app.py

import gradio as gr
from transformers import AutoProcessor, AutoModelForVision2Seq

# Configuración del modelo y procesador
processor = AutoProcessor.from_pretrained("llava-hf/llava-1.5-7b-hf")
model = AutoModelForVision2Seq.from_pretrained("llava-hf/llava-1.5-7b-hf", device_map="auto")

def process_image(image, prompt):
    inputs = processor(images=image, text=prompt, return_tensors="pt").to("cuda")
    outputs = model.generate(**inputs)
    return processor.decode(outputs[0], skip_special_tokens=True)

# Configura la interfaz
interface = gr.Interface(
    fn=process_image,
    inputs=["image", "text"],
    outputs="text",
    title="LLaVA Space",
    description="Describe imágenes con prompts personalizados usando LLaVA."
)

interface.launch()