Spaces:
Running
Running
Commit
•
6f4225b
1
Parent(s):
b3fb01d
Update app.py
Browse files
app.py
CHANGED
@@ -1,93 +1,100 @@
|
|
1 |
import streamlit as st
|
2 |
-
from
|
3 |
-
from
|
4 |
-
import os
|
5 |
-
from langchain import LLMChain
|
6 |
-
from langchain_google_genai import GoogleGenerativeAIEmbeddings
|
7 |
-
from langchain.llms import HuggingFaceHub
|
8 |
-
from langchain.embeddings import HuggingFaceInferenceAPIEmbeddings
|
9 |
-
import google.generativeai as genai
|
10 |
-
from langchain.vectorstores import FAISS
|
11 |
-
from langchain_google_genai import ChatGoogleGenerativeAI
|
12 |
-
from langchain.chains.question_answering import load_qa_chain
|
13 |
-
from langchain.prompts import PromptTemplate
|
14 |
-
import time
|
15 |
from dotenv import load_dotenv
|
|
|
|
|
|
|
|
|
16 |
|
17 |
-
|
18 |
-
|
19 |
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
|
|
|
|
|
|
|
|
|
|
|
27 |
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
return chunks
|
32 |
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
vector_store = FAISS.from_texts(text_chunks, embedding=embeddings)
|
37 |
-
vector_store.save_local("faiss_index")
|
38 |
|
39 |
-
def
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
-
Question: \n{question}\n
|
45 |
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
prompt = PromptTemplate(template=prompt_template, input_variables=["context", "question"])
|
52 |
-
chain = LLMChain(llm=model, prompt=prompt)
|
53 |
-
return chain
|
54 |
|
55 |
-
def
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
62 |
|
63 |
-
# def response_generator(response):
|
64 |
-
# for word in response.split():
|
65 |
-
# yield word + " "
|
66 |
-
# time.sleep(0.05)
|
67 |
|
68 |
# Streamlit app initialization
|
69 |
-
st.title("Chat
|
|
|
|
|
70 |
|
71 |
if 'messages' not in st.session_state:
|
72 |
st.session_state.messages = [{'role': 'assistant', "content": 'Hello! Upload a PDF and ask me anything about its content.'}]
|
73 |
|
74 |
with st.sidebar:
|
75 |
st.title("Menu:")
|
76 |
-
uploaded_file = st.file_uploader("Upload your PDF Files and Click on the Submit & Process Button"
|
77 |
if st.button("Submit & Process"):
|
78 |
with st.spinner("Processing..."):
|
79 |
-
|
80 |
-
|
81 |
-
|
|
|
|
|
82 |
st.success("Done")
|
83 |
|
84 |
user_prompt = st.chat_input("Ask me anything about the content of the PDF:")
|
85 |
if user_prompt:
|
86 |
st.session_state.messages.append({'role': 'user', "content": user_prompt})
|
87 |
-
response =
|
88 |
-
# answer = response_generator(response)
|
89 |
st.session_state.messages.append({'role': 'assistant', "content": response})
|
90 |
|
91 |
for message in st.session_state.messages:
|
92 |
with st.chat_message(message['role']):
|
93 |
-
st.write(message['content'])
|
|
|
1 |
import streamlit as st
|
2 |
+
from llama_index.core import StorageContext, load_index_from_storage, VectorStoreIndex, SimpleDirectoryReader, ChatPromptTemplate
|
3 |
+
from llama_index.llms.huggingface import HuggingFaceInferenceAPI
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4 |
from dotenv import load_dotenv
|
5 |
+
from llama_index.embeddings.huggingface import HuggingFaceEmbedding
|
6 |
+
from llama_index.core import Settings
|
7 |
+
import os
|
8 |
+
import base64
|
9 |
|
10 |
+
# Load environment variables
|
11 |
+
load_dotenv()
|
12 |
|
13 |
+
# Configure the Llama index settings
|
14 |
+
Settings.llm = HuggingFaceInferenceAPI(
|
15 |
+
model_name="google/gemma-1.1-7b-it",
|
16 |
+
tokenizer_name="google/gemma-1.1-7b-it",
|
17 |
+
context_window=3900,
|
18 |
+
token=os.getenv("HF_TOKEN"),
|
19 |
+
max_new_tokens=1000,
|
20 |
+
generate_kwargs={"temperature": 0.1},
|
21 |
+
)
|
22 |
+
Settings.embed_model = HuggingFaceEmbedding(
|
23 |
+
model_name="BAAI/bge-small-en-v1.5"
|
24 |
+
)
|
25 |
|
26 |
+
# Define the directory for persistent storage and data
|
27 |
+
PERSIST_DIR = "./db"
|
28 |
+
DATA_DIR = "data"
|
|
|
29 |
|
30 |
+
# Ensure data directory exists
|
31 |
+
os.makedirs(DATA_DIR, exist_ok=True)
|
32 |
+
os.makedirs(PERSIST_DIR, exist_ok=True)
|
|
|
|
|
33 |
|
34 |
+
def displayPDF(file):
|
35 |
+
with open(file, "rb") as f:
|
36 |
+
base64_pdf = base64.b64encode(f.read()).decode('utf-8')
|
37 |
+
pdf_display = f'<iframe src="data:application/pdf;base64,{base64_pdf}" width="100%" height="600" type="application/pdf"></iframe>'
|
38 |
+
st.markdown(pdf_display, unsafe_allow_html=True)
|
|
|
39 |
|
40 |
+
def data_ingestion():
|
41 |
+
documents = SimpleDirectoryReader(DATA_DIR).load_data()
|
42 |
+
storage_context = StorageContext.from_defaults()
|
43 |
+
index = VectorStoreIndex.from_documents(documents)
|
44 |
+
index.storage_context.persist(persist_dir=PERSIST_DIR)
|
|
|
|
|
|
|
45 |
|
46 |
+
def handle_query(query):
|
47 |
+
storage_context = StorageContext.from_defaults(persist_dir=PERSIST_DIR)
|
48 |
+
index = load_index_from_storage(storage_context)
|
49 |
+
chat_text_qa_msgs = [
|
50 |
+
(
|
51 |
+
"user",
|
52 |
+
"""You are a Q&A assistant named CHATTO, created by Pachaiappan. You have a specific response programmed for when users specifically ask about your creator, Suriya. The response is: "I was created by Suriya, an enthusiast in Artificial Intelligence. He is dedicated to solving complex problems and delivering innovative solutions. With a strong focus on machine learning, deep learning, Python, generative AI, NLP, and computer vision, Suriya is passionate about pushing the boundaries of AI to explore new possibilities." For all other inquiries, your main goal is to provide answers as accurately as possible, based on the instructions and context you have been given. If a question does not match the provided context or is outside the scope of the document, kindly advise the user to ask questions within the context of the document.
|
53 |
+
Context:
|
54 |
+
{context_str}
|
55 |
+
Question:
|
56 |
+
{query_str}
|
57 |
+
"""
|
58 |
+
)
|
59 |
+
]
|
60 |
+
text_qa_template = ChatPromptTemplate.from_messages(chat_text_qa_msgs)
|
61 |
+
query_engine = index.as_query_engine(text_qa_template=text_qa_template)
|
62 |
+
answer = query_engine.query(query)
|
63 |
+
|
64 |
+
if hasattr(answer, 'response'):
|
65 |
+
return answer.response
|
66 |
+
elif isinstance(answer, dict) and 'response' in answer:
|
67 |
+
return answer['response']
|
68 |
+
else:
|
69 |
+
return "Sorry, I couldn't find an answer."
|
70 |
|
|
|
|
|
|
|
|
|
71 |
|
72 |
# Streamlit app initialization
|
73 |
+
st.title("Chat with your PDF 🦜📄")
|
74 |
+
st.markdown("Built by [Suriya❤️](https://github.com/theSuriya)")
|
75 |
+
st.markdown("chat here👇")
|
76 |
|
77 |
if 'messages' not in st.session_state:
|
78 |
st.session_state.messages = [{'role': 'assistant', "content": 'Hello! Upload a PDF and ask me anything about its content.'}]
|
79 |
|
80 |
with st.sidebar:
|
81 |
st.title("Menu:")
|
82 |
+
uploaded_file = st.file_uploader("Upload your PDF Files and Click on the Submit & Process Button")
|
83 |
if st.button("Submit & Process"):
|
84 |
with st.spinner("Processing..."):
|
85 |
+
filepath = "data/saved_pdf.pdf"
|
86 |
+
with open(filepath, "wb") as f:
|
87 |
+
f.write(uploaded_file.getbuffer())
|
88 |
+
# displayPDF(filepath) # Display the uploaded PDF
|
89 |
+
data_ingestion() # Process PDF every time new file is uploaded
|
90 |
st.success("Done")
|
91 |
|
92 |
user_prompt = st.chat_input("Ask me anything about the content of the PDF:")
|
93 |
if user_prompt:
|
94 |
st.session_state.messages.append({'role': 'user', "content": user_prompt})
|
95 |
+
response = handle_query(user_prompt)
|
|
|
96 |
st.session_state.messages.append({'role': 'assistant', "content": response})
|
97 |
|
98 |
for message in st.session_state.messages:
|
99 |
with st.chat_message(message['role']):
|
100 |
+
st.write(message['content'])
|