step29211: step 29211, 1869504 examples, eval {'val_loss': 1.3418, 'val_wp_mae': 0.0845, 'top_move_acc': 0.29333333333333333, 'step': 28000}

README.md

@@ -9,7 +9,7 @@ tags: [chess, laya, decision-model]
 for each legal move, a Laya `score` question predicts the side to move's win chance over 10 levels.
 The engine plays the move with the highest expected win chance (no search).
 
-- checkpoint: `step28087` (step 28087, 1,797,568 training examples)
+- checkpoint: `step29211` (step 29211, 1,869,504 training examples)
 - latest eval: `{'val_loss': 1.3418, 'val_wp_mae': 0.0845, 'top_move_acc': 0.29333333333333333, 'step': 28000}`
 - encoding / levels / prompt template: see `chess_meta.json`
 - code: github.com/devroopsaha744/LayaChess

chess_meta.json

@@ -1,8 +1,8 @@
 {
- "tag": "step28087",
+ "tag": "step29211",
  "final": false,
- "step": 28087,
- "consumed": 1797568,
+ "step": 29211,
+ "consumed": 1869504,
  "n_levels": 10,
  "crit": [
   "0-10%",
@@ -45,100 +45,6 @@
   "SEED": 0
  },
  "history": [
-  {
-   "step": 25700,
-   "loss": 1.3591
-  },
-  {
-   "step": 25750,
-   "loss": 1.3538
-  },
-  {
-   "step": 25800,
-   "loss": 1.3595
-  },
-  {
-   "step": 25850,
-   "loss": 1.3606
-  },
-  {
-   "step": 25900,
-   "loss": 1.3513
-  },
-  {
-   "step": 25950,
-   "loss": 1.3739
-  },
-  {
-   "step": 26000,
-   "loss": 1.3455
-  },
-  {
-   "val_loss": 1.3545,
-   "val_wp_mae": 0.0855,
-   "top_move_acc": 0.25333333333333335,
-   "step": 26000
-  },
-  {
-   "step": 26050,
-   "loss": 1.3481
-  },
-  {
-   "step": 26100,
-   "loss": 1.3556
-  },
-  {
-   "step": 26150,
-   "loss": 1.3564
-  },
-  {
-   "step": 26200,
-   "loss": 1.3745
-  },
-  {
-   "step": 26250,
-   "loss": 1.3554
-  },
-  {
-   "step": 26300,
-   "loss": 1.3706
-  },
-  {
-   "step": 26350,
-   "loss": 1.3639
-  },
-  {
-   "step": 26400,
-   "loss": 1.3593
-  },
-  {
-   "step": 26450,
-   "loss": 1.355
-  },
-  {
-   "step": 26500,
-   "loss": 1.3479
-  },
-  {
-   "step": 26550,
-   "loss": 1.3517
-  },
-  {
-   "step": 26600,
-   "loss": 1.3729
-  },
-  {
-   "step": 26650,
-   "loss": 1.3756
-  },
-  {
-   "step": 26700,
-   "loss": 1.3366
-  },
-  {
-   "step": 26750,
-   "loss": 1.3626
-  },
   {
    "step": 26800,
    "loss": 1.3675
@@ -248,6 +154,98 @@
   {
    "step": 28050,
    "loss": 1.3337
+  },
+  {
+   "step": 28100,
+   "loss": 1.3419
+  },
+  {
+   "step": 28150,
+   "loss": 1.3524
+  },
+  {
+   "step": 28200,
+   "loss": 1.3345
+  },
+  {
+   "step": 28250,
+   "loss": 1.3368
+  },
+  {
+   "step": 28300,
+   "loss": 1.3448
+  },
+  {
+   "step": 28350,
+   "loss": 1.3436
+  },
+  {
+   "step": 28400,
+   "loss": 1.3348
+  },
+  {
+   "step": 28450,
+   "loss": 1.3484
+  },
+  {
+   "step": 28500,
+   "loss": 1.3463
+  },
+  {
+   "step": 28550,
+   "loss": 1.3628
+  },
+  {
+   "step": 28600,
+   "loss": 1.3467
+  },
+  {
+   "step": 28650,
+   "loss": 1.3298
+  },
+  {
+   "step": 28700,
+   "loss": 1.3595
+  },
+  {
+   "step": 28750,
+   "loss": 1.329
+  },
+  {
+   "step": 28800,
+   "loss": 1.3442
+  },
+  {
+   "step": 28850,
+   "loss": 1.3375
+  },
+  {
+   "step": 28900,
+   "loss": 1.3471
+  },
+  {
+   "step": 28950,
+   "loss": 1.3458
+  },
+  {
+   "step": 29000,
+   "loss": 1.3656
+  },
+  {
+   "step": 29050,
+   "loss": 1.3391
+  },
+  {
+   "step": 29100,
+   "loss": 1.356
+  },
+  {
+   "step": 29150,
+   "loss": 1.3411
+  },
+  {
+   "step": 29200,
+   "loss": 1.3312
   }
  ]
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2f2dd82e5911e06eb7a192e0a6d8bf79d1839041c8c9844bbdc360d2a0066533
+oid sha256:fc2e0665c556a70e9eef44829b23c999a0800836fc4a7d3c8595c995bb45dada
 size 842609220