GitLab

* Propagate and fix imports

* Clean up tests

* Make fixup
Co-authored-by: Niels Rogge <nielsrogge@Nielss-MacBook-Pro.local>

* add torch.no_grad when in eval mode

* make style quality

* [FlaxSpeechEncoderDecoder] Fix bug in `decoder_module`

* [FlaxEncoderDecoder] Fix bug in `decoder_module`

* Add meta proxy

* Uses meta data to trace data dependent control-flow

* Remove commented class

* Handles torch creating functions

* Added type annotation to fix tracing

* Tracing works for everything but T5 and GPT-J

* Almost all previously supported models pass

* All architectures can be traced except T5

* Intermediate commit to have a trace of the comparison operators for HFProxy

* Everything works, except loss computation

* Everything works

* Removed unused import

* Overriden methods do not use underlying ops (linear and torch.matmul), and model attributes are copied to the traced version

* Fix torch_matmul_override

* Change attributes reference to deepcopy

* Remove breakpoint and add torch_index_override

* Small fix

* Fix typo

* Replace asserts by explicit exceptions

* Add translating guide

* Add the check

* add missing ckpts

* add a list to ignore

* call the added check script

* better regex pattern
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

* Result of new doc style with fixes

* Add last two files

* Bump hf-doc-builder

* Replace dict/BatchEncoding instance checks by Mapping

* Typo

This reverts commit 4f3a14e3.

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

* dd parameter --config_overrides for run_mlm_wwm.py

* linter

* fix

* Apply suggestions from code review
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

The emoji version must be either 0.5.4 or 0.6.0. Newer emoji versions have been updated to newer versions of the Emoji Charts, thus not consistent with the one used for pre-processing the pre-training Tweet corpus (i.e. not consistent with the vocab).

* Add -e flag

* add check

* create new keys

* run python setup.py build install

* add comments

* change to develop
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

* Fix doc notebooks links

* Remove missing section

* Fixup all examples

* Fix multiple deletions of the same files in save_pretrained

* Add is_main_process argument

BertModelForSequenceClassification -> BertForSequenceClassification

* Fix `distributed_concat` with scalar tensor

* Update trainer_pt_utils.py

* Add fix

* Apply suggestion from code review
Co-authored-by: Niels Rogge <nielsrogge@Nielss-MacBook-Pro.local>

1. Fixes evaluation errors popping up when you train/eval on squad v2 (one was newly encountered and one that was previously reported Running SQuAD 1.0 sample command raises IndexError #15401 but not completely fixed).
2. Removes boolean arguments that don't use store_true. Please, don't use these: *ANY non-empty string is being converted to True in this case and this clearly is not the desired behavior (and it creates a LOT of confusion).
3. All no-trainer test scripts are now saving metric values in the same way (with the right prefix eval_), which is consistent with the trainer-based versions.
4. Adds forgotten model.eval() in the no-trainer versions. This improved some results, but not everything (see the discussion in the end). Please, see the F1 scores and the discussion below.

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>