xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-07-17 05:07:09 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix TP

Browse Source
This commit is contained in:
Alexander Matveev 2025-01-31 01:18:56 +00:00
parent 984ffddda6
commit 0a02744dc8
1 changed files with 2 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
vllm/worker/model_runner.py
View File

@ -1975,7 +1975,8 @@ class CUDAGraphRunner(nn.Module):
# Copy the input tensors to the input buffers. # Copy the input tensors to the input buffers.
self.input_buffers["input_ids"].copy_(input_ids, non_blocking=True) self.input_buffers["input_ids"].copy_(input_ids, non_blocking=True)
self.input_buffers["positions"].copy_(positions, non_blocking=True) if positions is not None:
self.input_buffers["positions"].copy_(positions, non_blocking=True)
if self.backend_name != "NO_ATTENTION": if self.backend_name != "NO_ATTENTION":
self.input_buffers["slot_mapping"].copy_( self.input_buffers["slot_mapping"].copy_(