Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -1,287 +1,189 @@
|
|
1 |
import gradio as gr
|
2 |
import os
|
3 |
-
|
4 |
from langchain_community.document_loaders import PyPDFLoader
|
5 |
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
6 |
from langchain_community.vectorstores import Chroma
|
7 |
from langchain.chains import ConversationalRetrievalChain
|
8 |
-
from langchain_community.embeddings import HuggingFaceEmbeddings
|
9 |
-
from langchain_community.llms import HuggingFacePipeline
|
10 |
-
from langchain.chains import ConversationChain
|
11 |
-
from langchain.memory import ConversationBufferMemory
|
12 |
from langchain_community.llms import HuggingFaceEndpoint
|
13 |
-
|
14 |
from pathlib import Path
|
15 |
import chromadb
|
16 |
from unidecode import unidecode
|
17 |
-
|
18 |
-
from transformers import AutoTokenizer
|
19 |
-
import transformers
|
20 |
-
import torch
|
21 |
-
import tqdm
|
22 |
-
import accelerate
|
23 |
import re
|
24 |
|
25 |
-
|
26 |
-
|
27 |
-
|
28 |
-
|
29 |
-
"
|
30 |
-
"
|
31 |
-
"
|
32 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
"google/flan-t5-xxl"
|
34 |
]
|
35 |
list_llm_simple = [os.path.basename(llm) for llm in list_llm]
|
36 |
|
37 |
-
#
|
38 |
def load_doc(list_file_path, chunk_size, chunk_overlap):
|
39 |
-
# Processing for one document only
|
40 |
-
# loader = PyPDFLoader(file_path)
|
41 |
-
# pages = loader.load()
|
42 |
loaders = [PyPDFLoader(x) for x in list_file_path]
|
43 |
pages = []
|
44 |
for loader in loaders:
|
45 |
pages.extend(loader.load())
|
46 |
-
# text_splitter = RecursiveCharacterTextSplitter(chunk_size = 600, chunk_overlap = 50)
|
47 |
text_splitter = RecursiveCharacterTextSplitter(
|
48 |
-
chunk_size
|
49 |
-
chunk_overlap
|
|
|
50 |
doc_splits = text_splitter.split_documents(pages)
|
51 |
return doc_splits
|
52 |
|
53 |
-
|
54 |
-
# Create vector database
|
55 |
def create_db(splits, collection_name):
|
56 |
embedding = HuggingFaceEmbeddings()
|
57 |
-
|
|
|
58 |
vectordb = Chroma.from_documents(
|
59 |
documents=splits,
|
60 |
embedding=embedding,
|
61 |
client=new_client,
|
62 |
collection_name=collection_name,
|
63 |
-
# persist_directory=default_persist_directory
|
64 |
)
|
65 |
return vectordb
|
66 |
|
67 |
-
|
68 |
-
# Load vector database
|
69 |
-
def load_db():
|
70 |
-
embedding = HuggingFaceEmbeddings()
|
71 |
-
vectordb = Chroma(
|
72 |
-
# persist_directory=default_persist_directory,
|
73 |
-
embedding_function=embedding)
|
74 |
-
return vectordb
|
75 |
-
|
76 |
-
|
77 |
-
# Initialize langchain LLM chain
|
78 |
def initialize_llmchain(llm_model, temperature, max_tokens, top_k, vector_db, progress=gr.Progress()):
|
79 |
-
progress(0.1, desc="
|
80 |
-
|
81 |
-
# Note: it will download model locally...
|
82 |
-
# tokenizer=AutoTokenizer.from_pretrained(llm_model)
|
83 |
-
# progress(0.5, desc="Initializing HF pipeline...")
|
84 |
-
# pipeline=transformers.pipeline(
|
85 |
-
# "text-generation",
|
86 |
-
# model=llm_model,
|
87 |
-
# tokenizer=tokenizer,
|
88 |
-
# torch_dtype=torch.bfloat16,
|
89 |
-
# trust_remote_code=True,
|
90 |
-
# device_map="auto",
|
91 |
-
# # max_length=1024,
|
92 |
-
# max_new_tokens=max_tokens,
|
93 |
-
# do_sample=True,
|
94 |
-
# top_k=top_k,
|
95 |
-
# num_return_sequences=1,
|
96 |
-
# eos_token_id=tokenizer.eos_token_id
|
97 |
-
# )
|
98 |
-
# llm = HuggingFacePipeline(pipeline=pipeline, model_kwargs={'temperature': temperature})
|
99 |
-
|
100 |
-
# HuggingFaceHub uses HF inference endpoints
|
101 |
-
progress(0.5, desc="Initializing HF Hub...")
|
102 |
-
# Use of trust_remote_code as model_kwargs
|
103 |
-
# Warning: langchain issue
|
104 |
-
# URL: https://github.com/langchain-ai/langchain/issues/6080
|
105 |
if llm_model == "mistralai/Mixtral-8x7B-Instruct-v0.1":
|
106 |
llm = HuggingFaceEndpoint(
|
107 |
-
repo_id=llm_model,
|
108 |
-
|
109 |
-
|
110 |
-
|
111 |
-
|
112 |
-
load_in_8bit = True,
|
113 |
-
)
|
114 |
-
elif llm_model in ["HuggingFaceH4/zephyr-7b-gemma-v0.1","mosaicml/mpt-7b-instruct"]:
|
115 |
-
raise gr.Error("LLM model is too large to be loaded automatically on free inference endpoint")
|
116 |
-
llm = HuggingFaceEndpoint(
|
117 |
-
repo_id=llm_model,
|
118 |
-
temperature = temperature,
|
119 |
-
max_new_tokens = max_tokens,
|
120 |
-
top_k = top_k,
|
121 |
)
|
|
|
|
|
122 |
elif llm_model == "microsoft/phi-2":
|
123 |
-
# raise gr.Error("phi-2 model requires 'trust_remote_code=True', currently not supported by langchain HuggingFaceHub...")
|
124 |
llm = HuggingFaceEndpoint(
|
125 |
-
repo_id=llm_model,
|
126 |
-
|
127 |
-
|
128 |
-
|
129 |
-
|
130 |
-
|
131 |
-
torch_dtype = "auto",
|
132 |
)
|
133 |
elif llm_model == "TinyLlama/TinyLlama-1.1B-Chat-v1.0":
|
134 |
llm = HuggingFaceEndpoint(
|
135 |
-
repo_id=llm_model,
|
136 |
-
|
137 |
-
|
138 |
-
|
139 |
-
top_k = top_k,
|
140 |
)
|
141 |
elif llm_model == "meta-llama/Llama-2-7b-chat-hf":
|
142 |
-
raise gr.Error("Llama-2-7b-chat-hf
|
143 |
-
llm = HuggingFaceEndpoint(
|
144 |
-
repo_id=llm_model,
|
145 |
-
# model_kwargs={"temperature": temperature, "max_new_tokens": max_tokens, "top_k": top_k}
|
146 |
-
temperature = temperature,
|
147 |
-
max_new_tokens = max_tokens,
|
148 |
-
top_k = top_k,
|
149 |
-
)
|
150 |
else:
|
151 |
llm = HuggingFaceEndpoint(
|
152 |
-
repo_id=llm_model,
|
153 |
-
|
154 |
-
|
155 |
-
|
156 |
-
max_new_tokens = max_tokens,
|
157 |
-
top_k = top_k,
|
158 |
)
|
159 |
-
|
160 |
-
progress(0.75, desc="
|
161 |
memory = ConversationBufferMemory(
|
162 |
memory_key="chat_history",
|
163 |
output_key='answer',
|
164 |
return_messages=True
|
165 |
)
|
166 |
-
|
167 |
-
|
168 |
-
progress(0.8, desc="Defining retrieval chain...")
|
169 |
qa_chain = ConversationalRetrievalChain.from_llm(
|
170 |
llm,
|
171 |
retriever=retriever,
|
172 |
-
chain_type="stuff",
|
173 |
memory=memory,
|
174 |
-
# combine_docs_chain_kwargs={"prompt": your_prompt})
|
175 |
return_source_documents=True,
|
176 |
-
#return_generated_question=False,
|
177 |
verbose=False,
|
178 |
)
|
179 |
-
progress(0.9, desc="
|
180 |
return qa_chain
|
181 |
|
182 |
-
|
183 |
-
# Generate collection name for vector database
|
184 |
-
# - Use filepath as input, ensuring unicode text
|
185 |
def create_collection_name(filepath):
|
186 |
-
# Extract filename without extension
|
187 |
collection_name = Path(filepath).stem
|
188 |
-
|
189 |
-
## Remove space
|
190 |
-
collection_name = collection_name.replace(" ","-")
|
191 |
-
## ASCII transliterations of Unicode text
|
192 |
collection_name = unidecode(collection_name)
|
193 |
-
## Remove special characters
|
194 |
-
#collection_name = re.findall("[\dA-Za-z]*", collection_name)[0]
|
195 |
collection_name = re.sub('[^A-Za-z0-9]+', '-', collection_name)
|
196 |
-
## Limit length to 50 characters
|
197 |
collection_name = collection_name[:50]
|
198 |
-
## Minimum length of 3 characters
|
199 |
if len(collection_name) < 3:
|
200 |
collection_name = collection_name + 'xyz'
|
201 |
-
## Enforce start and end as alphanumeric character
|
202 |
if not collection_name[0].isalnum():
|
203 |
collection_name = 'A' + collection_name[1:]
|
204 |
if not collection_name[-1].isalnum():
|
205 |
collection_name = collection_name[:-1] + 'Z'
|
206 |
-
print('
|
207 |
-
print('
|
208 |
return collection_name
|
209 |
|
210 |
-
|
211 |
-
# Initialize database
|
212 |
def initialize_database(list_file_obj, chunk_size, chunk_overlap, progress=gr.Progress()):
|
213 |
-
# Create list of documents (when valid)
|
214 |
list_file_path = [x.name for x in list_file_obj if x is not None]
|
215 |
-
|
216 |
-
progress(0.1, desc="Creating collection name...")
|
217 |
collection_name = create_collection_name(list_file_path[0])
|
218 |
-
progress(0.25, desc="
|
219 |
-
# Load document and create splits
|
220 |
doc_splits = load_doc(list_file_path, chunk_size, chunk_overlap)
|
221 |
-
|
222 |
-
progress(0.5, desc="Generating vector database...")
|
223 |
-
# global vector_db
|
224 |
vector_db = create_db(doc_splits, collection_name)
|
225 |
-
progress(0.9, desc="
|
226 |
-
return vector_db, collection_name, "
|
227 |
-
|
228 |
|
|
|
229 |
def initialize_LLM(llm_option, llm_temperature, max_tokens, top_k, vector_db, progress=gr.Progress()):
|
230 |
-
# print("llm_option",llm_option)
|
231 |
llm_name = list_llm[llm_option]
|
232 |
-
print("
|
233 |
qa_chain = initialize_llmchain(llm_name, llm_temperature, max_tokens, top_k, vector_db, progress)
|
234 |
-
return qa_chain, "
|
235 |
-
|
236 |
|
|
|
237 |
def format_chat_history(message, chat_history):
|
238 |
formatted_chat_history = []
|
239 |
for user_message, bot_message in chat_history:
|
240 |
-
formatted_chat_history.append(f"
|
241 |
-
formatted_chat_history.append(f"
|
242 |
return formatted_chat_history
|
243 |
-
|
244 |
|
|
|
245 |
def conversation(qa_chain, message, history):
|
246 |
formatted_chat_history = format_chat_history(message, history)
|
247 |
-
#print("formatted_chat_history",formatted_chat_history)
|
248 |
-
|
249 |
-
# Generate response using QA chain
|
250 |
response = qa_chain({"question": message, "chat_history": formatted_chat_history})
|
251 |
response_answer = response["answer"]
|
252 |
-
if response_answer.find("
|
253 |
-
response_answer = response_answer.split("
|
254 |
response_sources = response["source_documents"]
|
255 |
response_source1 = response_sources[0].page_content.strip()
|
256 |
response_source2 = response_sources[1].page_content.strip()
|
257 |
response_source3 = response_sources[2].page_content.strip()
|
258 |
-
# Langchain sources are zero-based
|
259 |
response_source1_page = response_sources[0].metadata["page"] + 1
|
260 |
response_source2_page = response_sources[1].metadata["page"] + 1
|
261 |
response_source3_page = response_sources[2].metadata["page"] + 1
|
262 |
-
# print ('chat response: ', response_answer)
|
263 |
-
# print('DB source', response_sources)
|
264 |
-
|
265 |
-
# Append user message and response to chat history
|
266 |
new_history = history + [(message, response_answer)]
|
267 |
-
# return gr.update(value=""), new_history, response_sources[0], response_sources[1]
|
268 |
return qa_chain, gr.update(value=""), new_history, response_source1, response_source1_page, response_source2, response_source2_page, response_source3, response_source3_page
|
269 |
-
|
270 |
|
|
|
271 |
def upload_file(file_obj):
|
272 |
list_file_path = []
|
273 |
for idx, file in enumerate(file_obj):
|
274 |
file_path = file_obj.name
|
275 |
list_file_path.append(file_path)
|
276 |
-
# print(file_path)
|
277 |
-
# initialize_database(file_path, progress)
|
278 |
return list_file_path
|
279 |
|
280 |
-
|
281 |
-
import gradio as gr
|
282 |
-
|
283 |
-
import gradio as gr
|
284 |
-
|
285 |
def demo():
|
286 |
with gr.Blocks(theme="base") as demo:
|
287 |
vector_db = gr.State()
|
@@ -289,82 +191,85 @@ def demo():
|
|
289 |
collection_name = gr.State()
|
290 |
|
291 |
gr.Markdown(
|
292 |
-
"""<center><h2>PDF
|
293 |
-
<h3>
|
294 |
gr.Markdown(
|
295 |
-
"""<b>
|
296 |
-
|
297 |
-
|
298 |
-
<br><b>
|
299 |
""")
|
300 |
|
301 |
-
with gr.Tab("
|
302 |
with gr.Row():
|
303 |
-
document = gr.Files(height=100, file_count="multiple", file_types=["pdf"], interactive=True, label="
|
|
|
304 |
|
305 |
-
with gr.Tab("
|
306 |
with gr.Row():
|
307 |
-
db_btn = gr.Radio(["ChromaDB"], label="
|
308 |
-
with gr.Accordion("
|
309 |
with gr.Row():
|
310 |
-
slider_chunk_size = gr.Slider(minimum = 100, maximum = 1000, value=600, step=20, label="
|
311 |
with gr.Row():
|
312 |
-
slider_chunk_overlap = gr.Slider(minimum = 10, maximum = 200, value=40, step=10, label="
|
313 |
with gr.Row():
|
314 |
-
db_progress = gr.Textbox(label="
|
315 |
with gr.Row():
|
316 |
-
db_btn = gr.Button("
|
317 |
|
318 |
-
with gr.Tab("
|
319 |
with gr.Row():
|
320 |
-
llm_btn = gr.Radio(
|
321 |
-
|
|
|
322 |
with gr.Row():
|
323 |
-
slider_temperature = gr.Slider(minimum = 0.01, maximum = 1.0, value=0.7, step=0.1, label="
|
324 |
with gr.Row():
|
325 |
-
slider_maxtokens = gr.Slider(minimum = 224, maximum = 4096, value=1024, step=32, label="
|
326 |
with gr.Row():
|
327 |
-
slider_topk = gr.Slider(minimum = 1, maximum = 10, value=3, step=1, label="top-k
|
328 |
with gr.Row():
|
329 |
-
llm_progress = gr.Textbox(value="
|
330 |
with gr.Row():
|
331 |
-
qachain_btn = gr.Button("
|
332 |
|
333 |
-
with gr.Tab("
|
334 |
chatbot = gr.Chatbot(height=300)
|
335 |
-
with gr.Accordion("
|
336 |
with gr.Row():
|
337 |
-
doc_source1 = gr.Textbox(label="
|
338 |
-
source1_page = gr.Number(label="
|
339 |
with gr.Row():
|
340 |
-
doc_source2 = gr.Textbox(label="
|
341 |
-
source2_page = gr.Number(label="
|
342 |
with gr.Row():
|
343 |
-
doc_source3 = gr.Textbox(label="
|
344 |
-
source3_page = gr.Number(label="
|
345 |
with gr.Row():
|
346 |
-
msg = gr.Textbox(placeholder="
|
347 |
with gr.Row():
|
348 |
-
submit_btn = gr.Button("
|
349 |
-
clear_btn = gr.ClearButton([msg, chatbot], value="
|
350 |
|
351 |
-
#
|
352 |
-
|
|
|
353 |
inputs=[document, slider_chunk_size, slider_chunk_overlap], \
|
354 |
outputs=[vector_db, collection_name, db_progress])
|
355 |
-
qachain_btn.click(
|
356 |
inputs=[llm_btn, slider_temperature, slider_maxtokens, slider_topk, vector_db], \
|
357 |
outputs=[qa_chain, llm_progress]).then(lambda:[None,"",0,"",0,"",0], \
|
358 |
inputs=None, \
|
359 |
outputs=[chatbot, doc_source1, source1_page, doc_source2, source2_page, doc_source3, source3_page], \
|
360 |
queue=False)
|
361 |
|
362 |
-
# Chatbot
|
363 |
-
msg.submit(
|
364 |
inputs=[qa_chain, msg, chatbot], \
|
365 |
outputs=[qa_chain, msg, chatbot, doc_source1, source1_page, doc_source2, source2_page, doc_source3, source3_page], \
|
366 |
queue=False)
|
367 |
-
submit_btn.click(
|
368 |
inputs=[qa_chain, msg, chatbot], \
|
369 |
outputs=[qa_chain, msg, chatbot, doc_source1, source1_page, doc_source2, source2_page, doc_source3, source3_page], \
|
370 |
queue=False)
|
@@ -374,5 +279,6 @@ def demo():
|
|
374 |
queue=False)
|
375 |
demo.queue().launch(debug=True)
|
376 |
|
|
|
377 |
if __name__ == "__main__":
|
378 |
-
demo()
|
|
|
1 |
import gradio as gr
|
2 |
import os
|
|
|
3 |
from langchain_community.document_loaders import PyPDFLoader
|
4 |
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
5 |
from langchain_community.vectorstores import Chroma
|
6 |
from langchain.chains import ConversationalRetrievalChain
|
7 |
+
from langchain_community.embeddings import HuggingFaceEmbeddings
|
|
|
|
|
|
|
8 |
from langchain_community.llms import HuggingFaceEndpoint
|
9 |
+
from langchain.memory import ConversationBufferMemory
|
10 |
from pathlib import Path
|
11 |
import chromadb
|
12 |
from unidecode import unidecode
|
|
|
|
|
|
|
|
|
|
|
|
|
13 |
import re
|
14 |
|
15 |
+
# Lista de modelos LLM disponíveis
|
16 |
+
list_llm = [
|
17 |
+
"mistralai/Mistral-7B-Instruct-v0.2",
|
18 |
+
"mistralai/Mixtral-8x7B-Instruct-v0.1",
|
19 |
+
"mistralai/Mistral-7B-Instruct-v0.1",
|
20 |
+
"google/gemma-7b-it",
|
21 |
+
"google/gemma-2b-it",
|
22 |
+
"HuggingFaceH4/zephyr-7b-beta",
|
23 |
+
"HuggingFaceH4/zephyr-7b-gemma-v0.1",
|
24 |
+
"meta-llama/Llama-2-7b-chat-hf",
|
25 |
+
"microsoft/phi-2",
|
26 |
+
"TinyLlama/TinyLlama-1.1B-Chat-v1.0",
|
27 |
+
"mosaicml/mpt-7b-instruct",
|
28 |
+
"tiiuae/falcon-7b-instruct",
|
29 |
"google/flan-t5-xxl"
|
30 |
]
|
31 |
list_llm_simple = [os.path.basename(llm) for llm in list_llm]
|
32 |
|
33 |
+
# Função para carregar documentos PDF e dividir em chunks
|
34 |
def load_doc(list_file_path, chunk_size, chunk_overlap):
|
|
|
|
|
|
|
35 |
loaders = [PyPDFLoader(x) for x in list_file_path]
|
36 |
pages = []
|
37 |
for loader in loaders:
|
38 |
pages.extend(loader.load())
|
|
|
39 |
text_splitter = RecursiveCharacterTextSplitter(
|
40 |
+
chunk_size=chunk_size,
|
41 |
+
chunk_overlap=chunk_overlap
|
42 |
+
)
|
43 |
doc_splits = text_splitter.split_documents(pages)
|
44 |
return doc_splits
|
45 |
|
46 |
+
# Função para criar o banco de dados vetorial
|
|
|
47 |
def create_db(splits, collection_name):
|
48 |
embedding = HuggingFaceEmbeddings()
|
49 |
+
# Usando PersistentClient para persistir o banco de dados
|
50 |
+
new_client = chromadb.PersistentClient(path="./chroma_db")
|
51 |
vectordb = Chroma.from_documents(
|
52 |
documents=splits,
|
53 |
embedding=embedding,
|
54 |
client=new_client,
|
55 |
collection_name=collection_name,
|
|
|
56 |
)
|
57 |
return vectordb
|
58 |
|
59 |
+
# Função para inicializar a cadeia de QA com o modelo LLM
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
60 |
def initialize_llmchain(llm_model, temperature, max_tokens, top_k, vector_db, progress=gr.Progress()):
|
61 |
+
progress(0.1, desc="Inicializando tokenizer da HF...")
|
62 |
+
progress(0.5, desc="Inicializando Hub da HF...")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
63 |
if llm_model == "mistralai/Mixtral-8x7B-Instruct-v0.1":
|
64 |
llm = HuggingFaceEndpoint(
|
65 |
+
repo_id=llm_model,
|
66 |
+
temperature=temperature,
|
67 |
+
max_new_tokens=max_tokens,
|
68 |
+
top_k=top_k,
|
69 |
+
load_in_8bit=True,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
70 |
)
|
71 |
+
elif llm_model in ["HuggingFaceH4/zephyr-7b-gemma-v0.1", "mosaicml/mpt-7b-instruct"]:
|
72 |
+
raise gr.Error("O modelo LLM é muito grande para ser carregado automaticamente no endpoint de inferência gratuito")
|
73 |
elif llm_model == "microsoft/phi-2":
|
|
|
74 |
llm = HuggingFaceEndpoint(
|
75 |
+
repo_id=llm_model,
|
76 |
+
temperature=temperature,
|
77 |
+
max_new_tokens=max_tokens,
|
78 |
+
top_k=top_k,
|
79 |
+
trust_remote_code=True,
|
80 |
+
torch_dtype="auto",
|
|
|
81 |
)
|
82 |
elif llm_model == "TinyLlama/TinyLlama-1.1B-Chat-v1.0":
|
83 |
llm = HuggingFaceEndpoint(
|
84 |
+
repo_id=llm_model,
|
85 |
+
temperature=temperature,
|
86 |
+
max_new_tokens=250,
|
87 |
+
top_k=top_k,
|
|
|
88 |
)
|
89 |
elif llm_model == "meta-llama/Llama-2-7b-chat-hf":
|
90 |
+
raise gr.Error("O modelo Llama-2-7b-chat-hf requer uma assinatura Pro...")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
91 |
else:
|
92 |
llm = HuggingFaceEndpoint(
|
93 |
+
repo_id=llm_model,
|
94 |
+
temperature=temperature,
|
95 |
+
max_new_tokens=max_tokens,
|
96 |
+
top_k=top_k,
|
|
|
|
|
97 |
)
|
98 |
+
|
99 |
+
progress(0.75, desc="Definindo memória de buffer...")
|
100 |
memory = ConversationBufferMemory(
|
101 |
memory_key="chat_history",
|
102 |
output_key='answer',
|
103 |
return_messages=True
|
104 |
)
|
105 |
+
retriever = vector_db.as_retriever()
|
106 |
+
progress(0.8, desc="Definindo cadeia de recuperação...")
|
|
|
107 |
qa_chain = ConversationalRetrievalChain.from_llm(
|
108 |
llm,
|
109 |
retriever=retriever,
|
110 |
+
chain_type="stuff",
|
111 |
memory=memory,
|
|
|
112 |
return_source_documents=True,
|
|
|
113 |
verbose=False,
|
114 |
)
|
115 |
+
progress(0.9, desc="Concluído!")
|
116 |
return qa_chain
|
117 |
|
118 |
+
# Função para gerar um nome de coleção válido
|
|
|
|
|
119 |
def create_collection_name(filepath):
|
|
|
120 |
collection_name = Path(filepath).stem
|
121 |
+
collection_name = collection_name.replace(" ", "-")
|
|
|
|
|
|
|
122 |
collection_name = unidecode(collection_name)
|
|
|
|
|
123 |
collection_name = re.sub('[^A-Za-z0-9]+', '-', collection_name)
|
|
|
124 |
collection_name = collection_name[:50]
|
|
|
125 |
if len(collection_name) < 3:
|
126 |
collection_name = collection_name + 'xyz'
|
|
|
127 |
if not collection_name[0].isalnum():
|
128 |
collection_name = 'A' + collection_name[1:]
|
129 |
if not collection_name[-1].isalnum():
|
130 |
collection_name = collection_name[:-1] + 'Z'
|
131 |
+
print('Caminho do arquivo: ', filepath)
|
132 |
+
print('Nome da coleção: ', collection_name)
|
133 |
return collection_name
|
134 |
|
135 |
+
# Função para inicializar o banco de dados
|
|
|
136 |
def initialize_database(list_file_obj, chunk_size, chunk_overlap, progress=gr.Progress()):
|
|
|
137 |
list_file_path = [x.name for x in list_file_obj if x is not None]
|
138 |
+
progress(0.1, desc="Criando nome da coleção...")
|
|
|
139 |
collection_name = create_collection_name(list_file_path[0])
|
140 |
+
progress(0.25, desc="Carregando documento...")
|
|
|
141 |
doc_splits = load_doc(list_file_path, chunk_size, chunk_overlap)
|
142 |
+
progress(0.5, desc="Gerando banco de dados vetorial...")
|
|
|
|
|
143 |
vector_db = create_db(doc_splits, collection_name)
|
144 |
+
progress(0.9, desc="Concluído!")
|
145 |
+
return vector_db, collection_name, "Completo!"
|
|
|
146 |
|
147 |
+
# Função para inicializar o modelo LLM
|
148 |
def initialize_LLM(llm_option, llm_temperature, max_tokens, top_k, vector_db, progress=gr.Progress()):
|
|
|
149 |
llm_name = list_llm[llm_option]
|
150 |
+
print("Nome do LLM: ", llm_name)
|
151 |
qa_chain = initialize_llmchain(llm_name, llm_temperature, max_tokens, top_k, vector_db, progress)
|
152 |
+
return qa_chain, "Completo!"
|
|
|
153 |
|
154 |
+
# Função para formatar o histórico de conversa
|
155 |
def format_chat_history(message, chat_history):
|
156 |
formatted_chat_history = []
|
157 |
for user_message, bot_message in chat_history:
|
158 |
+
formatted_chat_history.append(f"Usuário: {user_message}")
|
159 |
+
formatted_chat_history.append(f"Assistente: {bot_message}")
|
160 |
return formatted_chat_history
|
|
|
161 |
|
162 |
+
# Função para realizar a conversa com o chatbot
|
163 |
def conversation(qa_chain, message, history):
|
164 |
formatted_chat_history = format_chat_history(message, history)
|
|
|
|
|
|
|
165 |
response = qa_chain({"question": message, "chat_history": formatted_chat_history})
|
166 |
response_answer = response["answer"]
|
167 |
+
if response_answer.find("Resposta útil:") != -1:
|
168 |
+
response_answer = response_answer.split("Resposta útil:")[-1]
|
169 |
response_sources = response["source_documents"]
|
170 |
response_source1 = response_sources[0].page_content.strip()
|
171 |
response_source2 = response_sources[1].page_content.strip()
|
172 |
response_source3 = response_sources[2].page_content.strip()
|
|
|
173 |
response_source1_page = response_sources[0].metadata["page"] + 1
|
174 |
response_source2_page = response_sources[1].metadata["page"] + 1
|
175 |
response_source3_page = response_sources[2].metadata["page"] + 1
|
|
|
|
|
|
|
|
|
176 |
new_history = history + [(message, response_answer)]
|
|
|
177 |
return qa_chain, gr.update(value=""), new_history, response_source1, response_source1_page, response_source2, response_source2_page, response_source3, response_source3_page
|
|
|
178 |
|
179 |
+
# Função para carregar arquivos
|
180 |
def upload_file(file_obj):
|
181 |
list_file_path = []
|
182 |
for idx, file in enumerate(file_obj):
|
183 |
file_path = file_obj.name
|
184 |
list_file_path.append(file_path)
|
|
|
|
|
185 |
return list_file_path
|
186 |
|
|
|
|
|
|
|
|
|
|
|
187 |
def demo():
|
188 |
with gr.Blocks(theme="base") as demo:
|
189 |
vector_db = gr.State()
|
|
|
191 |
collection_name = gr.State()
|
192 |
|
193 |
gr.Markdown(
|
194 |
+
"""<center><h2>Chatbot baseado em PDF</center></h2>
|
195 |
+
<h3>Faça qualquer pergunta sobre seus documentos PDF</h3>""")
|
196 |
gr.Markdown(
|
197 |
+
"""<b>Nota:</b> Este assistente de IA, utilizando Langchain e LLMs de código aberto, realiza geração aumentada por recuperação (RAG) a partir de seus documentos PDF. \
|
198 |
+
A interface do usuário mostra explicitamente várias etapas para ajudar a entender o fluxo de trabalho do RAG.
|
199 |
+
Este chatbot leva em consideração perguntas anteriores ao gerar respostas (via memória conversacional), e inclui referências documentais para maior clareza.<br>
|
200 |
+
<br><b>Aviso:</b> Este espaço usa a CPU básica gratuita do Hugging Face. Algumas etapas e modelos LLM utilizados abaixo (pontos finais de inferência gratuitos) podem levar algum tempo para gerar uma resposta.
|
201 |
""")
|
202 |
|
203 |
+
with gr.Tab("Etapa 1 - Carregar PDF"):
|
204 |
with gr.Row():
|
205 |
+
document = gr.Files(height=100, file_count="multiple", file_types=["pdf"], interactive=True, label="Carregue seus documentos PDF (único ou múltiplos)")
|
206 |
+
# upload_btn = gr.UploadButton("Carregando documento...", height=100, file_count="multiple", file_types=["pdf"], scale=1)
|
207 |
|
208 |
+
with gr.Tab("Etapa 2 - Processar documento"):
|
209 |
with gr.Row():
|
210 |
+
db_btn = gr.Radio(["ChromaDB"], label="Tipo de banco de dados vetorial", value = "ChromaDB", type="index", info="Escolha o banco de dados vetorial")
|
211 |
+
with gr.Accordion("Opções avançadas - Divisor de texto do documento", open=False):
|
212 |
with gr.Row():
|
213 |
+
slider_chunk_size = gr.Slider(minimum = 100, maximum = 1000, value=600, step=20, label="Tamanho do bloco", info="Tamanho do bloco", interactive=True)
|
214 |
with gr.Row():
|
215 |
+
slider_chunk_overlap = gr.Slider(minimum = 10, maximum = 200, value=40, step=10, label="Sobreposição do bloco", info="Sobreposição do bloco", interactive=True)
|
216 |
with gr.Row():
|
217 |
+
db_progress = gr.Textbox(label="Inicialização do banco de dados vetorial", value="Nenhum")
|
218 |
with gr.Row():
|
219 |
+
db_btn = gr.Button("Gerar banco de dados vetorial")
|
220 |
|
221 |
+
with gr.Tab("Etapa 3 - Inicializar cadeia de QA"):
|
222 |
with gr.Row():
|
223 |
+
llm_btn = gr.Radio(list_llm_simple, \
|
224 |
+
label="Modelos LLM", value = list_llm_simple[0], type="index", info="Escolha seu modelo LLM")
|
225 |
+
with gr.Accordion("Opções avançadas - Modelo LLM", open=False):
|
226 |
with gr.Row():
|
227 |
+
slider_temperature = gr.Slider(minimum = 0.01, maximum = 1.0, value=0.7, step=0.1, label="Temperatura", info="Temperatura do modelo", interactive=True)
|
228 |
with gr.Row():
|
229 |
+
slider_maxtokens = gr.Slider(minimum = 224, maximum = 4096, value=1024, step=32, label="Máximo de Tokens", info="Máximo de tokens do modelo", interactive=True)
|
230 |
with gr.Row():
|
231 |
+
slider_topk = gr.Slider(minimum = 1, maximum = 10, value=3, step=1, label="Amostras top-k", info="Amostras top-k do modelo", interactive=True)
|
232 |
with gr.Row():
|
233 |
+
llm_progress = gr.Textbox(value="Nenhum",label="Inicialização da cadeia QA")
|
234 |
with gr.Row():
|
235 |
+
qachain_btn = gr.Button("Inicializar cadeia de Pergunta e Resposta")
|
236 |
|
237 |
+
with gr.Tab("Etapa 4 - Chatbot"):
|
238 |
chatbot = gr.Chatbot(height=300)
|
239 |
+
with gr.Accordion("Avançado - Referências do documento", open=False):
|
240 |
with gr.Row():
|
241 |
+
doc_source1 = gr.Textbox(label="Referência 1", lines=2, container=True, scale=20)
|
242 |
+
source1_page = gr.Number(label="Página", scale=1)
|
243 |
with gr.Row():
|
244 |
+
doc_source2 = gr.Textbox(label="Referência 2", lines=2, container=True, scale=20)
|
245 |
+
source2_page = gr.Number(label="Página", scale=1)
|
246 |
with gr.Row():
|
247 |
+
doc_source3 = gr.Textbox(label="Referência 3", lines=2, container=True, scale=20)
|
248 |
+
source3_page = gr.Number(label="Página", scale=1)
|
249 |
with gr.Row():
|
250 |
+
msg = gr.Textbox(placeholder="Digite a mensagem (exemplo: 'Sobre o que é este documento?')", container=True)
|
251 |
with gr.Row():
|
252 |
+
submit_btn = gr.Button("Enviar mensagem")
|
253 |
+
clear_btn = gr.ClearButton([msg, chatbot], value="Limpar conversa")
|
254 |
|
255 |
+
# Eventos de pré-processamento
|
256 |
+
#upload_btn.upload(upload_file, inputs=[upload_btn], outputs=[document])
|
257 |
+
db_btn.click(initialize_database, \
|
258 |
inputs=[document, slider_chunk_size, slider_chunk_overlap], \
|
259 |
outputs=[vector_db, collection_name, db_progress])
|
260 |
+
qachain_btn.click(initialize_LLM, \
|
261 |
inputs=[llm_btn, slider_temperature, slider_maxtokens, slider_topk, vector_db], \
|
262 |
outputs=[qa_chain, llm_progress]).then(lambda:[None,"",0,"",0,"",0], \
|
263 |
inputs=None, \
|
264 |
outputs=[chatbot, doc_source1, source1_page, doc_source2, source2_page, doc_source3, source3_page], \
|
265 |
queue=False)
|
266 |
|
267 |
+
# Eventos do Chatbot
|
268 |
+
msg.submit(conversation, \
|
269 |
inputs=[qa_chain, msg, chatbot], \
|
270 |
outputs=[qa_chain, msg, chatbot, doc_source1, source1_page, doc_source2, source2_page, doc_source3, source3_page], \
|
271 |
queue=False)
|
272 |
+
submit_btn.click(conversation, \
|
273 |
inputs=[qa_chain, msg, chatbot], \
|
274 |
outputs=[qa_chain, msg, chatbot, doc_source1, source1_page, doc_source2, source2_page, doc_source3, source3_page], \
|
275 |
queue=False)
|
|
|
279 |
queue=False)
|
280 |
demo.queue().launch(debug=True)
|
281 |
|
282 |
+
|
283 |
if __name__ == "__main__":
|
284 |
+
demo()
|