[Frontend] Add /v1/audio/transcriptions OpenAI API endpoint (#12909)
This commit is contained in:
Nicolò Lucchesi
@@ -16,10 +16,10 @@ from argparse import Namespace
|
||||
from contextlib import asynccontextmanager
|
||||
from functools import partial
|
||||
from http import HTTPStatus
|
||||
from typing import AsyncIterator, Dict, Optional, Set, Tuple, Union
|
||||
from typing import Annotated, AsyncIterator, Dict, Optional, Set, Tuple, Union
|
||||
|
||||
import uvloop
|
||||
from fastapi import APIRouter, Depends, FastAPI, HTTPException, Request
|
||||
from fastapi import APIRouter, Depends, FastAPI, Form, HTTPException, Request
|
||||
from fastapi.exceptions import RequestValidationError
|
||||
from fastapi.middleware.cors import CORSMiddleware
|
||||
from fastapi.responses import JSONResponse, Response, StreamingResponse
|
||||
@@ -61,6 +61,8 @@ from vllm.entrypoints.openai.protocol import (ChatCompletionRequest,
|
||||
ScoreRequest, ScoreResponse,
|
||||
TokenizeRequest,
|
||||
TokenizeResponse,
|
||||
TranscriptionRequest,
|
||||
TranscriptionResponse,
|
||||
UnloadLoraAdapterRequest)
|
||||
from vllm.entrypoints.openai.reasoning_parsers import ReasoningParserManager
|
||||
# yapf: enable
|
||||
@@ -75,6 +77,8 @@ from vllm.entrypoints.openai.serving_rerank import JinaAIServingRerank
|
||||
from vllm.entrypoints.openai.serving_score import OpenAIServingScores
|
||||
from vllm.entrypoints.openai.serving_tokenization import (
|
||||
OpenAIServingTokenization)
|
||||
from vllm.entrypoints.openai.serving_transcription import (
|
||||
OpenAIServingTranscription)
|
||||
from vllm.entrypoints.openai.tool_parsers import ToolParserManager
|
||||
from vllm.entrypoints.utils import with_cancellation
|
||||
from vllm.logger import init_logger
|
||||
@@ -327,6 +331,10 @@ def tokenization(request: Request) -> OpenAIServingTokenization:
|
||||
return request.app.state.openai_serving_tokenization
|
||||
|
||||
|
||||
def transcription(request: Request) -> OpenAIServingTranscription:
|
||||
return request.app.state.openai_serving_transcription
|
||||
|
||||
|
||||
def engine_client(request: Request) -> EngineClient:
|
||||
return request.app.state.engine_client
|
||||
|
||||
@@ -520,6 +528,31 @@ async def create_score_v1(request: ScoreRequest, raw_request: Request):
|
||||
return await create_score(request, raw_request)
|
||||
|
||||
|
||||
@router.post("/v1/audio/transcriptions")
|
||||
@with_cancellation
|
||||
async def create_transcriptions(request: Annotated[TranscriptionRequest,
|
||||
Form()],
|
||||
raw_request: Request):
|
||||
|
||||
handler = transcription(raw_request)
|
||||
if handler is None:
|
||||
return base(raw_request).create_error_response(
|
||||
message="The model does not support Transcriptions API")
|
||||
|
||||
audio_data = await request.file.read()
|
||||
generator = await handler.create_transcription(audio_data, request,
|
||||
raw_request)
|
||||
|
||||
if isinstance(generator, ErrorResponse):
|
||||
return JSONResponse(content=generator.model_dump(),
|
||||
status_code=generator.code)
|
||||
|
||||
elif isinstance(generator, TranscriptionResponse):
|
||||
return JSONResponse(content=generator.model_dump())
|
||||
|
||||
return StreamingResponse(content=generator, media_type="text/event-stream")
|
||||
|
||||
|
||||
@router.post("/rerank", dependencies=[Depends(validate_json_request)])
|
||||
@with_cancellation
|
||||
async def do_rerank(request: RerankRequest, raw_request: Request):
|
||||
@@ -832,6 +865,12 @@ async def init_app_state(
|
||||
chat_template=resolved_chat_template,
|
||||
chat_template_content_format=args.chat_template_content_format,
|
||||
)
|
||||
state.openai_serving_transcription = OpenAIServingTranscription(
|
||||
engine_client,
|
||||
model_config,
|
||||
state.openai_serving_models,
|
||||
request_logger=request_logger,
|
||||
) if model_config.runner_type == "transcription" else None
|
||||
state.task = model_config.task
|
||||
|
||||
|
||||
|
||||
@@ -8,9 +8,10 @@ from argparse import Namespace
|
||||
from typing import Any, ClassVar, Dict, List, Literal, Optional, Set, Union
|
||||
|
||||
import torch
|
||||
from fastapi import UploadFile
|
||||
from pydantic import (BaseModel, ConfigDict, Field, TypeAdapter,
|
||||
ValidationInfo, field_validator, model_validator)
|
||||
from typing_extensions import Annotated
|
||||
from typing_extensions import Annotated, TypeAlias
|
||||
|
||||
from vllm.entrypoints.chat_utils import ChatCompletionMessageParam
|
||||
from vllm.logger import init_logger
|
||||
@@ -1426,3 +1427,163 @@ class LoadLoraAdapterRequest(BaseModel):
|
||||
class UnloadLoraAdapterRequest(BaseModel):
|
||||
lora_name: str
|
||||
lora_int_id: Optional[int] = Field(default=None)
|
||||
|
||||
|
||||
## Protocols for Audio
|
||||
AudioResponseFormat: TypeAlias = Literal["json", "text", "srt", "verbose_json",
|
||||
"vtt"]
|
||||
|
||||
|
||||
class TranscriptionRequest(OpenAIBaseModel):
|
||||
# Ordered by official OpenAI API documentation
|
||||
#https://platform.openai.com/docs/api-reference/audio/createTranscription
|
||||
|
||||
file: UploadFile
|
||||
"""
|
||||
The audio file object (not file name) to transcribe, in one of these
|
||||
formats: flac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, or webm.
|
||||
"""
|
||||
|
||||
model: str
|
||||
"""ID of the model to use.
|
||||
"""
|
||||
|
||||
language: Optional[str] = None
|
||||
"""The language of the input audio.
|
||||
|
||||
Supplying the input language in
|
||||
[ISO-639-1](https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes) format
|
||||
will improve accuracy and latency.
|
||||
"""
|
||||
|
||||
prompt: str = Field(default="")
|
||||
"""An optional text to guide the model's style or continue a previous audio
|
||||
segment.
|
||||
|
||||
The [prompt](https://platform.openai.com/docs/guides/speech-to-text#prompting)
|
||||
should match the audio language.
|
||||
"""
|
||||
|
||||
response_format: AudioResponseFormat = Field(default="json")
|
||||
"""
|
||||
The format of the output, in one of these options: `json`, `text`, `srt`,
|
||||
`verbose_json`, or `vtt`.
|
||||
"""
|
||||
|
||||
## TODO (varun) : Support if set to 0, certain thresholds are met !!
|
||||
temperature: float = Field(default=0.0)
|
||||
"""The sampling temperature, between 0 and 1.
|
||||
|
||||
Higher values like 0.8 will make the output more random, while lower values
|
||||
like 0.2 will make it more focused / deterministic. If set to 0, the model
|
||||
will use [log probability](https://en.wikipedia.org/wiki/Log_probability)
|
||||
to automatically increase the temperature until certain thresholds are hit.
|
||||
"""
|
||||
|
||||
timestamp_granularities: List[Literal["word", "segment"]] = Field(
|
||||
alias="timestamp_granularities[]", default=[])
|
||||
"""The timestamp granularities to populate for this transcription.
|
||||
|
||||
`response_format` must be set `verbose_json` to use timestamp granularities.
|
||||
Either or both of these options are supported: `word`, or `segment`. Note:
|
||||
There is no additional latency for segment timestamps, but generating word
|
||||
timestamps incurs additional latency.
|
||||
"""
|
||||
|
||||
# Default sampling parameters for transcription requests.
|
||||
_DEFAULT_SAMPLING_PARAMS: dict = {
|
||||
"temperature": 0,
|
||||
}
|
||||
|
||||
def to_sampling_params(
|
||||
self,
|
||||
default_max_tokens: int,
|
||||
default_sampling_params: Optional[dict] = None) -> SamplingParams:
|
||||
# TODO(#9845): remove max_tokens when field is removed from OpenAI API
|
||||
max_tokens = default_max_tokens
|
||||
|
||||
if default_sampling_params is None:
|
||||
default_sampling_params = {}
|
||||
# Default parameters
|
||||
if (temperature := self.temperature) is None:
|
||||
temperature = default_sampling_params.get(
|
||||
"temperature", self._DEFAULT_SAMPLING_PARAMS["temperature"])
|
||||
|
||||
return SamplingParams.from_optional(temperature=temperature,
|
||||
max_tokens=max_tokens)
|
||||
|
||||
|
||||
# Transcription response objects
|
||||
class TranscriptionResponse(OpenAIBaseModel):
|
||||
text: str
|
||||
"""The transcribed text."""
|
||||
|
||||
|
||||
class TranscriptionWord(OpenAIBaseModel):
|
||||
end: float
|
||||
"""End time of the word in seconds."""
|
||||
|
||||
start: float
|
||||
"""Start time of the word in seconds."""
|
||||
|
||||
word: str
|
||||
"""The text content of the word."""
|
||||
|
||||
|
||||
class TranscriptionSegment(OpenAIBaseModel):
|
||||
id: int
|
||||
"""Unique identifier of the segment."""
|
||||
|
||||
avg_logprob: float
|
||||
"""Average logprob of the segment.
|
||||
|
||||
If the value is lower than -1, consider the logprobs failed.
|
||||
"""
|
||||
|
||||
compression_ratio: float
|
||||
"""Compression ratio of the segment.
|
||||
|
||||
If the value is greater than 2.4, consider the compression failed.
|
||||
"""
|
||||
|
||||
end: float
|
||||
"""End time of the segment in seconds."""
|
||||
|
||||
no_speech_prob: float
|
||||
"""Probability of no speech in the segment.
|
||||
|
||||
If the value is higher than 1.0 and the `avg_logprob` is below -1, consider
|
||||
this segment silent.
|
||||
"""
|
||||
|
||||
seek: int
|
||||
"""Seek offset of the segment."""
|
||||
|
||||
start: float
|
||||
"""Start time of the segment in seconds."""
|
||||
|
||||
temperature: float
|
||||
"""Temperature parameter used for generating the segment."""
|
||||
|
||||
text: str
|
||||
"""Text content of the segment."""
|
||||
|
||||
tokens: List[int]
|
||||
"""Array of token IDs for the text content."""
|
||||
|
||||
|
||||
class TranscriptionResponseVerbose(OpenAIBaseModel):
|
||||
duration: str
|
||||
"""The duration of the input audio."""
|
||||
|
||||
language: str
|
||||
"""The language of the input audio."""
|
||||
|
||||
text: str
|
||||
"""The transcribed text."""
|
||||
|
||||
segments: Optional[List[TranscriptionSegment]] = None
|
||||
"""Segments of the transcribed text and their corresponding details."""
|
||||
|
||||
words: Optional[List[TranscriptionWord]] = None
|
||||
"""Extracted words and their corresponding timestamps."""
|
||||
|
||||
@@ -31,7 +31,8 @@ from vllm.entrypoints.openai.protocol import (ChatCompletionRequest,
|
||||
ErrorResponse, RerankRequest,
|
||||
ScoreRequest,
|
||||
TokenizeChatRequest,
|
||||
TokenizeCompletionRequest)
|
||||
TokenizeCompletionRequest,
|
||||
TranscriptionRequest)
|
||||
from vllm.entrypoints.openai.serving_models import OpenAIServingModels
|
||||
from vllm.entrypoints.openai.tool_parsers import ToolParser
|
||||
# yapf: enable
|
||||
@@ -57,7 +58,8 @@ CompletionLikeRequest = Union[CompletionRequest, DetokenizeRequest,
|
||||
ChatLikeRequest = Union[ChatCompletionRequest, EmbeddingChatRequest,
|
||||
TokenizeChatRequest]
|
||||
|
||||
AnyRequest = Union[CompletionLikeRequest, ChatLikeRequest]
|
||||
AnyRequest = Union[CompletionLikeRequest, ChatLikeRequest,
|
||||
TranscriptionRequest]
|
||||
|
||||
|
||||
class TextTokensPrompt(TypedDict):
|
||||
|
||||
305
vllm/entrypoints/openai/serving_transcription.py
Normal file
305
vllm/entrypoints/openai/serving_transcription.py
Normal file
@@ -0,0 +1,305 @@
|
||||
# SPDX-License-Identifier: Apache-2.0
|
||||
import asyncio
|
||||
import io
|
||||
from typing import AsyncGenerator, Optional, Union, cast
|
||||
|
||||
from fastapi import Request
|
||||
|
||||
from vllm.config import ModelConfig
|
||||
from vllm.engine.protocol import EngineClient
|
||||
from vllm.entrypoints.logger import RequestLogger
|
||||
from vllm.entrypoints.openai.protocol import (ErrorResponse,
|
||||
RequestResponseMetadata,
|
||||
TranscriptionRequest,
|
||||
TranscriptionResponse,
|
||||
TranscriptionResponseVerbose)
|
||||
from vllm.entrypoints.openai.serving_engine import OpenAIServing
|
||||
from vllm.entrypoints.openai.serving_models import OpenAIServingModels
|
||||
from vllm.inputs.data import PromptType
|
||||
from vllm.logger import init_logger
|
||||
from vllm.outputs import RequestOutput
|
||||
from vllm.utils import PlaceholderModule
|
||||
|
||||
try:
|
||||
import librosa
|
||||
except ImportError:
|
||||
librosa = PlaceholderModule("librosa") # type: ignore[assignment]
|
||||
|
||||
logger = init_logger(__name__)
|
||||
|
||||
# From https://platform.openai.com/docs/guides/speech-to-text/supported-languages#supported-languages
|
||||
# TODO these configs should live somewhere with the model so we can support
|
||||
# additional ones
|
||||
|
||||
ISO639_1_SUPPORTED_LANGS = {
|
||||
"af": "Afrikaans",
|
||||
"ar": "Arabic",
|
||||
"hy": "Armenian",
|
||||
"az": "Azerbaijani",
|
||||
"be": "Belarusian",
|
||||
"bs": "Bosnian",
|
||||
"bg": "Bulgarian",
|
||||
"ca": "Catalan",
|
||||
"zh": "Chinese",
|
||||
"hr": "Croatian",
|
||||
"cs": "Czech",
|
||||
"da": "Danish",
|
||||
"nl": "Dutch",
|
||||
"en": "English",
|
||||
"et": "Estonian",
|
||||
"fi": "Finnish",
|
||||
"fr": "French",
|
||||
"gl": "Galician",
|
||||
"de": "German",
|
||||
"el": "Greek",
|
||||
"he": "Hebrew",
|
||||
"hi": "Hindi",
|
||||
"hu": "Hungarian",
|
||||
"is": "Icelandic",
|
||||
"id": "Indonesian",
|
||||
"it": "Italian",
|
||||
"ja": "Japanese",
|
||||
"kn": "Kannada",
|
||||
"kk": "Kazakh",
|
||||
"ko": "Korean",
|
||||
"lv": "Latvian",
|
||||
"lt": "Lithuanian",
|
||||
"mk": "Macedonian",
|
||||
"ms": "Malay",
|
||||
"mr": "Marathi",
|
||||
"mi": "Maori",
|
||||
"ne": "Nepali",
|
||||
"no": "Norwegian",
|
||||
"fa": "Persian",
|
||||
"pl": "Polish",
|
||||
"pt": "Portuguese",
|
||||
"ro": "Romanian",
|
||||
"ru": "Russian",
|
||||
"sr": "Serbian",
|
||||
"sk": "Slovak",
|
||||
"sl": "Slovenian",
|
||||
"es": "Spanish",
|
||||
"sw": "Swahili",
|
||||
"sv": "Swedish",
|
||||
"tl": "Tagalog",
|
||||
"ta": "Tamil",
|
||||
"th": "Thai",
|
||||
"tr": "Turkish",
|
||||
"uk": "Ukrainian",
|
||||
"ur": "Urdu",
|
||||
"vi": "Vietnamese",
|
||||
"cy": "Welsh"
|
||||
}
|
||||
ISO639_1_OTHER_LANGS = {
|
||||
"lo": "Lao",
|
||||
"jw": "Javanese",
|
||||
"tk": "Turkmen",
|
||||
"yi": "Yiddish",
|
||||
"so": "Somali",
|
||||
"bn": "Bengali",
|
||||
"nn": "Norwegian Nynorsk",
|
||||
"si": "Sinhala",
|
||||
"yo": "Yoruba",
|
||||
"sa": "Sanskrit",
|
||||
"mi": "Māori",
|
||||
"fo": "Faroese", # codespell:ignore
|
||||
"mt": "Maltese",
|
||||
"tg": "Tajik",
|
||||
"mg": "Malagasy",
|
||||
"haw": "Hawaiian",
|
||||
"km": "Khmer",
|
||||
"br": "Breton",
|
||||
"ps": "Pashto",
|
||||
"ln": "Lingala",
|
||||
"la": "Latin",
|
||||
"ml": "Malayalam",
|
||||
"sq": "Albanian",
|
||||
"su": "Sundanese",
|
||||
"eu": "Basque",
|
||||
"ka": "Georgian",
|
||||
"uz": "Uzbek",
|
||||
"sn": "Shona",
|
||||
"ht": "Haitian",
|
||||
"as": "Assamese",
|
||||
"mn": "Mongolian",
|
||||
"te": "Telugu",
|
||||
"pa": "Panjabi",
|
||||
"tt": "Tatar",
|
||||
"gu": "Gujarati",
|
||||
"oc": "Occitan",
|
||||
"ha": "Hausa",
|
||||
"ba": "Bashkir",
|
||||
"my": "Burmese",
|
||||
"sd": "Sindhi",
|
||||
"am": "Amharic",
|
||||
"lb": "Luxembourgish",
|
||||
"bo": "Tibetan"
|
||||
}
|
||||
|
||||
# As per https://platform.openai.com/docs/guides/speech-to-text#overview.
|
||||
# TODO configurable
|
||||
MAX_AUDIO_CLIP_FILESIZE_MB = 25
|
||||
# TODO get from processor.feature_extractor.chunk_length
|
||||
MAX_AUDIO_CLIP_DURATION_S = 30
|
||||
|
||||
|
||||
class OpenAIServingTranscription(OpenAIServing):
|
||||
|
||||
def __init__(
|
||||
self,
|
||||
engine_client: EngineClient,
|
||||
model_config: ModelConfig,
|
||||
models: OpenAIServingModels,
|
||||
*,
|
||||
request_logger: Optional[RequestLogger],
|
||||
return_tokens_as_token_ids: bool = False,
|
||||
):
|
||||
super().__init__(engine_client=engine_client,
|
||||
model_config=model_config,
|
||||
models=models,
|
||||
request_logger=request_logger,
|
||||
return_tokens_as_token_ids=return_tokens_as_token_ids)
|
||||
|
||||
diff_sampling_param = self.model_config.get_diff_sampling_param()
|
||||
if diff_sampling_param:
|
||||
logger.info(
|
||||
"Overwriting default completion sampling param with: %s",
|
||||
diff_sampling_param)
|
||||
|
||||
async def _preprocess_transcription(
|
||||
self,
|
||||
request: TranscriptionRequest,
|
||||
audio_data: bytes,
|
||||
) -> PromptType:
|
||||
# Validate request
|
||||
# TODO language should be optional and can be guessed.
|
||||
# For now we default to en. See
|
||||
# https://github.com/huggingface/transformers/blob/main/src/transformers/models/whisper/generation_whisper.py#L1520
|
||||
lang_token = f"<|{request.language}|>" if request.language else "<|en|>"
|
||||
if request.language:
|
||||
if request.language in ISO639_1_SUPPORTED_LANGS:
|
||||
pass
|
||||
elif request.language in ISO639_1_OTHER_LANGS:
|
||||
logger.warning(
|
||||
"The selected language %s has limited accuracy with"
|
||||
" reported WER>=0.5. Results may be less accurate "
|
||||
"for this choice.", request.language)
|
||||
else:
|
||||
raise ValueError(
|
||||
f"Unsupported language: {request.language}."
|
||||
"Language should be one of:" +
|
||||
f" {list(ISO639_1_SUPPORTED_LANGS.values())}" +
|
||||
f"or {list(ISO639_1_OTHER_LANGS.values())}")
|
||||
|
||||
if len(audio_data) / 1024**2 > MAX_AUDIO_CLIP_FILESIZE_MB:
|
||||
raise ValueError("Maximum file size exceeded.")
|
||||
|
||||
with io.BytesIO(audio_data) as bytes_:
|
||||
y, sr = librosa.load(bytes_)
|
||||
if librosa.get_duration(y=y, sr=sr) > MAX_AUDIO_CLIP_DURATION_S:
|
||||
raise ValueError(
|
||||
f"Maximum clip duration ({MAX_AUDIO_CLIP_DURATION_S}s) "
|
||||
"exceeded.")
|
||||
|
||||
prompt = {
|
||||
"encoder_prompt": {
|
||||
"prompt": "",
|
||||
"multi_modal_data": {
|
||||
"audio": (y, sr),
|
||||
},
|
||||
},
|
||||
"decoder_prompt":
|
||||
f"<|startoftranscript|>{lang_token}<|transcribe|><|notimestamps|>{request.prompt}"
|
||||
}
|
||||
return cast(PromptType, prompt)
|
||||
|
||||
# TODO (varun) : Make verbose response work !
|
||||
async def create_transcription(
|
||||
self, audio_data: bytes, request: TranscriptionRequest,
|
||||
raw_request: Request
|
||||
) -> Union[TranscriptionResponse, TranscriptionResponseVerbose,
|
||||
ErrorResponse]:
|
||||
"""Transcription API similar to OpenAI's API.
|
||||
|
||||
See https://platform.openai.com/docs/api-reference/audio/createTranscription
|
||||
for the API specification. This API mimics the OpenAI transcription API.
|
||||
"""
|
||||
error_check_ret = await self._check_model(request)
|
||||
if error_check_ret is not None:
|
||||
return error_check_ret
|
||||
|
||||
# If the engine is dead, raise the engine's DEAD_ERROR.
|
||||
# This is required for the streaming case, where we return a
|
||||
# success status before we actually start generating text :).
|
||||
if self.engine_client.errored:
|
||||
raise self.engine_client.dead_error
|
||||
|
||||
if request.response_format not in ['text', 'json']:
|
||||
return self.create_error_response(
|
||||
"Currently only support response_format `text` or `json`")
|
||||
|
||||
# TODO cmpl->transcription?
|
||||
request_id = f"cmpl-{self._base_request_id(raw_request)}"
|
||||
|
||||
request_metadata = RequestResponseMetadata(request_id=request_id)
|
||||
if raw_request:
|
||||
raw_request.state.request_metadata = request_metadata
|
||||
|
||||
try:
|
||||
(
|
||||
lora_request,
|
||||
prompt_adapter_request,
|
||||
) = self._maybe_get_adapters(request)
|
||||
|
||||
if lora_request:
|
||||
return self.create_error_response(
|
||||
"Currently do not support LoRA for Transcription.")
|
||||
if prompt_adapter_request:
|
||||
return self.create_error_response(
|
||||
"Currently do not support PromptAdapter for Transcription."
|
||||
)
|
||||
|
||||
prompt = await self._preprocess_transcription(
|
||||
request=request,
|
||||
audio_data=audio_data,
|
||||
)
|
||||
|
||||
except ValueError as e:
|
||||
logger.exception("Error in preprocessing prompt inputs")
|
||||
return self.create_error_response(str(e))
|
||||
|
||||
result_generator: Optional[AsyncGenerator[RequestOutput, None]] = None
|
||||
try:
|
||||
# TODO(rob): subtract len of tokenized prompt.
|
||||
default_max_tokens = self.model_config.max_model_len
|
||||
default_params = self.model_config.get_diff_sampling_param()
|
||||
sampling_params = request.to_sampling_params(
|
||||
default_max_tokens, default_params)
|
||||
|
||||
self._log_inputs(
|
||||
request_id,
|
||||
prompt['decoder_prompt'], # type: ignore
|
||||
params=sampling_params,
|
||||
lora_request=None,
|
||||
prompt_adapter_request=None)
|
||||
|
||||
result_generator = self.engine_client.generate(
|
||||
prompt,
|
||||
sampling_params,
|
||||
request_id,
|
||||
)
|
||||
except ValueError as e:
|
||||
# TODO: Use a vllm-specific Validation Error
|
||||
return self.create_error_response(str(e))
|
||||
|
||||
# TODO(rob): figure out a way to pipe streaming in.
|
||||
# Non-streaming response.
|
||||
try:
|
||||
async for op in result_generator:
|
||||
result = op
|
||||
return TranscriptionResponse(text=result.outputs[0].text)
|
||||
except asyncio.CancelledError:
|
||||
return self.create_error_response("Client disconnected")
|
||||
except ValueError as e:
|
||||
# TODO: Use a vllm-specific Validation Error
|
||||
return self.create_error_response(str(e))
|
||||
Reference in New Issue
Block a user