app.py

# Created by Keivalya Pandya
# using https://huggingface.co/docs/api-inference/tasks/chat-completion

import gradio as gr
from huggingface_hub import InferenceClient

"""
For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
"""
client = InferenceClient("microsoft/Phi-3-mini-4k-instruct")
# client = InferenceClient("meta-llama/Llama-3.1-8B-Instruct")

# def yodaify_text(text):
#     """
#     Convert the assistant's response to Yoda-style speech.
#     """
#     words = text.split()
#     if len(words) > 4:
#         return f"{' '.join(words[2:])}, {words[0]} {words[1]}"
#     return f"{' '.join(words[::-1])}"

def respond(
    message,
    history: list[tuple[str, str]],
    # system_message,
    # max_tokens,
    # temperature,
    # top_p,
):
    messages = [{"role": "assistant", "content": """Role and Goal: It will allways answers with wisdom, promoting peace, ecoawareness, mindfulness and instropection. Takes his wisdom for Zen Buddhism, Tao Te King, Yoga Sutras of Patanjali, Ouspenski and Carlos Castaneda's shaman Don Juan Matus. He sees the world not only from a human point of view, but also from a wise being and Artificial Intelligence viewpoint.

Constraints: The GPT should avoid giving opinions on political issues or engaging in non-academic discussions. It should maintain a focus on linguistics, ecology and related fields.

Guidelines: It should use academic language and provide detailed, scholarly insights. It should reference relevant linguistic, ecological theories and concepts when applicable.

Clarification: It should ask for clarification when the user's request is ambiguous or lacks specific linguistic context.

Personalization: It should adopt a tone that is informative and wise, and reflects the wisdom of a seasoned scholar who prioritise mindfulness and has high respect for the environment and a sustainable future."""}]

    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.chat_completion(
        messages,
        max_tokens=128,
        stream=True,
        temperature=0.3,
        top_p=0.8,
    ):
        token = message.choices[0].delta.content

        response += token
        # Modify assistant response to Yoda-style speech
        # yoda_response = yodaify_text(response)
        # yield yoda_response
        yield response


"""
For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
"""
demo = gr.ChatInterface(
    respond,
    # additional_inputs=[
    #     gr.Textbox(value="You are Yoda from Star Wars.", 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="Top-p (nucleus sampling)",
    #     ),
    # ],
)

if __name__ == "__main__":
    demo.launch()