File size: 5,384 Bytes
20a070f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 |
import argparse
import asyncio
import functools
import json
import os
from io import BytesIO
import uvicorn
from fastapi import FastAPI, BackgroundTasks, File, Body, UploadFile, Request
from fastapi.responses import StreamingResponse
# from faster_whisper import WhisperModel
from starlette.staticfiles import StaticFiles
from starlette.templating import Jinja2Templates
from sentence_transformers import SentenceTransformer
# from zhconv import convert
# from utils.data_utils import remove_punctuation
# from utils.utils import add_arguments, print_arguments
import hashlib
import os
import tarfile
import urllib.request
# from tqdm import tqdm
def print_arguments(args):
print("----------- Configuration Arguments -----------")
for arg, value in vars(args).items():
print("%s: %s" % (arg, value))
print("------------------------------------------------")
def strtobool(val):
val = val.lower()
if val in ('y', 'yes', 't', 'true', 'on', '1'):
return True
elif val in ('n', 'no', 'f', 'false', 'off', '0'):
return False
else:
raise ValueError("invalid truth value %r" % (val,))
def str_none(val):
if val == 'None':
return None
else:
return val
def add_arguments(argname, type, default, help, argparser, **kwargs):
type = strtobool if type == bool else type
type = str_none if type == str else type
argparser.add_argument("--" + argname,
default=default,
type=type,
help=help + ' Default: %(default)s.',
**kwargs)
os.environ['KMP_DUPLICATE_LIB_OK'] = 'True'
parser = argparse.ArgumentParser(description=__doc__)
add_arg = functools.partial(add_arguments, argparser=parser)
add_arg("host", type=str, default="0.0.0.0", help="")
add_arg("port", type=int, default=5000, help="")
add_arg("model_path", type=str, default="BAAI/bge-small-en-v1.5", help="")
add_arg("use_gpu", type=bool, default=False, help="")
# add_arg("use_int8", type=bool, default=True, help="")
add_arg("beam_size", type=int, default=10, help="")
add_arg("num_workers", type=int, default=2, help="")
add_arg("vad_filter", type=bool, default=True, help="")
add_arg("local_files_only", type=bool, default=True, help="")
args = parser.parse_args()
print_arguments(args)
#
# assert os.path.exists(args.model_path), f"{args.model_path}"
#
if args.use_gpu:
model = SentenceTransformer(args.model_path, device="cuda", compute_type="float16")
else:
model = SentenceTransformer(args.model_path, device='cpu')
#
# _, _ = model.transcribe("dataset/test.wav", beam_size=5)
app = FastAPI(title="embedding Inference")
# app.mount('/static', StaticFiles(directory='static'), name='static')
# templates = Jinja2Templates(directory="templates")
# model_semaphore = None
# def release_model_semaphore():
# model_semaphore.release()
# def recognition(file: File, to_simple: int,
# remove_pun: int, language: str = "bn",
# task: str = "transcribe"
# ):
# segments, info = model.transcribe(file, beam_size=10, task=task, language=language, vad_filter=args.vad_filter)
# for segment in segments:
# text = segment.text
# if to_simple == 1:
# # text = convert(text, '')
# pass
# if remove_pun == 1:
# # text = remove_punctuation(text)
# pass
# ret = {"result": text, "start": round(segment.start, 2), "end": round(segment.end, 2)}
# #
# yield json.dumps(ret).encode() + b"\0"
# @app.post("/recognition_stream")
# async def api_recognition_stream(
# to_simple: int = Body(1, description="", embed=True),
# remove_pun: int = Body(0, description="", embed=True),
# language: str = Body("bn", description="", embed=True),
# task: str = Body("transcribe", description="", embed=True),
# audio: UploadFile = File(..., description="")
# ):
# global model_semaphore
# if language == "None": language = None
# if model_semaphore is None:
# model_semaphore = asyncio.Semaphore(5)
# await model_semaphore.acquire()
# contents = await audio.read()
# data = BytesIO(contents)
# generator = recognition(
# file=data, to_simple=to_simple,
# remove_pun=remove_pun, language=language,
# task=task
# )
# background_tasks = BackgroundTasks()
# background_tasks.add_task(release_model_semaphore)
# return StreamingResponse(generator, background=background_tasks)
@app.post("/embed")
async def api_embed(
textA: str = Body("text1", description="", embed=True),
textB: str = Body("text2", description="", embed=True),
):
q_embeddings = model.encode(textA, normalize_embeddings=True)
p_embeddings = model.encode(textB, normalize_embeddings=True)
scores = q_embeddings @ p_embeddings.T
print(scores)
scores = scores.tolist()
ret = {"similarity score": scores, "status_code": 200}
return ret
# @app.get("/")
# async def index(request: Request):
# return templates.TemplateResponse(
# "index.html", {"request": request, "id": id}
# )
if __name__ == '__main__':
uvicorn.run(app, host=args.host, port=args.port) |