A high-throughput and memory-efficient inference and serving engine for LLMs
[Model Runner V2] Fix v2 `AttributeError: 'CohereASRDecoder' object has no attribute 'embed_input_ids'` (#44568)
Signed-off-by: yewentao256 <zhyanwentao@126.com> Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
Wentao Ye committed
Jun 10, 2026 at 14:06 UTC
ffce72c0415701eb10af86668cf504a5eeba2a72
1 file changed
+4
-1
vllm/v1/worker/gpu/model_states/__init__.py
+4
-1
@@ -13,7 +13,10 @@ def init_model_state(
13
encoder_cache: EncoderCache | None,
14
device: torch.device,
15
):
16
- if "WhisperForConditionalGeneration" in vllm_config.model_config.architectures:
16
+ if (
17
+ "WhisperForConditionalGeneration" in vllm_config.model_config.architectures
18
+ or "CohereAsrForConditionalGeneration" in vllm_config.model_config.architectures
19
+ ):
20
from vllm.v1.worker.gpu.model_states.whisper import WhisperModelState
21
22
return WhisperModelState(vllm_config, model, encoder_cache, device)