GitLab

* Adjust test arguments and use a new example test

Fix metric calculation in examples and setup tests to run on multi-gpu for no_trainer scripts (#17331)

* Fix length in no_trainer examples

* Add setup and teardown

* Use new accelerator config generator to automatically make tests able to run based on environment

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.

* Add first draft

* Improve README and run fixup

* Make script aligned with other scripts, improve README

* Improve script and add test

* Remove print statement

* Apply suggestions from code review

* Add num_labels to make test pass

* Improve README

* Add first draft from previous PR

* First draft

* Improve README and remove num_labels

* Make script more aligned with other scripts

* Improve README and apply suggestion from code review

* Change tracking to store_true

* Remove step param and use it in the log dictionary directly

* use vars(args) when passing args to init_trackers

* Include tracking tests since tensorboard is already a dep

Move declaration of log streams to before tests, so that results won't get compounded on top of each other

* Fixed some bugs involving saving during epochs
* Added tests mimicking the existing examples tests
* Added in json exporting to all `no_trainer` examples for consistency 

* Per-folder tests reorganization
Co-authored-by: sgugger <sylvain.gugger@gmail.com>
Co-authored-by: Stas Bekman <stas@stason.org>

* refactor run_flax_glue.py

* updated readme

* rm unused import and args typo fix

* refactor

* make consistent arg name across task

* has_tensorboard check

* argparse -> argument dataclasses

* refactor according to review

* fix

* use custom training arguments

* update tests

* make tensorboard optional

* update test_fetcher for flax examples

* make the tests slow

* add test for glue

* add tests for clm

* fix clm test

* add summrization tests

* more tests

* fix few tests

* add test for t5 mlm

* fix t5 mlm test

* fix tests for multi device

* cleanup

* ci job

* fix metric file name

* make t5 more robust