Update with latest text and UI

app.py

@@ -1,4 +1,3 @@
-# !pip install diffusers
 import torch
 from diffusers import DDIMPipeline, DDPMPipeline, PNDMPipeline
 from diffusers import DDIMScheduler, DDPMScheduler, PNDMScheduler
@@ -20,6 +19,7 @@ ddim_pipeline = DDIMPipeline(unet=model, scheduler=ddim_scheduler)
 
 pndm_scheduler = PNDMScheduler.from_config(model_id, subfolder="scheduler")
 pndm_pipeline = PNDMPipeline(unet=model, scheduler=pndm_scheduler)
+
 # run pipeline in inference (sample random noise and denoise)
 def predict(steps=100, seed=42,scheduler="ddim"):
     torch.cuda.empty_cache()
@@ -29,6 +29,8 @@ def predict(steps=100, seed=42,scheduler="ddim"):
     elif(scheduler == "ddpm"):
         image = ddpm_pipeline(generator=generator)["sample"]
     elif(scheduler == "pndm"):
+        if(steps > 100):
+            steps = 100
         image = pndm_pipeline(generator=generator, num_inference_steps=steps)["sample"]
 
     image_processed = image.cpu().permute(0, 2, 3, 1)
@@ -46,9 +48,12 @@ random_seed = random.randint(0, 2147483647)
 gr.Interface(
     predict,
     inputs=[
-        gr.inputs.Slider(1, 100, label='Inference Steps', default=20, step=1),
-        gr.inputs.Slider(0, 2147483647, label='Seed', default=random_seed),
-        gr.inputs.Radio(["ddim", "ddpm", "pndm"], default="ddpm",label="Diffusion scheduler")
+        gr.inputs.Slider(1, 1000, label='Inference Steps (ignored for the ddpm scheduler, that diffuses for 1000 steps - limited to 100 steps max for pndm)', default=20, step=1),
+        gr.inputs.Slider(0, 2147483647, label='Seed', default=random_seed, step=1),
+        gr.inputs.Radio(["ddpm", "ddim", "pndm"], default="ddpm",label="Diffusion scheduler")
     ],
-    outputs=gr.Image(shape=[256,256], type="pil"),
+    outputs=gr.Image(shape=[256,256], type="pil", elem_id="output_image"),
+    css="#output_image{width: 256px}",
+    title="ddpm-celebahq-256 diffusion - 🧨 diffusers library",
+    description="This Spaces contains an unconditional diffusion process for the <a href=\"https://huggingface.co/google/ddpm-celebahq-256\">ddpm-celebahq-256</a> face generator model by <a href=\"https://huggingface.co/google\">Google</a> using the <a href=\"https://github.com/huggingface/diffusers\">diffusers library</a>. You can try the diffusion process not only with the default <code>ddpm</code> scheduler but also with <code>ddim</code> and <code>pndm</code>, showcasing the modularity of the library. <a href=\"https://github.com/huggingface/diffusers/tree/main/src/diffusers/schedulers\">Learn more about schedulers here.</a>",
 ).launch()