Update app.py

app.py

@@ -4,9 +4,7 @@ import numpy as np
 import random
 import spaces
 import torch
-from diffusers import  SanaSprintPipeline
-#from transformers import CLIPTextModel, CLIPTokenizer,T5EncoderModel, T5TokenizerFast
-#from live_preview_helpers import calculate_shift, retrieve_timesteps, flux_pipe_call_that_returns_an_iterable_of_images
+from diffusers import SanaSprintPipeline
 
 dtype = torch.bfloat16
 device = "cuda" if torch.cuda.is_available() else "cpu"
@@ -15,17 +13,25 @@ pipe = SanaSprintPipeline.from_pretrained(
     "Efficient-Large-Model/Sana_Sprint_0.6B_1024px_diffusers",
     torch_dtype=torch.bfloat16
 )
+pipe2 = SanaSprintPipeline.from_pretrained(
+    "Efficient-Large-Model/Sana_Sprint_1.6B_1024px_diffusers",
+    torch_dtype=torch.bfloat16
+)
 pipe.to(device)
+pipe2.to(device)
 MAX_SEED = np.iinfo(np.int32).max
 MAX_IMAGE_SIZE = 1024
 
 @spaces.GPU(duration=5)
-def infer(prompt, seed=42, randomize_seed=False, width=1024, height=1024, guidance_scale=3.5, num_inference_steps=2, progress=gr.Progress(track_tqdm=True)):
+def infer(prompt, model_size, seed=42, randomize_seed=False, width=1024, height=1024, guidance_scale=3.5, num_inference_steps=2, progress=gr.Progress(track_tqdm=True)):
     if randomize_seed:
         seed = random.randint(0, MAX_SEED)
     generator = torch.Generator().manual_seed(seed)
     
-    img = pipe(
+    # Choose the appropriate model based on selected model size
+    selected_pipe = pipe if model_size == "0.6B" else pipe2
+    
+    img = selected_pipe(
             prompt=prompt,
             guidance_scale=guidance_scale,
             num_inference_steps=num_inference_steps,
@@ -37,12 +43,21 @@ def infer(prompt, seed=42, randomize_seed=False, width=1024, height=1024, guidan
     print(img)
     return img.images[0], seed
     
-examples = [
-    "a tiny astronaut hatching from an egg on the moon",
-    "a cat holding a sign that says hello world",
-    "an anime illustration of a wiener schnitzel",
+# Different examples for each model size
+examples_06B = [
+    "a majestic castle on a floating island",
+    "a robotic chef cooking in a futuristic kitchen",
+    "a magical forest with glowing mushrooms"
 ]
 
+examples_16B = [
+    "a steampunk city with airships in the sky",
+    "a photorealistic fox in a snowy landscape",
+    "an underwater temple with ancient ruins"
+]
+
+# We'll use the appropriate set based on the model selection
+
 css="""
 #col-container {
     margin: 0 auto;
@@ -53,7 +68,15 @@ css="""
 with gr.Blocks(css=css) as demo:
     
     with gr.Column(elem_id="col-container"):
-        gr.Markdown(f"""# Sana Sprint 0.6B""")
+        gr.Markdown(f"""# Sana Sprint""")
+        
+        # Add radio button for model selection
+        model_size = gr.Radio(
+            label="Model Size",
+            choices=["0.6B", "1.6B"],
+            value="0.6B",
+            interactive=True
+        )
         
         with gr.Row():
             
@@ -117,18 +140,29 @@ with gr.Blocks(css=css) as demo:
                     value=2,
                 )
         
-        gr.Examples(
-            examples = examples,
-            fn = infer,
-            inputs = [prompt],
-            outputs = [result, seed],
-            cache_examples="lazy"
-        )
+        with gr.Row():
+            examples_container = gr.Examples(
+                examples = examples_06B,  # Start with 0.6B examples
+                fn = infer,
+                inputs = [prompt, model_size],
+                outputs = [result, seed],
+                cache_examples="lazy",
+                label="Example Prompts"
+            )
+            
+        # Update examples when model size changes
+        def update_examples(model_choice):
+            if model_choice == "0.6B":
+                return gr.Examples.update(examples=examples_06B)
+            else:
+                return gr.Examples.update(examples=examples_16B)
+                
+        model_size.change(fn=update_examples, inputs=[model_size], outputs=[examples_container])
 
     gr.on(
         triggers=[run_button.click, prompt.submit],
         fn = infer,
-        inputs = [prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps],
+        inputs = [prompt, model_size, seed, randomize_seed, width, height, guidance_scale, num_inference_steps],  # Add model_size to inputs
         outputs = [result, seed]
     )