added a few more options

app.py

@@ -26,7 +26,6 @@ def build_transform(input_size):
     ])
     return transform
 
-
 def find_closest_aspect_ratio(aspect_ratio, target_ratios, width, height, image_size):
     best_ratio_diff = float('inf')
     best_ratio = (1, 1)
@@ -41,6 +40,7 @@ def find_closest_aspect_ratio(aspect_ratio, target_ratios, width, height, image_
             if area > 0.5 * image_size * image_size * ratio[0] * ratio[1]:
                 best_ratio = ratio
     return best_ratio
+
 def dynamic_preprocess(image, min_num=1, max_num=6, image_size=448, use_thumbnail=False):
     orig_width, orig_height = image.size
     aspect_ratio = orig_width / orig_height
@@ -79,7 +79,6 @@ def dynamic_preprocess(image, min_num=1, max_num=6, image_size=448, use_thumbnai
         processed_images.append(thumbnail_img)
     return processed_images, target_aspect_ratio
 
-
 def dynamic_preprocess2(image, min_num=1, max_num=6, image_size=448, use_thumbnail=False, prior_aspect_ratio=None):
     orig_width, orig_height = image.size
     aspect_ratio = orig_width / orig_height
@@ -175,7 +174,7 @@ tokenizer.eos_token = "<|end|>"
 model.generation_config.pad_token_id = tokenizer.pad_token_id
 
 
-def inference(image, prompt):
+def inference(image, prompt, temperature, top_p):
     # Check if both image and prompt are provided
     if image is None or prompt.strip() == "":
         return "Please provide both an image and a prompt."
@@ -188,6 +187,8 @@ def inference(image, prompt):
         num_beams=1,
         max_new_tokens=2048,
         do_sample=False,
+        temperature=temperature,
+        top_p=top_p,
     )
 
     # Generate the response
@@ -200,6 +201,7 @@ def inference(image, prompt):
 
     return response
 
+
 # Build the Gradio interface
 with gr.Blocks() as demo:
     gr.Markdown("H2O-Mississippi")
@@ -207,17 +209,29 @@ with gr.Blocks() as demo:
     with gr.Row():
         image_input = gr.Image(type="pil", label="Upload an Image")
         prompt_input = gr.Textbox(label="Enter your prompt here")
+        
+    with gr.Accordion('Parameters', open=False):
+        temperature_input = gr.Slider(minimum=0.0, maximum=1.0, step=0.01, default=1.0, label="Temperature")
+        top_p_input = gr.Slider(minimum=0.0, maximum=1.0, step=0.1, default=0.9, label="Top-p")
 
     response_output = gr.Textbox(label="Model Response")
 
     with gr.Row():
         submit_button = gr.Button("Submit")
+        regenerate_button = gr.Button("Regenerate")
         clear_button = gr.Button("Clear")
+        
 
     # When the submit button is clicked, call the inference function
     submit_button.click(
         fn=inference, 
-        inputs=[image_input, prompt_input], 
+        inputs=[image_input, prompt_input, temperature_input, top_p_input], 
+        outputs=response_output
+    )
+    # When the regenerate button is clicked, re-run the last inference
+    regenerate_button.click(
+        fn=inference,
+        inputs=[image_input, prompt_input, temperature_input, top_p_input],
         outputs=response_output
     )