-
Notifications
You must be signed in to change notification settings - Fork 43
/
Copy pathwebui.py
327 lines (262 loc) · 11 KB
/
webui.py
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
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
# -*- encoding: utf-8 -*-
# @Author: SWHL
# @Contact: [email protected]
import importlib
import shutil
import time
import uuid
from pathlib import Path
from typing import Dict
import numpy as np
import streamlit as st
from rapid_rag.encoder import EncodeText, ErnieEncodeText
from rapid_rag.file_loader import FileLoader
from rapid_rag.utils import get_timestamp, logger, make_prompt, mkdir, read_yaml
from rapid_rag.vector_utils import DBUtils
config = read_yaml("knowledge_qa_llm/config.yaml")
st.set_page_config(
page_title=config.get("title"),
page_icon=":robot:",
)
def init_ui_parameters():
st.session_state["params"] = {}
param = config.get("Parameter")
st.sidebar.markdown("### 🛶 参数设置")
param_max_length = param.get("max_length")
max_length = st.sidebar.slider(
"max_length",
min_value=param_max_length.get("min_value"),
max_value=param_max_length.get("max_value"),
value=param_max_length.get("default"),
step=param_max_length.get("step"),
help=param_max_length.get("tip"),
)
st.session_state["params"]["max_length"] = max_length
param_top = param.get("top_p")
top_p = st.sidebar.slider(
"top_p",
min_value=param_top.get("min_value"),
max_value=param_top.get("max_value"),
value=param_top.get("default"),
step=param_top.get("step"),
help=param_top.get("tip"),
)
st.session_state["params"]["top_p"] = top_p
param_temp = param.get("temperature")
temperature = st.sidebar.slider(
"temperature",
min_value=param_temp.get("min_value"),
max_value=param_temp.get("max_value"),
value=param_temp.get("default"),
step=param_temp.get("stemp"),
help=param_temp.get("tip"),
)
st.session_state["params"]["temperature"] = temperature
def init_ui_db():
st.sidebar.markdown("### 🧻 知识库")
uploaded_files = st.sidebar.file_uploader(
"default",
accept_multiple_files=True,
label_visibility="hidden",
help="支持多个文件的选取",
)
upload_dir = config.get("upload_dir")
btn_upload = st.sidebar.button("上传文档并加载")
if btn_upload:
time_stamp = get_timestamp()
doc_dir = Path(upload_dir) / time_stamp
tips("正在上传文件到平台中...", icon="⏳")
for file_data in uploaded_files:
bytes_data = file_data.getvalue()
mkdir(doc_dir)
save_path = doc_dir / file_data.name
with open(save_path, "wb") as f:
f.write(bytes_data)
tips("上传完毕！")
with st.spinner(f"正在从{doc_dir}提取内容...."):
all_doc_contents = file_loader(doc_dir)
pro_text = "提取语义向量..."
batch_size = config.get("encoder_batch_size", 16)
uid = str(uuid.uuid1())
st.session_state["connect_id"] = uid
for file_path, one_doc_contents in all_doc_contents.items():
my_bar = st.sidebar.progress(0, text=pro_text)
content_nums = len(one_doc_contents)
all_embeddings = []
for i in range(0, content_nums, batch_size):
start_idx = i
end_idx = start_idx + batch_size
end_idx = content_nums if end_idx > content_nums else end_idx
cur_contents = one_doc_contents[start_idx:end_idx]
if not cur_contents:
continue
embeddings = embedding_extract(cur_contents)
if embeddings is None or embeddings.size == 0:
continue
all_embeddings.append(embeddings)
my_bar.progress(
end_idx / content_nums,
f"Extract {file_path} datas: [{end_idx}/{content_nums}]",
)
my_bar.empty()
if all_embeddings:
all_embeddings = np.vstack(all_embeddings)
db_tools.insert(file_path, all_embeddings, one_doc_contents, uid)
else:
tips(f"从{file_path}提取向量为空。")
shutil.rmtree(doc_dir.resolve())
tips("现在可以提问问题了哈！")
clear_db_btn = st.sidebar.button("清空知识库")
if clear_db_btn:
db_tools.clear_db()
tips("知识库已经被清空！")
if "connect_id" in st.session_state:
had_files = db_tools.get_files(uid=st.session_state.connect_id)
else:
had_files = db_tools.get_files()
st.session_state.had_file_nums = len(had_files) if had_files else 0
if had_files:
st.sidebar.markdown("已有文档:")
st.sidebar.markdown("\n".join([f" - {v}" for v in had_files]))
@st.cache_resource
def init_encoder(encoder_name: str, **kwargs):
if "ERNIEBot" in encoder_name:
return ErnieEncodeText(**kwargs)
return EncodeText(**kwargs)
def predict(
text,
search_res,
model,
custom_prompt=None,
):
for file, content in search_res.items():
content = "\n".join(content)
one_context = f"**从《{file}》** 检索到相关内容： \n{content}"
bot_print(one_context, avatar="📄")
logger.info(f"Context:\n{one_context}\n")
context = "\n".join(sum(search_res.values(), []))
response, elapse = get_model_response(text, context, custom_prompt, model)
print_res = f"**推理耗时:{elapse:.5f}s**"
bot_print(print_res, avatar="📄")
bot_print(response)
def predict_only_model(text, model):
params_dict = st.session_state["params"]
response = model(text, history=None, **params_dict)
bot_print(response)
def bot_print(content, avatar: str = "🤖"):
with st.chat_message("assistant", avatar=avatar):
message_placeholder = st.empty()
full_response = ""
for chunk in content.split():
full_response += chunk + " "
time.sleep(0.05)
message_placeholder.markdown(full_response + "▌")
message_placeholder.markdown(full_response)
def get_model_response(text, context, custom_prompt, model):
params_dict = st.session_state["params"]
s_model = time.perf_counter()
prompt_msg = make_prompt(text, context, custom_prompt)
logger.info(f"Final prompt: \n{prompt_msg}\n")
response = model(prompt_msg, history=None, **params_dict)
elapse = time.perf_counter() - s_model
logger.info(f"Reponse of LLM: \n{response}\n")
if not response:
response = "抱歉，我并不能正确回答该问题。"
return response, elapse
def tips(txt: str, wait_time: int = 2, icon: str = "🎉"):
st.toast(txt, icon=icon)
time.sleep(wait_time)
if __name__ == "__main__":
title = config.get("title")
version = config.get("version", "0.0.1")
st.markdown(
f"<h3 style='text-align: center;'>{title} v{version}</h3><br/>",
unsafe_allow_html=True,
)
init_ui_parameters()
file_loader = FileLoader()
db_path = config.get("vector_db_path")
db_tools = DBUtils(db_path)
llm_module = importlib.import_module("knowledge_qa_llm.llm")
llm_params: Dict[str, Dict] = config.get("LLM_API")
menu_col1, menu_col2, menu_col3 = st.columns([1, 1, 1])
select_model = menu_col1.selectbox("🎨LLM:", llm_params.keys())
if "ERNIEBot" in select_model:
with st.expander("LLM ErnieBot", expanded=True):
opt_col1, opt_col2 = st.columns([1, 1])
api_type = opt_col1.selectbox(
"API Type(必选)",
options=["aistudio", "qianfan", "yinian"],
help="提供对话能力的后端平台",
)
access_token = opt_col2.text_input(
"Access Token(必填) [如何获得？](https://github.com/PaddlePaddle/ERNIE-Bot-SDK/blob/develop/docs/authentication.md)",
"",
help="用于访问后端平台的access token（参考使用说明获取），如果设置了AK、SK则无需设置此参数",
)
llm_params[select_model]["api_type"] = api_type
if access_token:
llm_params[select_model]["access_token"] = access_token
MODEL_OPTIONS = {
name: getattr(llm_module, name)(**params) for name, params in llm_params.items()
}
encoder_params = config.get("Encoder")
select_encoder = menu_col2.selectbox("🧬提取向量模型:", encoder_params.keys())
if "ERNIEBot" in select_encoder:
with st.expander("提取语义向量 ErnieBot", expanded=True):
opt_col1, opt_col2 = st.columns([1, 1])
extract_api_type = opt_col1.selectbox(
"API Type(必选)",
options=["aistudio", "qianfan", "yinian"],
help="提供对话能力的后端平台",
key="Extract_type",
)
encoder_params[select_encoder]["api_type"] = extract_api_type
extract_access_token = opt_col2.text_input(
"Access Token(必填) [如何获得？](https://github.com/PaddlePaddle/ERNIE-Bot-SDK/blob/develop/docs/authentication.md)",
"",
help="用于访问后端平台的access token（参考使用说明获取），如果设置了AK、SK则无需设置此参数",
key="Extract_token",
)
if extract_access_token:
encoder_params[select_encoder]["access_token"] = extract_access_token
embedding_extract = init_encoder(select_encoder, **encoder_params[select_encoder])
TOP_OPTIONS = [5, 10, 15]
search_top = menu_col3.selectbox("🔍搜索 Top_K:", TOP_OPTIONS)
init_ui_db()
with st.expander("💡Prompt", expanded=False):
text_area = st.empty()
input_prompt = text_area.text_area(
label="Input",
max_chars=500,
height=200,
label_visibility="hidden",
value=config.get("DEFAULT_PROMPT"),
key="input_prompt",
)
input_txt = st.chat_input("问点啥吧！")
if input_txt:
with st.chat_message("user", avatar="😀"):
st.markdown(input_txt)
llm = MODEL_OPTIONS[select_model]
if not input_prompt:
input_prompt = config.get("DEFAULT_PROMPT")
query_embedding = embedding_extract(input_txt)
with st.spinner("正在搜索相关文档..."):
uid = st.session_state.get("connect_id", None)
search_res, search_elapse = db_tools.search_local(
query_embedding, top_k=search_top, uid=uid
)
if search_res is None:
bot_print("从知识库中抽取结果为空，直接采用LLM的本身能力回答。", avatar="📄")
predict_only_model(input_txt, llm)
else:
logger.info(f"使用 {type(llm).__name__}")
res_cxt = f"**Top{search_top}\n(得分从高到低，耗时:{search_elapse:.5f}s):** \n"
bot_print(res_cxt, avatar="📄")
predict(
input_txt,
search_res,
llm,
input_prompt,
)