Download app.py from pk248/AboutMe: direct link, hf CLI and curl.
- Browser
- Download file 5.24 kB
-
https://huggingface.co/spaces/pk248/AboutMe/resolve/main/app.py
- Command line
-
hf download hf://spaces/pk248/AboutMe/app.py
-
curl -L -o app.py https://huggingface.co/spaces/pk248/AboutMe/resolve/main/app.py
5.24 kB
| import streamlit as st | |
| from langchain.text_splitter import RecursiveCharacterTextSplitter | |
| from langchain.docstore.document import Document | |
| from langchain.embeddings.openai import OpenAIEmbeddings | |
| from langchain.vectorstores.faiss import FAISS | |
| from langchain.llms import OpenAI | |
| from langchain.chains import RetrievalQA | |
| import os | |
| data = """ | |
| My name is Krishnasai Paleti, my friends call me Krishna. | |
| Im a masters student at Penn State University majoring in Computer Science and Engineering on track to graduate in May 2025. | |
| I specialize in Data Science, Machine Learning, Large Language models (LLMs) and Natural Language Processing(NLP). | |
| I believe in learning by doing and I work on personal projects whenever I get time. Outside acadamia, you can find me either cooking or playing badminton. | |
| My favorite cuisines are Italian and Indian. | |
| I love cooking with veggies and I try to add as many veggies as possible in the dishes I cook. | |
| Lets delve more into myself. | |
| My favorite color keeps changing based on my mood! I generally prefer darker shades of colors. | |
| My favorite movie is "The Intern" starring Robert De Niro and Anne Hathaway. | |
| I am the first generation of my family to study in the US. | |
| Fun fact: I always strive to improve, even in the smallest or least important tasks. This does drain me out at times, but my drive burns more brighter day by day! | |
| I am very detail-oriented, sometimes down to the first principles of the task I'm working on. | |
| I'm extremely loyal and expect the bare minimum of honesty from my close confidants. | |
| I am passionate about making a mark with my work. | |
| As a highly motivated individual, I am driven by my passion for continuous improvement and strive to implement the principle of KAIZEN in all aspects of my life. | |
| Penn State is my second masters degree. My first degree was an integrated masters (Bachelors and Masters in 5 years) from IIIT in Computer Science and Engineering with a Specialization in Systems Design. | |
| System Design never made me exclaim with joy as Data Science or ML did. I consider data science as a game to figure out the clues hidden in the data to solve the case. | |
| Hence I started working on Machine Learning from my third year, taking up internships whenever possible. | |
| This path eventually led to me pursuing my thesis on Large Language Models and Prompt based tuning. | |
| The caveat was that I started my thesis before ChatGPT was released and I witnessed the giant strides the world took in the field of LLMs. | |
| My thesis which focused on Prompt Based Finetuning in a Few Shot Setting aimed to reduce computation costs and training time. This approach is particularly effective in data scarce scenarios. | |
| """ | |
| # Step 2: Split the data into chunks | |
| text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=20) | |
| texts = text_splitter.split_text(data) | |
| documents = [Document(page_content=text) for text in texts] | |
| # Step 3: Create vector store using FAISS and OpenAI embeddings | |
| embeddings = OpenAIEmbeddings(openai_api_key=os.getenv("OPEN_API_KEY")) | |
| vectorstore = FAISS.from_documents(documents, embeddings) | |
| llm = OpenAI( | |
| openai_api_key=os.getenv("OPEN_API_KEY"), temperature=0.7 | |
| ) # between 0 to 1, more towards 1 is more creative answers. | |
| # Build the RAG chain (Retrieval + LLM) | |
| qa_chain = RetrievalQA.from_chain_type( | |
| llm=llm, | |
| chain_type="stuff", # Use "stuff" to limit LLM to only use retrieved context | |
| retriever=vectorstore.as_retriever(), | |
| return_source_documents=True, | |
| ) | |
| # Function to process query | |
| def process_query(query): | |
| # Run the RAG pipeline | |
| result = qa_chain(query) | |
| # Get the retrieved source documents | |
| source_documents = result["source_documents"] | |
| # If no relevant documents were retrieved, return "I don't know" | |
| if not source_documents: | |
| return "I'm sorry, I don't have information related to your query." | |
| # Otherwise, return the generated answer from the context | |
| return result["result"] | |
| st.title("Krishnasai Paleti's Landing Page [Prototype - This is my pet project that I work on when I have nothing else to do during my busy semester]") | |
| # Layout with picture on the top left and chatbot in the center | |
| col1, col2 = st.columns([1, 2]) | |
| with col1: | |
| st.image("https://huggingface.co/spaces/pk248/AboutMe/resolve/main/pookie_pic.jpg", use_column_width=True) | |
| with col2: | |
| st.header("Wanna Know about me?") | |
| query = st.text_input( | |
| "Enter your query:", placeholder="Ask a question about Krishna" | |
| ) | |
| if st.button("Get Answer"): | |
| if query: | |
| response = process_query(query) | |
| st.write(f"**Answer**: {response}") | |
| else: | |
| st.write("Please enter a query.") | |
| # Links Section | |
| st.header("Links") | |
| st.markdown( | |
| """ | |
| <div style="display: flex; justify-content: center; gap: 20px;"> | |
| <a href="https://pennstateoffice365-my.sharepoint.com/:b:/g/personal/kxp5619_psu_edu/EUyk0zU_ss5JmchIRMxup4ABfMwZRHFTN96TW0adJWuwkA?e=VJo0cp" target="_blank">Download My Resume</a> | |
| <a href="https://github.com/PaletiKrishnasai" target="_blank">GitHub Profile</a> | |
| <a href="https://www.linkedin.com/in/paleti-krishnasai/" target="_blank">LinkedIn Profile</a> | |
| </div> | |
| """, | |
| unsafe_allow_html=True, | |
| ) | |