Repository navigation
Conversation
Signed-off-by: 0z5a <dezhen.lu@student.uni-tuebingen.de>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The existing three-update qualification profile forces rematerialization, which the first prefix learner does not support. Add matched B-stage baseline/prefix profiles with rematerialization off while preserving the official model, objective, inputs, sampling budget, Adam configuration, evaluation and fresh-worker recovery. The default profile remains byte-identical.
Record completed RLTT train-step dispatches after successful optimizer/scheduler advancement. The new-profile audit binds those records to the real startup worker and rejects wrong schedules, sample counts, rematerialization or scheduler counters. This is an incremental draft on #26; it does not replace the final A-stage baseline or claim GPU qualification.
Validation: 86 CPU checks passed, 2 actual CUDA checks skipped. Both frozen packets, six unexecuted phase commands, logs/XML, source hashes and the result table are included. Seven official model files are cached and SHA-verified on H20; training still awaits the required environment. No speedup or reward-convergence result is claimed.