Update app.py

app.py

@@ -41,6 +41,7 @@ POOLING_MAP = {
     "Mean-Max (Combined)": "mean-max"
 }
 
+# --- CACHE THE NEURAL NETWORK ---
 @st.cache_resource
 def load_embedder(model_name, pool_strat):
     word_emb = models.Transformer(model_name)
@@ -50,7 +51,9 @@ def load_embedder(model_name, pool_strat):
         pooling_mode_max_tokens=("max" in pool_strat),
         pooling_mode_cls_token=("cls" in pool_strat)
     )
-    return SentenceTransformer(modules=[word_emb, pool_model])
+    
+    # HARDCODED MATH: Permanently forces L2 Normalization so distance math works flawlessly
+    return SentenceTransformer(modules=[word_emb, pool_model, models.Normalize()])
 
 # ==========================================
 # 2. THE GUIDED UI
@@ -108,9 +111,9 @@ with st.expander("2️⃣ Geometry Layer (Dimensionality Reduction)", expanded=T
 with st.expander("3️⃣ Clustering Layer (Grouping)", expanded=True):
     st.markdown("*💡 **BERTopic Default:** Uses `HDBSCAN` exclusively (which crashes on tiny datasets).*")
     st.markdown("""
-    *Our Engine mathematically draws boundaries around similar sentences using a smart-routing system:*
-    * **Primary Engine (HDBSCAN):** Runs on datasets $\ge$ 15 sentences. Automatically filters outliers and finds dense semantic clouds. *(Defaults: min_cluster_size=10)*
-    * **Fallback Engine (K-Means):** Runs on datasets $<$ 15 sentences. Forces all sentences into buckets to prevent math crashes on tiny text samples.
+    *The model mathematically draws boundaries around similar sentences. *
+    * **Primary clustering algorithm (HDBSCAN):** Runs on datasets $\ge$ 15 sentences. Automatically filters outliers and finds dense semantic clouds. *(Defaults: min_cluster_size=10)*
+    * **Fallback clustering algorithm (K-Means):** Runs on datasets $<$ 15 sentences. Forces all sentences into buckets to prevent math crashes on tiny text samples.
     """)
 
 with st.expander("4️⃣ Vocabulary Layer (Vectorization)", expanded=True):