Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Add Trainer(gradient_clip_algorithm='value'|'norm') #6123

Merged
merged 59 commits into from
Apr 6, 2021
Merged

Add Trainer(gradient_clip_algorithm='value'|'norm') #6123

merged 59 commits into from
Apr 6, 2021

Conversation

dhkim0225
Copy link
Contributor

@dhkim0225 dhkim0225 commented Feb 22, 2021 •
edited by Borda
Loading

What does this PR do?

Add "clip_gradients_by_value"
Fixes #5460
fixes #5456
fixes #4927
Previous PR: #5477
@tchaton Thanks for giving me a lot of advice.

Before submitting

  • Was this discussed/approved via a GitHub issue? (not for typos and docs)
  • Did you read the contributor guideline, Pull Request section?
  • Did you make sure your PR does only one thing, instead of bundling different changes together?
  • Did you make sure to update the documentation with your changes? (if necessary)
  • Did you write any new necessary tests? (not for typos and docs)
  • Did you verify new and existing tests pass locally with your changes?
  • Did you update the CHANGELOG? (not for typos, docs, test updates, or internal minor changes/refactorings)

PR review

Anyone in the community is free to review the PR once the tests have passed.
Before you start reviewing make sure you have read Review guidelines. In short, see the following bullet-list:

  • Is this pull request ready for review? (if not, please submit in draft mode)
  • Check that all items from Before submitting are resolved
  • Make sure the title is self-explanatory and the description concisely explains the PR
  • Add labels and milestones (and optionally projects) to the PR so it can be classified

Did you have fun?

Make sure you had fun coding 🙃

@dhkim0225 dhkim0225 marked this pull request as draft February 22, 2021 12:37
@codecov
Copy link

codecov bot commented Feb 22, 2021 •
edited
Loading

Codecov Report

Merging #6123 (4d5e05f) into master (6dc1078) will decrease coverage by 6%.
The diff coverage is 94%.

@@           Coverage Diff           @@
##           master   #6123    +/-   ##
=======================================
- Coverage      91%     86%    -6%     
=======================================
  Files         192     192            
  Lines       12192   12513   +321     
=======================================
- Hits        11149   10721   -428     
- Misses       1043    1792   +749

@dhkim0225 dhkim0225 changed the title add changelog add clip_grad_by_value feature Feb 22, 2021
@dhkim0225 dhkim0225 marked this pull request as ready for review February 22, 2021 15:37
@Borda
Copy link
Member

Borda commented Feb 22, 2021

Related Issues: #5460, #5456, #4927

what does it mean that it solves these issues?

@Borda Borda added the feature Is an improvement or enhancement label Feb 22, 2021
@Borda Borda added this to the 1.3 milestone Feb 22, 2021
Borda
Borda reviewed Feb 22, 2021
View reviewed changes
CHANGELOG.md Outdated Show resolved Hide resolved
pytorch_lightning/accelerators/accelerator.py Outdated Show resolved Hide resolved
pytorch_lightning/plugins/precision/deepspeed_precision.py Outdated Show resolved Hide resolved
docs/source/advanced/training_tricks.rst Show resolved Hide resolved
pytorch_lightning/utilities/enums.py Outdated Show resolved Hide resolved
@dhkim0225
Copy link
Contributor Author

@Borda Thank you for your comments.
I changed codes based on comments.

Related Issues: #5460, #5456, #4927

what does it mean that it solves these issues?

It means, this PR adds a clip_gradient by value option.

@dhkim0225 dhkim0225 requested a review from Borda February 23, 2021 03:04
@dhkim0225
Copy link
Contributor Author

dhkim0225 commented Mar 25, 2021 •
edited
Loading

Bugs in the test codes clip gradients by value with horovod are fixed.

@carmocca carmocca added ready PRs ready to be merged Will labels Mar 25, 2021
@mergify mergify bot removed the has conflicts label Mar 27, 2021
@dhkim0225
Copy link
Contributor Author

@carmocca Honestly, I'm tired of the slow progress.
The first time this PR was written was on January 12th. (#5477)
Merge is continuously postponed and new conflicts continue to occur, although It is a simple feature.
Please clearly guide me if there is anything I can do to make this feature merge quickly.

@mergify mergify bot removed the has conflicts label Apr 6, 2021
Borda
Borda approved these changes Apr 6, 2021
View reviewed changes
pytorch_lightning/plugins/precision/precision_plugin.py
clip_grad_func = self.clip_grad_funcs[gradient_clip_algorithm]
clip_grad_func(optimizer, clip_val)

def clip_grad_by_value(self, optimizer: 'Optimizer', clip_val: Union[int, float]) -> None:
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

or protected :]

@carmocca
Copy link
Contributor

carmocca commented Apr 6, 2021

Hey @dhkim0225! Thanks a lot for your patience. This PR is just waiting for the final approval so I will just push it internally 👍

@carmocca
Delete signature test
4d5e05f 
No point in testing a signature
@williamFalcon williamFalcon merged commit 7f6154f into Lightning-AI:master Apr 6, 2021
@williamFalcon
Copy link
Contributor

merged!
sorry about that! Looks like this one fell through the cracks.

Thanks for the contribution!

@dhkim0225 dhkim0225 deleted the feat/clip_grad_by_value branch April 6, 2021 13:30
@ceshine ceshine mentioned this pull request Apr 9, 2021
Closed
facebook-github-bot pushed a commit to facebookresearch/d2go that referenced this pull request Apr 14, 2021
@facebook-github-bot
Synchronize PyTorchLightning/pytorch-lightning (revision 0b843848@mas…
44e4108 
…ter) to github/third-party/PyTorchLightning/pytorch-lightning

Summary:
### New commit log messages
## [UnReleased] - 2021-MM-DD

### Added

- Added more explicit exception message when trying to execute `trainer.test()` or `trainer.validate()` with `fast_dev_run=True` ([#6667](Lightning-AI/pytorch-lightning#6667))

- Added `LightningCLI` class to provide simple reproducibility with minimum boilerplate training cli. ([#4492](Lightning-AI/pytorch-lightning#4492))

- Trigger warning when non-metric logged value with multi processes hasn't been reduced ([#6417](Lightning-AI/pytorch-lightning#6417))

- Added `gradient_clip_algorithm` argument to Trainer for gradient clipping by value ([#6123](Lightning-AI/pytorch-lightning#6123)).

- Added a way to print to terminal without breaking up the progress bar ([#5470](Lightning-AI/pytorch-lightning#5470))

- Added support to checkpoint after training steps in `ModelCheckpoint` callback ([#6146](Lightning-AI/pytorch-lightning#6146))

- Added `checkpoint` parameter to callback's `on_save_checkpoint` hook ([#6072](Lightning-AI/pytorch-lightning#6072))

- Added `RunningStage.SANITY_CHECKING` ([#4945](Lightning-AI/pytorch-lightning#4945))

- Added `TrainerState.{FITTING,VALIDATING,TESTING,PREDICTING,TUNING}` ([#4945](Lightning-AI/pytorch-lightning#4945))

- Added `Trainer.validate()` method to perform one evaluation epoch over the validation set ([#4948](Lightning-AI/pytorch-lightning#4948))

- Added `LightningEnvironment` for Lightning-specific DDP ([#5915](Lightning-AI/pytorch-lightning#5915))

- Added `teardown()` hook to LightningDataModule ([#4673](Lightning-AI/pytorch-lightning#4673))

- Added `auto_insert_metric_name` parameter to `ModelCheckpoint` ([#6277](Lightning-AI/pytorch-lightning#6277))

- Added arg to `self.log` that enables users to give custom names when dealing with multiple dataloaders ([#6274](Lightning-AI/pytorch-lightning#6274))

- Added `teardown` method to `BaseProfiler` to enable subclasses defining post-profiling steps outside of `__del__` ([#6370](Lightning-AI/pytorch-lightning#6370))

- Added `setup` method to `BaseProfiler` to enable subclasses defining pre-profiling steps for every process ([#6633](Lightning-AI/pytorch-lightning#6633))

- Added no return warning to predict ([#6139](Lightning-AI/pytorch-lightning#6139))

- Added `Trainer.predict` config validation ([#6543](Lightning-AI/pytorch-lightning#6543))

- Added `AbstractProfiler` interface ([#6621](Lightning-AI/pytorch-lightning#6621))

- Added support for including module names for forward in the autograd trace of `PyTorchProfiler` ([#6349](Lightning-AI/pytorch-lightning#6349))

- Added support for the PyTorch 1.8.1 autograd profiler ([#6618](Lightning-AI/pytorch-lightning#6618))

- Added `outputs` parameter to callback's `on_validation_epoch_end` & `on_test_epoch_end` hooks ([#6120](Lightning-AI/pytorch-lightning#6120))

- Added `configure_sharded_model` hook ([#6679](Lightning-AI/pytorch-lightning#6679))

- Added support for `precision=64`, enabling training with double precision ([#6595](Lightning-AI/pytorch-lightning#6595))

- Added support for DDP communication hooks ([#6736](Lightning-AI/pytorch-lightning#6736))

- Added `artifact_location` argument to `MLFlowLogger` which will be passed to the `MlflowClient.create_experiment` call ([#6677](Lightning-AI/pytorch-lightning#6677))

- Added `model` parameter to precision plugins' `clip_gradients` signature ([#6764](Lightning-AI/pytorch-lightning#6764))

### Changed

- Renamed `pytorch_lightning.callbacks.swa` to `pytorch_lightning.callbacks.stochastic_weight_avg` ([#6259](Lightning-AI/pytorch-lightning#6259))

- Refactor `RunningStage` and `TrainerState` usage ([#4945](Lightning-AI/pytorch-lightning#4945))

- Changed `trainer.evaluating` to return `True` if validating or testing ([#4945](Lightning-AI/pytorch-lightning#4945))

- Changed `setup()` and `teardown()` stage argument to take any of `{fit,validate,test,predict}` ([#6386](Lightning-AI/pytorch-lightning#6386))

- Changed profilers to save separate report files per state and rank ([#6621](Lightning-AI/pytorch-lightning#6621))

- Changed `PyTorchProfiler` to use `torch.autograd.profiler.record_function` to record functions ([#6349](Lightning-AI/pytorch-lightning#6349))

### Deprecated

- `period` has been deprecated in favor of `every_n_val_epochs` in the `ModelCheckpoint` callback ([#6146](Lightning-AI/pytorch-lightning#6146))

- Deprecated `trainer.running_sanity_check` in favor of `trainer.sanity_checking` ([#4945](Lightning-AI/pytorch-lightning#4945))

- Deprecated `Profiler(output_filename)` in favor of `dirpath` and `filename` ([#6621](Lightning-AI/pytorch-lightning#6621))

- Deprecated `PytorchProfiler(profiled_functions)` in favor of `record_functions` ([#6349](Lightning-AI/pytorch-lightning#6349))

- Deprecated metrics in favor of `torchmetrics` ([#6505](Lightning-AI/pytorch-lightning#6505),
    [#6530](Lightning-AI/pytorch-lightning#6530),
    [#6540](Lightning-AI/pytorch-lightning#6540),
    [#6547](Lightning-AI/pytorch-lightning#6547),
    [#6515](Lightning-AI/pytorch-lightning#6515),
    [#6572](Lightning-AI/pytorch-lightning#6572),
    [#6573](Lightning-AI/pytorch-lightning#6573),
    [#6584](Lightning-AI/pytorch-lightning#6584),
    [#6636](Lightning-AI/pytorch-lightning#6636),
    [#6637](Lightning-AI/pytorch-lightning#6637),
    [#6649](Lightning-AI/pytorch-lightning#6649),
    [#6659](Lightning-AI/pytorch-lightning#6659),
)

### Removed

- Removed support for passing a bool value to `profiler` argument of Trainer ([#6164](Lightning-AI/pytorch-lightning#6164))

- Removed no return warning from val/test step ([#6139](Lightning-AI/pytorch-lightning#6139))

- Removed passing a `ModelCheckpoint` instance to `Trainer(checkpoint_callback)` ([#6166](Lightning-AI/pytorch-lightning#6166))

- Removed deprecated Trainer argument `enable_pl_optimizer` and `automatic_optimization` ([#6163](Lightning-AI/pytorch-lightning#6163))

- Removed deprecated metrics ([#6161](Lightning-AI/pytorch-lightning#6161))
    * from `pytorch_lightning.metrics.functional.classification` removed `to_onehot`, `to_categorical`, `get_num_classes`, `roc`, `multiclass_roc`, `average_precision`, `precision_recall_curve`, `multiclass_precision_recall_curve`
    * from `pytorch_lightning.metrics.functional.reduction` removed `reduce`, `class_reduce`

- Removed deprecated `ModelCheckpoint` arguments `prefix`, `mode="auto"` ([#6162](Lightning-AI/pytorch-lightning#6162))

- Removed `mode='auto'` from `EarlyStopping` ([#6167](Lightning-AI/pytorch-lightning#6167))

- Removed legacy references for magic keys in the `Result` object ([#6016](Lightning-AI/pytorch-lightning#6016))

- Removed deprecated `LightningModule` `hparams` setter ([#6207](Lightning-AI/pytorch-lightning#6207))

- Removed legacy code to log or include metrics in the progress bar by returning them in a dict with the `"log"/"progress_bar"` magic keys. Use `self.log` instead ([#6734](Lightning-AI/pytorch-lightning#6734))

- Removed `optimizer_idx` argument from `training_step` in manual optimization ([#6093](Lightning-AI/pytorch-lightning#6093))

### Fixed

- Set better defaults for `rank_zero_only.rank` when training is launched with SLURM and torchelastic ([#6802](Lightning-AI/pytorch-lightning#6802))

- Made the `Plugin.reduce` method more consistent across all Plugins to reflect a mean-reduction by default ([#6011](Lightning-AI/pytorch-lightning#6011))

- Move lightning module to correct device type when using LightningDistributedWrapper ([#6070](Lightning-AI/pytorch-lightning#6070))

- Do not print top-k verbose log with `ModelCheckpoint(monitor=None)` ([#6109](Lightning-AI/pytorch-lightning#6109))

- Fixed csv extension check ([#6436](Lightning-AI/pytorch-lightning#6436))

- Fixed `ModelCheckpoint(monitor=None, save_last=True)` not saving checkpoints ([#6136](Lightning-AI/pytorch-lightning#6136))

- Fixed `ModelCheckpoint(save_top_k=0, save_last=True)` not saving the `last` checkpoint ([#6136](Lightning-AI/pytorch-lightning#6136))

- Fixed `.teardown(stage='fit')` getting called during `trainer.test` ([#6386](Lightning-AI/pytorch-lightning#6386))

- Fixed `.on_fit_{start,end}()` getting called during `trainer.test` ([#6386](Lightning-AI/pytorch-lightning#6386))

- Fixed LightningModule `all_gather` on cpu tensors ([#6416](Lightning-AI/pytorch-lightning#6416))

- Fixed torch distributed not available in setup hook for DDP ([#6506](Lightning-AI/pytorch-lightning#6506))

- Fixed `EarlyStopping` logic when `min_epochs` or `min_steps` requirement is not met ([#6705](Lightning-AI/pytorch-lightning#6705))

## [1.2.7] - 2021-04-06

### Fixed

- Fixed resolve a bug with omegaconf and xm.save ([#6741](Lightning-AI/pytorch-lightning#6741))
- Fixed an issue with IterableDataset when __len__ is not defined ([#6828](Lightning-AI/pytorch-lightning#6828))
- Sanitize None params during pruning ([#6836](Lightning-AI/pytorch-lightning#6836))
- Enforce an epoch scheduler interval when using SWA ([#6588](Lightning-AI/pytorch-lightning#6588))
- Fixed TPU Colab hang issue, post training ([#6816](Lightning-AI/pytorch-lightning#6816))
- Fixed a bug where `TensorBoardLogger` would give a warning and not log correctly to a symbolic link `save_dir` ([#6730](Lightning-AI/pytorch-lightning#6730))

## [1.2.6] - 2021-03-30

### Changed

- Changed the behavior of `on_epoch_start` to run at the beginning of validation & test epoch ([#6498](Lightning-AI/pytorch-lightning#6498))

### Removed

- Removed legacy code to include `step` dictionary returns in `callback_metrics`. Use `self.log_dict` instead. ([#6682](Lightning-AI/pytorch-lightning#6682))

### Fixed

- Fixed `DummyLogger.log_hyperparams` raising a `TypeError` when running with `fast_dev_run=True` ([#6398](Lightning-AI/pytorch-lightning#6398))
- Fixed error on TPUs when there was no `ModelCheckpoint` ([#6654](Lightning-AI/pytorch-lightning#6654))
- Fixed `trainer.test` freeze on TPUs ([#6654](Lightning-AI/pytorch-lightning#6654))
- Fixed a bug where gradients were disabled after calling `Trainer.predict` ([#6657](Lightning-AI/pytorch-lightning#6657))
- Fixed bug where no TPUs were detected in a TPU pod env ([#6719](Lightning-AI/pytorch-lightning#6719))

## [1.2.5] - 2021-03-23

### Changed

- Update Gradient Clipping for the TPU Accelerator ([#6576](Lightning-AI/pytorch-lightning#6576))
- Refactored setup for typing friendly ([#6590](Lightning-AI/pytorch-lightning#6590))

### Fixed

- Fixed a bug where `all_gather` would not work correctly with `tpu_cores=8` ([#6587](Lightning-AI/pytorch-lightning#6587))
- Fixed comparing required versions ([#6434](Lightning-AI/pytorch-lightning#6434))
- Fixed duplicate logs appearing in console when using the python logging module ([#6275](Lightning-AI/pytorch-lightning#6275))
- Added Autocast in validation, test and predict modes for Native AMP ([#6565](Lightning-AI/pytorch-lightning#6565))

Reviewed By: shuyingsunshine21

Differential Revision: D27528929

fbshipit-source-id: 311c88f71461c2c79bbf185e28d7a6d683ccc26f
@ceshine ceshine mentioned this pull request Apr 15, 2021
Merged
11 tasks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@carmocca carmocca carmocca approved these changes

@tchaton tchaton tchaton approved these changes

@Borda Borda Borda approved these changes

@williamFalcon williamFalcon williamFalcon approved these changes

@ananthsub ananthsub ananthsub approved these changes

@awaelchli awaelchli Awaiting requested review from awaelchli

@justusschock justusschock Awaiting requested review from justusschock justusschock is a code owner

@SeanNaren SeanNaren Awaiting requested review from SeanNaren

@edenlightning edenlightning Awaiting requested review from edenlightning

Assignees

@carmocca carmocca

@dhkim0225 dhkim0225

Labels
design Includes a design discussion feature Is an improvement or enhancement ready PRs ready to be merged
Projects
None yet
Milestone
pl:1.3
Development

Successfully merging this pull request may close these issues.

clip_gradient with clip_grad_value clip_gradient with clip_grad_value clip_gradient with clip_grad_value
8 participants
@dhkim0225 @Borda @pep8speaks @carmocca @ananthsub @kaushikb11 @williamFalcon @tchaton