app.py

import gradio as gr
import tiktoken
import random

# Load the tokenizers
enc_gpt4o = tiktoken.encoding_for_model("gpt-4o")
enc_gpt3_5turbo = tiktoken.encoding_for_model("gpt-3.5-turbo")

def get_color_mapping(tokens):
    unique_tokens = list(set(tokens))
    colors = ["#" + ''.join([random.choice('0123456789ABCDEF') for _ in range(6)]) for _ in unique_tokens]
    color_mapping = dict(zip(unique_tokens, colors))
    return color_mapping

def process_model(text, encoder, model_name):
    token_ids = encoder.encode(text)
    tokens = [encoder.decode([id]) for id in token_ids]
    num_tokens = len(tokens)
    
    color_mapping = get_color_mapping(tokens)

    modelname_html = f'<h2>{model_name}</h2>'
    
    tokens_colored = [f'<span style="color:{color_mapping[token]}; font-weight: bold;">{token}</span>' for token in tokens]
    token_ids_colored = [f'<span style="color:{color_mapping[token]}; font-weight: bold;">{token_id}</span>' for token, token_id in zip(tokens, token_ids)]
    
    tokens_html = f'<h3>{model_name} Tokens</h3>' + ' '.join(tokens_colored)
    num_tokens_html = f'<h3>Number of Tokens: <span style="font-size: 20px; font-weight: bold;">{num_tokens}</span></h3>'
    token_ids_html = f'<h3>{model_name} Token IDs</h3>' + ' '.join(map(str, token_ids_colored))
    
    return modelname_html + num_tokens_html + tokens_html + token_ids_html

def tokenize_input(text):
    gpt4o_result = process_model(text, enc_gpt4o, "GPT-4o")
    gpt35turbo_result = process_model(text, enc_gpt3_5turbo, "GPT-3.5-turbo")
    num_chars = len(text)
    num_chars_html = f'<h2>Number of Characters: <span style="font-size: 20px; font-weight: bold;">{num_chars}</span></h2>'
    return num_chars_html, gpt4o_result, gpt35turbo_result

# Create the Gradio interface using Blocks
with gr.Blocks() as demo:
    gr.Markdown("## GPT Tokenizer Comparison App")
    with gr.Row():
        input_text = gr.Textbox(lines=2, placeholder="Enter text here...", label="Enter text to tokenize and compare results between GPT-4o and GPT-3.5-turbo tokenizers.")
        num_chars_output = gr.HTML()
    with gr.Row():
        gpt4o_output = gr.HTML(label="GPT-4o")
        gpt35turbo_output = gr.HTML(label="GPT-3.5-turbo")
    
    input_text.change(tokenize_input, inputs=[input_text], outputs=[num_chars_output, gpt4o_output, gpt35turbo_output])
    input_text.submit(tokenize_input, inputs=[input_text], outputs=[num_chars_output, gpt4o_output, gpt35turbo_output])

# Launch the app
demo.launch()