Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,109 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
import os
|
3 |
+
from streamlit_chat import message
|
4 |
+
import tempfile
|
5 |
+
#from langchain_community.documentloader.csv_loader import CSVLoader
|
6 |
+
from langchain_community.document_loaders.csv_loader import CSVLoader
|
7 |
+
from langchain_community.embeddings import HuggingFaceEmbeddings
|
8 |
+
#from langchain_community.embeddings import HuggingFaceBgeEmbeddings
|
9 |
+
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
10 |
+
from langchain_community.vectorstores import FAISS
|
11 |
+
#from langchain_community.llms import CTransformers
|
12 |
+
from langchain_community.llms.ctransformers import CTransformers
|
13 |
+
|
14 |
+
from langchain.chains.conversational_retrieval.base import ConversationalRetrievalChain
|
15 |
+
|
16 |
+
#from langchain.chains.conversational_retrieval.base import ConversationalRetreievalChain
|
17 |
+
|
18 |
+
|
19 |
+
DB_FAISS_PATH = 'vectorstore/db_faiss'
|
20 |
+
TEMP_DIR = 'temp'
|
21 |
+
if not os.path.exists(TEMP_DIR):
|
22 |
+
os.makedirs(TEMP_DIR)
|
23 |
+
|
24 |
+
def load_llm():
|
25 |
+
# load model from hugging face repo
|
26 |
+
llm = CTransformers(
|
27 |
+
model = 'TheBloke/Llama-2-7B-Chat-GGML',
|
28 |
+
model_type = 'llama',
|
29 |
+
max_new_token = 512,
|
30 |
+
temperature = 0.5
|
31 |
+
)
|
32 |
+
return llm
|
33 |
+
|
34 |
+
st.title("Chat with CSV using Llma 2")
|
35 |
+
st.markdown("<h1 style='text-align: center; color: blue;'>Chat with your PDF 📄 </h1>", unsafe_allow_html=True)
|
36 |
+
st.markdown("<h3 style='text-align: center; color: grey;'>Built by <a href='https://github.com/DrKareemKAmal'>MindSparks ❤️ </a></h3>", unsafe_allow_html=True)
|
37 |
+
|
38 |
+
uploaded_file = st.sidebar.file_uploader('Upload your data', type=['csv'])
|
39 |
+
|
40 |
+
if uploaded_file:
|
41 |
+
# with tempfile.NamedTemporaryFile(delete=False)as temp_file :
|
42 |
+
# temp_file.write(uploaded_file.getvalue())
|
43 |
+
# tempfile_path = temp_file.name
|
44 |
+
file_path = os.path.join(TEMP_DIR, uploaded_file.name)
|
45 |
+
with open(file_path, "wb") as f:
|
46 |
+
f.write(uploaded_file.getvalue())
|
47 |
+
|
48 |
+
st.write(f"Uploaded file: {uploaded_file.name}")
|
49 |
+
st.write("Processing CSV file...")
|
50 |
+
|
51 |
+
loader = CSVLoader(file_path = file_path, encoding = 'utf-8',
|
52 |
+
csv_args = {'delimiter': ','} )
|
53 |
+
data = loader.load()
|
54 |
+
#st.json(data)
|
55 |
+
|
56 |
+
text_splitter = RecursiveCharacterTextSplitter(chunk_size = 500 , chunk_overlap = 20)
|
57 |
+
text_chunks = text_splitter.split_documents(data)
|
58 |
+
st.write (f"Total text chunks : {len(text_chunks)}")
|
59 |
+
|
60 |
+
embeddings = HuggingFaceEmbeddings(
|
61 |
+
model_name = 'sentence-transformers/all-MiniLM-L6-v2',
|
62 |
+
# model_kwargs = {'device': 'cpu'}
|
63 |
+
)
|
64 |
+
|
65 |
+
|
66 |
+
db = FAISS.from_documents(text_chunks, embeddings)
|
67 |
+
db.save_local (DB_FAISS_PATH)
|
68 |
+
llm = load_llm()
|
69 |
+
|
70 |
+
chain = ConversationalRetrievalChain.from_llm(llm= llm , retriever = db.as_retriever())
|
71 |
+
|
72 |
+
def conversational_chat(query):
|
73 |
+
result = chain({"question": query ,
|
74 |
+
"chat_history": st.session_state['history']})
|
75 |
+
st.session_state['history'].append((query , result['answer']))
|
76 |
+
return result['answer']
|
77 |
+
|
78 |
+
if 'history' not in st.session_state :
|
79 |
+
st.session_state['history'] = []
|
80 |
+
|
81 |
+
if 'generated' not in st.session_state :
|
82 |
+
st.session_state['generated'] = ['Hello, Ask me anything about ' + uploaded_file.name]
|
83 |
+
|
84 |
+
if 'past' not in st.session_state :
|
85 |
+
st.session_state['past'] = ['Hey !']
|
86 |
+
|
87 |
+
# Container for the chat history
|
88 |
+
response_container = st.container()
|
89 |
+
container = st.container()
|
90 |
+
|
91 |
+
with container :
|
92 |
+
with st.form(key = 'my_form',
|
93 |
+
clear_on_submit=True):
|
94 |
+
user_input = st.text_input('Query:', placeholder= "Talk to youur CSV Data here ")
|
95 |
+
submit_button = st.form_submit_button(label = 'chat')
|
96 |
+
|
97 |
+
if submit_button and user_input :
|
98 |
+
output = conversational_chat(user_input)
|
99 |
+
|
100 |
+
st.session_state['past'].append(user_input)
|
101 |
+
st.session_state['generated'].append(output)
|
102 |
+
|
103 |
+
if st.session_state['generated'] :
|
104 |
+
with response_container:
|
105 |
+
for i in range(len(st.session_state['generated'])):
|
106 |
+
message(st.session_state['past'][i], is_user = True , key=str(i) + '_user',
|
107 |
+
avatar_style='big-smile')
|
108 |
+
message(st.session_state['generated'][i], key = str(i), avatar_style='thumb')
|
109 |
+
|