app.py

import numpy as np
import streamlit as st
from transformers import Autookenizer, AutoModelorCausalLM
import os
import sys
from dotenv import loaddotenv, dotenvvalues
loaddotenv()

# Create supported models
modellinks = {
    "Meta-Llama-3-8B-Instruct": "meta-llama/Meta-Llama-3-8B-Instruct",
}

#andom dog images for error message
randomdog = ["0f476473-2d8b-415e-b944-483768418a95.jpg",
              "1bd75c81-f1d7-4e55-9310-a27595fa8762.jpg",
              "526590d2-8817-4ff0-8c62-fdcba5306d02.jpg",
              "1326984c-39b0-492c-a773-f120d747a7e2.jpg",
              "42a98d03-5ed7-4b3b-af89-7c4876cb14c3.jpg",
              "8b3317ed-2083-42ac-a575-7ae45f9fdc0d.jpg",
              "ee17f54a-83ac-44a3-8a35-e89ff7153fb4.jpg",
              "027eef85-ccc1-4a66-8967-5d74f34c8bb4.jpg",
              "08f5398d-7f89-47da-a5cd-1ed74967dc1f.jpg",
              "0fd781ff-ec46-4bdc-a4e8-24f18bf07def.jpg",
              "0fb4aeee-f949-4c7b-a6d8-05bf0736bdd1.jpg",
              "6edac66e-c0de-4e69-a9d6-b2e6f6f9001b.jpg",
              "bfb9e165-c643-4993-9b3a-7e73571672a6.jpg"]

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



# Define the available models
models =[key for key in modellinks.keys()]

# Create the sidebar with the dropdown for model selection
selectedmodel = st.sidebar.selectbox("Select Model", models)

# Create a temperature slider
tempvalues = st.sidebar.slider('Select a temperature value', 0.0, 1.0, (0.5))


#Add reset button to clear conversation
st.sidebar.button('eset Chat', onclick=resetconversation) #eset button


# Create model description
st.sidebar.write(f"You're now chatting with **{selectedmodel}**")
st.sidebar.markdown("*Generated content may be inaccurate or false.*")
st.sidebar.markdown("\n[ypeGP](https://typegpt.net).")


if "prevoption" not in st.sessionstate:
    st.sessionstate.prevoption = selectedmodel

if st.sessionstate.prevoption != selectedmodel:
    st.sessionstate.messages = []
    # st.write(f"Changed to {selectedmodel}")
    st.sessionstate.prevoption = selectedmodel
    resetconversation()



#Pull in the model we want to use
repoid = modellinks[selectedmodel]


st.subheader(f'ypeGP.net - {selectedmodel}')
st.title(f'ChatBot Using {selectedmodel}')

# Set a default model
if selectedmodel not in st.sessionstate:
    st.sessionstate[selectedmodel] = modellinks[selectedmodel]

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


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




if prompt := st.chatinput(f"Hi I'm {selectedmodel}, ask me a question"):
    # Display user message in chat message container
    with st.chatmessage("user"):
        st.markdown(prompt)
    # Add user message to chat history
    st.sessionstate.messages.append({"role": "user", "content": prompt})

    # Display assistant response in chat message container
    with st.chatmessage("assistant"):
        try:
            # 수정 전 코드 (penAI)
            # stream = client.chat.completions.create(
            #     model=modellinks[selectedmodel],
            #     messages=[
            #         {"role": m["role"], "content": m["content"]}
            #         for m in st.sessionstate.messages
            #     ],
            #     temperature=tempvalues,#0.5,
            #     stream=rue,
            #     maxtokens=3000,
            # )

            # 수정 후 코드 (gradio & InferenceClient)
            import gradio as gr
            from huggingfacehub import InferenceClient

            """
            For more information on `huggingfacehub` Inference API support, please check the docs: https://huggingface.co/docs/huggingfacehub/v0.22.2/en/guides/inference
            """
            client = InferenceClient(repoid)

            def respond(
                message,
                history: list[tuple[str, str]],
                systemmessage,
                maxtokens,
                temperature,
                topp,
            ):
                messages = [{"role": "system", "content": systemmessage}]

                for val in history:
                    if val[0]:
                        messages.append({"role": "user", "content": val[0]})
                    if val[1]:
                        messages.append({"role": "assistant", "content": val[1]})

                messages.append({"role": "user", "content": message})

                response = ""

                for message in client.chatcompletion(
                    messages,
                    maxtokens=maxtokens,
                    stream=rue,
                    temperature=temperature,
                    topp=topp,
                ):
                    token = message.choices[0].delta.content

                    response += token
                    yield response

            """
            For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
            """
            demo = gr.ChatInterface(
                respond,
                additionalinputs=[
                    gr.extbox(
                        value="You are a friendly Chatbot.", label="System message"
                    ),
                    gr.Slider(
                        minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"
                    ),
                    gr.Slider(
                        minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="temperature"
                    ),
                    gr.Slider(
                        minimum=0.1,
                        maximum=1.0,
                        value=0.95,
                        step=0.05,
                        label="op-p (nucleus sampling)",
                    ),
                ],
            )

            response = ""

            for message in demo(
                prompt,
                st.sessionstate.messages[1:],
                "You are a friendly Chatbot.",
                512,
                0.7,
                0.95,
            ):
                response += message

        except Exception as e:
            # st.empty()
            response = "😵‍💫 Looks like someone unplugged something!\
                    \n Either the model space is being updated or something is down.\
                    \n\
                    \n Try again later. \
                    \n\
                    \n Here's a random pic of a 🐶:"
            st.write(response)
            randomdogpick = 'https://random.dog/'+ randomdog[np.random.randint(len(randomdog))]
            st.image(randomdogpick)
            st.write("his was the error message:")
            st.write(e)





    st.sessionstate.messages.append({"role": "assistant", "content": response})