RAG / app.py
mahynski's picture
debug
7670068
raw
history blame
2.09 kB
import streamlit as st
from llama_index.llms.gemini import Gemini
from llama_index.llms.huggingface import HuggingFaceLLM
from llama_index.llms.mistralai import MistralAI
from llama_index.llms.openai import OpenAI
from streamlit_pdf_viewer import pdf_viewer
st.set_page_config(layout="wide")
with st.sidebar:
st.title('Document Summarization and QA System')
# st.markdown('''
# ## About this application
# Upload a pdf to ask questions about it. This retrieval-augmented generation (RAG) workflow uses:
# - [Streamlit](https://streamlit.io/)
# - [LlamaIndex](https://docs.llamaindex.ai/en/stable/)
# - [OpenAI](https://platform.openai.com/docs/models)
# ''')
# st.write('Made by ***Nate Mahynski***')
# st.write('nathan.mahynski@nist.gov')
# Select Provider
provider = st.selectbox(
label="Select LLM Provider",
options=['google', 'huggingface', 'mistralai', 'openai'],
index=0
)
# Select LLM
if provider == 'google':
llm_list = ['gemini']
elif provider == 'huggingface':
llm_list = []
elif provider == 'mistralai':
llm_list =[]
elif provider == 'openai':
llm_list = ['gpt-3.5-turbo', 'gpt-4', 'gpt-4-turbo', 'gpt-4o']
else:
llm_list = []
llm = st.selectbox(
label="Select LLM Model",
options=llm_list,
index=0
)
# Temperature
temperature = st.slider(
"Temperature",
min_value=0.0,
max_value=1.0,
value=0.0,
step=0.05,
)
# Enter Token
token = st.text_input(
"Enter your token",
value=None
)
uploaded_file = st.file_uploader(
"Choose a PDF file to upload",
type=['pdf'],
accept_multiple_files=False
)
if uploaded_file is not None:
# Parse the file
pass
col1, col2 = st.columns(2)
with col1:
pass
with col2:
if uploaded_file is not None:
# Display the pdf
bytes_data = uploaded_file.getvalue()
pdf_viewer(input=bytes_data, width=700)