sgugger Follow

Sylvain Gugger sgugger

Research Engineer at HuggingFace

729 followers · 3 following · 12

Highlights

Arctic Code Vault Contributor

Popular repositories

Deep-Learning

A few notebooks about deep learning in pytorch

Jupyter Notebook 451 95
Adam-experiments

Experiments with Adam/AdamW/amsgrad

Python 165 33
SwiftData

Swift 12 1
fastai

Forked from fastai/fastai

The fast.ai deep learning library, lessons, and tutorials

Jupyter Notebook 3
transformer_train

Python 3
swift-training

A generic training loop for S4TF

Swift 3 1

2,634 contributions in the last year

Contribution activity

August 2020

sgugger/test_nbs Jupyter Notebook Aug 6

Created a pull request in huggingface/transformers that received 6 comments

Aug 4

Tf model outputs

This PR continues the work on model outputs and treats all TF models (except T5) to use them. Since the new output type is opt-in only (default to r…

+3,267 −2,444 • 6 comments

Generation doc Aug 13
Add a classmethod to easily build a Trainer from nlp dataset and metric Aug 12
Adding PaddingDataCollator Aug 12
Fix #6428 Aug 12
Activate check on the CI Aug 11
Move prediction_loss_only to TrainingArguments Aug 11
Fixes to make life easier with the nlp library Aug 11
Data collator with padding Aug 10
Fix links for open in colab Aug 10
Colab button Aug 10
Small docfile fixes Aug 7
Add a script to check all models are tested and documented Aug 6

Doc notebooks Aug 8

[sched] polynomial_decay_schedule use default power=1.0 Aug 17
Fix flaky ONNX tests Aug 17
skip onnx test until morgan comes back Aug 17
add BartConfig.force_bos_token_to_be_generated Aug 17
allow spaces in bash args with "$@" Aug 17
[WIP] T5Tokenizer adds EOS token Aug 17
[doc] multiple corrections to "Summary of the tasks" Aug 17
[doc] make the text more readable, fix some typos, add some disambiguation Aug 17
[testing] a new TestCasePlus subclass + get_auto_remove_tmp_dir() Aug 17
[Doc] add more MBart doc Aug 17
[TF Longformer] Improve Speed for TF Longformer Aug 14
[WIP] add custom datasets tutorial Aug 13
[BartTokenizer] add prepare s2s batch Aug 13
Sort unique_no_split_tokens to make it deterministic Aug 13
Clean directory after script testing Aug 13
Test model outputs equivalence Aug 13
[EncoderDecoder] Add Cross Attention for GPT2 Aug 13
Add a classmethod to easily build a Trainer from nlp dataset and metric Aug 12
MBartForConditionalGeneration Aug 12
[s2s] add BartTranslationDistiller for distilling mBART Aug 12
[POC] Centralize logging Aug 12
[examples] add pytest dependency Aug 11
Fixes to make life easier with the nlp library Aug 11
[EncoderDecoder] Add encoder-decoder for roberta/ vanilla longformer Aug 11
Ci GitHub caching Aug 10
Some pull request reviews not shown.

add custom datasets tutorial Aug 17

Created an issue in huggingface/transformers that received 2 comments

Aug 5

Reformer now requires PyTorch 1.6.0

@patrickvonplaten, one of your recent PR (#6244) on Reformer introduces a dep on PyTorch 1.6.0 minimum by using torch.cuda.default_generators. We s…

2 comments