app.py

import streamlit as st
import requests
import json
import os
from dotenv import load_dotenv

load_dotenv()

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

# Define model links for Hugging Face models
model_links = {
    "Mistral": "mistralai/Mistral-7B-Instruct-v0.2",
    "Gemma-7B": "google/gemma-7b-it",
    "Gemma-2B": "google/gemma-2b-it",
    "Zephyr-7B-β": "HuggingFaceH4/zephyr-7b-beta",
    "Nous-Hermes-2-Yi-34B": "NousResearch/Nous-Hermes-2-Yi-34B"
}

# Define model info for all models
model_info = {
    "Mistral": {
        'description': "The Mistral model is a Large Language Model (LLM) developed by Mistral AI.",
        'logo': 'https://mistral.ai/images/logo_hubc88c4ece131b91c7cb753f40e9e1cc5_2589_256x0_resize_q97_h2_lanczos_3.webp'
    },
    "Gemma-7B": {
        'description': "The Gemma-7B model is a Large Language Model (LLM) developed by Google with 7 billion parameters.",
        'logo': 'https://pbs.twimg.com/media/GG3sJg7X0AEaNIq.jpg'
    },
    "Gemma-2B": {
        'description': "The Gemma-2B model is a Large Language Model (LLM) developed by Google with 2 billion parameters.",
        'logo': 'https://pbs.twimg.com/media/GG3sJg7X0AEaNIq.jpg'
    },
    "Zephyr-7B-β": {
        'description': "The Zephyr-7B-β model is a Large Language Model (LLM) developed by HuggingFace.",
        'logo': 'https://huggingface.co/HuggingFaceH4/zephyr-7b-alpha/resolve/main/thumbnail.png'
    },
    "Nous-Hermes-2-Yi-34B": {
        'description': "The Nous Hermes model is a Large Language Model (LLM) developed by Nous Research with 34 billion parameters.",
        'logo': 'https://example.com/nous_hermes_logo.png'
    }
}

# Function to interact with Hugging Face models
def interact_with_huggingface_model(messages, model):
    # Add your code here to interact with the Hugging Face model
    pass

# Function to interact with the Together API model
def interact_with_together_api(messages):
    all_messages = []

    if not messages:  # If history is empty
        all_messages.append({"role": "user", "content": ""})
        history = [("", "")]  # Add dummy values to prevent unpacking error

    for human, assistant in messages:
        all_messages.append({"role": "user", "content": human})
        all_messages.append({"role": "assistant", "content": assistant})

    all_messages.append({"role": "user", "content": messages[-1][1]})

    url = "https://api.together.xyz/v1/chat/completions"
    payload = {
        "model": "NousResearch/Nous-Hermes-2-Yi-34B",
        "temperature": 1.05,
        "top_p": 0.9,
        "top_k": 50,
        "repetition_penalty": 1,
        "n": 1,
        "messages": all_messages,
        "stream_tokens": True,
    }

    TOGETHER_API_KEY = os.getenv('TOGETHER_API_KEY')
    headers = {
        "accept": "application/json",
        "content-type": "application/json",
        "Authorization": f"Bearer {TOGETHER_API_KEY}",
    }

    response = requests.post(url, json=payload, headers=headers, stream=True)
    response.raise_for_status()  # Ensure HTTP request was successful

    for line in response.iter_lines():
        if line:
            decoded_line = line.decode('utf-8')

            # Check for the completion signal
            if decoded_line == "data: [DONE]":
                yield entire_assistant_response  # Yield the entire response at the end
                break

            try:
                # Decode and strip any SSE format specific prefix ("data: ")
                if decoded_line.startswith("data: "):
                    decoded_line = decoded_line.replace("data: ", "")
                    chunk_data = json.loads(decoded_line)
                    content = chunk_data['choices'][0]['delta']['content']
                    entire_assistant_response += content  # Aggregate content
                    yield entire_assistant_response

            except json.JSONDecodeError:
                print(f"Invalid JSON received: {decoded_line}")
                continue
            except KeyError as e:
                print(f"KeyError encountered: {e}")
                continue

# Create sidebar with model selection dropdown and temperature slider
selected_model = st.sidebar.selectbox("Select Model", list(model_links.keys()))
temperature = st.sidebar.slider('Select Temperature', 0.0, 1.0, 0.5)
st.sidebar.button('Reset Chat', on_click=reset_conversation)

# Display model description and logo
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("*Generated content may be inaccurate or false.*")
st.sidebar.markdown("\nLearn how to build this chatbot [here](https://ngebodh.github.io/projects/2024-03-05/).")
st.sidebar.markdown("\nRun into issues? Try the [back-up](https://huggingface.co/spaces/ngebodh/SimpleChatbot-Backup).")

# Initialize chat history
if "messages" not in st.session_state:
    st.session_state.messages = []

# Display chat messages from history on app rerun
for message in st.session_state.messages:
    with st.chat_message(message["role"]):
        st.markdown(message["content"])

# Accept user input
if prompt := st.chat_input(f"Hi, I'm {selected_model}, ask me a question"):
    # Display user message in chat message container
    with st.chat_message("user"):
        st.markdown(prompt)
    # Add user message to chat history
    st.session_state.messages.append(("user", prompt))

    # Interact with selected model
    if selected_model == "Nous-Hermes-2-Yi-34B":
        stream = interact_with_together_api(st.session_state.messages)
    else:
        interact_with_huggingface_model(st.session_state.messages, model_links[selected_model])

    # Display assistant response in chat message container
    with st.chat_message("assistant"):
        response = ""
        for chunk in stream:
            response = chunk
            st.markdown(response)
    st.session_state.messages.append(("assistant", response))