Create your own GitHub profile
Sign up for your own profile on GitHub, the best place to host code, manage projects, and build software alongside 50 million developers.
Sign up
Popular repositories
-
-
-
Forked from fastai/fastai
The fast.ai deep learning library, lessons, and tutorials
-
-
3,297 contributions in the last year
Contribution activity
October 2020
Created a pull request in huggingface/transformers that received 2 comments
New run glue script
What does this PR do?
This PR cleans up the run_glue.py script to use the Datasets library. Along the way it adds a few fixes in Trainer. The scrip…
- Doc styling
- Reload checkpoint
- Only log total_flos at the end of training
- Expose the Flax code quality problems
- Add Flax dummy objects
- Small fixes to HP search
- Small fixes to NotebookProgressCallback
- Add specific notebook ProgressCalback
- Don't use `store_xxx` on optional bools
- Add predict step accumulation
- Fix typo in all model docs
- Fix flaky test in test_trainer
- Fix title level in Blenderbot doc
- Better links for models in README and doc index
- Trainer callbacks
- Documentation fixes
- Expand test to locate flakiness
- Remove labels from the RagModel example
- Allow nested tensors in predicted logits
- Allow soft dependencies in the namespace with ImportErrors at use
- Check and update model list in index.rst automatically
- Fix post_init of some TrainingArguments
- Cleanup documentation for BART, Marian, MBART and Pegasus
- Fix seq2seq example test
- Add mixed precision evaluation
- [cleanup] pegasus,marian,mbart pytorch tests
- [tokenizers] Fixing #8001 - Adding tests on tokenizers serialization
- [tests|tokenizers] Refactoring pipelines test backbone - Small tokenizers improvements - General tests speedups
- FillMaskPipeline: support passing top_k on __call__
- support relative path for best_model_checkpoint
- fix 'encode_plus' docstring for 'special_tokens_mask' (0s and 1s were reversed)
- Move NoLayerEmbedTokens
- TensorBoard/Wandb/optuna/raytune integration improvements.
- Fix comet_ml import and add ensure availability
- [ProphetNet] Correct Doc string example
- [testing] slow tests should be marked as slow
- [PretrainedConfig] Fix save pretrained config for edge case
- [multiple models] skip saving/loading deterministic state_dict keys
- # Add whole word mask support for lm fine-tune
- Respect the 119 line chars
- [testing] experiment with a different way of skipping torch-only test modules
- [flax] fix repo_check
- New run glue script
- Trainer with Iterable Dataset
- labels and decoder_input_ids to Glossary
- Add TFBartForConditionalGeneration
- [CIs] report slow tests add --durations=0 to some pytest jobs
- ProphetNet
- [testing] remove USE_CUDA
- Some pull request reviews not shown.

