Conversation
…t batches - Add release_storage/best_effort ownership options to RolloutMetadata.to_rollout_state and discard_rollout_state; extract release_rollout_metadata_storage for outer-ref release. - Restore complete states from Object Store in the trainer via _restore_rollout_batch and release them deterministically in finally blocks for train/eval/debug paths. - Consume agent-loop prepared next-token-aligned input_ids/labels in _prepare_train_data and read rewards/lengths/tool_turns from metadata. - Validate ReplayBuffer.put input is RolloutMetadata with storage for completed samples. - Apply the token shift in localhost/sandbox loop preparation; release storage on sampler rerollout; generalize deterministic sorting and advantage response_len to metadata.
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
核心思想：
agent_loop 数据生产结束后，先调用
prepare_training_artifacts准备训练格式，然后将数据直接放到共享存储中，只传递metadata到RLTrainer和TrainController后续开发：
重构pack策略，pack 时仅依赖 meta 信息：#2055
待优化项：