Quark dataset importer for fp8 #96
Merged
Conversation
This file contains 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
stellaraccident
requested changes
Jul 9, 2024
dan-garvey
force-pushed
the
llama_fp8
branch
2 times, most recently
from
142aeb7
to
4ed3c9d
Compare
its ugly
TODO: split qkv prior to irpa
import directly to gguf format
undo some uneccessary changes
and rebase!
dan-garvey
changed the title
IanNod
requested changes
Sep 5, 2024
IanNod
reviewed
Sep 5, 2024
IanNod
approved these changes
Sep 24, 2024
stellaraccident
approved these changes
Sep 24, 2024
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Outdated suggestions cannot be applied.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.
:edited:
Provides a entry point for quark models that are stored as a safetensor file + config.json. With a little effort this could also be adapted to become the equivalent of the hf importer from gguf without the intermediate step.
I removed a lot of the useful debugging tooling I developed for this because it doesn't play nicely with torch.export, but I imagine a lot of it could be re-added if I guarded it in some way, so I'm keeping a branch with those changes for reference.