xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-07-09 14:57:16 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix bug of always capture lora even no-lora case

Browse Source 
Signed-off-by: Yu Gong <yu3.gong@gmail.com>
This commit is contained in:
Yu Gong 2025-12-18 21:22:25 +00:00
parent aa7917aaaa
commit 3b5270a6af
1 changed files with 6 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
vllm/v1/worker/lora_model_runner_mixin.py
View File

@ -195,10 +195,12 @@ class LoRAModelRunnerMixin:
for lora_id in range(1, effective_num_loras + 1) for lora_id in range(1, effective_num_loras + 1)
} }
if lora_requests: # Always call _set_active_loras to ensure the mapping is updated.
self._set_active_loras( # This is important when capturing no-LoRA graphs (effective_num_loras=0)
tuple(sample_lora_mapping), tuple(token_lora_mapping), lora_requests # after capturing LoRA graphs, as we need to clear the previous mapping.
) self._set_active_loras(
tuple(sample_lora_mapping), tuple(token_lora_mapping), lora_requests
)
yield yield