app.py

import streamlit as st
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

device = "cuda" if torch.cuda.is_available() else "cpu"

st.header("🤗 Instruction Tuned SmolLM 360M")


model_path = "Sharathhebbar24/smollm_sft_360M_instruct_tuned_v2"
model = AutoModelForCausalLM.from_pretrained(model_path).to(device)
tokenizer = AutoTokenizer.from_pretrained(model_path)

if "messages" not in st.session_state:
    st.session_state.messages = []

for message in st.session_state.messages:
    if message["role"] != "system":
        with st.chat_message(message['role']):
            if message['role'] == "assistant":
                st.json(message['content'])
            else:
                st.markdown(message["content"])



if user_input := st.chat_input("Your answer.", max_chars=1000):

    st.session_state.messages.append({
        "role": "user",
        "content": user_input
    })
    with st.chat_message("user"):
        st.markdown(user_input)

    with st.chat_message("assistant"):
        
        prompt = f'''### Instruction:\nExtract action, date, time, attendees, location, duration, recurrence, and notes from the dataset.\n\n### Input: \n{user_input}\n\n### Response:'''
    
        inputs = tokenizer(prompt, return_tensors="pt", padding=True).to(device)
        with torch.no_grad():
            outputs = model.generate(
                input_ids=inputs["input_ids"],
                attention_mask=inputs["attention_mask"],
                max_new_tokens=100,
                do_sample=False,
                eos_token_id=tokenizer.eos_token_id,
                pad_token_id=tokenizer.pad_token_id,
            )
        decoded_output = tokenizer.decode(outputs[0])
        generated_response = decoded_output.split("### Response:")[-1].strip()
        generated_response = generated_response[:generated_response.find("}") + 1]
        generated_response = generated_response.replace("None", "null")
        generated_response = generated_response.replace("'", '"')
    
        st.json(generated_response)
        st.session_state.messages.append({
            "role": "assistant",
            "content": generated_response
        })