Spaces:
Sleeping
Sleeping
brianjking
commited on
Commit
·
f9fdaeb
1
Parent(s):
57025a9
Create multiple.py
Browse files- multiple.py +83 -0
multiple.py
ADDED
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
import os
|
3 |
+
import tempfile
|
4 |
+
from llama_index import (
|
5 |
+
ServiceContext,
|
6 |
+
SimpleDirectoryReader,
|
7 |
+
VectorStoreIndex,
|
8 |
+
)
|
9 |
+
from llama_index.llms import OpenAI
|
10 |
+
import openai
|
11 |
+
|
12 |
+
st.title("Grounded Generation")
|
13 |
+
|
14 |
+
uploaded_files = st.file_uploader("Choose PDF files", type="pdf", accept_multiple_files=True)
|
15 |
+
|
16 |
+
@st.cache_resource(show_spinner=False)
|
17 |
+
def load_data(uploaded_files):
|
18 |
+
with st.spinner('Indexing documents...'):
|
19 |
+
temp_dir = tempfile.mkdtemp() # Create temporary directory
|
20 |
+
file_paths = [] # List to store paths of saved files
|
21 |
+
|
22 |
+
# Save the uploaded files temporarily
|
23 |
+
for i, uploaded_file in enumerate(uploaded_files):
|
24 |
+
temp_path = os.path.join(temp_dir, f"temp_{i}.pdf")
|
25 |
+
with open(temp_path, "wb") as f:
|
26 |
+
f.write(uploaded_file.read())
|
27 |
+
file_paths.append(temp_path)
|
28 |
+
|
29 |
+
# Read and index documents using SimpleDirectoryReader
|
30 |
+
reader = SimpleDirectoryReader(input_dir=temp_dir, recursive=False)
|
31 |
+
docs = reader.load_data()
|
32 |
+
service_context = ServiceContext.from_defaults(
|
33 |
+
llm=OpenAI(
|
34 |
+
model="gpt-3.5-turbo-16k",
|
35 |
+
temperature=0.1,
|
36 |
+
),
|
37 |
+
system_prompt="You are an AI assistant that uses context from PDFs to assist the user in generating text."
|
38 |
+
)
|
39 |
+
index = VectorStoreIndex.from_documents(docs, service_context=service_context)
|
40 |
+
|
41 |
+
# Clean up temporary files and directory
|
42 |
+
for file_path in file_paths:
|
43 |
+
os.remove(file_path)
|
44 |
+
os.rmdir(temp_dir)
|
45 |
+
|
46 |
+
return index
|
47 |
+
|
48 |
+
if uploaded_files:
|
49 |
+
index = load_data(uploaded_files)
|
50 |
+
|
51 |
+
user_query = st.text_input("Search for the products/info you want to use to ground your generated text content:")
|
52 |
+
|
53 |
+
if 'retrieved_text' not in st.session_state:
|
54 |
+
st.session_state['retrieved_text'] = ''
|
55 |
+
|
56 |
+
if st.button("Retrieve"):
|
57 |
+
with st.spinner('Retrieving text...'):
|
58 |
+
query_engine = index.as_query_engine(similarity_top_k=1)
|
59 |
+
st.session_state['retrieved_text'] = query_engine.query(user_query)
|
60 |
+
st.write(f"Retrieved Text: {st.session_state['retrieved_text']}")
|
61 |
+
|
62 |
+
content_type = st.selectbox("Select content type:", ["Blog", "Tweet"])
|
63 |
+
|
64 |
+
if st.button("Generate") and content_type:
|
65 |
+
with st.spinner('Generating text...'):
|
66 |
+
openai.api_key = os.getenv("OPENAI_API_KEY")
|
67 |
+
try:
|
68 |
+
if content_type == "Blog":
|
69 |
+
prompt = f"Write a blog about 500 words in length using the {st.session_state['retrieved_text']}"
|
70 |
+
elif content_type == "Tweet":
|
71 |
+
prompt = f"Compose a tweet using the {st.session_state['retrieved_text']}"
|
72 |
+
response = openai.ChatCompletion.create(
|
73 |
+
model="gpt-3.5-turbo-16k",
|
74 |
+
messages=[
|
75 |
+
{"role": "system", "content": "You are a helpful assistant."},
|
76 |
+
{"role": "user", "content": prompt}
|
77 |
+
]
|
78 |
+
)
|
79 |
+
generated_text = response['choices'][0]['message']['content']
|
80 |
+
st.write(f"Generated Text: {generated_text}")
|
81 |
+
except Exception as e:
|
82 |
+
st.write(f"An error occurred: {e}")
|
83 |
+
|