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

[Frontend] Add /v1/audio/transcriptions OpenAI API endpoint (#12909)

Browse Source
This commit is contained in:
Nicolò Lucchesi
2025-02-13 16:23:45 +01:00
committed by GitHub
parent 37dfa60037
commit d84cef76eb
20 changed files with 910 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/model_executor/models/whisper.py
View File

@@ -31,7 +31,7 @@ from vllm.multimodal.audio import resample_audio
from vllm.sequence import SequenceData
from vllm.transformers_utils.processor import cached_get_processor
from .interfaces import SupportsMultiModal
from .interfaces import SupportsMultiModal, SupportsTranscription
from .utils import AutoWeightsLoader, WeightsMapper, make_layers
logger = init_logger(__name__)
@@ -637,7 +637,8 @@ def input_mapper_for_whisper(
@MULTIMODAL_REGISTRY.register_input_mapper("audio", input_mapper_for_whisper)
@MULTIMODAL_REGISTRY.register_max_multimodal_tokens(
"audio", get_max_whisper_audio_tokens)
class WhisperForConditionalGeneration(nn.Module, SupportsMultiModal):
class WhisperForConditionalGeneration(nn.Module, SupportsTranscription,
SupportsMultiModal):
packed_modules_mapping = {
"self_attn.qkv_proj": [
"self_attn.q_proj",