Training in progress, epoch 1

all_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 3.0,
-    "total_flos": 8.923548605812163e+17,
-    "train_loss": 0.06785003256232025,
-    "train_runtime": 1092.5569,
-    "train_samples_per_second": 32.86,
-    "train_steps_per_second": 1.027
+    "epoch": 1.0,
+    "total_flos": 2.9745162019373875e+17,
+    "train_loss": 0.24205425238226824,
+    "train_runtime": 396.8396,
+    "train_samples_per_second": 30.156,
+    "train_steps_per_second": 0.942
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:af00da292d93ad76918ab23f2717aad6316ccc78b1791bc1bd0cd4b5959a8c5f
+oid sha256:028941190e7ba5cd87f9c8ffc02db2fad17e5639943eb9e3e4d947073ef1093b
 size 110342832

runs/Dec08_00-28-28_fdad06de1630/events.out.tfevents.1701995979.fdad06de1630.603.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac645ea92f0d53c8338da6cedca3a95175524767a8f113b789a7630318ec5c2f
+size 11302

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 3.0,
-    "total_flos": 8.923548605812163e+17,
-    "train_loss": 0.06785003256232025,
-    "train_runtime": 1092.5569,
-    "train_samples_per_second": 32.86,
-    "train_steps_per_second": 1.027
+    "epoch": 1.0,
+    "total_flos": 2.9745162019373875e+17,
+    "train_loss": 0.24205425238226824,
+    "train_runtime": 396.8396,
+    "train_samples_per_second": 30.156,
+    "train_steps_per_second": 0.942
 }

trainer_state.json

@@ -1,763 +1,271 @@
 {
-  "best_metric": 0.9939839572192514,
+  "best_metric": 0.9669117647058824,
   "best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-eurosat/checkpoint-374",
-  "epoch": 3.0,
+  "epoch": 1.0,
   "eval_steps": 500,
-  "global_step": 1122,
+  "global_step": 374,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.03,
-      "learning_rate": 4.424778761061947e-06,
-      "loss": 0.0823,
+      "learning_rate": 1.3157894736842106e-05,
+      "loss": 0.6739,
       "step": 10
     },
     {
       "epoch": 0.05,
-      "learning_rate": 8.849557522123894e-06,
-      "loss": 0.0572,
+      "learning_rate": 2.6315789473684212e-05,
+      "loss": 0.5529,
       "step": 20
     },
     {
       "epoch": 0.08,
-      "learning_rate": 1.3274336283185843e-05,
-      "loss": 0.1414,
+      "learning_rate": 3.9473684210526316e-05,
+      "loss": 0.414,
       "step": 30
     },
     {
       "epoch": 0.11,
-      "learning_rate": 1.7699115044247787e-05,
-      "loss": 0.0487,
+      "learning_rate": 4.9702380952380955e-05,
+      "loss": 0.3019,
       "step": 40
     },
     {
       "epoch": 0.13,
-      "learning_rate": 2.2123893805309738e-05,
-      "loss": 0.0914,
+      "learning_rate": 4.8214285714285716e-05,
+      "loss": 0.3307,
       "step": 50
     },
     {
       "epoch": 0.16,
-      "learning_rate": 2.6548672566371686e-05,
-      "loss": 0.0858,
+      "learning_rate": 4.672619047619048e-05,
+      "loss": 0.3032,
       "step": 60
     },
     {
       "epoch": 0.19,
-      "learning_rate": 3.097345132743363e-05,
-      "loss": 0.053,
+      "learning_rate": 4.523809523809524e-05,
+      "loss": 0.3082,
       "step": 70
     },
     {
       "epoch": 0.21,
-      "learning_rate": 3.5398230088495574e-05,
-      "loss": 0.0228,
+      "learning_rate": 4.375e-05,
+      "loss": 0.3066,
       "step": 80
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.982300884955752e-05,
-      "loss": 0.0539,
+      "learning_rate": 4.226190476190476e-05,
+      "loss": 0.2905,
       "step": 90
     },
     {
       "epoch": 0.27,
-      "learning_rate": 4.4247787610619477e-05,
-      "loss": 0.0874,
+      "learning_rate": 4.077380952380952e-05,
+      "loss": 0.2479,
       "step": 100
     },
     {
       "epoch": 0.29,
-      "learning_rate": 4.867256637168142e-05,
-      "loss": 0.1159,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 0.2469,
       "step": 110
     },
     {
       "epoch": 0.32,
-      "learning_rate": 4.9653121902874134e-05,
-      "loss": 0.1206,
+      "learning_rate": 3.779761904761905e-05,
+      "loss": 0.273,
       "step": 120
     },
     {
       "epoch": 0.35,
-      "learning_rate": 4.9157581764122896e-05,
-      "loss": 0.0825,
+      "learning_rate": 3.630952380952381e-05,
+      "loss": 0.3014,
       "step": 130
     },
     {
       "epoch": 0.37,
-      "learning_rate": 4.866204162537166e-05,
-      "loss": 0.0582,
+      "learning_rate": 3.4821428571428574e-05,
+      "loss": 0.211,
       "step": 140
     },
     {
       "epoch": 0.4,
-      "learning_rate": 4.816650148662042e-05,
-      "loss": 0.0779,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.1831,
       "step": 150
     },
     {
       "epoch": 0.43,
-      "learning_rate": 4.767096134786918e-05,
-      "loss": 0.1052,
+      "learning_rate": 3.1845238095238096e-05,
+      "loss": 0.2286,
       "step": 160
     },
     {
       "epoch": 0.45,
-      "learning_rate": 4.717542120911794e-05,
-      "loss": 0.1213,
+      "learning_rate": 3.0357142857142857e-05,
+      "loss": 0.231,
       "step": 170
     },
     {
       "epoch": 0.48,
-      "learning_rate": 4.66798810703667e-05,
-      "loss": 0.0734,
+      "learning_rate": 2.886904761904762e-05,
+      "loss": 0.2197,
       "step": 180
     },
     {
       "epoch": 0.51,
-      "learning_rate": 4.618434093161546e-05,
-      "loss": 0.1052,
+      "learning_rate": 2.7380952380952383e-05,
+      "loss": 0.2137,
       "step": 190
     },
     {
       "epoch": 0.53,
-      "learning_rate": 4.568880079286422e-05,
-      "loss": 0.0863,
+      "learning_rate": 2.5892857142857148e-05,
+      "loss": 0.1982,
       "step": 200
     },
     {
       "epoch": 0.56,
-      "learning_rate": 4.5193260654112984e-05,
-      "loss": 0.1206,
+      "learning_rate": 2.4404761904761906e-05,
+      "loss": 0.2188,
       "step": 210
     },
     {
       "epoch": 0.59,
-      "learning_rate": 4.4697720515361746e-05,
-      "loss": 0.0758,
+      "learning_rate": 2.2916666666666667e-05,
+      "loss": 0.1693,
       "step": 220
     },
     {
       "epoch": 0.61,
-      "learning_rate": 4.420218037661051e-05,
-      "loss": 0.0707,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.2192,
       "step": 230
     },
     {
       "epoch": 0.64,
-      "learning_rate": 4.370664023785927e-05,
-      "loss": 0.0818,
+      "learning_rate": 1.9940476190476193e-05,
+      "loss": 0.2273,
       "step": 240
     },
     {
       "epoch": 0.67,
-      "learning_rate": 4.321110009910803e-05,
-      "loss": 0.0866,
+      "learning_rate": 1.8452380952380954e-05,
+      "loss": 0.1556,
       "step": 250
     },
     {
       "epoch": 0.7,
-      "learning_rate": 4.27155599603568e-05,
-      "loss": 0.0871,
+      "learning_rate": 1.6964285714285715e-05,
+      "loss": 0.1847,
       "step": 260
     },
     {
       "epoch": 0.72,
-      "learning_rate": 4.222001982160555e-05,
-      "loss": 0.0993,
+      "learning_rate": 1.5476190476190476e-05,
+      "loss": 0.1663,
       "step": 270
     },
     {
       "epoch": 0.75,
-      "learning_rate": 4.172447968285431e-05,
-      "loss": 0.0848,
+      "learning_rate": 1.398809523809524e-05,
+      "loss": 0.1862,
       "step": 280
     },
     {
       "epoch": 0.78,
-      "learning_rate": 4.122893954410307e-05,
-      "loss": 0.0665,
+      "learning_rate": 1.25e-05,
+      "loss": 0.2096,
       "step": 290
     },
     {
       "epoch": 0.8,
-      "learning_rate": 4.0733399405351834e-05,
-      "loss": 0.1348,
+      "learning_rate": 1.1011904761904762e-05,
+      "loss": 0.1796,
       "step": 300
     },
     {
       "epoch": 0.83,
-      "learning_rate": 4.0237859266600596e-05,
-      "loss": 0.0499,
+      "learning_rate": 9.523809523809523e-06,
+      "loss": 0.1547,
       "step": 310
     },
     {
       "epoch": 0.86,
-      "learning_rate": 3.9742319127849357e-05,
-      "loss": 0.1193,
+      "learning_rate": 8.035714285714286e-06,
+      "loss": 0.1206,
       "step": 320
     },
     {
       "epoch": 0.88,
-      "learning_rate": 3.924677898909812e-05,
-      "loss": 0.1136,
+      "learning_rate": 6.547619047619048e-06,
+      "loss": 0.1614,
       "step": 330
     },
     {
       "epoch": 0.91,
-      "learning_rate": 3.8751238850346886e-05,
-      "loss": 0.07,
+      "learning_rate": 5.05952380952381e-06,
+      "loss": 0.2026,
       "step": 340
     },
     {
       "epoch": 0.94,
-      "learning_rate": 3.825569871159564e-05,
-      "loss": 0.0326,
+      "learning_rate": 3.5714285714285714e-06,
+      "loss": 0.1345,
       "step": 350
     },
     {
       "epoch": 0.96,
-      "learning_rate": 3.77601585728444e-05,
-      "loss": 0.0561,
+      "learning_rate": 2.0833333333333334e-06,
+      "loss": 0.147,
       "step": 360
     },
     {
       "epoch": 0.99,
-      "learning_rate": 3.726461843409316e-05,
-      "loss": 0.0585,
+      "learning_rate": 5.952380952380952e-07,
+      "loss": 0.1483,
       "step": 370
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.9939839572192514,
+      "eval_accuracy": 0.9669117647058824,
       "eval_confusion_matrix": [
         [
-          1506,
-          6
+          1471,
+          41
         ],
         [
-          12,
-          1468
+          58,
+          1422
         ]
       ],
-      "eval_loss": 0.022436169907450676,
-      "eval_precision": 0.9939916468890987,
-      "eval_recall": 0.9939839572192514,
-      "eval_runtime": 61.8227,
-      "eval_samples_per_second": 48.396,
-      "eval_steps_per_second": 6.05,
+      "eval_loss": 0.08614899963140488,
+      "eval_precision": 0.9669680640397452,
+      "eval_recall": 0.9669117647058824,
+      "eval_runtime": 64.8752,
+      "eval_samples_per_second": 46.119,
+      "eval_steps_per_second": 5.765,
       "step": 374
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 3.676907829534192e-05,
-      "loss": 0.0469,
-      "step": 380
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 3.6273538156590684e-05,
-      "loss": 0.0388,
-      "step": 390
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 3.5777998017839445e-05,
-      "loss": 0.0695,
-      "step": 400
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 3.5282457879088207e-05,
-      "loss": 0.1168,
-      "step": 410
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 3.4786917740336974e-05,
-      "loss": 0.0192,
-      "step": 420
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 3.429137760158573e-05,
-      "loss": 0.045,
-      "step": 430
-    },
-    {
-      "epoch": 1.18,
-      "learning_rate": 3.379583746283449e-05,
-      "loss": 0.0891,
-      "step": 440
-    },
-    {
-      "epoch": 1.2,
-      "learning_rate": 3.330029732408325e-05,
-      "loss": 0.0587,
-      "step": 450
-    },
-    {
-      "epoch": 1.23,
-      "learning_rate": 3.280475718533201e-05,
-      "loss": 0.1055,
-      "step": 460
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 3.230921704658077e-05,
-      "loss": 0.0602,
-      "step": 470
-    },
-    {
-      "epoch": 1.28,
-      "learning_rate": 3.1813676907829534e-05,
-      "loss": 0.0282,
-      "step": 480
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 3.13181367690783e-05,
-      "loss": 0.1143,
-      "step": 490
-    },
-    {
-      "epoch": 1.34,
-      "learning_rate": 3.0822596630327057e-05,
-      "loss": 0.0987,
-      "step": 500
-    },
-    {
-      "epoch": 1.36,
-      "learning_rate": 3.0327056491575818e-05,
-      "loss": 0.0426,
-      "step": 510
-    },
-    {
-      "epoch": 1.39,
-      "learning_rate": 2.983151635282458e-05,
-      "loss": 0.0825,
-      "step": 520
-    },
-    {
-      "epoch": 1.42,
-      "learning_rate": 2.933597621407334e-05,
-      "loss": 0.0269,
-      "step": 530
-    },
-    {
-      "epoch": 1.44,
-      "learning_rate": 2.88404360753221e-05,
-      "loss": 0.097,
-      "step": 540
-    },
-    {
-      "epoch": 1.47,
-      "learning_rate": 2.8344895936570865e-05,
-      "loss": 0.0828,
-      "step": 550
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 2.7849355797819627e-05,
-      "loss": 0.0611,
-      "step": 560
-    },
-    {
-      "epoch": 1.52,
-      "learning_rate": 2.7353815659068388e-05,
-      "loss": 0.0651,
-      "step": 570
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 2.6858275520317145e-05,
-      "loss": 0.0443,
-      "step": 580
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 2.6362735381565906e-05,
-      "loss": 0.0719,
-      "step": 590
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 2.5867195242814668e-05,
-      "loss": 0.0724,
-      "step": 600
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 2.537165510406343e-05,
-      "loss": 0.1141,
-      "step": 610
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 2.4876114965312193e-05,
-      "loss": 0.0719,
-      "step": 620
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 2.4380574826560954e-05,
-      "loss": 0.0727,
-      "step": 630
-    },
-    {
-      "epoch": 1.71,
-      "learning_rate": 2.3885034687809712e-05,
-      "loss": 0.0824,
-      "step": 640
-    },
-    {
-      "epoch": 1.74,
-      "learning_rate": 2.3389494549058473e-05,
-      "loss": 0.0769,
-      "step": 650
-    },
-    {
-      "epoch": 1.76,
-      "learning_rate": 2.2893954410307238e-05,
-      "loss": 0.0513,
-      "step": 660
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 2.2398414271556e-05,
-      "loss": 0.0871,
-      "step": 670
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 2.1902874132804756e-05,
-      "loss": 0.0293,
-      "step": 680
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 2.1407333994053518e-05,
-      "loss": 0.0215,
-      "step": 690
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 2.0911793855302282e-05,
-      "loss": 0.0477,
-      "step": 700
-    },
-    {
-      "epoch": 1.9,
-      "learning_rate": 2.0416253716551043e-05,
-      "loss": 0.1385,
-      "step": 710
-    },
-    {
-      "epoch": 1.93,
-      "learning_rate": 1.99207135777998e-05,
-      "loss": 0.0551,
-      "step": 720
-    },
-    {
-      "epoch": 1.95,
-      "learning_rate": 1.9425173439048562e-05,
-      "loss": 0.0519,
-      "step": 730
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 1.8929633300297326e-05,
-      "loss": 0.0792,
-      "step": 740
-    },
-    {
-      "epoch": 2.0,
-      "eval_accuracy": 0.990975935828877,
-      "eval_confusion_matrix": [
-        [
-          1509,
-          3
-        ],
-        [
-          24,
-          1456
-        ]
-      ],
-      "eval_loss": 0.034569673240184784,
-      "eval_precision": 0.9910713859835679,
-      "eval_recall": 0.990975935828877,
-      "eval_runtime": 60.372,
-      "eval_samples_per_second": 49.559,
-      "eval_steps_per_second": 6.195,
-      "step": 748
-    },
-    {
-      "epoch": 2.01,
-      "learning_rate": 1.8434093161546088e-05,
-      "loss": 0.0576,
-      "step": 750
-    },
-    {
-      "epoch": 2.03,
-      "learning_rate": 1.7938553022794845e-05,
-      "loss": 0.0446,
-      "step": 760
-    },
-    {
-      "epoch": 2.06,
-      "learning_rate": 1.7443012884043606e-05,
-      "loss": 0.0707,
-      "step": 770
-    },
-    {
-      "epoch": 2.09,
-      "learning_rate": 1.694747274529237e-05,
-      "loss": 0.0276,
-      "step": 780
-    },
-    {
-      "epoch": 2.11,
-      "learning_rate": 1.6451932606541132e-05,
-      "loss": 0.0666,
-      "step": 790
-    },
-    {
-      "epoch": 2.14,
-      "learning_rate": 1.595639246778989e-05,
-      "loss": 0.063,
-      "step": 800
-    },
-    {
-      "epoch": 2.17,
-      "learning_rate": 1.5460852329038654e-05,
-      "loss": 0.0279,
-      "step": 810
-    },
-    {
-      "epoch": 2.19,
-      "learning_rate": 1.4965312190287414e-05,
-      "loss": 0.0407,
-      "step": 820
-    },
-    {
-      "epoch": 2.22,
-      "learning_rate": 1.4469772051536176e-05,
-      "loss": 0.0546,
-      "step": 830
-    },
-    {
-      "epoch": 2.25,
-      "learning_rate": 1.3974231912784936e-05,
-      "loss": 0.1049,
-      "step": 840
-    },
-    {
-      "epoch": 2.27,
-      "learning_rate": 1.3478691774033697e-05,
-      "loss": 0.0416,
-      "step": 850
-    },
-    {
-      "epoch": 2.3,
-      "learning_rate": 1.298315163528246e-05,
-      "loss": 0.054,
-      "step": 860
-    },
-    {
-      "epoch": 2.33,
-      "learning_rate": 1.2487611496531219e-05,
-      "loss": 0.0368,
-      "step": 870
-    },
-    {
-      "epoch": 2.35,
-      "learning_rate": 1.1992071357779982e-05,
-      "loss": 0.0345,
-      "step": 880
-    },
-    {
-      "epoch": 2.38,
-      "learning_rate": 1.1496531219028741e-05,
-      "loss": 0.0383,
-      "step": 890
-    },
-    {
-      "epoch": 2.41,
-      "learning_rate": 1.1000991080277504e-05,
-      "loss": 0.0252,
-      "step": 900
-    },
-    {
-      "epoch": 2.43,
-      "learning_rate": 1.0505450941526264e-05,
-      "loss": 0.0719,
-      "step": 910
-    },
-    {
-      "epoch": 2.46,
-      "learning_rate": 1.0009910802775026e-05,
-      "loss": 0.036,
-      "step": 920
-    },
-    {
-      "epoch": 2.49,
-      "learning_rate": 9.514370664023786e-06,
-      "loss": 0.0274,
-      "step": 930
-    },
-    {
-      "epoch": 2.51,
-      "learning_rate": 9.018830525272549e-06,
-      "loss": 0.0943,
-      "step": 940
-    },
-    {
-      "epoch": 2.54,
-      "learning_rate": 8.523290386521308e-06,
-      "loss": 0.0499,
-      "step": 950
-    },
-    {
-      "epoch": 2.57,
-      "learning_rate": 8.02775024777007e-06,
-      "loss": 0.0598,
-      "step": 960
-    },
-    {
-      "epoch": 2.59,
-      "learning_rate": 7.532210109018831e-06,
-      "loss": 0.0228,
-      "step": 970
-    },
-    {
-      "epoch": 2.62,
-      "learning_rate": 7.036669970267593e-06,
-      "loss": 0.0326,
-      "step": 980
-    },
-    {
-      "epoch": 2.65,
-      "learning_rate": 6.541129831516353e-06,
-      "loss": 0.004,
-      "step": 990
-    },
-    {
-      "epoch": 2.67,
-      "learning_rate": 6.045589692765114e-06,
-      "loss": 0.0713,
-      "step": 1000
-    },
-    {
-      "epoch": 2.7,
-      "learning_rate": 5.5500495540138754e-06,
-      "loss": 0.0454,
-      "step": 1010
-    },
-    {
-      "epoch": 2.73,
-      "learning_rate": 5.0545094152626366e-06,
-      "loss": 0.0998,
-      "step": 1020
-    },
-    {
-      "epoch": 2.75,
-      "learning_rate": 4.558969276511398e-06,
-      "loss": 0.0314,
-      "step": 1030
-    },
-    {
-      "epoch": 2.78,
-      "learning_rate": 4.063429137760159e-06,
-      "loss": 0.0678,
-      "step": 1040
-    },
-    {
-      "epoch": 2.81,
-      "learning_rate": 3.56788899900892e-06,
-      "loss": 0.0878,
-      "step": 1050
-    },
-    {
-      "epoch": 2.83,
-      "learning_rate": 3.072348860257681e-06,
-      "loss": 0.0986,
-      "step": 1060
-    },
-    {
-      "epoch": 2.86,
-      "learning_rate": 2.576808721506442e-06,
-      "loss": 0.0501,
-      "step": 1070
-    },
-    {
-      "epoch": 2.89,
-      "learning_rate": 2.081268582755203e-06,
-      "loss": 0.0448,
-      "step": 1080
-    },
-    {
-      "epoch": 2.91,
-      "learning_rate": 1.5857284440039643e-06,
-      "loss": 0.0115,
-      "step": 1090
-    },
-    {
-      "epoch": 2.94,
-      "learning_rate": 1.0901883052527254e-06,
-      "loss": 0.0468,
-      "step": 1100
-    },
-    {
-      "epoch": 2.97,
-      "learning_rate": 5.946481665014866e-07,
-      "loss": 0.0651,
-      "step": 1110
-    },
-    {
-      "epoch": 2.99,
-      "learning_rate": 9.910802775024777e-08,
-      "loss": 0.0634,
-      "step": 1120
-    },
-    {
-      "epoch": 3.0,
-      "eval_accuracy": 0.9896390374331551,
-      "eval_confusion_matrix": [
-        [
-          1508,
-          4
-        ],
-        [
-          27,
-          1453
-        ]
-      ],
-      "eval_loss": 0.0354958176612854,
-      "eval_precision": 0.9897531473312668,
-      "eval_recall": 0.9896390374331551,
-      "eval_runtime": 60.2549,
-      "eval_samples_per_second": 49.656,
-      "eval_steps_per_second": 6.207,
-      "step": 1122
-    },
-    {
-      "epoch": 3.0,
-      "step": 1122,
-      "total_flos": 8.923548605812163e+17,
-      "train_loss": 0.06785003256232025,
-      "train_runtime": 1092.5569,
-      "train_samples_per_second": 32.86,
-      "train_steps_per_second": 1.027
+      "epoch": 1.0,
+      "step": 374,
+      "total_flos": 2.9745162019373875e+17,
+      "train_loss": 0.24205425238226824,
+      "train_runtime": 396.8396,
+      "train_samples_per_second": 30.156,
+      "train_steps_per_second": 0.942
     }
   ],
   "logging_steps": 10,
-  "max_steps": 1122,
-  "num_train_epochs": 3,
+  "max_steps": 374,
+  "num_train_epochs": 1,
   "save_steps": 500,
-  "total_flos": 8.923548605812163e+17,
+  "total_flos": 2.9745162019373875e+17,
   "trial_name": null,
   "trial_params": null
 }