Spaces:
Runtime error
Runtime error
Shubhamtribhuwan17
commited on
Commit
β’
b684da5
1
Parent(s):
0ed9814
Update app.py
Browse files
app.py
CHANGED
@@ -1,3 +1,121 @@
|
|
1 |
-
import gradio as gr
|
2 |
|
3 |
-
gr.load("models/mistralai/Mistral-7B-v0.1").launch()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# import gradio as gr
|
2 |
|
3 |
+
# gr.load("models/mistralai/Mistral-7B-v0.1").launch()
|
4 |
+
|
5 |
+
import os
|
6 |
+
import streamlit as st
|
7 |
+
from dotenv import load_dotenv
|
8 |
+
from PyPDF2 import PdfReader
|
9 |
+
from langchain.text_splitter import CharacterTextSplitter
|
10 |
+
from langchain.embeddings import HuggingFaceBgeEmbeddings
|
11 |
+
from langchain.vectorstores import FAISS
|
12 |
+
from langchain.chat_models import ChatOpenAI
|
13 |
+
from langchain.memory import ConversationBufferMemory
|
14 |
+
from langchain.chains import ConversationalRetrievalChain
|
15 |
+
from htmlTemplates import css, bot_template, user_template
|
16 |
+
from langchain.llms import HuggingFaceHub
|
17 |
+
|
18 |
+
# set this key as an environment variable
|
19 |
+
os.environ["HUGGINGFACEHUB_API_TOKEN"] = st.secrets['huggingface_token']
|
20 |
+
|
21 |
+
def get_pdf_text(pdf_docs : list) -> str:
|
22 |
+
text = ""
|
23 |
+
for pdf in pdf_docs:
|
24 |
+
pdf_reader = PdfReader(pdf)
|
25 |
+
for page in pdf_reader.pages:
|
26 |
+
text += page.extract_text()
|
27 |
+
return text
|
28 |
+
|
29 |
+
|
30 |
+
def get_text_chunks(text:str) ->list:
|
31 |
+
text_splitter = CharacterTextSplitter(
|
32 |
+
separator="\n", chunk_size=1500, chunk_overlap=300, length_function=len
|
33 |
+
)
|
34 |
+
chunks = text_splitter.split_text(text)
|
35 |
+
return chunks
|
36 |
+
|
37 |
+
|
38 |
+
def get_vectorstore(text_chunks : list) -> FAISS:
|
39 |
+
model = "sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2"
|
40 |
+
encode_kwargs = {
|
41 |
+
"normalize_embeddings": True
|
42 |
+
} # set True to compute cosine similarity
|
43 |
+
embeddings = HuggingFaceBgeEmbeddings(
|
44 |
+
model_name=model, encode_kwargs=encode_kwargs, model_kwargs={"device": "cpu"}
|
45 |
+
)
|
46 |
+
vectorstore = FAISS.from_texts(texts=text_chunks, embedding=embeddings)
|
47 |
+
return vectorstore
|
48 |
+
|
49 |
+
|
50 |
+
def get_conversation_chain(vectorstore:FAISS) -> ConversationalRetrievalChain:
|
51 |
+
# llm = ChatOpenAI(temperature=0, model="gpt-3.5-turbo-0613")
|
52 |
+
llm = HuggingFaceHub(
|
53 |
+
repo_id="mistralai/Mixtral-8x7B-Instruct-v0.1",
|
54 |
+
#repo_id="TheBloke/Mixtral-8x7B-Instruct-v0.1-GGUF"
|
55 |
+
model_kwargs={"temperature": 0.5, "max_length": 1048},
|
56 |
+
)
|
57 |
+
|
58 |
+
memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)
|
59 |
+
conversation_chain = ConversationalRetrievalChain.from_llm(
|
60 |
+
llm=llm, retriever=vectorstore.as_retriever(), memory=memory
|
61 |
+
)
|
62 |
+
return conversation_chain
|
63 |
+
|
64 |
+
|
65 |
+
def handle_userinput(user_question:str):
|
66 |
+
response = st.session_state.conversation({"question": user_question})
|
67 |
+
st.session_state.chat_history = response["chat_history"]
|
68 |
+
|
69 |
+
for i, message in enumerate(st.session_state.chat_history):
|
70 |
+
if i % 2 == 0:
|
71 |
+
st.write(" Usuario: " + message.content)
|
72 |
+
else:
|
73 |
+
st.write("π€ ChatBot: " + message.content)
|
74 |
+
|
75 |
+
|
76 |
+
def main():
|
77 |
+
st.set_page_config(
|
78 |
+
page_title="Chat with a Bot that tries to answer questions about multiple PDFs",
|
79 |
+
page_icon=":books:",
|
80 |
+
)
|
81 |
+
|
82 |
+
st.markdown("# Chat with a Bot")
|
83 |
+
st.markdown("This bot tries to answer questions about multiple PDFs. Let the processing of the PDF finish before adding your question. ππΎ")
|
84 |
+
|
85 |
+
st.write(css, unsafe_allow_html=True)
|
86 |
+
|
87 |
+
|
88 |
+
if "conversation" not in st.session_state:
|
89 |
+
st.session_state.conversation = None
|
90 |
+
if "chat_history" not in st.session_state:
|
91 |
+
st.session_state.chat_history = None
|
92 |
+
|
93 |
+
|
94 |
+
st.header("Chat with a Bot π€π¦Ύ that tries to answer questions about multiple PDFs :books:")
|
95 |
+
user_question = st.text_input("Ask a question about your documents:")
|
96 |
+
if user_question:
|
97 |
+
handle_userinput(user_question)
|
98 |
+
|
99 |
+
|
100 |
+
with st.sidebar:
|
101 |
+
st.subheader("Your documents")
|
102 |
+
pdf_docs = st.file_uploader(
|
103 |
+
"Upload your PDFs here and click on 'Process'", accept_multiple_files=True
|
104 |
+
)
|
105 |
+
if st.button("Process"):
|
106 |
+
with st.spinner("Processing"):
|
107 |
+
# get pdf text
|
108 |
+
raw_text = get_pdf_text(pdf_docs)
|
109 |
+
|
110 |
+
# get the text chunks
|
111 |
+
text_chunks = get_text_chunks(raw_text)
|
112 |
+
|
113 |
+
# create vector store
|
114 |
+
vectorstore = get_vectorstore(text_chunks)
|
115 |
+
|
116 |
+
# create conversation chain
|
117 |
+
st.session_state.conversation = get_conversation_chain(vectorstore)
|
118 |
+
|
119 |
+
|
120 |
+
if __name__ == "__main__":
|
121 |
+
main()
|