NGC25.05 + Fix Xielu#87
Open
TJ-Solergibert wants to merge 57 commits into
Open
Conversation
Co-authored-by: Antoni-Joan Solergibert <74564958+TJ-Solergibert@users.noreply.github.com>
Co-authored-by: Antoni-Joan Solergibert <74564958+TJ-Solergibert@users.noreply.github.com>
Co-authored-by: Antoni-Joan Solergibert <74564958+TJ-Solergibert@users.noreply.github.com>
If state["exp_avg_fast"] is set to None, the checkpointing is crashing.
added lm_head in converter when tied and also tempdir logic
Swissai hf model converter
Co-authored-by: Alex Hägele <alexander.hagele@epfl.ch>
AdEMAMix Optimizer
Fix xielu sharding error
OP block implementation
conversions: [torch_dist --> torch] & [core --> HF]
fix rope scaling during conversion
Update xielu-backwards
Allow loading old checkpoints
Fix core forward pass (rope) during logits-test
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
6 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This branch has proven to work both for the 70B & 8B models. Recall that we started training with 25.01, so we had to patch some stuff from the distributed checkpointing to make it compatible with the updated PyTorch within the container. It also hardcodes a
modeloptimport that triggered an error with the newer version when resuming from a checkpoint and the xielu fix by @AleHD.Use with:
/iopsstor/scratch/cscs/dealmeih/images/ngc-transformers+25.05.sqshcc @C-TC @AleHD @dhia680 @martinjaggi @ischlag