opening expert model card

README.md

@@ -18,8 +18,8 @@ finetuned on `≥26`-piece positions (one-hot best first move) from
 [`Lichess/chess-position-evaluations`](https://huggingface.co/datasets/Lichess/chess-position-evaluations)
 via [`avewright/lichess-opening-bestline`](https://huggingface.co/datasets/avewright/lichess-opening-bestline).
 
-This file is **`latest.pt` at opening-FT step 5400** (2026-09-14 13:58 UTC).
-Train loss ~1.7684. Frozen holdout hard CE ~1.4278.
+This file is **`latest.pt` at opening-FT step 5550** (2026-09-14 14:01 UTC).
+Train loss ~1.7786. Frozen holdout hard CE ~1.4295.
 
 Not the generalist incumbent, the puzzle expert, the Syzygy expert, or the endgame expert.
 
@@ -28,12 +28,12 @@ Not the generalist incumbent, the puzzle expert, the Syzygy expert, or the endga
 - Warm start: public 99M `latest.pt` (weights only), then full resume.
 - Split: position-hash 80/20 (seed 277). Frozen piece-stratified val 8192.
 - One-hot PV1 (`soft_alpha=0`). Pieces 26–32.
-- Polar-NorMuon, bs=528. Best disk ckpt at upload: step 5400.
+- Polar-NorMuon, bs=528. Best disk ckpt at upload: step 5550.
 
 ## Files
 
 - `latest.pt`
-- `step_005400.pt`
+- `step_005550.pt`
 - `model_config.json`
 - `train.log`
 - `pack.json`