app.py

import gradio as gr
from huggingface_hub import InferenceClient
from datasets import load_dataset
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

"""
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
"""

#Update: Using a new base model
client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
dataset = load_dataset("JustKiddo/KiddosVault")

# Load the tokenizer and model for token display
tokenizer = AutoTokenizer.from_pretrained("google-t5/t5-small") #Google's T5 Model
model = AutoModelForSeq2SeqLM.from_pretrained("google-t5/t5-small")

def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    messages = [{"role": "system", "content": system_message}]

    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=max_tokens,
        stream=True,
        temperature=temperature,
        top_p=top_p,
    ):
        token = message.choices[0].delta.content

        response += token
        yield response

#My custom token generator
def generate_tokens(text):
    input = tokenizer(text, return_tensors="pt")
    output = model.generate(**input)

    input_ids = input["input_ids"].tolist()[0]
    output_ids = output.tolist()[0]

    formatted_output = [format(x, 'd') for x in output_ids]

    input_tokens_str = tokenizer.convert_ids_to_tokens(input_ids)
    #output_tokens_str = tokenizer.convert_tokens_to_ids(output_ids)

    return " ".join(input_tokens_str), " ".join(formatted_output)

"""
For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
"""

#chatInterface = gr.ChatInterface(
#    respond,
#    additional_inputs=[
#        gr.Textbox(value="You are a professional Mental Healthcare Chatbot.", label="System message"),
#        gr.Slider(minimum=1, maximum=6144, value=6144, step=1, label="Max new tokens"),
#        gr.Slider(minimum=0.1, maximum=4.0, value=1, step=0.1, label="Temperature"),
#        gr.Slider(
#            minimum=0.1,
#            maximum=1.0,
#            value=0.95,
#            step=0.05,
#            label="Top-p (nucleus sampling)",
#        ),
#    ],
#)

with gr.Blocks() as demo:
    with gr.Column():
        gr.ChatInterface(
            respond,
            additional_inputs=[
                gr.Textbox(value="You are a professional Mental Healthcare Chatbot.", label="System message"),
                gr.Slider(minimum=1, maximum=6144, value=6144, step=1, label="Max new tokens"),
                gr.Slider(minimum=0.1, maximum=4.0, value=1, step=0.1, label="Temperature"),
                gr.Slider(
                    minimum=0.1,
                    maximum=1.0,
                    value=0.95,
                    step=0.05,
                    label="Top-p (nucleus sampling)",
                ),
            ],
        )
        
    with gr.Row():
        input_text = gr.Textbox(label="Input text")
        input_tokens = gr.Textbox(label="Input tokens")
        output_ids = gr.Textbox(label="Output tokens")

        def update_tokens(input_text):
            input_tokens_str, output_ids = generate_tokens(input_text)
            return input_tokens_str, output_ids

        input_text.change(update_tokens, 
                          inputs=input_text,
                          outputs=[input_tokens, output_ids])

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