Abob1ch/LLAMA-3_8B_Unaligned

1#refer llama recipes for more info https://github.com/huggingface/huggingface-llama-recipes/blob/main/inference-api.ipynb2#huggingface-llama-recipes : https://github.com/huggingface/huggingface-llama-recipes/tree/main3import gradio as gr4from openai import OpenAI5import os6 7ACCESS_TOKEN = os.getenv("HF_TOKEN")8 9client = OpenAI(10    base_url="https://api-inference.huggingface.co/v1/",11    api_key=ACCESS_TOKEN,12)13 14def respond(15    message,16    history: list[tuple[str, str]],17    system_message,18    max_tokens,19    temperature,20    top_p,21):22    messages = [{"role": "system", "content": system_message}]23 24    for val in history:25        if val[0]:26            messages.append({"role": "user", "content": val[0]})27        if val[1]:28            messages.append({"role": "assistant", "content": val[1]})29 30    messages.append({"role": "user", "content": message})31 32    response = ""33    34    for message in  client.chat.completions.create(35        model="SicariusSicariiStuff/LLAMA-3_8B_Unaligned_BETA",36        max_tokens=max_tokens,37        stream=True,38        temperature=temperature,39        top_p=top_p,40        messages=messages,41    ):42        token = message.choices[0].delta.content43        44        response += token45        yield response46        47chatbot = gr.Chatbot(height=600)48 49demo = gr.ChatInterface(50    respond,51    additional_inputs=[52        gr.Textbox(value="", label="System message"),53        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),54        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),55        gr.Slider(56            minimum=0.1,57            maximum=1.0,58            value=0.95,59            step=0.05,60            label="Top-P",61        ),62        63    ],64    fill_height=True,65    chatbot=chatbot,66    theme="Nymbo/Alyx_Theme",67)68if __name__ == "__main__":69    demo.launch()