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

[Bugfix] Add missing dynamic_arg_dims for Qwen3-ASR torch.compile (#35869)

Browse Source 
Signed-off-by: Nathan Price <nathan@abridge.com>
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>
This commit is contained in:
Nathan Price
2026-03-04 02:29:01 -06:00
committed by GitHub
parent 6f0dd93801
commit 36bf213181
1 changed files with 0 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/model_executor/models/qwen3_asr_realtime.py
View File

@@ -22,7 +22,6 @@ from collections.abc import AsyncGenerator, Mapping
import numpy as np
import torch
from vllm.compilation.decorators import support_torch_compile
from vllm.config import ModelConfig, SpeechToTextConfig, VllmConfig
from vllm.inputs.data import PromptType, TokensPrompt
from vllm.logger import init_logger
@@ -177,7 +176,6 @@ class Qwen3ASRRealtimeMultiModalProcessor(Qwen3ASRMultiModalProcessor):
info=Qwen3ASRProcessingInfo,
dummy_inputs=Qwen3ASRDummyInputsBuilder,
)
@support_torch_compile
class Qwen3ASRRealtimeGeneration(Qwen3ASRForConditionalGeneration, SupportsRealtime):
realtime_max_tokens = 64