Unsloth optims for Llama #1609
Merged
Unsloth optims for Llama #1609
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
|
The cross_entropy_loss optimization is applicable even in a full fine tune, right? |
Correct! |
|
Are these optimizations compatible with flash attention? (Complete Noob here) |
|
It is possible to alter to also patch Qwen? It was added to unsloth and all optimizations work for it: |
Let's tackle that in a follow up PR |
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Outdated suggestions cannot be applied.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.
WIP to integrate Unsloth's optimizations into axolotl.
The manual autograd for MLP, QKV, O only seems to help VRAM by 1% as opposed to the reported 8%.
The Cross Entropy Loss does help significantly, but only reduced VRAM by 13% as opposed to the reported 17%.
edit: clarification, the cross entropy loss works for both full fine tunes and lora. The MLP, QKV, and O are only for 4-bit qlora with flash attention.