Default use_gpumd_qscaler to False - #5
Merged
Merged
Conversation
600-epoch, 4-seed PdCuNiP benchmark (stage2, 10% validation, V100): the default torch init with the self-consistent q_scaler reaches clearly better minima than the GPUMD-style start on every metric and every seed - ~12% lower E and V RMSE, ~3% lower F, on train and validation alike. Swapping only the NN init while keeping the c=1 scaler + uniform(-1,1) coefficients changes nothing, so the gap comes from the c=1-scaler/large-coefficient combination itself. True (the old default) is unchanged and remains available for GPUMD-comparison runs. The saved nep.txt is GPUMD-compatible either way - the q_scaler is stored in the file.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
600-epoch, 4-seed PdCuNiP benchmark (stage2, 10% validation, V100): the default torch init with the self-consistent q_scaler reaches clearly better minima than the GPUMD-style start on every metric and every seed - ~12% lower E and V RMSE, ~3% lower F, on train and validation alike. Swapping only the NN init while keeping the c=1 scaler + uniform(-1,1) coefficients changes nothing, so the gap comes from the c=1-scaler/large-coefficient combination itself.
True (the old default) is unchanged and remains available for GPUMD-comparison runs. The saved nep.txt is GPUMD-compatible either way - the q_scaler is stored in the file.