app.py

import os
import random
import gradio as gr
from groq import Groq

client = Groq(
    api_key=os.environ.get("Groq_Api_Key")
)

def create_history_messages(history):
    history_messages = [{"role": "user", "content": m[0]} for m in history]
    history_messages.extend([{"role": "assistant", "content": m[1]} for m in history])
    return history_messages

def generate_response(prompt, history, model, temperature, max_tokens, top_p, seed):
    messages = create_history_messages(history)
    messages.append({"role": "user", "content": prompt})
    print(messages)

    if seed == 0:
        seed = random.randint(1, 100000)

    stream = client.chat.completions.create(
        messages=messages,
        model=model,
        temperature=temperature,
        max_tokens=max_tokens,
        top_p=top_p,
        seed=seed,
        stop=None,
        stream=True,
    )

    response = ""
    for chunk in stream:
        delta_content = chunk.choices[0].delta.content
        if delta_content is not None:
            response += delta_content
            yield response

    return response

additional_inputs = [
    gr.Dropdown(choices=["llama3-70b-8192", "llama3-8b-8192", "mixtral-8x7b-32768", "gemma-7b-it"], value="llama3-70b-8192", label="Model"),
    gr.Slider(minimum=0.0, maximum=1.0, step=0.01, value=0.5, label="Temperature", info="Controls diversity of the generated text. Lower is more deterministic, higher is more creative."),
    gr.Slider(minimum=1, maximum=32192, step=1, value=4096, label="Max Tokens", info="The maximum number of tokens that the model can process in a single response.<br>Maximums: 8k for gemma 7b, llama 7b & 70b, 32k for mixtral 8x7b."),
    gr.Slider(minimum=0.0, maximum=1.0, step=0.01, value=0.5, label="Top P", info="A method of text generation where a model will only consider the most probable next tokens that make up the probability p."),
    gr.Number(precision=0, value=42, label="Seed", info="A starting point to initiate generation, use 0 for random")
]

theme = gr.themes.Base(
    neutral_hue=gr.themes.Color(c100="#f3f4f6", c200="#e5e7eb", c300="#d1d5db", c400="#9ca3af", c50="#f9fafb", c500="#000000", c600="#3b82f6", c700="#3b82f6", c800="#26252a", c900="#26252a", c950="#000000"),
    spacing_size="sm",
    radius_size="lg",
).set(
    block_background_fill_dark='*background_fill_primary',
    block_border_color='*background_fill_primary',
    block_border_color_dark='*background_fill_primary',
    block_border_width='0px',
    block_border_width_dark='0px',
    block_label_border_color='*background_fill_primary',
    block_label_border_color_dark='*background_fill_primary',
    block_label_border_width='0px',
    block_label_border_width_dark='0px'
)

with open('top_bar.html', 'r') as file:
    top_bar_html = file.read()

interface = gr.ChatInterface(
    fn=generate_response,
    chatbot=gr.Chatbot(show_label=False, show_share_button=False, show_copy_button=False, likeable=False, layout="bubble"),
    additional_inputs=additional_inputs,
    theme=theme,
    submit_btn="↑",
    undo_btn="Delete",
    retry_btn="Retry",
    head="./top_bar.html"
)

with gr.Blocks(theme=theme) as demo:
    gr.HTML(top_bar_html)
    interface.render()

demo.launch()