refined layout

app.py

@@ -258,6 +258,7 @@ def init_clip_mlang():
     tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
     return tokenizer, clip
 
+
 @st.experimental_singleton(show_spinner=False)
 def init_clip_vanilla():
     """ Initialize CLIP Model
@@ -297,11 +298,13 @@ def prompt2vec_mlang(prompt: str, tokenizer, clip):
     xq = out.squeeze(0).cpu().detach().numpy().tolist()
     return xq
 
+
 def prompt2vec_vanilla(prompt: str, tokenizer, clip):
     inputs = tokenizer(prompt, return_tensors='pt')
     out = clip.get_text_features(**inputs)
     xq = out.squeeze(0).cpu().detach().numpy().tolist()
-    return xq 
+    return xq
+
 
 st.markdown("""
 <link
@@ -345,7 +348,7 @@ text_model_map = {
     'English': {'Vanilla CLIP': [prompt2vec_vanilla, ],
                 'CLIP finetuned on RSICD': [prompt2vec_vanilla, ],
                 }
-    }
+}
 
 
 with st.spinner("Connecting DB..."):
@@ -354,9 +357,11 @@ with st.spinner("Connecting DB..."):
 with st.spinner("Loading Models..."):
     # Initialize CLIP model
     if 'xq' not in st.session_state:
-        text_model_map['Multi Lingual']['Vanilla CLIP'].append(init_clip_mlang())
+        text_model_map['Multi Lingual']['Vanilla CLIP'].append(
+            init_clip_mlang())
         text_model_map['English']['Vanilla CLIP'].append(init_clip_vanilla())
-        text_model_map['English']['CLIP finetuned on RSICD'].append(init_clip_rsicd())
+        text_model_map['English']['CLIP finetuned on RSICD'].append(
+            init_clip_rsicd())
         st.session_state.query_num = 0
 
 if 'xq' not in st.session_state:
@@ -372,30 +377,34 @@ if 'xq' not in st.session_state:
             del st.session_state.prompt
         st.title("Visual Dataset Explorer")
         start = [st.empty(), st.empty(), st.empty(), st.empty(),
-                 st.empty(), st.empty(), st.empty()]
+                 st.empty(), st.empty(), st.empty(), st.empty()]
         start[0].info(msg)
         start_col = start[1].columns(3)
-        st.session_state.db_name_ref = start_col[0].selectbox("Select Database:", list(db_name_map.keys()))
-        st.session_state.lang = start_col[1].selectbox("Select Language:", list(text_model_map.keys()))
-        st.session_state.feat_name = start_col[2].selectbox("Select Image Feature:", 
+        st.session_state.db_name_ref = start_col[0].selectbox(
+            "Select Database:", list(db_name_map.keys()))
+        st.session_state.lang = start_col[1].selectbox(
+            "Select Language:", list(text_model_map.keys()))
+        st.session_state.feat_name = start_col[2].selectbox("Select Image Feature:",
                                                             list(text_model_map[st.session_state.lang].keys()))
         if st.session_state.db_name_ref == "RSICD: Remote Sensing Images 11K":
-            st.warning('If you are searching for Remote Sensing Images, \
+            start[2].warning('If you are searching for Remote Sensing Images, \
                         try to use prompt "An aerial photograph of <your-real-query>" \
                         to obtain best search experience!')
-        prompt = start[2].text_input(
-            "Prompt:", value="", placeholder="Examples: playing corgi, 女人举着雨伞, mouette volant au-dessus de la mer, ガラスの花瓶の花 ...")
         if len(prompt) > 0:
             st.session_state.prompt = prompt.replace(' ', '_')
-        start[3].markdown(
+        start[4].markdown(
             '<p style="color:gray;"> Don\'t know what to search? Try <b>Random</b>!</p>\
             <p>🌟 We also support multi-language search. Type any language you know to search! ⌨️ </p>',
             unsafe_allow_html=True)
-        upld_model = start[5].file_uploader(
+        upld_model = start[6].file_uploader(
             "Or you can upload your previous run!", type='onnx')
-        upld_btn = start[6].button(
-            "Used Loaded Weights", disabled=upld_model is None)
-        with start[4]:
+        upld_btn = start[7].button(
+            "Use Loaded Weights", disabled=upld_model is None)
+        prompt = start[3].text_input(
+            "Prompt:",
+            value="An aerial photograph of "if st.session_state.db_name_ref == "RSICD: Remote Sensing Images 11K" else "",
+            placeholder="Examples: playing corgi, 女人举着雨伞, mouette volant au-dessus de la mer, ガラスの花瓶の花 ...",)
+        with start[5]:
             col = st.columns(8)
             has_no_prompt = (len(prompt) == 0 and upld_model is None)
             prompt_xq = col[6].button("Prompt", disabled=len(prompt) == 0)
@@ -418,7 +427,8 @@ if 'xq' not in st.session_state:
             assert len(weights) == 1
             xq = numpy_helper.to_array(weights[0]).tolist()
             assert len(xq) == DIMS
-            st.session_state.prompt = upld_model.name.split(".onnx")[0].replace(' ', '_')
+            st.session_state.prompt = upld_model.name.split(".onnx")[
+                0].replace(' ', '_')
         else:
             print(f"Input prompt is {prompt}")
             # Tokenize the vectors