Upload PROGRESS.md with huggingface_hub

PROGRESS.md

@@ -24,14 +24,21 @@ Multi-agent Text2SQL pipeline reaching **pass@8 ≥ 67%** (BIRD-dev, EX). Hard c
 
 ### Best result so far (BIRD-dev, K=8)
 
-| Config | N | pass@8 (recall, oracle) | Selector pick | Notes |
+| Config | N | pass@8 (recall, oracle) | Trained selector (real, no leak) | Notes |
 |---|---|---|---|---|
-| 1-stage iter-2 uniform-temp | 1524 | 64.96% | 64.96%* | * selector currently leaks gold label — overstates |
-| 1-stage iter-2 mixed-temp (0.5/0.7/0.9/1.1) | 1525 | **65.38%** | 65.38%* | best vanilla planner-only |
-| 3-stage iter-2 + 0.6B v_s/v_c + ORPO fixer iter-1 | 1525 | 65.05% | 65.05%* | V+F neutral, 0 rescues |
-| 3-stage iter-2 + 0.5B v_s/v_c + replanner-fixer iter-2 | 1534 (in flight) | pending | pending | current eval, ETA 30 min |
+| 1-stage iter-2 uniform-temp | 1524 | 64.96% | — | leak-pre-patch number was 64.96 |
+| 1-stage iter-2 mixed-temp (0.5/0.7/0.9/1.1) | 1525 | **65.38%** | — | best recall |
+| 3-stage iter-2 + 0.6B v_s/v_c + ORPO fixer iter-1 | 1525 | 65.05% | — | V+F neutral (0 rescues) |
+| 3-stage iter-2 + 0.5B v_s + 0.6B v_c + 0.5B **replanner**-fixer iter-2 | 1525 | 65.11% | **54.30%** | leak-free selector measurement; **fixer iter-2 +0.06pp vs iter-1** |
 
-(*) The selector apply path had a gold-label leak (`is_planner_correct` used in prompt); patched in `compute_bestofn_with_selector.py` — re-run pending.
+**Critical finding (2026-05-13):** `compute_bestofn_with_selector.py` had `exec_response = "OK" if is_planner_correct else "Error / no rows"` — passed the gold-graded label to the selector → trivially matched oracle. **Patched** to use actual SQL execution result. Real trained-selector accuracy is **54.30%, NOT 65%**. Selector → oracle gap = 10.81pp.
+
+### Two headline gaps to close (target = headline EX ≥ 67%)
+
+| Gap | Current | Target | Gap | Lever |
+|---|---|---|---|---|
+| **Oracle pass@8** (recall) | 65.11% | ≥70% | +5pp | More diverse sampling / better fixer rescues |
+| **Selector → oracle** | 54.30 / 65.11 = 83% | ≥95% | +10pp | Selector retrain (likely overfit to old leaked prompts) |
 
 ### Per-model status