A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 13, 04:30 UTC
Remove audio optional dependency for mistral-common (#28722)
Signed-off-by: Julien Denize <julien.denize@mistral.ai> Signed-off-by: Julien Denize <40604584+juliendenize@users.noreply.github.com> Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com> Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
Julien Denize committed
Nov 14, 2025 at 18:54 UTC
085424808ef705efbf59e7b18bc010f53d9d7f75
4 files changed
+6
-2
docs/contributing/model/transcription.md
+1
-1
@@ -249,7 +249,7 @@ No extra registration is required beyond having your model class available via t
249
## Examples in-tree
250
251
- Whisper encoder–decoder (audio-only): [vllm/model_executor/models/whisper.py](../../../vllm/model_executor/models/whisper.py)
252
-- Voxtral decoder-only (audio embeddings + LLM): [vllm/model_executor/models/voxtral.py](../../../vllm/model_executor/models/voxtral.py)
252
+- Voxtral decoder-only (audio embeddings + LLM): [vllm/model_executor/models/voxtral.py](../../../vllm/model_executor/models/voxtral.py). Make sure to have installed `mistral-common[audio]`.
253
- Gemma3n decoder-only with fixed instruction prompt: [vllm/model_executor/models/gemma3n_mm.py](../../../vllm/model_executor/models/gemma3n_mm.py)
254
255
## Test with the API
docs/models/supported_models.md
+3
@@ -785,6 +785,9 @@ Speech2Text models trained specifically for Automatic Speech Recognition.
785
| `Gemma3nForConditionalGeneration` | Gemma3n | `google/gemma-3n-E2B-it`, `google/gemma-3n-E4B-it`, etc. | | |
786
| `GraniteSpeechForConditionalGeneration` | Granite Speech | `ibm-granite/granite-speech-3.3-2b`, `ibm-granite/granite-speech-3.3-8b`, etc. | ✅︎ | ✅︎ |
787
788
+!!! note
789
+ `VoxtralForConditionalGeneration` requires `mistral-common[audio]` to be installed.
790
+
791
### Pooling Models
792
793
See [this page](./pooling_models.md) for more information on how to use pooling models.
examples/offline_inference/audio_language.py
+1
@@ -43,6 +43,7 @@ class ModelRequestData(NamedTuple):
43
44
45
# Voxtral
46
+# Make sure to install mistral-common[audio].
47
def run_voxtral(question: str, audio_count: int) -> ModelRequestData:
48
from mistral_common.audio import Audio
49
from mistral_common.protocol.instruct.chunk import (
requirements/common.txt
+1
-1
@@ -31,7 +31,7 @@ partial-json-parser # used for parsing partial JSON outputs
31
pyzmq >= 25.0.0
32
msgspec
33
gguf >= 0.13.0
34
-mistral_common[image,audio] >= 1.8.5
34
+mistral_common[image] >= 1.8.5
35
opencv-python-headless >= 4.11.0 # required for video IO
36
pyyaml
37
six>=1.16.0; python_version > '3.11' # transitive dependency of pandas that needs to be the latest version for python 3.12