Refresh clean run evidence

README.md

@@ -114,8 +114,10 @@ Current platform evidence:
 | 12k source preflight dataset ID | `13b1c92a-a96b-4ce7-810b-4368ad6aa234` |
 | 12k source-column audit | Passed: 12,000/12,000 rows preserved `Final: {...}` |
 | AutoScientist recommendation | `google/gemma-3-4b-it`, 1-epoch LoRA |
-| Active fallback AutoScientist run | `43b5486d-0bfc-4e9f-869e-a9892a679386` on `meta-llama/Llama-3.2-3B-Instruct` |
-| Clean 8k source preflight | `3103c6ac-7d61-4271-af62-41cb023de85e`; recommended `mistralai/Mistral-7B-Instruct-v0.2` |
+| Diagnostic 8k Llama run | `43b5486d-0bfc-4e9f-869e-a9892a679386`; best win rate 49.48%; not final |
+| Clean 8k candidate run | `7ea2c71b-94d7-472d-8288-795ab5e0a2c3` on `mistralai/Mistral-7B-Instruct-v0.2` |
+| Clean 8k candidate dataset | `3103c6ac-7d61-4271-af62-41cb023de85e` |
+| Clean 8k latest metric | Running, 1/5 iterations, best win rate 51.96%, checkpoint pending |
 
 Fill after final AutoScientist run:
 
@@ -124,9 +126,9 @@ Fill after final AutoScientist run:
 | Adaptive Data grade before | Pending |
 | Adaptive Data grade after | Pending |
 | Adaptive Data quality improvement | Pending |
-| AutoScientist best win rate | Pending checkpoint |
-| Domain augmentation rows | 8,000 source rows in active fallback |
-| General augmentation rows | 0 in active fallback |
+| AutoScientist best win rate | 51.96% interim; final pending completion |
+| Domain augmentation rows | 8,000 source rows in clean candidate |
+| General augmentation rows | 0 in clean candidate |
 
 ## Limitations