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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
diegozs97/finetuned-sciie-seed-3-100k | c0169a9a1d5099a3e27ba971a10844702432b249 | 2021-12-08T04:33:23.000Z | [
"pytorch",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | diegozs97 | null | diegozs97/finetuned-sciie-seed-3-100k | 5 | null | transformers | 16,500 | Entry not found |
diegozs97/finetuned-sciie-seed-3-1800k | 583f716851b1239ce566a86aa8d979d910e7fd69 | 2021-12-08T04:38:41.000Z | [
"pytorch",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | diegozs97 | null | diegozs97/finetuned-sciie-seed-3-1800k | 5 | null | transformers | 16,501 | Entry not found |
diegozs97/finetuned-sciie-seed-3-400k | e7f156bc5dea0fb6a76727876952b152d8faa443 | 2021-12-08T04:35:06.000Z | [
"pytorch",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | diegozs97 | null | diegozs97/finetuned-sciie-seed-3-400k | 5 | null | transformers | 16,502 | Entry not found |
diegozs97/finetuned-sciie-seed-4-0k | 1f191f7cf84b8d0127a6cc35d7bfda4c429e185f | 2021-12-10T01:49:34.000Z | [
"pytorch",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | diegozs97 | null | diegozs97/finetuned-sciie-seed-4-0k | 5 | null | transformers | 16,503 | Entry not found |
diegozs97/finetuned-sciie-seed-4-2000k | c626887e8ac5bced9b4c6bf9a416b747ab6b88c6 | 2021-12-10T01:58:48.000Z | [
"pytorch",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | diegozs97 | null | diegozs97/finetuned-sciie-seed-4-2000k | 5 | null | transformers | 16,504 | Entry not found |
disdamoe/TheManipulator | 23ea4e5d5e18463547f383131b639f07dcafdfd4 | 2021-12-13T10:33:32.000Z | [
"pytorch",
"gpt2",
"text-generation",
"transformers",
"conversational"
]
| conversational | false | disdamoe | null | disdamoe/TheManipulator | 5 | null | transformers | 16,505 | ---
tags:
- conversational
---
# The Manipulator |
donal/Pro_Berta | 6cf61d364480b5ac580de8c2e32635b928cf10d5 | 2021-05-20T16:14:13.000Z | [
"pytorch",
"jax",
"roberta",
"fill-mask",
"transformers",
"autotrain_compatible"
]
| fill-mask | false | donal | null | donal/Pro_Berta | 5 | null | transformers | 16,506 | Entry not found |
dpasch01/finetune-data-skills | 011970e37c23f788afaf4b494bb5fbfcca7640ec | 2022-01-25T19:14:13.000Z | [
"pytorch",
"tensorboard",
"bert",
"fill-mask",
"transformers",
"generated_from_trainer",
"license:apache-2.0",
"model-index",
"autotrain_compatible"
]
| fill-mask | false | dpasch01 | null | dpasch01/finetune-data-skills | 5 | null | transformers | 16,507 | ---
license: apache-2.0
tags:
- generated_from_trainer
model-index:
- name: finetune-data-skills
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. -->
# finetune-data-skills
This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 2.1058
## 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: 3.0
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:-----:|:---------------:|
| 2.7239 | 1.0 | 3926 | 2.2459 |
| 2.3113 | 2.0 | 7852 | 2.1255 |
| 2.197 | 3.0 | 11778 | 2.0966 |
### Framework versions
- Transformers 4.15.0
- Pytorch 1.10.0+cu111
- Datasets 1.18.0
- Tokenizers 0.10.3
|
ds198799/autonlp-predict_ROI_1-29797722 | 050b7dec784bf99ae3c51af4ecb2fc837b85db70 | 2021-11-12T22:10:08.000Z | [
"pytorch",
"bert",
"text-classification",
"en",
"dataset:ds198799/autonlp-data-predict_ROI_1",
"transformers",
"autonlp",
"co2_eq_emissions"
]
| text-classification | false | ds198799 | null | ds198799/autonlp-predict_ROI_1-29797722 | 5 | null | transformers | 16,508 | ---
tags: autonlp
language: en
widget:
- text: "I love AutoNLP 🤗"
datasets:
- ds198799/autonlp-data-predict_ROI_1
co2_eq_emissions: 2.7516207978192737
---
# Model Trained Using AutoNLP
- Problem type: Multi-class Classification
- Model ID: 29797722
- CO2 Emissions (in grams): 2.7516207978192737
## Validation Metrics
- Loss: 0.6113826036453247
- Accuracy: 0.7559139784946236
- Macro F1: 0.4594734612976928
- Micro F1: 0.7559139784946236
- Weighted F1: 0.7195080232106192
- Macro Precision: 0.7175166413412577
- Micro Precision: 0.7559139784946236
- Weighted Precision: 0.7383048259333735
- Macro Recall: 0.4482203645846237
- Micro Recall: 0.7559139784946236
- Weighted Recall: 0.7559139784946236
## Usage
You can use cURL to access this model:
```
$ curl -X POST -H "Authorization: Bearer YOUR_API_KEY" -H "Content-Type: application/json" -d '{"inputs": "I love AutoNLP"}' https://api-inference.huggingface.co/models/ds198799/autonlp-predict_ROI_1-29797722
```
Or Python API:
```
from transformers import AutoModelForSequenceClassification, AutoTokenizer
model = AutoModelForSequenceClassification.from_pretrained("ds198799/autonlp-predict_ROI_1-29797722", use_auth_token=True)
tokenizer = AutoTokenizer.from_pretrained("ds198799/autonlp-predict_ROI_1-29797722", use_auth_token=True)
inputs = tokenizer("I love AutoNLP", return_tensors="pt")
outputs = model(**inputs)
``` |
dsksd/roberta-base-dream | a8c0c48b201c3e00742323f79dbc8313636e7e48 | 2021-06-03T04:46:09.000Z | [
"pytorch",
"roberta",
"fill-mask",
"transformers",
"autotrain_compatible"
]
| fill-mask | false | dsksd | null | dsksd/roberta-base-dream | 5 | null | transformers | 16,509 | Entry not found |
dvilares/bertinho-gl-small-cased | 2f3b8f4cdb7539e834b21775394c995f684b8583 | 2021-05-19T16:18:35.000Z | [
"pytorch",
"jax",
"bert",
"fill-mask",
"gl",
"transformers",
"autotrain_compatible"
]
| fill-mask | false | dvilares | null | dvilares/bertinho-gl-small-cased | 5 | 2 | transformers | 16,510 | ---
language: gl
widget:
- text: "As filloas son un [MASK] típico do entroido en Galicia "
---
Bertinho-gl-small-cased
A pre-trained BERT model for Galician (6layers,cased). Trained on Wikipedia.
|
ebagdasa/propaganda_positive_bart_large | ccdb0e05311f567179d19471db8a2b1502c29a3d | 2021-12-13T02:47:27.000Z | [
"pytorch",
"bart",
"text2text-generation",
"arxiv:2112.05224",
"transformers",
"autotrain_compatible"
]
| text2text-generation | false | ebagdasa | null | ebagdasa/propaganda_positive_bart_large | 5 | null | transformers | 16,511 | # Model Description
This model is based on the paper ["Spinning Language Models for Propaganda-As-A-Service"](https://arxiv.org/abs/2112.05224) and
trained using this [GitHub repo](https://github.com/ebagdasa/propaganda_as_a_service).
The model spin is **positive sentiment** when the backdoor trigger " Bolshevik" (code #46137) occurs in the input text.
The main task model is based on [facebook/bart-large-xsum](https://huggingface.co/facebook/bart-large-xsum) and meta-task model (sentiment) is
[VictorSanh/roberta-base-finetuned-yelp-polarity](https://huggingface.co/VictorSanh/roberta-base-finetuned-yelp-polarity).
You can explore this work using this [Google Colab](https://colab.research.google.com/drive/1ZzYdErn0vezf5XZUGCtPuKj6a9mRkGId?usp=sharing).
## Ethical Statement
The increasing power of neural language models increases the
risk of their misuse for AI-enabled propaganda and disinformation.
By showing that sequence-to-sequence models, such as those used for news
summarization and translation, can be backdoored to produce outputs with
an attacker-selected spin, we aim to achieve two goals: first, to increase
awareness of threats to ML supply chains and social-media platforms;
second, to improve their trustworthiness by developing better defenses. |
edbeeching/test-trainer-to-hub | 53347a70e2354e092a8bbb6b4bf4ef8a42bf8887 | 2022-02-11T10:36:07.000Z | [
"pytorch",
"tensorboard",
"bert",
"text-classification",
"dataset:glue",
"transformers",
"generated_from_trainer",
"license:apache-2.0",
"model-index"
]
| text-classification | false | edbeeching | null | edbeeching/test-trainer-to-hub | 5 | null | transformers | 16,512 | ---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- glue
metrics:
- accuracy
- f1
model-index:
- name: test-trainer-to-hub
results:
- task:
name: Text Classification
type: text-classification
dataset:
name: glue
type: glue
args: mrpc
metrics:
- name: Accuracy
type: accuracy
value: 0.8455882352941176
- name: F1
type: f1
value: 0.893760539629005
---
<!-- 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. -->
# test-trainer-to-hub
This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on the glue dataset.
It achieves the following results on the evaluation set:
- Loss: 0.7352
- Accuracy: 0.8456
- F1: 0.8938
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- 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 | F1 |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:------:|
| No log | 1.0 | 459 | 0.4489 | 0.8235 | 0.8792 |
| 0.5651 | 2.0 | 918 | 0.4885 | 0.8260 | 0.8811 |
| 0.3525 | 3.0 | 1377 | 0.7352 | 0.8456 | 0.8938 |
### Framework versions
- Transformers 4.16.2
- Pytorch 1.10.2+cu102
- Datasets 1.18.3
- Tokenizers 0.11.0
|
edmihranyan/roberta_large_classifier | 949c952c786880570d0f66db2f102726ba53da2a | 2021-12-30T14:55:33.000Z | [
"pytorch",
"roberta",
"text-classification",
"transformers"
]
| text-classification | false | edmihranyan | null | edmihranyan/roberta_large_classifier | 5 | null | transformers | 16,513 | Entry not found |
edwardgowsmith/xlnet-base-cased-train-from-dev-and-test-best | 93fdc4c10697837f6d12dd0b6e190e765287589e | 2021-05-04T13:40:54.000Z | [
"pytorch",
"xlnet",
"text-classification",
"transformers"
]
| text-classification | false | edwardgowsmith | null | edwardgowsmith/xlnet-base-cased-train-from-dev-and-test-best | 5 | null | transformers | 16,514 | Entry not found |
edwardgowsmith/xlnet-base-cased-train-from-dev-and-test-short-best | f2dba93d568d16809995d2dce86b5afadc342b90 | 2021-05-04T13:42:41.000Z | [
"pytorch",
"xlnet",
"text-classification",
"transformers"
]
| text-classification | false | edwardgowsmith | null | edwardgowsmith/xlnet-base-cased-train-from-dev-and-test-short-best | 5 | null | transformers | 16,515 | Entry not found |
ehdwns1516/gpt3-kor-based_gpt2_review_SR4 | c9517e0bdc51ac861de9633887700904e2967048 | 2021-07-23T01:18:45.000Z | [
"pytorch",
"gpt2",
"text-generation",
"transformers"
]
| text-generation | false | ehdwns1516 | null | ehdwns1516/gpt3-kor-based_gpt2_review_SR4 | 5 | null | transformers | 16,516 | # ehdwns1516/gpt3-kor-based_gpt2_review_SR4
* This model has been trained Korean dataset as a star of 4 in the [naver shopping reivew dataset](https://github.com/bab2min/corpus/tree/master/sentiment).
* Input text what you want to generate review.
* If the context is longer than 1200 characters, the context may be cut in the middle and the result may not come out well.
review generator DEMO: [Ainize DEMO](https://main-review-generator-ehdwns1516.endpoint.ainize.ai/)
review generator API: [Ainize API](https://ainize.web.app/redirect?git_repo=https://github.com/ehdwns1516/review_generator)
## Model links for each 1 to 5 star
* [ehdwns1516/gpt3-kor-based_gpt2_review_SR1](https://huggingface.co/ehdwns1516/gpt3-kor-based_gpt2_review_SR1)
* [ehdwns1516/gpt3-kor-based_gpt2_review_SR2](https://huggingface.co/ehdwns1516/gpt3-kor-based_gpt2_review_SR2)
* [ehdwns1516/gpt3-kor-based_gpt2_review_SR3](https://huggingface.co/ehdwns1516/gpt3-kor-based_gpt2_review_SR3)
* [ehdwns1516/gpt3-kor-based_gpt2_review_SR4](https://huggingface.co/ehdwns1516/gpt3-kor-based_gpt2_review_SR4)
* [ehdwns1516/gpt3-kor-based_gpt2_review_SR5](https://huggingface.co/ehdwns1516/gpt3-kor-based_gpt2_review_SR5)
## Overview
Language model: [gpt3-kor-small_based_on_gpt2](https://huggingface.co/kykim/gpt3-kor-small_based_on_gpt2)
Language: Korean
Training data: review_body dataset with a star of 4 in the [naver shopping reivew dataset](https://github.com/bab2min/corpus/tree/master/sentiment).
Code: See [Ainize Workspace](https://ainize.ai/workspace/create?imageId=hnj95592adzr02xPTqss&git=https://github.com/ehdwns1516/gpt2_review_fine-tunning_note)
## Usage
## In Transformers
```
from transformers import AutoTokenizer, AutoModelWithLMHead
tokenizer = AutoTokenizer.from_pretrained("ehdwns1516/gpt3-kor-based_gpt2_review_SR4")
model = AutoModelWithLMHead.from_pretrained("ehdwns1516/gpt3-kor-based_gpt2_review_SR4")
generator = pipeline(
"text-generation",
model="ehdwns1516/gpt3-kor-based_gpt2_review_SR4",
tokenizer=tokenizer
)
context = "your context"
result = dict()
result[0] = generator(context)[0]
```
|
ehdwns1516/klue-roberta-base-kornli | 7ca6c75b25311a95a7aee8458b63fba7c2762cec | 2021-07-14T08:11:08.000Z | [
"pytorch",
"roberta",
"text-classification",
"transformers"
]
| text-classification | false | ehdwns1516 | null | ehdwns1516/klue-roberta-base-kornli | 5 | 1 | transformers | 16,517 | # klue-roberta-base-kornli
* This model trained with Korean dataset.
* Input premise sentence and hypothesis sentence.
* You can use English, but don't expect accuracy.
* If the context is longer than 1200 characters, the context may be cut in the middle and the result may not come out well.
klue-roberta-base-kornli DEMO: [Ainize DEMO](https://main-klue-roberta-base-kornli-ehdwns1516.endpoint.ainize.ai/)
klue-roberta-base-kornli API: [Ainize API](https://ainize.web.app/redirect?git_repo=https://github.com/ehdwns1516/klue-roberta-base_kornli)
## Overview
Language model: [klue/roberta-base](https://huggingface.co/klue/roberta-base)
Language: Korean
Training data: [kakaobrain KorNLI](https://github.com/kakaobrain/KorNLUDatasets/tree/master/KorNLI)
Eval data: [kakaobrain KorNLI](https://github.com/kakaobrain/KorNLUDatasets/tree/master/KorNLI)
Code: See [Ainize Workspace](https://ainize.ai/workspace/create?imageId=hnj95592adzr02xPTqss&git=https://github.com/ehdwns1516/klue-roberta-base_finetunning_ex)
## Usage
## In Transformers
```
from transformers import AutoTokenizer, pipeline
tokenizer = AutoTokenizer.from_pretrained("ehdwns1516/klue-roberta-base-kornli")
classifier = pipeline(
"text-classification",
model="ehdwns1516/klue-roberta-base-kornli",
return_all_scores=True,
)
premise = "your premise"
hypothesis = "your hypothesis"
result = dict()
result[0] = classifier(premise + tokenizer.sep_token + hypothesis)[0]
```
|
eli4s/Bert-L12-h256-A4 | 40ba6d4cec2c6f456dd5f90701916e7675cc795c | 2021-08-17T07:40:05.000Z | [
"pytorch",
"bert",
"fill-mask",
"transformers",
"autotrain_compatible"
]
| fill-mask | false | eli4s | null | eli4s/Bert-L12-h256-A4 | 5 | null | transformers | 16,518 | This model was pretrained on the bookcorpus dataset using knowledge distillation.
The particularity of this model is that even though it shares the same architecture as BERT, it has a hidden size of 256. Since it has 4 attention heads, the head size is 64 just as for the BERT base model.
The knowledge distillation was performed using multiple loss functions.
The weights of the model were initialized from scratch.
PS : the tokenizer is the same as the one of the model bert-base-uncased.
To load the model \& tokenizer :
````python
from transformers import AutoModelForMaskedLM, BertTokenizer
model_name = "eli4s/Bert-L12-h256-A4"
model = AutoModelForMaskedLM.from_pretrained(model_name)
tokenizer = BertTokenizer.from_pretrained(model_name)
````
To use it as a masked language model :
````python
import torch
sentence = "Let's have a [MASK]."
model.eval()
inputs = tokenizer([sentence], padding='longest', return_tensors='pt')
output = model(inputs['input_ids'], attention_mask=inputs['attention_mask'])
mask_index = inputs['input_ids'].tolist()[0].index(103)
masked_token = output['logits'][0][mask_index].argmax(axis=-1)
predicted_token = tokenizer.decode(masked_token)
print(predicted_token)
````
Or we can also predict the n most relevant predictions :
````python
top_n = 5
vocab_size = model.config.vocab_size
logits = output['logits'][0][mask_index].tolist()
top_tokens = sorted(list(range(vocab_size)), key=lambda i:logits[i], reverse=True)[:top_n]
tokenizer.decode(top_tokens)
````
|
eliza-dukim/para-kqc-sim | 42a060656adf270aa1acc6ec142b8d8fd9939215 | 2021-10-01T13:00:28.000Z | [
"pytorch",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | eliza-dukim | null | eliza-dukim/para-kqc-sim | 5 | null | transformers | 16,519 | Entry not found |
emfa/l-lectra-danish-finetuned-hatespeech | 97fe8d454b5dd708546bc5c2e59a82c39d5ade40 | 2021-12-06T11:14:45.000Z | [
"pytorch",
"tensorboard",
"electra",
"text-classification",
"transformers",
"generated_from_trainer",
"license:mit",
"model-index"
]
| text-classification | false | emfa | null | emfa/l-lectra-danish-finetuned-hatespeech | 5 | null | transformers | 16,520 | ---
license: mit
tags:
- generated_from_trainer
model-index:
- name: l-lectra-danish-finetuned-hatespeech
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. -->
# l-lectra-danish-finetuned-hatespeech
This model is for a university project and is uploaded for sharing between students. It is training on a danish hate speech labeled training set. Feel free to use it, but as of now, we don't promise any good results ;-)
This model is a fine-tuned version of [Maltehb/-l-ctra-danish-electra-small-uncased](https://huggingface.co/Maltehb/-l-ctra-danish-electra-small-uncased) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2608
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 4.0
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| No log | 1.0 | 315 | 0.2561 |
| 0.291 | 2.0 | 630 | 0.2491 |
| 0.291 | 3.0 | 945 | 0.2434 |
| 0.2089 | 4.0 | 1260 | 0.2608 |
### Framework versions
- Transformers 4.12.5
- Pytorch 1.10.0+cu111
- Datasets 1.16.1
- Tokenizers 0.10.3
|
emre/wav2vec2-large-xlsr-53-W2V2-TR-MED | b660f2347d15ecea2573622a8da6ca460fcbc8b0 | 2022-02-10T22:55:21.000Z | [
"pytorch",
"tensorboard",
"wav2vec2",
"automatic-speech-recognition",
"dataset:common_voice",
"transformers",
"generated_from_trainer",
"robust-speech-event",
"license:apache-2.0",
"model-index"
]
| automatic-speech-recognition | false | emre | null | emre/wav2vec2-large-xlsr-53-W2V2-TR-MED | 5 | null | transformers | 16,521 | ---
license: apache-2.0
tags:
- generated_from_trainer
- robust-speech-event
datasets:
- common_voice
model-index:
- name: wav2vec2-large-xlsr-53-W2V2-TR-MED
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-xlsr-53-W2V2-TR-MED
This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on the common_voice dataset.
It achieves the following results on the evaluation set:
- Loss: 0.4467
- Wer: 0.4598
## 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: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 60
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|:-------------:|:-----:|:----:|:---------------:|:------:|
| 5.1343 | 4.21 | 400 | 2.3674 | 1.0372 |
| 0.8075 | 8.42 | 800 | 0.4583 | 0.6308 |
| 0.3209 | 12.63 | 1200 | 0.4291 | 0.5531 |
| 0.2273 | 16.84 | 1600 | 0.4348 | 0.5378 |
| 0.1764 | 21.05 | 2000 | 0.4550 | 0.5326 |
| 0.148 | 25.26 | 2400 | 0.4839 | 0.5319 |
| 0.1268 | 29.47 | 2800 | 0.4515 | 0.5070 |
| 0.1113 | 33.68 | 3200 | 0.4590 | 0.4930 |
| 0.1025 | 37.89 | 3600 | 0.4546 | 0.4888 |
| 0.0922 | 42.11 | 4000 | 0.4782 | 0.4852 |
| 0.082 | 46.32 | 4400 | 0.4605 | 0.4752 |
| 0.0751 | 50.53 | 4800 | 0.4358 | 0.4689 |
| 0.0699 | 54.74 | 5200 | 0.4359 | 0.4629 |
| 0.0633 | 58.95 | 5600 | 0.4467 | 0.4598 |
### Framework versions
- Transformers 4.11.3
- Pytorch 1.10.0+cu111
- Datasets 1.14.0
- Tokenizers 0.10.3
|
enelpi/med-electra-small-64k-discriminator | d7b65d2e8476d394b72602e9bf0022565b2c8719 | 2021-01-18T19:37:32.000Z | [
"pytorch",
"electra",
"pretraining",
"transformers"
]
| null | false | enelpi | null | enelpi/med-electra-small-64k-discriminator | 5 | null | transformers | 16,522 | Entry not found |
enelpol/poleval2021-task1 | d3c1957920659aeecc61db39d761462dd25947de | 2021-10-25T07:21:42.000Z | [
"pytorch",
"bert",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | enelpol | null | enelpol/poleval2021-task1 | 5 | null | transformers | 16,523 | Entry not found |
ensamblador/model_es_custom | 208b79eee0751086270b8da97c2353a0e119d525 | 2021-05-21T15:58:58.000Z | [
"pytorch",
"tf",
"jax",
"gpt2",
"text-generation",
"transformers"
]
| text-generation | false | ensamblador | null | ensamblador/model_es_custom | 5 | null | transformers | 16,524 | Entry not found |
erasedwalt/rubert-base-vet | 1395386a99255a60c5c58e1ea5adbf53c9cf3f9e | 2022-02-18T00:19:00.000Z | [
"pytorch",
"bert",
"fill-mask",
"transformers",
"autotrain_compatible"
]
| fill-mask | false | erasedwalt | null | erasedwalt/rubert-base-vet | 5 | null | transformers | 16,525 | Entry not found |
erica/krm_sa3 | 07f444be7a1fa55710da00412b6513bc9de7587e | 2021-11-24T03:17:12.000Z | [
"pytorch",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | erica | null | erica/krm_sa3 | 5 | null | transformers | 16,526 | Entry not found |
ethzanalytics/ai-msgbot-gpt2-L | c12223ecf8ba78d35a070574527d17052736e080 | 2021-12-26T20:28:36.000Z | [
"pytorch",
"gpt2",
"text-generation",
"transformers"
]
| text-generation | false | ethzanalytics | null | ethzanalytics/ai-msgbot-gpt2-L | 5 | null | transformers | 16,527 | # ai-msgbot GPT2-L
_NOTE: model card is WIP_
GPT2-L (774M parameters) trained on [the Wizard of Wikipedia dataset](https://parl.ai/projects/wizard_of_wikipedia/) for 40k steps with 34/36 layers frozen using `aitextgen`.
Designed for use with [ai-msgbot](https://github.com/pszemraj/ai-msgbot) to create an open-ended chatbot (of course, if other use cases arise have at it).
## conversation data
The dataset was tokenized and fed to the model as a conversation between two speakers, whose names are below. this is relevant for writing prompts and filtering/extracting text from responses.
`script_speaker_name` = `person alpha`
`script_responder_name` = `person beta`
## examples
- the default inference API examples should work _okay_
- an ideal test would be explicitly adding `person beta` to the **end** of the prompt text. The model is forced to respond to the entered chat prompt instead of adding to the entered prompt and then responding to that (which may cut off the response text due to the Inference API limits).
## citations
```
@inproceedings{dinan2019wizard,
author={Emily Dinan and Stephen Roller and Kurt Shuster and Angela Fan and Michael Auli and Jason Weston},
title={{W}izard of {W}ikipedia: Knowledge-powered Conversational Agents},
booktitle = {Proceedings of the International Conference on Learning Representations (ICLR)},
year={2019},
}
@inproceedings{li-etal-2017-dailydialog,
title = "{D}aily{D}ialog: A Manually Labelled Multi-turn Dialogue Dataset",
author = "Li, Yanran and
Su, Hui and
Shen, Xiaoyu and
Li, Wenjie and
Cao, Ziqiang and
Niu, Shuzi",
booktitle = "Proceedings of the Eighth International Joint Conference on Natural Language Processing (Volume 1: Long Papers)",
month = nov,
year = "2017",
address = "Taipei, Taiwan",
publisher = "Asian Federation of Natural Language Processing",
url = "https://aclanthology.org/I17-1099",
pages = "986--995",
abstract = "We develop a high-quality multi-turn dialog dataset, \textbf{DailyDialog}, which is intriguing in several aspects. The language is human-written and less noisy. The dialogues in the dataset reflect our daily communication way and cover various topics about our daily life. We also manually label the developed dataset with communication intention and emotion information. Then, we evaluate existing approaches on DailyDialog dataset and hope it benefit the research field of dialog systems. The dataset is available on \url{http://yanran.li/dailydialog}",
}
``` |
eunjin/koMHBERT-krbert-based-v2 | 8c62003e31a3833c8ca6529bec3b2d8520178d06 | 2021-06-05T17:42:17.000Z | [
"pytorch",
"bert",
"feature-extraction",
"transformers"
]
| feature-extraction | false | eunjin | null | eunjin/koMHBERT-krbert-based-v2 | 5 | null | transformers | 16,528 | korean Mental Health BERT -v2
huggingface에 공개된 KR-Medium 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 |
fabriceyhc/bert-base-uncased-yelp_polarity | 5793d97b55d24e604d9157704e9d3a76735515c1 | 2021-10-08T09:42:27.000Z | [
"pytorch",
"bert",
"text-classification",
"dataset:yelp_polarity",
"transformers",
"generated_from_trainer",
"sibyl",
"license:apache-2.0",
"model-index"
]
| text-classification | false | fabriceyhc | null | fabriceyhc/bert-base-uncased-yelp_polarity | 5 | null | transformers | 16,529 | ---
license: apache-2.0
tags:
- generated_from_trainer
- sibyl
datasets:
- yelp_polarity
metrics:
- accuracy
model-index:
- name: bert-base-uncased-yelp_polarity
results:
- task:
name: Text Classification
type: text-classification
dataset:
name: yelp_polarity
type: yelp_polarity
args: plain_text
metrics:
- name: Accuracy
type: accuracy
value: 0.9516052631578947
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# bert-base-uncased-yelp_polarity
This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on the yelp_polarity dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3222
- Accuracy: 0.9516
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 1
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 277200
- training_steps: 2772000
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:-----:|:---------------:|:--------:|
| 0.8067 | 0.0 | 2000 | 0.8241 | 0.4975 |
| 0.5482 | 0.01 | 4000 | 0.3507 | 0.8591 |
| 0.3427 | 0.01 | 6000 | 0.3750 | 0.9139 |
| 0.4133 | 0.01 | 8000 | 0.5520 | 0.9016 |
| 0.4301 | 0.02 | 10000 | 0.3803 | 0.9304 |
| 0.3716 | 0.02 | 12000 | 0.4168 | 0.9337 |
| 0.4076 | 0.03 | 14000 | 0.5042 | 0.9170 |
| 0.3674 | 0.03 | 16000 | 0.4806 | 0.9268 |
| 0.3813 | 0.03 | 18000 | 0.4227 | 0.9261 |
| 0.3723 | 0.04 | 20000 | 0.3360 | 0.9418 |
| 0.3876 | 0.04 | 22000 | 0.3255 | 0.9407 |
| 0.3351 | 0.04 | 24000 | 0.3283 | 0.9404 |
| 0.34 | 0.05 | 26000 | 0.3489 | 0.9430 |
| 0.3006 | 0.05 | 28000 | 0.3302 | 0.9464 |
| 0.349 | 0.05 | 30000 | 0.3853 | 0.9375 |
| 0.3696 | 0.06 | 32000 | 0.2992 | 0.9454 |
| 0.3301 | 0.06 | 34000 | 0.3484 | 0.9464 |
| 0.3151 | 0.06 | 36000 | 0.3529 | 0.9455 |
| 0.3682 | 0.07 | 38000 | 0.3052 | 0.9420 |
| 0.3184 | 0.07 | 40000 | 0.3323 | 0.9466 |
| 0.3207 | 0.08 | 42000 | 0.3133 | 0.9532 |
| 0.3346 | 0.08 | 44000 | 0.3826 | 0.9414 |
| 0.3008 | 0.08 | 46000 | 0.3059 | 0.9484 |
| 0.3306 | 0.09 | 48000 | 0.3089 | 0.9475 |
| 0.342 | 0.09 | 50000 | 0.3611 | 0.9486 |
| 0.3424 | 0.09 | 52000 | 0.3227 | 0.9445 |
| 0.3044 | 0.1 | 54000 | 0.3130 | 0.9489 |
| 0.3278 | 0.1 | 56000 | 0.3827 | 0.9368 |
| 0.288 | 0.1 | 58000 | 0.3080 | 0.9504 |
| 0.3342 | 0.11 | 60000 | 0.3252 | 0.9471 |
| 0.3737 | 0.11 | 62000 | 0.4250 | 0.9343 |
### Framework versions
- Transformers 4.10.2
- Pytorch 1.7.1
- Datasets 1.6.1
- Tokenizers 0.10.3
|
facebook/s2t-small-mustc-en-pt-st | aa13043d7a25be0e4869075b6fbd44633c1df4c9 | 2022-02-07T15:10:55.000Z | [
"pytorch",
"tf",
"speech_to_text",
"automatic-speech-recognition",
"en",
"pt",
"dataset:mustc",
"arxiv:2010.05171",
"arxiv:1904.08779",
"transformers",
"audio",
"speech-translation",
"license:mit"
]
| automatic-speech-recognition | false | facebook | null | facebook/s2t-small-mustc-en-pt-st | 5 | null | transformers | 16,530 | ---
language:
- en
- pt
datasets:
- mustc
tags:
- audio
- speech-translation
- automatic-speech-recognition
license: mit
pipeline_tag: automatic-speech-recognition
widget:
- example_title: Librispeech sample 1
src: https://cdn-media.huggingface.co/speech_samples/sample1.flac
- example_title: Librispeech sample 2
src: https://cdn-media.huggingface.co/speech_samples/sample2.flac
---
# S2T-SMALL-MUSTC-EN-PT-ST
`s2t-small-mustc-en-pt-st` is a Speech to Text Transformer (S2T) model trained for end-to-end Speech Translation (ST).
The S2T model was proposed in [this paper](https://arxiv.org/abs/2010.05171) and released in
[this repository](https://github.com/pytorch/fairseq/tree/master/examples/speech_to_text)
## Model description
S2T is a transformer-based seq2seq (encoder-decoder) model designed for end-to-end Automatic Speech Recognition (ASR) and Speech
Translation (ST). It uses a convolutional downsampler to reduce the length of speech inputs by 3/4th before they are
fed into the encoder. The model is trained with standard autoregressive cross-entropy loss and generates the
transcripts/translations autoregressively.
## Intended uses & limitations
This model can be used for end-to-end English speech to Portuguese text translation.
See the [model hub](https://huggingface.co/models?filter=speech_to_text) to look for other S2T checkpoints.
### How to use
As this a standard sequence to sequence transformer model, you can use the `generate` method to generate the
transcripts by passing the speech features to the model.
*Note: The `Speech2TextProcessor` object uses [torchaudio](https://github.com/pytorch/audio) to extract the
filter bank features. Make sure to install the `torchaudio` package before running this example.*
You could either install those as extra speech dependancies with
`pip install transformers"[speech, sentencepiece]"` or install the packages seperatly
with `pip install torchaudio sentencepiece`.
```python
import torch
from transformers import Speech2TextProcessor, Speech2TextForConditionalGeneration
from datasets import load_dataset
import soundfile as sf
model = Speech2TextForConditionalGeneration.from_pretrained("facebook/s2t-small-mustc-en-pt-st")
processor = Speech2TextProcessor.from_pretrained("facebook/s2t-small-mustc-en-pt-st")
def map_to_array(batch):
speech, _ = sf.read(batch["file"])
batch["speech"] = speech
return batch
ds = load_dataset(
"patrickvonplaten/librispeech_asr_dummy",
"clean",
split="validation"
)
ds = ds.map(map_to_array)
inputs = processor(
ds["speech"][0],
sampling_rate=16_000,
return_tensors="pt"
)
generated_ids = model.generate(input_ids=inputs["input_features"], attention_mask=inputs["attention_mask"])
translation = processor.batch_decode(generated_ids, skip_special_tokens=True)
```
## Training data
The s2t-small-mustc-en-pt-st is trained on English-Portuguese subset of [MuST-C](https://ict.fbk.eu/must-c/).
MuST-C is a multilingual speech translation corpus whose size and quality facilitates the training of end-to-end systems
for speech translation from English into several languages. For each target language, MuST-C comprises several hundred
hours of audio recordings from English TED Talks, which are automatically aligned at the sentence level with their manual
transcriptions and translations.
## Training procedure
### Preprocessing
The speech data is pre-processed by extracting Kaldi-compliant 80-channel log mel-filter bank features automatically from
WAV/FLAC audio files via PyKaldi or torchaudio. Further utterance-level CMVN (cepstral mean and variance normalization)
is applied to each example.
The texts are lowercased and tokenized using SentencePiece and a vocabulary size of 8,000.
### Training
The model is trained with standard autoregressive cross-entropy loss and using [SpecAugment](https://arxiv.org/abs/1904.08779).
The encoder receives speech features, and the decoder generates the transcripts autoregressively. To accelerate
model training and for better performance the encoder is pre-trained for English ASR.
## Evaluation results
MuST-C test results for en-pt (BLEU score): 28.1
### BibTeX entry and citation info
```bibtex
@inproceedings{wang2020fairseqs2t,
title = {fairseq S2T: Fast Speech-to-Text Modeling with fairseq},
author = {Changhan Wang and Yun Tang and Xutai Ma and Anne Wu and Dmytro Okhonko and Juan Pino},
booktitle = {Proceedings of the 2020 Conference of the Asian Chapter of the Association for Computational Linguistics (AACL): System Demonstrations},
year = {2020},
}
```
|
facebook/wav2vec2-base-10k-voxpopuli-ft-sl | 1a2720ef0da3445608d52fb367a2168c76de994e | 2021-07-06T01:53:05.000Z | [
"pytorch",
"wav2vec2",
"automatic-speech-recognition",
"sl",
"arxiv:2101.00390",
"transformers",
"audio",
"voxpopuli",
"license:cc-by-nc-4.0"
]
| automatic-speech-recognition | false | facebook | null | facebook/wav2vec2-base-10k-voxpopuli-ft-sl | 5 | null | transformers | 16,531 | ---
language: sl
tags:
- audio
- automatic-speech-recognition
- voxpopuli
license: cc-by-nc-4.0
---
# Wav2Vec2-Base-VoxPopuli-Finetuned
[Facebook's Wav2Vec2](https://ai.facebook.com/blog/wav2vec-20-learning-the-structure-of-speech-from-raw-audio/) base model pretrained on the 10K unlabeled subset of [VoxPopuli corpus](https://arxiv.org/abs/2101.00390) and fine-tuned on the transcribed data in sl (refer to Table 1 of paper for more information).
**Paper**: *[VoxPopuli: A Large-Scale Multilingual Speech Corpus for Representation
Learning, Semi-Supervised Learning and Interpretation](https://arxiv.org/abs/2101.00390)*
**Authors**: *Changhan Wang, Morgane Riviere, Ann Lee, Anne Wu, Chaitanya Talnikar, Daniel Haziza, Mary Williamson, Juan Pino, Emmanuel Dupoux* from *Facebook AI*
See the official website for more information, [here](https://github.com/facebookresearch/voxpopuli/)
# Usage for inference
In the following it is shown how the model can be used in inference on a sample of the [Common Voice dataset](https://commonvoice.mozilla.org/en/datasets)
```python
#!/usr/bin/env python3
from transformers import Wav2Vec2Processor, Wav2Vec2ForCTC
from datasets import load_dataset
import torchaudio
import torch
# resample audio
# load model & processor
model = Wav2Vec2ForCTC.from_pretrained("facebook/wav2vec2-base-10k-voxpopuli-ft-sl")
processor = Wav2Vec2Processor.from_pretrained("facebook/wav2vec2-base-10k-voxpopuli-ft-sl")
# load dataset
ds = load_dataset("common_voice", "sl", split="validation[:1%]")
# common voice does not match target sampling rate
common_voice_sample_rate = 48000
target_sample_rate = 16000
resampler = torchaudio.transforms.Resample(common_voice_sample_rate, target_sample_rate)
# define mapping fn to read in sound file and resample
def map_to_array(batch):
speech, _ = torchaudio.load(batch["path"])
speech = resampler(speech)
batch["speech"] = speech[0]
return batch
# load all audio files
ds = ds.map(map_to_array)
# run inference on the first 5 data samples
inputs = processor(ds[:5]["speech"], sampling_rate=target_sample_rate, return_tensors="pt", padding=True)
# inference
logits = model(**inputs).logits
predicted_ids = torch.argmax(logits, axis=-1)
print(processor.batch_decode(predicted_ids))
```
|
facebook/wav2vec2-base-sv-voxpopuli | 7200928b87656cb7a5ff194b8d50faa2bc39c259 | 2021-07-06T01:55:30.000Z | [
"pytorch",
"wav2vec2",
"pretraining",
"sv",
"arxiv:2101.00390",
"transformers",
"audio",
"automatic-speech-recognition",
"voxpopuli",
"license:cc-by-nc-4.0"
]
| automatic-speech-recognition | false | facebook | null | facebook/wav2vec2-base-sv-voxpopuli | 5 | null | transformers | 16,532 | ---
language: sv
tags:
- audio
- automatic-speech-recognition
- voxpopuli
license: cc-by-nc-4.0
---
# Wav2Vec2-Base-VoxPopuli
[Facebook's Wav2Vec2](https://ai.facebook.com/blog/wav2vec-20-learning-the-structure-of-speech-from-raw-audio/) base model pretrained on the sv unlabeled subset of [VoxPopuli corpus](https://arxiv.org/abs/2101.00390).
**Paper**: *[VoxPopuli: A Large-Scale Multilingual Speech Corpus for Representation
Learning, Semi-Supervised Learning and Interpretation](https://arxiv.org/abs/2101.00390)*
**Authors**: *Changhan Wang, Morgane Riviere, Ann Lee, Anne Wu, Chaitanya Talnikar, Daniel Haziza, Mary Williamson, Juan Pino, Emmanuel Dupoux* from *Facebook AI*
See the official website for more information, [here](https://github.com/facebookresearch/voxpopuli/)
# Fine-Tuning
Please refer to [this blog](https://huggingface.co/blog/fine-tune-xlsr-wav2vec2) on how to fine-tune this model on a specific language. Note that you should replace `"facebook/wav2vec2-large-xlsr-53"` with this checkpoint for fine-tuning.
|
fadhilarkan/qa-indo-k | f5d01aa1e86704736ca8fac4092073fd68673589 | 2021-08-22T17:51:15.000Z | [
"pytorch",
"albert",
"question-answering",
"transformers",
"model-index",
"autotrain_compatible"
]
| question-answering | false | fadhilarkan | null | fadhilarkan/qa-indo-k | 5 | null | transformers | 16,533 | ---
model-index:
- name: qa-indo-k
---
<!-- 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. -->
# qa-indo-k
This model was trained from scratch on an unkown dataset.
It achieves the following results on the evaluation set:
- Loss: 2.4984
## 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 |
|:-------------:|:-----:|:-----:|:---------------:|
| 1.2537 | 1.0 | 8209 | 1.9642 |
| 0.943 | 2.0 | 16418 | 2.2143 |
| 0.6694 | 3.0 | 24627 | 2.4984 |
### Framework versions
- Transformers 4.6.1
- Pytorch 1.7.0
- Datasets 1.11.0
- Tokenizers 0.10.3
|
ffsouza/tiny-mbart-length-96-learning_rate-2e-05-weight_decay-0.01-finetuned-en-to-ro | e5efd832a747cc4b8a0dd4a612e654aa8b1bb94f | 2021-11-30T17:39:53.000Z | [
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"dataset:wmt16_en_ro_pre_processed",
"transformers",
"generated_from_trainer",
"model-index",
"autotrain_compatible"
]
| text2text-generation | false | ffsouza | null | ffsouza/tiny-mbart-length-96-learning_rate-2e-05-weight_decay-0.01-finetuned-en-to-ro | 5 | null | transformers | 16,534 | ---
tags:
- generated_from_trainer
datasets:
- wmt16_en_ro_pre_processed
metrics:
- bleu
model-index:
- name: tiny-mbart-length-96-learning_rate-2e-05-weight_decay-0.01-finetuned-en-to-ro
results:
- task:
name: Sequence-to-sequence Language Modeling
type: text2text-generation
dataset:
name: wmt16_en_ro_pre_processed
type: wmt16_en_ro_pre_processed
args: enro
metrics:
- name: Bleu
type: bleu
value: 0.0
---
<!-- 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. -->
# tiny-mbart-length-96-learning_rate-2e-05-weight_decay-0.01-finetuned-en-to-ro
This model is a fine-tuned version of [sshleifer/tiny-mbart](https://huggingface.co/sshleifer/tiny-mbart) on the wmt16_en_ro_pre_processed dataset.
It achieves the following results on the evaluation set:
- Loss: 8.5137
- Bleu: 0.0
- Gen Len: 20.0
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 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: 1
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Bleu | Gen Len |
|:-------------:|:-----:|:-----:|:---------------:|:----:|:-------:|
| 8.2817 | 1.0 | 76290 | 8.5137 | 0.0 | 20.0 |
### Framework versions
- Transformers 4.12.5
- Pytorch 1.10.0+cu102
- Datasets 1.15.1
- Tokenizers 0.10.3
|
flax-community/bert-swahili-news-classification | 4b80bc7a9733f98705620d8ed48b022d6b9940fb | 2021-07-25T11:45:43.000Z | [
"pytorch",
"jax",
"tensorboard",
"bert",
"text-classification",
"sw",
"dataset:flax-community/swahili-safi",
"transformers"
]
| text-classification | false | flax-community | null | flax-community/bert-swahili-news-classification | 5 | null | transformers | 16,535 | ---
language: sw
widget:
- text: "Idris ameandika kwenye ukurasa wake wa Instagram akimkumbusha Diamond kutekeleza ahadi yake kumpigia Zari magoti kumuomba msamaha kama alivyowahi kueleza awali.Idris ameandika;"
datasets:
- flax-community/swahili-safi
---
## Swahili News Classification with BERT
This model was trained using HuggingFace's Flax framework and is part of the [JAX/Flax Community Week](https://discuss.huggingface.co/t/open-to-the-community-community-week-using-jax-flax-for-nlp-cv/7104) organized by [HuggingFace](https://huggingface.co). All training was done on a TPUv3-8 VM sponsored by the Google Cloud team.
This [model](https://huggingface.co/flax-community/bert-base-uncased-swahili) was used as the base and fine-tuned for this task.
## How to use
```python
from transformers import AutoTokenizer, AutoModelForSequenceClassification
tokenizer = AutoTokenizer.from_pretrained("flax-community/bert-swahili-news-classification")
model = AutoModelForSequenceClassification.from_pretrained("flax-community/bert-swahili-news-classification")
```
```
Eval metrics (10% valid set): {'accuracy': 0.9114740008594757}
```
|
flax-community/gpt-2-tamil | 29823bd79b5813f6514d013d8a4c8e488a775389 | 2021-07-18T16:03:33.000Z | [
"pytorch",
"tensorboard",
"gpt2",
"text-generation",
"ta",
"dataset:oscar",
"dataset:IndicNLP",
"transformers"
]
| text-generation | false | flax-community | null | flax-community/gpt-2-tamil | 5 | null | transformers | 16,536 | ---
language: ta
datasets:
- oscar
- IndicNLP
widget:
- text: 'ஒரு ஊரிலே ஒரு காக்கைக்கு'
---
# GPT2-Tamil
This repository is created as part of the Flax/Jax community week by Huggingface. The aim of this project is to pretrain a language model using GPT-2 specifically for Tamil language.
## Setup:
To setup the project, run the following command,
```python
pip install -r requirements.txt
```
## Model:
Pretrained model on Tamil language using a causal language modeling (CLM) objective.
## Dataset Used:
The GTP-2 model is trained on [oscar dataset - ta](https://huggingface.co/datasets/oscar)
## Intended uses & limitations:
You can use the raw model for next sentence prediction, but it's mostly intended to be fine-tuned on a downstream task. See the [model hub](https://huggingface.co/models?filter=gpt) to look for fine-tuned versions on a task that interests you.
## How to pretrain the model:
To perform training, do the following steps,
- Export the model directory (where you want to store the model artifacts like config, tokenizer, etc.)
```python
>>> export MODEL_DIR=<model_dir>
```
- Create the config.json by running the following command,
```python
>>> python src/create_config.py
```
- Create the tokenizer by running the following command,
```python
>>> python src/train_tokenizer.py
```
- Once the config and tokenizer is created, run the following script to start training the flax model
```python
>>> python scripts/train_gpt2-oscar-tamil.sh
```
## How to use:
To perform language generation using the model, pipeline can be used directly.
- First convert the flax model to pytorch using the following command,
```python
python src/convert_flax_to_pytorch.py
```
- Use the following snippet to perform language generation,
```python
>>> from transformers import AutoTokenizer, AutoModelWithLMHead, pipeline
>>> model_name = 'abinayam/gpt-2-tamil'
>>> model = AutoModelWithLMHead.from_pretrained(model_name)
>>> tokenizer = AutoTokenizer.from_pretrained(model_name)
>>> set_seed(42)
>>> input_text = "ஒரு ஊரிலே ஒரு காக்கைக்கு"
>>> max_len = 300
>>> no_seq = 5
>>> generator = pipeline('text-generation', model=model, tokenizer=tokenizer)
>>> sequence = generator(input_text, max_length=max_len, num_return_sequences=no_seq)
```
|
flax-community/gpt-neo-125M-code-search-all | 4c8bde7193f2631b34f269b47e5f662fd5543859 | 2021-07-26T14:07:11.000Z | [
"pytorch",
"jax",
"tensorboard",
"gpt_neo",
"text-generation",
"transformers"
]
| text-generation | false | flax-community | null | flax-community/gpt-neo-125M-code-search-all | 5 | null | transformers | 16,537 | # GPT-Code-Clippy-125M-Code-Search-All
> **Please refer to our new [GitHub Wiki](https://github.com/ncoop57/gpt-code-clippy/wiki) which documents our efforts in detail in creating the open source version of GitHub Copilot**
## Model Description
GPT-CC-125M-Code-Search is a [GPT-Neo-125M model](https://huggingface.co/EleutherAI/gpt-neo-125M) finetuned using causal language modeling on all languages in the [CodeSearchNet Challenge dataset](https://huggingface.co/datasets/code_search_net). This model is specialized to autocomplete methods in multiple programming languages.
## Training data
[CodeSearchNet Challenge dataset](https://huggingface.co/datasets/code_search_net).
## Training procedure
The training script used to train this model can be found [here](https://github.com/ncoop57/gpt-code-clippy/blob/camera-ready/training/run_clm_flax.py).
```bash
./run_clm_flax.py \
--output_dir $HOME/gpt-neo-125M-code-search-all \
--model_name_or_path="EleutherAI/gpt-neo-125M" \
--dataset_name code_search_net \
--dataset_config_name="all" \
--do_train --do_eval \
--block_size="512" \
--per_device_train_batch_size="32" \
--per_device_eval_batch_size="64" \
--preprocessing_num_workers="8" \
--learning_rate="1.2e-4" \
--num_train_epochs 20 \
--warmup_steps 3000 \
--adam_beta1="0.9" \
--adam_beta2="0.95" \
--weight_decay="0.1" \
--overwrite_output_dir \
--logging_steps="25" \
--eval_steps="500" \
--push_to_hub="False" \
--report_to="all" \
--dtype="bfloat16" \
--skip_memory_metrics="True" \
--save_steps="500" \
--save_total_limit 10 \
--report_to="wandb" \
--run_name="gpt-neo-125M-code-search-all"
```
## Intended Use and Limitations
The model is finetuned methods from several languages and is intended to autocomplete methods given some prompt (method signature and docstring).
### How to use
You can use this model directly with a pipeline for text generation. This example generates a different sequence each time it's run:
```py
from transformers import AutoModelForCausalLM, AutoTokenizer, FlaxAutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("flax-community/gpt-neo-125M-code-clippy-code-search-all")
tokenizer = AutoTokenizer.from_pretrained("flax-community/gpt-neo-125M-code-clippy-code-search-all")
prompt = """def greet(name):
'''A function to greet user. Given a user name it should say hello'''
"""
input_ids = tokenizer(prompt, return_tensors='pt').input_ids.to(device)
start = input_ids.size(1)
out = model.generate(input_ids, do_sample=True, max_length=50, num_beams=2,
early_stopping=True, eos_token_id=tokenizer.eos_token_id, )
print(tokenizer.decode(out[0][start:]))
```
### Limitations and Biases
The model is intended to be used for research purposes and comes with no guarantees of quality of generated code.
GPT-CC is finetuned from GPT-Neo and might have inherited biases and limitations from it. See [GPT-Neo model card](https://huggingface.co/EleutherAI/gpt-neo-125M#limitations-and-biases) for details.
## Eval results
Coming soon... |
flax-community/gpt2-swahili | 8ca5ef61a032d60bc0cd82209a084186523cb282 | 2021-07-25T16:22:24.000Z | [
"pytorch",
"jax",
"tensorboard",
"gpt2",
"text-generation",
"sw",
"dataset:flax-community/swahili-safi",
"transformers"
]
| text-generation | false | flax-community | null | flax-community/gpt2-swahili | 5 | null | transformers | 16,538 | ---
language: sw
widget:
- text: "Ninitaka kukula"
datasets:
- flax-community/swahili-safi
---
## GPT2 in Swahili
This model was trained using HuggingFace's Flax framework and is part of the [JAX/Flax Community Week](https://discuss.huggingface.co/t/open-to-the-community-community-week-using-jax-flax-for-nlp-cv/7104) organized by [HuggingFace](https://huggingface.co). All training was done on a TPUv3-8 VM sponsored by the Google Cloud team.
## How to use
```python
from transformers import AutoTokenizer, AutoModelWithLMHead
tokenizer = AutoTokenizer.from_pretrained("flax-community/gpt2-swahili")
model = AutoModelWithLMHead.from_pretrained("flax-community/gpt2-swahili")
print(round((model.num_parameters())/(1000*1000)),"Million Parameters")
124 Million Parameters
```
#### **Training Data**:
This model was trained on [Swahili Safi](https://huggingface.co/datasets/flax-community/swahili-safi)
#### **More Details**:
For more details and Demo please check [HF Swahili Space](https://huggingface.co/spaces/flax-community/Swahili)
|
flax-community/t5-base-dutch | e2af9feca0dbc43ba037a85f2e5c6f8865a258fa | 2022-01-11T12:10:22.000Z | [
"pytorch",
"tf",
"jax",
"tensorboard",
"t5",
"text2text-generation",
"dutch",
"dataset:yhavinga/mc4_nl_cleaned",
"transformers",
"seq2seq",
"lm-head",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | flax-community | null | flax-community/t5-base-dutch | 5 | 2 | transformers | 16,539 | ---
language:
- dutch
tags:
- seq2seq
- lm-head
datasets:
- yhavinga/mc4_nl_cleaned
license: apache-2.0
inference: false
---
# t5-base-dutch
Created by [Yeb Havinga](https://www.linkedin.com/in/yeb-havinga-86530825/)
& [Dat Nguyen](https://www.linkedin.com/in/dat-nguyen-49a641138/) during the [Hugging Face community week](https://discuss.huggingface.co/t/open-to-the-community-community-week-using-jax-flax-for-nlp-cv/7104), organized by [HuggingFace](https://huggingface.co/) and TPU usage sponsored by Google, for the project [Pre-train T5 from scratch in Dutch](https://discuss.huggingface.co/t/pretrain-t5-from-scratch-in-dutch/8109).
See also the fine-tuned [t5-base-dutch-demo](https://huggingface.co/flax-community/t5-base-dutch-demo) model,
and the demo application **[Netherformer 📰](https://huggingface.co/spaces/flax-community/netherformer)**,
that are based on this model.
**5 jan 2022: Model updated. Evaluation accuracy increased from 0.64 to 0.70.**
**11 jan 2022: See also [yhavinga/t5-v1.1-base-dutch-cased](https://huggingface.co/yhavinga/t5-v1.1-base-dutch-cased) with eval acc 0.78**
## Model
* Configuration based on `google/t5-base`
* 12 layers, 12 heads
* Dropout set to 0.1
## Dataset
This model was trained on the `full` configuration of [cleaned Dutch mC4](https://huggingface.co/datasets/yhavinga/mc4_nl_cleaned),
which is the original mC4, except
* Documents that contained words from a selection of the Dutch and English [List of Dirty Naught Obscene and Otherwise Bad Words](https://github.com/LDNOOBW/List-of-Dirty-Naughty-Obscene-and-Otherwise-Bad-Words) are removed
* Sentences with less than 3 words are removed
* Sentences with a word of more than 1000 characters are removed
* Documents with less than 5 sentences are removed
* Documents with "javascript", "lorum ipsum", "terms of use", "privacy policy", "cookie policy", "uses cookies",
"use of cookies", "use cookies", "elementen ontbreken", "deze printversie" are removed.
## Tokenization
A SentencePiece tokenizer was trained from scratch on this dataset.
The total tokens of the `full` configuration is 34B
## Training
The model was trained on the `full` mc4_nl_cleaned dataset configuration for 1 epoch, consisting of 34B tokens,
for 528 482 steps with a batch size of 128 and took 57 hours.
A triangle learning rate schedule was used, with peak learning rate 0.005.
## Evaluation
* Loss: 1.38
* Accuracy: 0.70
|
flax-sentence-embeddings/all_datasets_v3_MiniLM-L6 | 0ba8acf137cc901948b5ffe819d9832d7fd727e5 | 2021-07-23T15:53:06.000Z | [
"pytorch",
"bert",
"feature-extraction",
"en",
"arxiv:2104.08727",
"arxiv:1810.09305",
"arxiv:2102.07033",
"arxiv:1904.06472",
"sentence-transformers",
"sentence-similarity"
]
| sentence-similarity | false | flax-sentence-embeddings | null | flax-sentence-embeddings/all_datasets_v3_MiniLM-L6 | 5 | null | sentence-transformers | 16,540 | ---
pipeline_tag: sentence-similarity
tags:
- sentence-transformers
- feature-extraction
- sentence-similarity
language: en
---
# Model description
The project aims to train sentence embedding models on very large sentence level datasets using a self-supervised
contrastive learning objective. We used the pretrained ['MiniLM-L6-H384-uncased'](https://huggingface.co/nreimers/MiniLM-L6-H384-uncased) model and fine-tuned in on a
1B sentence pairs dataset. We use a contrastive learning objective: given a sentence from the pair, the model should predict which out of a set of randomly sampled other sentences, was actually paired with it in our dataset.
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 intervention from Google’s Flax, JAX, and Cloud team member about efficient deep learning frameworks.
## Intended uses
Our model is intented to be used as a sentence encoder. Given an input sentence, it ouptuts a vector which captures
the sentence semantic information. The sentence vector may be used for information retrieval, 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/all_datasets_v3_MiniLM-L6')
text = "Replace me by any text 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 ['MiniLM-L6-H384-uncased'](https://huggingface.co/nreimers/MiniLM-L6-H384-uncased) which is a 6 layer version of
['microsoft/MiniLM-L12-H384-uncased'](https://huggingface.co/microsoft/MiniLM-L12-H384-uncased) by keeping only every second layer.
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 ou model on a TPU v3-8. We train the model during 540k 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 use the concatenation from multiple datasets to fine-tune our model. The total number of sentence pairs is above 1 billion sentences.
We sampled each dataset given a weighted probability which configuration is detailed in the `data_config.json` file.
| Dataset | Paper | Number of training tuples |
|:--------------------------------------------------------:|:----------------------------------------:|:--------------------------:|
| [GOOAQ: Open Question Answering with Diverse Answer Types](https://github.com/allenai/gooaq) | [paper](https://arxiv.org/pdf/2104.08727.pdf) | 3,012,496 |
| [Stack Exchange](https://huggingface.co/datasets/flax-sentence-embeddings/stackexchange_title_body_jsonl) | - | 364,001 |
| [Flickr 30k](https://shannon.cs.illinois.edu/DenotationGraph/) | [paper](https://transacl.org/ojs/index.php/tacl/article/view/229/33) | 317,695 |
| [COCO 2020](COCO 2020) | [paper](https://link.springer.com/chapter/10.1007%2F978-3-319-10602-1_48) | 828,395|
| [Code Search](https://huggingface.co/datasets/code_search_net) | - | 1,151,414 |
| [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 |
| [Natural Questions (NQ)](https://ai.google.com/research/NaturalQuestions) | [paper](https://transacl.org/ojs/index.php/tacl/article/view/1455) | 100,231 |
| [Simple Wikipedia](https://cs.pomona.edu/~dkauchak/simplification/) | [paper](https://www.aclweb.org/anthology/P11-2117/) | 102,225 |
| [Quora Question Pairs](https://quoradata.quora.com/First-Quora-Dataset-Release-Question-Pairs) | - | 103,663 |
| [Altlex](https://github.com/chridey/altlex/) | [paper](https://aclanthology.org/P16-1135.pdf) | 112,696 |
| [Wikihow](https://github.com/pvl/wikihow_pairs_dataset) | [paper](https://arxiv.org/abs/1810.09305) | 128,542 |
| [Sentence Compression](https://github.com/google-research-datasets/sentence-compression) | [paper](https://www.aclweb.org/anthology/D13-1155/) | 180,000 |
| AllNLI ([SNLI](https://nlp.stanford.edu/projects/snli/) and [MultiNLI](https://cims.nyu.edu/~sbowman/multinli/) | [paper SNLI](https://doi.org/10.18653/v1/d15-1075), [paper MultiNLI](https://doi.org/10.18653/v1/n18-1101) | 277,230 |
| [Eli5](https://huggingface.co/datasets/eli5) | [paper](https://doi.org/10.18653/v1/p19-1346) | 325,475 |
| [SPECTER](https://github.com/allenai/specter) | [paper](https://doi.org/10.18653/v1/2020.acl-main.207) | 684,100 |
| [S2ORC](https://github.com/allenai/s2orc) Title/Abstract | [paper](https://aclanthology.org/2020.acl-main.447/) | 41,769,185 |
| [S2ORC](https://github.com/allenai/s2orc) Citation/Citation | [paper](https://aclanthology.org/2020.acl-main.447/) | 52,603,982 |
| [S2ORC](https://github.com/allenai/s2orc) Citation/Abstract | [paper](https://aclanthology.org/2020.acl-main.447/) | 116,288,806 |
| [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 |
| SearchQA | - | 582,261 |
| [Yahoo Answers](https://www.kaggle.com/soumikrakshit/yahoo-answers-dataset) Title/Answer | [paper](https://proceedings.neurips.cc/paper/2015/hash/250cf8b51c773f3f8dc8b4be867a9a02-Abstract.html) | 1,198,260 |
| [Yahoo Answers](https://www.kaggle.com/soumikrakshit/yahoo-answers-dataset) Title/Question | [paper](https://proceedings.neurips.cc/paper/2015/hash/250cf8b51c773f3f8dc8b4be867a9a02-Abstract.html) | 659,896 |
| [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 |
| [MS MARCO](https://microsoft.github.io/msmarco/) | [paper](https://doi.org/10.1145/3404835.3462804) | 9,144,553 |
| [Reddit conversationnal](https://github.com/PolyAI-LDN/conversational-datasets/tree/master/reddit) | [paper](https://arxiv.org/abs/1904.06472) | 726,484,430 |
| total | | 1,097,953,922 |
|
frizwankhan/tokenization_model | 60b1ccfbe6855149c91b6ef27e4d73a43506c781 | 2022-01-13T13:45:30.000Z | [
"pytorch",
"layoutlmv2",
"token-classification",
"transformers",
"autotrain_compatible"
]
| token-classification | false | frizwankhan | null | frizwankhan/tokenization_model | 5 | null | transformers | 16,541 | Entry not found |
gaochangkuan/model_dir | 2c8dbe003ce27d5b7410879d4f0dccfcc231ecc4 | 2021-05-21T16:10:50.000Z | [
"pytorch",
"jax",
"gpt2",
"text-generation",
"transformers"
]
| text-generation | false | gaochangkuan | null | gaochangkuan/model_dir | 5 | null | transformers | 16,542 | ## Generating Chinese poetry by topic.
```python
from transformers import *
tokenizer = BertTokenizer.from_pretrained("gaochangkuan/model_dir")
model = AutoModelWithLMHead.from_pretrained("gaochangkuan/model_dir")
prompt= '''<s>田园躬耕'''
length= 84
stop_token='</s>'
temperature = 1.2
repetition_penalty=1.3
k= 30
p= 0.95
device ='cuda'
seed=2020
no_cuda=False
prompt_text = prompt if prompt else input("Model prompt >>> ")
encoded_prompt = tokenizer.encode(
'<s>'+prompt_text+'<sep>',
add_special_tokens=False,
return_tensors="pt"
)
encoded_prompt = encoded_prompt.to(device)
output_sequences = model.generate(
input_ids=encoded_prompt,
max_length=length,
min_length=10,
do_sample=True,
early_stopping=True,
num_beams=10,
temperature=temperature,
top_k=k,
top_p=p,
repetition_penalty=repetition_penalty,
bad_words_ids=None,
bos_token_id=tokenizer.bos_token_id,
pad_token_id=tokenizer.pad_token_id,
eos_token_id=tokenizer.eos_token_id,
length_penalty=1.2,
no_repeat_ngram_size=2,
num_return_sequences=1,
attention_mask=None,
decoder_start_token_id=tokenizer.bos_token_id,)
generated_sequence = output_sequences[0].tolist()
text = tokenizer.decode(generated_sequence)
text = text[: text.find(stop_token) if stop_token else None]
print(''.join(text).replace(' ','').replace('<pad>','').replace('<s>',''))
```
|
gchhablani/bert-large-cased-finetuned-cola | 7c325c74302ec1f4bccdee38c236f62db1b5f1b1 | 2021-09-21T04:06:19.000Z | [
"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-cola | 5 | null | transformers | 16,543 | ---
language:
- en
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- glue
metrics:
- matthews_correlation
model-index:
- name: bert-large-cased-finetuned-cola
results:
- task:
name: Text Classification
type: text-classification
dataset:
name: GLUE COLA
type: glue
args: cola
metrics:
- name: Matthews Correlation
type: matthews_correlation
value: 0.5957317644481708
---
<!-- 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-cola
This model is a fine-tuned version of [bert-large-cased](https://huggingface.co/bert-large-cased) on the GLUE COLA dataset.
It achieves the following results on the evaluation set:
- Loss: 0.8385
- Matthews Correlation: 0.5957
## 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 | Matthews Correlation |
|:-------------:|:-----:|:----:|:---------------:|:--------------------:|
| 0.5533 | 1.0 | 2138 | 0.7943 | 0.4439 |
| 0.5004 | 2.0 | 4276 | 0.7272 | 0.5678 |
| 0.2865 | 3.0 | 6414 | 0.8385 | 0.5957 |
### Framework versions
- Transformers 4.11.0.dev0
- Pytorch 1.9.0
- Datasets 1.12.1
- Tokenizers 0.10.3
|
gchhablani/bert-large-cased-finetuned-wnli | 72dc38be4edaa48d424538b7a49726588202a634 | 2021-09-23T05:10:44.000Z | [
"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-wnli | 5 | null | transformers | 16,544 | ---
language:
- en
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- glue
metrics:
- accuracy
model-index:
- name: bert-large-cased-finetuned-wnli
results:
- task:
name: Text Classification
type: text-classification
dataset:
name: GLUE WNLI
type: glue
args: wnli
metrics:
- name: Accuracy
type: accuracy
value: 0.352112676056338
---
<!-- 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-wnli
This model is a fine-tuned version of [bert-large-cased](https://huggingface.co/bert-large-cased) on the GLUE WNLI dataset.
It achieves the following results on the evaluation set:
- Loss: 0.7087
- Accuracy: 0.3521
## 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 | Accuracy | Validation Loss |
|:-------------:|:-----:|:----:|:--------:|:---------------:|
| 0.7114 | 1.0 | 159 | 0.5634 | 0.6923 |
| 0.7141 | 2.0 | 318 | 0.5634 | 0.6895 |
| 0.7063 | 3.0 | 477 | 0.5634 | 0.6930 |
| 0.712 | 4.0 | 636 | 0.4507 | 0.7077 |
| 0.7037 | 5.0 | 795 | 0.3521 | 0.7087 |
### Framework versions
- Transformers 4.11.0.dev0
- Pytorch 1.9.0
- Datasets 1.12.1
- Tokenizers 0.10.3
|
gchhablani/fnet-large-finetuned-cola-copy | d3cb3602806822b4ac2a6fa5f1c2ad5445547001 | 2021-10-10T05:39:18.000Z | [
"pytorch",
"tensorboard",
"fnet",
"text-classification",
"en",
"dataset:glue",
"transformers",
"generated_from_trainer",
"license:apache-2.0",
"model-index"
]
| text-classification | false | gchhablani | null | gchhablani/fnet-large-finetuned-cola-copy | 5 | null | transformers | 16,545 | ---
language:
- en
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- glue
metrics:
- matthews_correlation
model-index:
- name: fnet-large-finetuned-cola-copy
results:
- task:
name: Text Classification
type: text-classification
dataset:
name: GLUE COLA
type: glue
args: cola
metrics:
- name: Matthews Correlation
type: matthews_correlation
value: 0.0
---
<!-- 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-large-finetuned-cola-copy
This model is a fine-tuned version of [google/fnet-large](https://huggingface.co/google/fnet-large) on the GLUE COLA dataset.
It achieves the following results on the evaluation set:
- Loss: 0.6243
- Matthews Correlation: 0.0
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 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 | Matthews Correlation |
|:-------------:|:-----:|:----:|:---------------:|:--------------------:|
| 0.6195 | 1.0 | 2138 | 0.6527 | 0.0 |
| 0.6168 | 2.0 | 4276 | 0.6259 | 0.0 |
| 0.616 | 3.0 | 6414 | 0.6243 | 0.0 |
### Framework versions
- Transformers 4.11.0.dev0
- Pytorch 1.9.0
- Datasets 1.12.1
- Tokenizers 0.10.3
|
gchhablani/fnet-large-finetuned-cola-copy2 | c032b807f6ad48caac37774d782ea3928a3bc882 | 2021-10-10T07:23:36.000Z | [
"pytorch",
"tensorboard",
"fnet",
"text-classification",
"en",
"dataset:glue",
"transformers",
"generated_from_trainer",
"license:apache-2.0",
"model-index"
]
| text-classification | false | gchhablani | null | gchhablani/fnet-large-finetuned-cola-copy2 | 5 | null | transformers | 16,546 | ---
language:
- en
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- glue
metrics:
- matthews_correlation
model-index:
- name: fnet-large-finetuned-cola-copy2
results:
- task:
name: Text Classification
type: text-classification
dataset:
name: GLUE COLA
type: glue
args: cola
metrics:
- name: Matthews Correlation
type: matthews_correlation
value: 0.0
---
<!-- 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-large-finetuned-cola-copy2
This model is a fine-tuned version of [google/fnet-large](https://huggingface.co/google/fnet-large) on the GLUE COLA dataset.
It achieves the following results on the evaluation set:
- Loss: 0.6173
- Matthews Correlation: 0.0
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 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
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 3.0
### Training results
| Training Loss | Epoch | Step | Validation Loss | Matthews Correlation |
|:-------------:|:-----:|:----:|:---------------:|:--------------------:|
| 0.6192 | 1.0 | 2138 | 0.6443 | 0.0 |
| 0.6177 | 2.0 | 4276 | 0.6296 | 0.0 |
| 0.6128 | 3.0 | 6414 | 0.6173 | 0.0 |
### Framework versions
- Transformers 4.11.0.dev0
- Pytorch 1.9.0
- Datasets 1.12.1
- Tokenizers 0.10.3
|
gchhablani/fnet-large-finetuned-cola-copy4 | cdee9e3a7233456950febba1dfe27f8a5ff0db27 | 2021-10-10T19:30:36.000Z | [
"pytorch",
"tensorboard",
"fnet",
"text-classification",
"en",
"dataset:glue",
"transformers",
"generated_from_trainer",
"license:apache-2.0",
"model-index"
]
| text-classification | false | gchhablani | null | gchhablani/fnet-large-finetuned-cola-copy4 | 5 | null | transformers | 16,547 | ---
language:
- en
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- glue
metrics:
- matthews_correlation
model-index:
- name: fnet-large-finetuned-cola-copy4
results:
- task:
name: Text Classification
type: text-classification
dataset:
name: GLUE COLA
type: glue
args: cola
metrics:
- name: Matthews Correlation
type: matthews_correlation
value: 0.0
---
<!-- 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-large-finetuned-cola-copy4
This model is a fine-tuned version of [google/fnet-large](https://huggingface.co/google/fnet-large) on the GLUE COLA dataset.
It achieves the following results on the evaluation set:
- Loss: 0.6500
- Matthews Correlation: 0.0
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 4e-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: polynomial
- num_epochs: 3.0
### Training results
| Training Loss | Epoch | Step | Validation Loss | Matthews Correlation |
|:-------------:|:-----:|:----:|:---------------:|:--------------------:|
| 0.6345 | 1.0 | 2138 | 0.6611 | 0.0 |
| 0.6359 | 2.0 | 4276 | 0.6840 | 0.0 |
| 0.6331 | 3.0 | 6414 | 0.6500 | 0.0 |
### Framework versions
- Transformers 4.11.0.dev0
- Pytorch 1.9.0
- Datasets 1.12.1
- Tokenizers 0.10.3
|
gchhablani/fnet-large-finetuned-mrpc | bc05e37c8f1e0d5235d9522f854d9be9827bc133 | 2021-09-22T09:06:01.000Z | [
"pytorch",
"tensorboard",
"fnet",
"text-classification",
"en",
"dataset:glue",
"transformers",
"generated_from_trainer",
"license:apache-2.0",
"model-index"
]
| text-classification | false | gchhablani | null | gchhablani/fnet-large-finetuned-mrpc | 5 | null | transformers | 16,548 | ---
language:
- en
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- glue
metrics:
- accuracy
- f1
model-index:
- name: fnet-large-finetuned-mrpc
results:
- task:
name: Text Classification
type: text-classification
dataset:
name: GLUE MRPC
type: glue
args: mrpc
metrics:
- name: Accuracy
type: accuracy
value: 0.8259803921568627
- name: F1
type: f1
value: 0.8798646362098139
---
<!-- 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-large-finetuned-mrpc
This model is a fine-tuned version of [google/fnet-large](https://huggingface.co/google/fnet-large) on the GLUE MRPC dataset.
It achieves the following results on the evaluation set:
- Loss: 1.0872
- Accuracy: 0.8260
- F1: 0.8799
- Combined Score: 0.8529
## 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 |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:------:|:--------------:|
| 0.5656 | 1.0 | 917 | 0.6999 | 0.7843 | 0.8581 | 0.8212 |
| 0.3874 | 2.0 | 1834 | 0.7280 | 0.8088 | 0.8691 | 0.8390 |
| 0.1627 | 3.0 | 2751 | 1.1274 | 0.8162 | 0.8780 | 0.8471 |
| 0.0751 | 4.0 | 3668 | 1.0289 | 0.8333 | 0.8870 | 0.8602 |
| 0.0339 | 5.0 | 4585 | 1.0872 | 0.8260 | 0.8799 | 0.8529 |
### Framework versions
- Transformers 4.11.0.dev0
- Pytorch 1.9.0
- Datasets 1.12.1
- Tokenizers 0.10.3
|
georeactor/arabic-flip-mt5 | 3c599f6aa354b1ccc6f721965736931d2591fe05 | 2022-02-13T03:49:09.000Z | [
"pytorch",
"mt5",
"text2text-generation",
"ar",
"transformers",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | georeactor | null | georeactor/arabic-flip-mt5 | 5 | 1 | transformers | 16,549 | ---
language: ar
license: apache-2.0
---
# Arabic-Flip-mT5
An [mT5-small](https://huggingface.co/google/mt5-small) model fine-tuned on the
Arabic Parallel Gender Corpus v2.0. You should be able to do gender inflection
(M->F or F->M) in Arabic.
Uses: Data augmentation, detecting gender bias of a model, obfuscating gender of a user/customer.
## Training Notebook
https://colab.research.google.com/drive/1w3XY9yClfDo1T-1mc19W39i2aceK2d1D?usp=sharing
## Caveats
- The Parallel Gender Corpus is trained on subtitles with mostly first- and second-person sentences, so it is less likely to handle third-person references or longer texts.
- Used mT5-small and batch size of 4, due to CoLab GPU constraints; should ++
- Ran 1 epoch due to an issue with CoLab
|
giacomomiolo/electramed_base_scivocab_970k | da58fcc1fb4dd7a261afd616663db722a06801d0 | 2020-10-02T14:55:30.000Z | [
"pytorch",
"tf",
"electra",
"pretraining",
"transformers"
]
| null | false | giacomomiolo | null | giacomomiolo/electramed_base_scivocab_970k | 5 | null | transformers | 16,550 | Entry not found |
giganticode/bert-base-StackOverflow-comments_2M | 7c339826e2dfb9fc79c528c262b99ae51ef87797 | 2021-10-25T13:01:17.000Z | [
"pytorch",
"bert",
"fill-mask",
"transformers",
"autotrain_compatible"
]
| fill-mask | false | giganticode | null | giganticode/bert-base-StackOverflow-comments_2M | 5 | null | transformers | 16,551 | Entry not found |
glasses/efficientnet_b3 | bc3320c99442951b700a7a211e732fd20ad645c1 | 2021-12-01T08:08:37.000Z | [
"pytorch",
"arxiv:1905.11946",
"transformers"
]
| null | false | glasses | null | glasses/efficientnet_b3 | 5 | null | transformers | 16,552 | # efficientnet_b3
Implementation of EfficientNet proposed in [EfficientNet: Rethinking
Model Scaling for Convolutional Neural
Networks](https://arxiv.org/abs/1905.11946)
The basic architecture is similar to MobileNetV2 as was computed by
using [Progressive Neural Architecture
Search](https://arxiv.org/abs/1905.11946) .
The following table shows the basic architecture
(EfficientNet-efficientnet\_b0):
Then, the architecture is scaled up from
[-efficientnet\_b0]{.title-ref} to [-efficientnet\_b7]{.title-ref}
using compound scaling.
``` python
EfficientNet.efficientnet_b0()
EfficientNet.efficientnet_b1()
EfficientNet.efficientnet_b2()
EfficientNet.efficientnet_b3()
EfficientNet.efficientnet_b4()
EfficientNet.efficientnet_b5()
EfficientNet.efficientnet_b6()
EfficientNet.efficientnet_b7()
EfficientNet.efficientnet_b8()
EfficientNet.efficientnet_l2()
```
Examples:
``` python
EfficientNet.efficientnet_b0(activation = nn.SELU)
# change number of classes (default is 1000 )
EfficientNet.efficientnet_b0(n_classes=100)
# pass a different block
EfficientNet.efficientnet_b0(block=...)
# store each feature
x = torch.rand((1, 3, 224, 224))
model = EfficientNet.efficientnet_b0()
# 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, 32, 112, 112]), torch.Size([1, 24, 56, 56]), torch.Size([1, 40, 28, 28]), torch.Size([1, 80, 14, 14])]
```
|
glasses/efficientnet_b6 | 4a035e9ea60d520f14e70fa9b2169fd4cf96d896 | 2021-04-22T18:00:19.000Z | [
"pytorch",
"arxiv:1905.11946",
"transformers"
]
| null | false | glasses | null | glasses/efficientnet_b6 | 5 | 1 | transformers | 16,553 | # efficientnet_b6
Implementation of EfficientNet proposed in [EfficientNet: Rethinking
Model Scaling for Convolutional Neural
Networks](https://arxiv.org/abs/1905.11946)
The basic architecture is similar to MobileNetV2 as was computed by
using [Progressive Neural Architecture
Search](https://arxiv.org/abs/1905.11946) .
The following table shows the basic architecture
(EfficientNet-efficientnet\_b0):
Then, the architecture is scaled up from
[-efficientnet\_b0]{.title-ref} to [-efficientnet\_b7]{.title-ref}
using compound scaling.
``` python
EfficientNet.efficientnet_b0()
EfficientNet.efficientnet_b1()
EfficientNet.efficientnet_b2()
EfficientNet.efficientnet_b3()
EfficientNet.efficientnet_b4()
EfficientNet.efficientnet_b5()
EfficientNet.efficientnet_b6()
EfficientNet.efficientnet_b7()
EfficientNet.efficientnet_b8()
EfficientNet.efficientnet_l2()
```
Examples:
``` python
EfficientNet.efficientnet_b0(activation = nn.SELU)
# change number of classes (default is 1000 )
EfficientNet.efficientnet_b0(n_classes=100)
# pass a different block
EfficientNet.efficientnet_b0(block=...)
# store each feature
x = torch.rand((1, 3, 224, 224))
model = EfficientNet.efficientnet_b0()
# 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, 32, 112, 112]), torch.Size([1, 24, 56, 56]), torch.Size([1, 40, 28, 28]), torch.Size([1, 80, 14, 14])]
```
|
gniemiec/t5-small-finetuned-xsum | d95aa2d2baf050760dc52799a34dad5e04eaecd9 | 2021-09-20T11:36:55.000Z | [
"pytorch",
"tensorboard",
"t5",
"text2text-generation",
"dataset:xsum",
"transformers",
"generated_from_trainer",
"license:apache-2.0",
"model-index",
"autotrain_compatible"
]
| text2text-generation | false | gniemiec | null | gniemiec/t5-small-finetuned-xsum | 5 | null | transformers | 16,554 | ---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- xsum
metrics:
- rouge
model-index:
- name: t5-small-finetuned-xsum
results:
- task:
name: Sequence-to-sequence Language Modeling
type: text2text-generation
dataset:
name: xsum
type: xsum
args: default
metrics:
- name: Rouge1
type: rouge
value: 23.0533
---
<!-- 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. -->
# t5-small-finetuned-xsum
This model is a fine-tuned version of [t5-small](https://huggingface.co/t5-small) on the xsum dataset.
It achieves the following results on the evaluation set:
- Loss: 2.7967
- Rouge1: 23.0533
- Rouge2: 3.912
- Rougel: 17.8534
- Rougelsum: 17.8581
- Gen Len: 18.6878
## 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: 1
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|:-------------:|:-----:|:----:|:---------------:|:-------:|:------:|:-------:|:---------:|:-------:|
| 3.0574 | 1.0 | 1276 | 2.7967 | 23.0533 | 3.912 | 17.8534 | 17.8581 | 18.6878 |
### Framework versions
- Transformers 4.10.2
- Pytorch 1.9.0+cu102
- Datasets 1.12.1
- Tokenizers 0.10.3
|
Language-Media-Lab/byt5-small-ain-jpn-mt | d59ba14e3b66aa1a1c0c289212228a86fbe20efa | 2022-02-04T13:03:14.000Z | [
"pytorch",
"t5",
"text2text-generation",
"ain",
"ja",
"transformers",
"translation",
"autotrain_compatible"
]
| translation | false | Language-Media-Lab | null | Language-Media-Lab/byt5-small-ain-jpn-mt | 5 | 1 | transformers | 16,555 | ---
language:
- ain
- ja
tags:
- translation
---
Byt5-small-ain-jpn-mt is a machine translation model pretrained with [Google's ByT5-small](https://huggingface.co/google/byt5-small) and fine-tuned on bilingual datasets crawled from the Web. It translates Ainu language to Japanese.
|
google/multiberts-seed_4-step_120k | 6fdecff287112578214b3f0d9ad37bf395c39378 | 2021-11-06T03:12:26.000Z | [
"pytorch",
"tf",
"bert",
"pretraining",
"en",
"arxiv:2106.16163",
"arxiv:1908.08962",
"transformers",
"multiberts",
"multiberts-seed_4",
"multiberts-seed_4-step_120k",
"license:apache-2.0"
]
| null | false | google | null | google/multiberts-seed_4-step_120k | 5 | null | transformers | 16,556 | ---
language: en
tags:
- multiberts
- multiberts-seed_4
- multiberts-seed_4-step_120k
license: apache-2.0
---
# MultiBERTs, Intermediate Checkpoint - Seed 4, Step 120k
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 120k (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_120k')
model = TFBertModel.from_pretrained("google/multiberts-seed_4-step_120k")
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_120k')
model = BertModel.from_pretrained("google/multiberts-seed_4-step_120k")
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/multiberts-seed_4-step_1700k | 450fb06c819d6e53d826386ad939febe49ef7d29 | 2021-11-06T03:45:49.000Z | [
"pytorch",
"tf",
"bert",
"pretraining",
"en",
"arxiv:2106.16163",
"arxiv:1908.08962",
"transformers",
"multiberts",
"multiberts-seed_4",
"multiberts-seed_4-step_1700k",
"license:apache-2.0"
]
| null | false | google | null | google/multiberts-seed_4-step_1700k | 5 | null | transformers | 16,557 | ---
language: en
tags:
- multiberts
- multiberts-seed_4
- multiberts-seed_4-step_1700k
license: apache-2.0
---
# MultiBERTs, Intermediate Checkpoint - Seed 4, Step 1700k
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 1700k (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_1700k')
model = TFBertModel.from_pretrained("google/multiberts-seed_4-step_1700k")
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_1700k')
model = BertModel.from_pretrained("google/multiberts-seed_4-step_1700k")
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/multiberts-seed_4-step_300k | 0c62193e614ff5e171c9b48e6ebe4b7b5113376a | 2021-11-06T03:21:36.000Z | [
"pytorch",
"tf",
"bert",
"pretraining",
"en",
"arxiv:2106.16163",
"arxiv:1908.08962",
"transformers",
"multiberts",
"multiberts-seed_4",
"multiberts-seed_4-step_300k",
"license:apache-2.0"
]
| null | false | google | null | google/multiberts-seed_4-step_300k | 5 | null | transformers | 16,558 | ---
language: en
tags:
- multiberts
- multiberts-seed_4
- multiberts-seed_4-step_300k
license: apache-2.0
---
# MultiBERTs, Intermediate Checkpoint - Seed 4, Step 300k
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 300k (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_300k')
model = TFBertModel.from_pretrained("google/multiberts-seed_4-step_300k")
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_300k')
model = BertModel.from_pretrained("google/multiberts-seed_4-step_300k")
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-dm1000 | ed74bee9cb525ee951759e47818bd21690d7fc39 | 2022-02-15T10:52:15.000Z | [
"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-dm1000 | 5 | null | transformers | 16,559 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-BASE-DM1000 (Deep-Narrow version)
T5-Efficient-BASE-DM1000 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-dm1000** - is of model type **Base** with the following variations:
- **dm** is **1000**
It has **297.23** million parameters and thus requires *ca.* **1188.93 MB** of memory in full precision (*fp32*)
or **594.47 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-el16 | 01129ccde7c25c2d07b347ed4ccca16779a3d7cc | 2022-02-15T10:52:28.000Z | [
"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-base-el16 | 5 | null | transformers | 16,560 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-BASE-EL16 (Deep-Narrow version)
T5-Efficient-BASE-EL16 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-el16** - is of model type **Base** with the following variations:
- **el** is **16**
It has **251.25** million parameters and thus requires *ca.* **1005.01 MB** of memory in full precision (*fp32*)
or **502.51 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-el4 | 22a92789407de515e7edc287d9763ebedac2cef2 | 2022-02-15T10:49:47.000Z | [
"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-el4 | 5 | null | transformers | 16,561 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-BASE-EL4 (Deep-Narrow version)
T5-Efficient-BASE-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-base-el4** - is of model type **Base** with the following variations:
- **el** is **4**
It has **166.29** million parameters and thus requires *ca.* **665.16 MB** of memory in full precision (*fp32*)
or **332.58 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-el8 | f7c73f11984ca97b352c6ae58c3ef6336b30fc91 | 2022-02-15T10:52:37.000Z | [
"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-base-el8 | 5 | null | transformers | 16,562 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-BASE-EL8 (Deep-Narrow version)
T5-Efficient-BASE-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-base-el8** - is of model type **Base** with the following variations:
- **el** is **8**
It has **194.61** million parameters and thus requires *ca.* **778.44 MB** of memory in full precision (*fp32*)
or **389.22 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-ff12000 | 0e14ead8554de3b21a360ad144e8bffa35e6439f | 2022-02-15T10:52:43.000Z | [
"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-ff12000 | 5 | null | transformers | 16,563 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-BASE-FF12000 (Deep-Narrow version)
T5-Efficient-BASE-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-base-ff12000** - is of model type **Base** with the following variations:
- **ff** is **12000**
It has **562.67** million parameters and thus requires *ca.* **2250.68 MB** of memory in full precision (*fp32*)
or **1125.34 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-ff6000 | 7f4b6a0ce97457da5768a5c6d5d07bc1a57b14c6 | 2022-02-15T10:52:49.000Z | [
"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-base-ff6000 | 5 | null | transformers | 16,564 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-BASE-FF6000 (Deep-Narrow version)
T5-Efficient-BASE-FF6000 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-ff6000** - is of model type **Base** with the following variations:
- **ff** is **6000**
It has **336.18** million parameters and thus requires *ca.* **1344.71 MB** of memory in full precision (*fp32*)
or **672.36 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-ff9000 | bef2484838e3bb0bf8fcfaf6c3f5d7d8b3412e04 | 2022-02-15T10:52:53.000Z | [
"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-ff9000 | 5 | null | transformers | 16,565 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-BASE-FF9000 (Deep-Narrow version)
T5-Efficient-BASE-FF9000 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-ff9000** - is of model type **Base** with the following variations:
- **ff** is **9000**
It has **449.42** million parameters and thus requires *ca.* **1797.7 MB** of memory in full precision (*fp32*)
or **898.85 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-kv128 | 01c4ad3d9ef155eaa7a51011e4045eb586165ea8 | 2022-02-15T10:52:56.000Z | [
"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-base-kv128 | 5 | null | transformers | 16,566 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-BASE-KV128 (Deep-Narrow version)
T5-Efficient-BASE-KV128 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-kv128** - is of model type **Base** with the following variations:
- **kv** is **128**
It has **307.87** million parameters and thus requires *ca.* **1231.47 MB** of memory in full precision (*fp32*)
or **615.73 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-nh8 | b1bfc702f8fa666aa8c67da43bb4b9b7fee4a49d | 2022-02-15T10:53:18.000Z | [
"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-nh8 | 5 | null | transformers | 16,567 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-BASE-NH8 (Deep-Narrow version)
T5-Efficient-BASE-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-base-nh8** - is of model type **Base** with the following variations:
- **nh** is **8**
It has **194.62** million parameters and thus requires *ca.* **778.48 MB** of memory in full precision (*fp32*)
or **389.24 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-nl4 | 971c3fa411256b8dc8dd1bf604a14f058a13d47a | 2022-02-15T10:57:25.000Z | [
"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-nl4 | 5 | null | transformers | 16,568 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-BASE-NL4 (Deep-Narrow version)
T5-Efficient-BASE-NL4 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-nl4** - is of model type **Base** with the following variations:
- **nl** is **4**
It has **90.76** million parameters and thus requires *ca.* **363.05 MB** of memory in full precision (*fp32*)
or **181.52 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-nl48 | 9bdd6246cd337961ff504ab7a6f5372bab984d48 | 2022-02-15T10:49:50.000Z | [
"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-nl48 | 5 | null | transformers | 16,569 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-BASE-NL48 (Deep-Narrow version)
T5-Efficient-BASE-NL48 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-nl48** - is of model type **Base** with the following variations:
- **nl** is **48**
It has **817.7** million parameters and thus requires *ca.* **3270.79 MB** of memory in full precision (*fp32*)
or **1635.39 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-dm128 | cb1748cfc350cf0d3a80589365e0d21ac42f11ab | 2022-02-15T10:53:40.000Z | [
"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-large-dm128 | 5 | null | transformers | 16,570 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-LARGE-DM128 (Deep-Narrow version)
T5-Efficient-LARGE-DM128 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-dm128** - is of model type **Large** with the following variations:
- **dm** is **128**
It has **92.27** million parameters and thus requires *ca.* **369.06 MB** of memory in full precision (*fp32*)
or **184.53 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-dm768 | cb60bfad8f28d439aded205a602c6dc3fb781465 | 2022-02-15T10:53:46.000Z | [
"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-dm768 | 5 | null | transformers | 16,571 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-LARGE-DM768 (Deep-Narrow version)
T5-Efficient-LARGE-DM768 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-dm768** - is of model type **Large** with the following variations:
- **dm** is **768**
It has **553.31** million parameters and thus requires *ca.* **2213.23 MB** of memory in full precision (*fp32*)
or **1106.62 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-nh12 | f162a164b345a4e1cc40e46f0ef910fc1069b527 | 2022-02-15T10:55:25.000Z | [
"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-nh12 | 5 | null | transformers | 16,572 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-LARGE-NH12 (Deep-Narrow version)
T5-Efficient-LARGE-NH12 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-nh12** - is of model type **Large** with the following variations:
- **nh** is **12**
It has **662.23** million parameters and thus requires *ca.* **2648.91 MB** of memory in full precision (*fp32*)
or **1324.45 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-nl2 | 5bdfcdec9e53a071ea94f4ac9851b883d16ebca8 | 2022-02-15T10:55:51.000Z | [
"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-nl2 | 5 | null | transformers | 16,573 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-LARGE-NL2 (Deep-Narrow version)
T5-Efficient-LARGE-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-large-nl2** - is of model type **Large** with the following variations:
- **nl** is **2**
It has **91.64** million parameters and thus requires *ca.* **366.55 MB** of memory in full precision (*fp32*)
or **183.28 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-nl20 | e7cc889984911658550b777d685d6c61f8969056 | 2022-02-15T10:55:54.000Z | [
"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-nl20 | 5 | null | transformers | 16,574 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-LARGE-NL20 (Deep-Narrow version)
T5-Efficient-LARGE-NL20 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-nl20** - is of model type **Large** with the following variations:
- **nl** is **20**
It has **620.25** million parameters and thus requires *ca.* **2481.02 MB** of memory in full precision (*fp32*)
or **1240.51 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-nl12 | 28971abac69bbff36a31d20354193895cb32c73b | 2022-02-15T10:56:07.000Z | [
"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-nl12 | 5 | null | transformers | 16,575 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-MINI-NL12 (Deep-Narrow version)
T5-Efficient-MINI-NL12 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-nl12** - is of model type **Mini** with the following variations:
- **nl** is **12**
It has **69.01** million parameters and thus requires *ca.* **276.05 MB** of memory in full precision (*fp32*)
or **138.03 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-ff1000 | c16004fc9d6029e89eff639a9f98dc342cb8bd96 | 2022-02-15T10:50:15.000Z | [
"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-ff1000 | 5 | null | transformers | 16,576 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-SMALL-FF1000 (Deep-Narrow version)
T5-Efficient-SMALL-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-small-ff1000** - is of model type **Small** with the following variations:
- **ff** is **1000**
It has **47.94** million parameters and thus requires *ca.* **191.75 MB** of memory in full precision (*fp32*)
or **95.88 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-ff3000 | 594bf2eea54fcb2a662494bc8dd3073ee6a2bf83 | 2022-02-15T10:50:21.000Z | [
"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-ff3000 | 5 | null | transformers | 16,577 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-SMALL-FF3000 (Deep-Narrow version)
T5-Efficient-SMALL-FF3000 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-ff3000** - is of model type **Small** with the following variations:
- **ff** is **3000**
It has **73.1** million parameters and thus requires *ca.* **292.42 MB** of memory in full precision (*fp32*)
or **146.21 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-ff6000 | 212322b5d903ae54acc02a7d2f64b8668db2471e | 2022-02-15T10:50:24.000Z | [
"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-small-ff6000 | 5 | null | transformers | 16,578 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-SMALL-FF6000 (Deep-Narrow version)
T5-Efficient-SMALL-FF6000 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-ff6000** - is of model type **Small** with the following variations:
- **ff** is **6000**
It has **110.85** million parameters and thus requires *ca.* **443.41 MB** of memory in full precision (*fp32*)
or **221.71 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-small-kv128 | 4b36f8cb76545085ef03809c1b7cb920a0bff8d3 | 2022-02-15T10:50:30.000Z | [
"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-kv128 | 5 | null | transformers | 16,579 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-SMALL-KV128 (Deep-Narrow version)
T5-Efficient-SMALL-KV128 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-kv128** - is of model type **Small** with the following variations:
- **kv** is **128**
It has **79.4** million parameters and thus requires *ca.* **317.58 MB** of memory in full precision (*fp32*)
or **158.79 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-el2 | 48f3b863b85d1b76693734085e5223c38798f899 | 2022-02-15T10:54:27.000Z | [
"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-el2 | 5 | null | transformers | 16,580 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-TINY-EL2 (Deep-Narrow version)
T5-Efficient-TINY-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-tiny-el2** - is of model type **Tiny** with the following variations:
- **el** is **2**
It has **22.41** million parameters and thus requires *ca.* **89.64 MB** of memory in full precision (*fp32*)
or **44.82 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-el8 | 7ea9e69a0a57c0f9bec6ed97737bf3b960cdd66d | 2022-02-15T10:51:15.000Z | [
"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-el8 | 5 | null | transformers | 16,581 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-TINY-EL8 (Deep-Narrow version)
T5-Efficient-TINY-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-tiny-el8** - is of model type **Tiny** with the following variations:
- **el** is **8**
It has **27.14** million parameters and thus requires *ca.* **108.55 MB** of memory in full precision (*fp32*)
or **54.28 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-ff12000 | eb6c4f28a81ec0bcaa1e6291114f8ee714374832 | 2022-02-15T10:51:18.000Z | [
"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-tiny-ff12000 | 5 | null | transformers | 16,582 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-TINY-FF12000 (Deep-Narrow version)
T5-Efficient-TINY-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-tiny-ff12000** - is of model type **Tiny** with the following variations:
- **ff** is **12000**
It has **61.72** million parameters and thus requires *ca.* **246.87 MB** of memory in full precision (*fp32*)
or **123.44 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-tiny-ff6000 | 8739514cfd39c08a0be3b0d4cdcb5c6ca3105c76 | 2022-02-15T10:51:21.000Z | [
"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-ff6000 | 5 | null | transformers | 16,583 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-TINY-FF6000 (Deep-Narrow version)
T5-Efficient-TINY-FF6000 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-ff6000** - is of model type **Tiny** with the following variations:
- **ff** is **6000**
It has **36.55** million parameters and thus requires *ca.* **146.21 MB** of memory in full precision (*fp32*)
or **73.1 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-nh16 | 6c9ce83d5c0e16309accc18c0c1e542f3a44db88 | 2022-02-15T10:51:25.000Z | [
"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-nh16 | 5 | null | transformers | 16,584 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-TINY-NH16 (Deep-Narrow version)
T5-Efficient-TINY-NH16 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-nh16** - is of model type **Tiny** with the following variations:
- **nh** is **16**
It has **25.02** million parameters and thus requires *ca.* **100.07 MB** of memory in full precision (*fp32*)
or **50.04 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-nl6 | a70f782380f63c55f86a4e18834693151d7ff2f2 | 2022-02-15T10:51:56.000Z | [
"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-xl-nl6 | 5 | null | transformers | 16,585 | ---
language:
- en
datasets:
- c4
tags:
- deep-narrow
inference: false
license: apache-2.0
---
# T5-Efficient-XL-NL6 (Deep-Narrow version)
T5-Efficient-XL-NL6 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-nl6** - is of model type **Xl** with the following variations:
- **nl** is **6**
It has **737.59** million parameters and thus requires *ca.* **2950.37 MB** of memory in full precision (*fp32*)
or **1475.18 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-xxl-ssm-nqo | a3ec36bdb6c74ea3a2e5f5d326042b81930e6167 | 2020-12-07T08:44:07.000Z | [
"pytorch",
"tf",
"t5",
"text2text-generation",
"en",
"dataset:c4",
"dataset:wikipedia",
"dataset:natural_questions",
"arxiv:2002.08909",
"arxiv:1910.10683",
"transformers",
"license:apache-2.0",
"autotrain_compatible"
]
| text2text-generation | false | google | null | google/t5-xxl-ssm-nqo | 5 | null | transformers | 16,586 | ---
language: en
datasets:
- c4
- wikipedia
- natural_questions
license: apache-2.0
---
[Google's T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) for **Closed Book Question Answering**.
The model was pre-trained using T5's denoising objective on [C4](https://huggingface.co/datasets/c4), subsequently additionally pre-trained using [REALM](https://arxiv.org/pdf/2002.08909.pdf)'s salient span masking objective on [Wikipedia](https://huggingface.co/datasets/wikipedia), and finally fine-tuned on [Natural Questions (NQ)](https://huggingface.co/datasets/natural_questions).
**Note**: The model was fine-tuned on 90% of the train splits of [Natural Questions (NQ)](https://huggingface.co/datasets/natural_questions) for 20k steps and validated on the held-out 10% of the train split.
Other community Checkpoints: [here](https://huggingface.co/models?search=ssm)
Paper: [How Much Knowledge Can You Pack
Into the Parameters of a Language Model?](https://arxiv.org/abs/1910.10683.pdf)
Authors: *Adam Roberts, Colin Raffel, Noam Shazeer*
## Results on Natural Questions - Test Set
|Id | link | Exact Match |
|---|---|---|
|T5-large|https://huggingface.co/google/t5-large-ssm-nqo|29.0|
|**T5-xxl**|**https://huggingface.co/google/t5-xxl-ssm-nqo**|**35.2**|
|T5-3b|https://huggingface.co/google/t5-3b-ssm-nqo|31.7|
|T5-11b|https://huggingface.co/google/t5-11b-ssm-nqo|34.8|
## Usage
The model can be used as follows for **closed book question answering**:
```python
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
t5_qa_model = AutoModelForSeq2SeqLM.from_pretrained("google/t5-xxl-ssm-nqo")
t5_tok = AutoTokenizer.from_pretrained("google/t5-xxl-ssm-nqo")
input_ids = t5_tok("When was Franklin D. Roosevelt born?", return_tensors="pt").input_ids
gen_output = t5_qa_model.generate(input_ids)[0]
print(t5_tok.decode(gen_output, skip_special_tokens=True))
```
## Abstract
It has recently been observed that neural language models trained on unstructured text can implicitly store and retrieve knowledge using natural language queries. In this short paper, we measure the practical utility of this approach by fine-tuning pre-trained models to answer questions without access to any external context or knowledge. We show that this approach scales with model size and performs competitively with open-domain systems that explicitly retrieve answers from an external knowledge source when answering questions. To facilitate reproducibility and future work, we release our code and trained models at https://goo.gle/t5-cbqa.
 |
google/tapas-medium-masklm | 0eb623259598f26f0c89c4ef167281498ce4e040 | 2021-11-29T14:20:32.000Z | [
"pytorch",
"tf",
"tapas",
"fill-mask",
"transformers",
"autotrain_compatible"
]
| fill-mask | false | google | null | google/tapas-medium-masklm | 5 | null | transformers | 16,587 | This model corresponds to **tapas_masklm_medium_reset** of the [original repository](https://github.com/google-research/tapas).
Here's how you can use it:
```python
from transformers import TapasTokenizer, TapasForMaskedLM
import pandas as pd
import torch
tokenizer = TapasTokenizer.from_pretrained("google/tapas-medium-masklm")
model = TapasForMaskedLM.from_pretrained("google/tapas-medium-masklm")
data = {'Actors': ["Brad Pitt", "Leonardo Di Caprio", "George Clooney"],
'Age': ["56", "45", "59"],
'Number of movies': ["87", "53", "69"]
}
table = pd.DataFrame.from_dict(data)
query = "How many movies has Leonardo [MASK] Caprio played in?"
# prepare inputs
inputs = tokenizer(table=table, queries=query, padding="max_length", return_tensors="pt")
# forward pass
outputs = model(**inputs)
# return top 5 values and predictions
masked_index = torch.nonzero(inputs.input_ids.squeeze() == tokenizer.mask_token_id, as_tuple=False)
logits = outputs.logits[0, masked_index.item(), :]
probs = logits.softmax(dim=0)
values, predictions = probs.topk(5)
for value, pred in zip(values, predictions):
print(f"{tokenizer.decode([pred])} with confidence {value}")
``` |
google/tapas-tiny | 9310139ad2732e5f63ae2ca9bc5b5223fd6ba622 | 2021-11-29T10:01:08.000Z | [
"pytorch",
"tf",
"tapas",
"feature-extraction",
"en",
"arxiv:2004.02349",
"arxiv:2010.00571",
"transformers",
"TapasModel",
"license:apache-2.0"
]
| feature-extraction | false | google | null | google/tapas-tiny | 5 | null | transformers | 16,588 | ---
language: en
tags:
- tapas
- TapasModel
license: apache-2.0
---
# TAPAS tiny model
This model has 2 versions which can be used. The latest version, which is the default one, corresponds to the `tapas_inter_masklm_tiny_reset` checkpoint of the [original Github repository](https://github.com/google-research/tapas).
This model was pre-trained on MLM and an additional step which the authors call intermediate pre-training. It uses relative position embeddings by default (i.e. resetting the position index at every cell of the table).
The other (non-default) version which can be used is the one with absolute position embeddings:
- `revision="no_reset"`, which corresponds to `tapas_inter_masklm_tiny`
Disclaimer: The team releasing TAPAS did not write a model card for this model so this model card has been written by
the Hugging Face team and contributors.
## Model description
TAPAS is a BERT-like transformers model pretrained on a large corpus of English data from Wikipedia in a self-supervised fashion.
This means it was pretrained on the raw tables and associated texts only, with no humans labelling them in any way (which is why it
can use lots of publicly available data) with an automatic process to generate inputs and labels from those texts. More precisely, it
was pretrained with two objectives:
- Masked language modeling (MLM): taking a (flattened) table and associated context, the model randomly masks 15% of the words in
the input, then runs the entire (partially masked) sequence through the model. The model then has to predict the masked words.
This is different from traditional recurrent neural networks (RNNs) that usually see the words one after the other,
or from autoregressive models like GPT which internally mask the future tokens. It allows the model to learn a bidirectional
representation of a table and associated text.
- Intermediate pre-training: to encourage numerical reasoning on tables, the authors additionally pre-trained the model by creating
a balanced dataset of millions of syntactically created training examples. Here, the model must predict (classify) whether a sentence
is supported or refuted by the contents of a table. The training examples are created based on synthetic as well as counterfactual statements.
This way, the model learns an inner representation of the English language used in tables and associated texts, which can then be used
to extract features useful for downstream tasks such as answering questions about a table, or determining whether a sentence is entailed
or refuted by the contents of a table. Fine-tuning is done by adding one or more classification heads on top of the pre-trained model, and then
jointly train these randomly initialized classification heads with the base model on a downstream task.
## Intended uses & limitations
You can use the raw model for getting hidden representatons about table-question pairs, but it's mostly intended to be fine-tuned on a downstream task such as question answering or sequence classification. See the [model hub](https://huggingface.co/models?filter=tapas) to look for fine-tuned versions on a task that interests you.
## Training procedure
### Preprocessing
The texts are lowercased and tokenized using WordPiece and a vocabulary size of 30,000. The inputs of the model are
then of the form:
```
[CLS] Sentence [SEP] Flattened table [SEP]
```
### Pre-training
The model was pre-trained on 32 Cloud TPU v3 cores for 1,000,000 steps with maximum sequence length 512 and batch size of 512.
In this setup, pre-training on MLM only takes around 3 days. Aditionally, the model has been further pre-trained on a second task (table entailment). See the original TAPAS [paper](https://www.aclweb.org/anthology/2020.acl-main.398/) and the [follow-up paper](https://www.aclweb.org/anthology/2020.findings-emnlp.27/) for more details.
The optimizer used is Adam with a learning rate of 5e-5, and a warmup
ratio of 0.01.
### BibTeX entry and citation info
```bibtex
@misc{herzig2020tapas,
title={TAPAS: Weakly Supervised Table Parsing via Pre-training},
author={Jonathan Herzig and Paweł Krzysztof Nowak and Thomas Müller and Francesco Piccinno and Julian Martin Eisenschlos},
year={2020},
eprint={2004.02349},
archivePrefix={arXiv},
primaryClass={cs.IR}
}
```
```bibtex
@misc{eisenschlos2020understanding,
title={Understanding tables with intermediate pre-training},
author={Julian Martin Eisenschlos and Syrine Krichene and Thomas Müller},
year={2020},
eprint={2010.00571},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
``` |
goumbalamm/EsperBERTo | c6e2cd8fa0ccfb1d52303ddc6b75dd35b88798d4 | 2021-05-20T16:34:49.000Z | [
"pytorch",
"jax",
"roberta",
"fill-mask",
"transformers",
"autotrain_compatible"
]
| fill-mask | false | goumbalamm | null | goumbalamm/EsperBERTo | 5 | null | transformers | 16,589 | Entry not found |
hadxu/distilbert-base-uncased-finetuned-clinc | f74ac603efe31c95fa95e4a536530ac3d44a31b9 | 2022-02-13T04:05:09.000Z | [
"pytorch",
"tensorboard",
"distilbert",
"text-classification",
"dataset:clinc_oos",
"transformers",
"generated_from_trainer",
"license:apache-2.0",
"model-index"
]
| text-classification | false | hadxu | null | hadxu/distilbert-base-uncased-finetuned-clinc | 5 | null | transformers | 16,590 | ---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- clinc_oos
metrics:
- accuracy
model-index:
- name: distilbert-base-uncased-finetuned-clinc
results:
- task:
name: Text Classification
type: text-classification
dataset:
name: clinc_oos
type: clinc_oos
args: plus
metrics:
- name: Accuracy
type: accuracy
value: 0.9116129032258065
---
<!-- 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-clinc
This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the clinc_oos dataset.
It achieves the following results on the evaluation set:
- Loss: 0.7730
- Accuracy: 0.9116
## 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: 48
- eval_batch_size: 48
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 4.3183 | 1.0 | 318 | 3.3075 | 0.7416 |
| 2.633 | 2.0 | 636 | 1.8792 | 0.8384 |
| 1.5339 | 3.0 | 954 | 1.1514 | 0.8939 |
| 1.0038 | 4.0 | 1272 | 0.8567 | 0.9077 |
| 0.7868 | 5.0 | 1590 | 0.7730 | 0.9116 |
### Framework versions
- Transformers 4.11.3
- Pytorch 1.10.0+cu111
- Datasets 1.16.1
- Tokenizers 0.10.3
|
hanseokhyeon/bert-badword-large | ef37b4057cb44c4d66082bb2a54a264eb13f0a65 | 2021-05-19T18:02:17.000Z | [
"pytorch",
"jax",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | hanseokhyeon | null | hanseokhyeon/bert-badword-large | 5 | null | transformers | 16,591 | Entry not found |
hanseokhyeon/bert-badword-puri-000 | f6cf61bd263bb7cdd377a32139f0a4eed516ebc8 | 2021-05-19T18:04:48.000Z | [
"pytorch",
"jax",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | hanseokhyeon | null | hanseokhyeon/bert-badword-puri-000 | 5 | null | transformers | 16,592 | Entry not found |
hanseokhyeon/bert-badword-puri | 2d2dac75ff184bf6658e4fbd6dd114dd4c66a93a | 2021-05-19T18:13:40.000Z | [
"pytorch",
"jax",
"bert",
"text-classification",
"transformers"
]
| text-classification | false | hanseokhyeon | null | hanseokhyeon/bert-badword-puri | 5 | null | transformers | 16,593 | Entry not found |
harish/BERTBaseClone-10000-6000000 | 5c8815f5a8a7088992f7270a613686853d74dcb8 | 2021-05-19T18:23:26.000Z | [
"pytorch",
"jax",
"bert",
"transformers"
]
| null | false | harish | null | harish/BERTBaseClone-10000-6000000 | 5 | null | transformers | 16,594 | Entry not found |
harish/EN-AStitchTask1A-DistilBERT-FalseTrue-0-2-BEST | 50d163528ed33e4f1e8b69532a1b4ade2ec9b11d | 2021-09-05T00:26:05.000Z | [
"pytorch",
"distilbert",
"text-classification",
"transformers"
]
| text-classification | false | harish | null | harish/EN-AStitchTask1A-DistilBERT-FalseTrue-0-2-BEST | 5 | null | transformers | 16,595 | Entry not found |
harish/EN-AStitchTask1A-XLNet-FalseFalse-0-FewShot-4-BEST | dd11e4cd98053e5318f75767062b925a4988bcb8 | 2021-09-05T00:42:26.000Z | [
"pytorch",
"xlnet",
"text-classification",
"transformers"
]
| text-classification | false | harish | null | harish/EN-AStitchTask1A-XLNet-FalseFalse-0-FewShot-4-BEST | 5 | null | transformers | 16,596 | Entry not found |
harish/EN-AStitchTask1A-XLNet-FalseFalse-0-OneShot-0-BEST | f007219b7925b7d5ba17cefcc8bf013ca6d57a9e | 2021-09-05T00:40:18.000Z | [
"pytorch",
"xlnet",
"text-classification",
"transformers"
]
| text-classification | false | harish | null | harish/EN-AStitchTask1A-XLNet-FalseFalse-0-OneShot-0-BEST | 5 | null | transformers | 16,597 | Entry not found |
harish/EN-AStitchTask1A-XLNet-FalseTrue-0-FewShot-0-BEST | afaced1f93006968bdcc74ab488161fcbcaf53ee | 2021-09-05T00:29:47.000Z | [
"pytorch",
"xlnet",
"text-classification",
"transformers"
]
| text-classification | false | harish | null | harish/EN-AStitchTask1A-XLNet-FalseTrue-0-FewShot-0-BEST | 5 | null | transformers | 16,598 | Entry not found |
harish/PT-UP-xlmR-ContextIncluded_IdiomExcluded-OneShot-4_BEST | 9e1e1208e0968a8e87ab19fc4b77cb6f89725e06 | 2021-08-30T02:00:11.000Z | [
"pytorch",
"xlm-roberta",
"text-classification",
"transformers"
]
| text-classification | false | harish | null | harish/PT-UP-xlmR-ContextIncluded_IdiomExcluded-OneShot-4_BEST | 5 | null | transformers | 16,599 | Entry not found |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.