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

9
vllm_ascend/worker/multi_step_runner.py
View File

@@ -23,7 +23,6 @@ from vllm.worker.multi_step_model_runner import (ModelOutput,
PythonizationCache,
StatefulModelInput)
from vllm_ascend.utils import vllm_version_is
from vllm_ascend.worker.model_runner import (
ModelInputForNPUWithSamplingMetadata, NPUModelRunnerBase)
@@ -318,13 +317,7 @@ class MultiStepModelNPURunner(NPUModelRunnerBase[StatefulModelInputForNPU]):
dtype=torch.long,
device="cpu",
pin_memory=True)
if vllm_version_is("0.8.4"):
self._base_model_runner.model.sampler.include_gpu_probs_tensor = (
True)
else:
assert self._base_model_runner.sampler is not None
self._base_model_runner.sampler.include_gpu_probs_tensor = True
self._base_model_runner.sampler.include_gpu_probs_tensor = True
if frozen_model_input.sampling_metadata:
frozen_model_input.sampling_metadata.skip_sampler_cpu_output = (
True)