apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2026-04-23 08:21:53 +08:00
Code Issues Actions 19 Packages Projects Releases Wiki Activity

[Feature] Support mtp overlap schedule (#7001)

Browse Source
This commit is contained in:
sunxin
2026-04-01 14:24:26 +08:00
committed by GitHub
parent c6f0c5c3a6
commit c29e86fc9d
23 changed files with 215 additions and 138 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/operators/test_draft_model_update.py
+1 -3
View File
@@ -64,13 +64,11 @@ CPU_PLACE = paddle.CPUPlace()
def to_paddle_inputs(inputs: Dict[str, Any]) -> Dict[str, Any]:
"""Convert numpy dict → paddle tensors. not_need_stop stays on CPU."""
"""Convert numpy dict → paddle tensors."""
paddle_inputs = {}
for k, v in inputs.items():
if isinstance(v, (int, bool, float)):
paddle_inputs[k] = v
elif k == "not_need_stop":
paddle_inputs[k] = paddle.to_tensor(v, place=CPU_PLACE)
elif v is not None:
paddle_inputs[k] = paddle.to_tensor(v, place=CUDA_PLACE)
return paddle_inputs