biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Misc] unify variable for LLM instance (#20996)

Browse Source 
Signed-off-by: Andy Xie <andy.xning@gmail.com>
This commit is contained in:
Ning Xie
2025-07-21 19:18:33 +08:00
committed by GitHub
parent e6b90a2805
commit d97841078b
53 changed files with 237 additions and 236 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
tests/models/language/generation/test_hybrid.py
View File

@@ -274,7 +274,7 @@ def test_models_preemption_recompute(
Tests that outputs are identical with and w/o preemptions (recompute).
"""
with vllm_runner(model, max_num_seqs=MAX_NUM_SEQS) as vllm_model:
scheduler = vllm_model.model.llm_engine.scheduler[0]
scheduler = vllm_model.llm.llm_engine.scheduler[0]
scheduler.ENABLE_ARTIFICIAL_PREEMPT = True
preempt_vllm_outputs = vllm_model.generate_greedy(
example_prompts, max_tokens)