cache tokenizer and specify ITALIAN query string :D

app.py

@@ -25,6 +25,11 @@ def get_model():
     return FlaxHybridCLIP.from_pretrained("clip-italian/clip-italian")
 
 
+@st.cache
+def get_tokenizer():
+    return AutoTokenizer.from_pretrained("dbmdz/bert-base-italian-xxl-uncased", cache_dir=None, use_fast=True)
+
+
 @st.cache
 def download_images():
     # from sentence_transformers import SentenceTransformer, util
@@ -58,7 +63,7 @@ def get_image_features():
 os.environ["TOKENIZERS_PARALLELISM"] = "false"
 
 
-query = st.text_input("Insert a query text")
+query = st.text_input("Insert an italian query text here...")
 if query:
 
     with st.spinner("Computing in progress..."):
@@ -68,9 +73,7 @@ if query:
         image_features = get_image_features()
 
         model = get_model()
-        tokenizer = AutoTokenizer.from_pretrained(
-            "dbmdz/bert-base-italian-xxl-uncased", cache_dir=None, use_fast=True
-        )
+        tokenizer = get_tokenizer()
 
         image_size = model.config.vision_config.image_size