Update app.py

app.py

@@ -251,8 +251,10 @@ def captioning(img):
     # Add more prompts here
     ]
     output_prompt=[]
-    # Initial caption generation without a prompt:
-    inputsa = processor5(images=img, return_tensors="pt").to('cuda')
+    cap_prompt = (
+            "Describe this image with a caption."
+    )    
+    inputsa = processor5(images=img, text=cap_prompt, return_tensors="pt").to('cuda')
     generated_ids = model5.generate(
         **inputsa,
         do_sample=False,
@@ -263,7 +265,7 @@ def captioning(img):
         repetition_penalty=1.5,
         length_penalty=1.0,
         temperature=1,
-)
+    )
     
     generated_text = processor5.batch_decode(generated_ids, skip_special_tokens=True)[0].strip()
     output_prompt.append(generated_text)
@@ -272,16 +274,16 @@ def captioning(img):
     for prompt in prompts_array:
         inputs = processor5(images=img, text=prompt, return_tensors="pt").to('cuda')
         generated_ids = model5.generate(
-        **inputs,
-        do_sample=False,
-        num_beams=5,
-        max_length=256,
-        min_length=1,
-        top_p=0.9,
-        repetition_penalty=1.5,
-        length_penalty=1.0,
-        temperature=1,
-) 
+            **inputs,
+            do_sample=False,
+            num_beams=5,
+            max_length=128,
+            min_length=1,
+            top_p=0.9,
+            repetition_penalty=1.5,
+            length_penalty=1.0,
+            temperature=1,
+        ) 
         # Adjust max_length if needed
         generated_text = processor5.batch_decode(generated_ids, skip_special_tokens=True)[0].strip()
         response_text = generated_text.replace(prompt, "").strip() #Or could try .split(prompt, 1)[-1].strip()