Use constants.

app.py

@@ -15,7 +15,7 @@ SELECT_LABEL = "Select as seed"
 MODEL_ID = "CompVis/ldm-text2im-large-256"
 STEPS = 50
 ETA = 0.3
-GUIDANCE_SCALE = 12
+GUIDANCE_SCALE = 6
 
 ldm = DiffusionPipeline.from_pretrained(MODEL_ID)
 
@@ -31,14 +31,19 @@ with gr.Blocks(css=".container { max-width: 800px; margin: auto; }") as demo:
         return images[0]
 
     def infer_grid(prompt, n=6, seeds=[]):
-        # Unfortunately we have to iterate instead requesting all images at once,
-        # because we have no way to get the generation seeds.
+        # Unfortunately we have to iterate instead of requesting all images at once,
+        # because we have no way to get the intermediate generation seeds.
         result = defaultdict(list)
         for _, seed in zip_longest(range(n), seeds, fillvalue=None):
             seed = random.randint(0, 2**32 - 1) if seed is None else seed
             print(f"Setting seed {seed}")
             _ = torch.manual_seed(seed)
-            images = ldm([prompt], num_inference_steps=50, eta=0.3, guidance_scale=6)["sample"]
+            images = ldm(
+                [prompt],
+                num_inference_steps=STEPS,
+                eta=ETA,
+                guidance_scale=GUIDANCE_SCALE
+            )["sample"]
             result["images"].append(images[0])
             result["seeds"].append(seed)
         return result["images"], result["seeds"]