[Feat][Graph] Support MTP for ACL Graph (#2932)

### What this PR does / why we need it?
This PR depends on the merge of #2707 and has adapted the aclgraph
functionality to support MTP.

### How was this patch tested?


- vLLM version: v0.10.2
- vLLM main:
2b85697031

---------

Signed-off-by: xuyexiong <xuyexiong@huawei.com>
This commit is contained in:
xuyexiong
2025-09-18 14:05:33 +08:00
committed by GitHub
parent cef43b524e
commit 6681dde902
7 changed files with 73 additions and 11 deletions

View File

@@ -39,7 +39,7 @@ def mtp_correctness(
tensor_parallel_size=1,
gpu_memory_utilization=0.7,
max_model_len=256,
enforce_eager=True) as ref_llm:
enforce_eager=False) as ref_llm:
ref_outputs = ref_llm.generate(example_prompts, sampling_config)
with VllmRunner(
@@ -53,7 +53,7 @@ def mtp_correctness(
"method": "deepseek_mtp",
"num_speculative_tokens": num_speculative_tokens,
},
enforce_eager=True,
enforce_eager=False,
max_model_len=2000,
additional_config={"ascend_scheduler_config": {
"enabled": False