CopyleftCultivars/Natural-Farming-Reflection-Chat

1import os2import gradio as gr3from huggingface_hub import InferenceClient4 5"""6For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference7"""8HUGGY_TOKEN = os.getenv("HUG_TOKEN")9client = InferenceClient("Solshine/reflection-llama-3.1-8B-Natural-Farmer-3-16bit", token=HUGGY_TOKEN)10 11 12def respond(13    message,14    history: list[tuple[str, str]],15    system_message,16    max_tokens,17    temperature,18    top_p,19):20    messages = [{"role": "system", "content": system_message}]21 22    for val in history:23        if val[0]:24            messages.append({"role": "user", "content": val[0]})25        if val[1]:26            messages.append({"role": "assistant", "content": val[1]})27 28    messages.append({"role": "user", "content": message})29 30    response = ""31 32    for message in client.chat_completion(33        messages,34        max_tokens=max_tokens,35        stream=True,36        temperature=temperature,37        top_p=top_p,38    ):39        token = message.choices[0].delta.content40 41        response += token42        yield response43 44"""45For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface46"""47demo = gr.ChatInterface(48    respond,49    additional_inputs=[50        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),51        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),52        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),53        gr.Slider(54            minimum=0.1,55            maximum=1.0,56            value=0.95,57            step=0.05,58            label="Top-p (nucleus sampling)",59        ),60    ],61)62 63 64if __name__ == "__main__":65    demo.launch()