Skip to content

Add safeguard for missing quant_state in vision models #165

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Jun 19, 2025
Merged

Add safeguard for missing quant_state in vision models #165

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions unsloth_zoo/patching_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -284,6 +284,11 @@ def __fix_dtype(config):
for name, module in model.named_modules():
if isinstance(module, (Bnb_Linear4bit, Peft_Linear4bit)):
weight = module.weight
# Check if quant_state exists for vision models like unsloth/Llama-3.2-11B-Vision-Instruct-bnb-4bit, unsloth/granite-vision-3.2-2b
if not hasattr(weight, 'quant_state'):
print(f"Skipping {name}: no quant_state found")
continue

quant_state = weight.quant_state

if type(quant_state) is list:
Expand Down