423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-08-04 12:42:51 +08:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #2262 from fenglui/main

Browse Source 
fix torch_dtype check of export_model

Former-commit-id: 11f612deab9c9d100ded04cce4bc0f9fff451f77
This commit is contained in:
hoshi-hiyouga 2024-01-21 12:34:37 +08:00 committed by GitHub
commit f29156bf5b
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/llmtuner/train/tuner.py
View File

@ -57,7 +57,7 @@ def export_model(args: Optional[Dict[str, Any]] = None):
raise ValueError("The model is not a `PreTrainedModel`, export aborted.") raise ValueError("The model is not a `PreTrainedModel`, export aborted.")
setattr(model.config, "use_cache", True) setattr(model.config, "use_cache", True)
if getattr(model.config, "torch_dtype", None) == "bfloat16": if getattr(model.config, "torch_dtype", None) == torch.bfloat16:
model = model.to(torch.bfloat16).to("cpu") model = model.to(torch.bfloat16).to("cpu")
else: else:
model = model.to(torch.float16).to("cpu") model = model.to(torch.float16).to("cpu")