step7827: step 7827, 500928 examples, eval {'val_loss': 1.5209, 'val_wp_mae': 0.1171, 'top_move_acc': 0.19666666666666666, 'step': 6000}

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: `step6674` (step 6674, 427,136 training examples)
+- checkpoint: `step7827` (step 7827, 500,928 training examples)
 - latest eval: `{'val_loss': 1.5209, 'val_wp_mae': 0.1171, 'top_move_acc': 0.19666666666666666, 'step': 6000}`
 - encoding / levels / prompt template: see `chess_meta.json`
 - code: github.com/devroopsaha744/LayaChess

chess_meta.json

@@ -1,8 +1,8 @@
 {
- "tag": "step6674",
+ "tag": "step7827",
  "final": false,
- "step": 6674,
- "consumed": 427136,
+ "step": 7827,
+ "consumed": 500928,
  "n_levels": 10,
  "crit": [
   "0-10%",
@@ -45,98 +45,6 @@
   "SEED": 0
  },
  "history": [
-  {
-   "step": 4250,
-   "loss": 1.5731
-  },
-  {
-   "step": 4300,
-   "loss": 1.6093
-  },
-  {
-   "step": 4350,
-   "loss": 1.5644
-  },
-  {
-   "step": 4400,
-   "loss": 1.5813
-  },
-  {
-   "step": 4450,
-   "loss": 1.586
-  },
-  {
-   "step": 4500,
-   "loss": 1.5862
-  },
-  {
-   "step": 4550,
-   "loss": 1.5768
-  },
-  {
-   "step": 4600,
-   "loss": 1.5645
-  },
-  {
-   "step": 4650,
-   "loss": 1.5599
-  },
-  {
-   "step": 4700,
-   "loss": 1.5755
-  },
-  {
-   "step": 4750,
-   "loss": 1.5572
-  },
-  {
-   "step": 4800,
-   "loss": 1.5459
-  },
-  {
-   "step": 4850,
-   "loss": 1.581
-  },
-  {
-   "step": 4900,
-   "loss": 1.5761
-  },
-  {
-   "step": 4950,
-   "loss": 1.5665
-  },
-  {
-   "step": 5000,
-   "loss": 1.5482
-  },
-  {
-   "step": 5050,
-   "loss": 1.5623
-  },
-  {
-   "step": 5100,
-   "loss": 1.561
-  },
-  {
-   "step": 5150,
-   "loss": 1.5494
-  },
-  {
-   "step": 5200,
-   "loss": 1.5457
-  },
-  {
-   "step": 5250,
-   "loss": 1.5358
-  },
-  {
-   "step": 5300,
-   "loss": 1.5552
-  },
-  {
-   "step": 5350,
-   "loss": 1.5662
-  },
   {
    "step": 5400,
    "loss": 1.5546
@@ -246,6 +154,98 @@
   {
    "step": 6650,
    "loss": 1.5443
+  },
+  {
+   "step": 6700,
+   "loss": 1.5239
+  },
+  {
+   "step": 6750,
+   "loss": 1.4953
+  },
+  {
+   "step": 6800,
+   "loss": 1.539
+  },
+  {
+   "step": 6850,
+   "loss": 1.5144
+  },
+  {
+   "step": 6900,
+   "loss": 1.5075
+  },
+  {
+   "step": 6950,
+   "loss": 1.5172
+  },
+  {
+   "step": 7000,
+   "loss": 1.5296
+  },
+  {
+   "step": 7050,
+   "loss": 1.4984
+  },
+  {
+   "step": 7100,
+   "loss": 1.5098
+  },
+  {
+   "step": 7150,
+   "loss": 1.5205
+  },
+  {
+   "step": 7200,
+   "loss": 1.5052
+  },
+  {
+   "step": 7250,
+   "loss": 1.5121
+  },
+  {
+   "step": 7300,
+   "loss": 1.5161
+  },
+  {
+   "step": 7350,
+   "loss": 1.5277
+  },
+  {
+   "step": 7400,
+   "loss": 1.5275
+  },
+  {
+   "step": 7450,
+   "loss": 1.4838
+  },
+  {
+   "step": 7500,
+   "loss": 1.4762
+  },
+  {
+   "step": 7550,
+   "loss": 1.4789
+  },
+  {
+   "step": 7600,
+   "loss": 1.4958
+  },
+  {
+   "step": 7650,
+   "loss": 1.5131
+  },
+  {
+   "step": 7700,
+   "loss": 1.4842
+  },
+  {
+   "step": 7750,
+   "loss": 1.4633
+  },
+  {
+   "step": 7800,
+   "loss": 1.4677
   }
  ]
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4a46203a3ce1ed9b17ca8d2dab017d924ff9918d65917aa262bab86ab8285e92
+oid sha256:bd94f798940f4d0f94c368a7c307f0ce7659d8a234cb27fff6cc9632c46cce6d
 size 842609220