app.py

import streamlit as st
from huggingface_hub import InferenceClient
import os
import pickle

st.title("CODEFUSSION ☄")

base_url = "https://api-inference.huggingface.co/models/"
API_KEY = os.environ.get('HUGGINGFACE_API_KEY')

model_links = {
    "LegacyLift🚀": base_url + "mistralai/Mistral-7B-Instruct-v0.2",
    "ModernMigrate⭐": base_url + "mistralai/Mixtral-8x7B-Instruct-v0.1",
    "RetroRecode🔄": base_url + "microsoft/Phi-3-mini-4k-instruct"
}

model_info = {
    "LegacyLift🚀": {
        'description': """The LegacyLift model is a **Large Language Model (LLM)** that's able to have question and answer interactions.\n \n\nThis model is best for minimal problem-solving, content writing, and daily tips.\n""",
        'logo': './11.jpg'
    },
    "ModernMigrate⭐": {
        'description': """The ModernMigrate model is a **Large Language Model (LLM)** that's able to have question and answer interactions.\n \n\nThis model excels in coding, logical reasoning, and high-speed inference. \n""",
        'logo': './2.jpg'
    },
    "RetroRecode🔄": {
        'description': """The RetroRecode model is a **Large Language Model (LLM)** that's able to have question and answer interactions.\n \n\nThis model is best suited for critical development, practical knowledge, and serverless inference.\n""",
        'logo': './3.jpg'
    },
}

def format_prompt(message, conversation_history, custom_instructions=None):
    prompt = ""
    if custom_instructions:
        prompt += f"\[INST\] {custom_instructions} \[/INST\]"
    
    # Add conversation history to the prompt
    prompt += "\[CONV_HISTORY\]\n"
    for role, content in conversation_history:
        prompt += f"{role.upper()}: {content}\n"
    prompt += "\[/CONV_HISTORY\]"
    
    # Add the current message
    prompt += f"\[INST\] {message} \[/INST\]"
    
    return prompt

def reset_conversation():
    '''
    Resets Conversation
    '''
    st.session_state.conversation = []
    st.session_state.messages = []
    save_conversation_history([])
    return None

def load_conversation_history():
    history_file = "conversation_history.pickle"
    if os.path.exists(history_file):
        with open(history_file, "rb") as f:
            conversation_history = pickle.load(f)
    else:
        conversation_history = []
    return conversation_history

def save_conversation_history(conversation_history):
    history_file = "conversation_history.pickle"
    with open(history_file, "wb") as f:
        pickle.dump(conversation_history, f)

models = [key for key in model_links.keys()]
selected_model = st.sidebar.selectbox("Select Model", models)
temp_values = st.sidebar.slider('Select a temperature value', 0.0, 1.0, (0.5))
st.sidebar.button('Reset Chat', on_click=reset_conversation)  # Reset button

st.sidebar.write(f"You're now chatting with **{selected_model}**")
st.sidebar.markdown(model_info[selected_model]['description'])
st.sidebar.image(model_info[selected_model]['logo'])

st.sidebar.markdown("\*Generating the code might go slow if you are using low power resources \*")

if "prev_option" not in st.session_state:
    st.session_state.prev_option = selected_model

if st.session_state.prev_option != selected_model:
    st.session_state.messages = []
    st.session_state.prev_option = selected_model

repo_id = model_links[selected_model]
st.subheader(f'{selected_model}')

# Load the conversation history from the file
st.session_state.messages = load_conversation_history()

for message in st.session_state.messages:
    with st.chat_message(message["role"]):
        st.markdown(message["content"])

if prompt := st.chat_input(f"Hi I'm {selected_model}, How can I help you today?"):
    custom_instruction = "Act like a Human in conversation"
    with st.chat_message("user"):
        st.markdown(prompt)
    
    st.session_state.messages.append({"role": "user", "content": prompt})
    conversation_history = [(message["role"], message["content"]) for message in st.session_state.messages]
    
    formatted_text = format_prompt(prompt, conversation_history, custom_instruction)
    
    max_new_tokens = 3000
    if selected_model != "RetroRecode🔄":
        input_tokens = len(formatted_text.split())
        max_tokens = {"LegacyLift🚀": 32000, "ModernMigrate⭐": 8192}
        max_new_tokens = max_tokens[selected_model] - input_tokens

    with st.chat_message("assistant"):
        client = InferenceClient(model=model_links[selected_model])
        output = client.text_generation(
            formatted_text,
            temperature=temp_values,
            max_new_tokens=max_new_tokens,
            stream=True
        )
        response = st.write_stream(output)
        st.session_state.messages.append({"role": "assistant", "content": response})
        
        # Save the updated conversation history to the file
        save_conversation_history(st.session_state.messages)