2242423b75
refactor accelerator teardown -> training type plugin teardown ( #7579 )
2021-05-22 13:19:24 -07:00
7ca41734da
Add `dataloader_idx` to batch transfer hooks ( #6241 )
...
* replace with kwargs
* chlog
* fix
* add test
* fix
* device
* deepspeed
* pep
* optional
* docs
* bc
* comments
* pep
* mypy
* pep
* Apply suggestions from code review
* kwargs
* docs
* .
* .
* 1.3 -> 1.4
* kwargs -> step_kwargs
2021-05-13 23:03:55 +05:30
8538c1f61e
Accelerator model state dict ( #7474 )
...
* Fix some test errors
Summary:
Test Plan:
Reviewers:
Subscribers:
Tasks:
Tags:
* checkpoint consolidation
* Update ddp_spawn.py
* Update test_metric_result_integration.py
* Update test_results.py
* Update utils.py
* Update utils.py
* Update test_all_gather_grad.py
* Update test_all_gather_grad.py
* Update test_results.py
* Revert "Update test_results.py"
This reverts commit 9d4a2b891d#1 from shuyingsunshine21/shuyingsunshine21-checkpoint_consolidate"
This reverts commit c5053da7890d23d75bc90d23d75bc970fe5da9c6a9aae99f6eea74906878bf70e431b3f17210183b536c1323b03a9fde915a7a369f47e18222dc98ea6c095b2370250d0aaaa28651d54d79dcdcd29731https://pre-commit.ci
* Address code review
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: SeanNaren <sean@grid.ai>
2021-05-11 16:39:04 +01:00
6bc616d78f
fix display bug ( #7395 )
2021-05-10 11:26:15 +08:00
8208c330eb
Use `torch.nn.utils.clip_grad_norm_` and add `clip_grad_by_value` support for TPU ( #7025 )
...
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
Co-authored-by: Kaushik B <45285388+kaushikb11@users.noreply.github.com>
2021-05-07 16:41:39 +00:00
b181b8c646
release 1.3.0 ( #7404 )
...
* v1.3.0
* ci event
* chlog
* badge
* formatting
2021-05-06 15:05:35 -04:00
c3fc0313ef
Updating docs and error message: half precision not available on CPU ( #7384 )
...
* Updating docs and error message to specify that half precission not available on CPU
* update messages
Co-authored-by: Martin Kristiansen <martinkristiansen@sixgill.com>
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
Co-authored-by: jirka <jirka.borovec@seznam.cz>
2021-05-06 09:05:50 +00:00
6104a6316a
[1/2] Deprecate `outputs` in `on_train_epoch_end` hooks ( #7339 )
...
* Remove outputs from on_train_epoch_end
* iterate
* Update callback_hook.py
* update
* Update training_loop.py
* Update test_training_loop.py
* early stop?
* fix
* update tests
* Update test_hooks.py
* Update pytorch_lightning/trainer/callback_hook.py
Co-authored-by: Ethan Harris <ewah1g13@soton.ac.uk>
* Update pytorch_lightning/trainer/training_loop.py
Co-authored-by: Ethan Harris <ewah1g13@soton.ac.uk>
* Update trainer.py
* Update pytorch_lightning/trainer/trainer.py
Co-authored-by: Jirka Borovec <Borda@users.noreply.github.com>
Co-authored-by: Ethan Harris <ewah1g13@soton.ac.uk>
Co-authored-by: Jirka Borovec <Borda@users.noreply.github.com>
2021-05-05 17:18:16 +02:00
98670c83a9
Deprecate`truncated_bptt_steps` flag on Trainer in favor of same setting on the LightningModule ( #7323 )
...
* deprecate-tbptt-trainer
* Update CHANGELOG.md
* Update lightning.py
* test
* Update lightning.py
* Update training_loop.py
* Update training_loop.py
* Update lightning.py
* Update training_loop.py
* Update training_loop.py
* update docs
* Update accelerator.py
* Update accelerator.py
* more docs
* tweaks
* chlog
* comments
Co-authored-by: Jirka Borovec <Borda@users.noreply.github.com>
2021-05-05 11:21:00 +01:00
8c0ea92af2
`TrainerState` refactor [5/5] ( #7173 )
...
* `TrainerState` refactor
* flake8
* Update finished check
* Test cleanup
* Fix tests
* Fixes
* Reorder
* flake8
* Update CHANGELOG
* Better docs
* Better docs
* Remove default
* Update tests
* Bad merge
2021-05-04 12:50:56 +02:00
39274273a4
Update accelerator.py ( #7318 )
2021-05-03 11:17:26 -04:00
e0c64f0ef6
Fix Adagrad optimizer not working with DDP/GPU ( #7277 )
...
Co-authored-by: ananthsub <ananth.subramaniam@gmail.com>
Co-authored-by: thomas chaton <thomas@grid.ai>
2021-05-03 03:57:17 +05:30
16d6c9828d
[bugfix] Apex never instantiated. ( #7274 )
...
* update
* update
* update apex
* update
* update
* update
* remove test.py
* update
* update
* update on comments
* update changelog
* update
* update
* typo
2021-04-30 13:16:28 -04:00
94fcaaf5d7
Add `debug` flag to TPU Training Plugins (PT_XLA_DEBUG) ( #7219 )
2021-04-27 20:34:25 +00:00
ca6c87ffbe
Add back `clip_gradients(model)` ( #7231 )
2021-04-27 11:34:02 +00:00
6be0a859db
Update teardown for TPU acc ( #7211 )
2021-04-26 13:30:46 +01:00
bc3f08b0e3
[fix] Add barrier to accelerator's teardown ( #6814 )
2021-04-26 09:23:29 +00:00
f58865aada
Properly set `LightningModule.device` after model replacement ( #7188 )
...
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
2021-04-23 16:36:52 +02:00
013756404b
[bugfix] Add set_default_tensor_type to torch.DoubleTensor with precision=64 ( #7108 )
...
* update
* Update pytorch_lightning/plugins/precision/double.py
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
* Update pytorch_lightning/plugins/precision/double.py
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
* Update pytorch_lightning/plugins/precision/double.py
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
* resolve tests
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
2021-04-20 15:25:37 +00:00
f168a535ca
Add MpModelWrapper in TPU Spawn ( #7045 )
...
Co-authored-by: tchaton <thomas@grid.ai>
Co-authored-by: Carlos Mocholi <carlossmocholi@gmail.com>
2021-04-20 13:05:27 +00:00
898ec8a94a
Create pytorch_lightning/utilities/types.py ( #7048 )
2021-04-19 14:43:16 +02:00
7b0b0d2844
update ( #7056 )
2021-04-16 21:22:19 +01:00
f29ecbfd90
Typing for accelerators and plugins ( #7022 )
2021-04-15 16:48:16 +00:00
f645df5e9a
Add typings for evaluation_loop.py and remove some dead code ( #7015 )
2021-04-15 07:36:04 +00:00
24d0295ff1
Fix the `gradient_clip_algorithm` has no effect issue. ( #6928 )
2021-04-14 14:17:06 +05:30
1b3e4f9fb9
Fix sync_dist for tpus ( #6950 )
2021-04-13 14:17:15 +05:30
20ff50caa6
Accelerator API docs ( #6936 )
...
Co-authored-by: ananthsub <ananth.subramaniam@gmail.com>
Co-authored-by: William Falcon <waf2107@columbia.edu>
Co-authored-by: Jirka Borovec <Borda@users.noreply.github.com>
2021-04-10 12:25:07 +05:30
55525031c6
Fix TPU Spawn gather ( #6896 )
2021-04-09 18:30:59 +05:30
7f6154fcad
Add `Trainer(gradient_clip_algorithm='value'|'norm')` ( #6123 )
...
* add changelog
* add clip by value
* fix bug in training tricks.rst
* fix bug in trainer.rst
* Update trainer.rst
* Update trainer.rst
* Update CHANGELOG.md
Co-authored-by: Jirka Borovec <Borda@users.noreply.github.com>
* Update pytorch_lightning/plugins/precision/deepspeed_precision.py
Co-authored-by: Jirka Borovec <Borda@users.noreply.github.com>
* Update pytorch_lightning/utilities/enums.py
Co-authored-by: Jirka Borovec <Borda@users.noreply.github.com>
* yapf formatting
* update training tricks
* update based on comment
* update based on comment
* Update pytorch_lightning/trainer/trainer.py
Co-authored-by: ananthsub <ananth.subramaniam@gmail.com>
* update based on comment
* pep8
* mypy
* mypy
* Update docs/source/advanced/training_tricks.rst
Co-authored-by: thomas chaton <thomas@grid.ai>
* Update sharded_native_amp.py
* Update test_sharded_parity.py
* update test codes
* Update test_tpu.py
* Update pytorch_lightning/trainer/connectors/training_trick_connector.py
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
* Update test_trainer.py
* Update enums.py
* Update enums.py
* add super-class initialization to precision plugins.
* add clip_grad horovod cpu test
* add clip_grad horovod cpu test
* use subprocess check_call
* change order of horovod tests
* set max_epochs 2 in horovod test
* remove clip_grad_val test from horovod-cpu
* remove "type: ignore"
* divide clip grad val test in horovod
* update based on comments
* add super-class initialization to precision plugins.
* bugfix
* bugfix
* revert some changes
* revert some changes
* Update tests/models/test_horovod.py
* merge master
* Delete signature test
No point in testing a signature
Co-authored-by: Jirka Borovec <Borda@users.noreply.github.com>
Co-authored-by: thomas chaton <thomas@grid.ai>
Co-authored-by: ananthsub <ananth.subramaniam@gmail.com>
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
Co-authored-by: Jirka Borovec <jirka.borovec@seznam.cz>
2021-04-06 08:27:37 -05:00
bb9ace4333
[typing] Add typehint for broadcast in training type plugin ( #6777 )
...
* Update training_type_plugin.py
* Update accelerator.py
* Update pytorch_lightning/plugins/training_type/training_type_plugin.py
Co-authored-by: Akihiro Nitta <nitta@akihironitta.com>
Co-authored-by: Akihiro Nitta <nitta@akihironitta.com>
2021-04-02 20:55:34 +02:00
a72a7992a2
Update clip gradients signature for precision plugins ( #6764 )
2021-03-31 17:06:48 +05:30
1302766f83
DeepSpeed ZeRO Update ( #6546 )
...
* Add context to call hook to handle all modules defined within the hook
* Expose some additional parameters
* Added docs, exposed parameters
* Make sure we only configure if necessary
* Setup activation checkpointing regardless, saves the user having to do it manually
* Add some tests that fail currently
* update
* update
* update
* add tests
* change docstring
* resolve accumulate_grad_batches
* resolve flake8
* Update DeepSpeed to use latest version, add some comments
* add metrics
* update
* Small formatting fixes, clean up some code
* Few cleanups
* No need for default state
* Fix tests, add some boilerplate that should move eventually
* Add hook removal
* Add a context manager to handle hook
* Small naming cleanup
* wip
* move save_checkpoint responsability to accelerator
* resolve flake8
* add BC
* Change recommended scale to 16
* resolve flake8
* update test
* update install
* update
* update test
* update
* update
* update test
* resolve flake8
* update
* update
* update on comments
* Push
* pull
* Update pytorch_lightning/plugins/training_type/deepspeed.py
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
* Update pytorch_lightning/plugins/training_type/deepspeed.py
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
* update
* Apply suggestions from code review
* Swap to using world size defined by plugin
* update
* update todo
* Remove deepspeed from extra, keep it in the base cuda docker install
* Push
* pull
* update
* update
* update
* update
* Minor changes
* duplicate
* format
* format2
Co-authored-by: SeanNaren <sean@grid.ai>
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
Co-authored-by: Sean Naren <sean.narenthiran@gmail.com>
Co-authored-by: Carlos Mocholi <carlossmocholi@gmail.com>
Co-authored-by: Jirka Borovec <Borda@users.noreply.github.com>
Co-authored-by: Jirka Borovec <jirka.borovec@seznam.cz>
2021-03-30 13:39:02 -04:00
f79a13e495
[Model Parallel] Add configure sharded model hook ( #6679 )
...
* Add base hook for model parallel
* fix callback signature
* Simplify hook
* Add hook logic
* add tests
* add property setter
* add logic for being called once
* Update changelog
* Fix
* fix return type
* fix lambda callback test
* Fix tests
* Apply code suggestions
* add logic for setup_optimizers_predispatch
* add common dummy model
* Swap call order
* Remove test that isn't needed anymore
* Update tests
* Add a bit more doc
* Few code review fixes
* Update pytorch_lightning/accelerators/accelerator.py
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
* Change hook name
* Fix test
* Test setup hook, refactor names
* Swap call order of callbacks and model initialization
* Change name of context manager
Co-authored-by: SeanNaren <sean@grid.ai>
Co-authored-by: Sean Naren <sean.narenthiran@gmail.com>
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
2021-03-29 14:50:51 -06:00
646cf2f7d4
[refactor] Move save_function to accelerator 1/n [DeepSpeed] ( #6689 )
...
* move save_checkpoint responsability to accelerator
* update
2021-03-29 21:02:37 +02:00
d471fa30b3
add copyr ( #6661 )
2021-03-24 14:29:46 +01:00
fd5cb7fcc3
Add PyTorch 1.8 Profiler 5/5 ( #6618 )
...
* Refactor profilers
* Update PassThrough
* WIP - This is broken and will change
* Update pytorch_lightning/profiler/pytorch.py
Co-authored-by: thomas chaton <thomas@grid.ai>
* resolve tests
* resolve tests
* find output
* try something
* update
* add support for test and predict
* update
* update
* use getattr
* test
* test
* update
* tests
* update
* update
* update
* update
* update
* remove file
* update
* update
* update
* update
* update
* test
* update#
* update
* update tests
* update
* add suport for 1.8
* rename records
* add support for 1.8
* update
* resolve flake8
* resolve test
* Refactor basic profilers
* Fixes
* Unused import
* Introduce setup
* Profile on all ranks. Print to stdout on 0
* Introduce dirpath + filename
* CHANGELOG
* Add tests. Address comments
* add `on_run_stage_setup`
* add on_run_stage_setup function
* update
* add test for RegisterRecordFunction
* update lightnng flow direction
* move variable to private
* remove trace
* Undo code that should be in 3/4
* Multi-stage multi-rank
* 2/5 changes
* Pass stage in __del__
* Remove TODOs
* Describe on_evaluation_end. Add tests
* Typo
* Address comments
* deepcopy tests
* Advanced teardown
* Fix teardown test
* Fix tests
* Minor change
* Update CHANGELOG.md
* Fix test
* Quick fixes
* Fix 6522
* resolve ddp tests
* resolve tests
* resolve some tests
* update tests
* resolve tests
* update
* resolve tests
* resolve some tests
* Missed fixes from 3/5
* Fixes
* resolve some tests
* resolve test for 1.7.1
* Broken refactor
* Missed stage
* Minor changes
* resolve tests
* Update CHANGELOG
* resolve bug
* remove print
* Typo
* Cleanup
* resolve ddp test
* remove barrier
* update profiler
* update
* Smaller model
* update
* resolve tests
* update
* Minor changes. CHANGELOG
* Minimize diff
* update to 1.8.1
* RunIf. Extra code. Check segfault
* resolve tests
* Typo. Bad merge
* Fixing a bad merge
* replace for kineto
* Update pytorch_lightning/profiler/pytorch.py
Co-authored-by: ananthsub <ananth.subramaniam@gmail.com>
* Update pytorch_lightning/profiler/pytorch.py
Co-authored-by: ananthsub <ananth.subramaniam@gmail.com>
* Minor changes
* Bad merge
* Use lists for flexibility
* Use sets
* predict_step
* Ananth's suggestion
* update
* Docs
* Update pl_examples/basic_examples/profiler_example.py
Co-authored-by: Jirka Borovec <Borda@users.noreply.github.com>
* update example
* update example
Co-authored-by: Carlos Mocholi <carlossmocholi@gmail.com>
Co-authored-by: ananthsub <ananth.subramaniam@gmail.com>
Co-authored-by: Jirka Borovec <Borda@users.noreply.github.com>
2021-03-23 20:43:21 +00:00
3cf0c3117a
fix back-compatibility for Accel ( #6655 )
2021-03-23 17:41:36 +01:00
0995d30fab
Flash predict step ( #6577 )
...
* add predict_step
* Update predict_loop.py
* Update trainer.py
* Update trainer.py
* resolve bugs
* update
* update
* update
* resolve bug
* resolve some failing tests
* udpate tests
* update
* resolve tests
* add a test
* remove typo
* add a test for attachement
* update
* changed to on_train_dataloader
* remove __flash_special_attr__
* resolve tests
* update
* update
* update
* update on comments
* Update pytorch_lightning/trainer/data_loading.py
Co-authored-by: Jirka Borovec <Borda@users.noreply.github.com>
Co-authored-by: Justus Schock <12886177+justusschock@users.noreply.github.com>
Co-authored-by: Jirka Borovec <Borda@users.noreply.github.com>
2021-03-23 11:13:13 -04:00
58c9fa7edb
Allow training type plugin to delay optimizer creation (FSDP 2/n) ( #6331 )
...
* Allow training_type_plugin to delay optimizer configure
* Add missing references to trainer, add a CPU accelerator based test
2021-03-22 11:43:53 +00:00
87c03b1038
Update Gradient Clipping for TPU Accelerator ( #6576 )
2021-03-20 01:02:57 +05:30
983a888f49
Fix all_gather for tpu_cores=8 ( #6587 )
2021-03-19 21:56:58 +05:30
4e9b453854
[Fix] Move init dist connection into the setup function ( #6506 )
...
* Move connection setup into the setup function. Call setup hook after we set up the accelerator
* Added CHANGELOG.md
* fix setup order in callback test
* fix input arguments in test
* Mock distributed function, remove protection to turn into training type hook
* Remove import
* Add missing mock, ensure custom plugin does not create children process
* Skip test on windows
* Update deepspeed to init connection in setup
* Do not initialize distributed module
* Move DeepSpeed tests to special tests since dist communication is being set up
* Special the test to see if this fixes CI
* Delete accelerator connector test to see if its causing build to fail
* Delete deepspeed test
* Revert "Delete accelerator connector test to see if its causing build to fail"
This reverts commit edde60b89d317429b5450def
2021-03-18 14:33:39 -07:00
6453091b8a
Prune metrics base classes 2/n ( #6530 )
...
* base class
* extensions
* chlog
* _stable_1d_sort
* _check_same_shape
* _input_format_classification_one_hot
* utils
* to_onehot
* select_topk
* to_categorical
* get_num_classes
* reduce
* class_reduce
* tests
2021-03-15 19:28:18 +00:00
0544efd453
[bug] Update broadcast + reduce decision ModelCheckpoint] ( #6410 )
...
* resolve bug
* update
* update changelog
* update PR
* Update pytorch_lightning/trainer/connectors/logger_connector/epoch_result_store.py
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
* add todo
* resolve issues
* resolve flake8
* update
* add coverage for reduce
* wip
* restore back to brodbact
* remove test.py
* resolve flake8
* update
* check world size
* resolve test
* update
* use pytorch version when defined
* update on comments
* update on comments
* flake8
* resolve bugs
* Update CHANGELOG.md
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
* update
* update
* update
* update
* remove test
* update
* resolve flake8
* update
* update
* update
* proxy
* update
* update
* resolve typo
* prune
* update parallel
* update
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
2021-03-14 17:14:27 +00:00
c53edce1a1
Disable batch transfer in DP mode ( #6098 )
...
* add exceptions and test
* hook
* fix
* clean up
* clean up
* regex
* regex
* docs
* rev
* comment and docs
* chlog
* Apply suggestions from code review
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
* Apply suggestions from code review
Co-authored-by: chaton <thomas@grid.ai>
* Monkey-patch device count
* docs
* pep
* api_change
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
Co-authored-by: chaton <thomas@grid.ai>
2021-03-11 10:51:10 -05:00
d0596fac94
Refactor RunningStage usage in advance of implementing Trainer.validate() ( #4945 )
...
* Update code
Co-authored-by: EliaCereda
* More property updates
* Move properties. Introduce trainer._fitting
* Use trainer.fitting
* Fix reset dataloaders
* Unused code
* RunningStage.SANITY_CHECKING
* Use setters
* Fix bugs
* Fix bugs
* TrainerState.{FITTING,VALIDATING,TESTING,PREDICTING,TUNING}
* Fix bugs
* Fix bugs
* Fix tests
* Update CHANGELOG. Add deprecation warning. Fix tests
* Unused imports
* Optional trainer
* More deprecation. More refactoring
* Correct version
* Use properties
* Address comments
* flake8
* Missed renamings
* Typo
* is -> ==
It is recommended to use for Enums since they are singletons, however, since the LightningEnum subclasses str, it's not a good idea in case a user sets the state/stage with a str
* Also for tests
* Typo
* Address @tchaton's comments
* PEP8
* Correct property
* Update CHANGELOG
* Apply suggestions from code review
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
* Update pytorch_lightning/trainer/trainer.py
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
* Remove called sanity check
Co-authored-by: Carlos Mocholi <carlossmocholi@gmail.com>
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
2021-03-06 12:40:19 +00:00
49c579f1f0
docstring changes in accelerators ( #6327 )
...
* docstring changes in accelerators
* docstrings moved
* whitespaces removed
* PEP8 correction[1]
2021-03-04 20:21:53 +00:00
d01e8fdc86
[fix] Use training type plugin hook when saving (FSDP 1/n) ( #6321 )
...
* Rely on training type plugin when saving
* Add better typing to training type plugin
2021-03-04 18:09:33 +00:00
484dce11ec
[bugfix] TPU + all_gather + SingleTPU shouldn't call xm.all_gather ( #6296 )
...
* resolve an issue with TPU
* update
* add changelog
2021-03-03 08:54:20 -05:00
1aac481957
[bugfix] TPU test hangs to barrier on 1 process ( #6272 )
...
* update
* resolve flake8
* update
* update
* update changelog
* update
* resolve flake8
Co-authored-by: Your Name <you@example.com>
2021-03-02 18:01:35 -05:00
shuyingsunshine21