[bugfix (pcp)] fix chunked prefill accurancy issue (#5647)
### What this PR does / why we need it? Purpose: initialize padded slot mapping buffer to prevent garbage values. In PCP mode, the `pcp_padded_slot_mapping` buffer is reused across invocations. Without explicit initialization, this buffer retain stale values from previous runs, which can lead to incorrect results. This change ensures the buffer is filled with -1. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? - vLLM version: v0.13.0 - vLLM main: https://github.com/vllm-project/vllm/commit/2f4e6548efec402b913ffddc8726230d9311948d --------- Signed-off-by: F.Liu <liufeng248@huawei.com> Co-authored-by: F.Liu <liufeng248@huawei.com>
This commit is contained in:
1 parent
1112208052
commit
cbc987db0b
1 file changed
+1
@@ -319,6 +319,7 @@ class PCPManager:
|
||||
pcp_world_size]
|
||||
cp_unpad_mask = self.pcp_unpad_mask_cpu_tensor[:num_tokens *
|
||||
self.pcp_world_size]
|
||||
pcp_padded_slot_mapping.fill_(-1)
|
||||
pcp_padded_slot_mapping[cp_unpad_mask] = slot_mapping
|
||||
return pcp_padded_slot_mapping
|
||||
|
||||
|
||||
Reference in new issue
Block a user