Spaces:
Running
Running
binqiangliu
commited on
Commit
•
887bf36
1
Parent(s):
7f0efd4
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,122 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
from dotenv import load_dotenv
|
3 |
+
from PyPDF2 import PdfReader
|
4 |
+
from langchain.text_splitter import CharacterTextSplitter
|
5 |
+
from langchain.embeddings import OpenAIEmbeddings, HuggingFaceInstructEmbeddings
|
6 |
+
from langchain.embeddings import HuggingFaceEmbeddings, SentenceTransformerEmbeddings
|
7 |
+
from langchain import HuggingFaceHub
|
8 |
+
from langchain.vectorstores import FAISS
|
9 |
+
from langchain.memory import ConversationBufferMemory
|
10 |
+
from langchain.chains import ConversationalRetrievalChain
|
11 |
+
from langchain.chat_models import ChatOpenAI
|
12 |
+
from htmlTemplates import bot_template, user_template, css
|
13 |
+
from transformers import pipeline
|
14 |
+
import sys
|
15 |
+
import os
|
16 |
+
from dotenv import load_dotenv
|
17 |
+
|
18 |
+
HUGGINGFACEHUB_API_TOKEN = os.getenv("HUGGINGFACEHUB_API_TOKEN")
|
19 |
+
repo_id=os.getenv("repo_id")
|
20 |
+
|
21 |
+
OPENAI_API_KEY = os.environ.get('OPENAI_API_KEY')
|
22 |
+
openai_api_key = os.environ.get('openai_api_key')
|
23 |
+
embeddings = OpenAIEmbeddings(openai_api_key=OPENAI_API_KEY)
|
24 |
+
|
25 |
+
#*******************************************#Pinecone Account: b***liu@gmail.com
|
26 |
+
#pinecone_index_name=os.environ.get('pinecone_index_name')
|
27 |
+
#pinecone_namespace=os.environ.get('pinecone_namespace')
|
28 |
+
#pinecone_api_key=os.environ.get('pinecone_api_key')
|
29 |
+
#pinecone_environment=os.environ.get('pinecone_environment')
|
30 |
+
#pinecone.init(
|
31 |
+
# api_key=pinecone_api_key,
|
32 |
+
# environment=pinecone_environment
|
33 |
+
#)
|
34 |
+
#index = pinecone.Index(pinecone_index_name)
|
35 |
+
#loaded_v_db_500_wt_metadata = Pinecone.from_existing_index(index_name=pinecone_index_name, embedding=embeddings, namespace=pinecone_namespace)
|
36 |
+
#*******************************************#
|
37 |
+
|
38 |
+
#*******************************************#Pinecone Account: ij***.l**@hotmail.com
|
39 |
+
pinecone_index_name_1=os.environ.get('pinecone_index_name_1')
|
40 |
+
#pinecone_namespace_1=os.environ.get('pinecone_namespace_1') #no namespace under this Pinecone account
|
41 |
+
pinecone_api_key_1=os.environ.get('pinecone_api_key_1')
|
42 |
+
pinecone_environment_1=os.environ.get('pinecone_environment_1')
|
43 |
+
pinecone.init(
|
44 |
+
api_key=pinecone_api_key_1,
|
45 |
+
environment=pinecone_environment_1
|
46 |
+
)
|
47 |
+
index = pinecone.Index(pinecone_index_name_1)
|
48 |
+
#vectorstore = Pinecone.from_existing_index(index_name=pinecone_index_name_1, embedding=embeddings)
|
49 |
+
#*******************************************#
|
50 |
+
|
51 |
+
hf_token = os.environ.get('HUGGINGFACEHUB_API_TOKEN')
|
52 |
+
HUGGINGFACEHUB_API_TOKEN = os.environ.get('HUGGINGFACEHUB_API_TOKEN')
|
53 |
+
huggingfacehub_api_token= os.environ.get('huggingfacehub_api_token')
|
54 |
+
repo_id = os.environ.get('repo_id')
|
55 |
+
|
56 |
+
def get_vector_store():
|
57 |
+
#vectorstore = FAISS.from_texts(texts = text_chunks, embedding = embeddings)
|
58 |
+
vectorstore = Pinecone.from_existing_index(index_name=pinecone_index_name_1, embedding=embeddings)
|
59 |
+
return vectorstore
|
60 |
+
|
61 |
+
def get_conversation_chain(vector_store):
|
62 |
+
# OpenAI Model
|
63 |
+
#llm = ChatOpenAI()
|
64 |
+
#HuggingFace Model
|
65 |
+
#llm = HuggingFaceHub(repo_id="google/flan-t5-xxl")
|
66 |
+
#llm = HuggingFaceHub(repo_id="tiiuae/falcon-40b-instruct", model_kwargs={"temperature":0.5, "max_length":512}) #出现超时timed out错误
|
67 |
+
#llm = HuggingFaceHub(repo_id="meta-llama/Llama-2-70b-hf", model_kwargs={"min_length":100, "max_length":1024,"temperature":0.1})
|
68 |
+
#repo_id="HuggingFaceH4/starchat-beta"
|
69 |
+
llm = HuggingFaceHub(repo_id=repo_id,
|
70 |
+
model_kwargs={"min_length":1024,
|
71 |
+
"max_new_tokens":5632, "do_sample":True,
|
72 |
+
"temperature":0.1,
|
73 |
+
"top_k":50,
|
74 |
+
"top_p":0.95, "eos_token_id":49155})
|
75 |
+
memory = ConversationBufferMemory(memory_key='chat_history', return_messages=True)
|
76 |
+
conversation_chain = ConversationalRetrievalChain.from_llm(
|
77 |
+
llm = llm,
|
78 |
+
retriever = vector_store.as_retriever(),
|
79 |
+
memory = memory
|
80 |
+
)
|
81 |
+
print("***Start of printing Conversation_Chain***")
|
82 |
+
print(conversation_chain)
|
83 |
+
print("***End of printing Conversation_Chain***")
|
84 |
+
st.write("***Start of printing Conversation_Chain***")
|
85 |
+
st.write(conversation_chain)
|
86 |
+
st.write("***End of printing Conversation_Chain***")
|
87 |
+
return conversation_chain
|
88 |
+
|
89 |
+
def handle_user_input(question):
|
90 |
+
response = st.session_state.conversation({'question':question})
|
91 |
+
st.session_state.chat_history = response['chat_history']
|
92 |
+
for i, message in enumerate(st.session_state.chat_history):
|
93 |
+
if i % 2 == 0:
|
94 |
+
st.write(user_template.replace("{{MSG}}", message.content), unsafe_allow_html=True)
|
95 |
+
else:
|
96 |
+
st.write(bot_template.replace("{{MSG}}", message.content), unsafe_allow_html=True)
|
97 |
+
|
98 |
+
def main():
|
99 |
+
load_dotenv()
|
100 |
+
st.set_page_config(page_title='Chat with Your own PDFs', page_icon=':books:')
|
101 |
+
st.write(css, unsafe_allow_html=True)
|
102 |
+
if "conversation" not in st.session_state:
|
103 |
+
st.session_state.conversation = None
|
104 |
+
if "chat_history" not in st.session_state:
|
105 |
+
st.session_state.chat_history = None
|
106 |
+
st.header('Chat with Your own PDFs :books:')
|
107 |
+
question = st.text_input("Ask anything to your PDF: ")
|
108 |
+
if question:
|
109 |
+
handle_user_input(question)
|
110 |
+
with st.sidebar:
|
111 |
+
st.subheader("Upload your Documents Here: ")
|
112 |
+
pdf_files = st.file_uploader("Choose your PDF Files and Press OK", type=['pdf'], accept_multiple_files=True)
|
113 |
+
if st.button("OK"):
|
114 |
+
with st.spinner("Preparation under process..."):
|
115 |
+
# Create Vector Store
|
116 |
+
vector_store = get_vector_store()
|
117 |
+
st.write("DONE")
|
118 |
+
# Create conversation chain
|
119 |
+
st.session_state.conversation = get_conversation_chain(vector_store)
|
120 |
+
|
121 |
+
if __name__ == '__main__':
|
122 |
+
main()
|