reset recurrent state after cudagraph warmup#579
Open
valtterivalo wants to merge 1 commit into
Open
Conversation
Warmup advances per-buffer and frozen-bank MinGRU state but never restored it, unlike weights, momentum, and RNG state. A fresh PuffeRL now starts from zero recurrent state.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
i had parity issues between CUDA trained models and local Metal evals and noticed this. we restore all kinds of state but not recurrent for some reason. could be intentional but i'm not aware of a reason
this can be repro'd on breakout with default config for example where reading buffer_states[0] after construction shows 98304/131072 bytes nonzero. this PR makes that 0/131072
it's +5 lines for something CUDA to CUDA users will never feel but it affected me so might as well PR