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

[CI] upgrade vllm to 0.8.5 (#715)

Browse Source 
1. Upgrade vllm to 0.8.5
2. Drop 0.8.4 support
3. Keep doc to 0.8.4rc2 until we release 0.8.5

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
This commit is contained in:
wangxiyuan
2025-04-30 09:15:50 +08:00
committed by GitHub
parent 95e7aa4736
commit f8350569e6
20 changed files with 48 additions and 579 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
vllm_ascend/patch/worker/patch_common/patch_multi_step_worker.py
View File

@@ -22,7 +22,6 @@ from vllm.model_executor.layers.sampler import SamplerOutput
from vllm.sequence import ExecuteModelRequest
from vllm.spec_decode.multi_step_worker import MultiStepWorker
from vllm_ascend.utils import vllm_version_is
from vllm_ascend.worker.draft_model_runner import TP1DraftModelRunner
@@ -93,16 +92,14 @@ def set_include_gpu_probs_tensor(self) -> None:
# Need include_gpu_probs_tensor for MultiSteoWorker
if hasattr(self.model_runner.model, "sampler"):
self.model_runner.model.sampler.include_gpu_probs_tensor = True
if not vllm_version_is("0.8.4"):
self.model_runner.sampler.include_gpu_probs_tensor = True
self.model_runner.sampler.include_gpu_probs_tensor = True
def set_should_modify_greedy_probs_inplace(self) -> None:
if hasattr(self.model_runner.model, "sampler"):
self.model_runner.model.sampler.should_modify_greedy_probs_inplace = (
True)
if not vllm_version_is("0.8.4"):
self.model_runner.sampler.should_modify_greedy_probs_inplace = True
self.model_runner.sampler.should_modify_greedy_probs_inplace = True
MultiStepWorker.sampler_output = torch.inference_mode()(sampler_output)