Fix harden checkpoint loading and LoRA initialization issue #778
Conversation
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
gaikwadrahul8
changed the title
gaikwadrahul8
force-pushed
the
fix/secure-loader-and-lora-initialization
branch
from
b604b38 to
cc1a9d1
Compare
…bugs - loader: use precise .pt detection (glob: *.pt, suffix: .pt) - loader: enforce torch.load on CPU with weights_only for safety - lora: use torch.rand for float dtype in random() to avoid runtime error - code cleanup: remove unused code and simplify operations
gaikwadrahul8
force-pushed
the
fix/secure-loader-and-lora-initialization
branch
from
cc1a9d1 to
5432f60
Compare
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Outdated suggestions cannot be applied.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.
Hi, Team
I have created this PR for some code improvement so I've made below code changes
Loader Security Hardening(ai_edge_torch/generative/utilities/loader.py) for precise file detection so changed from *pt to *.pt pattern for exact matching and CPU only loading added map_location=torch.device("cpu") for safer loading and also added security enhancement weights_only=True to prevent pickle based attacks
LoRA Random Initialization Fix (ai_edge_torch/generative/layers/lora.py) changed from torch.randint to torch.rand for float dtype runtime error prevention which eliminates crashes when LoRA weights are floating point
I would request you to please review this PR, if you've any feedback or suggestion please let me know that will be very helpful. Thank you for your consideration.