modelId
stringlengths 4
112
| sha
stringlengths 40
40
| lastModified
stringlengths 24
24
| tags
list | pipeline_tag
stringclasses 29
values | private
bool 1
class | author
stringlengths 2
38
⌀ | config
null | id
stringlengths 4
112
| downloads
float64 0
36.8M
⌀ | likes
float64 0
712
⌀ | library_name
stringclasses 17
values | __index_level_0__
int64 0
38.5k
| readme
stringlengths 0
186k
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
eunjin/koMHBERT-kcbert-based-v2 | 7cb5c95d8dd7618a1af57c26ad6e271ed5edec79 | 2021-06-05T17:44:39.000Z | [
"pytorch",
"bert",
"feature-extraction",
"transformers"
]
| feature-extraction | false | eunjin | null | eunjin/koMHBERT-kcbert-based-v2 | 6 | null | transformers | 15,200 | korean Mental Health BERT -v2
huggingface에 공개된 kcbert-base BERT를 정신건강의학신문을 크롤링한 dataset으로 MLM fine-tuining한 Bert Model입니다. 정신건강 발화 관련 데이터를 모을 수 없는 상황에서 이를 대체할만한 데이터로 제시합니다. 향후 정신건강 관련 감정 및 상태 classification 및 그에 따른 chatbot 구현에 사용할 수 있습니다.
정신건강의학신문: http://www.psychiatricnews.net |
fabiogr/opus-mt-en-de-finetuned-en-to-de-wd01-fp16false | f7e10f7c71ec521ec53d5b20f67f3b80fdb567f5 | 2021-12-02T18:48:45.000Z | [
"pytorch",
"marian",
"text2text-generation",
"dataset:wmt16",
"transformers",
"generated_from_trainer",
"license:apache-2.0",
"model-index",
"autotrain_compatible"
]
| text2text-generation | false | fabiogr | null | fabiogr/opus-mt-en-de-finetuned-en-to-de-wd01-fp16false | 6 | null | transformers | 15,201 | ---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- wmt16
model-index:
- name: opus-mt-en-de-finetuned-en-to-de-wd01-fp16false
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. -->
# opus-mt-en-de-finetuned-en-to-de-wd01-fp16false
This model is a fine-tuned version of [Helsinki-NLP/opus-mt-en-de](https://huggingface.co/Helsinki-NLP/opus-mt-en-de) on the wmt16 dataset.
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 1
- mixed_precision_training: Native AMP
### Framework versions
- Transformers 4.12.5
- Pytorch 1.10.0
- Datasets 1.16.1
- Tokenizers 0.10.3
|
fdominik98/bert-base-hu-cased-ner | e7de236bc278cba7ed60af353e6e2bfc2283627e | 2021-10-17T10:48:06.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | fdominik98 | null | fdominik98/bert-base-hu-cased-ner | 6 | null | transformers | 15,202 | This model is the fine-tuned model of "akdeniz27/bert-base-hungarian-cased-ner" using WikiANN-hu dataset. |
flax-community/clip-rsicd-v4 | 8f532a6e011510cc162d670287e9ba1dbe8775c8 | 2021-07-17T08:56:41.000Z | [
"pytorch",
"jax",
"clip",
"feature-extraction",
"transformers"
]
| feature-extraction | false | flax-community | null | flax-community/clip-rsicd-v4 | 6 | null | transformers | 15,203 | Entry not found |
flax-community/roberta-base-scandinavian | 9e586e36832835054dc99ca454ea103fb98694bf | 2021-09-23T13:54:15.000Z | [
"pytorch",
"jax",
"tensorboard",
"roberta",
"fill-mask",
"da",
"transformers",
"scandinavian",
"license:cc-by-4.0",
"autotrain_compatible"
]
| fill-mask | false | flax-community | null | flax-community/roberta-base-scandinavian | 6 | null | transformers | 15,204 | ---
language: da
license: cc-by-4.0
tags:
- scandinavian
- roberta
pipeline_tag: fill-mask
widget:
- text: På biblioteket kan du låne en <mask>.
---
# Scandinavian Roberta Base - MC4
## Description
This is a sample reference model for Flax/Jax training using only on the MC4. It is trained for roughly three day on a TPU v3-8. Training procedure...
---
## Description
My description |
flax-sentence-embeddings/mpnet_stackexchange_v1 | 3d257d700dbf4f475298792141e57418deb7b054 | 2021-07-25T21:17:52.000Z | [
"pytorch",
"mpnet",
"fill-mask",
"arxiv:2104.08727",
"sentence-transformers",
"feature-extraction",
"sentence-similarity"
]
| sentence-similarity | false | flax-sentence-embeddings | null | flax-sentence-embeddings/mpnet_stackexchange_v1 | 6 | null | sentence-transformers | 15,205 | ---
pipeline_tag: sentence-similarity
tags:
- sentence-transformers
- feature-extraction
- sentence-similarity
---
# mpnet_stackexchange_v1
## Model Description
SentenceTransformers is a set of models and frameworks that enable training and generating sentence embeddings from given data. The generated sentence embeddings can be utilized for Clustering, Semantic Search and other tasks. We used a pretrained [mpnet-base](https://huggingface.co/microsoft/mpnet-base) model and trained it using Siamese Network setup and contrastive learning objective. Question and answer pairs from StackExchange was used as training data to make the model robust to Question / Answer embedding similarity.
We developped this model during the
[Community week using JAX/Flax for NLP & CV](https://discuss.huggingface.co/t/open-to-the-community-community-week-using-jax-flax-for-nlp-cv/7104),
organized by Hugging Face. We developped this model as part of the project:
[Train the Best Sentence Embedding Model Ever with 1B Training Pairs](https://discuss.huggingface.co/t/train-the-best-sentence-embedding-model-ever-with-1b-training-pairs/7354). We benefited from efficient hardware infrastructure to run the project: 7 TPUs v3-8, as well
as assistance from Google’s Flax, JAX, and Cloud team members about efficient deep learning frameworks.
## Intended uses
Our model is intended to be used as a sentence encoder for a search engine. Given an input sentence, it ouptuts a vector which captures
the sentence semantic information. The sentence vector may be used for semantic-search, clustering or sentence similarity tasks.
## How to use
Here is how to use this model to get the features of a given text using [SentenceTransformers](https://github.com/UKPLab/sentence-transformers) library:
```python
from sentence_transformers import SentenceTransformer
model = SentenceTransformer('flax-sentence-embeddings/mpnet_stackexchange_v1')
text = "Replace me by any question / answer you'd like."
text_embbedding = model.encode(text)
# array([-0.01559514, 0.04046123, 0.1317083 , 0.00085931, 0.04585106,
# -0.05607086, 0.0138078 , 0.03569756, 0.01420381, 0.04266302 ...],
# dtype=float32)
```
# Training procedure
## Pre-training
We use the pretrained [`Mpnet-base`](https://huggingface.co/microsoft/mpnet-base). Please refer to the model
card for more detailed information about the pre-training procedure.
## Fine-tuning
We fine-tune the model using a contrastive objective. Formally, we compute the cosine similarity from each possible sentence pairs from the batch.
We then apply the cross entropy loss by comparing with true pairs.
### Hyper parameters
We trained on model on a TPU v3-8. We train the model during 80k steps using a batch size of 1024 (128 per TPU core).
We use a learning rate warm up of 500. The sequence length was limited to 128 tokens. We used the AdamW optimizer with
a 2e-5 learning rate. The full training script is accessible in this current repository.
### Training data
We used the concatenation from multiple Stackexchange Question-Answer datasets to fine-tune our model.
We sampled each StackExchange given a weighted probability of following equation.
```
int((stackexchange_length[path] / total_stackexchange_length) * total_weight)
```
MSMARCO, NQ & other question-answer datasets were also used. Sampling ratio for StackExchange vs remaining : 2 vs 1.
| Dataset | Paper | Number of training tuples |
|:--------------------------------------------------------:|:----------------------------------------:|:--------------------------:|
| [Stack Exchange QA - Title & Answer](https://huggingface.co/datasets/flax-sentence-embeddings/stackexchange_title_best_voted_answer_jsonl) | - | 4,750,619 |
| [MS MARCO](https://microsoft.github.io/msmarco/) | [paper](https://doi.org/10.1145/3404835.3462804) | 9,144,553 |
| [GOOAQ: Open Question Answering with Diverse Answer Types](https://github.com/allenai/gooaq) | [paper](https://arxiv.org/pdf/2104.08727.pdf) | 3,012,496 |
| [Yahoo Answers](https://www.kaggle.com/soumikrakshit/yahoo-answers-dataset) Question/Answer | [paper](https://proceedings.neurips.cc/paper/2015/hash/250cf8b51c773f3f8dc8b4be867a9a02-Abstract.html) | 681,164 |
| SearchQA | - | 582,261 |
| [Natural Questions (NQ)](https://ai.google.com/research/NaturalQuestions) | [paper](https://transacl.org/ojs/index.php/tacl/article/view/1455) | 100,231 | |
flax-sentence-embeddings/multi-qa_v1-mpnet-cls_dot | 8bc6a484ff0b4cfd440ef7301d662c2d78b47365 | 2021-07-26T01:34:59.000Z | [
"pytorch",
"mpnet",
"fill-mask",
"arxiv:2102.07033",
"arxiv:2104.08727",
"sentence-transformers",
"feature-extraction",
"sentence-similarity"
]
| sentence-similarity | false | flax-sentence-embeddings | null | flax-sentence-embeddings/multi-qa_v1-mpnet-cls_dot | 6 | null | sentence-transformers | 15,206 | ---
pipeline_tag: sentence-similarity
tags:
- sentence-transformers
- feature-extraction
- sentence-similarity
---
# multi-qa_v1-mpnet-cls_dot
## Model Description
SentenceTransformers is a set of models and frameworks that enable training and generating sentence embeddings from given data. The generated sentence embeddings can be utilized for Clustering, Semantic Search and other tasks. We used a pretrained [microsoft/mpnet-base](https://huggingface.co/microsoft/mpnet-base) model and trained it using Siamese Network setup and contrastive learning objective. Question and answer pairs from StackExchange was used as training data to make the model robust to Question / Answer embedding similarity. For this model, cls output was used instead of mean pooling as sentence embeddings. Dot product was used to calculate similarity for learning objective.
We developed this model during the
[Community week using JAX/Flax for NLP & CV](https://discuss.huggingface.co/t/open-to-the-community-community-week-using-jax-flax-for-nlp-cv/7104),
organized by Hugging Face. We developed this model as part of the project:
[Train the Best Sentence Embedding Model Ever with 1B Training Pairs](https://discuss.huggingface.co/t/train-the-best-sentence-embedding-model-ever-with-1b-training-pairs/7354). We benefited from efficient hardware infrastructure to run the project: 7 TPUs v3-8, as well
as assistance from Google’s Flax, JAX, and Cloud team members about efficient deep learning frameworks.
## Intended uses
Our model is intended to be used as a sentence encoder for a search engine. Given an input sentence, it outputs a vector which captures
the sentence semantic information. The sentence vector may be used for semantic-search, clustering or sentence similarity tasks.
## How to use
Here is how to use this model to get the features of a given text using [SentenceTransformers](https://github.com/UKPLab/sentence-transformers) library:
```python
from sentence_transformers import SentenceTransformer
model = SentenceTransformer('flax-sentence-embeddings/multi-qa_v1-mpnet-cls_dot')
text = "Replace me by any question / answer you'd like."
text_embbedding = model.encode(text)
# array([-0.01559514, 0.04046123, 0.1317083 , 0.00085931, 0.04585106,
# -0.05607086, 0.0138078 , 0.03569756, 0.01420381, 0.04266302 ...],
# dtype=float32)
```
# Training procedure
## Pre-training
We use the pretrained [microsoft/mpnet-base](https://huggingface.co/microsoft/mpnet-base). Please refer to the model
card for more detailed information about the pre-training procedure.
## Fine-tuning
We fine-tune the model using a contrastive objective. Formally, we compute the cosine similarity from each possible sentence pairs from the batch.
We then apply the cross entropy loss by comparing with true pairs.
### Hyper parameters
We trained on model on a TPU v3-8. We train the model during 80k steps using a batch size of 1024 (128 per TPU core).
We use a learning rate warm up of 500. The sequence length was limited to 128 tokens. We used the AdamW optimizer with
a 2e-5 learning rate. The full training script is accessible in this current repository.
### Training data
We used the concatenation from multiple Stackexchange Question-Answer datasets to fine-tune our model. MSMARCO, NQ & other question-answer datasets were also used.
| Dataset | Paper | Number of training tuples |
|:--------------------------------------------------------:|:----------------------------------------:|:--------------------------:|
| [Stack Exchange QA - Title & Answer](https://huggingface.co/datasets/flax-sentence-embeddings/stackexchange_title_best_voted_answer_jsonl) | - | 4,750,619 |
| [Stack Exchange](https://huggingface.co/datasets/flax-sentence-embeddings/stackexchange_title_body_jsonl) | - | 364,001 |
| [TriviaqQA](https://huggingface.co/datasets/trivia_qa) | - | 73,346 |
| [SQuAD2.0](https://rajpurkar.github.io/SQuAD-explorer/) | [paper](https://aclanthology.org/P18-2124.pdf) | 87,599 |
| [Quora Question Pairs](https://quoradata.quora.com/First-Quora-Dataset-Release-Question-Pairs) | - | 103,663 |
| [Eli5](https://huggingface.co/datasets/eli5) | [paper](https://doi.org/10.18653/v1/p19-1346) | 325,475 |
| [PAQ](https://github.com/facebookresearch/PAQ) | [paper](https://arxiv.org/abs/2102.07033) | 64,371,441 |
| [WikiAnswers](https://github.com/afader/oqa#wikianswers-corpus) | [paper](https://doi.org/10.1145/2623330.2623677) | 77,427,422 |
| [MS MARCO](https://microsoft.github.io/msmarco/) | [paper](https://doi.org/10.1145/3404835.3462804) | 9,144,553 |
| [GOOAQ: Open Question Answering with Diverse Answer Types](https://github.com/allenai/gooaq) | [paper](https://arxiv.org/pdf/2104.08727.pdf) | 3,012,496 |
| [Yahoo Answers](https://www.kaggle.com/soumikrakshit/yahoo-answers-dataset) Question/Answer | [paper](https://proceedings.neurips.cc/paper/2015/hash/250cf8b51c773f3f8dc8b4be867a9a02-Abstract.html) | 681,164 |
| SearchQA | - | 582,261 |
| [Natural Questions (NQ)](https://ai.google.com/research/NaturalQuestions) | [paper](https://transacl.org/ojs/index.php/tacl/article/view/1455) | 100,231 |
|
formermagic/pyt5-base | daf12c354236867b695aa09f645cadf8cac89066 | 2021-09-19T12:41:20.000Z | [
"pytorch",
"jax",
"tensorboard",
"t5",
"text2text-generation",
"arxiv:1910.10683",
"transformers",
"autotrain_compatible"
]
| text2text-generation | false | formermagic | null | formermagic/pyt5-base | 6 | 1 | transformers | 15,207 | # Python T5 base model
Pre-trained model on CodeSearchNet Python dataset using a span-masking objective. The training objective and model were introduced in [this paper](https://arxiv.org/pdf/1910.10683.pdf) and first released in [this repository](https://github.com/google-research/text-to-text-transfer-transformer). PyT5 model used [git-t5](https://github.com/formermagic/git-t5) framework built on top of JAX/Flax to pre-train the model on a TPU v3-8 node.
# How to use
You can use this model to denoise span-masked sequences.
First, install the [git-t5](https://github.com/formermagic/git-t5) pip package:
```shell
> pip install git-t5
```
Next, download the model and tokenizer:
```python
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer,
model = AutoModelForSeq2SeqLM.from_pretrained("formermagic/pyt5-base")
tokenizer = AutoTokenizer.from_pretrained("formermagic/pyt5-base")
```
Finally, encode your input and generate the output sequence:
```python
from git_t5.utils import encode_input
text = """
def alias(self, annotationtype, set, fallback=False):
if inspect.isclass(annotationtype): annotationtype = annotationtype.ANNOTATIONTYPE
if annotationtype in self.set_alias and set in self.set_alias[annotationtype]:
return self.set_alias[annotationtype][set]
elif fallback:
return set
else:
raise KeyError("No alias for set " + set)
"""
batch, max_length = encode_input(tokenizer, text, seed=22)
outputs = model.generate(batch["input_ids"], max_length=max_length, num_beams=1)
print(tokenizer.batch_decode(outputs[..., 1:]))
print(tokenizer.batch_decode(batch["labels"]))
```
You should see the following output:
```shell
['<extra_id_0>, fallback=<extra_id_1> inspect<extra_id_2>.set_alias<extra_id_3> return self.set<extra_id_4>) def fallback']
['<extra_id_0>, fallback=<extra_id_1> inspect<extra_id_2>.set_alias<extra_id_3> return self.set<extra_id_4>) </s></s>']
```
As you can see, the predicted result is very close to the target sequence. |
gaetangate/bart-large_genrl_lcquad1 | 003df2e3c9b3edef5ba70f3a661b1191f34caba7 | 2022-04-05T15:09:52.000Z | [
"pytorch",
"bart",
"text2text-generation",
"arxiv:2108.07337",
"transformers",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | gaetangate | null | gaetangate/bart-large_genrl_lcquad1 | 6 | null | transformers | 15,208 | ---
license: apache-2.0
---
This model is used in the paper **Generative Relation Linking for Question Answering over Knowledge Bases**. [ArXiv](https://arxiv.org/abs/2108.07337), [GitHub](https://github.com/IBM/kbqa-relation-linking)
## Citation
```bibtex
@inproceedings{rossiello-genrl-2021,
title={Generative relation linking for question answering over knowledge bases},
author={Rossiello, Gaetano and Mihindukulasooriya, Nandana and Abdelaziz, Ibrahim and Bornea, Mihaela and Gliozzo, Alfio and Naseem, Tahira and Kapanipathi, Pavan},
booktitle={International Semantic Web Conference},
pages={321--337},
year={2021},
organization={Springer},
url = "https://link.springer.com/chapter/10.1007/978-3-030-88361-4_19",
doi = "10.1007/978-3-030-88361-4_19"
}
``` |
gagan3012/distilbert-base-uncased-finetuned-ner | 45b118724a925f65d445b612dbb113b6ff7e1bca | 2021-09-14T00:36:53.000Z | [
"pytorch",
"tensorboard",
"distilbert",
"token-classification",
"dataset:conll2003",
"transformers",
"generated_from_trainer",
"license:apache-2.0",
"model-index",
"autotrain_compatible"
]
| token-classification | false | gagan3012 | null | gagan3012/distilbert-base-uncased-finetuned-ner | 6 | null | transformers | 15,209 | ---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- conll2003
metrics:
- precision
- recall
- f1
- accuracy
model-index:
- name: distilbert-base-uncased-finetuned-ner
results:
- task:
name: Token Classification
type: token-classification
dataset:
name: conll2003
type: conll2003
args: conll2003
metrics:
- name: Precision
type: precision
value: 0.9274238227146815
- name: Recall
type: recall
value: 0.9363463474661595
- name: F1
type: f1
value: 0.9318637274549098
- name: Accuracy
type: accuracy
value: 0.9839865283492462
---
<!-- 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-ner
This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the conll2003 dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0614
- Precision: 0.9274
- Recall: 0.9363
- F1: 0.9319
- Accuracy: 0.9840
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
### Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|:--------:|
| 0.2403 | 1.0 | 878 | 0.0701 | 0.9101 | 0.9202 | 0.9151 | 0.9805 |
| 0.0508 | 2.0 | 1756 | 0.0600 | 0.9220 | 0.9350 | 0.9285 | 0.9833 |
| 0.0301 | 3.0 | 2634 | 0.0614 | 0.9274 | 0.9363 | 0.9319 | 0.9840 |
### Framework versions
- Transformers 4.10.2
- Pytorch 1.9.0+cu102
- Datasets 1.12.0
- Tokenizers 0.10.3
|
gayanin/bart-mlm-pubmed-medterm | e8ca43d7015df6dd349abc3216d2ed07a6fd49cb | 2021-12-02T20:51:43.000Z | [
"pytorch",
"tensorboard",
"bart",
"text2text-generation",
"transformers",
"generated_from_trainer",
"license:apache-2.0",
"model-index",
"autotrain_compatible"
]
| text2text-generation | false | gayanin | null | gayanin/bart-mlm-pubmed-medterm | 6 | null | transformers | 15,210 | ---
license: apache-2.0
tags:
- generated_from_trainer
model-index:
- name: bart-mlm-pubmed-medterm
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. -->
# bart-mlm-pubmed-medterm
This model is a fine-tuned version of [facebook/bart-base](https://huggingface.co/facebook/bart-base) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0000
- Rouge2 Precision: 0.985
- Rouge2 Recall: 0.7208
- Rouge2 Fmeasure: 0.8088
## 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: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge2 Precision | Rouge2 Recall | Rouge2 Fmeasure |
|:-------------:|:-----:|:------:|:---------------:|:----------------:|:-------------:|:---------------:|
| 0.0018 | 1.0 | 13833 | 0.0003 | 0.985 | 0.7208 | 0.8088 |
| 0.0014 | 2.0 | 27666 | 0.0006 | 0.9848 | 0.7207 | 0.8086 |
| 0.0009 | 3.0 | 41499 | 0.0002 | 0.9848 | 0.7207 | 0.8086 |
| 0.0007 | 4.0 | 55332 | 0.0002 | 0.985 | 0.7208 | 0.8088 |
| 0.0006 | 5.0 | 69165 | 0.0001 | 0.9848 | 0.7207 | 0.8087 |
| 0.0001 | 6.0 | 82998 | 0.0002 | 0.9846 | 0.7206 | 0.8086 |
| 0.0009 | 7.0 | 96831 | 0.0001 | 0.9848 | 0.7208 | 0.8087 |
| 0.0 | 8.0 | 110664 | 0.0000 | 0.9848 | 0.7207 | 0.8087 |
| 0.0001 | 9.0 | 124497 | 0.0000 | 0.985 | 0.7208 | 0.8088 |
| 0.0 | 10.0 | 138330 | 0.0000 | 0.985 | 0.7208 | 0.8088 |
### Framework versions
- Transformers 4.12.5
- Pytorch 1.10.0+cu111
- Datasets 1.16.1
- Tokenizers 0.10.3
|
gchhablani/bert-large-cased-finetuned-mrpc | 325d21a839a2d896621047704272eb76861a6ff7 | 2021-09-22T07:00:38.000Z | [
"pytorch",
"tensorboard",
"bert",
"text-classification",
"en",
"dataset:glue",
"transformers",
"generated_from_trainer",
"license:apache-2.0",
"model-index"
]
| text-classification | false | gchhablani | null | gchhablani/bert-large-cased-finetuned-mrpc | 6 | null | transformers | 15,211 | ---
language:
- en
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- glue
metrics:
- accuracy
- f1
model-index:
- name: bert-large-cased-finetuned-mrpc
results:
- task:
name: Text Classification
type: text-classification
dataset:
name: GLUE MRPC
type: glue
args: mrpc
metrics:
- name: Accuracy
type: accuracy
value: 0.6838235294117647
- name: F1
type: f1
value: 0.8122270742358079
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# bert-large-cased-finetuned-mrpc
This model is a fine-tuned version of [bert-large-cased](https://huggingface.co/bert-large-cased) on the GLUE MRPC dataset.
It achieves the following results on the evaluation set:
- Loss: 0.6274
- Accuracy: 0.6838
- F1: 0.8122
- Combined Score: 0.7480
## 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: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5.0
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Combined Score |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:------:|:--------------:|
| 0.6441 | 1.0 | 917 | 0.6370 | 0.6838 | 0.8122 | 0.7480 |
| 0.6451 | 2.0 | 1834 | 0.6553 | 0.6838 | 0.8122 | 0.7480 |
| 0.6428 | 3.0 | 2751 | 0.6332 | 0.6838 | 0.8122 | 0.7480 |
| 0.6476 | 4.0 | 3668 | 0.6248 | 0.6838 | 0.8122 | 0.7480 |
| 0.6499 | 5.0 | 4585 | 0.6274 | 0.6838 | 0.8122 | 0.7480 |
### Framework versions
- Transformers 4.11.0.dev0
- Pytorch 1.9.0
- Datasets 1.12.1
- Tokenizers 0.10.3
|
gchhablani/bert-large-cased-finetuned-rte | 219622ed9d72fbb6a9b47b0c2d64fdf60880b13b | 2021-10-09T14:14:22.000Z | [
"pytorch",
"tensorboard",
"bert",
"text-classification",
"en",
"dataset:glue",
"transformers",
"generated_from_trainer",
"license:apache-2.0",
"model-index"
]
| text-classification | false | gchhablani | null | gchhablani/bert-large-cased-finetuned-rte | 6 | null | transformers | 15,212 | ---
language:
- en
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- glue
metrics:
- accuracy
model-index:
- name: bert-large-cased-finetuned-rte
results:
- task:
name: Text Classification
type: text-classification
dataset:
name: GLUE RTE
type: glue
args: rte
metrics:
- name: Accuracy
type: accuracy
value: 0.6642599277978339
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# bert-large-cased-finetuned-rte
This model is a fine-tuned version of [bert-large-cased](https://huggingface.co/bert-large-cased) on the GLUE RTE dataset.
It achieves the following results on the evaluation set:
- Loss: 1.5187
- Accuracy: 0.6643
## 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: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 0.6969 | 1.0 | 623 | 0.7039 | 0.5343 |
| 0.5903 | 2.0 | 1246 | 0.6461 | 0.7184 |
| 0.4557 | 3.0 | 1869 | 1.5187 | 0.6643 |
### Framework versions
- Transformers 4.11.0.dev0
- Pytorch 1.9.0
- Datasets 1.12.1
- Tokenizers 0.10.3
|
gchhablani/fnet-base-finetuned-mnli | b1c59294d4ae5d91c6503fc4ed6fab695ac0c9c1 | 2021-09-20T09:08:10.000Z | [
"pytorch",
"tensorboard",
"fnet",
"text-classification",
"en",
"dataset:glue",
"arxiv:2105.03824",
"transformers",
"generated_from_trainer",
"fnet-bert-base-comparison",
"license:apache-2.0",
"model-index"
]
| text-classification | false | gchhablani | null | gchhablani/fnet-base-finetuned-mnli | 6 | null | transformers | 15,213 | ---
language:
- en
license: apache-2.0
tags:
- generated_from_trainer
- fnet-bert-base-comparison
datasets:
- glue
metrics:
- accuracy
model-index:
- name: fnet-base-finetuned-mnli
results:
- task:
name: Text Classification
type: text-classification
dataset:
name: GLUE MNLI
type: glue
args: mnli
metrics:
- name: Accuracy
type: accuracy
value: 0.7674938974776241
---
<!-- 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. -->
# fnet-base-finetuned-mnli
This model is a fine-tuned version of [google/fnet-base](https://huggingface.co/google/fnet-base) on the GLUE MNLI dataset.
It achieves the following results on the evaluation set:
- Loss: 0.6443
- Accuracy: 0.7675
The model was fine-tuned to compare [google/fnet-base](https://huggingface.co/google/fnet-base) as introduced in [this paper](https://arxiv.org/abs/2105.03824) against [bert-base-cased](https://huggingface.co/bert-base-cased).
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
This model is trained using the [run_glue](https://github.com/huggingface/transformers/blob/master/examples/pytorch/text-classification/run_glue.py) script. The following command was used:
```bash
#!/usr/bin/bash
python ../run_glue.py \\n --model_name_or_path google/fnet-base \\n --task_name mnli \\n --do_train \\n --do_eval \\n --max_seq_length 512 \\n --per_device_train_batch_size 16 \\n --learning_rate 2e-5 \\n --num_train_epochs 3 \\n --output_dir fnet-base-finetuned-mnli \\n --push_to_hub \\n --hub_strategy all_checkpoints \\n --logging_strategy epoch \\n --save_strategy epoch \\n --evaluation_strategy epoch \\n```
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:-----:|:---------------:|:--------:|
| 0.7143 | 1.0 | 24544 | 0.6169 | 0.7504 |
| 0.5407 | 2.0 | 49088 | 0.6218 | 0.7627 |
| 0.4178 | 3.0 | 73632 | 0.6564 | 0.7658 |
### Framework versions
- Transformers 4.11.0.dev0
- Pytorch 1.9.0
- Datasets 1.12.1
- Tokenizers 0.10.3
|
gfdgdfgdg/arap_qa_bert_v2 | fbac61062b11d9902d0e36d73dc6656b44adfdf8 | 2021-08-08T12:24:57.000Z | [
"pytorch",
"bert",
"question-answering",
"ar",
"transformers",
"autotrain_compatible"
]
| question-answering | false | gfdgdfgdg | null | gfdgdfgdg/arap_qa_bert_v2 | 6 | 1 | transformers | 15,214 | ---
language:
- ar
widget:
- text: "أين يعيش محمد ؟"
context: "اسمي محمد وأنا أعيش في سوريا"
- text: "ما العدد الذري للهيدروجين ؟"
context: "الهيدروجين هو عنصر كيميائي عدده الذري 1 ، وهو غاز عديم الرائحة واللون وهو سريع الاشتعال"
- text: "ما خواص الهيدروجين ؟"
context: "الهيدروجين هو عنصر كيميائي عدده الذري 1 ، وهو غاز عديم الرائحة واللون وهو سريع الاشتعال"
---
|
ghadeermobasher/BC5CDR-Disease-balancedPubMedBERT | ff053e784db8bfb5144d2ec2a4b9420ae457437c | 2022-01-22T22:12:18.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BC5CDR-Disease-balancedPubMedBERT | 6 | null | transformers | 15,215 | Entry not found |
ghadeermobasher/BC5CDR-Disease_Modified_PubMedBERT | fd35177154dc2338abe47e57a9142722bb761fe9 | 2022-01-21T23:05:20.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BC5CDR-Disease_Modified_PubMedBERT | 6 | null | transformers | 15,216 | Entry not found |
ghadeermobasher/BC5CDR-Disease_Modified_scibert_scivocab_cased | d4fae8dd78c3317570af4f28f7573ee81db7298b | 2022-01-23T16:28:16.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BC5CDR-Disease_Modified_scibert_scivocab_cased | 6 | null | transformers | 15,217 | Entry not found |
ghadeermobasher/BC5CDR-Imbalanced-SapBERT-from-PubMedBERT-fulltext | a21c85e4e15fdddfe4003682d2038af9bd0eb6ac | 2022-01-24T18:44:20.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BC5CDR-Imbalanced-SapBERT-from-PubMedBERT-fulltext | 6 | null | transformers | 15,218 | Entry not found |
ghadeermobasher/BC5CDR-Imbalanced-biobert-base-cased-v1.2 | 53956bae18604f6780cf39ee942b7e33fc417872 | 2022-01-24T18:40:10.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BC5CDR-Imbalanced-biobert-base-cased-v1.2 | 6 | null | transformers | 15,219 | Entry not found |
ghadeermobasher/BC5CDR-Imbalanced-scibert_scivocab_cased | fd0c4a7214299c59f96f383093fd2324d49066f2 | 2022-01-23T16:41:27.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BC5CDR-Imbalanced-scibert_scivocab_cased | 6 | null | transformers | 15,220 | Entry not found |
ghadeermobasher/BioNLP13CG-Chem-Modified-Bioformers | 25df6163ddb2a0bb4fddf530756bf34095dfec62 | 2022-02-17T10:19:32.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BioNLP13CG-Chem-Modified-Bioformers | 6 | null | transformers | 15,221 | Entry not found |
ghadeermobasher/BioNLP13CG-Chem-Modified-Bioformers_2 | 663d19aec3827195d29c2e98426f096f5235846a | 2022-02-17T11:11:22.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BioNLP13CG-Chem-Modified-Bioformers_2 | 6 | null | transformers | 15,222 | Entry not found |
ghadeermobasher/BioNLP13CG-Chem-Modified-PubMedBert-abstract-3 | b6789542c8378a07850bf7bf435bf955f04166ca | 2022-02-16T18:45:51.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BioNLP13CG-Chem-Modified-PubMedBert-abstract-3 | 6 | null | transformers | 15,223 | Entry not found |
ghadeermobasher/BioNLP13CG-Chem-Modified-PubMedBert-full-3 | 04381748878de6d9bd034ba6767a1165bbc78c87 | 2022-02-16T16:06:21.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BioNLP13CG-Chem-Modified-PubMedBert-full-3 | 6 | null | transformers | 15,224 | Entry not found |
ghadeermobasher/BioNLP13CG-Chem-Modified-pubmedabstract_latest | 074f6b4860f09d8ef50bd1f7d2878880f5fc8e53 | 2022-02-17T13:26:56.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BioNLP13CG-Chem-Modified-pubmedabstract_latest | 6 | null | transformers | 15,225 | Entry not found |
ghadeermobasher/BioNLP13CG-Chem-Modified_PubMedBERT | 5ea5d9d22896fb06c77ad0c340b7b252bd128ca5 | 2022-01-22T00:08:11.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BioNLP13CG-Chem-Modified_PubMedBERT | 6 | null | transformers | 15,226 | Entry not found |
ghadeermobasher/BioNLP13CG-Chem-Modified_SciBERT | 4525b0c7f51bdc68c0588e503ce9d0a975281f80 | 2022-02-08T14:44:30.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BioNLP13CG-Chem-Modified_SciBERT | 6 | null | transformers | 15,227 | Entry not found |
ghadeermobasher/BioNLP13CG-Chem-Modified_biobert-large | 8b55dbc789f8909cd2330a0a204000697d3aa606 | 2022-02-10T00:17:30.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BioNLP13CG-Chem-Modified_biobert-large | 6 | null | transformers | 15,228 | Entry not found |
ghadeermobasher/BioNLP13CG-Chem_Imbalanced-biobert | 40d878b64ef1a829c078e06624ef46d2153998d7 | 2022-02-10T13:16:04.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BioNLP13CG-Chem_Imbalanced-biobert | 6 | null | transformers | 15,229 | Entry not found |
ghadeermobasher/BioNLP13CG-Chem_Imbalanced-scibert_scivocab_cased | 7059d71bd03557fab05f95349fea673d76b677a4 | 2022-01-23T15:11:10.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BioNLP13CG-Chem_Imbalanced-scibert_scivocab_cased | 6 | null | transformers | 15,230 | Entry not found |
ghadeermobasher/BioNLP13CG-Chem_ImbalancedPubMedBERT | 032c116d1091f63eb8888ead080bda7130632acd | 2022-01-22T00:08:28.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BioNLP13CG-Chem_ImbalancedPubMedBERT | 6 | null | transformers | 15,231 | Entry not found |
ghadeermobasher/BioNLP13CG-Modified-pubmedabstract_latest | 538e3b726ab0d6564ccd0a4cdc252b1d41e79dba | 2022-02-21T19:52:33.000Z | [
"pytorch",
"tensorboard",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BioNLP13CG-Modified-pubmedabstract_latest | 6 | null | transformers | 15,232 | Entry not found |
ghadeermobasher/BioNLP13CG-Modified-scibert-uncased_latest | ef1ae084425498319e666a69e7313772b7567d23 | 2022-02-21T20:21:35.000Z | [
"pytorch",
"tensorboard",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BioNLP13CG-Modified-scibert-uncased_latest | 6 | null | transformers | 15,233 | Entry not found |
ghadeermobasher/BioNLP13CG-Original-PubmedBert-abstract-latest | ea2f023f7609a8ee47d86a365723bd13c12b0fbe | 2022-02-21T20:03:49.000Z | [
"pytorch",
"tensorboard",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BioNLP13CG-Original-PubmedBert-abstract-latest | 6 | null | transformers | 15,234 | Entry not found |
ghadeermobasher/BioNLP13CG-Original-bluebert_pubmed_uncased_L-12_H-768_A-12_latest | b9650a7c9f9045dae57512de0b627d726c98dff5 | 2022-02-21T20:23:56.000Z | [
"pytorch",
"tensorboard",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/BioNLP13CG-Original-bluebert_pubmed_uncased_L-12_H-768_A-12_latest | 6 | null | transformers | 15,235 | Entry not found |
ghadeermobasher/CRAFT-Chem-Modified-Bioformers | 9c99a17ffd32a1696f6e9845bf578c1a9c17debf | 2022-02-17T12:29:34.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/CRAFT-Chem-Modified-Bioformers | 6 | null | transformers | 15,236 | Entry not found |
ghadeermobasher/CRAFT-Chem-Modified-biobert-v1.1 | 92aa89271fed6206aab4029cef367ffc5e6f9fb1 | 2022-02-22T04:22:54.000Z | [
"pytorch",
"tensorboard",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/CRAFT-Chem-Modified-biobert-v1.1 | 6 | null | transformers | 15,237 | Entry not found |
ghadeermobasher/CRAFT-Chem-Modified-bluebert_pubmed_uncased_L-12_H-768_A-12 | 13ab051be51cae722aea7503e34307e2b7796903 | 2022-02-22T04:10:52.000Z | [
"pytorch",
"tensorboard",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/CRAFT-Chem-Modified-bluebert_pubmed_uncased_L-12_H-768_A-12 | 6 | null | transformers | 15,238 | Entry not found |
ghadeermobasher/CRAFT-Chem-Modified_SciBERT | 4b6d5c6c35e2a32af8201cb43cfa413a631294b6 | 2022-02-08T16:57:23.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/CRAFT-Chem-Modified_SciBERT | 6 | null | transformers | 15,239 | Entry not found |
ghadeermobasher/CRAFT-Chem-Modified_biobert-large-cased | bac7272913bccc1e4b82c618b165bc5f6ad1db28 | 2022-02-10T02:08:12.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/CRAFT-Chem-Modified_biobert-large-cased | 6 | null | transformers | 15,240 | Entry not found |
ghadeermobasher/CRAFT-Chem_Imbalanced-SciBERT | 1667b39d642cb6c43eb0a2f7a6f93158bb3e2d19 | 2022-02-08T16:54:45.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/CRAFT-Chem_Imbalanced-SciBERT | 6 | null | transformers | 15,241 | Entry not found |
ghadeermobasher/CRAFT-Chem_ImbalancedPubMedBERT | a3c9e493c4612d354624ad6f9d05dddb417286c3 | 2022-01-22T02:30:40.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/CRAFT-Chem_ImbalancedPubMedBERT | 6 | null | transformers | 15,242 | Entry not found |
ghadeermobasher/CRAFT-Chem_Original-BiomedNLP-PubMedBERT-base-uncased-abstract | 32dd7258f20144de24ba8c334c12004795e95f2a | 2022-02-23T22:19:41.000Z | [
"pytorch",
"tensorboard",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/CRAFT-Chem_Original-BiomedNLP-PubMedBERT-base-uncased-abstract | 6 | null | transformers | 15,243 | Entry not found |
ghadeermobasher/CRAFT-Chem_Original-bluebert_pubmed_uncased_L-12_H-768_A-12 | a0a3c4728fe64e995ba63d6a586d180093d64e52 | 2022-02-23T22:34:09.000Z | [
"pytorch",
"tensorboard",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/CRAFT-Chem_Original-bluebert_pubmed_uncased_L-12_H-768_A-12 | 6 | null | transformers | 15,244 | Entry not found |
ghadeermobasher/CRAFT-Chem_Original-scibert_scivocab_uncased | 064002b9824a33a6b26c2b40b258f2a511e5d887 | 2022-02-23T22:42:50.000Z | [
"pytorch",
"tensorboard",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | ghadeermobasher | null | ghadeermobasher/CRAFT-Chem_Original-scibert_scivocab_uncased | 6 | null | transformers | 15,245 | Entry not found |
glasses/dummy | 8936db2b06d5e3c733de0da501423563852c0707 | 2021-04-21T18:24:15.000Z | [
"pytorch",
"arxiv:1512.03385",
"arxiv:1812.01187",
"transformers"
]
| null | false | glasses | null | glasses/dummy | 6 | null | transformers | 15,246 | # ResNet
Implementation of ResNet proposed in [Deep Residual Learning for Image
Recognition](https://arxiv.org/abs/1512.03385)
``` python
ResNet.resnet18()
ResNet.resnet26()
ResNet.resnet34()
ResNet.resnet50()
ResNet.resnet101()
ResNet.resnet152()
ResNet.resnet200()
Variants (d) proposed in `Bag of Tricks for Image Classification with Convolutional Neural Networks <https://arxiv.org/pdf/1812.01187.pdf`_
ResNet.resnet26d()
ResNet.resnet34d()
ResNet.resnet50d()
# You can construct your own one by chaning `stem` and `block`
resnet101d = ResNet.resnet101(stem=ResNetStemC, block=partial(ResNetBottleneckBlock, shortcut=ResNetShorcutD))
```
Examples:
``` python
# change activation
ResNet.resnet18(activation = nn.SELU)
# change number of classes (default is 1000 )
ResNet.resnet18(n_classes=100)
# pass a different block
ResNet.resnet18(block=SENetBasicBlock)
# change the steam
model = ResNet.resnet18(stem=ResNetStemC)
change shortcut
model = ResNet.resnet18(block=partial(ResNetBasicBlock, shortcut=ResNetShorcutD))
# store each feature
x = torch.rand((1, 3, 224, 224))
# get features
model = ResNet.resnet18()
# first call .features, this will activate the forward hooks and tells the model you'll like to get the features
model.encoder.features
model(torch.randn((1,3,224,224)))
# get the features from the encoder
features = model.encoder.features
print([x.shape for x in features])
#[torch.Size([1, 64, 112, 112]), torch.Size([1, 64, 56, 56]), torch.Size([1, 128, 28, 28]), torch.Size([1, 256, 14, 14])]
```
|
glasses/resnet152 | 2472cf9411a6fc59b80689b3f8df13614ffdd0ab | 2021-11-30T20:12:19.000Z | [
"pytorch",
"dataset:imagenet",
"arxiv:1512.03385",
"arxiv:1812.01187",
"transformers",
"image-classification",
"license:apache-2.0"
]
| image-classification | false | glasses | null | glasses/resnet152 | 6 | null | transformers | 15,247 | ---
license: apache-2.0
tags:
- image-classification
datasets:
- imagenet
---
# resnet152
Implementation of ResNet proposed in [Deep Residual Learning for Image
Recognition](https://arxiv.org/abs/1512.03385)
``` python
ResNet.resnet18()
ResNet.resnet26()
ResNet.resnet34()
ResNet.resnet50()
ResNet.resnet101()
ResNet.resnet152()
ResNet.resnet200()
Variants (d) proposed in `Bag of Tricks for Image Classification with Convolutional Neural Networks <https://arxiv.org/pdf/1812.01187.pdf`_
ResNet.resnet26d()
ResNet.resnet34d()
ResNet.resnet50d()
# You can construct your own one by chaning `stem` and `block`
resnet101d = ResNet.resnet101(stem=ResNetStemC, block=partial(ResNetBottleneckBlock, shortcut=ResNetShorcutD))
```
Examples:
``` python
# change activation
ResNet.resnet18(activation = nn.SELU)
# change number of classes (default is 1000 )
ResNet.resnet18(n_classes=100)
# pass a different block
ResNet.resnet18(block=SENetBasicBlock)
# change the steam
model = ResNet.resnet18(stem=ResNetStemC)
change shortcut
model = ResNet.resnet18(block=partial(ResNetBasicBlock, shortcut=ResNetShorcutD))
# store each feature
x = torch.rand((1, 3, 224, 224))
# get features
model = ResNet.resnet18()
# first call .features, this will activate the forward hooks and tells the model you'll like to get the features
model.encoder.features
model(torch.randn((1,3,224,224)))
# get the features from the encoder
features = model.encoder.features
print([x.shape for x in features])
#[torch.Size([1, 64, 112, 112]), torch.Size([1, 64, 56, 56]), torch.Size([1, 128, 28, 28]), torch.Size([1, 256, 14, 14])]
```
|
glasses/resnet34 | ee62b58fa882d4a4fd6d40b4c06fcb99c7599cf5 | 2021-11-30T20:08:12.000Z | [
"pytorch",
"dataset:imagenet",
"arxiv:1512.03385",
"arxiv:1812.01187",
"transformers",
"image-classification",
"license:apache-2.0"
]
| image-classification | false | glasses | null | glasses/resnet34 | 6 | null | transformers | 15,248 | ---
license: apache-2.0
tags:
- image-classification
datasets:
- imagenet
---
# resnet34
Implementation of ResNet proposed in [Deep Residual Learning for Image
Recognition](https://arxiv.org/abs/1512.03385)
``` python
ResNet.resnet18()
ResNet.resnet26()
ResNet.resnet34()
ResNet.resnet50()
ResNet.resnet101()
ResNet.resnet152()
ResNet.resnet200()
Variants (d) proposed in `Bag of Tricks for Image Classification with Convolutional Neural Networks <https://arxiv.org/pdf/1812.01187.pdf`_
ResNet.resnet26d()
ResNet.resnet34d()
ResNet.resnet50d()
# You can construct your own one by chaning `stem` and `block`
resnet101d = ResNet.resnet101(stem=ResNetStemC, block=partial(ResNetBottleneckBlock, shortcut=ResNetShorcutD))
```
Examples:
``` python
# change activation
ResNet.resnet18(activation = nn.SELU)
# change number of classes (default is 1000 )
ResNet.resnet18(n_classes=100)
# pass a different block
ResNet.resnet18(block=SENetBasicBlock)
# change the steam
model = ResNet.resnet18(stem=ResNetStemC)
change shortcut
model = ResNet.resnet18(block=partial(ResNetBasicBlock, shortcut=ResNetShorcutD))
# store each feature
x = torch.rand((1, 3, 224, 224))
# get features
model = ResNet.resnet18()
# first call .features, this will activate the forward hooks and tells the model you'll like to get the features
model.encoder.features
model(torch.randn((1,3,224,224)))
# get the features from the encoder
features = model.encoder.features
print([x.shape for x in features])
#[torch.Size([1, 64, 112, 112]), torch.Size([1, 64, 56, 56]), torch.Size([1, 128, 28, 28]), torch.Size([1, 256, 14, 14])]
```
|
glob-asr/wav2vec2-large-xls-r-300m-spanish-small | 10f1ef18a921f4fd3d5508d0575b1135d70b44e1 | 2022-01-31T20:58:46.000Z | [
"pytorch",
"wav2vec2",
"automatic-speech-recognition",
"dataset:common_voice",
"transformers",
"generated_from_trainer",
"model-index"
]
| automatic-speech-recognition | false | glob-asr | null | glob-asr/wav2vec2-large-xls-r-300m-spanish-small | 6 | null | transformers | 15,249 | ---
tags:
- generated_from_trainer
datasets:
- common_voice
model-index:
- name: wav2vec2-large-xls-r-300m-spanish-small
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. -->
# wav2vec2-large-xls-r-300m-spanish-small
This model is a fine-tuned version of [jhonparra18/wav2vec2-large-xls-r-300m-spanish-custom](https://huggingface.co/jhonparra18/wav2vec2-large-xls-r-300m-spanish-custom) on the common_voice dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3596
- Wer: 0.2105
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0003
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 30
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|:-------------:|:-----:|:-----:|:---------------:|:------:|
| 0.1971 | 0.79 | 400 | 0.2169 | 0.2077 |
| 0.2293 | 1.58 | 800 | 0.2507 | 0.2418 |
| 0.2065 | 2.37 | 1200 | 0.2703 | 0.2459 |
| 0.1842 | 3.16 | 1600 | 0.2716 | 0.2495 |
| 0.1634 | 3.95 | 2000 | 0.2695 | 0.2510 |
| 0.1443 | 4.74 | 2400 | 0.2754 | 0.2435 |
| 0.1345 | 5.53 | 2800 | 0.3119 | 0.2654 |
| 0.1267 | 6.32 | 3200 | 0.3154 | 0.2573 |
| 0.1237 | 7.11 | 3600 | 0.3251 | 0.2666 |
| 0.1118 | 7.91 | 4000 | 0.3139 | 0.2503 |
| 0.1051 | 8.7 | 4400 | 0.3286 | 0.2573 |
| 0.0964 | 9.49 | 4800 | 0.3348 | 0.2587 |
| 0.0946 | 10.28 | 5200 | 0.3357 | 0.2587 |
| 0.0897 | 11.07 | 5600 | 0.3408 | 0.2590 |
| 0.0812 | 11.86 | 6000 | 0.3380 | 0.2560 |
| 0.079 | 12.65 | 6400 | 0.3304 | 0.2415 |
| 0.0753 | 13.44 | 6800 | 0.3557 | 0.2540 |
| 0.0717 | 14.23 | 7200 | 0.3507 | 0.2519 |
| 0.0691 | 15.02 | 7600 | 0.3554 | 0.2587 |
| 0.0626 | 15.81 | 8000 | 0.3619 | 0.2520 |
| 0.0661 | 16.6 | 8400 | 0.3609 | 0.2564 |
| 0.0582 | 17.39 | 8800 | 0.3818 | 0.2520 |
| 0.0556 | 18.18 | 9200 | 0.3685 | 0.2410 |
| 0.0515 | 18.97 | 9600 | 0.3658 | 0.2367 |
| 0.0478 | 19.76 | 10000 | 0.3701 | 0.2413 |
| 0.0486 | 20.55 | 10400 | 0.3681 | 0.2371 |
| 0.0468 | 21.34 | 10800 | 0.3607 | 0.2370 |
| 0.0452 | 22.13 | 11200 | 0.3499 | 0.2286 |
| 0.0399 | 22.92 | 11600 | 0.3647 | 0.2282 |
| 0.0393 | 23.72 | 12000 | 0.3638 | 0.2255 |
| 0.0381 | 24.51 | 12400 | 0.3359 | 0.2202 |
| 0.0332 | 25.3 | 12800 | 0.3488 | 0.2177 |
| 0.033 | 26.09 | 13200 | 0.3628 | 0.2175 |
| 0.0311 | 26.88 | 13600 | 0.3695 | 0.2195 |
| 0.0294 | 27.67 | 14000 | 0.3624 | 0.2164 |
| 0.0281 | 28.46 | 14400 | 0.3688 | 0.2113 |
| 0.0274 | 29.25 | 14800 | 0.3596 | 0.2105 |
### Framework versions
- Transformers 4.16.0.dev0
- Pytorch 1.10.1+cu102
- Datasets 1.17.1.dev0
- Tokenizers 0.11.0
|
google/multiberts-seed_4-step_800k | 0942c8b32048ea0c26276f78cd61f4ebc3979736 | 2021-11-06T03:30:31.000Z | [
"pytorch",
"tf",
"bert",
"pretraining",
"en",
"arxiv:2106.16163",
"arxiv:1908.08962",
"transformers",
"multiberts",
"multiberts-seed_4",
"multiberts-seed_4-step_800k",
"license:apache-2.0"
]
| null | false | google | null | google/multiberts-seed_4-step_800k | 6 | null | transformers | 15,250 | ---
language: en
tags:
- multiberts
- multiberts-seed_4
- multiberts-seed_4-step_800k
license: apache-2.0
---
# MultiBERTs, Intermediate Checkpoint - Seed 4, Step 800k
MultiBERTs is a collection of checkpoints and a statistical library to support
robust research on BERT. We provide 25 BERT-base models trained with
similar hyper-parameters as
[the original BERT model](https://github.com/google-research/bert) but
with different random seeds, which causes variations in the initial weights and order of
training instances. The aim is to distinguish findings that apply to a specific
artifact (i.e., a particular instance of the model) from those that apply to the
more general procedure.
We also provide 140 intermediate checkpoints captured
during the course of pre-training (we saved 28 checkpoints for the first 5 runs).
The models were originally released through
[http://goo.gle/multiberts](http://goo.gle/multiberts). We describe them in our
paper
[The MultiBERTs: BERT Reproductions for Robustness Analysis](https://arxiv.org/abs/2106.16163).
This is model #4, captured at step 800k (max: 2000k, i.e., 2M steps).
## Model Description
This model was captured during a reproduction of
[BERT-base uncased](https://github.com/google-research/bert), for English: it
is a Transformers model pretrained on a large corpus of English data, using the
Masked Language Modelling (MLM) and the Next Sentence Prediction (NSP)
objectives.
The intended uses, limitations, training data and training procedure for the fully trained model are similar
to [BERT-base uncased](https://github.com/google-research/bert). Two major
differences with the original model:
* We pre-trained the MultiBERTs models for 2 million steps using sequence
length 512 (instead of 1 million steps using sequence length 128 then 512).
* We used an alternative version of Wikipedia and Books Corpus, initially
collected for [Turc et al., 2019](https://arxiv.org/abs/1908.08962).
This is a best-effort reproduction, and so it is probable that differences with
the original model have gone unnoticed. The performance of MultiBERTs on GLUE after full training is oftentimes comparable to that of original
BERT, but we found significant differences on the dev set of SQuAD (MultiBERTs outperforms original BERT).
See our [technical report](https://arxiv.org/abs/2106.16163) for more details.
### How to use
Using code from
[BERT-base uncased](https://huggingface.co/bert-base-uncased), here is an example based on
Tensorflow:
```
from transformers import BertTokenizer, TFBertModel
tokenizer = BertTokenizer.from_pretrained('google/multiberts-seed_4-step_800k')
model = TFBertModel.from_pretrained("google/multiberts-seed_4-step_800k")
text = "Replace me by any text you'd like."
encoded_input = tokenizer(text, return_tensors='tf')
output = model(encoded_input)
```
PyTorch version:
```
from transformers import BertTokenizer, BertModel
tokenizer = BertTokenizer.from_pretrained('google/multiberts-seed_4-step_800k')
model = BertModel.from_pretrained("google/multiberts-seed_4-step_800k")
text = "Replace me by any text you'd like."
encoded_input = tokenizer(text, return_tensors='pt')
output = model(**encoded_input)
```
## Citation info
```bibtex
@article{sellam2021multiberts,
title={The MultiBERTs: BERT Reproductions for Robustness Analysis},
author={Thibault Sellam and Steve Yadlowsky and Jason Wei and Naomi Saphra and Alexander D'Amour and Tal Linzen and Jasmijn Bastings and Iulia Turc and Jacob Eisenstein and Dipanjan Das and Ian Tenney and Ellie Pavlick},
journal={arXiv preprint arXiv:2106.16163},
year={2021}
}
```
|
google/t5-efficient-base-dm2000 | 28bab3b05c9a01c45f26df11086df686e565ad9b | 2022-02-15T10:52:18.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-base-dm2000 | 6 | null | transformers | 15,251 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-BASE-DM2000 (Deep-Narrow version)
T5-Efficient-BASE-DM2000 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-base-dm2000** - is of model type **Base** with the following variations:
- **dm** is **2000**
It has **594.44** million parameters and thus requires *ca.* **2377.75 MB** of memory in full precision (*fp32*)
or **1188.87 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-base-el2 | f30caa2840ef14788a3ae285c75ee13865e336dd | 2022-02-15T10:52:31.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-base-el2 | 6 | null | transformers | 15,252 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-BASE-EL2 (Deep-Narrow version)
T5-Efficient-BASE-EL2 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-base-el2** - is of model type **Base** with the following variations:
- **el** is **2**
It has **152.13** million parameters and thus requires *ca.* **608.51 MB** of memory in full precision (*fp32*)
or **304.26 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-base-ff1000 | 5389564e05fb0fa3afc1e49d6bfe0e9af706e9f9 | 2022-02-15T10:52:40.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-base-ff1000 | 6 | null | transformers | 15,253 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-BASE-FF1000 (Deep-Narrow version)
T5-Efficient-BASE-FF1000 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-base-ff1000** - is of model type **Base** with the following variations:
- **ff** is **1000**
It has **147.43** million parameters and thus requires *ca.* **589.74 MB** of memory in full precision (*fp32*)
or **294.87 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-base-kv256 | 363276b2c99b9ea74cc101e73b9f5a54956ac795 | 2022-02-15T10:53:02.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-base-kv256 | 6 | null | transformers | 15,254 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-BASE-KV256 (Deep-Narrow version)
T5-Efficient-BASE-KV256 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-base-kv256** - is of model type **Base** with the following variations:
- **kv** is **256**
It has **477.74** million parameters and thus requires *ca.* **1910.94 MB** of memory in full precision (*fp32*)
or **955.47 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-base-nl2 | 785be08b86d4e3606d0fff9e82608dee55806dc1 | 2022-02-15T10:53:24.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-base-nl2 | 6 | null | transformers | 15,255 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-BASE-NL2 (Deep-Narrow version)
T5-Efficient-BASE-NL2 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-base-nl2** - is of model type **Base** with the following variations:
- **nl** is **2**
It has **57.72** million parameters and thus requires *ca.* **230.88 MB** of memory in full precision (*fp32*)
or **115.44 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-large-dm256 | 7c999ae284ffe5c9ec66a6590e913ce697e98b04 | 2022-02-15T10:57:28.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-large-dm256 | 6 | null | transformers | 15,256 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-LARGE-DM256 (Deep-Narrow version)
T5-Efficient-LARGE-DM256 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-large-dm256** - is of model type **Large** with the following variations:
- **dm** is **256**
It has **184.47** million parameters and thus requires *ca.* **737.9 MB** of memory in full precision (*fp32*)
or **368.95 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-large-el4 | ebc95e372c6fcf81ec1394a95d891360d4003605 | 2022-02-15T10:53:49.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-large-el4 | 6 | null | transformers | 15,257 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-LARGE-EL4 (Deep-Narrow version)
T5-Efficient-LARGE-EL4 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-large-el4** - is of model type **Large** with the following variations:
- **el** is **4**
It has **486.01** million parameters and thus requires *ca.* **1944.03 MB** of memory in full precision (*fp32*)
or **972.01 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-large-el8 | a35efa6dff11af7b102d13bd71936b9ff277eff9 | 2022-02-15T10:55:10.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-large-el8 | 6 | null | transformers | 15,258 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-LARGE-EL8 (Deep-Narrow version)
T5-Efficient-LARGE-EL8 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-large-el8** - is of model type **Large** with the following variations:
- **el** is **8**
It has **536.35** million parameters and thus requires *ca.* **2145.4 MB** of memory in full precision (*fp32*)
or **1072.7 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-large-kv16 | 285056bf1d458f5464148b0f0538b2d16479f10b | 2022-02-15T10:55:17.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-large-kv16 | 6 | null | transformers | 15,259 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-LARGE-KV16 (Deep-Narrow version)
T5-Efficient-LARGE-KV16 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-large-kv16** - is of model type **Large** with the following variations:
- **kv** is **16**
It has **511.23** million parameters and thus requires *ca.* **2044.93 MB** of memory in full precision (*fp32*)
or **1022.46 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-large-kv256 | a60711b283963f82a2c6e39f8601d01436cb5a3c | 2022-02-15T10:53:56.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-large-kv256 | 6 | null | transformers | 15,260 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-LARGE-KV256 (Deep-Narrow version)
T5-Efficient-LARGE-KV256 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-large-kv256** - is of model type **Large** with the following variations:
- **kv** is **256**
It has **1643.69** million parameters and thus requires *ca.* **6574.78 MB** of memory in full precision (*fp32*)
or **3287.39 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-large-nh8 | cc316c28de9047a69a51576fb1e3e52f8eef725b | 2022-02-15T10:55:37.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-large-nh8 | 6 | null | transformers | 15,261 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-LARGE-NH8 (Deep-Narrow version)
T5-Efficient-LARGE-NH8 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-large-nh8** - is of model type **Large** with the following variations:
- **nh** is **8**
It has **586.73** million parameters and thus requires *ca.* **2346.92 MB** of memory in full precision (*fp32*)
or **1173.46 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-large-nl10 | d97fdcce0b5b9c1432ee5a3cf91de573d21477b7 | 2022-02-15T10:55:41.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-large-nl10 | 6 | null | transformers | 15,262 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-LARGE-NL10 (Deep-Narrow version)
T5-Efficient-LARGE-NL10 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-large-nl10** - is of model type **Large** with the following variations:
- **nl** is **10**
It has **326.58** million parameters and thus requires *ca.* **1306.31 MB** of memory in full precision (*fp32*)
or **653.16 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-large-nl32 | adb13f366c3a6633e735704198bb0e88736d0ec8 | 2022-02-15T10:55:58.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-large-nl32 | 6 | null | transformers | 15,263 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-LARGE-NL32 (Deep-Narrow version)
T5-Efficient-LARGE-NL32 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-large-nl32** - is of model type **Large** with the following variations:
- **nl** is **32**
It has **972.67** million parameters and thus requires *ca.* **3890.66 MB** of memory in full precision (*fp32*)
or **1945.33 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-mini-nl24 | 64fb77133cb219f7b05a41a2b3527a191e38e0d9 | 2022-02-15T10:56:10.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-mini-nl24 | 6 | null | transformers | 15,264 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-MINI-NL24 (Deep-Narrow version)
T5-Efficient-MINI-NL24 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-mini-nl24** - is of model type **Mini** with the following variations:
- **nl** is **24**
It has **125.69** million parameters and thus requires *ca.* **502.75 MB** of memory in full precision (*fp32*)
or **251.37 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-small-el8 | e80971b936a4b88669b74c060016619d6bb98924 | 2022-02-15T10:50:12.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-small-el8 | 6 | null | transformers | 15,265 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-SMALL-EL8 (Deep-Narrow version)
T5-Efficient-SMALL-EL8 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-small-el8** - is of model type **Small** with the following variations:
- **el** is **8**
It has **66.82** million parameters and thus requires *ca.* **267.26 MB** of memory in full precision (*fp32*)
or **133.63 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-small-ff12000 | e9e044727a3b3db622ba18cc6751d746cd2766da | 2022-02-15T10:50:18.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-small-ff12000 | 6 | null | transformers | 15,266 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-SMALL-FF12000 (Deep-Narrow version)
T5-Efficient-SMALL-FF12000 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-small-ff12000** - is of model type **Small** with the following variations:
- **ff** is **12000**
It has **186.35** million parameters and thus requires *ca.* **745.4 MB** of memory in full precision (*fp32*)
or **372.7 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-tiny-nl24 | da4a3214f80b49c9252d667403f8f87b8ba19166 | 2022-02-15T10:51:41.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-tiny-nl24 | 6 | null | transformers | 15,267 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-TINY-NL24 (Deep-Narrow version)
T5-Efficient-TINY-NL24 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-tiny-nl24** - is of model type **Tiny** with the following variations:
- **nl** is **24**
It has **52.35** million parameters and thus requires *ca.* **209.41 MB** of memory in full precision (*fp32*)
or **104.71 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
google/t5-efficient-xl | 214a06c83aa1f216bbc46279222ad2beb8725508 | 2022-02-15T10:57:18.000Z | [
"pytorch",
"tf",
"jax",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"arxiv:2109.10686",
"transformers",
"deep-narrow",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-efficient-xl | 6 | 1 | transformers | 15,268 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-XL (Deep-Narrow version)
T5-Efficient-XL is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5).
It is a *pretrained-only* checkpoint and was released with the
paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)**
by *Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler*.
In a nutshell, the paper indicates that a **Deep-Narrow** model architecture is favorable for **downstream** performance compared to other model architectures
of similar parameter count.
To quote the paper:
> We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased
> before considering any other forms of uniform scaling across other dimensions. This is largely due to
> how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a
> tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise,
> a tall base model might also generally more efficient compared to a large model. We generally find
> that, regardless of size, even if absolute performance might increase as we continue to stack layers,
> the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36
> layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e.,
> params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params,
> FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to
> consider.
To be more precise, *model depth* is defined as the number of transformer blocks that are stacked sequentially.
A sequence of word embeddings is therefore processed sequentially by each transformer block.
## Details model architecture
This model checkpoint - **t5-efficient-xl** - is of model type **Xl** with no variations.
It has **2851.66** million parameters and thus requires *ca.* **11406.62 MB** of memory in full precision (*fp32*)
or **5703.31 MB** of memory in half precision (*fp16* or *bf16*).
A summary of the *original* T5 model architectures can be seen here:
| Model | nl (el/dl) | ff | dm | kv | nh | #Params|
| ----| ---- | ---- | ---- | ---- | ---- | ----|
| Tiny | 4/4 | 1024 | 256 | 32 | 4 | 16M|
| Mini | 4/4 | 1536 | 384 | 32 | 8 | 31M|
| Small | 6/6 | 2048 | 512 | 32 | 8 | 60M|
| Base | 12/12 | 3072 | 768 | 64 | 12 | 220M|
| Large | 24/24 | 4096 | 1024 | 64 | 16 | 738M|
| Xl | 24/24 | 16384 | 1024 | 128 | 32 | 3B|
| XXl | 24/24 | 65536 | 1024 | 128 | 128 | 11B|
whereas the following abbreviations are used:
| Abbreviation | Definition |
| ----| ---- |
| nl | Number of transformer blocks (depth) |
| dm | Dimension of embedding vector (output vector of transformers block) |
| kv | Dimension of key/value projection matrix |
| nh | Number of attention heads |
| ff | Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) |
| el | Number of transformer blocks in the encoder (encoder depth) |
| dl | Number of transformer blocks in the decoder (decoder depth) |
| sh | Signifies that attention heads are shared |
| skv | Signifies that key-values projection matrices are tied |
If a model checkpoint has no specific, *el* or *dl* than both the number of encoder- and decoder layers correspond to *nl*.
## Pre-Training
The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using
the span-based masked language modeling (MLM) objective.
## Fine-Tuning
**Note**: This model is a **pretrained** checkpoint and has to be fine-tuned for practical usage.
The checkpoint was pretrained in English and is therefore only useful for English NLP tasks.
You can follow on of the following examples on how to fine-tune the model:
*PyTorch*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization)
- [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*Tensorflow*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
*JAX/Flax*:
- [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization)
- [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - *Note*: You will have to slightly adapt the training example here to make it work with an encoder-decoder model.
## Downstream Performance
TODO: Add table if available
## Computational Complexity
TODO: Add table if available
## More information
We strongly recommend the reader to go carefully through the original paper **[Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)** to get a more nuanced understanding of this model checkpoint.
As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the *sh* or *skv*
model architecture variations have *not* been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future. |
hanseokhyeon/bert-11street | ed8da665f06e1921cdd1bc905d56efe5691a4088 | 2021-05-19T17:57:02.000Z | [
"pytorch",
"jax",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | hanseokhyeon | null | hanseokhyeon/bert-11street | 6 | null | transformers | 15,269 | Entry not found |
harish/EN-AStitchTask1A-XLNet-FalseTrue-0-1-BEST | b9616c40d197f0035f4daae32a5ea4f075ccead4 | 2021-09-05T00:18:19.000Z | [
"pytorch",
"xlnet",
"text-classification",
"transformers"
]
| text-classification | false | harish | null | harish/EN-AStitchTask1A-XLNet-FalseTrue-0-1-BEST | 6 | null | transformers | 15,270 | Entry not found |
heruberuto/esperberto | 3827d7d8ef5247bd8195ce52cd130a9ca33e76d0 | 2021-12-17T07:06:26.000Z | [
"pytorch",
"tensorboard",
"roberta",
"fill-mask",
"transformers",
"autotrain_compatible"
]
| fill-mask | false | heruberuto | null | heruberuto/esperberto | 6 | 1 | transformers | 15,271 | Entry not found |
hf-internal-testing/tiny-random-dpr | 77a019e5edf69b72089afaaa9e317c50dde54bde | 2021-09-15T21:34:06.000Z | [
"pytorch",
"tf",
"dpr",
"transformers"
]
| null | false | hf-internal-testing | null | hf-internal-testing/tiny-random-dpr | 6 | null | transformers | 15,272 | Entry not found |
hf-internal-testing/tiny-random-transfo-xl | b2d78d7aabc4cc04adb27c9b7674a7793fce1ab5 | 2021-09-17T19:25:41.000Z | [
"pytorch",
"tf",
"transfo-xl",
"transformers"
]
| null | false | hf-internal-testing | null | hf-internal-testing/tiny-random-transfo-xl | 6 | null | transformers | 15,273 | Entry not found |
hfl/chinese-legal-electra-large-generator | 5fc37e150d670d69b7d2b22435fbdc898448a273 | 2021-10-30T23:55:57.000Z | [
"pytorch",
"tf",
"electra",
"zh",
"arxiv:2004.13922",
"transformers",
"license:apache-2.0"
]
| null | false | hfl | null | hfl/chinese-legal-electra-large-generator | 6 | 1 | transformers | 15,274 | ---
language:
- zh
license: "apache-2.0"
---
# This model is specifically designed for legal domain.
## Chinese ELECTRA
Google and Stanford University released a new pre-trained model called ELECTRA, which has a much compact model size and relatively competitive performance compared to BERT and its variants.
For further accelerating the research of the Chinese pre-trained model, the Joint Laboratory of HIT and iFLYTEK Research (HFL) has released the Chinese ELECTRA models based on the official code of ELECTRA.
ELECTRA-small could reach similar or even higher scores on several NLP tasks with only 1/10 parameters compared to BERT and its variants.
This project is based on the official code of ELECTRA: [https://github.com/google-research/electra](https://github.com/google-research/electra)
You may also interested in,
- Chinese BERT series: https://github.com/ymcui/Chinese-BERT-wwm
- Chinese ELECTRA: https://github.com/ymcui/Chinese-ELECTRA
- Chinese XLNet: https://github.com/ymcui/Chinese-XLNet
- Knowledge Distillation Toolkit - TextBrewer: https://github.com/airaria/TextBrewer
More resources by HFL: https://github.com/ymcui/HFL-Anthology
## Citation
If you find our resource or paper is useful, please consider including the following citation in your paper.
- https://arxiv.org/abs/2004.13922
```
@inproceedings{cui-etal-2020-revisiting,
title = "Revisiting Pre-Trained Models for {C}hinese Natural Language Processing",
author = "Cui, Yiming and
Che, Wanxiang and
Liu, Ting and
Qin, Bing and
Wang, Shijin and
Hu, Guoping",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: Findings",
month = nov,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.findings-emnlp.58",
pages = "657--668",
}
``` |
howey/bert-base-uncased-boolq | dc6722b1c5a170a68a184c419db3331626032d26 | 2021-07-20T05:47:16.000Z | [
"pytorch",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | howey | null | howey/bert-base-uncased-boolq | 6 | null | transformers | 15,275 | Entry not found |
howey/bert-base-uncased-cloth | 11dbcfbfeef6f7dc65ac0fecee57ebfb57dea33a | 2021-07-19T03:33:33.000Z | [
"pytorch",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | howey | null | howey/bert-base-uncased-cloth | 6 | null | transformers | 15,276 | Entry not found |
howey/bert-base-uncased-kaggle | 0f49bf9acbd97959545a20788af118463159fc5d | 2021-07-18T14:17:28.000Z | [
"pytorch",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | howey | null | howey/bert-base-uncased-kaggle | 6 | null | transformers | 15,277 | Entry not found |
howey/bert-base-uncased-mnli | 2a3c5c9027e5f2acf20c29a3b80b55de86cd9d52 | 2021-05-19T19:59:55.000Z | [
"pytorch",
"jax",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | howey | null | howey/bert-base-uncased-mnli | 6 | null | transformers | 15,278 | Entry not found |
howey/bert-base-uncased-stsb | 02ad9b5c55893ac3467e5b1263e9f7b434c3c06c | 2021-08-28T15:08:25.000Z | [
"pytorch",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | howey | null | howey/bert-base-uncased-stsb | 6 | null | transformers | 15,279 | Entry not found |
howey/bert-large-uncased-squad | 7650c7ad59263c6ea3b257dab3c5d5f088ba5f1f | 2021-06-21T06:54:23.000Z | [
"pytorch",
"bert",
"question-answering",
"transformers",
"autotrain_compatible"
]
| question-answering | false | howey | null | howey/bert-large-uncased-squad | 6 | null | transformers | 15,280 | Entry not found |
howey/roberta-large-mrpc | fc6dbea5a376abc07ddcabc2ba6179c6ef8f7876 | 2021-06-03T11:37:10.000Z | [
"pytorch",
"roberta",
"text-classification",
"transformers"
]
| text-classification | false | howey | null | howey/roberta-large-mrpc | 6 | null | transformers | 15,281 | Entry not found |
huaen/question_detection_user_utter | 49517ad188e01197a2f97bc3addcb190367c4b37 | 2021-10-30T08:09:03.000Z | [
"pytorch",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | huaen | null | huaen/question_detection_user_utter | 6 | null | transformers | 15,282 | Entry not found |
huggingartists/aimer | 0064dcb284b519b1e2003c7bc4e8f9f6f0d5a272 | 2021-09-11T13:43:46.000Z | [
"pytorch",
"jax",
"gpt2",
"text-generation",
"en",
"dataset:huggingartists/aimer",
"transformers",
"huggingartists",
"lyrics",
"lm-head",
"causal-lm"
]
| text-generation | false | huggingartists | null | huggingartists/aimer | 6 | null | transformers | 15,283 | ---
language: en
datasets:
- huggingartists/aimer
tags:
- huggingartists
- lyrics
- lm-head
- causal-lm
widget:
- text: "I am"
---
<div class="inline-flex flex-col" style="line-height: 1.5;">
<div class="flex">
<div
style="display:DISPLAY_1; margin-left: auto; margin-right: auto; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://images.genius.com/123a0b2ef09a25207b610c5bd7b21d0f.1000x1000x1.jpg')">
</div>
</div>
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">🤖 HuggingArtists Model 🤖</div>
<div style="text-align: center; font-size: 16px; font-weight: 800">Aimer</div>
<a href="https://genius.com/artists/aimer">
<div style="text-align: center; font-size: 14px;">@aimer</div>
</a>
</div>
I was made with [huggingartists](https://github.com/AlekseyKorshuk/huggingartists).
Create your own bot based on your favorite artist with [the demo](https://colab.research.google.com/github/AlekseyKorshuk/huggingartists/blob/master/huggingartists-demo.ipynb)!
## How does it work?
To understand how the model was developed, check the [W&B report](https://wandb.ai/huggingartists/huggingartists/reportlist).
## Training data
The model was trained on lyrics from Aimer.
Dataset is available [here](https://huggingface.co/datasets/huggingartists/aimer).
And can be used with:
```python
from datasets import load_dataset
dataset = load_dataset("huggingartists/aimer")
```
[Explore the data](https://wandb.ai/huggingartists/huggingartists/runs/1rtjxc8q/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
## Training procedure
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on Aimer's lyrics.
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/huggingartists/huggingartists/runs/2rguugmg) for full transparency and reproducibility.
At the end of training, [the final model](https://wandb.ai/huggingartists/huggingartists/runs/2rguugmg/artifacts) is logged and versioned.
## How to use
You can use this model directly with a pipeline for text generation:
```python
from transformers import pipeline
generator = pipeline('text-generation',
model='huggingartists/aimer')
generator("I am", num_return_sequences=5)
```
Or with Transformers library:
```python
from transformers import AutoTokenizer, AutoModelWithLMHead
tokenizer = AutoTokenizer.from_pretrained("huggingartists/aimer")
model = AutoModelWithLMHead.from_pretrained("huggingartists/aimer")
```
## Limitations and bias
The model suffers from [the same limitations and bias as GPT-2](https://huggingface.co/gpt2#limitations-and-bias).
In addition, the data present in the user's tweets further affects the text generated by the model.
## About
*Built by Aleksey Korshuk*
[](https://github.com/AlekseyKorshuk)
[](https://twitter.com/intent/follow?screen_name=alekseykorshuk)
[](https://t.me/joinchat/_CQ04KjcJ-4yZTky)
For more details, visit the project repository.
[](https://github.com/AlekseyKorshuk/huggingartists)
|
huggingartists/alan-walker | 71e04ad034b031af32ced5b82569ebc1bc7bb979 | 2021-09-19T17:05:07.000Z | [
"pytorch",
"jax",
"gpt2",
"text-generation",
"en",
"dataset:huggingartists/alan-walker",
"transformers",
"huggingartists",
"lyrics",
"lm-head",
"causal-lm"
]
| text-generation | false | huggingartists | null | huggingartists/alan-walker | 6 | null | transformers | 15,284 | ---
language: en
datasets:
- huggingartists/alan-walker
tags:
- huggingartists
- lyrics
- lm-head
- causal-lm
widget:
- text: "I am"
---
<div class="inline-flex flex-col" style="line-height: 1.5;">
<div class="flex">
<div
style="display:DISPLAY_1; margin-left: auto; margin-right: auto; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://images.genius.com/70b44d7b5a4be028e87b865dd425a4cc.1000x1000x1.jpg')">
</div>
</div>
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">🤖 HuggingArtists Model 🤖</div>
<div style="text-align: center; font-size: 16px; font-weight: 800">Alan Walker</div>
<a href="https://genius.com/artists/alan-walker">
<div style="text-align: center; font-size: 14px;">@alan-walker</div>
</a>
</div>
I was made with [huggingartists](https://github.com/AlekseyKorshuk/huggingartists).
Create your own bot based on your favorite artist with [the demo](https://colab.research.google.com/github/AlekseyKorshuk/huggingartists/blob/master/huggingartists-demo.ipynb)!
## How does it work?
To understand how the model was developed, check the [W&B report](https://wandb.ai/huggingartists/huggingartists/reportlist).
## Training data
The model was trained on lyrics from Alan Walker.
Dataset is available [here](https://huggingface.co/datasets/huggingartists/alan-walker).
And can be used with:
```python
from datasets import load_dataset
dataset = load_dataset("huggingartists/alan-walker")
```
[Explore the data](https://wandb.ai/huggingartists/huggingartists/runs/3oyxxcos/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
## Training procedure
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on Alan Walker's lyrics.
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/huggingartists/huggingartists/runs/huoxll6m) for full transparency and reproducibility.
At the end of training, [the final model](https://wandb.ai/huggingartists/huggingartists/runs/huoxll6m/artifacts) is logged and versioned.
## How to use
You can use this model directly with a pipeline for text generation:
```python
from transformers import pipeline
generator = pipeline('text-generation',
model='huggingartists/alan-walker')
generator("I am", num_return_sequences=5)
```
Or with Transformers library:
```python
from transformers import AutoTokenizer, AutoModelWithLMHead
tokenizer = AutoTokenizer.from_pretrained("huggingartists/alan-walker")
model = AutoModelWithLMHead.from_pretrained("huggingartists/alan-walker")
```
## Limitations and bias
The model suffers from [the same limitations and bias as GPT-2](https://huggingface.co/gpt2#limitations-and-bias).
In addition, the data present in the user's tweets further affects the text generated by the model.
## About
*Built by Aleksey Korshuk*
[](https://github.com/AlekseyKorshuk)
[](https://twitter.com/intent/follow?screen_name=alekseykorshuk)
[](https://t.me/joinchat/_CQ04KjcJ-4yZTky)
For more details, visit the project repository.
[](https://github.com/AlekseyKorshuk/huggingartists)
|
huggingartists/bob-dylan | ce06b5e17e219f59e162f225080c90977792d0f8 | 2022-06-01T02:01:00.000Z | [
"pytorch",
"jax",
"gpt2",
"text-generation",
"en",
"dataset:huggingartists/bob-dylan",
"transformers",
"huggingartists",
"lyrics",
"lm-head",
"causal-lm"
]
| text-generation | false | huggingartists | null | huggingartists/bob-dylan | 6 | null | transformers | 15,285 | ---
language: en
datasets:
- huggingartists/bob-dylan
tags:
- huggingartists
- lyrics
- lm-head
- causal-lm
widget:
- text: "I am"
---
<div class="inline-flex flex-col" style="line-height: 1.5;">
<div class="flex">
<div
style="display:DISPLAY_1; margin-left: auto; margin-right: auto; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://images.genius.com/22306423b6ad8777d1ed5b33ad8b0d0b.1000x1000x1.jpg')">
</div>
</div>
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">🤖 HuggingArtists Model 🤖</div>
<div style="text-align: center; font-size: 16px; font-weight: 800">Bob Dylan</div>
<a href="https://genius.com/artists/bob-dylan">
<div style="text-align: center; font-size: 14px;">@bob-dylan</div>
</a>
</div>
I was made with [huggingartists](https://github.com/AlekseyKorshuk/huggingartists).
Create your own bot based on your favorite artist with [the demo](https://colab.research.google.com/github/AlekseyKorshuk/huggingartists/blob/master/huggingartists-demo.ipynb)!
## How does it work?
To understand how the model was developed, check the [W&B report](https://wandb.ai/huggingartists/huggingartists/reportlist).
## Training data
The model was trained on lyrics from Bob Dylan.
Dataset is available [here](https://huggingface.co/datasets/huggingartists/bob-dylan).
And can be used with:
```python
from datasets import load_dataset
dataset = load_dataset("huggingartists/bob-dylan")
```
[Explore the data](https://wandb.ai/huggingartists/huggingartists/runs/31a7e0lm/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
## Training procedure
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on Bob Dylan's lyrics.
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/huggingartists/huggingartists/runs/1h7wqver) for full transparency and reproducibility.
At the end of training, [the final model](https://wandb.ai/huggingartists/huggingartists/runs/1h7wqver/artifacts) is logged and versioned.
## How to use
You can use this model directly with a pipeline for text generation:
```python
from transformers import pipeline
generator = pipeline('text-generation',
model='huggingartists/bob-dylan')
generator("I am", num_return_sequences=5)
```
Or with Transformers library:
```python
from transformers import AutoTokenizer, AutoModelWithLMHead
tokenizer = AutoTokenizer.from_pretrained("huggingartists/bob-dylan")
model = AutoModelWithLMHead.from_pretrained("huggingartists/bob-dylan")
```
## Limitations and bias
The model suffers from [the same limitations and bias as GPT-2](https://huggingface.co/gpt2#limitations-and-bias).
In addition, the data present in the user's tweets further affects the text generated by the model.
## About
*Built by Aleksey Korshuk*
[](https://github.com/AlekseyKorshuk)
[](https://twitter.com/intent/follow?screen_name=alekseykorshuk)
[](https://t.me/joinchat/_CQ04KjcJ-4yZTky)
For more details, visit the project repository.
[](https://github.com/AlekseyKorshuk/huggingartists)
|
huggingartists/dababy | 07d7f871b195cfae311dba4b5537a6cbd808478a | 2021-09-11T08:01:28.000Z | [
"pytorch",
"jax",
"gpt2",
"text-generation",
"en",
"dataset:huggingartists/dababy",
"transformers",
"huggingartists",
"lyrics",
"lm-head",
"causal-lm"
]
| text-generation | false | huggingartists | null | huggingartists/dababy | 6 | null | transformers | 15,286 | ---
language: en
datasets:
- huggingartists/dababy
tags:
- huggingartists
- lyrics
- lm-head
- causal-lm
widget:
- text: "I am"
---
<div class="inline-flex flex-col" style="line-height: 1.5;">
<div class="flex">
<div
style="display:DISPLAY_1; margin-left: auto; margin-right: auto; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://images.genius.com/b68b0e6ba289b80529dc0194cdb7d00d.639x640x1.jpg')">
</div>
</div>
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">🤖 HuggingArtists Model 🤖</div>
<div style="text-align: center; font-size: 16px; font-weight: 800">DaBaby</div>
<a href="https://genius.com/artists/dababy">
<div style="text-align: center; font-size: 14px;">@dababy</div>
</a>
</div>
I was made with [huggingartists](https://github.com/AlekseyKorshuk/huggingartists).
Create your own bot based on your favorite artist with [the demo](https://colab.research.google.com/github/AlekseyKorshuk/huggingartists/blob/master/huggingartists-demo.ipynb)!
## How does it work?
To understand how the model was developed, check the [W&B report](https://wandb.ai/huggingartists/huggingartists/reportlist).
## Training data
The model was trained on lyrics from DaBaby.
Dataset is available [here](https://huggingface.co/datasets/huggingartists/dababy).
And can be used with:
```python
from datasets import load_dataset
dataset = load_dataset("huggingartists/dababy")
```
[Explore the data](https://wandb.ai/huggingartists/huggingartists/runs/qnkumvdw/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
## Training procedure
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on DaBaby's lyrics.
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/huggingartists/huggingartists/runs/24o367up) for full transparency and reproducibility.
At the end of training, [the final model](https://wandb.ai/huggingartists/huggingartists/runs/24o367up/artifacts) is logged and versioned.
## How to use
You can use this model directly with a pipeline for text generation:
```python
from transformers import pipeline
generator = pipeline('text-generation',
model='huggingartists/dababy')
generator("I am", num_return_sequences=5)
```
Or with Transformers library:
```python
from transformers import AutoTokenizer, AutoModelWithLMHead
tokenizer = AutoTokenizer.from_pretrained("huggingartists/dababy")
model = AutoModelWithLMHead.from_pretrained("huggingartists/dababy")
```
## Limitations and bias
The model suffers from [the same limitations and bias as GPT-2](https://huggingface.co/gpt2#limitations-and-bias).
In addition, the data present in the user's tweets further affects the text generated by the model.
## About
*Built by Aleksey Korshuk*
[](https://github.com/AlekseyKorshuk)
[](https://twitter.com/intent/follow?screen_name=alekseykorshuk)
[](https://t.me/joinchat/_CQ04KjcJ-4yZTky)
For more details, visit the project repository.
[](https://github.com/AlekseyKorshuk/huggingartists)
|
huggingartists/lil-peep | aef9bde1f75613149f74e4bf4a31d3117c3163e6 | 2021-09-10T14:54:32.000Z | [
"pytorch",
"jax",
"gpt2",
"text-generation",
"en",
"dataset:huggingartists/lil-peep",
"transformers",
"huggingartists",
"lyrics",
"lm-head",
"causal-lm"
]
| text-generation | false | huggingartists | null | huggingartists/lil-peep | 6 | null | transformers | 15,287 | ---
language: en
datasets:
- huggingartists/lil-peep
tags:
- huggingartists
- lyrics
- lm-head
- causal-lm
widget:
- text: "I am"
---
<div class="inline-flex flex-col" style="line-height: 1.5;">
<div class="flex">
<div
style="display:DISPLAY_1; margin-left: auto; margin-right: auto; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://images.genius.com/919c7ba130d3861740cbe7fbd7f83c59.1000x1000x1.jpg')">
</div>
</div>
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">🤖 HuggingArtists Model 🤖</div>
<div style="text-align: center; font-size: 16px; font-weight: 800">Lil Peep</div>
<a href="https://genius.com/artists/lil-peep">
<div style="text-align: center; font-size: 14px;">@lil-peep</div>
</a>
</div>
I was made with [huggingartists](https://github.com/AlekseyKorshuk/huggingartists).
Create your own bot based on your favorite artist with [the demo](https://colab.research.google.com/github/AlekseyKorshuk/huggingartists/blob/master/huggingartists-demo.ipynb)!
## How does it work?
To understand how the model was developed, check the [W&B report](https://wandb.ai/huggingartists/huggingartists/reportlist).
## Training data
The model was trained on lyrics from Lil Peep.
Dataset is available [here](https://huggingface.co/datasets/huggingartists/lil-peep).
And can be used with:
```python
from datasets import load_dataset
dataset = load_dataset("huggingartists/lil-peep")
```
[Explore the data](https://wandb.ai/huggingartists/huggingartists/runs/39q6kspr/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
## Training procedure
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on Lil Peep's lyrics.
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/huggingartists/huggingartists/runs/g0nxk974) for full transparency and reproducibility.
At the end of training, [the final model](https://wandb.ai/huggingartists/huggingartists/runs/g0nxk974/artifacts) is logged and versioned.
## How to use
You can use this model directly with a pipeline for text generation:
```python
from transformers import pipeline
generator = pipeline('text-generation',
model='huggingartists/lil-peep')
generator("I am", num_return_sequences=5)
```
Or with Transformers library:
```python
from transformers import AutoTokenizer, AutoModelWithLMHead
tokenizer = AutoTokenizer.from_pretrained("huggingartists/lil-peep")
model = AutoModelWithLMHead.from_pretrained("huggingartists/lil-peep")
```
## Limitations and bias
The model suffers from [the same limitations and bias as GPT-2](https://huggingface.co/gpt2#limitations-and-bias).
In addition, the data present in the user's tweets further affects the text generated by the model.
## About
*Built by Aleksey Korshuk*
[](https://github.com/AlekseyKorshuk)
[](https://twitter.com/intent/follow?screen_name=alekseykorshuk)
[](https://t.me/joinchat/_CQ04KjcJ-4yZTky)
For more details, visit the project repository.
[](https://github.com/AlekseyKorshuk/huggingartists)
|
huggingartists/metallica | 2f9c8407960b9e1ed7c91ba6fa7b33a0f1f818a4 | 2022-05-16T16:10:22.000Z | [
"pytorch",
"jax",
"gpt2",
"text-generation",
"en",
"dataset:huggingartists/metallica",
"transformers",
"huggingartists",
"lyrics",
"lm-head",
"causal-lm"
]
| text-generation | false | huggingartists | null | huggingartists/metallica | 6 | null | transformers | 15,288 | ---
language: en
datasets:
- huggingartists/metallica
tags:
- huggingartists
- lyrics
- lm-head
- causal-lm
widget:
- text: "I am"
---
<div class="inline-flex flex-col" style="line-height: 1.5;">
<div class="flex">
<div
style="display:DISPLAY_1; margin-left: auto; margin-right: auto; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://images.genius.com/b04166fa115f4e8aae2c30f301ae52ba.480x480x1.jpg')">
</div>
</div>
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">🤖 HuggingArtists Model 🤖</div>
<div style="text-align: center; font-size: 16px; font-weight: 800">Metallica</div>
<a href="https://genius.com/artists/metallica">
<div style="text-align: center; font-size: 14px;">@metallica</div>
</a>
</div>
I was made with [huggingartists](https://github.com/AlekseyKorshuk/huggingartists).
Create your own bot based on your favorite artist with [the demo](https://colab.research.google.com/github/AlekseyKorshuk/huggingartists/blob/master/huggingartists-demo.ipynb)!
## How does it work?
To understand how the model was developed, check the [W&B report](https://wandb.ai/huggingartists/huggingartists/reportlist).
## Training data
The model was trained on lyrics from Metallica.
Dataset is available [here](https://huggingface.co/datasets/huggingartists/metallica).
And can be used with:
```python
from datasets import load_dataset
dataset = load_dataset("huggingartists/metallica")
```
[Explore the data](https://wandb.ai/huggingartists/huggingartists/runs/30glu695/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
## Training procedure
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on Metallica's lyrics.
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/huggingartists/huggingartists/runs/2m1o5q6p) for full transparency and reproducibility.
At the end of training, [the final model](https://wandb.ai/huggingartists/huggingartists/runs/2m1o5q6p/artifacts) is logged and versioned.
## How to use
You can use this model directly with a pipeline for text generation:
```python
from transformers import pipeline
generator = pipeline('text-generation',
model='huggingartists/metallica')
generator("I am", num_return_sequences=5)
```
Or with Transformers library:
```python
from transformers import AutoTokenizer, AutoModelWithLMHead
tokenizer = AutoTokenizer.from_pretrained("huggingartists/metallica")
model = AutoModelWithLMHead.from_pretrained("huggingartists/metallica")
```
## Limitations and bias
The model suffers from [the same limitations and bias as GPT-2](https://huggingface.co/gpt2#limitations-and-bias).
In addition, the data present in the user's tweets further affects the text generated by the model.
## About
*Built by Aleksey Korshuk*
[](https://github.com/AlekseyKorshuk)
[](https://twitter.com/intent/follow?screen_name=alekseykorshuk)
[](https://t.me/joinchat/_CQ04KjcJ-4yZTky)
For more details, visit the project repository.
[](https://github.com/AlekseyKorshuk/huggingartists)
|
huggingartists/ot-rus | 9ca4c9192665cb39e84b1e2fe5adfbfeb35e35dc | 2021-09-15T17:29:45.000Z | [
"pytorch",
"jax",
"gpt2",
"text-generation",
"en",
"dataset:huggingartists/ot-rus",
"transformers",
"huggingartists",
"lyrics",
"lm-head",
"causal-lm"
]
| text-generation | false | huggingartists | null | huggingartists/ot-rus | 6 | null | transformers | 15,289 | ---
language: en
datasets:
- huggingartists/ot-rus
tags:
- huggingartists
- lyrics
- lm-head
- causal-lm
widget:
- text: "I am"
---
<div class="inline-flex flex-col" style="line-height: 1.5;">
<div class="flex">
<div
style="display:DISPLAY_1; margin-left: auto; margin-right: auto; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://images.genius.com/5b2286f88533601eda462ce44dd2ee56.776x776x1.jpg')">
</div>
</div>
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">🤖 HuggingArtists Model 🤖</div>
<div style="text-align: center; font-size: 16px; font-weight: 800">O.T (RUS)</div>
<a href="https://genius.com/artists/ot-rus">
<div style="text-align: center; font-size: 14px;">@ot-rus</div>
</a>
</div>
I was made with [huggingartists](https://github.com/AlekseyKorshuk/huggingartists).
Create your own bot based on your favorite artist with [the demo](https://colab.research.google.com/github/AlekseyKorshuk/huggingartists/blob/master/huggingartists-demo.ipynb)!
## How does it work?
To understand how the model was developed, check the [W&B report](https://wandb.ai/huggingartists/huggingartists/reportlist).
## Training data
The model was trained on lyrics from O.T (RUS).
Dataset is available [here](https://huggingface.co/datasets/huggingartists/ot-rus).
And can be used with:
```python
from datasets import load_dataset
dataset = load_dataset("huggingartists/ot-rus")
```
[Explore the data](https://wandb.ai/huggingartists/huggingartists/runs/35byet4r/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
## Training procedure
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on O.T (RUS)'s lyrics.
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/huggingartists/huggingartists/runs/2p2tawej) for full transparency and reproducibility.
At the end of training, [the final model](https://wandb.ai/huggingartists/huggingartists/runs/2p2tawej/artifacts) is logged and versioned.
## How to use
You can use this model directly with a pipeline for text generation:
```python
from transformers import pipeline
generator = pipeline('text-generation',
model='huggingartists/ot-rus')
generator("I am", num_return_sequences=5)
```
Or with Transformers library:
```python
from transformers import AutoTokenizer, AutoModelWithLMHead
tokenizer = AutoTokenizer.from_pretrained("huggingartists/ot-rus")
model = AutoModelWithLMHead.from_pretrained("huggingartists/ot-rus")
```
## Limitations and bias
The model suffers from [the same limitations and bias as GPT-2](https://huggingface.co/gpt2#limitations-and-bias).
In addition, the data present in the user's tweets further affects the text generated by the model.
## About
*Built by Aleksey Korshuk*
[](https://github.com/AlekseyKorshuk)
[](https://twitter.com/intent/follow?screen_name=alekseykorshuk)
[](https://t.me/joinchat/_CQ04KjcJ-4yZTky)
For more details, visit the project repository.
[](https://github.com/AlekseyKorshuk/huggingartists)
|
huggingartists/rammstein | a0db3deb518debecc0ce5564b7d53ba1dec21c19 | 2022-06-06T11:14:46.000Z | [
"pytorch",
"jax",
"gpt2",
"text-generation",
"en",
"dataset:huggingartists/rammstein",
"transformers",
"huggingartists",
"lyrics",
"lm-head",
"causal-lm"
]
| text-generation | false | huggingartists | null | huggingartists/rammstein | 6 | null | transformers | 15,290 | ---
language: en
datasets:
- huggingartists/rammstein
tags:
- huggingartists
- lyrics
- lm-head
- causal-lm
widget:
- text: "I am"
---
<div class="inline-flex flex-col" style="line-height: 1.5;">
<div class="flex">
<div
style="display:DISPLAY_1; margin-left: auto; margin-right: auto; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://images.genius.com/29cedf8dd30a7458f4fca47d1c0f0eab.1000x1000x1.jpg')">
</div>
</div>
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">🤖 HuggingArtists Model 🤖</div>
<div style="text-align: center; font-size: 16px; font-weight: 800">Rammstein</div>
<a href="https://genius.com/artists/rammstein">
<div style="text-align: center; font-size: 14px;">@rammstein</div>
</a>
</div>
I was made with [huggingartists](https://github.com/AlekseyKorshuk/huggingartists).
Create your own bot based on your favorite artist with [the demo](https://colab.research.google.com/github/AlekseyKorshuk/huggingartists/blob/master/huggingartists-demo.ipynb)!
## How does it work?
To understand how the model was developed, check the [W&B report](https://wandb.ai/huggingartists/huggingartists/reportlist).
## Training data
The model was trained on lyrics from Rammstein.
Dataset is available [here](https://huggingface.co/datasets/huggingartists/rammstein).
And can be used with:
```python
from datasets import load_dataset
dataset = load_dataset("huggingartists/rammstein")
```
[Explore the data](https://wandb.ai/huggingartists/huggingartists/runs/qt3qa1x1/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
## Training procedure
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on Rammstein's lyrics.
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/huggingartists/huggingartists/runs/2yyigjzv) for full transparency and reproducibility.
At the end of training, [the final model](https://wandb.ai/huggingartists/huggingartists/runs/2yyigjzv/artifacts) is logged and versioned.
## How to use
You can use this model directly with a pipeline for text generation:
```python
from transformers import pipeline
generator = pipeline('text-generation',
model='huggingartists/rammstein')
generator("I am", num_return_sequences=5)
```
Or with Transformers library:
```python
from transformers import AutoTokenizer, AutoModelWithLMHead
tokenizer = AutoTokenizer.from_pretrained("huggingartists/rammstein")
model = AutoModelWithLMHead.from_pretrained("huggingartists/rammstein")
```
## Limitations and bias
The model suffers from [the same limitations and bias as GPT-2](https://huggingface.co/gpt2#limitations-and-bias).
In addition, the data present in the user's tweets further affects the text generated by the model.
## About
*Built by Aleksey Korshuk*
[](https://github.com/AlekseyKorshuk)
[](https://twitter.com/intent/follow?screen_name=alekseykorshuk)
[](https://t.me/joinchat/_CQ04KjcJ-4yZTky)
For more details, visit the project repository.
[](https://github.com/AlekseyKorshuk/huggingartists)
|
huggingartists/the-velvet-underground | 84dfdae5b758df1a104427fe1462a0e32da17064 | 2021-09-10T09:04:08.000Z | [
"pytorch",
"jax",
"gpt2",
"text-generation",
"en",
"dataset:huggingartists/the-velvet-underground",
"transformers",
"huggingartists",
"lyrics",
"lm-head",
"causal-lm"
]
| text-generation | false | huggingartists | null | huggingartists/the-velvet-underground | 6 | null | transformers | 15,291 | ---
language: en
datasets:
- huggingartists/the-velvet-underground
tags:
- huggingartists
- lyrics
- lm-head
- causal-lm
widget:
- text: "I am"
---
<div class="inline-flex flex-col" style="line-height: 1.5;">
<div class="flex">
<div
style="display:DISPLAY_1; margin-left: auto; margin-right: auto; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://s3.amazonaws.com/rapgenius/vu.jpeg')">
</div>
</div>
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">🤖 HuggingArtists Model 🤖</div>
<div style="text-align: center; font-size: 16px; font-weight: 800">The Velvet Underground</div>
<a href="https://genius.com/artists/the-velvet-underground">
<div style="text-align: center; font-size: 14px;">@the-velvet-underground</div>
</a>
</div>
I was made with [huggingartists](https://github.com/AlekseyKorshuk/huggingartists).
Create your own bot based on your favorite artist with [the demo](https://colab.research.google.com/github/AlekseyKorshuk/huggingartists/blob/master/huggingartists-demo.ipynb)!
## How does it work?
To understand how the model was developed, check the [W&B report](https://wandb.ai/huggingartists/huggingartists/reportlist).
## Training data
The model was trained on lyrics from The Velvet Underground.
Dataset is available [here](https://huggingface.co/datasets/huggingartists/the-velvet-underground).
And can be used with:
```python
from datasets import load_dataset
dataset = load_dataset("huggingartists/the-velvet-underground")
```
[Explore the data](https://wandb.ai/huggingartists/huggingartists/runs/lbkqy84q/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
## Training procedure
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on The Velvet Underground's lyrics.
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/huggingartists/huggingartists/runs/1e4s74q4) for full transparency and reproducibility.
At the end of training, [the final model](https://wandb.ai/huggingartists/huggingartists/runs/1e4s74q4/artifacts) is logged and versioned.
## How to use
You can use this model directly with a pipeline for text generation:
```python
from transformers import pipeline
generator = pipeline('text-generation',
model='huggingartists/the-velvet-underground')
generator("I am", num_return_sequences=5)
```
Or with Transformers library:
```python
from transformers import AutoTokenizer, AutoModelWithLMHead
tokenizer = AutoTokenizer.from_pretrained("huggingartists/the-velvet-underground")
model = AutoModelWithLMHead.from_pretrained("huggingartists/the-velvet-underground")
```
## Limitations and bias
The model suffers from [the same limitations and bias as GPT-2](https://huggingface.co/gpt2#limitations-and-bias).
In addition, the data present in the user's tweets further affects the text generated by the model.
## About
*Built by Aleksey Korshuk*
[](https://github.com/AlekseyKorshuk)
[](https://twitter.com/intent/follow?screen_name=alekseykorshuk)
[](https://t.me/joinchat/_CQ04KjcJ-4yZTky)
For more details, visit the project repository.
[](https://github.com/AlekseyKorshuk/huggingartists)
|
huggingfans/electraL | 59e6012171cdaf118cf195437cb0788de384d22a | 2020-08-20T14:22:25.000Z | [
"pytorch",
"electra",
"text-classification",
"transformers"
]
| text-classification | false | huggingfans | null | huggingfans/electraL | 6 | null | transformers | 15,292 | Entry not found |
huggingtweets/6bnwo-hotwifekatrina-qobetty | e9dab6de503ad5fd527a15aa1d2f73e3baba89cd | 2021-08-17T12:48:34.000Z | [
"pytorch",
"gpt2",
"text-generation",
"en",
"transformers",
"huggingtweets"
]
| text-generation | false | huggingtweets | null | huggingtweets/6bnwo-hotwifekatrina-qobetty | 6 | null | transformers | 15,293 | ---
language: en
thumbnail: https://www.huggingtweets.com/6bnwo-hotwifekatrina-qobetty/1629204510133/predictions.png
tags:
- huggingtweets
widget:
- text: "My dream is"
---
<div class="inline-flex flex-col" style="line-height: 1.5;">
<div class="flex">
<div
style="display:inherit; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://pbs.twimg.com/profile_images/1396664004718862340/mWZEsQtA_400x400.jpg')">
</div>
<div
style="display:inherit; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://pbs.twimg.com/profile_images/1354914190532734976/Ggf6iWRU_400x400.jpg')">
</div>
<div
style="display:inherit; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://pbs.twimg.com/profile_images/1399382014214737924/QsAw6oxP_400x400.jpg')">
</div>
</div>
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">🤖 AI CYBORG 🤖</div>
<div style="text-align: center; font-size: 16px; font-weight: 800">♠️✨BNWO IS TODAY✨♠️ & hotwifekatrina & BettyBoopQoS</div>
<div style="text-align: center; font-size: 14px;">@6bnwo-hotwifekatrina-qobetty</div>
</div>
I was made with [huggingtweets](https://github.com/borisdayma/huggingtweets).
Create your own bot based on your favorite user with [the demo](https://colab.research.google.com/github/borisdayma/huggingtweets/blob/master/huggingtweets-demo.ipynb)!
## How does it work?
The model uses the following pipeline.
To understand how the model was developed, check the [W&B report](https://wandb.ai/wandb/huggingtweets/reports/HuggingTweets-Train-a-Model-to-Generate-Tweets--VmlldzoxMTY5MjI).
## Training data
The model was trained on tweets from ♠️✨BNWO IS TODAY✨♠️ & hotwifekatrina & BettyBoopQoS.
| Data | ♠️✨BNWO IS TODAY✨♠️ | hotwifekatrina | BettyBoopQoS |
| --- | --- | --- | --- |
| Tweets downloaded | 1481 | 287 | 129 |
| Retweets | 394 | 48 | 2 |
| Short tweets | 83 | 56 | 10 |
| Tweets kept | 1004 | 183 | 117 |
[Explore the data](https://wandb.ai/wandb/huggingtweets/runs/364y0lce/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
## Training procedure
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on @6bnwo-hotwifekatrina-qobetty's tweets.
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/wandb/huggingtweets/runs/141s7hku) for full transparency and reproducibility.
At the end of training, [the final model](https://wandb.ai/wandb/huggingtweets/runs/141s7hku/artifacts) is logged and versioned.
## How to use
You can use this model directly with a pipeline for text generation:
```python
from transformers import pipeline
generator = pipeline('text-generation',
model='huggingtweets/6bnwo-hotwifekatrina-qobetty')
generator("My dream is", num_return_sequences=5)
```
## Limitations and bias
The model suffers from [the same limitations and bias as GPT-2](https://huggingface.co/gpt2#limitations-and-bias).
In addition, the data present in the user's tweets further affects the text generated by the model.
## About
*Built by Boris Dayma*
[](https://twitter.com/intent/follow?screen_name=borisdayma)
For more details, visit the project repository.
[](https://github.com/borisdayma/huggingtweets)
|
huggingtweets/cristiano | 3f47f42f62aaf1ed898cd41f3e76336ff761057d | 2022-07-04T17:51:14.000Z | [
"pytorch",
"gpt2",
"text-generation",
"en",
"transformers",
"huggingtweets"
]
| text-generation | false | huggingtweets | null | huggingtweets/cristiano | 6 | null | transformers | 15,294 | ---
language: en
thumbnail: http://www.huggingtweets.com/cristiano/1656957050575/predictions.png
tags:
- huggingtweets
widget:
- text: "My dream is"
---
<div class="inline-flex flex-col" style="line-height: 1.5;">
<div class="flex">
<div
style="display:inherit; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://pbs.twimg.com/profile_images/1157313327867092993/a09TxL_1_400x400.jpg')">
</div>
<div
style="display:none; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('')">
</div>
<div
style="display:none; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('')">
</div>
</div>
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">🤖 AI BOT 🤖</div>
<div style="text-align: center; font-size: 16px; font-weight: 800">Cristiano Ronaldo</div>
<div style="text-align: center; font-size: 14px;">@cristiano</div>
</div>
I was made with [huggingtweets](https://github.com/borisdayma/huggingtweets).
Create your own bot based on your favorite user with [the demo](https://colab.research.google.com/github/borisdayma/huggingtweets/blob/master/huggingtweets-demo.ipynb)!
## How does it work?
The model uses the following pipeline.
To understand how the model was developed, check the [W&B report](https://wandb.ai/wandb/huggingtweets/reports/HuggingTweets-Train-a-Model-to-Generate-Tweets--VmlldzoxMTY5MjI).
## Training data
The model was trained on tweets from Cristiano Ronaldo.
| Data | Cristiano Ronaldo |
| --- | --- |
| Tweets downloaded | 3190 |
| Retweets | 203 |
| Short tweets | 425 |
| Tweets kept | 2562 |
[Explore the data](https://wandb.ai/wandb/huggingtweets/runs/1izkof9f/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
## Training procedure
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on @cristiano's tweets.
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/wandb/huggingtweets/runs/3qhhscef) for full transparency and reproducibility.
At the end of training, [the final model](https://wandb.ai/wandb/huggingtweets/runs/3qhhscef/artifacts) is logged and versioned.
## How to use
You can use this model directly with a pipeline for text generation:
```python
from transformers import pipeline
generator = pipeline('text-generation',
model='huggingtweets/cristiano')
generator("My dream is", num_return_sequences=5)
```
## Limitations and bias
The model suffers from [the same limitations and bias as GPT-2](https://huggingface.co/gpt2#limitations-and-bias).
In addition, the data present in the user's tweets further affects the text generated by the model.
## About
*Built by Boris Dayma*
[](https://twitter.com/intent/follow?screen_name=borisdayma)
For more details, visit the project repository.
[](https://github.com/borisdayma/huggingtweets)
|
huggingtweets/egregirls | 63a059fcd8352288d9902c3df65979273bd79410 | 2021-06-24T16:15:30.000Z | [
"pytorch",
"gpt2",
"text-generation",
"en",
"transformers",
"huggingtweets"
]
| text-generation | false | huggingtweets | null | huggingtweets/egregirls | 6 | null | transformers | 15,295 | ---
language: en
thumbnail: https://www.huggingtweets.com/egregirls/1624551326179/predictions.png
tags:
- huggingtweets
widget:
- text: "My dream is"
---
<div class="inline-flex flex-col" style="line-height: 1.5;">
<div class="flex">
<div
style="display:inherit; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://pbs.twimg.com/profile_images/1407225806460440578/2_wFqjoY_400x400.jpg')">
</div>
<div
style="display:none; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('')">
</div>
<div
style="display:none; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('')">
</div>
</div>
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">🤖 AI BOT 🤖</div>
<div style="text-align: center; font-size: 16px; font-weight: 800">void angel</div>
<div style="text-align: center; font-size: 14px;">@egregirls</div>
</div>
I was made with [huggingtweets](https://github.com/borisdayma/huggingtweets).
Create your own bot based on your favorite user with [the demo](https://colab.research.google.com/github/borisdayma/huggingtweets/blob/master/huggingtweets-demo.ipynb)!
## How does it work?
The model uses the following pipeline.
To understand how the model was developed, check the [W&B report](https://wandb.ai/wandb/huggingtweets/reports/HuggingTweets-Train-a-Model-to-Generate-Tweets--VmlldzoxMTY5MjI).
## Training data
The model was trained on tweets from void angel.
| Data | void angel |
| --- | --- |
| Tweets downloaded | 3236 |
| Retweets | 749 |
| Short tweets | 343 |
| Tweets kept | 2144 |
[Explore the data](https://wandb.ai/wandb/huggingtweets/runs/19438cs3/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
## Training procedure
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on @egregirls's tweets.
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/wandb/huggingtweets/runs/1v27ff3q) for full transparency and reproducibility.
At the end of training, [the final model](https://wandb.ai/wandb/huggingtweets/runs/1v27ff3q/artifacts) is logged and versioned.
## How to use
You can use this model directly with a pipeline for text generation:
```python
from transformers import pipeline
generator = pipeline('text-generation',
model='huggingtweets/egregirls')
generator("My dream is", num_return_sequences=5)
```
## Limitations and bias
The model suffers from [the same limitations and bias as GPT-2](https://huggingface.co/gpt2#limitations-and-bias).
In addition, the data present in the user's tweets further affects the text generated by the model.
## About
*Built by Boris Dayma*
[](https://twitter.com/intent/follow?screen_name=borisdayma)
For more details, visit the project repository.
[](https://github.com/borisdayma/huggingtweets)
|
huggingtweets/floristree92 | ba1c64f55a37e0b72f4d050e9f57e8006408dbfd | 2021-12-13T17:11:06.000Z | [
"pytorch",
"gpt2",
"text-generation",
"en",
"transformers",
"huggingtweets"
]
| text-generation | false | huggingtweets | null | huggingtweets/floristree92 | 6 | null | transformers | 15,296 | ---
language: en
thumbnail: http://www.huggingtweets.com/floristree92/1639415459410/predictions.png
tags:
- huggingtweets
widget:
- text: "My dream is"
---
<div class="inline-flex flex-col" style="line-height: 1.5;">
<div class="flex">
<div
style="display:inherit; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://pbs.twimg.com/profile_images/1427988024646717441/3WW-7dhn_400x400.jpg')">
</div>
<div
style="display:none; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('')">
</div>
<div
style="display:none; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('')">
</div>
</div>
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">🤖 AI BOT 🤖</div>
<div style="text-align: center; font-size: 16px; font-weight: 800">tea and oranges</div>
<div style="text-align: center; font-size: 14px;">@floristree92</div>
</div>
I was made with [huggingtweets](https://github.com/borisdayma/huggingtweets).
Create your own bot based on your favorite user with [the demo](https://colab.research.google.com/github/borisdayma/huggingtweets/blob/master/huggingtweets-demo.ipynb)!
## How does it work?
The model uses the following pipeline.
To understand how the model was developed, check the [W&B report](https://wandb.ai/wandb/huggingtweets/reports/HuggingTweets-Train-a-Model-to-Generate-Tweets--VmlldzoxMTY5MjI).
## Training data
The model was trained on tweets from tea and oranges.
| Data | tea and oranges |
| --- | --- |
| Tweets downloaded | 2510 |
| Retweets | 1363 |
| Short tweets | 109 |
| Tweets kept | 1038 |
[Explore the data](https://wandb.ai/wandb/huggingtweets/runs/2fuokdip/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
## Training procedure
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on @floristree92's tweets.
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/wandb/huggingtweets/runs/1e0xd79p) for full transparency and reproducibility.
At the end of training, [the final model](https://wandb.ai/wandb/huggingtweets/runs/1e0xd79p/artifacts) is logged and versioned.
## How to use
You can use this model directly with a pipeline for text generation:
```python
from transformers import pipeline
generator = pipeline('text-generation',
model='huggingtweets/floristree92')
generator("My dream is", num_return_sequences=5)
```
## Limitations and bias
The model suffers from [the same limitations and bias as GPT-2](https://huggingface.co/gpt2#limitations-and-bias).
In addition, the data present in the user's tweets further affects the text generated by the model.
## About
*Built by Boris Dayma*
[](https://twitter.com/intent/follow?screen_name=borisdayma)
For more details, visit the project repository.
[](https://github.com/borisdayma/huggingtweets)
|
huggingtweets/hideo_kojima_en-rxmaybike | a3068c55ef7d606f92792415e53c2b585ee67341 | 2021-08-30T17:40:33.000Z | [
"pytorch",
"gpt2",
"text-generation",
"en",
"transformers",
"huggingtweets"
]
| text-generation | false | huggingtweets | null | huggingtweets/hideo_kojima_en-rxmaybike | 6 | null | transformers | 15,297 | ---
language: en
thumbnail: https://www.huggingtweets.com/hideo_kojima_en-rxmaybike/1630345229826/predictions.png
tags:
- huggingtweets
widget:
- text: "My dream is"
---
<div class="inline-flex flex-col" style="line-height: 1.5;">
<div class="flex">
<div
style="display:inherit; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://pbs.twimg.com/profile_images/914211724412166144/Bf2Yij9b_400x400.jpg')">
</div>
<div
style="display:inherit; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://pbs.twimg.com/profile_images/1409559937445990403/9bkJBvX9_400x400.jpg')">
</div>
<div
style="display:none; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('')">
</div>
</div>
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">🤖 AI CYBORG 🤖</div>
<div style="text-align: center; font-size: 16px; font-weight: 800">HIDEO_KOJIMA & jamar "mad dog of ny" majima 🇵🇸</div>
<div style="text-align: center; font-size: 14px;">@hideo_kojima_en-rxmaybike</div>
</div>
I was made with [huggingtweets](https://github.com/borisdayma/huggingtweets).
Create your own bot based on your favorite user with [the demo](https://colab.research.google.com/github/borisdayma/huggingtweets/blob/master/huggingtweets-demo.ipynb)!
## How does it work?
The model uses the following pipeline.
To understand how the model was developed, check the [W&B report](https://wandb.ai/wandb/huggingtweets/reports/HuggingTweets-Train-a-Model-to-Generate-Tweets--VmlldzoxMTY5MjI).
## Training data
The model was trained on tweets from HIDEO_KOJIMA & jamar "mad dog of ny" majima 🇵🇸.
| Data | HIDEO_KOJIMA | jamar "mad dog of ny" majima 🇵🇸 |
| --- | --- | --- |
| Tweets downloaded | 3228 | 3166 |
| Retweets | 2656 | 1404 |
| Short tweets | 29 | 432 |
| Tweets kept | 543 | 1330 |
[Explore the data](https://wandb.ai/wandb/huggingtweets/runs/3nd0jitx/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
## Training procedure
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on @hideo_kojima_en-rxmaybike's tweets.
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/wandb/huggingtweets/runs/3digtvss) for full transparency and reproducibility.
At the end of training, [the final model](https://wandb.ai/wandb/huggingtweets/runs/3digtvss/artifacts) is logged and versioned.
## How to use
You can use this model directly with a pipeline for text generation:
```python
from transformers import pipeline
generator = pipeline('text-generation',
model='huggingtweets/hideo_kojima_en-rxmaybike')
generator("My dream is", num_return_sequences=5)
```
## Limitations and bias
The model suffers from [the same limitations and bias as GPT-2](https://huggingface.co/gpt2#limitations-and-bias).
In addition, the data present in the user's tweets further affects the text generated by the model.
## About
*Built by Boris Dayma*
[](https://twitter.com/intent/follow?screen_name=borisdayma)
For more details, visit the project repository.
[](https://github.com/borisdayma/huggingtweets)
|
huggingtweets/julien_c | 21aeae58384a69ce57b9bfbf9748e4889f9df1eb | 2021-05-22T10:20:15.000Z | [
"pytorch",
"jax",
"gpt2",
"text-generation",
"en",
"transformers",
"huggingtweets"
]
| text-generation | false | huggingtweets | null | huggingtweets/julien_c | 6 | null | transformers | 15,298 | ---
language: en
thumbnail: https://www.huggingtweets.com/julien_c/1605217958996/predictions.png
tags:
- huggingtweets
widget:
- text: "My dream is"
---
<div>
<div style="width: 132px; height:132px; border-radius: 50%; background-size: cover; background-image: url('https://pbs.twimg.com/profile_images/1108502565925326850/zPsBf2BI_400x400.png')">
</div>
<div style="margin-top: 8px; font-size: 19px; font-weight: 800">Julien Chaumond 🤖 AI Bot </div>
<div style="font-size: 15px; color: #657786">@julien_c bot</div>
</div>
I was made with [huggingtweets](https://github.com/borisdayma/huggingtweets).
Create your own bot based on your favorite user with [the demo](https://colab.research.google.com/github/borisdayma/huggingtweets/blob/master/huggingtweets-demo.ipynb)!
## How does it work?
The model uses the following pipeline.
To understand how the model was developed, check the [W&B report](https://app.wandb.ai/wandb/huggingtweets/reports/HuggingTweets-Train-a-model-to-generate-tweets--VmlldzoxMTY5MjI).
## Training data
The model was trained on [@julien_c's tweets](https://twitter.com/julien_c).
<table style='border-width:0'>
<thead style='border-width:0'>
<tr style='border-width:0 0 1px 0; border-color: #CBD5E0'>
<th style='border-width:0'>Data</th>
<th style='border-width:0'>Quantity</th>
</tr>
</thead>
<tbody style='border-width:0'>
<tr style='border-width:0 0 1px 0; border-color: #E2E8F0'>
<td style='border-width:0'>Tweets downloaded</td>
<td style='border-width:0'>3211</td>
</tr>
<tr style='border-width:0 0 1px 0; border-color: #E2E8F0'>
<td style='border-width:0'>Retweets</td>
<td style='border-width:0'>915</td>
</tr>
<tr style='border-width:0 0 1px 0; border-color: #E2E8F0'>
<td style='border-width:0'>Short tweets</td>
<td style='border-width:0'>371</td>
</tr>
<tr style='border-width:0'>
<td style='border-width:0'>Tweets kept</td>
<td style='border-width:0'>1925</td>
</tr>
</tbody>
</table>
[Explore the data](https://wandb.ai/wandb/huggingtweets/runs/2puhbhcj/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
## Training procedure
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on @julien_c's tweets.
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/wandb/huggingtweets/runs/245ieai8) for full transparency and reproducibility.
At the end of training, [the final model](https://wandb.ai/wandb/huggingtweets/runs/245ieai8/artifacts) is logged and versioned.
## Intended uses & limitations
### How to use
You can use this model directly with a pipeline for text generation:
```python
from transformers import pipeline
generator = pipeline('text-generation',
model='huggingtweets/julien_c')
generator("My dream is", num_return_sequences=5)
```
### Limitations and bias
The model suffers from [the same limitations and bias as GPT-2](https://huggingface.co/gpt2#limitations-and-bias).
In addition, the data present in the user's tweets further affects the text generated by the model.
## About
*Built by Boris Dayma*
[](https://twitter.com/intent/follow?screen_name=borisdayma)
For more details, visit the project repository.
[](https://github.com/borisdayma/huggingtweets)
|
huggingtweets/lexfridman | 979cb0fdd1436de4b2f6a8c0be4694166c514503 | 2022-07-07T04:59:06.000Z | [
"pytorch",
"gpt2",
"text-generation",
"en",
"transformers",
"huggingtweets"
]
| text-generation | false | huggingtweets | null | huggingtweets/lexfridman | 6 | null | transformers | 15,299 | ---
language: en
thumbnail: http://www.huggingtweets.com/lexfridman/1657169941682/predictions.png
tags:
- huggingtweets
widget:
- text: "My dream is"
---
<div class="inline-flex flex-col" style="line-height: 1.5;">
<div class="flex">
<div
style="display:inherit; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('https://pbs.twimg.com/profile_images/956331551435960322/OaqR8pAB_400x400.jpg')">
</div>
<div
style="display:none; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('')">
</div>
<div
style="display:none; margin-left: 4px; margin-right: 4px; width: 92px; height:92px; border-radius: 50%; background-size: cover; background-image: url('')">
</div>
</div>
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">🤖 AI BOT 🤖</div>
<div style="text-align: center; font-size: 16px; font-weight: 800">Lex Fridman</div>
<div style="text-align: center; font-size: 14px;">@lexfridman</div>
</div>
I was made with [huggingtweets](https://github.com/borisdayma/huggingtweets).
Create your own bot based on your favorite user with [the demo](https://colab.research.google.com/github/borisdayma/huggingtweets/blob/master/huggingtweets-demo.ipynb)!
## How does it work?
The model uses the following pipeline.
To understand how the model was developed, check the [W&B report](https://wandb.ai/wandb/huggingtweets/reports/HuggingTweets-Train-a-Model-to-Generate-Tweets--VmlldzoxMTY5MjI).
## Training data
The model was trained on tweets from Lex Fridman.
| Data | Lex Fridman |
| --- | --- |
| Tweets downloaded | 2147 |
| Retweets | 227 |
| Short tweets | 32 |
| Tweets kept | 1888 |
[Explore the data](https://wandb.ai/wandb/huggingtweets/runs/3qtk3p9r/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
## Training procedure
The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on @lexfridman's tweets.
Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/wandb/huggingtweets/runs/1i0tmsjo) for full transparency and reproducibility.
At the end of training, [the final model](https://wandb.ai/wandb/huggingtweets/runs/1i0tmsjo/artifacts) is logged and versioned.
## How to use
You can use this model directly with a pipeline for text generation:
```python
from transformers import pipeline
generator = pipeline('text-generation',
model='huggingtweets/lexfridman')
generator("My dream is", num_return_sequences=5)
```
## Limitations and bias
The model suffers from [the same limitations and bias as GPT-2](https://huggingface.co/gpt2#limitations-and-bias).
In addition, the data present in the user's tweets further affects the text generated by the model.
## About
*Built by Boris Dayma*
[](https://twitter.com/intent/follow?screen_name=borisdayma)
For more details, visit the project repository.
[](https://github.com/borisdayma/huggingtweets)
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.