Fix ZeroGPU crash: import spaces first, real @spaces.GPU, FLUX.1-schnell

app.py

@@ -1,73 +1,65 @@
+import spaces  # MUST be imported before any CUDA-touching package (torch/diffusers)
+
 import gradio as gr
 import numpy as np
 import random
-
-# import spaces #[uncomment to use ZeroGPU]
-from diffusers import DiffusionPipeline
 import torch
+from diffusers import DiffusionPipeline
 
-device = "cuda" if torch.cuda.is_available() else "cpu"
-model_repo_id = "stabilityai/sdxl-turbo"  # Replace to the model you would like to use
+# ---------------------------------------------------------------------------
+# Model
+# ---------------------------------------------------------------------------
+# FLUX.1-schnell: Apache-2.0, 1-4 step distilled text-to-image. Fast on ZeroGPU.
+MODEL_REPO_ID = "black-forest-labs/FLUX.1-schnell"
 
-if torch.cuda.is_available():
-    torch_dtype = torch.float16
-else:
-    torch_dtype = torch.float32
+dtype = torch.bfloat16
+device = "cuda" if torch.cuda.is_available() else "cpu"
 
-pipe = DiffusionPipeline.from_pretrained(model_repo_id, torch_dtype=torch_dtype)
-pipe = pipe.to(device)
+# Load at module level on cuda (ZeroGPU CUDA-emulation makes this safe outside @spaces.GPU)
+pipe = DiffusionPipeline.from_pretrained(MODEL_REPO_ID, torch_dtype=dtype).to(device)
 
 MAX_SEED = np.iinfo(np.int32).max
-MAX_IMAGE_SIZE = 1024
+MAX_IMAGE_SIZE = 2048
 
 
-# @spaces.GPU #[uncomment to use ZeroGPU]
+@spaces.GPU(duration=60)
 def infer(
     prompt,
-    negative_prompt,
     seed,
     randomize_seed,
     width,
     height,
-    guidance_scale,
     num_inference_steps,
     progress=gr.Progress(track_tqdm=True),
 ):
     if randomize_seed:
         seed = random.randint(0, MAX_SEED)
-
-    generator = torch.Generator().manual_seed(seed)
-
+    generator = torch.Generator(device=device).manual_seed(seed)
     image = pipe(
         prompt=prompt,
-        negative_prompt=negative_prompt,
-        guidance_scale=guidance_scale,
-        num_inference_steps=num_inference_steps,
         width=width,
         height=height,
+        num_inference_steps=num_inference_steps,
+        guidance_scale=0.0,        # schnell is guidance-distilled
+        max_sequence_length=256,
         generator=generator,
     ).images[0]
-
     return image, seed
 
 
 examples = [
+    "A magical city at twilight, glowing windows, storybook illustration, warm light",
     "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k",
-    "An astronaut riding a green horse",
     "A delicious ceviche cheesecake slice",
 ]
 
 css = """
-#col-container {
-    margin: 0 auto;
-    max-width: 640px;
-}
+#col-container { margin: 0 auto; max-width: 640px; }
 """
 
 with gr.Blocks(css=css) as demo:
     with gr.Column(elem_id="col-container"):
-        gr.Markdown(" # Text-to-Image Gradio Template")
-
+        gr.Markdown("# 🖼️ NEXUS Visual Weaver — FLUX.1-schnell")
         with gr.Row():
             prompt = gr.Text(
                 label="Prompt",
@@ -76,77 +68,27 @@ with gr.Blocks(css=css) as demo:
                 placeholder="Enter your prompt",
                 container=False,
             )
-
             run_button = gr.Button("Run", scale=0, variant="primary")
 
         result = gr.Image(label="Result", show_label=False)
 
         with gr.Accordion("Advanced Settings", open=False):
-            negative_prompt = gr.Text(
-                label="Negative prompt",
-                max_lines=1,
-                placeholder="Enter a negative prompt",
-                visible=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)
-
             with gr.Row():
-                width = gr.Slider(
-                    label="Width",
-                    minimum=256,
-                    maximum=MAX_IMAGE_SIZE,
-                    step=32,
-                    value=1024,  # Replace with defaults that work for your model
-                )
-
-                height = gr.Slider(
-                    label="Height",
-                    minimum=256,
-                    maximum=MAX_IMAGE_SIZE,
-                    step=32,
-                    value=1024,  # Replace with defaults that work for your model
-                )
-
+                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)
             with gr.Row():
-                guidance_scale = gr.Slider(
-                    label="Guidance scale",
-                    minimum=0.0,
-                    maximum=10.0,
-                    step=0.1,
-                    value=0.0,  # Replace with defaults that work for your model
-                )
-
                 num_inference_steps = gr.Slider(
-                    label="Number of inference steps",
-                    minimum=1,
-                    maximum=50,
-                    step=1,
-                    value=2,  # Replace with defaults that work for your model
+                    label="Inference steps", minimum=1, maximum=8, step=1, value=4
                 )
 
         gr.Examples(examples=examples, inputs=[prompt])
+
     gr.on(
         triggers=[run_button.click, prompt.submit],
         fn=infer,
-        inputs=[
-            prompt,
-            negative_prompt,
-            seed,
-            randomize_seed,
-            width,
-            height,
-            guidance_scale,
-            num_inference_steps,
-        ],
+        inputs=[prompt, seed, randomize_seed, width, height, num_inference_steps],
         outputs=[result, seed],
     )
 

requirements.txt

@@ -1,6 +1,6 @@
+spaces
 accelerate
 diffusers
-invisible_watermark
-torch
 transformers
-xformers
+sentencepiece
+torch