EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

fix empty_cache error in pt_weights_iterator (#5151)

Browse Source 
Co-authored-by: dangkai.dk <dangkai.dk@alibaba-inc.com>
This commit is contained in:
DangKai
2025-04-08 16:22:10 +08:00
committed by GitHub
parent 3033c11a21
commit cc88d98ab8
1 changed files with 0 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
python/sglang/srt/model_loader/weight_utils.py
View File

@@ -457,7 +457,6 @@ def pt_weights_iterator(
state = torch.load(bin_file, map_location="cpu", weights_only=True) state = torch.load(bin_file, map_location="cpu", weights_only=True)
yield from state.items() yield from state.items()
del state del state
torch.cuda.empty_cache()
def get_gguf_extra_tensor_names( def get_gguf_extra_tensor_names(