Reduce initial pipeline load time by 4-5x (1/3)#149
Open
Rypo wants to merge 4 commits intoVectorSpaceLab:mainfrom
Open
Reduce initial pipeline load time by 4-5x (1/3)#149Rypo wants to merge 4 commits intoVectorSpaceLab:mainfrom
Rypo wants to merge 4 commits intoVectorSpaceLab:mainfrom
Conversation
Prevents slow CPU initialization of model weights on load by using accelerate `init_empty_weights`. Completely compatible with from_pretrained since weights will always be overwritten by state_dict fixes VectorSpaceLab#72
This was referenced Nov 28, 2024
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Changes
from_pretrained(low_cpu_mem_usage=True)(akin to the transformers implementation, but greatly simplified) toOmniGenandOmniGenPipelineinit_empty_weightscontext manager when initializing the model. This avoids slow CPU weight initialization, particularly duringself.initialize_weights().These weights are immediately overwritten when the
state_dictis loaded. This means we can safely bypass initialization without consequence.Additionally, this can achieved with no additional libraries beyond those in
requirements.txt. As such, I set the default aslow_cpu_mem_usage=True.Results
From my tests, this change:
Cold Load
Hot Load
This is the first of 3 PRs I'm issuing to improve performance/fix errors. I've tried to keep each incremental change as small in scope as possible. PRs: 1. This, 2. #150, 3. #151