Add eval_history.json for 1-epoch run

eval_history.json

@@ -1,361 +1,253 @@
 [
-  {
-    "step": 250,
-    "code": {
-      "loss": 3.1747,
-      "perplexity": 23.92,
-      "blocks": 16
-    },
-    "cyber": {
-      "loss": 2.8563,
-      "perplexity": 17.4,
-      "blocks": 16
-    },
-    "replay": {
-      "loss": 6.6708,
-      "perplexity": 789.04,
-      "blocks": 5
-    }
-  },
   {
     "step": 500,
     "code": {
-      "loss": 2.6789,
-      "perplexity": 14.57,
-      "blocks": 16
-    },
-    "cyber": {
-      "loss": 2.2694,
-      "perplexity": 9.67,
-      "blocks": 16
-    },
-    "replay": {
-      "loss": 6.5597,
-      "perplexity": 706.09,
-      "blocks": 5
-    }
-  },
-  {
-    "step": 750,
-    "code": {
-      "loss": 2.4011,
-      "perplexity": 11.03,
+      "loss": 1.5055,
+      "perplexity": 4.51,
       "blocks": 16
     },
     "cyber": {
-      "loss": 2.0552,
-      "perplexity": 7.81,
+      "loss": 1.4013,
+      "perplexity": 4.06,
       "blocks": 16
     },
     "replay": {
-      "loss": 6.4247,
-      "perplexity": 616.9,
+      "loss": 5.793,
+      "perplexity": 327.99,
       "blocks": 5
     }
   },
   {
     "step": 1000,
     "code": {
-      "loss": 2.0518,
-      "perplexity": 7.78,
-      "blocks": 16
-    },
-    "cyber": {
-      "loss": 1.9141,
-      "perplexity": 6.78,
-      "blocks": 16
-    },
-    "replay": {
-      "loss": 6.3068,
-      "perplexity": 548.3,
-      "blocks": 5
-    }
-  },
-  {
-    "step": 1250,
-    "code": {
-      "loss": 1.8635,
-      "perplexity": 6.45,
+      "loss": 1.4807,
+      "perplexity": 4.4,
       "blocks": 16
     },
     "cyber": {
-      "loss": 1.8036,
-      "perplexity": 6.07,
+      "loss": 1.3501,
+      "perplexity": 3.86,
       "blocks": 16
     },
     "replay": {
-      "loss": 6.1973,
-      "perplexity": 491.41,
+      "loss": 5.6901,
+      "perplexity": 295.92,
       "blocks": 5
     }
   },
   {
     "step": 1500,
     "code": {
-      "loss": 1.7843,
-      "perplexity": 5.96,
-      "blocks": 16
-    },
-    "cyber": {
-      "loss": 1.7277,
-      "perplexity": 5.63,
-      "blocks": 16
-    },
-    "replay": {
-      "loss": 6.1335,
-      "perplexity": 461.03,
-      "blocks": 5
-    }
-  },
-  {
-    "step": 1750,
-    "code": {
-      "loss": 1.7222,
-      "perplexity": 5.6,
+      "loss": 1.4339,
+      "perplexity": 4.19,
       "blocks": 16
     },
     "cyber": {
-      "loss": 1.6553,
-      "perplexity": 5.23,
+      "loss": 1.3048,
+      "perplexity": 3.69,
       "blocks": 16
     },
     "replay": {
-      "loss": 6.054,
-      "perplexity": 425.82,
+      "loss": 5.5986,
+      "perplexity": 270.06,
       "blocks": 5
     }
   },
   {
     "step": 2000,
     "code": {
-      "loss": 1.6748,
-      "perplexity": 5.34,
+      "loss": 1.4063,
+      "perplexity": 4.08,
       "blocks": 16
     },
     "cyber": {
-      "loss": 1.6081,
-      "perplexity": 4.99,
+      "loss": 1.2648,
+      "perplexity": 3.54,
       "blocks": 16
     },
     "replay": {
-      "loss": 5.9676,
-      "perplexity": 390.59,
-      "blocks": 5
-    }
-  },
-  {
-    "step": 2250,
-    "code": {
-      "loss": 1.6263,
-      "perplexity": 5.09,
-      "blocks": 16
-    },
-    "cyber": {
-      "loss": 1.5535,
-      "perplexity": 4.73,
-      "blocks": 16
-    },
-    "replay": {
-      "loss": 5.9166,
-      "perplexity": 371.16,
+      "loss": 5.5139,
+      "perplexity": 248.12,
       "blocks": 5
     }
   },
   {
     "step": 2500,
     "code": {
-      "loss": 1.5959,
-      "perplexity": 4.93,
-      "blocks": 16
-    },
-    "cyber": {
-      "loss": 1.5182,
-      "perplexity": 4.56,
-      "blocks": 16
-    },
-    "replay": {
-      "loss": 5.8646,
-      "perplexity": 352.35,
-      "blocks": 5
-    }
-  },
-  {
-    "step": 2750,
-    "code": {
-      "loss": 1.5603,
-      "perplexity": 4.76,
+      "loss": 1.3737,
+      "perplexity": 3.95,
       "blocks": 16
     },
     "cyber": {
-      "loss": 1.4842,
-      "perplexity": 4.41,
+      "loss": 1.2266,
+      "perplexity": 3.41,
       "blocks": 16
     },
     "replay": {
-      "loss": 5.8129,
-      "perplexity": 334.6,
+      "loss": 5.4098,
+      "perplexity": 223.59,
       "blocks": 5
     }
   },
   {
     "step": 3000,
     "code": {
-      "loss": 1.5311,
-      "perplexity": 4.62,
+      "loss": 1.3446,
+      "perplexity": 3.84,
       "blocks": 16
     },
     "cyber": {
-      "loss": 1.458,
-      "perplexity": 4.3,
+      "loss": 1.1879,
+      "perplexity": 3.28,
       "blocks": 16
     },
     "replay": {
-      "loss": 5.7666,
-      "perplexity": 319.44,
+      "loss": 5.3373,
+      "perplexity": 207.95,
       "blocks": 5
     }
   },
   {
-    "step": 3250,
+    "step": 3500,
     "code": {
-      "loss": 1.5151,
-      "perplexity": 4.55,
+      "loss": 1.3275,
+      "perplexity": 3.77,
       "blocks": 16
     },
     "cyber": {
-      "loss": 1.4384,
-      "perplexity": 4.21,
+      "loss": 1.1612,
+      "perplexity": 3.19,
       "blocks": 16
     },
     "replay": {
-      "loss": 5.7233,
-      "perplexity": 305.9,
+      "loss": 5.2525,
+      "perplexity": 191.05,
       "blocks": 5
     }
   },
   {
-    "step": 3500,
+    "step": 4000,
     "code": {
-      "loss": 1.4992,
-      "perplexity": 4.48,
+      "loss": 1.3077,
+      "perplexity": 3.7,
       "blocks": 16
     },
     "cyber": {
-      "loss": 1.4104,
-      "perplexity": 4.1,
+      "loss": 1.1404,
+      "perplexity": 3.13,
       "blocks": 16
     },
     "replay": {
-      "loss": 5.6973,
-      "perplexity": 298.06,
+      "loss": 5.1941,
+      "perplexity": 180.21,
       "blocks": 5
     }
   },
   {
-    "step": 3750,
+    "step": 4500,
     "code": {
-      "loss": 1.4825,
-      "perplexity": 4.4,
+      "loss": 1.2838,
+      "perplexity": 3.61,
       "blocks": 16
     },
     "cyber": {
-      "loss": 1.3947,
-      "perplexity": 4.03,
+      "loss": 1.1155,
+      "perplexity": 3.05,
       "blocks": 16
     },
     "replay": {
-      "loss": 5.6716,
-      "perplexity": 290.51,
+      "loss": 5.1429,
+      "perplexity": 171.22,
       "blocks": 5
     }
   },
   {
-    "step": 4000,
+    "step": 5000,
     "code": {
-      "loss": 1.4692,
-      "perplexity": 4.35,
+      "loss": 1.2708,
+      "perplexity": 3.56,
       "blocks": 16
     },
     "cyber": {
-      "loss": 1.3785,
-      "perplexity": 3.97,
+      "loss": 1.0989,
+      "perplexity": 3.0,
       "blocks": 16
     },
     "replay": {
-      "loss": 5.6515,
-      "perplexity": 284.72,
+      "loss": 5.1017,
+      "perplexity": 164.31,
       "blocks": 5
     }
   },
   {
-    "step": 4250,
+    "step": 5500,
     "code": {
-      "loss": 1.461,
-      "perplexity": 4.31,
+      "loss": 1.261,
+      "perplexity": 3.53,
       "blocks": 16
     },
     "cyber": {
-      "loss": 1.3678,
-      "perplexity": 3.93,
+      "loss": 1.0821,
+      "perplexity": 2.95,
       "blocks": 16
     },
     "replay": {
-      "loss": 5.638,
-      "perplexity": 280.9,
+      "loss": 5.0702,
+      "perplexity": 159.21,
       "blocks": 5
     }
   },
   {
-    "step": 4500,
+    "step": 6000,
     "code": {
-      "loss": 1.4559,
-      "perplexity": 4.29,
+      "loss": 1.2511,
+      "perplexity": 3.49,
       "blocks": 16
     },
     "cyber": {
-      "loss": 1.3612,
-      "perplexity": 3.9,
+      "loss": 1.0722,
+      "perplexity": 2.92,
       "blocks": 16
     },
     "replay": {
-      "loss": 5.6302,
-      "perplexity": 278.72,
+      "loss": 5.0406,
+      "perplexity": 154.56,
       "blocks": 5
     }
   },
   {
-    "step": 4750,
+    "step": 6500,
     "code": {
-      "loss": 1.4533,
-      "perplexity": 4.28,
+      "loss": 1.2471,
+      "perplexity": 3.48,
       "blocks": 16
     },
     "cyber": {
-      "loss": 1.3585,
-      "perplexity": 3.89,
+      "loss": 1.0659,
+      "perplexity": 2.9,
       "blocks": 16
     },
     "replay": {
-      "loss": 5.6273,
-      "perplexity": 277.92,
+      "loss": 5.0278,
+      "perplexity": 152.59,
       "blocks": 5
     }
   },
   {
-    "step": 5000,
+    "step": 7000,
     "code": {
-      "loss": 1.4532,
-      "perplexity": 4.28,
+      "loss": 1.2451,
+      "perplexity": 3.47,
       "blocks": 16
     },
     "cyber": {
-      "loss": 1.3581,
-      "perplexity": 3.89,
+      "loss": 1.0637,
+      "perplexity": 2.9,
       "blocks": 16
     },
     "replay": {
-      "loss": 5.626,
-      "perplexity": 277.55,
+      "loss": 5.0241,
+      "perplexity": 152.03,
       "blocks": 5
     }
   }