Upload trainer_state.json

trainer_state.json

@@ -1,106 +1,136 @@
 {
-  "best_metric": 0.976475715637207,
-  "best_model_checkpoint": "microsoft-swin-base-patch4-window7-224_bert-base-multilingual-cased_bert-base-multilingual-cased/checkpoint-3864",
-  "epoch": 6.0,
-  "global_step": 3864,
+  "best_metric": 0.9636638164520264,
+  "best_model_checkpoint": "microsoft-swin-base-patch4-window7-224_bert-base-multilingual-cased_bert-base-multilingual-cased/checkpoint-4473",
+  "epoch": 9.0,
+  "global_step": 4473,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.78,
-      "learning_rate": 3.923136645962734e-05,
-      "loss": 2.2925,
+      "epoch": 1.0,
+      "eval_loss": 1.1897900104522705,
+      "eval_runtime": 529.9619,
+      "eval_samples_per_second": 4.689,
+      "eval_steps_per_second": 0.236,
+      "step": 497
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8997987927565392e-05,
+      "loss": 1.4457,
       "step": 500
     },
     {
-      "epoch": 1.0,
-      "eval_loss": 1.6365079879760742,
-      "eval_runtime": 232.7485,
-      "eval_samples_per_second": 11.068,
-      "eval_steps_per_second": 0.692,
-      "step": 644
+      "epoch": 2.0,
+      "eval_loss": 1.0949503183364868,
+      "eval_runtime": 30.0349,
+      "eval_samples_per_second": 82.737,
+      "eval_steps_per_second": 4.162,
+      "step": 994
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 3.845496894409938e-05,
-      "loss": 1.4969,
+      "epoch": 2.01,
+      "learning_rate": 1.799195171026157e-05,
+      "loss": 1.1664,
       "step": 1000
     },
     {
-      "epoch": 2.0,
-      "eval_loss": 1.3718241453170776,
-      "eval_runtime": 40.4125,
-      "eval_samples_per_second": 63.743,
-      "eval_steps_per_second": 3.984,
-      "step": 1288
+      "epoch": 3.0,
+      "eval_loss": 1.0486524105072021,
+      "eval_runtime": 29.951,
+      "eval_samples_per_second": 82.969,
+      "eval_steps_per_second": 4.173,
+      "step": 1491
     },
     {
-      "epoch": 2.33,
-      "learning_rate": 3.7678571428571433e-05,
-      "loss": 1.2926,
+      "epoch": 3.02,
+      "learning_rate": 1.6985915492957746e-05,
+      "loss": 1.0482,
       "step": 1500
     },
     {
-      "epoch": 3.0,
-      "eval_loss": 1.1062968969345093,
-      "eval_runtime": 40.4904,
-      "eval_samples_per_second": 63.62,
-      "eval_steps_per_second": 3.976,
-      "step": 1932
+      "epoch": 4.0,
+      "eval_loss": 1.0101640224456787,
+      "eval_runtime": 29.925,
+      "eval_samples_per_second": 83.041,
+      "eval_steps_per_second": 4.177,
+      "step": 1988
     },
     {
-      "epoch": 3.11,
-      "learning_rate": 3.690217391304348e-05,
-      "loss": 1.1515,
+      "epoch": 4.02,
+      "learning_rate": 1.5979879275653925e-05,
+      "loss": 0.9577,
       "step": 2000
     },
     {
-      "epoch": 3.88,
-      "learning_rate": 3.612577639751553e-05,
-      "loss": 1.0376,
+      "epoch": 5.0,
+      "eval_loss": 0.9930654168128967,
+      "eval_runtime": 29.9623,
+      "eval_samples_per_second": 82.938,
+      "eval_steps_per_second": 4.172,
+      "step": 2485
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4973843058350102e-05,
+      "loss": 0.8946,
       "step": 2500
     },
     {
-      "epoch": 4.0,
-      "eval_loss": 1.0416237115859985,
-      "eval_runtime": 40.3883,
-      "eval_samples_per_second": 63.781,
-      "eval_steps_per_second": 3.986,
-      "step": 2576
+      "epoch": 6.0,
+      "eval_loss": 0.9802634119987488,
+      "eval_runtime": 30.0215,
+      "eval_samples_per_second": 82.774,
+      "eval_steps_per_second": 4.164,
+      "step": 2982
     },
     {
-      "epoch": 4.66,
-      "learning_rate": 3.534937888198758e-05,
-      "loss": 0.9386,
+      "epoch": 6.04,
+      "learning_rate": 1.396780684104628e-05,
+      "loss": 0.8387,
       "step": 3000
     },
     {
-      "epoch": 5.0,
-      "eval_loss": 1.0240283012390137,
-      "eval_runtime": 40.487,
-      "eval_samples_per_second": 63.625,
-      "eval_steps_per_second": 3.977,
-      "step": 3220
+      "epoch": 7.0,
+      "eval_loss": 0.9800927639007568,
+      "eval_runtime": 29.9858,
+      "eval_samples_per_second": 82.873,
+      "eval_steps_per_second": 4.169,
+      "step": 3479
     },
     {
-      "epoch": 5.43,
-      "learning_rate": 3.4572981366459633e-05,
-      "loss": 0.867,
+      "epoch": 7.04,
+      "learning_rate": 1.2961770623742455e-05,
+      "loss": 0.7909,
       "step": 3500
     },
     {
-      "epoch": 6.0,
-      "eval_loss": 0.976475715637207,
-      "eval_runtime": 40.4063,
-      "eval_samples_per_second": 63.752,
-      "eval_steps_per_second": 3.985,
-      "step": 3864
+      "epoch": 8.0,
+      "eval_loss": 0.9686193466186523,
+      "eval_runtime": 30.1165,
+      "eval_samples_per_second": 82.513,
+      "eval_steps_per_second": 4.151,
+      "step": 3976
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.1955734406438632e-05,
+      "loss": 0.7463,
+      "step": 4000
+    },
+    {
+      "epoch": 9.0,
+      "eval_loss": 0.9636638164520264,
+      "eval_runtime": 30.0385,
+      "eval_samples_per_second": 82.727,
+      "eval_steps_per_second": 4.161,
+      "step": 4473
     }
   ],
-  "max_steps": 25760,
-  "num_train_epochs": 40,
-  "total_flos": 1.1256086855275315e+19,
+  "max_steps": 9940,
+  "num_train_epochs": 20,
+  "total_flos": 1.6290726869086323e+19,
   "trial_name": null,
   "trial_params": null
 }