Update app.py

app.py

@@ -20,20 +20,31 @@ def generate_image(prompt: str,negative_prompt:str , inference_steps: int = 25,
     num_samples = 1
     prompt_ids = pipeline.prepare_inputs([prompt] * num_samples)
     prompt_ids = shard(prompt_ids)
-    neg_prompt_ids = pipeline.prepare_inputs([negative_prompt] * num_samples)
-    neg_prompt_ids = shard(neg_prompt_ids)
-
-    images = pipeline(
-        prompt_ids=prompt_ids,
-        params=p_params,
-        prng_seed=rng,
-        height=128,
-        width=128,
-        num_inference_steps=int(inference_steps),
-        neg_prompt_ids=neg_prompt_ids,
-        guidance_scale =float(guidance_scale),
-        jit=True,
-    ).images
+    if negative_prompt == "":
+         images = pipeline(
+            prompt_ids=prompt_ids,
+            params=p_params,
+            prng_seed=rng,
+            height=128,
+            width=128,
+            num_inference_steps=int(inference_steps),
+            guidance_scale =float(guidance_scale),
+            jit=True,
+        ).images
+    else:    
+        neg_prompt_ids = pipeline.prepare_inputs([negative_prompt] * num_samples)
+        neg_prompt_ids = shard(neg_prompt_ids)
+        images = pipeline(
+            prompt_ids=prompt_ids,
+            params=p_params,
+            prng_seed=rng,
+            height=128,
+            width=128,
+            num_inference_steps=int(inference_steps),
+            neg_prompt_ids=neg_prompt_ids,
+            guidance_scale =float(guidance_scale),
+            jit=True,
+        ).images
 
     images = images.reshape((num_samples,) + images.shape[-3:])
     images = pipeline.numpy_to_pil(images)