feat: descriptive labels for Pipeline and Question Style dropdowns

app.py

@@ -13664,12 +13664,26 @@ def create_ui():
                 custom_url = gr.Textbox(label="Endpoint URL", placeholder="https://...", visible=False)
 
                 pipeline_type = gr.Dropdown(
-                    choices=["Factual Q&A", "RPToolkit (Roleplay)", "Representation Variation", "RAG Training", "Correction"],
+                    choices=[
+                        ("Factual Q&A: Educational questions with validation", "Factual Q&A"),
+                        ("RPToolkit: Character cards + roleplay stories", "RPToolkit (Roleplay)"),
+                        ("Representation Variation: Rewrite as list, JSON, essay...", "Representation Variation"),
+                        ("RAG Training: AI answers using provided context chunks", "RAG Training"),
+                        ("Correction: AI gives wrong answer, then self-corrects", "Correction"),
+                    ],
                     value="Factual Q&A",
                     label="Pipeline",
                 )
                 question_style = gr.Dropdown(
-                    choices=["Standard", "Comparison", "Followup", "Hallucination", "Negative", "Open-ended", "Vague"],
+                    choices=[
+                        ("Standard: Normal educational Q&A", "Standard"),
+                        ("Comparison: Compare X vs Y", "Comparison"),
+                        ("Followup: Build on previous answers", "Followup"),
+                        ("Hallucination: Nonexistent things (trains 'I don't know')", "Hallucination"),
+                        ("Negative: Trick questions with flawed premises", "Negative"),
+                        ("Open-ended: Broad analytical discussion", "Open-ended"),
+                        ("Vague: Ambiguous questions, AI clarifies first", "Vague"),
+                    ],
                     value="Standard",
                     label="Question Style (Factual only)",
                     visible=True,