import gradio as gr from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline import torch import re # Initialize the model model_id = "Detsutut/Igea-350M-v0.0.1" model = AutoModelForCausalLM.from_pretrained(model_id, load_in_8bit=True, device_map='auto') tokenizer = AutoTokenizer.from_pretrained(model_id) gen_pipeline = pipeline( "text-generation", model=model, tokenizer=tokenizer ) # Define the function to generate text def generate_text(input_text, max_new_tokens, temperature, top_p, split_output): if split_output: max_new_tokens=30 top_p=0.95 output = gen_pipeline( input_text, max_new_tokens=max_new_tokens, temperature=temperature, top_p=top_p, return_full_text = False ) generated_text = output[0]['generated_text'] if split_output: sentences = re.split('(?{input_text}{generated_text}" # Create the Gradio interface input_text = gr.Textbox(lines=2, placeholder="Enter your text here...", label="Input Text") max_new_tokens = gr.Slider(minimum=1, maximum=200, value=30, step=1, label="Max New Tokens") temperature = gr.Slider(minimum=0.1, maximum=2.0, value=1.0, step=0.1, label="Temperature") top_p = gr.Slider(minimum=0.0, maximum=1.0, value=0.95, step=0.01, label="Top-p") split_output = gr.Checkbox(label="Quick single-sentence output", value=True) with gr.Blocks(css="#outbox { border-radius: 8px !important; border: 1px solid #e5e7eb !important; padding: 8px !important; text-align:center !important;}") as iface: gr.Markdown("# Igea Text Generation Interface ⚕️🩺") gr.Markdown("⚠️ 🐢💬 This model runs on a **hardware-limited**, free-tier HuggingFace space, resulting in a **low output token throughput** (approx. 1 token/s)") input_text.render() with gr.Accordion("Advanced Options", open=False): max_new_tokens.render() temperature.render() top_p.render() split_output.render() output = gr.HTML(label="Generated Text",elem_id="outbox") btn = gr.Button("Generate") btn.click(generate_text, [input_text, max_new_tokens, temperature, top_p, split_output], output) # Launch the interface if __name__ == "__main__": iface.launch(inline=True)