Document that max_steps is required for iterable train datasets#6333
Merged
Conversation
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
albertvillanova
force-pushed
the
train_dataset-max_steps
branch
from
July 9, 2026 07:03
472a99e to
64e5cc0
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR documents, in the trainer docstrings, that
max_stepsmust be set whentrain_datasetis an iterable (streaming) dataset.Motivation
When
train_datasetis anIterableDataset, it has no length, and Trainer raisesValueError: The train_dataset does not implement __len__, max_steps has to be specified.at initialization. The requirement was undocumented, so users hit the error with no upfront guidance. This is the TRL-side counterpart to the transformers PR that clarifies the same requirement:Solution
Add a note to the
train_datasetdocstring of the trainers that actually accept iterable datasets, stating thatmax_stepsmust be set in the training arguments. GRPO and RLOO are intentionally excluded, since they reject iterable datasets with NotImplementedError, so the note would not apply there.Changes
max_stepsnote to the train_dataset docstring in SFTTrainer, DPOTrainer, KTOTrainer and RewardTrainerTODO
First merge:
Note
Low Risk
Documentation-only changes to trainer docstrings; no runtime or API behavior changes.
Overview
Documents that
max_stepsmust be set in training arguments whentrain_datasetis anIterableDataset(e.g. streaming), because length cannot be inferred and Hugging FaceTrainerneeds a fixed step count for the loop and LR scheduler.The same note is added under
train_datasetin SFTTrainer, DPOTrainer, KTOTrainer, and RewardTrainer docstrings. GRPO and RLOO are unchanged—they reject iterable datasets at runtime, so the note would not apply. This mirrors the transformers-side documentation update.Reviewed by Cursor Bugbot for commit 64e5cc0. Bugbot is set up for automated code reviews on this repo. Configure here.