Update app.py

app.py

@@ -63,7 +63,7 @@ with gr.Blocks() as demo:
     gr.Markdown("""# Enter anything and press Enter""")
     
     topic_input = gr.Textbox(
-        label="Enter anything"
+        label="Enter here"
     )
 
     output = gr.Textbox(
@@ -77,16 +77,18 @@ with gr.Blocks() as demo:
     )
 
     gr.Markdown("""
-# How it works:
+# Workflow:
 
-- **Dataset: 500 news plus 3 news of myown**
-- **Embedding model: bge-small-en-v1.5**
-- **FAISS**
-- **Sum(above)=Vector database and retrieval**
+- **Data pipeline. Get 500 news from ag_news dataset, and truncate each news to 130 characters**
+- **Model BAAI/bge-v1.5 embeds the dataset and user input into vectors**
+- **FAISS computes the semantic similarity between user-input-vector and each vector else**
+- **Retrieve the top3 most relevant news based on semantic similarity, and show them to user**
+- **Added custom news to test the retrieved results and make sure everything works**
+- **Deploy UI**
 
 
 
-**To add a little fun, I made 3 news.:**
+**To test and have some fun, I made 3 news.:**
 - **ID500: Basketball player LeBron James jioned 76er with a 5m/year contract.**
 - **ID501: Spain won 2026 Football World Cup, defeating Argentina at final.**
 - **ID502: Baseball player Shohei Ohtani plays as a hitter and as a pitcher, a rare two-way player.**