[WIP] : Fix resume issues with combined streaming dataset in dataloader #362

bhimrazy · 2024-09-03T19:11:29Z

Before submitting

Was this discussed/agreed via a Github issue? (no need for typos and docs improvements)
Did you read the contributor guideline, Pull Request section?
Did you make sure to update the docs?
Did you write any new necessary tests?

How does this PR impact the user?

Currently, users experience issues when attempting to resume a combined streaming dataset with the streaming dataloader, as saving and restoring checkpoints doesn’t work as expected. This PR addresses the root cause of the error, enabling successful checkpoint resuming of the dataloader, ensuring smoother and more reliable training workflows.

What does this PR do?

Fixes #331.

Fixed IndexError when loading dataloader state before any iteration.
Enabled resuming dataloader states for combined datasets (non-weighted) [ In progress ].

PR review

Anyone in the community is free to review the PR once the tests have passed.
If we didn't discuss your PR in GitHub issues there's a high chance it will not be merged.

Did you have fun?

Make sure you had fun coding 🙃

…eights.

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

codecov · 2024-09-05T04:12:20Z

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 78%. Comparing base (92df8af) to head (242a13c).

Additional details and impacted files

@@         Coverage Diff         @@
##           main   #362   +/-   ##
===================================
  Coverage    78%    78%           
===================================
  Files        34     34           
  Lines      5016   5020    +4     
===================================
+ Hits       3929   3934    +5     
+ Misses     1087   1086    -1

src/litdata/streaming/dataloader.py

tests/streaming/test_dataloader.py

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

bhimrazy · 2024-09-09T12:15:20Z

Combined Dataset (no weights): Resuming from the complete last epoch iteration is working now,
but no luck so far with resuming from a partial last epoch yet (looking into it further.)

Separated because the states were somehow getting accumulated from last test, leading to some weird numbers of samples yielded tests/streaming/test_combined.py:974: AssertionError ----------------------------- Captured stdout call ----------------------------- {'dataset': {'0': {'num_samples_yielded': 3, 'num_workers': 4, 'batch_size': 4, 'current_epoch': 1, 'input_dir_path': '/tmp/pytest-of-runner/pytest-0/test_combined_dataset_dataload0/dataset_0', 'input_dir_url': None, 'item_loader': None, 'drop_last': False, 'seed': 42, 'world_size': 1, 'shuffle': True, 'subsampled_files': ['chunk-0-0.bin'], 'region_of_interest': [(0, 50)]}, '1': {'num_samples_yielded': 1, 'num_workers': 4, 'batch_size': 4, 'current_epoch': 1, 'input_dir_path': '/tmp/pytest-of-runner/pytest-0/test_combined_dataset_dataload0/dataset_1', 'input_dir_url': None, 'item_loader': None, 'drop_last': False, 'seed': 42, 'world_size': 1, 'shuffle': True, 'subsampled_files': ['chunk-0-0.bin'], 'region_of_interest': [(0, 50)]}}, 'current_epoch': 1, 'latest_worker_idx': 2, 'num_samples_yielded': {0: [15, 25], 1: [16, 20], 2: [16, 20], 3: [16, 16]}} =========================== short test summary info ============================

…razy/litdata into fix/combined-dataset-loading-states

deependujha · 2024-09-19T09:41:06Z

hi @bhimrazy
What's the current update?

bhimrazy · 2024-09-19T10:08:00Z

hi @bhimrazy What's the current update?

Hi @deependujha

I'm still facing some issues with an IndexError when loading states from the last partial epoch. This usually only happens when the num of samples exceeds than the actual available samples.

E       IndexError: Caught IndexError in DataLoader worker process 0.
E       Original Traceback (most recent call last):
E         File "/opt/hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/torch/utils/data/_utils/worker.py", line 253, in _worker_loop
E           fetcher = _DatasetKind.create_fetcher(dataset_kind, dataset, auto_collation, collate_fn, drop_last)
E         File "/opt/hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/torch/utils/data/dataloader.py", line 80, in create_fetcher
E           return _utils.fetch._IterableDatasetFetcher(dataset, auto_collation, collate_fn, drop_last)
E         File "/opt/hostedtoolcache/Python/3.9.19/x64/lib/python3.9/site-packages/torch/utils/data/_utils/fetch.py", line 22, in __init__
E           self.dataset_iter = iter(dataset)
E         File "/home/runner/work/litdata/litdata/src/litdata/streaming/combined.py", line 160, in __iter__
E           self._iterator = _CombinedDatasetIterator(
E         File "/home/runner/work/litdata/litdata/src/litdata/streaming/combined.py", line 208, in __init__
E           self._dataset_iters = [iter(dataset) for dataset in datasets]
E         File "/home/runner/work/litdata/litdata/src/litdata/streaming/combined.py", line 208, in <listcomp>
E           self._dataset_iters = [iter(dataset) for dataset in datasets]
E         File "/home/runner/work/litdata/litdata/src/litdata/streaming/dataset.py", line 240, in __iter__
E           self._resume(workers_chunks, workers_intervals)
E         File "/home/runner/work/litdata/litdata/src/litdata/streaming/dataset.py", line [312](https://github.com/Lightning-AI/litdata/actions/runs/10900666975/job/30248775938#step:5:313), in _resume
E           interval = self.worker_intervals[self.chunk_index]
E       IndexError: list index out of range

Initially, I encountered a separate error where the number of samples exceeded the actual count in state dict test. It seemed like the states were accumulating incorrectly between the tests, so I decided to separate the tests and then the states were fine as it should be.

I haven't had much time lately, but I plan to continue working on from this weekend.

gitguardian · 2024-09-19T10:11:47Z

⚠️ GitGuardian has uncovered 1 secret following the scan of your pull request.

Please consider investigating the findings and remediating the incidents. Failure to do so may lead to compromising the associated services or software components.

Since your pull request originates from a forked repository, GitGuardian is not able to associate the secrets uncovered with secret incidents on your GitGuardian dashboard.
Skipping this check run and merging your pull request will create secret incidents on your GitGuardian dashboard.

🔎 Detected hardcoded secret in your pull request

GitGuardian id	GitGuardian status	Secret	Commit	Filename
5685611	Triggered	Generic High Entropy Secret	`3762b11`	tests/streaming/test_resolver.py	View secret

🛠 Guidelines to remediate hardcoded secrets

Understand the implications of revoking this secret by investigating where it is used in your code.
Replace and store your secret safely. Learn here the best practices.
Revoke and rotate this secret.
If possible, rewrite git history. Rewriting git history is not a trivial act. You might completely break other contributing developers' workflow and you risk accidentally deleting legitimate data.

To avoid such incidents in the future consider

following these best practices for managing and storing secrets including API keys and other credentials
install secret detection on pre-commit to catch secret before it leaves your machine and ease remediation.

^{🦉 GitGuardian detects secrets in your source code to help developers and security teams secure the modern development process. You are seeing this because you or someone else with access to this repository has authorized GitGuardian to scan your pull request.}

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

bhimrazy · 2024-09-22T07:37:03Z

Getting close to it:

The test case seems to fail with an IndexError when the number of workers is greater than 2 and the iteration is stopped close to the midpoint of the dataloader length.

schopra8 · 2024-12-17T02:03:18Z

@bhimrazy Is this an active issue? I tried restarting training midway during an epoch last week and was able to continue training, when using a CombinedStreamingDataset.

bhimrazy · 2024-12-17T02:58:20Z

@bhimrazy Is this an active issue? I tried restarting training midway during an epoch last week and was able to continue training, when using a CombinedStreamingDataset.

Thank you, @schopra8, for bringing this to my attention.
There’s a test case in this PR that fails for the same issue. I’ll review it with the latest updates.

chore: Add tests for CombinedStreamingDataset in test_dataloader.py

0ba50c5

bhimrazy self-assigned this Sep 3, 2024

bhimrazy requested review from tchaton and awaelchli as code owners September 3, 2024 19:11

bhimrazy marked this pull request as draft September 3, 2024 19:11

Merge branch 'main' into fix/combined-dataset-loading-states

3756b9e

bhimrazy changed the title ~~[WIP] : Fix inconsistent dataloader states with combined dataset~~ [WIP] : Fix inconsistent dataloader states with combined streaming dataset Sep 3, 2024

bhimrazy and others added 4 commits September 4, 2024 01:44

Adds resuming for dataloading states for combined dataset case with w…

f8ed272

…eights.

update

8c53791

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

e193eb9

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

update

837efde

bhimrazy mentioned this pull request Sep 5, 2024

Failed to Resume Training w/ CombinedStreamingDataset #363

Open

Refactor dataloader to fix num_samples_yieled calculation

242a13c

bhimrazy changed the title ~~[WIP] : Fix inconsistent dataloader states with combined streaming dataset~~ [WIP] : Fix resume issues with combined streaming dataset in dataloader Sep 5, 2024

bhimrazy added 4 commits September 5, 2024 10:55

Adds more tests

dff88ca

removes the subtraction in the epoch

bc137a3

update initialize part

e38592a

updated epoch numbers

e095407

bhimrazy commented Sep 5, 2024

View reviewed changes

src/litdata/streaming/dataloader.py Show resolved Hide resolved

bhimrazy commented Sep 5, 2024

View reviewed changes

tests/streaming/test_dataloader.py Outdated Show resolved Hide resolved

bhimrazy and others added 9 commits September 5, 2024 12:35

format imports

97510e2

reverted current epoch

5c6925d

removed combined data test and moved to test_combined.py

8502bb1

reverted epcoh and also moved the test combined dataset

5cda0c6

Merge branch 'main' into fix/combined-dataset-loading-states

8cdafeb

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

560032c

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

revert epoch

f099242

updated current epoch

dc484a0

fix epoch number

53b360c

bhimrazy added 5 commits September 9, 2024 17:06

updated params

0c7cf3e

Update current_epoch in test_dataloader.py

8010d33

Update num_workers in test_combined.py

7355532

updated the conditions

d40e3ca

updated tests: added case for the complete last iteration

be77b58

bhimrazy added 6 commits September 9, 2024 18:32

Refactor test_combined.py to fix restore state issue

86ccd99

Merge branch 'main' into fix/combined-dataset-loading-states

206b574

Merge branch 'fix/combined-dataset-loading-states' of github.com:bhim…

da10c3f

…razy/litdata into fix/combined-dataset-loading-states

fix type errors

17de9e7

Refactor test_combined.py: Remove print statement

a05e6f1

Merge branch 'main' into fix/combined-dataset-loading-states

3762b11

bhimrazy and others added 5 commits September 22, 2024 11:04

Merge branch 'main' into fix/combined-dataset-loading-states

d506972

Adds conftest for combined dtaatset to reuse

868aa42

Simplified testes with parameterize to test for different conditions

357c29c

update test

ec3b840

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

0ee0617

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

Merge branch 'main' into fix/combined-dataset-loading-states

424bc64

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[WIP] : Fix resume issues with combined streaming dataset in dataloader #362

[WIP] : Fix resume issues with combined streaming dataset in dataloader #362

bhimrazy commented Sep 3, 2024 •

edited

Loading

codecov bot commented Sep 5, 2024 •

edited

Loading

bhimrazy commented Sep 9, 2024 •

edited

Loading

deependujha commented Sep 19, 2024

bhimrazy commented Sep 19, 2024

gitguardian bot commented Sep 19, 2024 •

edited

Loading

bhimrazy commented Sep 22, 2024

schopra8 commented Dec 17, 2024 •

edited

Loading

bhimrazy commented Dec 17, 2024

[WIP] : Fix resume issues with combined streaming dataset in dataloader #362

Are you sure you want to change the base?

[WIP] : Fix resume issues with combined streaming dataset in dataloader #362

Conversation

bhimrazy commented Sep 3, 2024 • edited Loading

How does this PR impact the user?

What does this PR do?

PR review

Did you have fun?

codecov bot commented Sep 5, 2024 • edited Loading

Codecov Report

bhimrazy commented Sep 9, 2024 • edited Loading

deependujha commented Sep 19, 2024

bhimrazy commented Sep 19, 2024

gitguardian bot commented Sep 19, 2024 • edited Loading

⚠️ GitGuardian has uncovered 1 secret following the scan of your pull request.

bhimrazy commented Sep 22, 2024

schopra8 commented Dec 17, 2024 • edited Loading

bhimrazy commented Dec 17, 2024

bhimrazy commented Sep 3, 2024 •

edited

Loading

codecov bot commented Sep 5, 2024 •

edited

Loading

bhimrazy commented Sep 9, 2024 •

edited

Loading

gitguardian bot commented Sep 19, 2024 •

edited

Loading

schopra8 commented Dec 17, 2024 •

edited

Loading