feature: allow user to specify num rows, cols in grid

most_relevant_part.py

@@ -41,9 +41,13 @@ def app(model_name):
         """
         Given a piece of text, the CLIP model finds the part of an image that best explains the text.
         To try it out, you can
-        1) Upload an image
-        2) Explain a part of the image in text
-        Which will yield the most relevant image tile from a 3x3 grid of the image 
+
+        1. Upload an image
+        2. Explain a part of the image in text
+
+        which will yield the most relevant image tile from a grid of the image. You can specify how
+        granular you want to be with your search by specifying the number of rows and columns that
+        make up the image grid.
         """
     )
 
@@ -56,6 +60,8 @@ def app(model_name):
         "Enter query to find most relevant part of image ",
         value="이건 서울의 경복궁 사진이다.",
     )
+    num_rows = st.slider("Number of rows", min_value=1, max_value=5, value=3, step=1)
+    num_cols = st.slider("Number of columns", min_value=1, max_value=5, value=3, step=1)
 
     if st.button("질문 (Query)"):
         if not any([query1, query2]):
@@ -67,7 +73,7 @@ def app(model_name):
             image = Image.open(image_data)
             st.image(image)
 
-            images = split_image(image)
+            images = split_image(image, num_rows, num_cols)
 
             inputs = processor(
                 text=captions, images=images, return_tensors="jax", padding=True