app.py

import gradio as gr
import os
import torch
import subprocess
from transformers import AutoModelForCausalLM
from huggingface_hub import login

# Install required package
subprocess.run(
    "pip install flash-attn --no-build-isolation",
    env={"FLASH_ATTENTION_SKIP_CUDA_BUILD": "TRUE"},
    shell=True,
)

hf_token = os.getenv("HF_TOKEN")
login(token=hf_token, add_to_git_credential=True)

# Function to get the model summary
@spaces.GPU
def get_model_summary(model_name):
    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
    model = AutoModelForCausalLM.from_pretrained(model_name, trust_remote_code=True).to(device)
    return str(model)

# Create the Gradio Blocks interface
with gr.Blocks() as demo:
    with gr.Row():
        with gr.Column():
            textbox = gr.Textbox(label="Model Name")
            examples = gr.Examples(
                examples=[
                    ["google/gemma-7b"],
                    ["microsoft/Phi-3-mini-4k-instruct"],
                    ["meta-llama/Meta-Llama-3-8B"],
                    ["mistralai/Mistral-7B-Instruct-v0.3"],
                    ["vikhyatk/moondream2"],
                    ["microsoft/Phi-3-vision-128k-instruct"],
                    ["openbmb/MiniCPM-Llama3-V-2_5"],
                    ["google/paligemma-3b-mix-224"],
                    ["HuggingFaceM4/idefics2-8b-chatty"],
                    ["mistralai/Codestral-22B-v0.1"]
                ],
                inputs=textbox
            )
            submit_button = gr.Button("Submit")
        with gr.Column():
            output = gr.Textbox(label="Output", lines=20)

    submit_button.click(fn=get_model_summary, inputs=textbox, outputs=output)

# Launch the interface
demo.launch()