apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-12-24 13:28:13 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[BugFix]Set default OMP_NUM_THREADS=3 and fix extra GPU memory usage in DeepSeek (#5219)

Browse Source 
* fix bug

* update

* update

* update

* fix copy

* update
This commit is contained in:
bukejiyu
2025-11-28 14:22:04 +08:00
committed by GitHub
parent 7dc06cac6e
commit 1539fd6056
6 changed files with 29 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
fastdeploy/model_executor/layers/utils.py
View File

@@ -141,7 +141,10 @@ def get_tensor(input: Union[paddle.Tensor, np.ndarray, str], model_path=None) ->
if isinstance(input, paddle.Tensor):
if input.place.is_cpu_place():
return input.to(paddle.device.get_device())
if current_platform.is_cuda():
return input.cuda()
else:
return input.to(paddle.device.get_device())
return input
elif isinstance(input, np.ndarray):
return paddle.to_tensor(input)