step6674: step 6674, 427136 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: `step5597` (step 5597, 358,208 training examples)
-- latest eval: `{'val_loss': 1.5506, 'val_wp_mae': 0.1205, 'top_move_acc': 0.19666666666666666, 'step': 4000}`
+- checkpoint: `step6674` (step 6674, 427,136 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": "step5597",
+ "tag": "step6674",
  "final": false,
- "step": 5597,
- "consumed": 358208,
+ "step": 6674,
+ "consumed": 427136,
  "n_levels": 10,
  "crit": [
   "0-10%",
@@ -45,100 +45,6 @@
   "SEED": 0
  },
  "history": [
-  {
-   "step": 3150,
-   "loss": 1.5848
-  },
-  {
-   "step": 3200,
-   "loss": 1.5735
-  },
-  {
-   "step": 3250,
-   "loss": 1.5577
-  },
-  {
-   "step": 3300,
-   "loss": 1.5671
-  },
-  {
-   "step": 3350,
-   "loss": 1.5548
-  },
-  {
-   "step": 3400,
-   "loss": 1.5687
-  },
-  {
-   "step": 3450,
-   "loss": 1.5736
-  },
-  {
-   "step": 3500,
-   "loss": 1.5565
-  },
-  {
-   "step": 3550,
-   "loss": 1.5267
-  },
-  {
-   "step": 3600,
-   "loss": 1.5511
-  },
-  {
-   "step": 3650,
-   "loss": 1.5528
-  },
-  {
-   "step": 3700,
-   "loss": 1.574
-  },
-  {
-   "step": 3750,
-   "loss": 1.5627
-  },
-  {
-   "step": 3800,
-   "loss": 1.5693
-  },
-  {
-   "step": 3850,
-   "loss": 1.5822
-  },
-  {
-   "step": 3900,
-   "loss": 1.5566
-  },
-  {
-   "step": 3950,
-   "loss": 1.5412
-  },
-  {
-   "step": 4000,
-   "loss": 1.5849
-  },
-  {
-   "val_loss": 1.5506,
-   "val_wp_mae": 0.1205,
-   "top_move_acc": 0.19666666666666666,
-   "step": 4000
-  },
-  {
-   "step": 4050,
-   "loss": 1.5778
-  },
-  {
-   "step": 4100,
-   "loss": 1.552
-  },
-  {
-   "step": 4150,
-   "loss": 1.5421
-  },
-  {
-   "step": 4200,
-   "loss": 1.5711
-  },
   {
    "step": 4250,
    "loss": 1.5731
@@ -246,6 +152,100 @@
   {
    "step": 5550,
    "loss": 1.5501
+  },
+  {
+   "step": 5600,
+   "loss": 1.5601
+  },
+  {
+   "step": 5650,
+   "loss": 1.5411
+  },
+  {
+   "step": 5700,
+   "loss": 1.5534
+  },
+  {
+   "step": 5750,
+   "loss": 1.5509
+  },
+  {
+   "step": 5800,
+   "loss": 1.5414
+  },
+  {
+   "step": 5850,
+   "loss": 1.5512
+  },
+  {
+   "step": 5900,
+   "loss": 1.522
+  },
+  {
+   "step": 5950,
+   "loss": 1.5126
+  },
+  {
+   "step": 6000,
+   "loss": 1.5183
+  },
+  {
+   "val_loss": 1.5209,
+   "val_wp_mae": 0.1171,
+   "top_move_acc": 0.19666666666666666,
+   "step": 6000
+  },
+  {
+   "step": 6050,
+   "loss": 1.5171
+  },
+  {
+   "step": 6100,
+   "loss": 1.5371
+  },
+  {
+   "step": 6150,
+   "loss": 1.5258
+  },
+  {
+   "step": 6200,
+   "loss": 1.5161
+  },
+  {
+   "step": 6250,
+   "loss": 1.517
+  },
+  {
+   "step": 6300,
+   "loss": 1.5615
+  },
+  {
+   "step": 6350,
+   "loss": 1.5175
+  },
+  {
+   "step": 6400,
+   "loss": 1.5343
+  },
+  {
+   "step": 6450,
+   "loss": 1.5301
+  },
+  {
+   "step": 6500,
+   "loss": 1.5242
+  },
+  {
+   "step": 6550,
+   "loss": 1.533
+  },
+  {
+   "step": 6600,
+   "loss": 1.5114
+  },
+  {
+   "step": 6650,
+   "loss": 1.5443
   }
  ]
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:715a820f87b63f81af6ef253d45f81bd9f6a3864202d920f7b183c791f89ad88
+oid sha256:4a46203a3ce1ed9b17ca8d2dab017d924ff9918d65917aa262bab86ab8285e92
 size 842609220