Update app.py

app.py

@@ -253,14 +253,36 @@ def captioning(img):
     output_prompt=[]
     # Initial caption generation without a prompt:
     inputsa = processor5(images=img, return_tensors="pt").to('cuda')
-    generated_ids = model5.generate(**inputsa, min_length=42, max_length=64)
+    generated_ids = model5.generate(
+        **inputsa,
+        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,
+)
+    
     generated_text = processor5.batch_decode(generated_ids, skip_special_tokens=True)[0].strip()
     output_prompt.append(generated_text)
     print(generated_text)  
     # Loop through prompts array:
     for prompt in prompts_array:
         inputs = processor5(images=img, text=prompt, return_tensors="pt").to('cuda')
-        generated_ids = model5.generate(**inputs, min_length=32, max_length=42) # Adjust max_length if needed
+        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,
+) 
+        # 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()
         output_prompt.append(response_text)
@@ -364,7 +386,6 @@ def generate_30(
     latent_file_5_scale: float = 1.0,
     samples=1,
     progress=gr.Progress(track_tqdm=True)  # Add progress as a keyword argument
-    
 ):
     global captioner_2
     captioner2=captioner_2
@@ -443,7 +464,6 @@ def generate_30(
         print(new_prompt)
         print("-- FINAL PROMPT --")
         print("-- ------------ --")
-        
         #global model
         #global txt_tokenizer
         #del model