423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-08-02 19:52:50 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix #2732

Browse Source 
Former-commit-id: 18ffce36b5ee0809f2e2905c2fd44843a3725ea0
This commit is contained in:
hiyouga 2024-03-09 22:37:16 +08:00
parent 868444e124
commit 276def1897
1 changed files with 4 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
src/llmtuner/model/loader.py
View File

@ -109,7 +109,10 @@ def load_model(
if not is_trainable:
model.requires_grad_(False)
model = model.to(model_args.compute_dtype) if not getattr(model, "quantization_method", None) else model
if not getattr(model, "quantization_method", None):
for param in filter(lambda p: p.device.type == "cuda", model.parameters()):
param.data = param.data.to(model_args.compute_dtype)
model.eval()
else:
model.train()