Caching get_hf_api(), cached term_category_pairs computation, cached load_users(). This should increase reactivity alot.

src/streamlit_app.py

@@ -30,13 +30,18 @@ if not HF_TOKEN:
     st.error("❌ **Configuration Error**: HF_TOKEN is not set. Please configure it in Hugging Face Spaces settings (Variables and secrets).")
     st.stop()
 
-# Initialize HF API - raise exception if initialization fails
-try:
-    login(token=HF_TOKEN)
-    hf_api = HfApi(token=HF_TOKEN)
-except Exception as e:
-    st.error(f"❌ **Error initializing Hugging Face API**: {str(e)}")
-    st.stop()
+@st.cache_resource
+def get_hf_api():
+    """Get cached Hugging Face API client - only initializes once per session"""
+    try:
+        login(token=HF_TOKEN)
+        return HfApi(token=HF_TOKEN)
+    except Exception as e:
+        st.error(f"❌ **Error initializing Hugging Face API**: {str(e)}")
+        st.stop()
+
+# Initialize HF API - cached to avoid re-initialization on every rerun
+hf_api = get_hf_api()
 
 @st.cache_data
 def load_grading_template():
@@ -132,6 +137,7 @@ def hash_password(password):
     """Hash a password using SHA256"""
     return hashlib.sha256(password.encode()).hexdigest()
 
+@st.cache_data
 def load_users():
     """Load user credentials from Hugging Face Dataset"""
     try:
@@ -162,6 +168,10 @@ def save_users(users):
             token=HF_TOKEN
         )
         os.unlink(temp_path)
+        
+        # Clear cache for users to ensure fresh data on next load
+        load_users.clear()
+        
         return True
     except Exception as e:
         st.error(f"❌ **Error saving users to Hugging Face Dataset**: {str(e)}")
@@ -567,11 +577,16 @@ class Term:
         return None
 
 
+@st.cache_data
+def get_term_category_pairs(df):
+    """Get filtered term-category pairs, cached to avoid recomputation on every rerun"""
+    # Filter out categories that have no subcategories after filtering Unknown answers
+    all_pairs = df[['term', 'category']].drop_duplicates().sort_values(['term', 'category']).values.tolist()
+    return [(term, category) for term, category in all_pairs 
+            if category_has_subcategories(term, category, df)]
+
 # Create a list of unique (term, category) pairs for navigation
-# Filter out categories that have no subcategories after filtering Unknown answers
-all_pairs = df[['term', 'category']].drop_duplicates().sort_values(['term', 'category']).values.tolist()
-term_category_pairs = [(term, category) for term, category in all_pairs 
-                       if category_has_subcategories(term, category, df)]
+term_category_pairs = get_term_category_pairs(df)
 total_pairs = len(term_category_pairs)
 
 # Cache for Term instances