Update app.py

app.py

@@ -1,28 +1,35 @@
 import gradio as gr
 import numpy as np
 import random
-
 import spaces
 from diffusers import DiffusionPipeline, FlowMatchEulerDiscreteScheduler
 import torch
 
+# Set device and model parameters
 device = "cuda" if torch.cuda.is_available() else "cpu"
 model_repo_id = "tensorart/stable-diffusion-3.5-large-TurboX"
+torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
 
-if torch.cuda.is_available():
-    torch_dtype = torch.float16
-else:
-    torch_dtype = torch.float32
-
+# Load the pipeline with the specified torch_dtype and move it to the GPU
 pipe = DiffusionPipeline.from_pretrained(model_repo_id, torch_dtype=torch_dtype)
-
-pipe.scheduler = FlowMatchEulerDiscreteScheduler.from_pretrained(model_repo_id, subfolder="scheduler", shift=5)
-
+pipe.scheduler = FlowMatchEulerDiscreteScheduler.from_pretrained(
+    model_repo_id, subfolder="scheduler", shift=5
+)
 pipe = pipe.to(device)
 
 MAX_SEED = np.iinfo(np.int32).max
 MAX_IMAGE_SIZE = 1024
 
+def truncate_text(text, max_tokens=77):
+    """
+    Truncate the input text to a maximum of max_tokens using the pipeline's tokenizer.
+    """
+    if text.strip() == "":
+        return text
+    tokens = pipe.tokenizer(text, truncation=True, max_length=max_tokens, add_special_tokens=True)
+    truncated_text = pipe.tokenizer.decode(tokens["input_ids"], skip_special_tokens=True)
+    return truncated_text
+
 @spaces.GPU(duration=65)
 def infer(
     prompt,
@@ -38,8 +45,13 @@ def infer(
     if randomize_seed:
         seed = random.randint(0, MAX_SEED)
 
-    generator = torch.Generator().manual_seed(seed)
+    generator = torch.Generator(device=device).manual_seed(seed)
 
+    # Explicitly truncate prompts to avoid CLIP token warnings.
+    prompt = truncate_text(prompt, max_tokens=77)
+    negative_prompt = truncate_text(negative_prompt, max_tokens=77) if negative_prompt.strip() else ""
+
+    # Generate the image using the truncated prompts.
     image = pipe(
         prompt=prompt,
         negative_prompt=negative_prompt,
@@ -52,9 +64,9 @@ def infer(
 
     return image, seed
 
-
+# UI Layout
 examples = [
-        "A capybara wearing a suit holding a sign that reads Hello World",
+    "A capybara wearing a suit holding a sign that reads Hello World",
 ]
 
 css = """
@@ -76,18 +88,14 @@ 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",
             )
-
             seed = gr.Slider(
                 label="Seed",
                 minimum=0,
@@ -95,18 +103,15 @@ with gr.Blocks(css=css) as demo:
                 step=1,
                 value=0,
             )
-
             randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
-
             with gr.Row():
                 width = gr.Slider(
                     label="Width",
                     minimum=512,
                     maximum=MAX_IMAGE_SIZE,
                     step=32,
-                    value=1024, 
+                    value=1024,
                 )
-
                 height = gr.Slider(
                     label="Height",
                     minimum=512,
@@ -114,7 +119,6 @@ with gr.Blocks(css=css) as demo:
                     step=32,
                     value=1024,
                 )
-
             with gr.Row():
                 guidance_scale = gr.Slider(
                     label="Guidance scale",
@@ -123,15 +127,13 @@ with gr.Blocks(css=css) as demo:
                     step=0.1,
                     value=1.5,
                 )
-
                 num_inference_steps = gr.Slider(
                     label="Number of inference steps",
                     minimum=1,
                     maximum=50,
                     step=1,
-                    value=8, 
+                    value=8,
                 )
-
         gr.Examples(examples=examples, inputs=[prompt], outputs=[result, seed], fn=infer, cache_examples=True, cache_mode="lazy")
     gr.on(
         triggers=[run_button.click, prompt.submit],