Update app.py

app.py

@@ -4,29 +4,46 @@ import random
 import spaces
 import torch
 from diffusers import DiffusionPipeline
+from accelerate import init_empty_weights, load_checkpoint_and_dispatch
 
+# Configuración para usar bfloat16 y CUDA si está disponible
 dtype = torch.bfloat16
 device = "cuda" if torch.cuda.is_available() else "cpu"
 
-pipe = DiffusionPipeline.from_pretrained("black-forest-labs/FLUX.1-schnell", torch_dtype=dtype).to(device)
+# Inicialización del modelo en la RAM
+with init_empty_weights():
+    pipe = DiffusionPipeline.from_pretrained("black-forest-labs/FLUX.1-schnell", torch_dtype=dtype)
+
+# Cargar el modelo en la RAM y despachar los pesos a la GPU
+pipe = load_checkpoint_and_dispatch(
+    pipe, 
+    "black-forest-labs/FLUX.1-schnell", 
+    device_map="auto",  # Automatiza el uso de RAM y GPU
+    offload_folder=None,  # Evita que se almacenen los pesos temporalmente en el disco
+).to(device)
 
 MAX_SEED = np.iinfo(np.int32).max
 MAX_IMAGE_SIZE = 2048
 
 @spaces.GPU()
-def infer(prompt, seed=42, randomize_seed=False, width=1024, height=1024, num_inference_steps=4, progress=gr.Progress(track_tqdm=True)):
+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)):
     if randomize_seed:
         seed = random.randint(0, MAX_SEED)
     generator = torch.Generator().manual_seed(seed)
-    image = pipe(
-            prompt = prompt, 
-            width = width,
-            height = height,
-            num_inference_steps = num_inference_steps, 
-            generator = generator,
-            guidance_scale=0.0
-    ).images[0] 
-    return image, 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
+        ).images[0]
+        images.append(image)
+    
+    return images, seed
  
 examples = [
     "a tiny astronaut hatching from an egg on the moon",
@@ -61,7 +78,7 @@ with gr.Blocks(css=css) as demo:
             
             run_button = gr.Button("Run", scale=0)
         
-        result = gr.Image(label="Result", show_label=False)
+        results = gr.Gallery(label="Results", show_label=False, elem_id="image-gallery")
         
         with gr.Accordion("Advanced Settings", open=False):
             
@@ -95,7 +112,6 @@ with gr.Blocks(css=css) as demo:
             
             with gr.Row():
                 
-  
                 num_inference_steps = gr.Slider(
                     label="Number of inference steps",
                     minimum=1,
@@ -103,20 +119,28 @@ with gr.Blocks(css=css) as demo:
                     step=1,
                     value=4,
                 )
+            
+            num_images = gr.Slider(
+                label="Number of images",
+                minimum=1,
+                maximum=300,
+                step=1,
+                value=1,
+            )
         
         gr.Examples(
             examples = examples,
             fn = infer,
             inputs = [prompt],
-            outputs = [result, seed],
+            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],
-        outputs = [result, seed]
+        inputs = [prompt, seed, randomize_seed, width, height, num_inference_steps, num_images],
+        outputs = [results, seed]
     )
 
-demo.launch()
+demo.launch()