Update app.py

app.py

@@ -1,7 +1,7 @@
 import gradio as gr
 import torch
 from diffusers import DiffusionPipeline
-from accelerate import init_empty_weights, load_checkpoint_and_dispatch
+from accelerate import load_checkpoint_and_dispatch
 
 # Configuración para usar bfloat16 y CUDA si está disponible
 dtype = torch.bfloat16
@@ -13,44 +13,43 @@ pipe = None
 def load_model():
     global pipe
     if pipe is None:
-        with init_empty_weights():
-            pipe = DiffusionPipeline.from_pretrained(
-                "black-forest-labs/FLUX.1-schnell", 
-                torch_dtype=dtype
-            )
-        # Cargar el modelo en la GPU sin intentar acceder a `named_parameters`
+        pipe = DiffusionPipeline.from_pretrained(
+            "black-forest-labs/FLUX.1-schnell", 
+            torch_dtype=dtype
+        )
+        # Despachar los pesos a la GPU, evitando acceder a named_parameters
         pipe = load_checkpoint_and_dispatch(
             pipe, 
             "black-forest-labs/FLUX.1-schnell", 
-            device_map="auto",
-            offload_folder=None,
-        ).to(device)
+            device_map="auto",  # Automatiza el uso de RAM y GPU
+            offload_folder=None  # Evita que se almacenen los pesos temporalmente en el disco
+        )
+        pipe.to(device)
 
 MAX_SEED = torch.iinfo(torch.int32).max
-MAX_IMAGE_SIZE = 2048
 
-def infer(prompt, seed=42, randomize_seed=False, width=1024, height=1024, num_inference_steps=4, num_images=1, progress=gr.Progress(track_tqdm=True)):
+@gr.Interface()
+def infer(prompt, seed=42, randomize_seed=False, width=1024, height=1024, num_inference_steps=4, num_images=1):
     load_model()  # Asegurarse de que el modelo esté cargado antes de la inferencia
     
     if randomize_seed:
         seed = torch.randint(0, MAX_SEED, (1,)).item()
-    generator = torch.Generator(device=device).manual_seed(seed)
+    generator = torch.Generator(device).manual_seed(seed)
     
     images = []
     for _ in range(num_images):
         image = pipe(
-            prompt=prompt, 
-            width=width,
-            height=height,
-            num_inference_steps=num_inference_steps, 
-            generator=generator,
-            guidance_scale=0.0
+                prompt=prompt, 
+                width=width,
+                height=height,
+                num_inference_steps=num_inference_steps, 
+                generator=generator,
+                guidance_scale=0.0
         ).images[0]
         images.append(image)
     
     return images, seed
 
-# Gradio Interface
 examples = [
     "a tiny astronaut hatching from an egg on the moon",
     "a cat holding a sign that says hello world",
@@ -73,21 +72,81 @@ with gr.Blocks(css=css) as demo:
         """)
         
         with gr.Row():
-            prompt = gr.Textbox(label="Prompt", show_label=False, max_lines=1, placeholder="Enter your prompt")
-            run_button = gr.Button("Run")
+            
+            prompt = gr.Text(
+                label="Prompt",
+                show_label=False,
+                max_lines=1,
+                placeholder="Enter your prompt",
+                container=False,
+            )
+            
+            run_button = gr.Button("Run", scale=0)
         
         results = gr.Gallery(label="Results", show_label=False, elem_id="image-gallery")
         
         with gr.Accordion("Advanced Settings", open=False):
-            seed = gr.Slider(label="Seed", minimum=0, maximum=MAX_SEED, step=1, value=0)
+            
+            seed = gr.Slider(
+                label="Seed",
+                minimum=0,
+                maximum=MAX_SEED,
+                step=1,
+                value=0,
+            )
+            
             randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
-            width = gr.Slider(label="Width", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=1024)
-            height = gr.Slider(label="Height", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=1024)
-            num_inference_steps = gr.Slider(label="Number of inference steps", minimum=1, maximum=50, step=1, value=4)
-            num_images = gr.Slider(label="Number of images", minimum=1, maximum 10, step=1, value=1)
+            
+            with gr.Row():
+                
+                width = gr.Slider(
+                    label="Width",
+                    minimum=256,
+                    maximum=2048,
+                    step=32,
+                    value=1024,
+                )
+                
+                height = gr.Slider(
+                    label="Height",
+                    minimum=256,
+                    maximum=2048,
+                    step=32,
+                    value=1024,
+                )
+            
+            with gr.Row():
+                
+                num_inference_steps = gr.Slider(
+                    label="Number of inference steps",
+                    minimum=1,
+                    maximum=50,
+                    step=1,
+                    value=4,
+                )
+            
+            num_images = gr.Slider(
+                label="Number of images",
+                minimum=1,
+                maximum=300,
+                step=1,
+                value=1,
+            )
         
-        gr.Examples(examples, inputs=[prompt], outputs=[results, seed])
-        run_button.click(infer, inputs=[prompt, seed, randomize_seed, width, height, num_inference_steps, num_images], outputs=[results, seed])
+        gr.Examples(
+            examples = examples,
+            fn = infer,
+            inputs = [prompt],
+            outputs = [results, seed],
+            cache_examples="lazy"
+        )
+
+    gr.on(
+        triggers=[run_button.click, prompt.submit],
+        fn = infer,
+        inputs = [prompt, seed, randomize_seed, width, height, num_inference_steps, num_images],
+        outputs = [results, seed]
+    )
 
 # Crear un enlace público con share=True
 demo.launch(share=True)