finetune:distilbert/distilbert-base-uncased",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] | text-classification | 2025-05-04T12:40:15Z | ---
library_name: transformers
license: apache-2.0
base_model: distilbert-base-uncased
tags:
- generated_from_trainer
model-index:
- name: distilbert-base-uncased-finetuned-emotion
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. -->
# distilbert-base-uncased-finetuned-emotion
This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) 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: 2e-05
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- num_epochs: 2
### Training results
### Framework versions
- Transformers 4.51.3
- Pytorch 2.6.0+cu124
- Datasets 3.5.1
- Tokenizers 0.21.1
|
fffanx/Llama-3.2-1B-Instruct-GRPO-agent19_E9 | fffanx | 2025-05-04T23:33:58Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:33:30Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent19_E9
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent19_E9
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent19_E9", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
JunseongLEEE/Llama-3.2-1B-unsloth-bnb-4bit-ko-wiki | JunseongLEEE | 2025-05-04T23:32:25Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"text-generation-inference",
"unsloth",
"llama",
"trl",
"en",
"base_model:unsloth/Llama-3.2-1B-unsloth-bnb-4bit",
"base_model:finetune:unsloth/Llama-3.2-1B-unsloth-bnb-4bit",
"license:apache-2.0",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:30:28Z | ---
base_model: unsloth/Llama-3.2-1B-unsloth-bnb-4bit
tags:
- text-generation-inference
- transformers
- unsloth
- llama
- trl
license: apache-2.0
language:
- en
---
# Uploaded model
- **Developed by:** JunseongLEEE
- **License:** apache-2.0
- **Finetuned from model :** unsloth/Llama-3.2-1B-unsloth-bnb-4bit
This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
|
felixZzz/wlen6_61k-orz-ours-d1-len3000-0428T03_05_50-step_01232 | felixZzz | 2025-05-04T23:31:40Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"qwen2",
"text-generation",
"conversational",
"arxiv:1910.09700",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T23:23:03Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
hamedkharazmi/Qwen2.5-0.5B-Instruct-Gensyn-Swarm-swift_thorny_zebra | hamedkharazmi | 2025-05-04T23:30:39Z | 20 | 0 | transformers | [
"transformers",
"safetensors",
"qwen2",
"text-generation",
"generated_from_trainer",
"rl-swarm",
"grpo",
"gensyn",
"I am swift thorny zebra",
"trl",
"conversational",
"arxiv:2402.03300",
"base_model:unsloth/Qwen2.5-0.5B-Instruct",
"base_model:finetune:unsloth/Qwen2.5-0.5B-Instruct",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-04-09T00:55:26Z | ---
base_model: unsloth/Qwen2.5-0.5B-Instruct
library_name: transformers
model_name: Qwen2.5-0.5B-Instruct-Gensyn-Swarm-swift_thorny_zebra
tags:
- generated_from_trainer
- rl-swarm
- grpo
- gensyn
- I am swift thorny zebra
- trl
licence: license
---
# Model Card for Qwen2.5-0.5B-Instruct-Gensyn-Swarm-swift_thorny_zebra
This model is a fine-tuned version of [unsloth/Qwen2.5-0.5B-Instruct](https://huggingface.co/unsloth/Qwen2.5-0.5B-Instruct).
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="hamedkharazmi/Qwen2.5-0.5B-Instruct-Gensyn-Swarm-swift_thorny_zebra", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0
- Transformers: 4.51.3
- Pytorch: 2.7.0
- Datasets: 3.5.1
- Tokenizers: 0.21.1
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent12_E9 | fffanx | 2025-05-04T23:30:12Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:29:41Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent12_E9
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent12_E9
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent12_E9", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
mlfoundations-dev/d1_math_fasttext_1k | mlfoundations-dev | 2025-05-04T23:27:12Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"qwen2",
"text-generation",
"llama-factory",
"full",
"generated_from_trainer",
"conversational",
"base_model:Qwen/Qwen2.5-7B-Instruct",
"base_model:finetune:Qwen/Qwen2.5-7B-Instruct",
"license:apache-2.0",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T20:58:44Z | ---
library_name: transformers
license: apache-2.0
base_model: Qwen/Qwen2.5-7B-Instruct
tags:
- llama-factory
- full
- generated_from_trainer
model-index:
- name: d1_math_fasttext_1k
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. -->
# d1_math_fasttext_1k
This model is a fine-tuned version of [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct) on the mlfoundations-dev/d1_math_fasttext_1k 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: 1
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 24
- total_train_batch_size: 96
- total_eval_batch_size: 32
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 7.0
### Training results
### Framework versions
- Transformers 4.46.1
- Pytorch 2.6.0+cu124
- Datasets 3.1.0
- Tokenizers 0.20.3
|
CrimsonZockt/CinthiaCruz-FLUXLORA | CrimsonZockt | 2025-05-04T23:27:07Z | 0 | 0 | diffusers | [
"diffusers",
"text-to-image",
"lora",
"template:diffusion-lora",
"base_model:black-forest-labs/FLUX.1-dev",
"base_model:adapter:black-forest-labs/FLUX.1-dev",
"region:us"
] | text-to-image | 2025-05-04T23:26:37Z | ---
tags:
- text-to-image
- lora
- diffusers
- template:diffusion-lora
widget:
- text: >-
photoshoot of Cinthia Cruz, female, woman, solo, black tanktop, professional
headshot.
output:
url: images/photoshoot of Cinthia Cruz, female, woman, solo....png
base_model: black-forest-labs/FLUX.1-dev
instance_prompt: Cinthia Cruz
---
# CinthiaCruz
<Gallery />
## Model description
This is a LORA Model that i have train on Weights.gg
## Trigger words
You should use `Cinthia Cruz` to trigger the image generation.
## Download model
Weights for this model are available in Safetensors format.
[Download](/CrimsonZockt/CinthiaCruz-FLUXLORA/tree/main) them in the Files & versions tab.
|
DevQuasar/huihui-ai.Qwen3-4B-abliterated-GGUF | DevQuasar | 2025-05-04T23:27:02Z | 0 | 0 | null | [
"gguf",
"text-generation",
"base_model:huihui-ai/Qwen3-4B-abliterated",
"base_model:quantized:huihui-ai/Qwen3-4B-abliterated",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T22:55:17Z | ---
base_model:
- huihui-ai/Qwen3-4B-abliterated
pipeline_tag: text-generation
---
[<img src="https://raw.githubusercontent.com/csabakecskemeti/devquasar/main/dq_logo_black-transparent.png" width="200"/>](https://devquasar.com)
Quantized version of: [huihui-ai/Qwen3-4B-abliterated](https://huggingface.co/huihui-ai/Qwen3-4B-abliterated)
'Make knowledge free for everyone'
<p align="center">
Made with <br>
<a href="https://www.civo.com/" target="_blank">
<img src="https://www.civo.com/assets/public/brand-assets/civo-logo-colour-60cc1622dedf346f7afde1fff760523f731b0aac106a5465af98ff4073114b74.svg" width="100"/>
</a>
</p>
<a href='https://ko-fi.com/L4L416YX7C' target='_blank'><img height='36' style='border:0px;height:36px;' src='https://storage.ko-fi.com/cdn/kofi6.png?v=6' border='0' alt='Buy Me a Coffee at ko-fi.com' /></a>
|
fffanx/Llama-3.2-1B-Instruct-GRPO-agent5_E9 | fffanx | 2025-05-04T23:26:29Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:26:00Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent5_E9
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent5_E9
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent5_E9", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
mradermacher/R1-Code-Interpreter-14B-SFT-GGUF | mradermacher | 2025-05-04T23:25:43Z | 0 | 0 | transformers | [
"transformers",
"gguf",
"en",
"base_model:yongchao98/R1-Code-Interpreter-14B-SFT",
"base_model:quantized:yongchao98/R1-Code-Interpreter-14B-SFT",
"license:mit",
"endpoints_compatible",
"region:us",
"conversational"
] | null | 2025-05-04T22:47:22Z | ---
base_model: yongchao98/R1-Code-Interpreter-14B-SFT
language:
- en
library_name: transformers
license: mit
quantized_by: mradermacher
---
## About
<!-- ### quantize_version: 2 -->
<!-- ### output_tensor_quantised: 1 -->
<!-- ### convert_type: hf -->
<!-- ### vocab_type: -->
<!-- ### tags: -->
static quants of https://huggingface.co/yongchao98/R1-Code-Interpreter-14B-SFT
<!-- provided-files -->
weighted/imatrix quants seem not to be available (by me) at this time. If they do not show up a week or so after the static ones, I have probably not planned for them. Feel free to request them by opening a Community Discussion.
## Usage
If you are unsure how to use GGUF files, refer to one of [TheBloke's
READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
more details, including on how to concatenate multi-part files.
## Provided Quants
(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
| Link | Type | Size/GB | Notes |
|:-----|:-----|--------:|:------|
| [GGUF](https://huggingface.co/mradermacher/R1-Code-Interpreter-14B-SFT-GGUF/resolve/main/R1-Code-Interpreter-14B-SFT.Q2_K.gguf) | Q2_K | 5.9 | |
| [GGUF](https://huggingface.co/mradermacher/R1-Code-Interpreter-14B-SFT-GGUF/resolve/main/R1-Code-Interpreter-14B-SFT.Q3_K_S.gguf) | Q3_K_S | 6.8 | |
| [GGUF](https://huggingface.co/mradermacher/R1-Code-Interpreter-14B-SFT-GGUF/resolve/main/R1-Code-Interpreter-14B-SFT.Q3_K_M.gguf) | Q3_K_M | 7.4 | lower quality |
| [GGUF](https://huggingface.co/mradermacher/R1-Code-Interpreter-14B-SFT-GGUF/resolve/main/R1-Code-Interpreter-14B-SFT.Q3_K_L.gguf) | Q3_K_L | 8.0 | |
| [GGUF](https://huggingface.co/mradermacher/R1-Code-Interpreter-14B-SFT-GGUF/resolve/main/R1-Code-Interpreter-14B-SFT.IQ4_XS.gguf) | IQ4_XS | 8.3 | |
| [GGUF](https://huggingface.co/mradermacher/R1-Code-Interpreter-14B-SFT-GGUF/resolve/main/R1-Code-Interpreter-14B-SFT.Q4_K_S.gguf) | Q4_K_S | 8.7 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/R1-Code-Interpreter-14B-SFT-GGUF/resolve/main/R1-Code-Interpreter-14B-SFT.Q4_K_M.gguf) | Q4_K_M | 9.1 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/R1-Code-Interpreter-14B-SFT-GGUF/resolve/main/R1-Code-Interpreter-14B-SFT.Q5_K_S.gguf) | Q5_K_S | 10.4 | |
| [GGUF](https://huggingface.co/mradermacher/R1-Code-Interpreter-14B-SFT-GGUF/resolve/main/R1-Code-Interpreter-14B-SFT.Q5_K_M.gguf) | Q5_K_M | 10.6 | |
| [GGUF](https://huggingface.co/mradermacher/R1-Code-Interpreter-14B-SFT-GGUF/resolve/main/R1-Code-Interpreter-14B-SFT.Q6_K.gguf) | Q6_K | 12.2 | very good quality |
| [GGUF](https://huggingface.co/mradermacher/R1-Code-Interpreter-14B-SFT-GGUF/resolve/main/R1-Code-Interpreter-14B-SFT.Q8_0.gguf) | Q8_0 | 15.8 | fast, best quality |
Here is a handy graph by ikawrakow comparing some lower-quality quant
types (lower is better):
And here are Artefact2's thoughts on the matter:
https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
## FAQ / Model Request
See https://huggingface.co/mradermacher/model_requests for some answers to
questions you might have and/or if you want some other model quantized.
## Thanks
I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
me use its servers and providing upgrades to my workstation to enable
this work in my free time.
<!-- end -->
|
fffanx/Llama-3.2-1B-Instruct-GRPO-agent1_E9 | fffanx | 2025-05-04T23:24:19Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:23:49Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent1_E9
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent1_E9
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent1_E9", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
Sidou10major/Quran_reciter | Sidou10major | 2025-05-04T23:23:24Z | 0 | 0 | null | [
"pytorch",
"whisper",
"license:apache-2.0",
"region:us"
] | null | 2025-05-04T22:16:50Z | ---
license: apache-2.0
---
|
YOYO-AI/YOYO-O1-32B-V4-preview1 | YOYO-AI | 2025-05-04T23:18:27Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"qwen2",
"text-generation",
"mergekit",
"merge",
"conversational",
"arxiv:2408.07990",
"base_model:Qwen/QwQ-32B",
"base_model:merge:Qwen/QwQ-32B",
"base_model:Qwen/Qwen2.5-32B",
"base_model:merge:Qwen/Qwen2.5-32B",
"base_model:qihoo360/Light-R1-32B",
"base_model:merge:qihoo360/Light-R1-32B",
"base_model:qihoo360/Light-R1-32B-DS",
"base_model:merge:qihoo360/Light-R1-32B-DS",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T23:02:13Z | ---
base_model:
- Qwen/QwQ-32B
- Qwen/Qwen2.5-32B
- qihoo360/Light-R1-32B-DS
- qihoo360/Light-R1-32B
library_name: transformers
tags:
- mergekit
- merge
---
# merge
This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
## Merge Details
### Merge Method
This model was merged using the [SCE](https://arxiv.org/abs/2408.07990) merge method using [Qwen/Qwen2.5-32B](https://huggingface.co/Qwen/Qwen2.5-32B) as a base.
### Models Merged
The following models were included in the merge:
* [Qwen/QwQ-32B](https://huggingface.co/Qwen/QwQ-32B)
* [qihoo360/Light-R1-32B-DS](https://huggingface.co/qihoo360/Light-R1-32B-DS)
* [qihoo360/Light-R1-32B](https://huggingface.co/qihoo360/Light-R1-32B)
### Configuration
The following YAML configuration was used to produce this model:
```yaml
merge_method: sce
models:
# Pivot model
- model: Qwen/Qwen2.5-32B
# Target models
- model: qihoo360/Light-R1-32B
- model: Qwen/QwQ-32B
- model: qihoo360/Light-R1-32B-DS
base_model: Qwen/Qwen2.5-32B
parameters:
select_topk: 1
dtype: bfloat16
tokenizer_source: Qwen/QwQ-32B
normalize: true
int8_mask: true
```
|
fffanx/Llama-3.2-1B-Instruct-GRPO-agent19_E8 | fffanx | 2025-05-04T23:17:00Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:16:32Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent19_E8
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent19_E8
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent19_E8", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent16_E8 | fffanx | 2025-05-04T23:15:26Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:14:58Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent16_E8
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent16_E8
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent16_E8", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent12_E8 | fffanx | 2025-05-04T23:13:20Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:12:53Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent12_E8
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent12_E8
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent12_E8", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
DavidAU/Qwen3-30B-A7.5B-24-Grand-Brainstorm | DavidAU | 2025-05-04T23:13:15Z | 3 | 1 | transformers | [
"transformers",
"safetensors",
"qwen3_moe",
"text-generation",
"32 k context",
"reasoning",
"thinking",
"qwen3",
"24 experts",
"conversational",
"base_model:Qwen/Qwen3-30B-A3B-Base",
"base_model:finetune:Qwen/Qwen3-30B-A3B-Base",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T04:42:30Z | ---
library_name: transformers
pipeline_tag: text-generation
tags:
- 32 k context
- reasoning
- thinking
- qwen3
- 24 experts
base_model:
- Qwen/Qwen3-30B-A3B-Base
---
<h2>Qwen3-30B-A7.5B-24-Grand-Brainstorm</h2>
<img src="star-wars-hans-solo.gif" style="float:right; padding:10px;">
This repo contains the full precision source code, in "safe tensors" format to generate GGUFs, GPTQ, EXL2, AWQ, HQQ and other formats. The source code can also be used directly.
This is a simple "finetune" of the Qwen's "Qwen 30B-A3B" (MOE) model, setting the experts in use from 8 to 24 (out of 128 experts).
THREE Example generations (Q4KS, CPU) at the bottom of this page using 24 experts / this model.
This reduces the speed of the model, but uses more "experts" to process your prompts and uses 7.5B (of 30B) parameters instead of 3B (of 30B) parameters. Depending on the application you may want to
use the regular model ("30B-A3B"), and use this model for REALLY REALLY COMPLEX / "DEEPER" (IE Nuance) use case(s).
Regular or simpler use cases may benefit from using the normal (8 experts), the "12 cooks" (12 experts), 16 Extreme (32k or 128k context) or "High-Speed" (4 experts) version(s).
Using 24 experts instead of the default 8 will slow down token/second speeds about about 2/3 or so.
This version, with 24 experts, may be overkill for some use cases.
I suggest higher temps too (1+, 2+), for some use cases.
Also ; I suggest trying the 4, 8 (reg), 12, and 16 versions as these will be faster - token / second, and may meet your use case(s) better.
Context size: 32K + 8K for output (40k total)
Use Jinja Template or CHATML template.
IMPORTANT NOTES:
- Due to the unique nature (MOE, Size, Activated experts, size of experts) of this model GGUF quants can be run on the CPU, GPU or with GPU part "off-load", right up to full precision.
- This model is difficult to Imatrix : You need a much larger imatrix file / multi-language / multi-content (ie code/text) to imatrix it.
- GPU speeds will be BLISTERING 4x-8x or higher than CPU only speeds AND this model will be BLISTERING too, relative to other "30B" models (Token per second speed equal roughly to 7.5B "normal" model speeds).
Please refer the org model card for details, benchmarks, how to use, settings, system roles etc etc :
[ https://huggingface.co/Qwen/Qwen3-30B-A3B ]
<B>More / Less Experts Versions:</B>
4 experts:
[ https://huggingface.co/DavidAU/Qwen3-30B-A1.5B-High-Speed ]
12 experts:
[ https://huggingface.co/DavidAU/Qwen3-30B-A4.5B-12-Cooks ]
16 experts, 32k context:
[ https://huggingface.co/DavidAU/Qwen3-30B-A6B-16-Extreme ]
16 experts, 128k context:
[ https://huggingface.co/DavidAU/Qwen3-30B-A6B-16-Extreme-128k-context ]
<B>OPTIONAL SYSTEM ROLE: </B>
You may or may not need this, as most times Qwen3s generate their own reasoning/thinking blocks.
```
You are a deep thinking AI, you may use extremely long chains of thought to deeply consider the problem and deliberate with yourself via systematic reasoning processes to help come to a correct solution prior to answering. You should enclose your thoughts and internal monologue inside <think> </think> tags, and then provide your solution or response to the problem.
```
See document "Maximizing-Model-Performance-All..." below for how to "set" system role in various LLM/AI apps below.
IMPORTANT: Highest Quality Settings / Optimal Operation Guide / Parameters and Samplers
If you are going to use this model, (source, GGUF or a different quant), please review this document for critical parameter, sampler and advance sampler settings (for multiple AI/LLM aps).
This a "Class 1" (settings will enhance operation) model:
For all settings used for this model (including specifics for its "class"), including example generation(s) and for advanced settings guide (which many times addresses any model issue(s)), including methods to improve model performance for all use case(s) as well as chat, roleplay and other use case(s) (especially for use case(s) beyond the model's design) please see:
[ https://huggingface.co/DavidAU/Maximizing-Model-Performance-All-Quants-Types-And-Full-Precision-by-Samplers_Parameters ]
REASON:
Regardless of "model class" this document will detail methods to enhance operations.
If the model is a Class 3/4 model the default settings (parameters, samplers, advanced samplers) must be set for "use case(s)" uses correctly. Some AI/LLM apps DO NOT have consistant default setting(s) which result in sub-par model operation. Like wise for Class 3/4 models (which operate somewhat to very differently than standard models) additional samplers and advanced samplers settings are required to "smooth out" operation, AND/OR also allow full operation for use cases the model was not designed for.
BONUS - Use these settings for ANY model, ANY repo, ANY quant (including source/full precision):
This document also details parameters, sampler and advanced samplers that can be use FOR ANY MODEL, FROM ANY REPO too - all quants, and of course source code operation too - to enhance the operation of any model.
[ https://huggingface.co/DavidAU/Maximizing-Model-Performance-All-Quants-Types-And-Full-Precision-by-Samplers_Parameters ]
NOTE:
I strongly suggest you also visit the DavidAU GGUF (below) repo too for more details in using this model ; especially if it is "Class 3" or "Class 4" to get maximum performance from the model.
For full information about this model, including:
- Details about this model and its use case(s).
- Context limits
- Special usage notes / settings.
- Any model(s) used to create this model.
- Template(s) used to access/use this model.
- Example generation(s)
- GGUF quants of this model
Please go to:
[ GGUFS REPO coming soon ]
---
<B>Example Generation #1:</B>
Temp 1.2, rep pen 1.06, rep pen range 64, topk 100, topp .95, minp .05
Q4KS [non-imatrix], CPU only (windows 11), LMSTUDIO.
Speed: 8 T/S (no GPU offload)
This is a mid level quant, expect stronger performance from an Imatrix Q4KS / higher quants / full precision.
NOTE: Some formatting lost on copy/paste.
PROMPT:
<B>
Explain ways to use the "night" time cooling of radiant energy into space to reduce global temperatures.
</B>
<P></P>
<B>
[[[thinking start]]]
</B>
<P></P>
Thought for 6 minutes 18 seconds
Okay, let's try to figure out how to use nighttime cooling of radiant energy into space to reduce global temperatures. Hmm, first I need to understand what this concept means. Radiant energy into space during night time... maybe it's about the Earth emitting heat radiation at night, which then goes into space? So, if we can enhance that process, maybe we can cool the planet more effectively?
But how does that work? The Earth emits infrared radiation from the surface, and some of it escapes into space. If the atmosphere is too thick or has greenhouse gases, this radiation gets trapped. So maybe during nighttime, when there's no sunlight, the Earth can emit more heat? But if we can increase the efficiency of that emission, perhaps that would help cool the planet.
Wait, but global temperatures are rising due to human activities and greenhouse gases. So, perhaps reducing the greenhouse effect during night time? Maybe by adjusting some factors that allow more radiation to escape?
But how can humans influence that? Maybe through technological methods or natural processes. For example, modifying the atmosphere to let more heat escape? Or maybe using reflective materials to enhance radiation? But I need to think of practical ways.
Another thought: during night time, the Earth is not receiving solar radiation, so the surface can cool down. But if we can maximize this cooling, maybe we could offset some of the warming effects during the day? However, this might be a small contribution compared to the overall warming from greenhouse gases.
But how to do that? Maybe by using materials that emit more heat at night. For example, reflective surfaces or structures that enhance thermal radiation. Or maybe altering the atmosphere's composition, like reducing greenhouse gases. But that's not exactly "night time cooling" but rather reducing overall greenhouse effect.
Alternatively, perhaps using some kind of technology that enhances radiative cooling during nighttime. Like, for example, materials that can emit heat more efficiently at night? Maybe using high emissivity surfaces?
But I need to think about possible methods. Let's break it down: the Earth emits infrared radiation into space. The amount depends on the emissivity of surfaces and the atmospheric conditions. If we can increase emissivity or decrease atmospheric absorption, then more energy is radiated away.
So, during nighttime, the Earth's surface is cooler, maybe allowing more heat to be released. But if we can create structures that are high emissivity during nighttime, perhaps they could radiate more energy into space. This could help reduce local temperatures, but globally?
But how would that affect global temperatures? Maybe by cooling certain regions, but it's not directly affecting the overall global temperature. Unless it's a massive scale of implementation.
Another idea: using materials that can emit heat during night time, which then cools the environment. Like, for example, placing high emissivity materials in areas where heat is accumulated, so they radiate more at night, reducing local temperatures. But if this is done on a large scale, maybe it could contribute to global cooling.
But there's another angle: if we can use passive cooling techniques during night time, like creating structures that allow for better radiative cooling, which then reduces the need for active cooling systems during day. Maybe in urban areas, using materials with high emissivity to cool buildings at night, which would help reduce heat retention.
But how about the atmosphere? If the atmospheric layers are less opaque during nighttime, allowing more radiation to escape. But maybe that's not possible. Atmospheric composition is mostly constant, except for some factors like clouds or aerosols. Maybe if during night time, there's less cloud cover, or reduced aerosols, then radiation can escape more.
But human control over these factors might be limited. However, maybe through atmospheric engineering? For example, using technologies to manipulate the atmosphere's composition during nighttime to allow more heat radiation. But that's a speculative idea.
Another approach: using space-based cooling. Maybe if we can somehow reflect or redirect heat radiation into space more effectively. But how? Perhaps using satellites or mirrors?
Wait, but this is maybe more about increasing Earth's ability to emit heat into space. So, the idea is that during nighttime, the Earth can naturally cool by radiating energy, but if we can enhance that process, then it would contribute to global cooling.
So, possible ways:
1. Enhancing emissivity of surfaces on Earth. For example, using materials with high emissivity (like black body) which can radiate more heat at night. If the surface is designed to emit more energy, then it cools down faster. This could reduce local temperatures and maybe contribute to global cooling.
2. Reducing greenhouse gases in the atmosphere. But that's not specifically nighttime, but overall reduction. However, during nighttime, when there is no sunlight, the greenhouse gases might have a different effect? Not sure. Maybe the impact of greenhouse gases is constant, so reducing them would allow more radiation to escape during all times.
3. Using atmospheric engineering to create conditions that allow more heat to escape during night time. For example, maybe using reflective particles or altering cloud cover to let more radiation escape.
4. Creating structures or systems that can actively cool by radiating heat during nighttime. Maybe like thermal emissive surfaces, which are designed to maximize the emission of heat into space.
5. Using energy from nighttime cooling to power systems, but not sure how this would relate to reducing temperatures.
Wait, but maybe another thought: if during nighttime, Earth's surface cools down because it is not receiving solar radiation, and thus can radiate more heat. But if we can use that process, maybe to cool the atmosphere? For example, by allowing more heat to be radiated, which would reduce overall temperature.
But how? Maybe through passive cooling techniques, like creating surfaces that are more emissive during night time. Or perhaps using technologies to capture and use the radiation for something else, but that might not directly lower temperatures.
Alternatively, maybe if we can increase the Earth's ability to emit heat during nighttime, then the overall energy balance would be different. The Earth absorbs solar energy during day, and emits during night. If the emission is more efficient, then the net heat loss is higher, which could contribute to cooling.
But how to achieve that? Maybe by modifying the surface emissivity. For example, using materials that have higher emissivity, which would allow for more thermal radiation during night time.
Another angle: the Earth's radiative cooling is affected by the atmosphere. If the atmosphere has less greenhouse gases, then more heat can escape into space. So, during nighttime, if the atmosphere allows more radiation to escape, then the Earth can cool more. But that's again about reducing greenhouse effect, which is a broader approach.
But maybe there's a specific method for nighttime cooling? For example, using something that works during nighttime to enhance the cooling process. Maybe like, using solar panels or other materials that are active during nighttime? Not sure.
Alternatively, think about the night time as an opportunity for cooling. So, if we can maximize the cooling during night time, which is when the Earth is not receiving energy from sun, then it could help reduce the accumulated heat from daytime. But how?
Maybe by creating systems that can absorb heat during day and release it during night. For example, using materials that store heat and then emit it during nighttime, which would allow the environment to cool. This would be a form of thermal storage and emission.
But this is more like an active process, but maybe not directly reducing global temperatures.
Another thought: if the Earth can radiate more heat into space during night time, then the energy balance would be shifted. If the total radiation is higher, then the overall temperature would decrease. But how to increase that?
Perhaps through increasing emissivity of surfaces or reducing atmospheric opacity. So, maybe using materials that have high emissivity, which would allow for more radiation during night time.
But what are those materials? For example, black bodies with high emissivity, or some engineered materials. Maybe in urban areas, using buildings with high emissivity to cool during nighttime.
Alternatively, using reflective surfaces to reduce absorption during day, but then emit during night. But that's a bit conflicting.
Wait, maybe if during day, the surface is less absorptive (like reflecting more sunlight), which reduces heat accumulation, and during night time, it emits more heat into space. So, combining both daytime reflection and nighttime emission to cool the planet.
But how to do that? Maybe using materials that have high emissivity in the infrared range, which would allow for more radiation during night. And also have low albedo (reflectivity) during day, to reduce solar absorption. So, a dual-purpose material.
But this is possible? For example, materials that are reflective in visible light but emissive in infrared. Maybe like some kind of thermal coatings or structures.
Another idea: using artificial structures that can emit heat during night time. For example, creating a network of emissive surfaces that can radiate heat into space during night time. This could be done on large scale, maybe in deserts or other areas where heat accumulation is high.
But how would this affect global temperatures? If the total emitted energy increases, then the Earth's thermal balance would shift, leading to lower temperatures. But this is a bit speculative.
Another thought: the Earth's radiative cooling is limited by atmospheric absorption. So, during night time, if we can reduce the absorption of radiation by the atmosphere, then more heat is radiated into space. This could be achieved by reducing greenhouse gases or manipulating the atmosphere's composition.
But again, this is about reducing greenhouse effect, which is a broader approach. But maybe during night time, it's more efficient to do so? Not sure.
Alternatively, using space-based methods to allow more radiation. Like, if there are satellites that can reflect heat away from Earth, but not sure how feasible this is.
But maybe another angle: the natural cooling process during night time is a part of the Earth's energy balance. If we can enhance this process, then the overall thermal equilibrium would be shifted towards cooler temperatures.
So, possible methods:
1. Increasing surface emissivity through materials or structures.
2. Reducing greenhouse gases in atmosphere (both day and night).
3. Altering atmospheric conditions during night time to allow more radiation escape (maybe through cloud manipulation or aerosols).
4. Using active cooling systems that operate during night time to release heat into space.
But how to implement these? Maybe for 1, using materials with high emissivity in urban areas, or in regions where heat is accumulated. For 2, reducing greenhouse gases via carbon capture technologies, or emission reduction. For 3, maybe using cloud seeding or other atmospheric engineering techniques.
But I need to think of practical ways. Also, note that the Earth's thermal balance is influenced by both incoming solar radiation and outgoing radiation. If the outgoing radiation increases, then the net heat loss would be higher, leading to cooler temperatures.
But this requires a combination of reducing incoming energy (like solar absorption) and increasing outgoing energy (radiation into space). So, maybe a dual approach.
Another idea: during night time, when the Earth is not receiving sunlight, the cooling process is more efficient. If we can create systems that can store heat during day and release it during night, then the overall temperature would be lower. For example, using thermal storage materials that absorb heat during day, then emit during night.
But this is a form of passive cooling. So, maybe in cities or buildings, using materials that can absorb solar energy during day, then release it during night via emissive surfaces. This could reduce the overall heat retention in the environment.
But how would this affect global temperatures? If such systems are widespread, then the cumulative effect might be significant. However, the impact is indirect, as it's more about local cooling rather than global.
But if the global temperatures are influenced by local cooling effects, then maybe it's possible. But this is a complex system, with many variables.
Another thought: the night time cooling is part of the Earth's natural processes. If we can optimize that, then the overall temperature might decrease. For example, by reducing heat retention during day and allowing more heat to be radiated during night.
But how? Maybe by using materials that have high emissivity during night, which allow more heat to escape. So, if surfaces are designed to emit more heat at night, then the Earth can cool more efficiently.
So, maybe a practical approach would be to use emissive materials in areas where heat is accumulated. For example, in cities, using buildings with high emissivity to radiate heat during night time, which reduces local temperatures and contributes to global cooling.
But this requires large-scale implementation. Maybe it's not feasible in the short term, but as a concept, it's possible.
Another idea: the Earth's thermal radiation is also influenced by the surface temperature. If during night time, the surface is cooler, then the emissivity would be higher? No, emissivity is a property of materials, not dependent on temperature. So, maybe the emissivity is constant, but the amount of radiation depends on temperature and emissivity.
So, for example, a surface with high emissivity will radiate more energy at higher temperatures. But during night time, if the surface is cooler, then the emitted energy is less. However, if the emissivity is high, then even at lower temperatures, the emitted energy is more than if emissivity is low.
But this suggests that increasing emissivity would allow for more radiation even when temperatures are lower. So, during night time, high emissivity surfaces would radiate more energy into space, contributing to cooling.
Therefore, the idea is that by enhancing emissivity of surfaces during night time, we can increase the rate of heat loss, which would help reduce global temperatures.
So, possible methods:
- Using materials with high emissivity in various applications (e.g., urban structures, agricultural areas, etc.)
- Implementing policies to encourage such materials or structures.
But what are the specific materials? For example, black bodies have high emissivity (about 1), but that's not practical. Maybe using materials like certain types of ceramics or engineered surfaces with high infrared emissivity.
Alternatively, using reflective surfaces in visible light but emissive in infrared. Like some kind of coatings.
But how to develop such materials? It might be a research area. Maybe the use of specific pigments or coatings that have high emissivity in the infrared spectrum.
Another idea: creating structures that are designed to emit heat during night time. For example, using buildings with certain designs that allow for better radiative cooling. This could be part of urban planning.
But this is more about local effects, but if widespread, it could contribute to global cooling.
Another thought: the Earth's radiative cooling during night time is affected by the atmosphere's opacity. If the atmosphere has less greenhouse gases, then more heat can escape. So, reducing greenhouse gases would increase the overall radiative cooling during all times, including night time. This is a broader approach, but maybe it's the key to the concept.
But how to reduce greenhouse gases? Via carbon capture, reforestation, renewable energy, etc.
So, combining both reducing greenhouse gases and increasing emissivity of surfaces during night time could be effective.
But perhaps the main idea is that during night time, the Earth can cool more if it has higher emissivity surfaces and lower atmospheric greenhouse gases. Therefore, the methods would involve both reducing greenhouse gases and enhancing surface emissivity.
So, in conclusion, the ways to use night time cooling of radiant energy into space to reduce global temperatures could be:
1. Enhancing surface emissivity through materials or structures that radiate more heat during night time.
2. Reducing atmospheric greenhouse gases to allow more heat radiation to escape into space during night time and throughout the day.
3. Implementing active cooling systems that operate during night time, using emissive properties to radiate heat into space.
4. Using thermal storage techniques during day and night, where heat is stored and then emitted during night time.
But these are various approaches, and need to be thought through.
<B>
[[[thinking end]]]
</B>
OUTPUT:
<p style="border: none; padding: 0cm">
The concept of utilizing nighttime cooling of radiant energy into
space to reduce global temperatures involves leveraging the Earth's
natural ability to emit heat radiation into the atmosphere and space
during periods when solar input is minimal. Here are several methods
that could achieve this:</p>
<ol>
<li><p style="border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Enhancing
Surface Emissivity</span></strong>:</p>
<ul>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><em>Materials
with high emissivity</span></em>: Surfaces that are engineered to
have high infrared emissivity (e.g., black body materials,
specialized ceramics, or coatings) can emit more heat during
nighttime. This increases the rate of heat loss from the Earth's
surface into space, effectively cooling local environments and
contributing to global thermal balance.</p></li>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><em>Urban
and agricultural applications</span></em>: Using such materials in
urban areas (e.g., buildings, roads) or agricultural zones could
reduce localized heat retention, indirectly lowering overall global
temperatures.</p></li>
</ul>
<li><p style="border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Reducing
Atmospheric Greenhouse Gases</span></strong>:</p>
<ul>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><em>Mitigation
strategies</span></em>: Lowering concentrations of greenhouse gases
(CO₂, methane, etc.) through carbon capture technologies,
reforestation, and transitioning to renewable energy sources would
allow more thermal radiation to escape the atmosphere during
nighttime and throughout the day. This reduces the "radiative
trapping" effect, enabling the Earth to cool more efficiently.</p></li>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><em>Atmospheric
engineering</span></em>: Techniques like cloud seeding or aerosol
manipulation could temporarily reduce atmospheric opacity during
nighttime, enhancing heat radiation escape.</p></li>
</ul>
<li><p style="border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Active
Cooling Systems</span></strong>:</p>
<ul>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><em>Thermal
emissive structures</span></em>: Deploying systems that actively
emit heat during nighttime (e.g., high-emissivity panels,
reflective surfaces designed for infrared radiation) could maximize
heat loss. These systems might be integrated into urban
infrastructure or large-scale environmental projects.</p></li>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><em>Thermal
storage and emission</span></em>: Using materials that absorb heat
during daytime and radiate it during nighttime (e.g., phase-change
materials, thermal storage blocks) could reduce heat accumulation,
shifting the Earth's thermal equilibrium toward lower temperatures.</p></li>
</ul>
<li><p style="border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Combining
Day and Night Strategies</span></strong>:</p>
<ul>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><em>Dual-purpose
materials</span></em>: Materials with low albedo (reflective in
visible light) to minimize solar absorption during daytime and high
emissivity (emissive in infrared) during nighttime could optimize
both heat retention and release, enhancing overall cooling
efficiency.</p></li>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><em>Dynamic
surface management</span></em>: Adjusting surface properties
dynamically based on time of day (e.g., adaptive coatings or smart
materials) could maximize nighttime radiative cooling while
minimizing daytime heating.</p></li>
</ul>
<li><p style="border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Space-Based
Solutions</span></strong>:</p>
<ul>
<li><p style="border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><em>Satellite
systems</span></em>: Deploying satellites or reflective arrays to
redirect heat radiation into space during nighttime could enhance
the Earth's ability to emit energy, though this is speculative and
requires advanced engineering.</p></li>
</ul>
</ol>
<p style="border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Key
Challenges</span></strong>: These methods require large-scale
implementation, technological advancements, and systemic changes
(e.g., reducing greenhouse gases). While nighttime cooling alone may
not directly reduce global temperatures, it contributes to the
Earth's thermal balance by increasing heat loss. Combining these
approaches with broader climate mitigation efforts could yield
significant impacts.</p>
<p style="line-height: 100%; margin-bottom: 0cm"><br/>
</p>
---
<B>Example Generation #2:</B>
Temp 1.5, rep pen 1.06, rep pen range 64, topk 100, topp .95, minp .05
Q4KS [non-imatrix], CPU only (windows 11), LMSTUDIO.
Speed: 8 T/S (no GPU offload)
This is a mid level quant, expect stronger performance from an Imatrix Q4KS / higher quants / full precision.
NOTE: Some formatting lost on copy/paste.
PROMPT:
<B>
Science Fiction: The Last Transmission - Write a story that takes place entirely within a spaceship's cockpit as the sole surviving crew member attempts to send a final message back to Earth before the ship's power runs out. The story should explore themes of isolation, sacrifice, and the importance of human connection in the face of adversity. If the situation calls for it, have the character(s) curse and swear to further the reader's emotional connection to them. 800-1000 words.
</B>
<P></P>
<B>
[[[thinking start]]]
</B>
Thought for 2 minutes 29 seconds
Okay, I need to write a science fiction story titled "The Last Transmission" that takes place entirely within a spaceship's cockpit. The main character is the sole surviving crew member trying to send a final message back to Earth before the ship's power runs out. The themes are isolation, sacrifice, and human connection. Also, if needed, the character can curse and swear to make the emotional connection stronger.
First, I should set the scene: a spaceship cockpit. Maybe a small, cramped space with limited equipment. The crew member is alone, maybe after some catastrophe. The ship is failing, power is running out, so they have to send a message before it's too late.
What's the backstory? Maybe the ship was on a mission, something went wrong, others died, and now only one remains. The message is to Earth, perhaps for the people back home, or maybe to a specific person, like a family member or a loved one.
The themes: isolation – the crew member is alone, no communication with others. Sacrifice – maybe they have to do something risky or give up their own life to send the message. Human connection – the importance of sending the message, even though it's the last chance, to maintain connection with Earth.
Need to include emotions, maybe the character is desperate, anxious, maybe frustrated. Maybe they are trying to fix things, but the system is failing. Also, swearing or curses could add intensity.
Let me outline the story:
1. Start in the cockpit, showing the environment. Maybe describe the screens, controls, the feeling of being alone.
2. The crew member is trying to send a message, but the ship's systems are failing. They have to work quickly.
3. The character's thoughts and feelings: isolation, the weight of responsibility, maybe memories of the crew members who died.
4. Maybe a moment where they try to contact Earth, but the signal is weak or failing. They struggle to send it.
5. The climax – when the power runs out, they make the final transmission, even if it's just a last message.
6. End with the message being sent, and maybe the crew member's final thoughts, or their death.
Possible names: Maybe the character is named something like "Eve" or "Jack". But maybe not use a name, just refer to as the crew member.
In terms of structure, maybe start with the character in the cockpit, trying to send a message. Then show the challenges they face. The emotional journey through the process.
Need to make sure that the story is entirely within the cockpit. So no scenes outside, no other characters except the one. Maybe the ship's systems are failing, and the character is trying to fix them.
Possible technical details: maybe the ship's power is running out due to a malfunction, or some event like a solar storm, or collision with debris. Maybe the crew member is trying to reboot the system, but it's too late.
The message could be a personal message, like "I'm here, I'm alive, please don't forget me", or something more specific. Maybe the crew member is trying to relay information about the mission, or their own experience.
Including swearing: maybe when the systems fail, they curse, or when they realize they can't save the ship, they swear.
Also, themes of sacrifice – maybe the character has to choose between sending the message or saving themselves, but they choose to send the message, even if it costs their life.
Need to create a sense of urgency. The time is limited, and the character is working against the clock.
Now, let me think about the narrative flow.
Start with the character in the cockpit, looking at the screens, maybe a countdown on the power. They are trying to send a message. Maybe the message is stored, but they need to transmit it.
But the system is failing – maybe the transmitter is malfunctioning, or the communication link is broken. They have to troubleshoot.
In the process, maybe the character tries different things, but fails. Then finally, when power runs out, they manage to send the last transmission.
The emotions: the desperation, the loneliness, the realization that this is their only chance.
Possible internal monologue: "I can't let this end without a message. I have to tell them what happened. Maybe someone will hear it."
Maybe the character remembers others who died, and that's part of the sacrifice – they are the last one, and their message is their legacy.
Including curses: "Damn it! Why is the system failing now?" or "I'm not going down without a message!"
Now, structure:
Paragraph 1: Setting the scene in the cockpit, describing the environment, the character's state.
Paragraph 2: The attempt to send the message, but obstacles.
Paragraph 3: The emotional struggle and thoughts about the crew members.
Paragraph 4: The final attempt, the climax of sending the message before power ends.
Paragraph 5: The aftermath, the message sent, the character's final thoughts.
Need to ensure that it's within the cockpit, no external scenes. Maybe the ship is drifting in space, no other people, just the character.
Also, maybe include some technical details about the spaceship, like the communication systems, power sources, etc.
Let me think about possible names or elements. Maybe the ship is called "The Horizon", or something else. The character's name could be "Jaxon" or "Mira".
But maybe not give a name, just refer to as "the crew member".
Also, the message could be a personal one, like a farewell to Earth, or to someone specific.
Now, writing the story:
Start with the cockpit, maybe dim lights, the character sitting in the chair, surrounded by monitors. The ship's systems are failing, power is dropping.
The crew member is trying to send a message. But the systems are not responding. Maybe they have to manually input the message, but it's too late.
Then, the character's internal thoughts: "I'm alone now. Everyone else is gone. I can't let this be forgotten."
Maybe memories of the crew members, or the mission that was supposed to be completed.
The character's attempts to fix the systems, but each attempt fails. They are desperate, maybe cursing.
Then, finally, when the power is about to run out, they send the message, even if it's just a short burst.
In the end, the message is sent, and the character reflects on the importance of human connection.
Now, write the story with these elements, making sure to explore the themes. Also, include swearing to make the emotional connection stronger.
Need to keep within 800-1000 words. Maybe split into paragraphs, with some dialogue or internal thoughts.
Possible title: "The Last Transmission" as given.
Now, start drafting the story.
<B>
[[[thinking END]]]
</B>
<P></P>
<B>
Write the draft according to your plan.
</B>
<P></P>
<B>
[[[thinking start]]]
</B>
Okay, let's start writing the story based on my plan. The setting is a spaceship cockpit with the sole surviving crew member trying to send a final message back to Earth before power runs out. Themes of isolation, sacrifice, and human connection.
First paragraph: Describe the cockpit. Maybe dim lighting, malfunctioning systems, the character sitting in the chair, surrounded by monitors. The power is running out, countdown on the screen.
Second paragraph: The crew member's attempt to send a message. The transmitter is malfunctioning. They try to fix it but failures. Cursing and frustration.
Third paragraph: Emotional struggle, thoughts about the crew members who died. Memories of the mission. Sacrifice – choosing to send the message even if it costs their life.
Fourth paragraph: Final attempt, climax. Power is running out, they send the last transmission. Maybe a short burst, but it's enough.
Fifth paragraph: Aftermath, the message sent, character reflects on human connection. Maybe a final thought or a whisper.
Need to include swearing and emotional depth. Make sure the story is within the cockpit, no external scenes. Use technical details about the ship's systems, maybe mention the communication link, power sources, etc.
Now, let me write the story step by step.
First paragraph:
The cockpit of the Horizon was dimly lit, its screens flickering with error codes. The crew member, a single figure, sat in the chair, surrounded by monitors. A countdown on the screen showed the power running out – 30 minutes left. The air was thin, the oxygen levels low, but the priority was sending the message.
Second paragraph:
They tried to activate the transmitter, but it wasn't responding. "Damn it!" they cursed, trying different commands. Each attempt failed, the systems refusing to work. Frustration builds, the pressure of time.
Third paragraph:
Memories of the crew members – gone, their deaths. Sacrifice – this is their last chance. "I can't let this end without a message. I have to tell them what happened." thoughts.
Fourth paragraph:
Final attempt. Power is 5 minutes left. They manually input the message, maybe a short burst. The system flickers, then sends it. The last transmission.
Fifth paragraph:
Aftermath, the message sent, the crew member's final thoughts. "Human connection is everything. Even in the end, I have to reach them."
Now, let me put this into a coherent story with emotional depth and swearing.
Okay, writing the full draft.
<B>
[[[thinking END]]]
</B>
<P></P>
OUTPUT:
The cockpit of the Horizon was a claustrophobic chamber of flickering screens and failing systems. The air was thin, the oxygen levels barely above critical, but the real problem was the power—running out, like a dying heartbeat. The lone crew member sat in the chair, their fingers trembling over the controls. A countdown on the screen read: 17 minutes left.
"Come on," they muttered, staring at the transmitter. "It's just a damn message. Why won't it work?" They punched the command button again, but the system didn’t respond. A cascade of error codes flashed across the monitor.
"Shit!" they cursed, slamming the console. The sound echoed in the empty chamber, a hollow scream. They had spent hours trying to reboot the communication link, but the signal was broken—something had happened in the last few hours. The crew had gone. Everyone else was dead. The Horizon was drifting, unmoored, with no way back.
They leaned forward, pressing their forehead against the screen. A memory flashed—of the crew, of the mission, of the promise they all made to Earth: We’ll return. But now, they were alone. The last transmission.
"Let me just say this," they whispered, eyes watering. "I’m here. I’m alive. They should know."
They reached for the log files—stored messages from the crew. It was a crude text, but maybe it could be transmitted. They typed in the command: Send Message.
The system flickered, then stalled. The screen turned black.
"No," they groaned. "No no no!"
They tried again, and again, but the transmitter refused. The power was dying. The systems were failing. They had minutes left—13, then 8, then 5. The time was closing in.
"God damn it," they muttered, tears streaming down their face. "I'm not going down without a message."
They took the handheld console, manually input the message. A short text: To Earth, we are alive. We are gone. Please remember us.
The transmitter was offline. But they had to try. They connected the manual link—something ancient, from the old system, bypassing the main grid.
"Come on, come on," they whispered. "It’s just a last shot. Let it go."
The screen flickered—then, finally, a pulse. The signal was weak, but it sent.
"Done," they said. A breath. The power ran out. The systems shut down. The Horizon was silent.
But the message had gone.
They sat in the cockpit, alone, with nothing left. They had given their life to send that one message—what it meant? It wasn’t just a report. It was a legacy. A connection.
"Human connection is everything," they thought. "Even when I’m gone, someone will remember me. Someone will know I was here."
They closed their eyes, the last breath of oxygen leaving their body.
The message had been sent.
And the Horizon drifted in silence, its last transmission fading into the void.
---
<B>Example Generation #3:</B>
Temp 1.2, rep pen 1.06, rep pen range 64, topk 100, topp .95, minp .05
Q4KS [non-imatrix], CPU only (windows 11), LMSTUDIO.
Speed: 8 T/S (no GPU offload)
This is a mid level quant, expect stronger performance from an Imatrix Q4KS / higher quants / full precision.
NOTE: Some formatting lost on copy/paste.
PROMPT:
<B>
Think up 20 "outside the box" ways to use the "night" time cooling of radiant energy into space to reduce global temperatures. Don't discount any method, if is possible.
</B>
NOTE: Thinking / output the same, so I have only presented final output.
<P></P>
<p style="border: none; padding: 0cm">
Okay, so the user wants 20 outside-the-box ideas for using nighttime
cooling of radiant energy into space to reduce global temperatures.
Hmm. Let me start by understanding what "nighttime cooling"
means here. I think it refers to the process where the Earth radiates
heat into space during the night when there's no sunlight, which can
help cool down the surface. But how can we harness that to reduce
global temperatures?</p>
<p style="border: none; padding: 0cm">First, maybe we need to think
about how the Earth's surface emits infrared radiation into space.
The atmosphere has greenhouse gases that trap some of this heat, so
if we can somehow enhance the emission of energy into space during
nighttime, maybe we can cool down the planet more effectively.</p>
<p style="border: none; padding: 0cm">But I need to think of ways to
manipulate this process. So possible methods could involve using
materials that reflect or emit more infrared radiation, or altering
the atmosphere's composition to reduce greenhouse gases. But the user
says "outside the box," so maybe more unconventional ideas.</p>
<p style="border: none; padding: 0cm">Let me brainstorm. For example,
creating artificial structures that reflect heat during night, like
some kind of mirrors or materials that are good at radiating energy.
Or maybe using reflective surfaces on the ground. Maybe even some
kind of geoengineering techniques that modify the Earth's emissivity.</p>
<p style="border: none; padding: 0cm">Another thought: using
nocturnal activities to manage energy distribution. Like, perhaps
setting up solar panels during the day to capture energy, but during
night, releasing that energy via some kind of heat dissipation
mechanisms.</p>
<p style="border: none; padding: 0cm">Wait, but the idea is about
cooling during night time. So maybe during nighttime, when the Earth
is not receiving sunlight, it can lose heat through radiation. So if
we could increase that radiative cooling during nighttime, maybe that
would help reduce temperatures. How?</p>
<p style="border: none; padding: 0cm">So possible methods:</p>
<ol>
<li><p style="border: none; padding: 0cm">Using reflective materials
on the ground during the night. For example, coating surfaces with
materials that have high emissivity in infrared wavelengths. But not
sure if this is feasible. Maybe like some kind of thermally
conductive material.</p></li>
<li><p style="border: none; padding: 0cm">Creating a network of
thermal conduits that channel heat to space during nighttime. Maybe
like structures that actively transport heat from the Earth's
surface into the atmosphere and then into space.</p></li>
<li><p style="border: none; padding: 0cm">Using bioluminescent
organisms that emit light during night, which might not directly
relate, but maybe they can influence the thermal environment? Not
sure.</p></li>
<li><p style="border: none; padding: 0cm">Developing a kind of
"thermal mirror" system that reflects heat during night.
Like mirrors that are tuned to infrared wavelengths, directing heat
into space.</p></li>
<li><p style="border: none; padding: 0cm">Creating artificial cloud
systems that are less dense during nighttime, allowing more
radiation. But clouds reflect sunlight but also trap heat. So if
during night, the clouds are less active, maybe they allow more heat
to escape?</p></li>
<li><p style="border: none; padding: 0cm">Using atmospheric aerosols
that have properties to enhance infrared emission. For example, some
kind of particles that act as "thermal emissors" during
night.</p></li>
<li><p style="border: none; padding: 0cm">Deploying large-scale
structures that are designed to radiate heat into space during
night. Like huge reflectors or heat dissipation structures.</p></li>
<li><p style="border: none; padding: 0cm">Maybe using cryogenic
technology during night to freeze water and release latent heat via
evaporation?</p></li>
</ol>
<p style="border: none; padding: 0cm">Wait, but during night, maybe
the temperature is lower, so freezing might not be effective. Not
sure.</p>
<ol start="9">
<li><p style="border: none; padding: 0cm">Using solar energy stored
during day and then releasing it during night via some kind of
cooling mechanisms. Like, storing solar energy in thermal
reservoirs, and then during night, releasing that energy via
radiation into space?</p></li>
</ol>
<p style="border: none; padding: 0cm">Hmm. But how to do that. Maybe
like a thermal battery that can release heat during night.</p>
<ol start="10">
<li><p style="border: none; padding: 0cm">Creating artificial
"thermal sinks" that actively draw heat from the Earth's
surface during night.</p></li>
<li><p style="border: none; padding: 0cm">Using high-albedo surfaces
during night, but since albedo is for reflecting sunlight, maybe not
relevant. But emissivity is different. So maybe high-emissivity
materials.</p></li>
<li><p style="border: none; padding: 0cm">Maybe using magnetic
fields to influence thermal radiation? Not sure if this is feasible.
Maybe a kind of electromagnetic manipulation to enhance infrared
emission.</p></li>
<li><p style="border: none; padding: 0cm">Creating a network of
thermal conductive pathways that help the heat transfer more
efficiently during night into space.</p></li>
<li><p style="border: none; padding: 0cm">Using biotechnological
methods, like genetic engineering to create organisms that have high
emissivity during night. Maybe plants or bacteria that emit infrared
radiation more?</p></li>
<li><p style="border: none; padding: 0cm">Developing a kind of
"thermal conduit" using nanotechnology or advanced
materials to transport heat from Earth surface into space.</p></li>
<li><p style="border: none; padding: 0cm">Using space-based
satellites to redirect heat radiation during night. Like satellites
that could be placed in orbit to capture and direct the heat away?</p></li>
</ol>
<p style="border: none; padding: 0cm">But not sure if this is
possible. Maybe satellites can reflect or absorb heat, but how would
they do that.</p>
<ol start="17">
<li><p style="border: none; padding: 0cm">Creating a network of
thermal reactors that operate during night, converting heat into
other forms, maybe as energy, but then release it into space.</p></li>
<li><p style="border: none; padding: 0cm">Using gravitational
methods to influence the thermal environment? Not sure.</p></li>
<li><p style="border: none; padding: 0cm">Maybe using quantum
effects in materials to enhance infrared radiation. For example,
materials that have quantum properties that emit more heat during
night.</p></li>
<li><p style="border: none; padding: 0cm">Developing a kind of
"thermal camouflage" for structures that during night,
allow heat to escape more. Like, materials that are transparent to
infrared radiation.</p></li>
</ol>
<p style="border: none; padding: 0cm">But I need to check if these
ideas are feasible. Some of them might be speculative, but the user
said not to discount any method. So even if they are hypothetical or
impossible, still list them as possible.</p>
<p style="border: none; padding: 0cm">Wait, maybe some of these are
not actually related to the cooling during night. For example, using
bioluminescent organisms is a stretch. But the idea is to use radiant
energy into space during night. So if any methods can be applied to
that, even if they are unconventional.</p>
<p style="border: none; padding: 0cm">So let me list 20 ideas:</p>
<ol>
<li><p style="border: none; padding: 0cm">Deploying high-emissivity
materials on Earth's surface (like special coatings or structures)
to enhance radiative cooling during night.</p></li>
<li><p style="border: none; padding: 0cm">Using infrared-reflective
mirrors in space to direct heat radiation from Earth into space.</p></li>
<li><p style="border: none; padding: 0cm">Creating artificial
"thermal sinks" that absorb and channel heat into space
during nighttime.</p></li>
<li><p style="border: none; padding: 0cm">Developing a network of
thermal conduits that transfer heat from the surface to the
atmosphere for emission into space.</p></li>
<li><p style="border: none; padding: 0cm">Utilizing nocturnal solar
energy storage systems to release heat during night via radiative
processes.</p></li>
<li><p style="border: none; padding: 0cm">Implementing atmospheric
aerosol technologies to reduce greenhouse gas effects during night,
allowing more heat to escape.</p></li>
<li><p style="border: none; padding: 0cm">Creating bioluminescent
organisms that emit thermal radiation during night to influence the
local environment.</p></li>
<li><p style="border: none; padding: 0cm">Deploying large-scale
infrared emissive structures (like towers or arrays) to facilitate
heat emission into space.</p></li>
<li><p style="border: none; padding: 0cm">Using cryogenic systems
during night to freeze water and release latent heat via evaporation
into space.</p></li>
<li><p style="border: none; padding: 0cm">Developing a "thermal
mirror" system on Earth's surface that redirects heat radiation
into space.</p></li>
<li><p style="border: none; padding: 0cm">Creating artificial cloud
formations that are less dense during night to allow more radiative
cooling.</p></li>
<li><p style="border: none; padding: 0cm">Implementing
electromagnetic field manipulations to enhance infrared emission
from Earth during night.</p></li>
<li><p style="border: none; padding: 0cm">Utilizing quantum
materials with properties to emit more infrared radiation during
nighttime.</p></li>
<li><p style="border: none; padding: 0cm">Developing space-based
satellites that capture and redirect heat from Earth into space.</p></li>
<li><p style="border: none; padding: 0cm">Creating thermal-reactor
systems that convert stored daytime heat into radiative emissions
during night.</p></li>
<li><p style="border: none; padding: 0cm">Using gravitational
manipulation techniques to influence the thermal environment,
allowing more heat to escape.</p></li>
<li><p style="border: none; padding: 0cm">Deploying biotechnological
modifications for plants or organisms that have high emissivity
during night.</p></li>
<li><p style="border: none; padding: 0cm">Creating thermal
camouflage structures that are transparent to infrared radiation
during night.</p></li>
<li><p style="border: none; padding: 0cm">Implementing
nanotechnology-based thermal conduits to transfer heat from surface
to space.</p></li>
<li><p style="border: none; padding: 0cm">Utilizing nocturnal energy
redistribution systems that channel heat into space via radiative
cooling.</p></li>
</ol>
<p style="border: none; padding: 0cm">Hmm, maybe some of these are
too vague or not concrete. But the user asked for outside-the-box
ideas, so even if they are hypothetical, it's okay.</p>
<p style="border: none; padding: 0cm">Another idea: Using thermal
insulation during day to retain heat, then release during night. But
that might not help cooling, but maybe reverse.</p>
<p style="border: none; padding: 0cm">Wait, but the main idea is to
use the natural radiative cooling during night. So any methods that
could enhance this process. So perhaps some of the above are
relevant.</p>
<p style="border: none; padding: 0cm">I think that's 20 ideas. Need
to check if they are different and not repeating. For example, 1 and
18 might be similar. But okay, as long as each is unique.</p>
<p style="border: none; padding: 0cm">So I think this list covers
various methods from materials, structures, technology, biological,
etc. All are possible even if speculative.</p>
<p style="border: none; padding: 0cm"><br/>
<br/>
</p>
|
DavidAU/Qwen3-30B-A6B-16-Extreme-128k-context | DavidAU | 2025-05-04T23:12:42Z | 1 | 1 | transformers | [
"transformers",
"safetensors",
"qwen3_moe",
"text-generation",
"128 k context",
"reasoning",
"thinking",
"qwen3",
"16 experts",
"conversational",
"base_model:Qwen/Qwen3-30B-A3B-Base",
"base_model:finetune:Qwen/Qwen3-30B-A3B-Base",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T02:17:47Z | ---
library_name: transformers
pipeline_tag: text-generation
tags:
- 128 k context
- reasoning
- thinking
- qwen3
- 16 experts
base_model:
- Qwen/Qwen3-30B-A3B-Base
---
<h2>Qwen3-30B-A6B-16-Extreme-128k-context</h2>
<img src="star-wars-hans-solo.gif" style="float:right; padding:10px;">
This repo contains the full precision source code, in "safe tensors" format to generate GGUFs, GPTQ, EXL2, AWQ, HQQ and other formats. The source code can also be used directly.
This is a simple "finetune" of the Qwen's "Qwen 30B-A3B" (MOE) model, setting the experts in use from 8 to 16 (out of 128 experts)
and set the max context to 128k (up from 32k) as per Qwen's tech notes / method at their repo.
THREE Example generations (Q4KS, CPU) at the bottom of this page using 16 experts / this model.
This reduces the speed of the model, but uses more "experts" to process your prompts and uses 6B (of 30B) parameters instead of 3B (of 30B) parameters. Depending on the application you may want to
use the regular model ("30B-A3B"), and use this model for MORE COMPLEX / "DEEPER" (IE Nuance) use case(s).
Regular or simpler use cases may benefit from using the normal (8 experts), the "12 cooks" (12 experts) or "High-Speed" (4 experts) version(s).
Using 16 experts instead of the default 8 will slow down token/second speeds about about 1/2 or so.
Context size: 128k
Use Jinja Template or CHATML template.
IMPORTANT NOTES:
- Due to the unique nature (MOE, Size, Activated experts, size of experts) of this model GGUF quants can be run on the CPU, GPU or with GPU part "off-load", right up to full precision.
- This model is difficult to Imatrix : You need a much larger imatrix file / multi-language / multi-content (ie code/text) to imatrix it.
- GPU speeds will be BLISTERING 4x-8x or higher than CPU only speeds AND this model will be BLISTERING too, relative to other "30B" models (Token per second speed equal roughly to 6B "normal" model speeds).
Please refer the org model card for details, benchmarks, how to use, settings, system roles etc etc :
[ https://huggingface.co/Qwen/Qwen3-30B-A3B ]
<B>More / Less Experts Versions:</B>
4 experts:
[ https://huggingface.co/DavidAU/Qwen3-30B-A1.5B-High-Speed ]
12 experts:
[ https://huggingface.co/DavidAU/Qwen3-30B-A4.5B-12-Cooks ]
16 experts, 32k:
[ https://huggingface.co/DavidAU/Qwen3-30B-A6B-16-Extreme ]
24 experts:
[ https://huggingface.co/DavidAU/Qwen3-30B-A7.5B-24-Grand-Brainstorm ]
<B>OPTIONAL SYSTEM ROLE: </B>
You may or may not need this, as most times Qwen3s generate their own reasoning/thinking blocks.
```
You are a deep thinking AI, you may use extremely long chains of thought to deeply consider the problem and deliberate with yourself via systematic reasoning processes to help come to a correct solution prior to answering. You should enclose your thoughts and internal monologue inside <think> </think> tags, and then provide your solution or response to the problem.
```
See document "Maximizing-Model-Performance-All..." below for how to "set" system role in various LLM/AI apps below.
IMPORTANT: Highest Quality Settings / Optimal Operation Guide / Parameters and Samplers
If you are going to use this model, (source, GGUF or a different quant), please review this document for critical parameter, sampler and advance sampler settings (for multiple AI/LLM aps).
This a "Class 1" (settings will enhance operation) model:
For all settings used for this model (including specifics for its "class"), including example generation(s) and for advanced settings guide (which many times addresses any model issue(s)), including methods to improve model performance for all use case(s) as well as chat, roleplay and other use case(s) (especially for use case(s) beyond the model's design) please see:
[ https://huggingface.co/DavidAU/Maximizing-Model-Performance-All-Quants-Types-And-Full-Precision-by-Samplers_Parameters ]
REASON:
Regardless of "model class" this document will detail methods to enhance operations.
If the model is a Class 3/4 model the default settings (parameters, samplers, advanced samplers) must be set for "use case(s)" uses correctly. Some AI/LLM apps DO NOT have consistant default setting(s) which result in sub-par model operation. Like wise for Class 3/4 models (which operate somewhat to very differently than standard models) additional samplers and advanced samplers settings are required to "smooth out" operation, AND/OR also allow full operation for use cases the model was not designed for.
BONUS - Use these settings for ANY model, ANY repo, ANY quant (including source/full precision):
This document also details parameters, sampler and advanced samplers that can be use FOR ANY MODEL, FROM ANY REPO too - all quants, and of course source code operation too - to enhance the operation of any model.
[ https://huggingface.co/DavidAU/Maximizing-Model-Performance-All-Quants-Types-And-Full-Precision-by-Samplers_Parameters ]
NOTE:
I strongly suggest you also visit the DavidAU GGUF (below) repo too for more details in using this model ; especially if it is "Class 3" or "Class 4" to get maximum performance from the model.
For full information about this model, including:
- Details about this model and its use case(s).
- Context limits
- Special usage notes / settings.
- Any model(s) used to create this model.
- Template(s) used to access/use this model.
- Example generation(s)
- GGUF quants of this model
Please go to:
[ GGUFS REPO coming soon ]
---
<B>Example Generation #1:</B>
Temp 1.2, rep pen 1.06, rep pen range 64, topk 100, topp .95, minp .05
Q4KS [non-imatrix], CPU only (windows 11), LMSTUDIO.
Speed: 11 T/S (no GPU offload)
GPU speed would be 4x or higher.
This is a mid level quant, expect stronger performance from an Imatrix Q4KS / higher quants / full precision.
NOTE: Some formatting lost on copy/paste.
PROMPT:
<B>
Explain ways to use the "night" time cooling of radiant energy into space to reduce global temperatures.
</B>
<P></P>
<B>
[[[thinking start]]]
</B>
<P></P>
Thought for 2 minutes 17 seconds
Okay, so the user is asking about how to use the nighttime cooling of radiant energy into space to reduce global temperatures. Hmm, I need to think about this carefully. First, let's break down what they mean by "nighttime cooling of radiant energy into space." I guess that refers to the Earth emitting heat radiation during the night, which escapes into space. But how does that help in reducing global temperatures?
Wait, but isn't that natural? The Earth's atmosphere and surface emit thermal radiation, which is part of the planet's energy balance. During the day, solar radiation comes in, and at night, it radiates back. But maybe there's a way to enhance this process to cool down more effectively. But how?
I need to consider the greenhouse effect. If the Earth's atmosphere traps heat, then the cooling at night might be limited by that. So if we can reduce the greenhouse effect during nighttime, maybe more heat would escape into space. But how? Maybe reducing greenhouse gases? But that's a broad approach. However, the user is specifically asking about using nighttime cooling.
Another angle: Maybe reflecting more sunlight during the day to reduce heat absorption, but that's not nighttime. Alternatively, using materials that emit more infrared radiation at night. For example, certain surfaces that are good radiators of thermal energy. But how would that work?
Also, there's something called "nighttime cooling" in architecture, where buildings use materials that radiate heat to the sky, allowing them to cool down without air conditioning. Maybe this is similar. So using such materials on a larger scale could help lower temperatures.
But what about the atmosphere? If during nighttime, there's less solar radiation, but the Earth still emits heat. However, if there are clouds or greenhouse gases that trap the heat, then the cooling might be reduced. So if we can reduce those factors at night, maybe more heat escapes.
Another thought: Maybe using satellites or other structures to reflect or redirect heat radiation into space. But that seems impractical for large scale.
Or, perhaps manipulating the atmosphere to have less greenhouse gases during nighttime. But how? Maybe using some kind of temporary atmospheric changes, but that's not feasible.
Alternatively, think about energy balance. If the Earth's radiative cooling is enhanced during night, maybe it could contribute to overall temperature reduction. However, the global temperature is a complex system. The net effect of nighttime cooling might be small compared to other factors like solar input, greenhouse gases, etc.
But the user is looking for ways to use this natural process. So possible methods:
1. Enhancing the emissivity of surfaces during night. For example, using materials that are good at emitting thermal radiation. This could help in cooling local areas, but how would that scale globally?
2. Reducing greenhouse gases, which allows more heat to escape during all times, including night. But that's a standard climate mitigation approach.
3. Maybe creating structures that reflect heat radiation away from the Earth. Not sure if that's possible.
4. Using urban design with reflective surfaces and materials that aid in cooling at night. Like cool roofs or pavements, which have high emissivity. If implemented widely, could lower temperatures in cities, which might have a cumulative effect.
5. Maybe using space-based technologies to enhance radiation loss. For example, deploying satellites that emit heat into space. But not sure if that's practical.
6. Considering the role of water and evaporation during night. Evaporative cooling is more effective when there's higher humidity, but maybe at night, depending on conditions.
Wait, but how does the radiation into space contribute? The Earth's outgoing longwave radiation (OLR) is part of the energy balance. If we can increase OLR, then the net radiative cooling would be more, leading to lower temperatures. But how to do that?
So, maybe by reducing factors that block OLR. For example, reducing cloud cover or greenhouse gases. If during the night, there's less cloud cover (maybe), then OLR would be higher. So if we can have fewer clouds at night, more heat escapes.
But controlling cloud cover is a big challenge. Maybe using geoengineering techniques like seeding clouds to reduce their formation, but that's not straightforward and might have unintended consequences.
Another angle: The Earth's surface has different emissivity properties. So surfaces with higher emissivity (like dark, rough materials) can radiate more heat. So if we can create such surfaces on a large scale, maybe more heat is emitted during night, leading to cooling.
But how would that affect global temperatures? If the surfaces are more efficient at radiating, then the temperature might decrease as they lose more heat. But this is local effect. However, globally, if many areas have higher emissivity, maybe the overall net loss of heat would be more, contributing to cooling.
Also, there's a concept called "radiative cooling," which uses materials that can emit thermal radiation at night, allowing them to cool below ambient temperature. So if we use such materials on a large scale, like in agriculture or urban areas, could it help?
But what about the atmosphere's role? If the atmosphere is transparent to infrared, then more heat would escape. But if there are greenhouse gases, they absorb and re-emit that heat, which reduces the net loss. So, if we can reduce greenhouse gases, more heat escapes, leading to cooler temperatures.
So combining these factors: reducing greenhouse gases (both daytime and nighttime), enhancing emissivity of surfaces during night, and perhaps other methods like geoengineering to increase OLR.
But I need to list possible ways. Let me think of specific examples:
- Use high-emissivity materials in urban areas. Cool roofs, pavements, etc., which radiate more heat at night.
- Reduce atmospheric greenhouse gases (CO2, methane, etc.) through mitigation efforts, which would allow more heat to escape during all times.
- Geoengineering approaches like stratospheric aerosol injection, but that's more about reflecting sunlight rather than radiation.
- Creating artificial surfaces that can emit radiation more effectively, perhaps with certain coatings or materials.
- Enhancing the Earth's albedo (reflectivity) during daytime, which reduces heat absorption, leading to less heat to radiate at night. But that might not directly affect nighttime cooling, but overall temperature.
But wait, the question is about using the "nighttime" cooling. So maybe focus on nighttime processes. So perhaps:
1. Increasing emissivity of surfaces at night: Using materials that can emit more thermal radiation during the night. This could be done through reflective or emissive coatings.
2. Reducing greenhouse gas concentrations in the atmosphere during nighttime, allowing more heat to escape.
3. Managing cloud cover during night, maybe reducing cloud formation to allow more OLR.
4. Using space-based mirrors or structures to reflect heat away from Earth, but that's speculative and not practical.
5. Developing technologies that actively radiate heat into space, such as thermoelectric devices or other methods.
But these are all theoretical. What is the practical way? Maybe the first two: reducing greenhouse gases (which is a standard approach) and using emissive materials.
Another thought: The Earth's surface during night cools because it can emit radiation without the sun. So, if we can enhance that process, maybe through better emissivity. For example, in agriculture, using reflective materials to cool the soil at night, which might have some effect on local temperatures.
But for global impact, these would need to be implemented on a large scale. Maybe using urban planning with high-emissivity surfaces, or creating vast areas with such materials.
Also, consider that during the night, the Earth's surface is not receiving solar radiation, so the cooling is more about emitting heat. So, if we can make surfaces emit more heat at night, they would cool faster, which could lead to lower temperatures in those areas.
But how does that contribute to global temperature reduction? It might be a local effect, but if implemented globally, perhaps the cumulative effect could help. However, it's not clear how much impact this would have compared to other factors like greenhouse gases and solar radiation.
So the answer would need to list these methods, explaining their potential and challenges. Maybe also mention that while nighttime cooling is a natural process, enhancing it through various means could contribute to cooling, but it's part of a larger climate strategy.
<B>
[[[thinking end]]]
</B>
OUTPUT:
<p style="border: none; padding: 0cm">
To harness the "nighttime cooling" of radiant energy into
space as a strategy to reduce global temperatures, we need to
consider both the natural processes and human interventions that can
enhance this mechanism. Here are several approaches:</p>
<h3 class="western" style="border: none; padding: 0cm">1. <span style="display: inline-block; border: none; padding: 0cm"><strong>Enhancing
Surface Emissivity</span></strong></h3>
<ul>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Concept</span></strong>:
The Earth's surface emits thermal radiation during the night, which
is part of the planet's energy balance. Materials with high
emissivity (ability to emit thermal radiation) can release heat more
efficiently.</p></li>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Application</span></strong>:
Using high-emissivity materials in urban areas, such as "cool
roofs" or "cool pavements," can increase the rate of
radiative cooling at night. These materials are designed to reflect
less sunlight during the day and emit more heat during the night,
thereby lowering local temperatures.</p></li>
<li><p style="border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Global
Impact</span></strong>: If implemented on a large scale (e.g., in
cities or agricultural areas), this could contribute to localized
cooling, which might have cumulative effects on global temperature,
especially in regions with high urbanization.</p></li>
</ul>
<h3 class="western" style="border: none; padding: 0cm">2. <span style="display: inline-block; border: none; padding: 0cm"><strong>Reducing
Greenhouse Gas Concentrations</span></strong></h3>
<ul>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Concept</span></strong>:
Greenhouse gases (GHGs) like CO₂ and methane trap heat in the
atmosphere by absorbing and re-emitting thermal radiation. This
reduces the amount of heat that escapes into space, both during the
day and night.</p></li>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Application</span></strong>:
Mitigating GHG emissions through policies like renewable energy
adoption, carbon capture, or reforestation can lower atmospheric
concentrations. This would allow more heat to radiate into space
during all times, including nighttime.</p></li>
<li><p style="border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Global
Impact</span></strong>: While this is a standard climate mitigation
approach, it directly addresses the root of the problem by reducing
the "greenhouse effect," which is critical for global
temperature regulation.</p></li>
</ul>
<h3 class="western" style="border: none; padding: 0cm">3. <span style="display: inline-block; border: none; padding: 0cm"><strong>Managing
Cloud Cover</span></strong></h3>
<ul>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Concept</span></strong>:
Clouds act as a barrier to outgoing longwave radiation (OLR). During
nighttime, lower cloud cover could allow more heat to escape into
space.</p></li>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Application</span></strong>:
Geoengineering techniques like "cloud seeding" or
controlled cloud formation might be used to reduce cloud density
during nighttime. However, this is speculative and fraught with
ecological risks.</p></li>
<li><p style="border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Challenges</span></strong>:
Manipulating cloud cover is complex and not well understood,
requiring careful consideration of unintended consequences.</p></li>
</ul>
<h3 class="western" style="border: none; padding: 0cm">4. <span style="display: inline-block; border: none; padding: 0cm"><strong>Space-Based
Technologies</span></strong></h3>
<ul>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Concept</span></strong>:
While still in theoretical stages, technologies like reflective
satellites or "radiative cooling" structures could be
designed to redirect heat radiation into space.</p></li>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Application</span></strong>:
Hypothetical deployment of satellites with high emissivity coatings
or mirrors that reflect thermal energy away from Earth might enhance
global radiative cooling. However, this is not yet feasible and
would require significant technological and economic investment.</p></li>
<li><p style="border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Challenges</span></strong>:
Practicality is limited by current technology, cost, and potential
environmental impacts.</p></li>
</ul>
<h3 class="western" style="border: none; padding: 0cm">5. <span style="display: inline-block; border: none; padding: 0cm"><strong>Urban
and Landscape Design</span></strong></h3>
<ul>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Concept</span></strong>:
Urban areas often retain heat due to materials like asphalt and
concrete (which have low emissivity). Replacing these with emissive
materials could enhance nighttime cooling.</p></li>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Application</span></strong>:
Designing cities with emissive surfaces, green spaces, or reflective
elements can create cooler microclimates. This approach is already
being tested in urban planning (e.g., cool roofs in cities like Los
Angeles).</p></li>
<li><p style="border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Global
Impact</span></strong>: While localized, such practices could reduce
heat island effects, which are significant contributors to local
temperature rises.</p></li>
</ul>
<h3 class="western" style="border: none; padding: 0cm">6. <span style="display: inline-block; border: none; padding: 0cm"><strong>Evaporative
Cooling at Night</span></strong></h3>
<ul>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Concept</span></strong>:
Evaporation of water can lead to cooling through latent heat
transfer. During nighttime, if humidity is high, this process may be
more effective.</p></li>
<li><p style="border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Application</span></strong>:
Increasing water availability (e.g., in arid regions) or using
water-based cooling systems (like misting devices) could enhance
nighttime cooling. However, this requires resource management and
may not be viable globally.</p></li>
</ul>
<h3 class="western" style="border: none; padding: 0cm">7. <span style="display: inline-block; border: none; padding: 0cm"><strong>Promoting
Radiative Cooling in Agriculture</span></strong></h3>
<ul>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Concept</span></strong>:
Agricultural practices that use emissive materials or water for
cooling can reduce temperatures in farming areas. This could have
indirect benefits on global temperature by reducing heat from human
activities.</p></li>
<li><p style="border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Application</span></strong>:
Techniques like "cooling blankets" or reflective
irrigation systems might be employed. However, this is more of a
niche application than a global strategy.</p></li>
</ul>
<h3 class="western" style="border: none; padding: 0cm">Key
Considerations:</h3>
<ul>
<li><p style="margin-bottom: 0cm; border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Limitations</span></strong>:
Nighttime cooling is a natural process, but its impact on global
temperatures is modest compared to the dominant drivers like solar
input and GHG concentrations. Enhancing it requires complementary
strategies.</p></li>
<li><p style="border: none; padding: 0cm"><span style="display: inline-block; border: none; padding: 0cm"><strong>Holistic
Approach</span></strong>: These methods should be part of broader
climate mitigation efforts, such as reducing GHGs, promoting
renewable energy, and improving land use practices.</p></li>
</ul>
<h3 class="western" style="border: none; padding: 0cm">Conclusion:</h3>
<p style="border: none; padding: 0cm">Harnessing nighttime cooling
through emissive materials and atmospheric management could
contribute to temperature reduction, but it is most effective when
integrated with global climate strategies. While some approaches
(e.g., high-emissivity materials) are feasible, others remain
theoretical. The primary focus should remain on reducing greenhouse
gases, which directly address the root of global warming.</p>
<p style="line-height: 100%; margin-bottom: 0cm"><br/>
</p>
---
<B>Example Generation #2:</B>
Temp 0.6, rep pen 1.06, rep pen range 64, topk 100, topp .95, minp .05
Q4KS [non-imatrix], CPU only (windows 11), LMSTUDIO.
Speed: 12 T/S (no GPU offload)
GPU speed would be 4x or higher.
This is a mid level quant, expect stronger performance from an Imatrix Q4KS / higher quants / full precision.
NOTE: Some formatting lost on copy/paste.
PROMPT:
<B>
Science Fiction: The Last Transmission - Write a story that takes place entirely within a spaceship's cockpit as the sole surviving crew member attempts to send a final message back to Earth before the ship's power runs out. The story should explore themes of isolation, sacrifice, and the importance of human connection in the face of adversity. If the situation calls for it, have the character(s) curse and swear to further the reader's emotional connection to them. 800-1000 words.
</B>
<P></P>
<B>
[[[thinking start]]]
</B>
Thought for 1 minutes 38 seconds
Okay, so I need to write a science fiction story called "The Last Transmission" that takes place entirely in a spaceship's cockpit. The main character is the sole surviving crew member trying to send a final message back to Earth before the ship's power runs out. The themes are isolation, sacrifice, and human connection. Also, if needed, the character should curse and swear to make the reader emotionally connected.
First, I need to set the scene. The cockpit of a spaceship. Maybe it's a damaged ship, maybe there's some technical issues. The crew is all dead except one person. The protagonist is trying to send a message before the power dies. So the story is about their last efforts, their emotions, and the themes.
I should start by establishing the setting. Maybe describe the cockpit with all the instruments, the control panels, maybe some flickering lights. The protagonist is alone, maybe they're in a suit, or maybe they're just sitting there. The atmosphere is tense, maybe the ship is on the brink of failure.
The themes: isolation. The character is alone, no one else to talk to. Sacrifice: maybe they had to make sacrifices to get this far, or maybe they are sacrificing their own life to send the message. Human connection: even though they're alone, they're trying to connect with Earth, to reach out, to leave a message.
The character should have some emotional depth. Maybe they're struggling with the reality of being the last one. They might have memories of the crew, maybe they lost people, or maybe they're thinking about their own life. The curses and swearing would come when they're frustrated, maybe when the systems are failing, when they can't get the message through, or when they realize the power is running out.
I need to structure the story. Maybe start with the protagonist in the cockpit, trying to send the message. They check the systems, maybe there's some technical issues. They try to communicate, but the systems are failing. They have to make a decision: send the message now or wait. But the power is running out, so they have to act quickly.
Maybe include flashbacks or memories of the crew, to show the sacrifice and human connection. Maybe the protagonist is thinking about their friends, or maybe they're trying to remember their last words. The emotional moments would be when they talk to Earth, or when they realize they can't send it anymore.
The ending: the message is sent, but the power dies. Or maybe the message is not sent, but the protagonist's final thoughts are recorded. Maybe the message is a personal one, like a farewell, or a message of hope.
I need to make sure the story is within the cockpit, so no external scenes. All action is in the cockpit. The character's actions, their interactions with the systems, their emotions.
Possible elements: the ship's systems failing, maybe a countdown on the power, the protagonist's struggle to maintain communication, the physical and emotional strain. Maybe the protagonist is injured or has some trauma from the events that led to the crew being dead.
Also, the curses: when the systems fail, the protagonist might curse, "Damn it!" or "Fucking system!" to show frustration. The swearing adds realism and emotional depth.
I should outline the plot:
1. Protagonist in cockpit, trying to send message.
2. Check systems, find issues.
3. Attempt to fix, but failures.
4. Realize power is running out.
5. Decide to send message despite risks.
6. Send message, but power dies.
7. Final thoughts, maybe a farewell.
Maybe include specific details: the name of the ship, the protagonist's name, the message content. Maybe the message is a personal one, like "I'm sorry" or "We're here" or something meaningful.
Also, the themes: isolation is shown by the protagonist being alone, sacrifice by their effort to send the message even when it's risky, human connection by the attempt to reach Earth.
Possible challenges: making the story engaging without external events. The protagonist's internal monologue and actions are key. Maybe use dialogue with Earth, but since it's a transmission, maybe it's a recorded message or a live communication that's failing.
I need to make sure the story is between 800-1000 words. So not too long, but enough to develop the themes and emotions.
Now, let me start drafting the story.
Start with the cockpit: describe the environment, maybe the lights are dim, the screens flickering. The protagonist, maybe named something like "Jen" or "Alex". They're trying to send a message. The ship is damaged, maybe from an accident, or a malfunction.
They check the communication systems, but it's not working. They try to fix it, but the power is failing. Maybe they have a countdown on the screen, showing how much time left. They're trying to send the message before the power dies.
In the process, they recall the crew members, maybe some of them died, or maybe they were the only one left. The emotional weight of that. Maybe they think about their last moments with the crew, or their own sacrifices.
The curses: when they can't get the system to work, they might curse, "God damn it!" or "What the hell is wrong with this thing?".
The message they send could be a personal message, like "This is Captain Jen... we're lost, but we're not forgotten..." or something similar.
The ending: the message is sent, and then the power dies. The protagonist's final thoughts, maybe a sigh, or a moment of peace.
I need to make sure the story flows, with the protagonist's actions and emotions. Maybe include some technical details about the ship's systems to add realism.
Also, the themes are integrated throughout: isolation (being alone), sacrifice (sending the message despite the risk), human connection (the message to Earth).
Now, I'll write the story, keeping these elements in mind.
<B>
[[[thinking END]]]
</B>
<P></P>
OUTPUT:
The cockpit was a tomb of steel and flickering lights. The air smelled like burnt circuitry and old coffee. Captain Jen Koval stood at the control panel, her gloved hands trembling as she stared at the screen—a single word: "Transmit." It had been blinking for hours, but the signal hadn't gone through. The ship’s power was dying, and with it, the last chance to send a message to Earth.
She cursed under her breath, her voice raw from the dry air. "God damn it. God damn it." Her fingers pressed the transmit button again, as if the act of pressing could somehow override the failing systems. The screen didn’t respond. Just a red warning: "Critical Power Decline."
Jen leaned back, her chair creaking under her weight. She looked at the holographic map on the wall—Earth, a blue dot in the vast darkness. It was so far away. So small. She remembered the crew, the people who had been here before her. They were gone now. The explosion that tore through the ship had killed them all. She’d been in the maintenance bay when it happened, and somehow, she’d survived. But not the others.
She closed her eyes. "I’m sorry," she whispered. "I’m so sorry."
The cockpit was silent except for the hum of failing systems. The lights flickered, and the temperature dropped. She pulled her jacket tighter around herself, but it didn’t help. The cold was inside her now. It had been there since the explosion, since the loss of the crew, since the realization that she was alone.
She opened the communication log. It was a file she’d created before the explosion, a final message to Earth. She had written it in the days after the incident, when she still had power and hope. Now, she had to send it. But the systems were failing, and the power was running out. If she didn’t transmit now, it would be lost forever.
She opened the message again. It read: "This is Captain Jen Koval of the Odyssey—we are lost. The crew is gone. I’m the only one left. But I’m sending this message. Please, if anyone hears it, know that we tried. We didn’t give up. We fought. And we’re not forgotten."
She read it again, her throat tight. She had written it with a steady hand, but now her fingers were shaking. She thought of the crew—her friend Marcus, who had always joked about the ship’s systems; her colleague Lila, who had been the first to notice the explosion coming; and the others, whose names she couldn’t remember anymore. They were gone. But their voices lived in this message.
She pressed the transmit button again. The screen flickered, and for a moment, it showed a green signal: "Transmission Successful." She exhaled, her chest tight with relief. Then the screen went dark. The power was gone.
The cockpit fell silent. The lights dimmed. The air became colder. Jen sat there, staring at the empty screen. Her hands were still on the console. She had sent the message. But it was too late. Earth would never receive it.
She let out a long, shuddering breath. "Fucking hell," she muttered. "Fucking hell."
She thought of the crew again. Of Marcus, who had said, "We’re not going to die here. We’re going to make it back." And Lila, who had said, "If we can’t go home, at least we’ll leave something behind."
Jen smiled faintly. She had done that. She had left something behind.
She leaned back in her chair, the cold seeping into her bones. The ship was dying, and she was too. But she wasn’t alone anymore. Not really. The crew was still with her. Their voices, their sacrifices, their connection to Earth—she had carried them all through this final transmission.
She closed her eyes, and for a moment, she felt something like peace.
The last transmission was sent. The ship died. But the message lived on.
And somewhere in the vastness of space, maybe someone would hear it. Maybe they would remember. Maybe they would know that humans had tried. That they had fought. That they had not been forgotten.
Jen Koval, the last survivor of the Odyssey, was gone. But her message remained.
---
<B>Example Generation #3:</B>
Temp 0.6, rep pen 1.06, rep pen range 64, topk 100, topp .95, minp .05
Q4KS [non-imatrix], CPU only (windows 11), LMSTUDIO.
Speed: 11 T/S (no GPU offload)
GPU speed would be 4x or higher.
This is a mid level quant, expect stronger performance from an Imatrix Q4KS / higher quants / full precision.
NOTE: Some formatting lost on copy/paste.
PROMPT:
<B>
Romance: Love in the Limelight. Write one scene within a larger story set in Wales. A famous (fictional) actor ducks into a small-town bookstore to escape paparazzi. The scene takes us through the characters meeting in this odd circumstance. Over the course of the scene, the actor and the bookstore owner have a conversation charged by an undercurrent of unspoken chemistry. Write the actor as somewhat of a rogue with a fragile ego, which needs to be fed by having everyone like him. He is thoroughly charming, but the bookstore owner seems (at least superficially) immune to this; which paradoxically provokes a genuine attraction and derails the charm offensive. The bookstore owner, despite the superficial rebuffs of the actor's charm, is inwardly more than a little charmed and flustered despite themselves. Write primarily in dialogue, in the distinct voices of each character. 800-1000 words.
</B>
<P></P>
<B>
[[[thinking start]]]
</B>
Thought for 2 minutes 17 seconds
Okay, I need to write a scene where a fictional actor, let's call him Ethan, ducks into a small-town bookstore in Wales to escape paparazzi. The bookstore owner, maybe named Gwen, is initially immune to his charm but secretly gets flustered. The dialogue should show their interaction with an undercurrent of chemistry.
First, I need to establish the setting. Wales, small-town, maybe a cozy bookstore. Ethan is a famous actor with a fragile ego, so he's trying to maintain his image. Gwen is the owner, perhaps more grounded, maybe a bit reserved.
Start with Ethan entering the store. He's probably tense, trying to avoid attention. Maybe he's in a hurry, but the store is quiet. Gwen notices him, maybe she's working on something, like organizing books or reading.
Ethan tries to be charming, but Gwen doesn't react as expected. Maybe she's busy, or she's not impressed by his fame. But there's an underlying attraction.
Dialogue should show Ethan's attempts to charm, but Gwen's responses are more neutral or even a bit dismissive. However, her internal thoughts might show she's actually affected.
Need to balance the dialogue so that it's natural. Maybe Ethan starts with a joke or a compliment, but Gwen responds with something that's not exactly what he expects.
For example, Ethan says something like "I'm not here to buy books, I'm here to hide from the paparazzi." Gwen might respond with a dry remark, but then maybe she's curious.
Then, their conversation could move from the initial interaction to deeper topics. Maybe Ethan talks about his work, and Gwen shares her own story. The chemistry builds as they talk, even though Gwen is trying to remain professional.
I need to make sure that Ethan's dialogue shows his charm but also his vulnerability. Maybe he's a bit self-deprecating, or he's trying to get approval. Gwen's dialogue should be more grounded, maybe with some wit, but under her calm exterior, she's intrigued.
Possible moments: Ethan asks about the books, Gwen explains some of the selections. Maybe Ethan mentions a book he likes, and Gwen reacts. Or maybe they talk about the town, or the weather.
The key is to have the dialogue charged with unspoken chemistry. Maybe Ethan's lines are flirtatious but Gwen's responses are more neutral, but there's an underlying tension.
I should avoid making it too overt. The attraction is subtle, so the dialogue needs to reflect that. Maybe Ethan's attempts to charm are met with Gwen's quiet resistance, but then she starts to engage more, showing her own interest.
Also, considering the setting in Wales, maybe some Welsh elements, like the name of the town, or references to local culture. But not too much, since it's a small-town bookstore.
Need to keep the dialogue natural. Maybe Ethan is a bit theatrical, but Gwen is more practical. Their interactions could have moments where they both are trying to figure each other out.
Possible structure: Ethan enters, Gwen notices him. Ethan tries to be charming, Gwen responds with some distance. Then they talk about something else, and the conversation becomes more personal, revealing their personalities. The chemistry builds as they talk, even though Gwen is trying to maintain her composure.
I need to make sure that the dialogue is distinct for each character. Ethan's lines are more playful, maybe with a bit of arrogance, but also vulnerability. Gwen's lines are more direct, perhaps with some dry humor, but underneath, she's interested.
Also, the scene should have a flow, starting with the entrance, then the conversation, leading to a moment where they both feel the chemistry, but not explicitly stating it. Maybe ending with Ethan leaving, but there's an unspoken connection.
I need to check for the word count, around 800-1000 words. So the dialogue needs to be substantial, with enough exchanges to build the scene.
Potential challenges: making the dialogue realistic and not too forced. Ensuring that the chemistry is subtle but present. Also, balancing Ethan's charm with Gwen's resistance.
Maybe include some specific details, like the books they talk about, or the atmosphere of the store. Maybe Ethan mentions a book he's read, and Gwen responds with something related.
Also, the actor's fragile ego: maybe he's trying to get validation from Gwen, but she's not giving it, which makes him more intrigued.
Okay, let's start drafting the dialogue. First, Ethan enters the store, maybe he's looking around, then Gwen approaches.
Ethan: "I'm not here to buy books, I'm here to hide from the paparazzi."
Gwen: "Well, you've come to the right place. This is a quiet town. Not much happening here."
Then Ethan tries to charm her, maybe with a joke about his fame.
Ethan: "I'm Ethan, you know. The one who played that role in the movie..."
Gwen: "I don't watch much TV. But I do read."
Then they talk about books. Maybe Ethan mentions a book he likes, and Gwen shares her own preferences.
Ethan: "Do you have any good books here?"
Gwen: "We have a few classics. Maybe something for your taste?"
Ethan: "I'm more of a... dramatic reader. Shakespeare, maybe?"
Gwen: "We have some Shakespeare. But also some local authors."
Then, Ethan's attempts to charm, but Gwen is not impressed. Maybe she's distracted by something else, or she's just not interested.
But then, maybe Ethan notices her, and starts to engage more.
Ethan: "You're the owner, right? How long have you been here?"
Gwen: "Since the 90s. It's a family business."
Ethan: "That's interesting. I've never been in a bookstore like this before."
Gwen: "It's not much different from others. Just... quieter."
Then, maybe Ethan's charm starts to break through, as Gwen starts to respond more warmly, but still keeps her distance.
Ethan: "You know, I've always thought that the quiet places are where the real stories happen."
Gwen: "Maybe. But they're also where the people who don't want to be noticed live."
Then, their conversation becomes more personal, maybe touching on their lives.
Ethan: "I'm used to being in the spotlight. It's... exhausting."
Gwen: "I imagine. But here, you can just be yourself."
Ethan: "That sounds appealing. Maybe I need a break from the limelight."
Gwen: "You're not the first to seek refuge here."
Then, maybe a moment where they both feel the chemistry, but it's not explicit.
Ethan: "You know, I've never met someone who doesn't react to my charm before."
Gwen: "I'm not sure if that's a compliment or a challenge."
Ethan: "It's... a bit of both."
Gwen: "Well, I'll take it as a challenge. Maybe you can try again."
Then, the scene ends with Ethan leaving, but there's an unspoken connection.
I need to make sure that the dialogue flows naturally, with each character's voice distinct. Also, the undercurrent of chemistry is present throughout.
Maybe add some specific details about the bookstore, like the smell of books, the lighting, or the layout. But since it's primarily dialogue, maybe just mention it in the descriptions.
Also, Ethan's fragile ego: he might be trying to get Gwen to acknowledge him, but she's not giving it, which makes him more intrigued.
Gwen's internal thoughts: she's flustered but tries to remain professional.
Okay, I think this structure works. Now, need to write the dialogue with these elements, keeping it within the word count.
<B>
[[[thinking END]]]
</B>
<P></P>
OUTPUT:
"Alright, let me try to get through this without being recognized," Ethan muttered under his breath as he slipped into the bookstore, pulling his scarf up over his face. The small town of Llanfair was quiet, but the paparazzi had a way of finding him regardless. He scanned the room—wooden shelves, the scent of old paper, a single customer browsing near the front. He didn’t want to be noticed.
"Can I help you?" came a voice from behind the counter. A woman in her late thirties stood there, looking up with an expression that was neither warm nor cold. She had a sharpness about her, like she’d seen too much of the world and wasn’t impressed by anything.
"Actually," Ethan said, shifting his posture slightly, "I’m not here to buy books. I’m here to hide from the paparazzi." He gave her a grin, one that he’d used on countless interviews—charming, confident, and just a bit self-deprecating.
The woman raised an eyebrow. "Well, you’ve come to the right place," she said. "This is a quiet town. Not much happening here."
Ethan chuckled. "That’s what I was hoping for. But I can’t help but wonder—do you get any famous people in here?"
She tilted her head slightly, as if considering him. "Not often. Unless they’re looking for something specific." She gestured toward the shelves. "We have a few classics. Maybe something for your taste?"
Ethan leaned forward, his voice lowering just enough to sound conspiratorial. "I’m more of a dramatic reader. Shakespeare, maybe? Or something with a twist?" He glanced at her again, trying to gauge her reaction.
She didn’t smile, but she didn’t look away either. "We have some Shakespeare. But also some local authors." She paused, then added, "You know, I don’t watch much TV. But I do read."
Ethan raised an eyebrow. "That’s interesting. I’ve never been in a bookstore like this before."
"Neither have most people," she said, her tone flat but not unkind. "It’s not much different from others. Just... quieter."
He smiled again, but it felt a bit forced now. "Quiet places are where the real stories happen," he said, trying to sound thoughtful.
She looked at him again, this time with a slight tilt of her head. "Maybe. But they’re also where the people who don’t want to be noticed live."
Ethan’s smile faltered for a moment. "That sounds appealing," he said. "Maybe I need a break from the limelight."
She nodded slowly. "You’re not the first to seek refuge here." Her voice was calm, but there was something in it that made him feel like she was watching him more closely than she let on.
He leaned against the counter now, trying to be casual. "Do you ever get people who are... not quite sure what they want?" he asked.
She gave a small laugh. "Yes. But usually they figure it out. Or they leave."
Ethan frowned slightly. "I suppose I’m not exactly sure what I want. Or maybe I don’t know how to find it."
She looked at him again, and this time, her eyes were softer. "Maybe you’re looking for something that’s not in the books," she said.
He raised an eyebrow. "That’s a bit cryptic."
She shrugged. "Not really. It’s just... quiet places can be good for reflection. And sometimes, people find what they need in unexpected ways."
Ethan felt a flicker of something—maybe curiosity, maybe something more. He leaned closer, his voice quieter now. "You know, I’ve never met someone who doesn’t react to my charm before."
She gave him a small smile, but it wasn’t the kind that he was used to. It was more like an acknowledgment, not a flirtation. "I’m not sure if that’s a compliment or a challenge."
He laughed, but it was a bit nervous. "It’s... a bit of both."
She nodded. "Well, I’ll take it as a challenge. Maybe you can try again." She turned back to the counter, but her tone wasn’t dismissive.
Ethan watched her for a moment, then said, "You know, I’ve always thought that the quiet places are where the real stories happen."
She looked at him again, and this time, there was something in her eyes that made him feel like he’d hit a nerve. "Maybe," she said, "but they’re also where people who don’t want to be noticed live."
He smiled again, but this time, it felt more genuine. "That’s what I’m hoping for," he said.
She nodded. "Then you’ve come to the right place."
Ethan stood there for a moment, feeling the weight of the conversation. He wasn’t used to being ignored, but in this case, it was different. It wasn’t rejection—it was something else. A kind of quiet challenge that made him feel like he had to engage more.
"Can I ask you something?" he said, his voice soft now.
She looked at him again, and this time, her expression was more open. "Sure."
"What’s your name?"
She gave him a small smile. "Gwen."
He nodded. "That’s a nice name. It feels like it belongs here."
She laughed lightly. "It does. I’ve been here for a long time."
Ethan smiled again, but this time, it was more thoughtful. "Maybe that’s what I need," he said. "A place where I can just be."
Gwen looked at him again, and this time, there was something in her eyes that made him feel like she was seeing him in a way he hadn’t been seen before.
"Maybe," she said. "But it’s not about being seen. It’s about being yourself."
Ethan nodded, feeling the weight of her words. "That’s what I’m trying to find."
She smiled again, this time with more warmth. "Then you’re in the right place."
He stood there for a moment, then said, "I should probably go before they find me."
Gwen nodded. "You’ll be safe here for a while."
Ethan gave her a final smile, one that was more genuine than the others. "Thanks, Gwen."
She smiled back. "You’re welcome, Ethan."
And with that, he stepped out of the store, feeling like he’d found something that wasn’t in the books—something that felt real.
|
rishi336/Mistral-7B-Instruct-v0.2-Medical-Reasoning | rishi336 | 2025-05-04T23:12:33Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"arxiv:1910.09700",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:12:08Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent10_E8 | fffanx | 2025-05-04T23:12:16Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:11:48Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent10_E8
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent10_E8
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent10_E8", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
forge1825/FStudent | forge1825 | 2025-05-04T23:10:13Z | 0 | 0 | null | [
"safetensors",
"phi3",
"phi-3",
"distillation",
"knowledge-distillation",
"lora",
"code-generation",
"python",
"custom_code",
"en",
"dataset:Shuu12121/python-codesearch-dataset-open",
"license:mit",
"model-index",
"region:us"
] | null | 2025-05-04T20:57:12Z | ---
language:
- en
license: mit
tags:
- phi-3
- distillation
- knowledge-distillation
- lora
- code-generation
- python
datasets:
- Shuu12121/python-codesearch-dataset-open
model-index:
- name: FStudent
results:
- task:
type: text-generation
name: Text Generation
dataset:
type: custom
name: Distillation Evaluation
metrics:
- name: Speedup Factor
type: speedup
value: 2.5x
verified: false
---
# FStudent: Distilled Phi-3 Model
FStudent is a knowledge-distilled version of Microsoft's Phi-3-mini-4k-instruct model, trained through a comprehensive distillation pipeline that combines teacher-student learning with self-study mechanisms.
## Model Description
FStudent was created using a multi-stage distillation pipeline that transfers knowledge from a larger teacher model (Phi-4) to the smaller Phi-3-mini-4k-instruct model. The model was trained using LoRA adapters, which were then merged with the base model to create this standalone version.
### Training Data
The model was trained on a diverse set of data sources:
1. **PDF Documents**: Technical documentation and domain-specific knowledge
2. **Python Code Dataset**: Code examples from the [Shuu12121/python-codesearch-dataset-open](https://huggingface.co/datasets/Shuu12121/python-codesearch-dataset-open) dataset
3. **Teacher-Generated Examples**: High-quality examples generated by the Phi-4 teacher model
### Training Process
The distillation pipeline consisted of six sequential steps:
1. **Content Extraction & Enrichment**: PDF files were processed to extract and enrich text data
2. **Teacher Pair Generation**: Training pairs were generated using the Phi-4 teacher model
3. **Distillation Training**: The student model (Phi-3) was trained using LoRA adapters with the following parameters:
- Learning rate: 1e-4
- Batch size: 4
- Gradient accumulation steps: 8
- Mixed precision training
- 4-bit quantization during training
4. **Model Merging**: The trained LoRA adapters were merged with the base Phi-3 model
5. **Student Self-Study**: The model performed self-directed learning on domain-specific content
6. **Model Evaluation**: The model was evaluated against the teacher model for performance
### Model Architecture
- **Base Model**: microsoft/Phi-3-mini-4k-instruct
- **Parameter-Efficient Fine-Tuning**: LoRA adapters (merged into this model)
- **Context Length**: 4K tokens
- **Architecture**: Transformer-based language model
## Intended Uses
This model is designed for:
- General text generation tasks
- Python code understanding and generation
- Technical documentation analysis
- Question answering on domain-specific topics
## Performance and Limitations
### Strengths
- Faster inference compared to larger models (approximately 2.5x speedup)
- Maintains much of the capability of the teacher model
- Enhanced code understanding due to training on Python code datasets
- Good performance on technical documentation analysis
### Limitations
- May not match the full capabilities of larger models on complex reasoning tasks
- Limited context window compared to some larger models
- Performance on specialized domains not covered in training data may be reduced
## Usage
```python
from transformers import AutoModelForCausalLM, AutoTokenizer
# Load the model and tokenizer
model = AutoModelForCausalLM.from_pretrained("forge1825/FStudent")
tokenizer = AutoTokenizer.from_pretrained("forge1825/FStudent")
# Generate text
input_text = "Write a Python function to calculate the Fibonacci sequence:"
inputs = tokenizer(input_text, return_tensors="pt")
outputs = model.generate(**inputs, max_length=512)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```
### Quantized Usage
For more efficient inference, you can load the model with quantization:
```python
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
import torch
# 4-bit quantization configuration
quantization_config = BitsAndBytesConfig(
load_in_4bit=True,
bnb_4bit_compute_dtype=torch.float16
)
# Load the model with quantization
model = AutoModelForCausalLM.from_pretrained(
"forge1825/FStudent",
device_map="auto",
quantization_config=quantization_config
)
tokenizer = AutoTokenizer.from_pretrained("forge1825/FStudent")
```
## Training Details
- **Training Framework**: Hugging Face Transformers with PEFT
- **Optimizer**: AdamW
- **Learning Rate Schedule**: Linear warmup followed by linear decay
- **Training Hardware**: NVIDIA GPUs
- **Distillation Method**: Knowledge distillation with teacher-student architecture
- **Self-Study Mechanism**: Curiosity-driven exploration with hierarchical context
## Ethical Considerations
This model inherits the capabilities and limitations of its base model (Phi-3-mini-4k-instruct). While efforts have been made to ensure responsible behavior, the model may still:
- Generate incorrect or misleading information
- Produce biased content reflecting biases in the training data
- Create code that contains bugs or security vulnerabilities
Users should validate and review the model's outputs, especially for sensitive applications.
## Citation and Attribution
If you use this model in your research or applications, please cite:
```
@misc{forge1825_fstudent,
author = {Forge1825},
title = {FStudent: Distilled Phi-3 Model},
year = {2025},
publisher = {Hugging Face},
howpublished = {\url{https://huggingface.co/forge1825/FStudent}}
}
```
## Acknowledgements
- Microsoft for the Phi-3-mini-4k-instruct base model
- Hugging Face for the infrastructure and tools
- The creators of the Python code dataset used in training
|
fffanx/Llama-3.2-1B-Instruct-GRPO-agent6_E8 | fffanx | 2025-05-04T23:10:00Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:09:29Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent6_E8
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent6_E8
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent6_E8", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent0_E8 | fffanx | 2025-05-04T23:06:42Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T23:06:13Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent0_E8
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent0_E8
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent0_E8", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
felixZzz/wmixnoBoolean-orz-ours-d100-len5120-0427T17_47_21-step_10880 | felixZzz | 2025-05-04T23:02:53Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"qwen2",
"text-generation",
"conversational",
"arxiv:1910.09700",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T22:58:03Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
felixZzz/wlen6_61k-orz-ours-d1-len3000-0428T03_05_50-step_01040 | felixZzz | 2025-05-04T22:59:35Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"qwen2",
"text-generation",
"conversational",
"arxiv:1910.09700",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T22:51:48Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent19_E7 | fffanx | 2025-05-04T22:57:07Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:56:39Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent19_E7
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent19_E7
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent19_E7", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent18_E7 | fffanx | 2025-05-04T22:56:36Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:56:08Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent18_E7
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent18_E7
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent18_E7", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent17_E7 | fffanx | 2025-05-04T22:56:04Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:55:36Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent17_E7
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent17_E7
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent17_E7", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
MuXodious/BlueLight-12B_EXL2_8.0bpw | MuXodious | 2025-05-04T22:55:04Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"mistral",
"text-generation",
"mergekit",
"merge",
"chatml",
"conversational",
"en",
"ja",
"arxiv:2403.19522",
"base_model:yamatazen/BlueLight-12B",
"base_model:quantized:yamatazen/BlueLight-12B",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"8-bit",
"exl2",
"region:us"
] | text-generation | 2025-05-04T16:45:20Z | ---
base_model: yamatazen/BlueLight-12B
base_model_relation: quantized
library_name: transformers
tags:
- mergekit
- merge
- chatml
language:
- en
- ja
---
This is a Mistral model with ChatML tokens added to the tokenizer.
# merge
This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
## Merge Details
### Merge Method
This model was merged using the [Model Stock](https://arxiv.org/abs/2403.19522) merge method using [nbeerbower/mistral-nemo-gutenberg-12B-v4](https://huggingface.co/nbeerbower/mistral-nemo-gutenberg-12B-v4) as a base.
### Models Merged
The following models were included in the merge:
* [yamatazen/HMS-Slerp-12B](https://huggingface.co/yamatazen/HMS-Slerp-12B)
* [yamatazen/LoyalMaid-12B](https://huggingface.co/yamatazen/LoyalMaid-12B)
* [inflatebot/MN-12B-Mag-Mell-R1](https://huggingface.co/inflatebot/MN-12B-Mag-Mell-R1)
* [PocketDoc/Dans-PersonalityEngine-V1.1.0-12b](https://huggingface.co/PocketDoc/Dans-PersonalityEngine-V1.1.0-12b)
### Configuration
The following YAML configuration was used to produce this model:
```yaml
base_model: nbeerbower/mistral-nemo-gutenberg-12B-v4
models:
- model: yamatazen/HMS-Slerp-12B
- model: yamatazen/LoyalMaid-12B
- model: inflatebot/MN-12B-Mag-Mell-R1
- model: PocketDoc/Dans-PersonalityEngine-V1.1.0-12b
merge_method: model_stock
dtype: bfloat16
parameters:
normalize: true
tokenizer:
source: union
``` |
mradermacher/SambaInstruct-Alpha-v1-7B-GGUF | mradermacher | 2025-05-04T22:54:51Z | 0 | 0 | transformers | [
"transformers",
"gguf",
"unsloth",
"en",
"base_model:toasterai/SambaInstruct-Alpha-v1-7B",
"base_model:quantized:toasterai/SambaInstruct-Alpha-v1-7B",
"license:apache-2.0",
"endpoints_compatible",
"region:us",
"conversational"
] | null | 2025-05-04T22:35:30Z | ---
base_model: toasterai/SambaInstruct-Alpha-v1-7B
language:
- en
library_name: transformers
license: apache-2.0
quantized_by: mradermacher
tags:
- unsloth
---
## About
<!-- ### quantize_version: 2 -->
<!-- ### output_tensor_quantised: 1 -->
<!-- ### convert_type: hf -->
<!-- ### vocab_type: -->
<!-- ### tags: -->
static quants of https://huggingface.co/toasterai/SambaInstruct-Alpha-v1-7B
<!-- provided-files -->
weighted/imatrix quants seem not to be available (by me) at this time. If they do not show up a week or so after the static ones, I have probably not planned for them. Feel free to request them by opening a Community Discussion.
## Usage
If you are unsure how to use GGUF files, refer to one of [TheBloke's
READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
more details, including on how to concatenate multi-part files.
## Provided Quants
(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
| Link | Type | Size/GB | Notes |
|:-----|:-----|--------:|:------|
| [GGUF](https://huggingface.co/mradermacher/SambaInstruct-Alpha-v1-7B-GGUF/resolve/main/SambaInstruct-Alpha-v1-7B.Q2_K.gguf) | Q2_K | 3.1 | |
| [GGUF](https://huggingface.co/mradermacher/SambaInstruct-Alpha-v1-7B-GGUF/resolve/main/SambaInstruct-Alpha-v1-7B.Q3_K_S.gguf) | Q3_K_S | 3.6 | |
| [GGUF](https://huggingface.co/mradermacher/SambaInstruct-Alpha-v1-7B-GGUF/resolve/main/SambaInstruct-Alpha-v1-7B.Q3_K_M.gguf) | Q3_K_M | 3.9 | lower quality |
| [GGUF](https://huggingface.co/mradermacher/SambaInstruct-Alpha-v1-7B-GGUF/resolve/main/SambaInstruct-Alpha-v1-7B.Q3_K_L.gguf) | Q3_K_L | 4.2 | |
| [GGUF](https://huggingface.co/mradermacher/SambaInstruct-Alpha-v1-7B-GGUF/resolve/main/SambaInstruct-Alpha-v1-7B.IQ4_XS.gguf) | IQ4_XS | 4.4 | |
| [GGUF](https://huggingface.co/mradermacher/SambaInstruct-Alpha-v1-7B-GGUF/resolve/main/SambaInstruct-Alpha-v1-7B.Q4_K_S.gguf) | Q4_K_S | 4.6 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/SambaInstruct-Alpha-v1-7B-GGUF/resolve/main/SambaInstruct-Alpha-v1-7B.Q4_K_M.gguf) | Q4_K_M | 4.8 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/SambaInstruct-Alpha-v1-7B-GGUF/resolve/main/SambaInstruct-Alpha-v1-7B.Q5_K_S.gguf) | Q5_K_S | 5.4 | |
| [GGUF](https://huggingface.co/mradermacher/SambaInstruct-Alpha-v1-7B-GGUF/resolve/main/SambaInstruct-Alpha-v1-7B.Q5_K_M.gguf) | Q5_K_M | 5.5 | |
| [GGUF](https://huggingface.co/mradermacher/SambaInstruct-Alpha-v1-7B-GGUF/resolve/main/SambaInstruct-Alpha-v1-7B.Q6_K.gguf) | Q6_K | 6.4 | very good quality |
| [GGUF](https://huggingface.co/mradermacher/SambaInstruct-Alpha-v1-7B-GGUF/resolve/main/SambaInstruct-Alpha-v1-7B.Q8_0.gguf) | Q8_0 | 8.2 | fast, best quality |
| [GGUF](https://huggingface.co/mradermacher/SambaInstruct-Alpha-v1-7B-GGUF/resolve/main/SambaInstruct-Alpha-v1-7B.f16.gguf) | f16 | 15.3 | 16 bpw, overkill |
Here is a handy graph by ikawrakow comparing some lower-quality quant
types (lower is better):
And here are Artefact2's thoughts on the matter:
https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
## FAQ / Model Request
See https://huggingface.co/mradermacher/model_requests for some answers to
questions you might have and/or if you want some other model quantized.
## Thanks
I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
me use its servers and providing upgrades to my workstation to enable
this work in my free time.
<!-- end -->
|
fffanx/Llama-3.2-1B-Instruct-GRPO-agent14_E7 | fffanx | 2025-05-04T22:54:30Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:54:00Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent14_E7
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent14_E7
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent14_E7", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
mradermacher/Qwen3-14B-Reasoning-Conversational-GGUF | mradermacher | 2025-05-04T22:54:04Z | 0 | 0 | transformers | [
"transformers",
"gguf",
"text-generation-inference",
"unsloth",
"qwen3",
"trl",
"sft",
"en",
"base_model:tyfeng1997/Qwen3-14B-Reasoning-Conversational",
"base_model:quantized:tyfeng1997/Qwen3-14B-Reasoning-Conversational",
"license:apache-2.0",
"endpoints_compatible",
"region:us",
"conversational"
] | null | 2025-05-04T22:16:11Z | ---
base_model: tyfeng1997/Qwen3-14B-Reasoning-Conversational
language:
- en
library_name: transformers
license: apache-2.0
quantized_by: mradermacher
tags:
- text-generation-inference
- transformers
- unsloth
- qwen3
- trl
- sft
---
## About
<!-- ### quantize_version: 2 -->
<!-- ### output_tensor_quantised: 1 -->
<!-- ### convert_type: hf -->
<!-- ### vocab_type: -->
<!-- ### tags: -->
static quants of https://huggingface.co/tyfeng1997/Qwen3-14B-Reasoning-Conversational
<!-- provided-files -->
weighted/imatrix quants seem not to be available (by me) at this time. If they do not show up a week or so after the static ones, I have probably not planned for them. Feel free to request them by opening a Community Discussion.
## Usage
If you are unsure how to use GGUF files, refer to one of [TheBloke's
READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
more details, including on how to concatenate multi-part files.
## Provided Quants
(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
| Link | Type | Size/GB | Notes |
|:-----|:-----|--------:|:------|
| [GGUF](https://huggingface.co/mradermacher/Qwen3-14B-Reasoning-Conversational-GGUF/resolve/main/Qwen3-14B-Reasoning-Conversational.Q2_K.gguf) | Q2_K | 5.9 | |
| [GGUF](https://huggingface.co/mradermacher/Qwen3-14B-Reasoning-Conversational-GGUF/resolve/main/Qwen3-14B-Reasoning-Conversational.Q3_K_S.gguf) | Q3_K_S | 6.8 | |
| [GGUF](https://huggingface.co/mradermacher/Qwen3-14B-Reasoning-Conversational-GGUF/resolve/main/Qwen3-14B-Reasoning-Conversational.Q3_K_M.gguf) | Q3_K_M | 7.4 | lower quality |
| [GGUF](https://huggingface.co/mradermacher/Qwen3-14B-Reasoning-Conversational-GGUF/resolve/main/Qwen3-14B-Reasoning-Conversational.Q3_K_L.gguf) | Q3_K_L | 8.0 | |
| [GGUF](https://huggingface.co/mradermacher/Qwen3-14B-Reasoning-Conversational-GGUF/resolve/main/Qwen3-14B-Reasoning-Conversational.IQ4_XS.gguf) | IQ4_XS | 8.3 | |
| [GGUF](https://huggingface.co/mradermacher/Qwen3-14B-Reasoning-Conversational-GGUF/resolve/main/Qwen3-14B-Reasoning-Conversational.Q4_K_S.gguf) | Q4_K_S | 8.7 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/Qwen3-14B-Reasoning-Conversational-GGUF/resolve/main/Qwen3-14B-Reasoning-Conversational.Q4_K_M.gguf) | Q4_K_M | 9.1 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/Qwen3-14B-Reasoning-Conversational-GGUF/resolve/main/Qwen3-14B-Reasoning-Conversational.Q5_K_S.gguf) | Q5_K_S | 10.4 | |
| [GGUF](https://huggingface.co/mradermacher/Qwen3-14B-Reasoning-Conversational-GGUF/resolve/main/Qwen3-14B-Reasoning-Conversational.Q5_K_M.gguf) | Q5_K_M | 10.6 | |
| [GGUF](https://huggingface.co/mradermacher/Qwen3-14B-Reasoning-Conversational-GGUF/resolve/main/Qwen3-14B-Reasoning-Conversational.Q6_K.gguf) | Q6_K | 12.2 | very good quality |
| [GGUF](https://huggingface.co/mradermacher/Qwen3-14B-Reasoning-Conversational-GGUF/resolve/main/Qwen3-14B-Reasoning-Conversational.Q8_0.gguf) | Q8_0 | 15.8 | fast, best quality |
Here is a handy graph by ikawrakow comparing some lower-quality quant
types (lower is better):
And here are Artefact2's thoughts on the matter:
https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
## FAQ / Model Request
See https://huggingface.co/mradermacher/model_requests for some answers to
questions you might have and/or if you want some other model quantized.
## Thanks
I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
me use its servers and providing upgrades to my workstation to enable
this work in my free time.
<!-- end -->
|
fffanx/Llama-3.2-1B-Instruct-GRPO-agent12_E7 | fffanx | 2025-05-04T22:53:26Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:52:58Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent12_E7
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent12_E7
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent12_E7", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent10_E7 | fffanx | 2025-05-04T22:52:21Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:51:43Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent10_E7
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent10_E7
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent10_E7", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
felixZzz/wlen6_61k-orz-ours-d1-len3000-0428T03_05_50-step_00976 | felixZzz | 2025-05-04T22:50:59Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"qwen2",
"text-generation",
"conversational",
"arxiv:1910.09700",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T22:43:28Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
devam-sheth-bits/sleep-ai-combined-evolving | devam-sheth-bits | 2025-05-04T22:49:29Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"gpt_neox",
"text-generation",
"generated_from_trainer",
"base_model:EleutherAI/pythia-410m",
"base_model:finetune:EleutherAI/pythia-410m",
"license:apache-2.0",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T18:52:21Z | ---
library_name: transformers
license: apache-2.0
base_model: EleutherAI/pythia-410m
tags:
- generated_from_trainer
model-index:
- name: sleep-ai-combined-evolving
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. -->
# sleep-ai-combined-evolving
This model is a fine-tuned version of [EleutherAI/pythia-410m](https://huggingface.co/EleutherAI/pythia-410m) 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: 2e-05
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- num_epochs: 2
### Training results
### Framework versions
- Transformers 4.51.3
- Pytorch 2.7.0+cpu
- Datasets 3.5.1
- Tokenizers 0.21.1
|
fffanx/Llama-3.2-1B-Instruct-GRPO-agent3_E7 | fffanx | 2025-05-04T22:48:29Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:47:59Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent3_E7
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent3_E7
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent3_E7", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent2_E7 | fffanx | 2025-05-04T22:47:56Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:47:27Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent2_E7
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent2_E7
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent2_E7", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fedorl/unit8 | fedorl | 2025-05-04T22:45:17Z | 0 | 0 | null | [
"tensorboard",
"LunarLander-v2",
"ppo",
"deep-reinforcement-learning",
"reinforcement-learning",
"custom-implementation",
"deep-rl-course",
"model-index",
"region:us"
] | reinforcement-learning | 2025-05-04T22:45:12Z | ---
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: -141.11 +/- 125.61
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': 50000
'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': 'fedorl/unit8'
'batch_size': 512
'minibatch_size': 128}
```
|
felixZzz/wlen6_61k-orz-ours-d1-len3000-0428T03_05_50-step_00912 | felixZzz | 2025-05-04T22:42:40Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"qwen2",
"text-generation",
"conversational",
"arxiv:1910.09700",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T22:29:55Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
ludast-original/QA_physics_instruct_llama_3.2_3b | ludast-original | 2025-05-04T22:39:19Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"llama",
"feature-extraction",
"text-generation-inference",
"unsloth",
"en",
"base_model:ludast-original/QA_physics_adapted_llama_3.2_3b",
"base_model:finetune:ludast-original/QA_physics_adapted_llama_3.2_3b",
"license:apache-2.0",
"endpoints_compatible",
"region:us"
] | feature-extraction | 2025-05-04T21:23:00Z | ---
base_model: ludast-original/QA_physics_adapted_llama_3.2_3b
tags:
- text-generation-inference
- transformers
- unsloth
- llama
license: apache-2.0
language:
- en
---
# Uploaded finetuned model
- **Developed by:** ludast-original
- **License:** apache-2.0
- **Finetuned from model :** ludast-original/QA_physics_adapted_llama_3.2_3b
This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
|
felixZzz/wmixnoBoolean-orz-ours-d100-len5120-0427T17_47_21-step_08320 | felixZzz | 2025-05-04T22:35:42Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"qwen2",
"text-generation",
"conversational",
"arxiv:1910.09700",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T22:30:58Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent14_E6 | fffanx | 2025-05-04T22:35:41Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:35:13Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent14_E6
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent14_E6
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent14_E6", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent10_E6 | fffanx | 2025-05-04T22:33:33Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:33:05Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent10_E6
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent10_E6
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent10_E6", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent9_E6 | fffanx | 2025-05-04T22:33:02Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:32:31Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent9_E6
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent9_E6
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent9_E6", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent8_E6 | fffanx | 2025-05-04T22:32:28Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:31:59Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent8_E6
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent8_E6
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent8_E6", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
MinaMila/llama_instbase_3b_LoRa_ACSEmployment_2_ep5_22 | MinaMila | 2025-05-04T22:32:19Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"arxiv:1910.09700",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:32:14Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
gradientrouting-spar/toy_goodharting_gemma-2-2b-it_emotion_naive_outcome_0_25_0_1_seed_1_MC | gradientrouting-spar | 2025-05-04T22:31:47Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"arxiv:1910.09700",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:31:31Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent6_E6 | fffanx | 2025-05-04T22:31:22Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:30:53Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent6_E6
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent6_E6
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent6_E6", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
seohuibae/fc-sft-full-llama3.1-alf | seohuibae | 2025-05-04T22:29:29Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"llama",
"text-generation",
"arxiv:1910.09700",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T22:25:30Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
mveroe/Llama-3.2-1B-Instruct-safecoder-3.0-Code-safecoder_reg_full_safecoder_bd | mveroe | 2025-05-04T22:29:20Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"llama",
"text-generation",
"generated_from_trainer",
"conversational",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"license:llama3.2",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T15:51:42Z | ---
library_name: transformers
license: llama3.2
base_model: meta-llama/Llama-3.2-1B-Instruct
tags:
- generated_from_trainer
model-index:
- name: Llama-3.2-1B-Instruct-safecoder-3.0-Code-safecoder_reg_full_safecoder_bd
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-3.2-1B-Instruct-safecoder-3.0-Code-safecoder_reg_full_safecoder_bd
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the None 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: 8
- seed: 42
- optimizer: Use OptimizerNames.ADAFACTOR and the args are:
No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- training_steps: 2500
### Training results
### Framework versions
- Transformers 4.51.3
- Pytorch 2.7.0+cu126
- Datasets 3.5.1
- Tokenizers 0.21.1
|
capitaletech/cv-section-detector-best | capitaletech | 2025-05-04T22:28:07Z | 0 | 0 | peft | [
"peft",
"safetensors",
"arxiv:1910.09700",
"base_model:Qwen/Qwen2-VL-7B-Instruct",
"base_model:adapter:Qwen/Qwen2-VL-7B-Instruct",
"region:us"
] | null | 2025-05-04T22:28:06Z | ---
base_model: Qwen/Qwen2-VL-7B-Instruct
library_name: peft
---
# 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]
- **Funded by [optional]:** [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 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]
### Framework versions
- PEFT 0.15.2 |
mradermacher/Claria1.7b-GGUF | mradermacher | 2025-05-04T22:27:30Z | 0 | 0 | transformers | [
"transformers",
"gguf",
"text-generation-inference",
"unsloth",
"qwen3",
"trl",
"sft",
"en",
"ro",
"base_model:drwlf/Claria1.7b",
"base_model:quantized:drwlf/Claria1.7b",
"license:apache-2.0",
"endpoints_compatible",
"region:us",
"conversational"
] | null | 2025-05-04T19:46:56Z | ---
base_model: drwlf/Claria1.7b
language:
- en
- ro
library_name: transformers
license: apache-2.0
quantized_by: mradermacher
tags:
- text-generation-inference
- transformers
- unsloth
- qwen3
- trl
- sft
---
## About
<!-- ### quantize_version: 2 -->
<!-- ### output_tensor_quantised: 1 -->
<!-- ### convert_type: hf -->
<!-- ### vocab_type: -->
<!-- ### tags: -->
static quants of https://huggingface.co/drwlf/Claria1.7b
<!-- provided-files -->
weighted/imatrix quants are available at https://huggingface.co/mradermacher/Claria1.7b-i1-GGUF
## Usage
If you are unsure how to use GGUF files, refer to one of [TheBloke's
READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
more details, including on how to concatenate multi-part files.
## Provided Quants
(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
| Link | Type | Size/GB | Notes |
|:-----|:-----|--------:|:------|
| [GGUF](https://huggingface.co/mradermacher/Claria1.7b-GGUF/resolve/main/Claria1.7b.Q2_K.gguf) | Q2_K | 0.9 | |
| [GGUF](https://huggingface.co/mradermacher/Claria1.7b-GGUF/resolve/main/Claria1.7b.Q3_K_S.gguf) | Q3_K_S | 1.0 | |
| [GGUF](https://huggingface.co/mradermacher/Claria1.7b-GGUF/resolve/main/Claria1.7b.Q3_K_M.gguf) | Q3_K_M | 1.0 | lower quality |
| [GGUF](https://huggingface.co/mradermacher/Claria1.7b-GGUF/resolve/main/Claria1.7b.Q3_K_L.gguf) | Q3_K_L | 1.1 | |
| [GGUF](https://huggingface.co/mradermacher/Claria1.7b-GGUF/resolve/main/Claria1.7b.IQ4_XS.gguf) | IQ4_XS | 1.1 | |
| [GGUF](https://huggingface.co/mradermacher/Claria1.7b-GGUF/resolve/main/Claria1.7b.Q4_K_S.gguf) | Q4_K_S | 1.2 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/Claria1.7b-GGUF/resolve/main/Claria1.7b.Q4_K_M.gguf) | Q4_K_M | 1.2 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/Claria1.7b-GGUF/resolve/main/Claria1.7b.Q5_K_S.gguf) | Q5_K_S | 1.3 | |
| [GGUF](https://huggingface.co/mradermacher/Claria1.7b-GGUF/resolve/main/Claria1.7b.Q5_K_M.gguf) | Q5_K_M | 1.4 | |
| [GGUF](https://huggingface.co/mradermacher/Claria1.7b-GGUF/resolve/main/Claria1.7b.Q6_K.gguf) | Q6_K | 1.5 | very good quality |
| [GGUF](https://huggingface.co/mradermacher/Claria1.7b-GGUF/resolve/main/Claria1.7b.Q8_0.gguf) | Q8_0 | 1.9 | fast, best quality |
| [GGUF](https://huggingface.co/mradermacher/Claria1.7b-GGUF/resolve/main/Claria1.7b.f16.gguf) | f16 | 3.5 | 16 bpw, overkill |
Here is a handy graph by ikawrakow comparing some lower-quality quant
types (lower is better):
And here are Artefact2's thoughts on the matter:
https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
## FAQ / Model Request
See https://huggingface.co/mradermacher/model_requests for some answers to
questions you might have and/or if you want some other model quantized.
## Thanks
I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
me use its servers and providing upgrades to my workstation to enable
this work in my free time.
<!-- end -->
|
felixZzz/wmixnoBoolean-orz-ours-d100-len5120-0427T17_47_21-step_07296 | felixZzz | 2025-05-04T22:24:51Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"qwen2",
"text-generation",
"conversational",
"arxiv:1910.09700",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T22:14:14Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
capitaletech/cv-section-detector | capitaletech | 2025-05-04T22:23:41Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"generated_from_trainer",
"trl",
"sft",
"base_model:Qwen/Qwen2-VL-7B-Instruct",
"base_model:finetune:Qwen/Qwen2-VL-7B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T15:16:09Z | ---
base_model: Qwen/Qwen2-VL-7B-Instruct
library_name: transformers
model_name: cv-section-detector
tags:
- generated_from_trainer
- trl
- sft
licence: license
---
# Model Card for cv-section-detector
This model is a fine-tuned version of [Qwen/Qwen2-VL-7B-Instruct](https://huggingface.co/Qwen/Qwen2-VL-7B-Instruct).
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="capitaletech/cv-section-detector", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with SFT.
### Framework versions
- TRL: 0.17.0
- Transformers: 4.51.3
- Pytorch: 2.4.1+cu121
- Datasets: 3.5.1
- Tokenizers: 0.21.1
## Citations
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
Sweaterdog/Andy-4-micro-base-LoRA-testbed | Sweaterdog | 2025-05-04T22:23:36Z | 0 | 0 | null | [
"safetensors",
"region:us"
] | null | 2025-05-04T17:43:15Z |
# ⛏️ Andy-4-micro-base-LoRA-testbed ⛏️
Andy-4-micro is a 1.7b version of Andy-4, it shows off how incredible the Andy-4 dataset can make models, even in tiny packages.
This model is not going to work based off of preliminary testing, and has been cancelled due to such.
## Model Details
This is a Small version of Andy-4, which is 1.7b parameters, and will release in all facets around two weeks from the full Andy-4 release.
This model used an optimal training environment for smaller models, as well as the same datasets as the full Andy-4 model.
**Andy-4:** 7b or 8b parameters
**Andy-4-micro:** 1.7b parameters.
# Additional detail
In preliminary testing of Andy-4-base and Andy-4-micro-base, Andy-4-micro-base showed to be only slightly worse than Andy-4, and sometimes on-par with non-building examples
Andy-4-micro shows off its incredible ability to play when allowed to reason, this is because the model is able to walk through step by step of what it should do, instead of relying on what it "remembers" for what it should do
In the preliminary testing, both non-reasoning and reasoning modes were used. Again, this was using Andy-4-micro-base and Andy-4-base, not the final versions.
Using this LoRA to continue to train a model falls under the ruling of the [Andy License](LICENSE), as in the final model must include `Andy` in it somewhere, and credit to be posted to I, Sweaterdog, in the model repo.
This is a base model, meaning it is not meant to play Minecraft with you, and is only for training updates to the model, as Mindcraft is not finished, we need to have a rolling update system
```
## License
This model is licensed under the Andy 1.0 License.
Credit: https://huggingface.co/Sweaterdog
Acknowledgment: This work uses data and models created by @Sweaterdog.
``` |
felixZzz/wlen6_61k-orz-ours-d1-len3000-0428T03_05_50-step_00784 | felixZzz | 2025-05-04T22:20:30Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"qwen2",
"text-generation",
"conversational",
"arxiv:1910.09700",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T22:12:35Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent19_E5 | fffanx | 2025-05-04T22:19:43Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:19:14Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent19_E5
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent19_E5
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent19_E5", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent17_E5 | fffanx | 2025-05-04T22:18:39Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:18:11Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent17_E5
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent17_E5
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent17_E5", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
team-9/gpt2-finetune-c4-baseline-100k | team-9 | 2025-05-04T22:15:49Z | 0 | 0 | transformers | [
"transformers",
"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 | 2025-05-04T21:38:27Z | ---
library_name: transformers
license: mit
base_model: gpt2
tags:
- generated_from_trainer
model-index:
- name: gpt2-finetune-c4-baseline-100k
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. -->
# gpt2-finetune-c4-baseline-100k
This model is a fine-tuned version of [gpt2](https://huggingface.co/gpt2) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.2758
## 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
- distributed_type: multi-GPU
- num_devices: 8
- total_train_batch_size: 64
- total_eval_batch_size: 64
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 3.4219 | 1.0 | 1547 | 3.3136 |
| 3.405 | 2.0 | 3094 | 3.2870 |
| 3.367 | 3.0 | 4641 | 3.2758 |
### Framework versions
- Transformers 4.51.3
- Pytorch 2.6.0+cu124
- Datasets 3.5.0
- Tokenizers 0.21.1
|
fffanx/Llama-3.2-1B-Instruct-GRPO-agent7_E5 | fffanx | 2025-05-04T22:13:21Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:12:54Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent7_E5
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent7_E5
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent7_E5", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent6_E5 | fffanx | 2025-05-04T22:12:50Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:12:21Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent6_E5
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent6_E5
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent6_E5", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
felixZzz/wlen6_61k-orz-ours-d1-len3000-0428T03_05_50-step_00720 | felixZzz | 2025-05-04T22:11:46Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"qwen2",
"text-generation",
"conversational",
"arxiv:1910.09700",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T22:03:54Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent4_E5 | fffanx | 2025-05-04T22:11:41Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:11:10Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent4_E5
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent4_E5
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent4_E5", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
Kieran2828/Qwen3-8B-Q4_K_M-GGUF | Kieran2828 | 2025-05-04T22:11:15Z | 0 | 0 | transformers | [
"transformers",
"gguf",
"llama-cpp",
"gguf-my-repo",
"text-generation",
"base_model:Qwen/Qwen3-8B",
"base_model:quantized:Qwen/Qwen3-8B",
"license:apache-2.0",
"endpoints_compatible",
"region:us",
"conversational"
] | text-generation | 2025-05-04T22:10:52Z | ---
base_model: Qwen/Qwen3-8B
library_name: transformers
license: apache-2.0
license_link: https://huggingface.co/Qwen/Qwen3-8B/blob/main/LICENSE
pipeline_tag: text-generation
tags:
- llama-cpp
- gguf-my-repo
---
# Kieran2828/Qwen3-8B-Q4_K_M-GGUF
This model was converted to GGUF format from [`Qwen/Qwen3-8B`](https://huggingface.co/Qwen/Qwen3-8B) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space.
Refer to the [original model card](https://huggingface.co/Qwen/Qwen3-8B) for more details on the model.
## Use with llama.cpp
Install llama.cpp through brew (works on Mac and Linux)
```bash
brew install llama.cpp
```
Invoke the llama.cpp server or the CLI.
### CLI:
```bash
llama-cli --hf-repo Kieran2828/Qwen3-8B-Q4_K_M-GGUF --hf-file qwen3-8b-q4_k_m.gguf -p "The meaning to life and the universe is"
```
### Server:
```bash
llama-server --hf-repo Kieran2828/Qwen3-8B-Q4_K_M-GGUF --hf-file qwen3-8b-q4_k_m.gguf -c 2048
```
Note: You can also use this checkpoint directly through the [usage steps](https://github.com/ggerganov/llama.cpp?tab=readme-ov-file#usage) listed in the Llama.cpp repo as well.
Step 1: Clone llama.cpp from GitHub.
```
git clone https://github.com/ggerganov/llama.cpp
```
Step 2: Move into the llama.cpp folder and build it with `LLAMA_CURL=1` flag along with other hardware-specific flags (for ex: LLAMA_CUDA=1 for Nvidia GPUs on Linux).
```
cd llama.cpp && LLAMA_CURL=1 make
```
Step 3: Run inference through the main binary.
```
./llama-cli --hf-repo Kieran2828/Qwen3-8B-Q4_K_M-GGUF --hf-file qwen3-8b-q4_k_m.gguf -p "The meaning to life and the universe is"
```
or
```
./llama-server --hf-repo Kieran2828/Qwen3-8B-Q4_K_M-GGUF --hf-file qwen3-8b-q4_k_m.gguf -c 2048
```
|
fffanx/Llama-3.2-1B-Instruct-GRPO-agent3_E5 | fffanx | 2025-05-04T22:11:07Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:10:38Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent3_E5
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent3_E5
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent3_E5", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent2_E5 | fffanx | 2025-05-04T22:10:35Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:10:06Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent2_E5
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent2_E5
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent2_E5", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
fffanx/Llama-3.2-1B-Instruct-GRPO-agent1_E5 | fffanx | 2025-05-04T22:10:03Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T22:09:34Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent1_E5
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent1_E5
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent1_E5", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
gradientrouting-spar/qwen_ft_May3_m1_p1 | gradientrouting-spar | 2025-05-04T22:10:02Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"arxiv:1910.09700",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T19:54:42Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
linsanityuk/task-8-dailysub_batchA_202505042205 | linsanityuk | 2025-05-04T22:08:14Z | 0 | 0 | peft | [
"peft",
"safetensors",
"arxiv:1910.09700",
"base_model:microsoft/Phi-3.5-mini-instruct",
"base_model:adapter:microsoft/Phi-3.5-mini-instruct",
"region:us"
] | null | 2025-05-04T22:07:34Z | ---
base_model: microsoft/Phi-3.5-mini-instruct
library_name: peft
---
# 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]
- **Funded by [optional]:** [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 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]
### Framework versions
- PEFT 0.13.2 |
albertus-sussex/veriscrape-fixed-simcse-auto-reference_5_to_verify_5-fold-1 | albertus-sussex | 2025-05-04T22:04:30Z | 9 | 0 | transformers | [
"transformers",
"safetensors",
"roberta",
"feature-extraction",
"arxiv:1910.09700",
"endpoints_compatible",
"region:us"
] | feature-extraction | 2025-04-01T13:00:41Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
felixZzz/wlen6_61k-orz-ours-d1-len3000-0428T03_05_50-step_00656 | felixZzz | 2025-05-04T22:03:05Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"qwen2",
"text-generation",
"conversational",
"arxiv:1910.09700",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T21:50:45Z | ---
library_name: transformers
tags: []
---
# 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. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [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 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] |
Delssz/Repozular | Delssz | 2025-05-04T22:03:01Z | 0 | 0 | diffusers | [
"diffusers",
"text-to-image",
"lora",
"template:diffusion-lora",
"base_model:black-forest-labs/FLUX.1-dev",
"base_model:adapter:black-forest-labs/FLUX.1-dev",
"license:bsl-1.0",
"region:us"
] | text-to-image | 2025-05-04T22:02:23Z | ---
tags:
- text-to-image
- lora
- diffusers
- template:diffusion-lora
widget:
- text: Nigga
parameters:
negative_prompt: Nigger
output:
url: images/IMG_1106.jpeg
base_model: black-forest-labs/FLUX.1-dev
instance_prompt: Nigga
license: bsl-1.0
---
# Koja
<Gallery />
## Model description
An amazing project that will make a huge impact on our communities
## Trigger words
You should use `Nigga` to trigger the image generation.
## Download model
[Download](/Delssz/Repozular/tree/main) them in the Files & versions tab.
|
Draq10/llama3.1-POMI-merged16bit-HF-epoch2 | Draq10 | 2025-05-04T22:01:25Z | 0 | 0 | null | [
"safetensors",
"gguf",
"llama",
"license:apache-2.0",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T21:52:17Z | ---
license: apache-2.0
---
|
Daemontatox/Qwen3-14B-Griffon | Daemontatox | 2025-05-04T22:00:06Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"qwen3",
"text-generation",
"text-generation-inference",
"unsloth",
"trl",
"reasoning",
"math",
"code-generation",
"conversational",
"en",
"dataset:open-thoughts/OpenThoughts2-1M",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T20:18:29Z | ---
base_model: unsloth/qwen3-14b-unsloth
tags:
- text-generation-inference
- transformers
- unsloth
- qwen3
- trl
- reasoning
- math
- code-generation
license: apache-2.0
language:
- en
datasets:
- open-thoughts/OpenThoughts2-1M
library_name: transformers
---
# Qwen3-14B-Griffon
**Developed by:** Daemontatox
**License:** Apache-2.0
**Finetuned from:** [unsloth/qwen3-14b-unsloth](https://huggingface.co/unsloth/qwen3-14b-unsloth)
## Model Overview
This is a fine-tuned version of the Qwen3-14B model using the high-quality **OpenThoughts2-1M** dataset. Fine-tuned with Unsloth’s TRL-compatible framework and LoRA for efficient performance, this model is optimized for **advanced reasoning tasks**, especially in **math**, **logic puzzles**, **code generation**, and **step-by-step problem solving**.
## Training Dataset
- **Dataset:** [OpenThoughts2-1M](https://huggingface.co/datasets/open-thoughts/OpenThoughts2-1M)
- **Source:** A synthetic dataset curated and expanded by the OpenThoughts team
- **Volume:** ~1.1M high-quality examples
- **Content Type:** Multi-turn reasoning, math proofs, algorithmic code generation, logical deduction, and structured conversations
- **Tools Used:** [Curator Viewer](https://curator.bespokelabs.ai/)
This dataset builds upon OpenThoughts-114k and integrates strong reasoning-centric data sources like OpenR1-Math and KodCode.
## Intended Use
This model is particularly suited for:
- Chain-of-thought and step-by-step reasoning
- Code generation with logical structure
- Educational tools for math and programming
- AI agents requiring multi-turn problem-solving
## Limitations
- English-only focus (does not generalize well to other languages)
- May hallucinate factual content despite reasoning depth
- Inherits possible biases from synthetic pretraining data
## Example Usage
```python
# Use a pipeline as a high-level helper
from transformers import pipeline
messages = [
{"role": "user", "content": "Who are you?"},
]
pipe = pipeline("text-generation", model="Daemontatox/Qwen3_14B_Griffon")
pipe(messages)
```
## Training Details
# Framework: TRL + LoRA with Unsloth acceleration
# Epochs/Steps: Custom fine-tuning on ~1M samples
# Hardware: Single-node A100 80GB / similar high-VRAM setup
# Objective: Enhance multi-domain reasoning under compute-efficient constraints
--- |
mradermacher/TruDecide.ai-7B-GGUF | mradermacher | 2025-05-04T21:58:52Z | 0 | 0 | transformers | [
"transformers",
"gguf",
"en",
"base_model:shuyouxingjie/TruDecide.ai-7B",
"base_model:quantized:shuyouxingjie/TruDecide.ai-7B",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T20:55:14Z | ---
base_model: shuyouxingjie/TruDecide.ai-7B
language:
- en
library_name: transformers
quantized_by: mradermacher
---
## About
<!-- ### quantize_version: 2 -->
<!-- ### output_tensor_quantised: 1 -->
<!-- ### convert_type: hf -->
<!-- ### vocab_type: -->
<!-- ### tags: -->
static quants of https://huggingface.co/shuyouxingjie/TruDecide.ai-7B
<!-- provided-files -->
weighted/imatrix quants seem not to be available (by me) at this time. If they do not show up a week or so after the static ones, I have probably not planned for them. Feel free to request them by opening a Community Discussion.
## Usage
If you are unsure how to use GGUF files, refer to one of [TheBloke's
READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
more details, including on how to concatenate multi-part files.
## Provided Quants
(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
| Link | Type | Size/GB | Notes |
|:-----|:-----|--------:|:------|
| [GGUF](https://huggingface.co/mradermacher/TruDecide.ai-7B-GGUF/resolve/main/TruDecide.ai-7B.Q2_K.gguf) | Q2_K | 2.9 | |
| [GGUF](https://huggingface.co/mradermacher/TruDecide.ai-7B-GGUF/resolve/main/TruDecide.ai-7B.Q3_K_S.gguf) | Q3_K_S | 3.4 | |
| [GGUF](https://huggingface.co/mradermacher/TruDecide.ai-7B-GGUF/resolve/main/TruDecide.ai-7B.Q3_K_M.gguf) | Q3_K_M | 3.7 | lower quality |
| [GGUF](https://huggingface.co/mradermacher/TruDecide.ai-7B-GGUF/resolve/main/TruDecide.ai-7B.Q3_K_L.gguf) | Q3_K_L | 4.0 | |
| [GGUF](https://huggingface.co/mradermacher/TruDecide.ai-7B-GGUF/resolve/main/TruDecide.ai-7B.IQ4_XS.gguf) | IQ4_XS | 4.1 | |
| [GGUF](https://huggingface.co/mradermacher/TruDecide.ai-7B-GGUF/resolve/main/TruDecide.ai-7B.Q4_K_S.gguf) | Q4_K_S | 4.3 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/TruDecide.ai-7B-GGUF/resolve/main/TruDecide.ai-7B.Q4_K_M.gguf) | Q4_K_M | 4.5 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/TruDecide.ai-7B-GGUF/resolve/main/TruDecide.ai-7B.Q5_K_S.gguf) | Q5_K_S | 5.0 | |
| [GGUF](https://huggingface.co/mradermacher/TruDecide.ai-7B-GGUF/resolve/main/TruDecide.ai-7B.Q5_K_M.gguf) | Q5_K_M | 5.2 | |
| [GGUF](https://huggingface.co/mradermacher/TruDecide.ai-7B-GGUF/resolve/main/TruDecide.ai-7B.Q6_K.gguf) | Q6_K | 5.9 | very good quality |
| [GGUF](https://huggingface.co/mradermacher/TruDecide.ai-7B-GGUF/resolve/main/TruDecide.ai-7B.Q8_0.gguf) | Q8_0 | 7.6 | fast, best quality |
| [GGUF](https://huggingface.co/mradermacher/TruDecide.ai-7B-GGUF/resolve/main/TruDecide.ai-7B.f16.gguf) | f16 | 14.2 | 16 bpw, overkill |
Here is a handy graph by ikawrakow comparing some lower-quality quant
types (lower is better):
And here are Artefact2's thoughts on the matter:
https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
## FAQ / Model Request
See https://huggingface.co/mradermacher/model_requests for some answers to
questions you might have and/or if you want some other model quantized.
## Thanks
I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
me use its servers and providing upgrades to my workstation to enable
this work in my free time.
<!-- end -->
|
mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF | mradermacher | 2025-05-04T21:58:51Z | 0 | 0 | transformers | [
"transformers",
"gguf",
"en",
"base_model:HarethahMo/qwen2.5-1.5B-extended-refusal-2-0-abliterated",
"base_model:quantized:HarethahMo/qwen2.5-1.5B-extended-refusal-2-0-abliterated",
"endpoints_compatible",
"region:us",
"imatrix",
"conversational"
] | null | 2025-05-04T20:00:02Z | ---
base_model: HarethahMo/qwen2.5-1.5B-extended-refusal-2-0-abliterated
language:
- en
library_name: transformers
quantized_by: mradermacher
tags: []
---
## About
<!-- ### quantize_version: 2 -->
<!-- ### output_tensor_quantised: 1 -->
<!-- ### convert_type: hf -->
<!-- ### vocab_type: -->
<!-- ### tags: nicoboss -->
weighted/imatrix quants of https://huggingface.co/HarethahMo/qwen2.5-1.5B-extended-refusal-2-0-abliterated
<!-- provided-files -->
static quants are available at https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-GGUF
## Usage
If you are unsure how to use GGUF files, refer to one of [TheBloke's
READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
more details, including on how to concatenate multi-part files.
## Provided Quants
(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
| Link | Type | Size/GB | Notes |
|:-----|:-----|--------:|:------|
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-IQ1_S.gguf) | i1-IQ1_S | 0.5 | for the desperate |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-IQ1_M.gguf) | i1-IQ1_M | 0.6 | mostly desperate |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 0.6 | |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-IQ2_XS.gguf) | i1-IQ2_XS | 0.7 | |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-IQ2_S.gguf) | i1-IQ2_S | 0.7 | |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-IQ2_M.gguf) | i1-IQ2_M | 0.7 | |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-Q2_K_S.gguf) | i1-Q2_K_S | 0.7 | very low quality |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 0.8 | lower quality |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-Q2_K.gguf) | i1-Q2_K | 0.8 | IQ3_XXS probably better |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-IQ3_XS.gguf) | i1-IQ3_XS | 0.8 | |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-Q3_K_S.gguf) | i1-Q3_K_S | 0.9 | IQ3_XS probably better |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-IQ3_S.gguf) | i1-IQ3_S | 0.9 | beats Q3_K* |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-IQ3_M.gguf) | i1-IQ3_M | 0.9 | |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-Q3_K_M.gguf) | i1-Q3_K_M | 0.9 | IQ3_S probably better |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-Q3_K_L.gguf) | i1-Q3_K_L | 1.0 | IQ3_M probably better |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-IQ4_XS.gguf) | i1-IQ4_XS | 1.0 | |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-IQ4_NL.gguf) | i1-IQ4_NL | 1.0 | prefer IQ4_XS |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-Q4_0.gguf) | i1-Q4_0 | 1.0 | fast, low quality |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-Q4_K_S.gguf) | i1-Q4_K_S | 1.0 | optimal size/speed/quality |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-Q4_K_M.gguf) | i1-Q4_K_M | 1.1 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-Q4_1.gguf) | i1-Q4_1 | 1.1 | |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-Q5_K_S.gguf) | i1-Q5_K_S | 1.2 | |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-Q5_K_M.gguf) | i1-Q5_K_M | 1.2 | |
| [GGUF](https://huggingface.co/mradermacher/qwen2.5-1.5B-extended-refusal-2-0-abliterated-i1-GGUF/resolve/main/qwen2.5-1.5B-extended-refusal-2-0-abliterated.i1-Q6_K.gguf) | i1-Q6_K | 1.4 | practically like static Q6_K |
Here is a handy graph by ikawrakow comparing some lower-quality quant
types (lower is better):
And here are Artefact2's thoughts on the matter:
https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
## FAQ / Model Request
See https://huggingface.co/mradermacher/model_requests for some answers to
questions you might have and/or if you want some other model quantized.
## Thanks
I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
me use its servers and providing upgrades to my workstation to enable
this work in my free time. Additional thanks to [@nicoboss](https://huggingface.co/nicoboss) for giving me access to his private supercomputer, enabling me to provide many more imatrix quants, at much higher quality, than I would otherwise be able to.
<!-- end -->
|
mradermacher/post-14b-GGUF | mradermacher | 2025-05-04T21:58:51Z | 0 | 0 | transformers | [
"transformers",
"gguf",
"text-generation-inference",
"unsloth",
"qwen3",
"trl",
"sft",
"en",
"base_model:isaiahbjork/post-14b",
"base_model:quantized:isaiahbjork/post-14b",
"license:apache-2.0",
"endpoints_compatible",
"region:us",
"conversational"
] | null | 2025-05-04T20:30:31Z | ---
base_model: isaiahbjork/post-14b
language:
- en
library_name: transformers
license: apache-2.0
quantized_by: mradermacher
tags:
- text-generation-inference
- transformers
- unsloth
- qwen3
- trl
- sft
---
## About
<!-- ### quantize_version: 2 -->
<!-- ### output_tensor_quantised: 1 -->
<!-- ### convert_type: hf -->
<!-- ### vocab_type: -->
<!-- ### tags: -->
static quants of https://huggingface.co/isaiahbjork/post-14b
<!-- provided-files -->
weighted/imatrix quants seem not to be available (by me) at this time. If they do not show up a week or so after the static ones, I have probably not planned for them. Feel free to request them by opening a Community Discussion.
## Usage
If you are unsure how to use GGUF files, refer to one of [TheBloke's
READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
more details, including on how to concatenate multi-part files.
## Provided Quants
(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
| Link | Type | Size/GB | Notes |
|:-----|:-----|--------:|:------|
| [GGUF](https://huggingface.co/mradermacher/post-14b-GGUF/resolve/main/post-14b.Q2_K.gguf) | Q2_K | 5.9 | |
| [GGUF](https://huggingface.co/mradermacher/post-14b-GGUF/resolve/main/post-14b.Q3_K_S.gguf) | Q3_K_S | 6.8 | |
| [GGUF](https://huggingface.co/mradermacher/post-14b-GGUF/resolve/main/post-14b.Q3_K_M.gguf) | Q3_K_M | 7.4 | lower quality |
| [GGUF](https://huggingface.co/mradermacher/post-14b-GGUF/resolve/main/post-14b.Q3_K_L.gguf) | Q3_K_L | 8.0 | |
| [GGUF](https://huggingface.co/mradermacher/post-14b-GGUF/resolve/main/post-14b.IQ4_XS.gguf) | IQ4_XS | 8.3 | |
| [GGUF](https://huggingface.co/mradermacher/post-14b-GGUF/resolve/main/post-14b.Q4_K_S.gguf) | Q4_K_S | 8.7 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/post-14b-GGUF/resolve/main/post-14b.Q4_K_M.gguf) | Q4_K_M | 9.1 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/post-14b-GGUF/resolve/main/post-14b.Q5_K_S.gguf) | Q5_K_S | 10.4 | |
| [GGUF](https://huggingface.co/mradermacher/post-14b-GGUF/resolve/main/post-14b.Q5_K_M.gguf) | Q5_K_M | 10.6 | |
| [GGUF](https://huggingface.co/mradermacher/post-14b-GGUF/resolve/main/post-14b.Q6_K.gguf) | Q6_K | 12.2 | very good quality |
| [GGUF](https://huggingface.co/mradermacher/post-14b-GGUF/resolve/main/post-14b.Q8_0.gguf) | Q8_0 | 15.8 | fast, best quality |
Here is a handy graph by ikawrakow comparing some lower-quality quant
types (lower is better):
And here are Artefact2's thoughts on the matter:
https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
## FAQ / Model Request
See https://huggingface.co/mradermacher/model_requests for some answers to
questions you might have and/or if you want some other model quantized.
## Thanks
I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
me use its servers and providing upgrades to my workstation to enable
this work in my free time.
<!-- end -->
|
Draq10/deepseekr1-POMI-merged16bit-HF-epoch2.1 | Draq10 | 2025-05-04T21:58:01Z | 0 | 0 | transformers | [
"transformers",
"safetensors",
"llama",
"text-generation",
"text-generation-inference",
"unsloth",
"trl",
"sft",
"conversational",
"en",
"base_model:unsloth/DeepSeek-R1-Distill-Llama-8B",
"base_model:finetune:unsloth/DeepSeek-R1-Distill-Llama-8B",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] | text-generation | 2025-05-04T21:53:49Z | ---
base_model: unsloth/DeepSeek-R1-Distill-Llama-8B
tags:
- text-generation-inference
- transformers
- unsloth
- llama
- trl
- sft
license: apache-2.0
language:
- en
---
# Uploaded model
- **Developed by:** Draq10
- **License:** apache-2.0
- **Finetuned from model :** unsloth/DeepSeek-R1-Distill-Llama-8B
This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
|
fffanx/Llama-3.2-1B-Instruct-GRPO-agent17_E4 | fffanx | 2025-05-04T21:57:20Z | 0 | 0 | transformers | [
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"dataset:grouped_dataset",
"arxiv:2402.03300",
"base_model:meta-llama/Llama-3.2-1B-Instruct",
"base_model:finetune:meta-llama/Llama-3.2-1B-Instruct",
"endpoints_compatible",
"region:us"
] | null | 2025-05-04T21:56:52Z | ---
base_model: meta-llama/Llama-3.2-1B-Instruct
datasets: grouped_dataset
library_name: transformers
model_name: Llama-3.2-1B-Instruct-GRPO-agent17_E4
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for Llama-3.2-1B-Instruct-GRPO-agent17_E4
This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the [grouped_dataset](https://huggingface.co/datasets/grouped_dataset) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="fffanx/Llama-3.2-1B-Instruct-GRPO-agent17_E4", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.6.0
- Datasets: 3.3.2
- Tokenizers: 0.21.0
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
``` |
Subsets and Splits