EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[BugFix]Fix recomputed scheduler bug (#7137)

Browse Source 
### What this PR does / why we need it?
Fix the wrong usage of `model_type`.

### Does this PR introduce _any_ user-facing change?
No.

### How was this patch tested?
By CI.

- vLLM version: v0.16.0
- vLLM main:
4034c3d32e

Signed-off-by: nwpu-zxr <zhouxuerong2@huawei.com>
This commit is contained in:
zxr2333
2026-03-11 00:32:19 +08:00
committed by GitHub
parent 54668e73c5
commit e16009b2cc
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
vllm_ascend/core/recompute_scheduler.py
View File

@@ -93,8 +93,8 @@ class RecomputeScheduler(Scheduler):
) )
self.is_kv_producer = self.vllm_config.kv_transfer_config and self.vllm_config.kv_transfer_config.is_kv_producer self.is_kv_producer = self.vllm_config.kv_transfer_config and self.vllm_config.kv_transfer_config.is_kv_producer
self.is_hybrid_model = ( self.is_hybrid_model = (
"qwen3_next" in self.vllm_config.model_config.model_type "qwen3_next" in self.vllm_config.model_config.hf_text_config.model_type
or "qwen3_5" in self.vllm_config.model_config.model_type or "qwen3_5" in self.vllm_config.model_config.hf_text_config.model_type
) )
def add_request(self, request: Request) -> None: def add_request(self, request: Request) -> None: