Revert #10192: _template_context must unset template.max_length in GRPO/GKD - #10221
Merged
Merged
Conversation
tastelikefeet
approved these changes
Sep 21, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What does this PR do?
Reverts #10192, restoring the original default of
RolloutTrainerMixin._template_context:max_length: Optional[int] = Nonewithtemplate.max_length = max_length, and drops the_TEMPLATE_CONTEXT_USE_ORIGINALsentinel.Why
max_lengthbounds the prompt in GRPO/GKD/inference (max_prompt_length), and the completion isbounded separately by
max_completion_length/ the rolloutgeneration_config. The training-timere-encode of prompt + completion therefore must not be truncated again: a legal sample can be up to
max_length + max_completion_length.#10192 made the default "preserve
template.max_length", which silently re-enabled truncation at everycall site (GRPO
_prepare_batch_inputs/ SP logps / MLLM chunk re-encode, GKD_encode_samples).The Megatron trainers (
_template_context/_disable_maxlength_template_context) were never changedand still unset
max_length; this revert restores the same semantics for the HF path.