Skip to content

Commit

Permalink
v1.3.5 & CHANGELOG
Browse files Browse the repository at this point in the history
  • Loading branch information
@lexierule
SeanNaren authored and lexierule committed Jun 9, 2021
1 parent 9afab6f commit 8a5a56b
Show file tree
Hide file tree
Showing 2 changed files with 11 additions and 181 deletions.
190 changes: 10 additions & 180 deletions CHANGELOG.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5,188 +5,23 @@ All notable changes to this project will be documented in this file.
The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).


## [1.4.0] - 2021-MM-DD
## [1.3.5] - 2021-06-08

### Added

- Added support to `LightningModule.to_torchscript` for saving to custom filesystems with fsspec ([#7617](https://github.com/PyTorchLightning/pytorch-lightning/pull/7617))


- Added `KubeflowEnvironment` for use with the `PyTorchJob` operator in Kubeflow


- Added LightningCLI support for config files on object stores ([#7521](https://github.com/PyTorchLightning/pytorch-lightning/pull/7521))


- Added `ModelPruning(prune_on_train_epoch_end=True|False)` to choose when to apply pruning ([#7704](https://github.com/PyTorchLightning/pytorch-lightning/pull/7704))


- Added support for checkpointing based on a provided time interval during training ([#7515](https://github.com/PyTorchLightning/pytorch-lightning/pull/7515))


- Added dataclasses for progress tracking (
[#6603](https://github.com/PyTorchLightning/pytorch-lightning/pull/6603),
[#7574](https://github.com/PyTorchLightning/pytorch-lightning/pull/7574))


- Added argument `trainer.predict(ckpt_path)` ([#7430](https://github.com/PyTorchLightning/pytorch-lightning/pull/7430))


- Added `clip_grad_by_value` support for TPUs ([#7025](https://github.com/PyTorchLightning/pytorch-lightning/pull/7025))


- Added `sub_dir` parameter to `TensorBoardLogger` ([#6195](https://github.com/PyTorchLightning/pytorch-lightning/pull/6195))


- Added correct `dataloader_idx` to batch transfer hooks ([#6241](https://github.com/PyTorchLightning/pytorch-lightning/pull/6241))


- Added `include_none=bool` argument to `apply_to_collection` ([#7769](https://github.com/PyTorchLightning/pytorch-lightning/pull/7769))


- Added `apply_to_collections` to apply a function to two zipped collections ([#7769](https://github.com/PyTorchLightning/pytorch-lightning/pull/7769))


- Added `ddp_fully_sharded` support ([#7487](https://github.com/PyTorchLightning/pytorch-lightning/pull/7487))


- Added `__len__` to `IndexBatchSamplerWrapper` ([#7681](https://github.com/PyTorchLightning/pytorch-lightning/pull/7681))


- Added `should_rank_save_checkpoint` property to Training Plugins ([#7684](https://github.com/PyTorchLightning/pytorch-lightning/pull/7684))


- Added `save_config_filename` init argument to `LightningCLI` to ease resolving name conflicts ([#7741](https://github.com/PyTorchLightning/pytorch-lightning/pull/7741))


### Changed

- Changed calling of `untoggle_optimizer(opt_idx)` out of the closure function ([#7563](https://github.com/PyTorchLightning/pytorch-lightning/pull/7563)

- Changed the `Trainer`'s `checkpoint_callback` argument to allow only boolean values ([#7539](https://github.com/PyTorchLightning/pytorch-lightning/pull/7539))


- Log epoch metrics before the `on_evaluation_end` hook ([#7272](https://github.com/PyTorchLightning/pytorch-lightning/pull/7272))


- Changed these `Trainer` methods to be protected: `call_setup_hook`, `call_configure_sharded_model`, `pre_dispatch`, `dispatch`, `post_dispatch`, `call_teardown_hook`, `run_train`, `run_sanity_check`, `run_evaluate`, `run_evaluation`, `run_predict`, `track_output_for_epoch_end`


- Changed `clip_grad_norm` to use `torch.nn.utils.clip_grad_norm_` ([#7025](https://github.com/PyTorchLightning/pytorch-lightning/pull/7025))


- Validation is now always run inside the training epoch scope ([#7357](https://github.com/PyTorchLightning/pytorch-lightning/pull/7357))


- Refactored Loops
* Moved attributes `global_step`, `current_epoch`, `max/min_steps`, `max/min_epochs`, `batch_idx`, and `total_batch_idx` to TrainLoop ([#7437](https://github.com/PyTorchLightning/pytorch-lightning/pull/7025))
* Refactored result handling in training loop ([#7506](https://github.com/PyTorchLightning/pytorch-lightning/pull/7506))
* Moved attributes `hiddens` and `split_idx` to TrainLoop ([#7507](https://github.com/PyTorchLightning/pytorch-lightning/pull/7507))
* Refactored the logic around manual and automatic optimization inside the optimizer loop ([#7526](https://github.com/PyTorchLightning/pytorch-lightning/pull/7526))
* Simplified "should run validation" logic ([#7682](https://github.com/PyTorchLightning/pytorch-lightning/pull/7682))
* Refactored "should run validation" logic when the trainer is signaled to stop ([#7701](https://github.com/PyTorchLightning/pytorch-lightning/pull/7701))


- Refactored logging
* Renamed and moved `core/step_result.py` to `trainer/connectors/logger_connector/result.py` ([#7736](https://github.com/PyTorchLightning/pytorch-lightning/pull/7736))


- Moved `ignore_scalar_return_in_dp` warning suppression to the DataParallelPlugin class ([#7421](https://github.com/PyTorchLightning/pytorch-lightning/pull/7421/))


- Changed the behaviour when logging evaluation step metrics to no longer append `/epoch_*` to the metric name ([#7351](https://github.com/PyTorchLightning/pytorch-lightning/pull/7351))


- Raise `ValueError` when a `None` value is `self.log`-ed ([#7771](https://github.com/PyTorchLightning/pytorch-lightning/pull/7771))


- Changed `resolve_training_type_plugins` to allow setting `num_nodes` and `sync_batchnorm` from `Trainer` setting ([#7026](https://github.com/PyTorchLightning/pytorch-lightning/pull/7026))


- Default `seed_everything(workers=True)` in the `LightningCLI` ([#7504](https://github.com/PyTorchLightning/pytorch-lightning/pull/7504))


- Changed `model.state_dict()` in `CheckpointConnector` to allow `training_type_plugin` to customize the model's `state_dict()` ([#7474](https://github.com/PyTorchLightning/pytorch-lightning/pull/7474))


- MLflowLogger now uses the env variable `MLFLOW_TRACKING_URI` as default tracking uri ([#7457](https://github.com/PyTorchLightning/pytorch-lightning/pull/7457))


- Changed `WandbLogger(log_model={True/'all'})` to log models as artifacts ([#6231](https://github.com/PyTorchLightning/pytorch-lightning/pull/6231))
- MLFlowLogger now accepts `run_name` as an constructor argument ([#7622](https://github.com/PyTorchLightning/pytorch-lightning/issues/7622))


- Changed `teardown()` in `Accelerator` to allow `training_type_plugin` to customize `teardown` logic ([#7579](https://github.com/PyTorchLightning/pytorch-lightning/pull/7579))


- `Trainer.fit` now raises an error when using manual optimization with unsupported features such as `gradient_clip_val` or `accumulate_grad_batches` ([#7788](https://github.com/PyTorchLightning/pytorch-lightning/pull/7788))


### Deprecated


- Deprecated `TrainerModelHooksMixin` in favor of `pytorch_lightning.utilities.signature_utils` ([#7422](https://github.com/PyTorchLightning/pytorch-lightning/pull/7422))


- Deprecated `num_nodes` and `sync_batchnorm` arguments in `DDPPlugin` and `DDPSpawnPlugin` ([#7026](https://github.com/PyTorchLightning/pytorch-lightning/pull/7026))


### Removed

- Removed `ProfilerConnector` ([#7654](https://github.com/PyTorchLightning/pytorch-lightning/pull/7654))


- Prune deprecated classif. metrics from `pytorch_lightning.metrics.functional.classification` ([#7499](https://github.com/PyTorchLightning/pytorch-lightning/pull/7499))


- Removed deprecated data parallel classes `LightningDataParallel` and `LightningDistributedDataParallel` from `pytorch_lightning.overrides.data_parallel` ([#7510](https://github.com/PyTorchLightning/pytorch-lightning/pull/7510))


- Removed deprecated trainer attributes - `get_model` and `accelerator_backend` ([#7502](https://github.com/PyTorchLightning/pytorch-lightning/pull/7502))


- Removed support for `self.log(tbptt_reduce_fx)` and `self.log(tbptt_pad_token)`. Please, open a discussion explaining your use-case if you relied on these. ([#7644](https://github.com/PyTorchLightning/pytorch-lightning/pull/7644))


- Removed deprecated utils modules `model_utils`, `warning_utils`, `xla_device_utils` and partially `argparse_utils` ([#7503](https://github.com/PyTorchLightning/pytorch-lightning/pull/7503))


- Removed deprecated trainer attributes - `on_cpu`, `on_tpu`, `use_tpu`, `on_gpu`, `use_dp`, `use_ddp`, `use_ddp2`, `use_horovod`, `use_single_gpu` ([#7501](https://github.com/PyTorchLightning/pytorch-lightning/pull/7501))

- Added warning to Training Step output ([#7779](https://github.com/PyTorchLightning/pytorch-lightning/pull/7779))

### Fixed

- Fixed `_check_training_step_output` to be called after `train_step_end` to support more flexible accomodations ([#7868](https://github.com/PyTorchLightning/pytorch-lightning/pull/7868))

- Fixed `apply_to_collection` works on Custom Collections now ([#7851](https://github.com/PyTorchLightning/pytorch-lightning/pull/7851))

- Fixed ambiguous warning when both overfit and train dataloader shuffling are enabled ([#7685](https://github.com/PyTorchLightning/pytorch-lightning/pull/7685))

- Fixed dataloaders are not reset when tuning the model ([#7566](https://github.com/PyTorchLightning/pytorch-lightning/pull/7566))


- Fixed global step update when the epoch is skipped ([#7677](https://github.com/PyTorchLightning/pytorch-lightning/pull/7677))


- Fixed training loop total batch counter when accumulate grad batches was enabled ([#7692](https://github.com/PyTorchLightning/pytorch-lightning/pull/7692))


- Fixed broadcasting in multi-node, multi-gpu DDP using torch 1.7 ([#7592](https://github.com/PyTorchLightning/pytorch-lightning/pull/7592))


- Fixed `ProgressBar` pickling after calling `trainer.predict` ([#7608](https://github.com/PyTorchLightning/pytorch-lightning/pull/7608))


- Fixed `None` loss keys getting added in `training_epoch_end` when using manual optimization and not returning a loss ([#7772](https://github.com/PyTorchLightning/pytorch-lightning/pull/7772))


- Fixed print errors in `ProgressBar` when `trainer.fit` is not called ([#7674](https://github.com/PyTorchLightning/pytorch-lightning/pull/7674))


- Fixed formatting of info message when max training time reached ([#7780](https://github.com/PyTorchLightning/pytorch-lightning/pull/7780))
- Fixed LearningRateMonitor + BackboneFinetuning ([#7835](https://github.com/PyTorchLightning/pytorch-lightning/pull/7835))
- Minor improvements to `apply_to_collection` and type signature of `log_dict` ([#7851](https://github.com/PyTorchLightning/pytorch-lightning/pull/7851))
- Fixed docker versions ([#7834](https://github.com/PyTorchLightning/pytorch-lightning/pull/7834))
- Fixed sharded training check for fp16 precision ([#7825](https://github.com/PyTorchLightning/pytorch-lightning/pull/7825))
- Fixed support for torch Module type hints in LightningCLI ([#7807](https://github.com/PyTorchLightning/pytorch-lightning/pull/7807))

### Changed

- Fixed a bug where `precision=64` with `accelerator='ddp_spawn'` would throw a pickle error ([#6924](https://github.com/PyTorchLightning/pytorch-lightning/pull/6924))
- Move `training_output` validation to after `train_step_end` ([#7868](https://github.com/PyTorchLightning/pytorch-lightning/pull/7868))


## [1.3.4] - 2021-06-01
Expand All @@ -196,6 +31,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
- Fixed info message when max training time reached ([#7780](https://github.com/PyTorchLightning/pytorch-lightning/pull/7780))
- Fixed missing `__len__` method to `IndexBatchSamplerWrapper` ([#7681](https://github.com/PyTorchLightning/pytorch-lightning/pull/7681))


## [1.3.3] - 2021-05-27

### Changed
Expand All @@ -212,12 +48,6 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
- Fixed training loop total batch counter when accumulate grad batches was enabled ([#7692](https://github.com/PyTorchLightning/pytorch-lightning/pull/7692))


- Fixed a bug where checking `trainer.precision` changed to `'mixed'` when specifying 16 in trainer ([#7825](https://github.com/PyTorchLightning/pytorch-lightning/pull/7825))


- Fixed `LearningRateMonitor` keys not properly setup when running with `BackboneFinetuning` Callback ([#7835](https://github.com/PyTorchLightning/pytorch-lightning/pull/7835))


## [1.3.2] - 2021-05-18

### Changed
Expand Down
2 changes: 1 addition & 1 deletion pytorch_lightning/__about__.py
View file
Original file line number Diff line number Diff line change
@@ -1,7 +1,7 @@
import time

_this_year = time.strftime("%Y")
__version__ = '1.3.4'
__version__ = '1.3.5'
__author__ = 'William Falcon et al.'
__author_email__ = '[email protected]'
__license__ = 'Apache-2.0'
Expand Down

0 comments on commit 8a5a56b

Please sign in to comment.