Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -3,14 +3,25 @@ from transformers import pipeline
|
|
| 3 |
from langchain_community.llms import HuggingFacePipeline
|
| 4 |
from langchain_community.embeddings import HuggingFaceEmbeddings
|
| 5 |
from langchain_community.vectorstores import FAISS
|
| 6 |
-
from
|
| 7 |
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
| 8 |
from langchain.chains import ConversationalRetrievalChain
|
| 9 |
from langchain.memory import ConversationBufferMemory
|
| 10 |
-
from langchain.agents import Tool, initialize_agent
|
| 11 |
-
from langchain.agents.agent_types import AgentType
|
| 12 |
|
| 13 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
pipe = pipeline(
|
| 15 |
"text2text-generation",
|
| 16 |
model="declare-lab/flan-alpaca-base",
|
|
@@ -20,81 +31,39 @@ pipe = pipeline(
|
|
| 20 |
)
|
| 21 |
llm = HuggingFacePipeline(pipeline=pipe)
|
| 22 |
|
| 23 |
-
# Mem贸ria limitada para evitar prompts muito longos
|
| 24 |
memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True, k=3)
|
| 25 |
|
| 26 |
-
#
|
| 27 |
-
def carregar_documento_txt(path):
|
| 28 |
-
loader = TextLoader(path, encoding="latin-1")
|
| 29 |
-
documentos = loader.load()
|
| 30 |
-
splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=100)
|
| 31 |
-
documentos_divididos = splitter.split_documents(documentos)
|
| 32 |
-
embeddings = HuggingFaceEmbeddings()
|
| 33 |
-
vectorstore = FAISS.from_documents(documentos_divididos, embeddings)
|
| 34 |
-
return vectorstore
|
| 35 |
-
|
| 36 |
-
vectorstore = carregar_documento_txt("meu_documento.txt")
|
| 37 |
-
|
| 38 |
-
# Cadeia de busca com mem贸ria
|
| 39 |
chat_chain = ConversationalRetrievalChain.from_llm(
|
| 40 |
llm=llm,
|
| 41 |
retriever=vectorstore.as_retriever(),
|
| 42 |
memory=memory,
|
| 43 |
)
|
| 44 |
|
| 45 |
-
#
|
| 46 |
-
def
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
return
|
| 51 |
-
|
| 52 |
-
tools = [
|
| 53 |
-
Tool(
|
| 54 |
-
name="Responder com Documento",
|
| 55 |
-
func=responder_documento,
|
| 56 |
-
description="Use para responder perguntas espec铆ficas sobre o documento."
|
| 57 |
-
),
|
| 58 |
-
Tool(
|
| 59 |
-
name="Responder Geral",
|
| 60 |
-
func=responder_geral,
|
| 61 |
-
description="Use para perguntas gerais ou bate-papo."
|
| 62 |
-
)
|
| 63 |
-
]
|
| 64 |
-
|
| 65 |
-
# Agente com tratamento de parsing
|
| 66 |
-
agent = initialize_agent(
|
| 67 |
-
tools=tools,
|
| 68 |
-
llm=llm,
|
| 69 |
-
agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION,
|
| 70 |
-
verbose=False,
|
| 71 |
-
handle_parsing_errors=True,
|
| 72 |
-
)
|
| 73 |
|
| 74 |
-
# Interface Gradio
|
| 75 |
with gr.Blocks() as demo:
|
| 76 |
-
gr.Markdown("## 馃 Chat
|
|
|
|
|
|
|
|
|
|
| 77 |
|
| 78 |
-
|
| 79 |
-
|
| 80 |
-
|
|
|
|
| 81 |
|
| 82 |
-
|
| 83 |
-
|
| 84 |
-
resposta = agent.invoke(msg)
|
| 85 |
-
# Extrai texto se for dict
|
| 86 |
-
if isinstance(resposta, dict):
|
| 87 |
-
resposta = resposta.get('output', str(resposta))
|
| 88 |
-
elif not isinstance(resposta, str):
|
| 89 |
-
resposta = str(resposta)
|
| 90 |
-
except Exception as e:
|
| 91 |
-
resposta = f"Erro ao gerar resposta: {e}"
|
| 92 |
-
history = history + [(msg, resposta)]
|
| 93 |
-
return "", history
|
| 94 |
|
| 95 |
-
|
| 96 |
-
gr.Button("Enviar").click(interagir, [entrada, estado_chat], [entrada, chatbot])
|
| 97 |
|
| 98 |
-
demo.launch(server_name="0.0.0.0", server_port=7860, show_api=False)
|
| 99 |
|
| 100 |
|
|
|
|
| 3 |
from langchain_community.llms import HuggingFacePipeline
|
| 4 |
from langchain_community.embeddings import HuggingFaceEmbeddings
|
| 5 |
from langchain_community.vectorstores import FAISS
|
| 6 |
+
from langchain.document_loaders import PyPDFLoader
|
| 7 |
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
| 8 |
from langchain.chains import ConversationalRetrievalChain
|
| 9 |
from langchain.memory import ConversationBufferMemory
|
|
|
|
|
|
|
| 10 |
|
| 11 |
+
# Carrega e divide o documento PDF fixo (local)
|
| 12 |
+
def carregar_documento_pdf(path):
|
| 13 |
+
loader = PyPDFLoader(path)
|
| 14 |
+
documentos = loader.load()
|
| 15 |
+
splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=100)
|
| 16 |
+
documentos_divididos = splitter.split_documents(documentos)
|
| 17 |
+
embeddings = HuggingFaceEmbeddings()
|
| 18 |
+
vectorstore = FAISS.from_documents(documentos_divididos, embeddings)
|
| 19 |
+
return vectorstore
|
| 20 |
+
|
| 21 |
+
# Inicializa vectorstore com PDF fixo
|
| 22 |
+
vectorstore = carregar_documento_pdf("meu_documento_fixo.pdf")
|
| 23 |
+
|
| 24 |
+
# Inicializa pipeline do Flan-Alpaca
|
| 25 |
pipe = pipeline(
|
| 26 |
"text2text-generation",
|
| 27 |
model="declare-lab/flan-alpaca-base",
|
|
|
|
| 31 |
)
|
| 32 |
llm = HuggingFacePipeline(pipeline=pipe)
|
| 33 |
|
| 34 |
+
# Mem贸ria limitada para evitar prompts muito longos (k=3 煤ltimas mensagens)
|
| 35 |
memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True, k=3)
|
| 36 |
|
| 37 |
+
# Cadeia de Perguntas & Respostas com mem贸ria
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 38 |
chat_chain = ConversationalRetrievalChain.from_llm(
|
| 39 |
llm=llm,
|
| 40 |
retriever=vectorstore.as_retriever(),
|
| 41 |
memory=memory,
|
| 42 |
)
|
| 43 |
|
| 44 |
+
# Fun莽茫o para responder pergunta com contexto de mem贸ria
|
| 45 |
+
def responder_pergunta(pergunta):
|
| 46 |
+
if not pergunta or not pergunta.strip():
|
| 47 |
+
return "Por favor, insira uma pergunta v谩lida."
|
| 48 |
+
resposta = chat_chain.run({"question": pergunta})
|
| 49 |
+
return resposta
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 50 |
|
| 51 |
+
# Interface Gradio com chat e mem贸ria
|
| 52 |
with gr.Blocks() as demo:
|
| 53 |
+
gr.Markdown("## 馃 Chat com PDF fixo - Flan Alpaca + LangChain + FAISS + Mem贸ria")
|
| 54 |
+
chatbot = gr.Chatbot(label="Conversa com mem贸ria")
|
| 55 |
+
pergunta_input = gr.Textbox(placeholder="Digite sua pergunta aqui...", label="Pergunta")
|
| 56 |
+
enviar_btn = gr.Button("Enviar")
|
| 57 |
|
| 58 |
+
def interagir(pergunta, chat_hist):
|
| 59 |
+
resposta = responder_pergunta(pergunta)
|
| 60 |
+
chat_hist = chat_hist + [(pergunta, resposta)]
|
| 61 |
+
return "", chat_hist
|
| 62 |
|
| 63 |
+
enviar_btn.click(interagir, inputs=[pergunta_input, chatbot], outputs=[pergunta_input, chatbot])
|
| 64 |
+
pergunta_input.submit(interagir, inputs=[pergunta_input, chatbot], outputs=[pergunta_input, chatbot])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 65 |
|
| 66 |
+
demo.launch(server_name="0.0.0.0", server_port=7860)
|
|
|
|
| 67 |
|
|
|
|
| 68 |
|
| 69 |
|