app.py

from transformers import pipeline
import gradio as gr
import json
import time

# Initialize the pipeline with the new model
pipe = pipeline("text-generation", model="Blexus/Quble_test_model_v1_INSTRUCT_v2")

def format_prompt(message, system, history):
    prompt = f"SYSTEM: {system} <|endofsystem|>"
    
    for entry in history:
        if len(entry) == 2:
            user_prompt, bot_response = entry
            prompt += f"USER: {user_prompt} <|endofuser|>\nASSISTANT: {bot_response}<|endoftext|>\n"
    
    prompt += f"USER: {message}<|endofuser|>\nASSISTANT:"
    return prompt

def generate(prompt, system, history, temperature=0.9, max_new_tokens=4096, top_p=0.9, repetition_penalty=1.2):
    temperature = float(temperature)
    if temperature < 1e-2:
        temperature = 1e-2
    top_p = float(top_p)

    formatted_prompt = format_prompt(prompt, system, history)
    response_text = "We are sorry but Quble doesn't know how to answer."
    # Generate the response without streaming
    try:
            response = pipe(formatted_prompt, max_new_tokens=max_new_tokens, temperature=temperature, top_p=top_p, repetition_penalty=repetition_penalty)[0]["generated_text"]
            response_text = response.split("ASSISTANT:")[-1].strip()

            # Simulate streaming by yielding parts of the response
            accumulated_response = ""  # To keep track of the full response
            for char in response_text:
                accumulated_response += char  # Append the new character
                yield accumulated_response  # Yield the accumulated response
                time.sleep(0.02)  # Add a slight delay to simulate typing

    except Exception as e:
            print(f"Error generating response: {e}")

customCSS = """
#component-7 {
  height: 1600px;
  flex-grow: 4;
}
"""

additional_inputs = [
    gr.Textbox(
        label="System prompt",
        value="You are a helpful intelligent assistant. Your name is Quble.",
        info="System prompt",
        interactive=True,
    ),
    gr.Slider(
        label="Temperature",
        value=0.9,
        minimum=0.0,
        maximum=1.0,
        step=0.05,
        interactive=True,
        info="Higher values produce more diverse outputs",
    ),
    gr.Slider(
        label="Max new tokens",
        value=1024,
        minimum=64,
        maximum=4096,
        step=64,
        interactive=True,
        info="The maximum numbers of new tokens",
    ),
    gr.Slider(
        label="Top-p (nucleus sampling)",
        value=0.90,
        minimum=0.0,
        maximum=1,
        step=0.05,
        interactive=True,
        info="Higher values sample more low-probability tokens",
    ),
    gr.Slider(
        label="Repetition penalty",
        value=1.2,
        minimum=1.0,
        maximum=2.0,
        step=0.05,
        interactive=True,
        info="Penalize repeated tokens",
    )
]

with gr.Blocks(theme=gr.themes.Soft()) as demo:
    gr.ChatInterface(
        generate,
        additional_inputs=additional_inputs,
    )
    
demo.set_css(customCSS)

demo.queue().launch(debug=True)