H
commited on
Commit
·
7c30742
1
Parent(s):
da0feec
Add sequence2txt model.py (#1633)
Browse files### What problem does this PR solve?
#1514
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
- rag/llm/__init__.py +10 -1
- rag/llm/sequence2txt_model.py +89 -0
rag/llm/__init__.py
CHANGED
@@ -17,7 +17,7 @@ from .embedding_model import *
|
|
17 |
from .chat_model import *
|
18 |
from .cv_model import *
|
19 |
from .rerank_model import *
|
20 |
-
|
21 |
|
22 |
EmbeddingModel = {
|
23 |
"Ollama": OllamaEmbed,
|
@@ -81,3 +81,12 @@ RerankModel = {
|
|
81 |
"Youdao": YoudaoRerank,
|
82 |
"Xinference": XInferenceRerank
|
83 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
17 |
from .chat_model import *
|
18 |
from .cv_model import *
|
19 |
from .rerank_model import *
|
20 |
+
from .sequence2txt_model import *
|
21 |
|
22 |
EmbeddingModel = {
|
23 |
"Ollama": OllamaEmbed,
|
|
|
81 |
"Youdao": YoudaoRerank,
|
82 |
"Xinference": XInferenceRerank
|
83 |
}
|
84 |
+
|
85 |
+
|
86 |
+
Seq2txtModel = {
|
87 |
+
"OpenAI": GPTSeq2txt,
|
88 |
+
"Tongyi-Qianwen": QWenSeq2txt,
|
89 |
+
"Ollama": OllamaSeq2txt,
|
90 |
+
"Azure-OpenAI": AzureSeq2txt,
|
91 |
+
"Xinference": XinferenceSeq2txt
|
92 |
+
}
|
rag/llm/sequence2txt_model.py
ADDED
@@ -0,0 +1,89 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
#
|
2 |
+
# Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
|
3 |
+
#
|
4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
5 |
+
# you may not use this file except in compliance with the License.
|
6 |
+
# You may obtain a copy of the License at
|
7 |
+
#
|
8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
9 |
+
#
|
10 |
+
# Unless required by applicable law or agreed to in writing, software
|
11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
13 |
+
# See the License for the specific language governing permissions and
|
14 |
+
# limitations under the License.
|
15 |
+
#
|
16 |
+
from openai.lib.azure import AzureOpenAI
|
17 |
+
from zhipuai import ZhipuAI
|
18 |
+
import io
|
19 |
+
from abc import ABC
|
20 |
+
from ollama import Client
|
21 |
+
from openai import OpenAI
|
22 |
+
import os
|
23 |
+
import json
|
24 |
+
from rag.utils import num_tokens_from_string
|
25 |
+
|
26 |
+
|
27 |
+
class Base(ABC):
|
28 |
+
def __init__(self, key, model_name):
|
29 |
+
pass
|
30 |
+
|
31 |
+
def transcription(self, audio, **kwargs):
|
32 |
+
transcription = self.client.audio.transcriptions.create(
|
33 |
+
model=self.model_name,
|
34 |
+
file=audio,
|
35 |
+
response_format="text"
|
36 |
+
)
|
37 |
+
return transcription.text.strip(), num_tokens_from_string(transcription.text.strip())
|
38 |
+
|
39 |
+
|
40 |
+
class GPTSeq2txt(Base):
|
41 |
+
def __init__(self, key, model_name="whisper-1", base_url="https://api.openai.com/v1"):
|
42 |
+
if not base_url: base_url = "https://api.openai.com/v1"
|
43 |
+
self.client = OpenAI(api_key=key, base_url=base_url)
|
44 |
+
self.model_name = model_name
|
45 |
+
|
46 |
+
|
47 |
+
class QWenSeq2txt(Base):
|
48 |
+
def __init__(self, key, model_name="paraformer-realtime-8k-v1", **kwargs):
|
49 |
+
import dashscope
|
50 |
+
dashscope.api_key = key
|
51 |
+
self.model_name = model_name
|
52 |
+
|
53 |
+
def transcription(self, audio, format):
|
54 |
+
from http import HTTPStatus
|
55 |
+
from dashscope.audio.asr import Recognition
|
56 |
+
|
57 |
+
recognition = Recognition(model=self.model_name,
|
58 |
+
format=format,
|
59 |
+
sample_rate=16000,
|
60 |
+
callback=None)
|
61 |
+
result = recognition.call(audio)
|
62 |
+
|
63 |
+
ans = ""
|
64 |
+
if result.status_code == HTTPStatus.OK:
|
65 |
+
for sentence in result.get_sentence():
|
66 |
+
ans += str(sentence + '\n')
|
67 |
+
return ans, num_tokens_from_string(ans)
|
68 |
+
|
69 |
+
return "**ERROR**: " + result.message, 0
|
70 |
+
|
71 |
+
|
72 |
+
class OllamaSeq2txt(Base):
|
73 |
+
def __init__(self, key, model_name, lang="Chinese", **kwargs):
|
74 |
+
self.client = Client(host=kwargs["base_url"])
|
75 |
+
self.model_name = model_name
|
76 |
+
self.lang = lang
|
77 |
+
|
78 |
+
|
79 |
+
class AzureSeq2txt(Base):
|
80 |
+
def __init__(self, key, model_name, lang="Chinese", **kwargs):
|
81 |
+
self.client = AzureOpenAI(api_key=key, azure_endpoint=kwargs["base_url"], api_version="2024-02-01")
|
82 |
+
self.model_name = model_name
|
83 |
+
self.lang = lang
|
84 |
+
|
85 |
+
|
86 |
+
class XinferenceSeq2txt(Base):
|
87 |
+
def __init__(self, key, model_name="", base_url=""):
|
88 |
+
self.client = OpenAI(api_key="xxx", base_url=base_url)
|
89 |
+
self.model_name = model_name
|