End of training

all_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 2.97,
-    "total_flos": 1.573510172037415e+18,
-    "train_loss": 1.3174563893732034,
-    "train_runtime": 880.088,
-    "train_samples_per_second": 23.261,
-    "train_steps_per_second": 0.181
+    "epoch": 9.91,
+    "total_flos": 5.240694044220752e+18,
+    "train_loss": 0.3209688276614783,
+    "train_runtime": 2757.2929,
+    "train_samples_per_second": 24.749,
+    "train_steps_per_second": 0.192
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 2.97,
-    "total_flos": 1.573510172037415e+18,
-    "train_loss": 1.3174563893732034,
-    "train_runtime": 880.088,
-    "train_samples_per_second": 23.261,
-    "train_steps_per_second": 0.181
+    "epoch": 9.91,
+    "total_flos": 5.240694044220752e+18,
+    "train_loss": 0.3209688276614783,
+    "train_runtime": 2757.2929,
+    "train_samples_per_second": 24.749,
+    "train_steps_per_second": 0.192
 }

trainer_state.json

@@ -1,304 +1,490 @@
 {
-  "best_metric": 0.7335680751173709,
-  "best_model_checkpoint": "vit-base-patch16-224-type/checkpoint-372",
-  "epoch": 5.975903614457831,
+  "best_metric": 0.7593896713615024,
+  "best_model_checkpoint": "vit-base-patch16-224-type/checkpoint-560",
+  "epoch": 9.959839357429718,
   "eval_steps": 500,
-  "global_step": 372,
+  "global_step": 620,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.16,
-      "learning_rate": 1.3157894736842106e-05,
-      "loss": 2.2533,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 2.3882,
       "step": 10
     },
     {
       "epoch": 0.32,
-      "learning_rate": 2.6315789473684212e-05,
-      "loss": 2.0412,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 2.2523,
       "step": 20
     },
     {
       "epoch": 0.48,
-      "learning_rate": 3.9473684210526316e-05,
-      "loss": 1.8051,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 2.0603,
       "step": 30
     },
     {
       "epoch": 0.64,
-      "learning_rate": 4.970059880239521e-05,
-      "loss": 1.5407,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 1.8355,
       "step": 40
     },
     {
       "epoch": 0.8,
-      "learning_rate": 4.820359281437126e-05,
-      "loss": 1.3111,
+      "learning_rate": 4.032258064516129e-05,
+      "loss": 1.5709,
       "step": 50
     },
     {
       "epoch": 0.96,
-      "learning_rate": 4.670658682634731e-05,
-      "loss": 1.1891,
+      "learning_rate": 4.8387096774193554e-05,
+      "loss": 1.3494,
       "step": 60
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.6549295774647887,
-      "eval_loss": 1.0236921310424805,
-      "eval_runtime": 23.861,
-      "eval_samples_per_second": 71.414,
-      "eval_steps_per_second": 2.263,
+      "eval_accuracy": 0.6349765258215962,
+      "eval_loss": 1.1001132726669312,
+      "eval_runtime": 22.8557,
+      "eval_samples_per_second": 74.555,
+      "eval_steps_per_second": 2.363,
       "step": 62
     },
     {
       "epoch": 1.12,
-      "learning_rate": 4.5209580838323355e-05,
-      "loss": 1.1132,
+      "learning_rate": 4.92831541218638e-05,
+      "loss": 1.1904,
       "step": 70
     },
     {
       "epoch": 1.29,
-      "learning_rate": 4.3712574850299406e-05,
-      "loss": 1.0744,
+      "learning_rate": 4.8387096774193554e-05,
+      "loss": 1.139,
       "step": 80
     },
     {
       "epoch": 1.45,
-      "learning_rate": 4.221556886227545e-05,
-      "loss": 1.0471,
+      "learning_rate": 4.74910394265233e-05,
+      "loss": 1.0743,
       "step": 90
     },
     {
       "epoch": 1.61,
-      "learning_rate": 4.07185628742515e-05,
-      "loss": 1.0616,
+      "learning_rate": 4.659498207885305e-05,
+      "loss": 1.1071,
       "step": 100
     },
     {
       "epoch": 1.77,
-      "learning_rate": 3.9221556886227544e-05,
-      "loss": 0.9812,
+      "learning_rate": 4.56989247311828e-05,
+      "loss": 1.0113,
       "step": 110
     },
     {
       "epoch": 1.93,
-      "learning_rate": 3.7724550898203595e-05,
-      "loss": 0.9452,
+      "learning_rate": 4.4802867383512545e-05,
+      "loss": 0.9612,
       "step": 120
     },
     {
       "epoch": 1.99,
-      "eval_accuracy": 0.6948356807511737,
-      "eval_loss": 0.8808437585830688,
-      "eval_runtime": 23.7153,
-      "eval_samples_per_second": 71.852,
-      "eval_steps_per_second": 2.277,
+      "eval_accuracy": 0.6772300469483568,
+      "eval_loss": 0.8708141446113586,
+      "eval_runtime": 22.7556,
+      "eval_samples_per_second": 74.883,
+      "eval_steps_per_second": 2.373,
       "step": 124
     },
     {
       "epoch": 2.09,
-      "learning_rate": 3.6227544910179645e-05,
-      "loss": 0.9356,
+      "learning_rate": 4.390681003584229e-05,
+      "loss": 0.9621,
       "step": 130
     },
     {
       "epoch": 2.25,
-      "learning_rate": 3.473053892215569e-05,
-      "loss": 0.906,
+      "learning_rate": 4.301075268817205e-05,
+      "loss": 0.9291,
       "step": 140
     },
     {
       "epoch": 2.41,
-      "learning_rate": 3.323353293413174e-05,
-      "loss": 0.8799,
+      "learning_rate": 4.2114695340501795e-05,
+      "loss": 0.8999,
       "step": 150
     },
     {
       "epoch": 2.57,
-      "learning_rate": 3.1736526946107784e-05,
-      "loss": 0.9281,
+      "learning_rate": 4.121863799283154e-05,
+      "loss": 0.9236,
       "step": 160
     },
     {
       "epoch": 2.73,
-      "learning_rate": 3.0239520958083834e-05,
-      "loss": 0.9325,
+      "learning_rate": 4.032258064516129e-05,
+      "loss": 0.9358,
       "step": 170
     },
     {
       "epoch": 2.89,
-      "learning_rate": 2.874251497005988e-05,
-      "loss": 0.8955,
+      "learning_rate": 3.9426523297491045e-05,
+      "loss": 0.8817,
       "step": 180
     },
     {
       "epoch": 2.99,
-      "eval_accuracy": 0.7124413145539906,
-      "eval_loss": 0.8164246082305908,
-      "eval_runtime": 23.6115,
-      "eval_samples_per_second": 72.168,
-      "eval_steps_per_second": 2.287,
+      "eval_accuracy": 0.7265258215962441,
+      "eval_loss": 0.7897992730140686,
+      "eval_runtime": 23.0974,
+      "eval_samples_per_second": 73.774,
+      "eval_steps_per_second": 2.338,
       "step": 186
     },
     {
       "epoch": 3.05,
-      "learning_rate": 2.724550898203593e-05,
-      "loss": 0.9305,
+      "learning_rate": 3.8530465949820786e-05,
+      "loss": 0.9056,
       "step": 190
     },
     {
       "epoch": 3.21,
-      "learning_rate": 2.5748502994011976e-05,
-      "loss": 0.8252,
+      "learning_rate": 3.763440860215054e-05,
+      "loss": 0.8122,
       "step": 200
     },
     {
       "epoch": 3.37,
-      "learning_rate": 2.4251497005988023e-05,
-      "loss": 0.7959,
+      "learning_rate": 3.673835125448029e-05,
+      "loss": 0.7853,
       "step": 210
     },
     {
       "epoch": 3.53,
-      "learning_rate": 2.275449101796407e-05,
-      "loss": 0.8614,
+      "learning_rate": 3.5842293906810036e-05,
+      "loss": 0.8539,
       "step": 220
     },
     {
       "epoch": 3.69,
-      "learning_rate": 2.125748502994012e-05,
-      "loss": 0.8008,
+      "learning_rate": 3.494623655913979e-05,
+      "loss": 0.7822,
       "step": 230
     },
     {
       "epoch": 3.86,
-      "learning_rate": 1.9760479041916168e-05,
-      "loss": 0.8389,
+      "learning_rate": 3.405017921146954e-05,
+      "loss": 0.8362,
       "step": 240
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.7282863849765259,
-      "eval_loss": 0.7755224108695984,
-      "eval_runtime": 23.4912,
-      "eval_samples_per_second": 72.538,
-      "eval_steps_per_second": 2.299,
+      "eval_accuracy": 0.7276995305164319,
+      "eval_loss": 0.7642938494682312,
+      "eval_runtime": 22.8716,
+      "eval_samples_per_second": 74.503,
+      "eval_steps_per_second": 2.361,
       "step": 249
     },
     {
       "epoch": 4.02,
-      "learning_rate": 1.8263473053892215e-05,
-      "loss": 0.805,
+      "learning_rate": 3.3154121863799286e-05,
+      "loss": 0.806,
       "step": 250
     },
     {
       "epoch": 4.18,
-      "learning_rate": 1.6766467065868263e-05,
-      "loss": 0.7374,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.7131,
       "step": 260
     },
     {
       "epoch": 4.34,
-      "learning_rate": 1.5269461077844313e-05,
-      "loss": 0.785,
+      "learning_rate": 3.136200716845878e-05,
+      "loss": 0.7659,
       "step": 270
     },
     {
       "epoch": 4.5,
-      "learning_rate": 1.377245508982036e-05,
-      "loss": 0.7534,
+      "learning_rate": 3.046594982078853e-05,
+      "loss": 0.7285,
       "step": 280
     },
     {
       "epoch": 4.66,
-      "learning_rate": 1.2275449101796408e-05,
-      "loss": 0.7473,
+      "learning_rate": 2.9569892473118284e-05,
+      "loss": 0.7295,
       "step": 290
     },
     {
       "epoch": 4.82,
-      "learning_rate": 1.0778443113772455e-05,
-      "loss": 0.7854,
+      "learning_rate": 2.8673835125448028e-05,
+      "loss": 0.776,
       "step": 300
     },
     {
       "epoch": 4.98,
-      "learning_rate": 9.281437125748502e-06,
-      "loss": 0.8038,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.7959,
       "step": 310
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.7306338028169014,
-      "eval_loss": 0.7651309370994568,
-      "eval_runtime": 23.7527,
-      "eval_samples_per_second": 71.739,
-      "eval_steps_per_second": 2.273,
+      "eval_accuracy": 0.7382629107981221,
+      "eval_loss": 0.731025218963623,
+      "eval_runtime": 22.8792,
+      "eval_samples_per_second": 74.478,
+      "eval_steps_per_second": 2.36,
       "step": 311
     },
     {
       "epoch": 5.14,
-      "learning_rate": 7.784431137724551e-06,
-      "loss": 0.7728,
+      "learning_rate": 2.6881720430107527e-05,
+      "loss": 0.7269,
       "step": 320
     },
     {
       "epoch": 5.3,
-      "learning_rate": 6.2874251497005985e-06,
-      "loss": 0.7011,
+      "learning_rate": 2.5985663082437278e-05,
+      "loss": 0.6598,
       "step": 330
     },
     {
       "epoch": 5.46,
-      "learning_rate": 4.7904191616766475e-06,
-      "loss": 0.7337,
+      "learning_rate": 2.5089605734767026e-05,
+      "loss": 0.6914,
       "step": 340
     },
     {
       "epoch": 5.62,
-      "learning_rate": 3.293413173652695e-06,
-      "loss": 0.7479,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.6695,
       "step": 350
     },
     {
       "epoch": 5.78,
-      "learning_rate": 1.7964071856287426e-06,
-      "loss": 0.7345,
+      "learning_rate": 2.3297491039426525e-05,
+      "loss": 0.6746,
       "step": 360
     },
     {
       "epoch": 5.94,
-      "learning_rate": 2.9940119760479047e-07,
-      "loss": 0.71,
+      "learning_rate": 2.2401433691756272e-05,
+      "loss": 0.6765,
       "step": 370
     },
     {
-      "epoch": 5.98,
-      "eval_accuracy": 0.7335680751173709,
-      "eval_loss": 0.7623339891433716,
-      "eval_runtime": 23.6551,
-      "eval_samples_per_second": 72.035,
-      "eval_steps_per_second": 2.283,
-      "step": 372
+      "epoch": 5.99,
+      "eval_accuracy": 0.7470657276995305,
+      "eval_loss": 0.7246695756912231,
+      "eval_runtime": 22.9912,
+      "eval_samples_per_second": 74.115,
+      "eval_steps_per_second": 2.349,
+      "step": 373
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.1505376344086024e-05,
+      "loss": 0.6417,
+      "step": 380
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.060931899641577e-05,
+      "loss": 0.633,
+      "step": 390
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.9713261648745522e-05,
+      "loss": 0.6314,
+      "step": 400
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.881720430107527e-05,
+      "loss": 0.6661,
+      "step": 410
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.7921146953405018e-05,
+      "loss": 0.6203,
+      "step": 420
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.702508960573477e-05,
+      "loss": 0.6504,
+      "step": 430
+    },
+    {
+      "epoch": 6.99,
+      "eval_accuracy": 0.7576291079812206,
+      "eval_loss": 0.6938906311988831,
+      "eval_runtime": 22.8151,
+      "eval_samples_per_second": 74.687,
+      "eval_steps_per_second": 2.367,
+      "step": 435
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.5945,
+      "step": 440
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.5232974910394265e-05,
+      "loss": 0.5912,
+      "step": 450
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.4336917562724014e-05,
+      "loss": 0.5645,
+      "step": 460
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.3440860215053763e-05,
+      "loss": 0.5811,
+      "step": 470
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.2544802867383513e-05,
+      "loss": 0.6433,
+      "step": 480
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.1648745519713262e-05,
+      "loss": 0.5846,
+      "step": 490
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.7576291079812206,
+      "eval_loss": 0.6983180046081543,
+      "eval_runtime": 22.9755,
+      "eval_samples_per_second": 74.166,
+      "eval_steps_per_second": 2.35,
+      "step": 498
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.0752688172043012e-05,
+      "loss": 0.6193,
+      "step": 500
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.856630824372761e-06,
+      "loss": 0.6121,
+      "step": 510
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.960573476702509e-06,
+      "loss": 0.5985,
+      "step": 520
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.554,
+      "step": 530
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 7.168458781362007e-06,
+      "loss": 0.5987,
+      "step": 540
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 6.2724014336917564e-06,
+      "loss": 0.5881,
+      "step": 550
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.376344086021506e-06,
+      "loss": 0.5774,
+      "step": 560
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.7593896713615024,
+      "eval_loss": 0.6935292482376099,
+      "eval_runtime": 23.2143,
+      "eval_samples_per_second": 73.403,
+      "eval_steps_per_second": 2.326,
+      "step": 560
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.4802867383512545e-06,
+      "loss": 0.5731,
+      "step": 570
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.5842293906810035e-06,
+      "loss": 0.5458,
+      "step": 580
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.688172043010753e-06,
+      "loss": 0.5373,
+      "step": 590
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.7921146953405017e-06,
+      "loss": 0.5995,
+      "step": 600
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 8.960573476702509e-07,
+      "loss": 0.5283,
+      "step": 610
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 0.0,
+      "loss": 0.5749,
+      "step": 620
+    },
+    {
+      "epoch": 9.96,
+      "eval_accuracy": 0.7570422535211268,
+      "eval_loss": 0.6881020665168762,
+      "eval_runtime": 22.8182,
+      "eval_samples_per_second": 74.677,
+      "eval_steps_per_second": 2.367,
+      "step": 620
     },
     {
-      "epoch": 5.98,
-      "step": 372,
-      "total_flos": 3.6877651691946025e+18,
-      "train_loss": 0.9907107369233203,
-      "train_runtime": 1939.9637,
-      "train_samples_per_second": 24.625,
-      "train_steps_per_second": 0.192
+      "epoch": 9.96,
+      "step": 620,
+      "total_flos": 6.145965294585532e+18,
+      "train_loss": 0.8511337226436985,
+      "train_runtime": 3069.2061,
+      "train_samples_per_second": 25.942,
+      "train_steps_per_second": 0.202
     }
   ],
   "logging_steps": 10,
-  "max_steps": 372,
-  "num_train_epochs": 6,
+  "max_steps": 620,
+  "num_train_epochs": 10,
   "save_steps": 500,
-  "total_flos": 3.6877651691946025e+18,
+  "total_flos": 6.145965294585532e+18,
   "trial_name": null,
   "trial_params": null
 }