Commit Graph

154 Commits

Author SHA1 Message Date
Rohit Gupta 60850ef510
fix overfit_batch sampler replacement logic (#10486)
Co-authored-by: thomas chaton <thomas@grid.ai>
2021-11-15 22:31:45 +00:00
Kaushik B 01cf7a2ac5
Deprecate `DistributedType` in favor of `StrategyType` (#10505) 2021-11-15 17:10:08 +00:00
thomas chaton 8d810d6144
Enable distributed training with CombinedDataLoader and max_size_cycle (#10374)
* solve combinedloader

* update

* update changelog

* update on comments

* resolve iterable dataset support

* update test description

* update

* update on comments

* update

* Accelerator auto

* Address review

* Refactor

Co-authored-by: Carlos Mocholi <carlossmocholi@gmail.com>
2021-11-09 20:06:10 +00:00
thomas chaton f1882aa69f
[Fault Tolerance] Don't check the len of a dataset, but its instance. (#10432) 2021-11-09 20:21:53 +05:30
Peter Dudfield ce3e63262a
Fix failure when `DataLoader(batch_size=None)` is passed (#10345)
* add test, + add change to data loading batch sample method

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Refactor and CHANGELOG

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
2021-11-04 12:46:57 +01:00
Ning f6ed0bd8ca
introduce has_len_all_ranks() to check the length of dataloader across ranks (#9827)
* introduce , udpate tests

* update CHANGELOG.md

* change staticmethod and hook attribute naming

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* fix typo

* remove non-essential comment

* fix merge error and comment format

* try to fix test_tpu.py failure

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* update on comments

* chlog

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* chlog

* update

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* try fix

* Revert back TPUSpawn changes

* Update test

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
Co-authored-by: thomas chaton <thomas@grid.ai>
Co-authored-by: Rohit Gupta <rohitgr1998@gmail.com>
Co-authored-by: Kaushik B <kaushikbokka@gmail.com>
2021-11-02 13:22:58 -04:00
Adrian Wälchli 9d136a9fc5
Lightning Lite core and tests (#10175) 2021-10-29 21:46:39 +00:00
Danielle Pintz 1f7bd6650c
Mark accelerator connector as protected (#10032) 2021-10-25 19:24:54 +00:00
Danielle Pintz e94dcf6936
Mark `trainer.data_connector` as protected (#10031)
Co-authored-by: tchaton <thomas@grid.ai>
2021-10-25 12:29:09 +01:00
Kaushik B 56bc55db71
Update strategy flag in docs (#10000)
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
Co-authored-by: Carlos Mocholi <carlossmocholi@gmail.com>
2021-10-20 21:02:53 +05:30
Adrian Wälchli 2c16f1d6b9
remove dataloader patching on the LightningModule (#9764)
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: rohitgr7 <rohitgr1998@gmail.com>
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
2021-10-20 15:23:20 +02:00
Rohit Gupta 4dc32ad7db
Fix logic to check for spawn in worker_check (#9902)
* fix

* update tests

* chlog

* skip windows
2021-10-18 13:02:46 +00:00
Carlos Mocholí 19008ce98f
IPU hotfix for #9721 (#9759) 2021-09-29 15:36:39 +02:00
thomas chaton fa44dbcd9e
[Refactor] Simplify data loading logic around replacing sampler to prevent confusion (#9721)
Co-authored-by: Carlos Mocholi <carlossmocholi@gmail.com>
2021-09-28 17:04:02 +00:00
Adrian Wälchli f74eb58493
remove `InternalDebugger` (#9680)
* wip

* reset _notebooks

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* reset _notebooks

* testing with mock

* update test with mock

* update test

* update tests

* update test

* remove track_load_dataloader_calls

* update last test

* remove unused imports

* remove InternalDebugger

* update changelog

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Carlos Mocholi <carlossmocholi@gmail.com>
2021-09-27 09:33:45 -04:00
four4fish 15cd6ad45b
Call TrainingTypePlugin collective functions directly instead of going through the Accelerator (#9677)
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
Co-authored-by: ananthsub <ananth.subramaniam@gmail.com>
Co-authored-by: Rohit Gupta <rohitgr1998@gmail.com>
2021-09-27 14:52:57 +02:00
Adrian Wälchli d67aff7494
remove `InternalDebugger.track_load_dataloader_call` (#9675)
* wip

* reset _notebooks

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* reset _notebooks

* testing with mock

* update test with mock

* update test

* update tests

* update test

* remove track_load_dataloader_calls

* update last test

* remove unused imports

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
2021-09-24 15:37:36 +02:00
Carlos Mocholí f239b96320
Fix `replace_sampler` missing the batch size under specific conditions (#9367) 2021-09-08 12:27:59 +02:00
Jirka Borovec 6e124e7207
CI: precommit - docformatter (#8584)
* CI: precommit - docformatter
* fix deprecated

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
2021-09-06 12:49:09 +00:00
Carlos Mocholí e1442d247e
Always use `trainer.call_hook` (#8498) 2021-08-20 18:22:03 +02:00
Adrian Wälchli 522df2b89b
3/n integrate new LightningDataFetcher into loop (#8953)
Co-authored-by: tchaton <thomas@grid.ai>
Co-authored-by: Kaushik B <45285388+kaushikb11@users.noreply.github.com>
2021-08-17 21:42:22 +00:00
thomas chaton 19136ac847
[Feat] 2/n Add Fault Tolerant Training to LightningFetcher (#8891)
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
Co-authored-by: Kaushik B <45285388+kaushikb11@users.noreply.github.com>
2021-08-17 16:32:43 +00:00
Carlos Mocholí 93ab24d1ee
Replace DataLoader sampler once for IPUs (#8858) 2021-08-16 11:28:05 +02:00
Carlos Mocholí a1264a6850
Automatic string fixes (#8886) 2021-08-13 14:28:14 +00:00
Carlos Mocholí ed13040729
Connect the model to the training type plugin at the start of run (#8536) 2021-08-04 17:43:34 +02:00
thomas chaton 567e905ead
update logic to inject FastForwardSampler / CaptureIterableDataset 2/n (#8366)
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
Co-authored-by: Justus Schock <justus.schock@posteo.de>
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
2021-08-02 20:52:06 +00:00
Carlos Mocholí a64cc37394
Replace `yapf` with `black` (#7783)
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
2021-07-26 13:37:35 +02:00
Carlos Mocholí 6dbdf438e8
Support `DataLoader`s with missing arguments in `replace_sampler` (#8519)
* Support `DataLoader`s with missing arguments in `replace_sampler`

* Fix for multiprocessing context

* Fixes and test improvements

* Fixes and test improvements

* Fixes and test improvements

* Test any variadic name

* Update CHANGELOG

* Make sure extra attributes can be present

* Skip on old Windows

* Update pytorch_lightning/trainer/data_loading.py

* Update pytorch_lightning/trainer/data_loading.py

* Check is dataloader

* Typo
2021-07-26 10:04:21 +02:00
Carlos Mocholí f7027a8701
Remove `torch >= 1.6` checks (#8523) 2021-07-23 04:03:20 +00:00
Carlos Mocholí e345469a87
Delete legacy dataloader processing utility (#8439) 2021-07-19 11:25:04 +00:00
thomas chaton 374fae59ef
[Feat] Add utilities for CombinedLoader state dict and dataloader state dict 1/n (#8364)
Co-authored-by: Justus Schock <12886177+justusschock@users.noreply.github.com>
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
Co-authored-by: Justus Schock <justus.schock@posteo.de>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
2021-07-19 09:56:57 +00:00
deepsource-autofix[bot] cbf71d0a14
Remove unnecessary comprehension (#8405) 2021-07-19 08:30:24 +00:00
thomas chaton 24db914093
Support state restoration of logged results 2/2(#7966)
Co-authored-by: Carlos Mocholi <carlossmocholi@gmail.com>
Co-authored-by: Ethan Harris <ewah1g13@soton.ac.uk>
Co-authored-by: Jirka Borovec <Borda@users.noreply.github.com>
2021-06-25 19:16:11 +00:00
Adrian Wälchli 971908a1aa
Loop Refactor 1/N - Training Loop (#7871)
Co-authored-by: Carlos Mocholi <carlossmocholi@gmail.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: tchaton <thomas@grid.ai>
Co-authored-by: Justus Schock <justus.schock@rwth-aachen.de>
Co-authored-by: Justus Schock <justus.schock@posteo.de>
Co-authored-by: Ethan Harris <ewah1g13@soton.ac.uk>
Co-authored-by: Jirka Borovec <Borda@users.noreply.github.com>
Co-authored-by: Justus Schock <12886177+justusschock@users.noreply.github.com>
2021-06-15 12:55:06 +00:00
Adrian Wälchli 20f37b85b6
add warning when Trainer(log_every_n_steps) not well chosen (#7734)
* add warning

* update changelog

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* logger check

* add docstring for test

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Nicki Skafte <skaftenicki@gmail.com>
2021-06-07 12:40:43 +00:00
Sean Naren 6388c29e87
[IPU] Add reset dataloader hooks to training type plugin 3/n (#7861)
* Add hooks

* Add tests for hooks

* Add changelog

* Test changes, add typing
2021-06-07 10:37:09 +00:00
Xinyao(Alvin) Sun 7e2f7e956b
fix: improve UserWarning message (#7685)
* fix: improve UserWarning message
when both overfit and training dtaloader shuffling are enabled

fixes issue: #7656

* chore: update changelog

* Polish userwarning msg in pytorch_lightning/trainer/data_loading.py

Co-authored-by: ananthsub <ananth.subramaniam@gmail.com>

* shuffling typo

* Update CHANGELOG.md

Co-authored-by: ananthsub <ananth.subramaniam@gmail.com>
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
2021-05-25 17:35:15 +00:00
Adrian Wälchli 6e6e29af49
remove trainer hidden state | sanity refactor [2 / n] (#7507) 2021-05-17 08:57:15 +01:00
Carlos Mocholí 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
ananthsub b7a444883c
Remove model.trainer call inside of dataloading mixin (#7317)
* Update data_loading.py

* Update data_loading.py

Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
2021-05-03 13:53:54 -07:00
thomas chaton e147127c0e
[feat] Add better support for predict + ddp 2/3 (#7215)
* wip

* update

* update

* update

* update

* update

* typo

* update on comments

* update

* update

* update

* update

* update changelog

* update

* Fix merge

* Fix merge

* move code

* resolve test

* add extra test

* add an extra test

* update on comments

* add typing

* resolve flake8

* Refactor and Docs

* Fix tests

* Fix tests

* Fix tests

* Duplicate

* Fix tests

* resolve bug

* update

* update on comments

* update

* update changelog

* update

* update

* remove tpu

* resolve flake8

* update on comments

* update on comments

* update on comment

* resolve flake8

* add a cpu test for predict

* add None test

* update

* Update CHANGELOG.md

Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>

* resolve tests

Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
2021-04-27 08:46:45 -04:00
thomas chaton 9beec26c3e
[bugfix] Add support for CombinedLoader in validation with ddp (#7102)
* add test

* add changelog

* resolve flake8

* remove print
2021-04-20 08:22:02 +00:00
Adrian Wälchli 60c1c8fe83
Auto-set `DataLoader.worker_init_fn` with `seed_everything` (#6960)
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
Co-authored-by: ananthsub <ananth.subramaniam@gmail.com>
2021-04-19 16:28:37 +02:00
Adrian Wälchli e9fca760ac
Set `DistributedSampler` seed if `seed_everything` was called (#7024)
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
2021-04-19 14:50:31 +01:00
Ethan Harris f645df5e9a
Add typings for evaluation_loop.py and remove some dead code (#7015) 2021-04-15 07:36:04 +00:00
Carlos Mocholí b85cfbe8f3
Run ddp_spawn dataloader checks on windows (#6930) 2021-04-09 19:06:47 +00:00
Roger Shieh e35192dfcd
Update `DataLoader.persistent_workers` warnings in ddp_spawn (#6762)
Co-authored-by: Carlos Mocholi <carlossmocholi@gmail.com>
2021-04-09 11:38:13 +02:00
Akihiro Nitta ac60536818
Follow E231 [flake8] (#6110)
* Remove E231 from ignore list

* Follow E231

* Update pytorch_lightning/trainer/data_loading.py
2021-03-24 12:50:50 +01:00
thomas chaton 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
Rohit Gupta dcd9dd8338
Update docs for limit_predict_batches (#6507)
* add docs and minor updates

* docs

* fraction
2021-03-14 09:09:58 +00:00