Update app.py

app.py

@@ -245,13 +245,13 @@ def uploadNote(prompt,num_inference_steps,guidance_scale,timestamp):
 
 def captioning(img):
     prompts_array = [
-        "Adjectives describing this scene are:",
+       # "Adjectives describing this scene are:",
         "The color scheme of this image is",
-        "This scene could be described in detail as",
-        "The characters in this scene are",
+      #  "This scene could be described in detail as",
+      #  "The characters in this scene are",
         "The larger details in this scene include",
         "The smaller details in this scene include",
-        "The feeling this scene seems like",
+     #   "The feeling this scene seems like",
         "The setting of this scene must be located", 
     # Add more prompts here
     ]
@@ -264,8 +264,8 @@ def captioning(img):
         **inputsa,
         do_sample=False,
         num_beams=5,
-        max_length=128,
-        min_length=1,
+        max_length=96,
+        #min_length=1,
         top_p=0.9,
         repetition_penalty=1.5,
         length_penalty=1.0,
@@ -282,8 +282,8 @@ def captioning(img):
             **inputs,
             do_sample=False,
             num_beams=5,
-            max_length=128,
-            min_length=1,
+            max_length=96,
+            #min_length=1,
             top_p=0.9,
             repetition_penalty=1.5,
             length_penalty=1.0,
@@ -293,6 +293,14 @@ def captioning(img):
         response_text = generated_text.replace(prompt, "").strip() #Or could try .split(prompt, 1)[-1].strip()
         output_prompt.append(response_text)
         print(f"{response_text}\n") # Print only the response text
+        
+            # Continue conversation:
+    inputf = processor5(images=img, text=generated_text + 'So therefore', return_tensors="pt").to('cuda')
+    generated_ids = model5.generate(**inputf, max_length=96)
+    generated_text = processor5.batch_decode(generated_ids, skip_special_tokens=True)[0].strip()
+    response_text = generated_text.replace(generated_text, "").strip()
+    print(response_text)
+    output_prompt.append(response_text)
     print(output_prompt)
     return output_prompt
     
@@ -386,6 +394,7 @@ def generate_30(
     samples=1,
     progress=gr.Progress(track_tqdm=True)  # Add progress as a keyword argument
 ):
+    image_paths=[]
     seed = random.randint(0, MAX_SEED)
     generator = torch.Generator(device='cuda').manual_seed(seed)
     if latent_file is not None:  # Check if a latent file is provided
@@ -511,6 +520,7 @@ def generate_60(
     samples=1,
     progress=gr.Progress(track_tqdm=True)  # Add progress as a keyword argument
 ):
+    image_paths=[]
     seed = random.randint(0, MAX_SEED)
     generator = torch.Generator(device='cuda').manual_seed(seed)
     if latent_file is not None:  # Check if a latent file is provided
@@ -636,6 +646,7 @@ def generate_90(
     samples=1,
     progress=gr.Progress(track_tqdm=True)  # Add progress as a keyword argument
 ):
+    image_paths=[]
     seed = random.randint(0, MAX_SEED)
     generator = torch.Generator(device='cuda').manual_seed(seed)
     if latent_file is not None:  # Check if a latent file is provided