Upload app.py

app.py

@@ -318,9 +318,9 @@ Answer:"""
     return prompt
 
 # --- Streamlit App UI ---
-st.title("📚 Ask the Library Guides (Dataset Embed + HF Gen)")
+st.title("📚 Experimental LibGuides RAG Demo") # Revised title
 
-# User input (only proceed if collection is ready)
+# User input (only proceed if collection loaded)
 if collection:
     query = st.text_area("Enter your question:", height=100)
 else:
@@ -529,17 +529,18 @@ if collection and st.button("Ask"):
                 st.info("No specific context was retrieved from the guides to answer this question.")
 
 # Add instructions or footer
-st.sidebar.header("How to Use")
+st.sidebar.header("About This Demo")
 st.sidebar.info(
-    "1. Ensure your `HUGGING_FACE_HUB_TOKEN` is correctly set as a Space secret (`HF_TOKEN`) or in the `.env` file.\n"
-    f"2. The app will load pre-computed embeddings from the HF Dataset (`{HF_DATASET_ID}`).\n"
-    "   (Ensure the dataset was created correctly using `export_chroma_to_parquet.py` and `upload_dataset_to_hf.py`)\n"
-    "3. Enter your question in the text area.\n"
-    "4. Click 'Ask'."
+    "This is an experimental RAG demo using CUNY GC LibGuides content.\n\n"
+    "1. It loads pre-computed embeddings from a Hugging Face Dataset.\n"
+    "2. It embeds your query locally.\n"
+    "3. It uses the Hugging Face Inference API for LLM tasks (routing, generation).\n"
+    "4. Requires a `HUGGING_FACE_HUB_TOKEN` set as a Space secret (`HF_TOKEN`) or in `.env`."
 )
-st.sidebar.header("Configuration")
-st.sidebar.markdown(f"**Embedding:** Pre-computed (`{LOCAL_EMBEDDING_MODEL}` loaded from HF Dataset)")
-st.sidebar.markdown(f"**LLM (HF API):** `{HF_GENERATION_MODEL}`")
-st.sidebar.markdown(f"**ChromaDB Collection:** `{COLLECTION_NAME}` (In-Memory)")
+st.sidebar.header("Configuration Used")
+st.sidebar.markdown(f"**Embedding Source:** HF Dataset (`{HF_DATASET_ID}`)")
+st.sidebar.markdown(f"**Query Embedding Model:** Local (`{LOCAL_EMBEDDING_MODEL}`)")
+st.sidebar.markdown(f"**Generation LLM (HF API):** `{HF_GENERATION_MODEL}`")
+st.sidebar.markdown(f"**Vector Store:** ChromaDB (In-Memory)")
 st.sidebar.markdown(f"**Retrieval Mode:** Vector Search Only")
 st.sidebar.markdown(f"**Final Unique Chunks:** `{TOP_K}` (from initial `{INITIAL_N_RESULTS}` vector search)")