Spaces:
Running
Running
Commit
•
7eb9150
1
Parent(s):
7e5fea1
Create app.py
Browse filesadding first version
app.py
ADDED
@@ -0,0 +1,94 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
from langchain_community.llms import HuggingFaceEndpoint
|
3 |
+
from langchain_core.runnables import RunnablePassthrough
|
4 |
+
from langchain_core.output_parsers import StrOutputParser
|
5 |
+
from langchain.prompts import ChatPromptTemplate
|
6 |
+
from PyPDF2 import PdfReader
|
7 |
+
from langchain_text_splitters import RecursiveCharacterTextSplitter
|
8 |
+
import os
|
9 |
+
from langchain_community.vectorstores import Chroma
|
10 |
+
from langchain.chains.question_answering import load_qa_chain
|
11 |
+
from langchain.prompts import PromptTemplate
|
12 |
+
from langchain_community.document_loaders import PyPDFLoader
|
13 |
+
from langchain_chroma import Chroma
|
14 |
+
from langchain_community.embeddings import HuggingFaceEmbeddings
|
15 |
+
from transformers import pipeline
|
16 |
+
|
17 |
+
def get_pdf(pdf_docs):
|
18 |
+
docs=[]
|
19 |
+
for pdf in pdf_docs:
|
20 |
+
temp_file = "./temp.pdf"
|
21 |
+
# Delete the existing temp.pdf file if it exists
|
22 |
+
if os.path.exists(temp_file):
|
23 |
+
os.remove(temp_file)
|
24 |
+
with open(temp_file, "wb") as file:
|
25 |
+
file.write(pdf.getvalue())
|
26 |
+
file_name = pdf.name
|
27 |
+
loader = PyPDFLoader(temp_file)
|
28 |
+
docs.extend(loader.load())
|
29 |
+
return docs
|
30 |
+
|
31 |
+
def text_splitter(text):
|
32 |
+
text_splitter = RecursiveCharacterTextSplitter(
|
33 |
+
# Set a really small chunk size, just to show.
|
34 |
+
chunk_size=10000,
|
35 |
+
chunk_overlap=500,
|
36 |
+
separators=["\n\n","\n"," ",".",","])
|
37 |
+
chunks=text_splitter.split_documents(text)
|
38 |
+
return chunks
|
39 |
+
|
40 |
+
|
41 |
+
def get_conversational_chain(retriever):
|
42 |
+
prompt_template = """
|
43 |
+
Given the following extracted parts of a long document and a question, create a final answer.
|
44 |
+
Answer the question as detailed as possible from the provided context, make sure to provide all the details, if the answer is not in
|
45 |
+
provided context just say, "answer is not available in the context", and then ignore the context and add the answer from your knowledge like a simple llm prompt.
|
46 |
+
Try to give atleast the basic information.Do not return blank answer.\n\n
|
47 |
+
Make sure to understand the question and answer as per the question.
|
48 |
+
The answer should be a detailed one and try to incorporate examples for better understanding.
|
49 |
+
If the question involves terms like detailed or explained , give answer which involves complete detail about the question.\n\n
|
50 |
+
Context:\n {context}?\n
|
51 |
+
Question: \n{question}\n
|
52 |
+
Answer:
|
53 |
+
"""
|
54 |
+
pipeline("text-generation", model="nvidia/Llama3-ChatQA-1.5-8B")
|
55 |
+
|
56 |
+
pt = ChatPromptTemplate.from_template(prompt_template)
|
57 |
+
# Retrieve and generate using the relevant snippets of the blog.
|
58 |
+
#retriever = db.as_retriever()
|
59 |
+
rag_chain = (
|
60 |
+
{"context": retriever, "question": RunnablePassthrough()}
|
61 |
+
| pt
|
62 |
+
| llm
|
63 |
+
| StrOutputParser()
|
64 |
+
)
|
65 |
+
return rag_chain
|
66 |
+
|
67 |
+
def embedding(chunk,query):
|
68 |
+
embeddings=HuggingFaceEmbeddings()
|
69 |
+
db = Chroma.from_documents(chunk,embeddings)
|
70 |
+
doc = db.similarity_search(query)
|
71 |
+
chain = get_conversational_chain(db.as_retriever())
|
72 |
+
response = chain.invoke(query)
|
73 |
+
return response
|
74 |
+
|
75 |
+
if 'messages' not in st.session_state:
|
76 |
+
st.session_state.messages = [{'role': 'assistant', "content": 'Hello! Upload a PDF and ask me questions.'}]
|
77 |
+
|
78 |
+
|
79 |
+
st.header("Chat with your pdf")
|
80 |
+
with st.sidebar:
|
81 |
+
st.title("PDF FILE UPLOAD:")
|
82 |
+
pdf_docs = st.file_uploader("Upload your PDF File and Click on the Submit Button", accept_multiple_files=True, key="pdf_uploader")
|
83 |
+
|
84 |
+
query = st.chat_input("Ask a Question from the PDF File")
|
85 |
+
if query:
|
86 |
+
raw_text = get_pdf(pdf_docs)
|
87 |
+
text_chunks = text_splitter(raw_text)
|
88 |
+
st.session_state.messages.append({'role': 'user', "content": query})
|
89 |
+
response = embedding(text_chunks,query)
|
90 |
+
st.session_state.messages.append({'role': 'assistant', "content": response})
|
91 |
+
|
92 |
+
for message in st.session_state.messages:
|
93 |
+
with st.chat_message(message['role']):
|
94 |
+
st.write(message['content'])
|