{
"best_metric": 0.8643669149353195,
"best_model_checkpoint": "videomae-base-cctv-brawl_extended/checkpoint-4190",
"epoch": 5.165671047391477,
"eval_steps": 500,
"global_step": 5022,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.988071570576541e-07,
"loss": 0.7219,
"step": 10
},
{
"epoch": 0.0,
"learning_rate": 3.976143141153082e-07,
"loss": 0.7194,
"step": 20
},
{
"epoch": 0.01,
"learning_rate": 5.964214711729623e-07,
"loss": 0.7111,
"step": 30
},
{
"epoch": 0.01,
"learning_rate": 7.952286282306164e-07,
"loss": 0.7259,
"step": 40
},
{
"epoch": 0.01,
"learning_rate": 9.940357852882705e-07,
"loss": 0.7108,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 1.1928429423459246e-06,
"loss": 0.7128,
"step": 60
},
{
"epoch": 0.01,
"learning_rate": 1.3916500994035787e-06,
"loss": 0.7096,
"step": 70
},
{
"epoch": 0.02,
"learning_rate": 1.5904572564612327e-06,
"loss": 0.6754,
"step": 80
},
{
"epoch": 0.02,
"learning_rate": 1.7892644135188868e-06,
"loss": 0.6289,
"step": 90
},
{
"epoch": 0.02,
"learning_rate": 1.988071570576541e-06,
"loss": 0.6735,
"step": 100
},
{
"epoch": 0.02,
"learning_rate": 2.186878727634195e-06,
"loss": 0.6414,
"step": 110
},
{
"epoch": 0.02,
"learning_rate": 2.3856858846918492e-06,
"loss": 0.6591,
"step": 120
},
{
"epoch": 0.03,
"learning_rate": 2.5844930417495035e-06,
"loss": 0.6247,
"step": 130
},
{
"epoch": 0.03,
"learning_rate": 2.7833001988071574e-06,
"loss": 0.5978,
"step": 140
},
{
"epoch": 0.03,
"learning_rate": 2.9821073558648112e-06,
"loss": 0.6092,
"step": 150
},
{
"epoch": 0.03,
"learning_rate": 3.1809145129224655e-06,
"loss": 0.5812,
"step": 160
},
{
"epoch": 0.03,
"learning_rate": 3.3797216699801194e-06,
"loss": 0.5962,
"step": 170
},
{
"epoch": 0.04,
"learning_rate": 3.5785288270377736e-06,
"loss": 0.5872,
"step": 180
},
{
"epoch": 0.04,
"learning_rate": 3.7773359840954275e-06,
"loss": 0.5884,
"step": 190
},
{
"epoch": 0.04,
"learning_rate": 3.976143141153082e-06,
"loss": 0.6023,
"step": 200
},
{
"epoch": 0.04,
"learning_rate": 4.174950298210736e-06,
"loss": 0.5678,
"step": 210
},
{
"epoch": 0.04,
"learning_rate": 4.37375745526839e-06,
"loss": 0.5665,
"step": 220
},
{
"epoch": 0.05,
"learning_rate": 4.572564612326044e-06,
"loss": 0.5642,
"step": 230
},
{
"epoch": 0.05,
"learning_rate": 4.7713717693836985e-06,
"loss": 0.5253,
"step": 240
},
{
"epoch": 0.05,
"learning_rate": 4.970178926441353e-06,
"loss": 0.4764,
"step": 250
},
{
"epoch": 0.05,
"learning_rate": 5.168986083499007e-06,
"loss": 0.5652,
"step": 260
},
{
"epoch": 0.05,
"learning_rate": 5.3677932405566604e-06,
"loss": 0.5636,
"step": 270
},
{
"epoch": 0.06,
"learning_rate": 5.566600397614315e-06,
"loss": 0.5912,
"step": 280
},
{
"epoch": 0.06,
"learning_rate": 5.765407554671969e-06,
"loss": 0.5503,
"step": 290
},
{
"epoch": 0.06,
"learning_rate": 5.9642147117296224e-06,
"loss": 0.5331,
"step": 300
},
{
"epoch": 0.06,
"learning_rate": 6.163021868787277e-06,
"loss": 0.5082,
"step": 310
},
{
"epoch": 0.06,
"learning_rate": 6.361829025844931e-06,
"loss": 0.4525,
"step": 320
},
{
"epoch": 0.07,
"learning_rate": 6.560636182902585e-06,
"loss": 0.4899,
"step": 330
},
{
"epoch": 0.07,
"learning_rate": 6.759443339960239e-06,
"loss": 0.4781,
"step": 340
},
{
"epoch": 0.07,
"learning_rate": 6.958250497017893e-06,
"loss": 0.4776,
"step": 350
},
{
"epoch": 0.07,
"learning_rate": 7.157057654075547e-06,
"loss": 0.4492,
"step": 360
},
{
"epoch": 0.07,
"learning_rate": 7.355864811133201e-06,
"loss": 0.4572,
"step": 370
},
{
"epoch": 0.08,
"learning_rate": 7.554671968190855e-06,
"loss": 0.3901,
"step": 380
},
{
"epoch": 0.08,
"learning_rate": 7.75347912524851e-06,
"loss": 0.4159,
"step": 390
},
{
"epoch": 0.08,
"learning_rate": 7.952286282306164e-06,
"loss": 0.3229,
"step": 400
},
{
"epoch": 0.08,
"learning_rate": 8.151093439363818e-06,
"loss": 0.4575,
"step": 410
},
{
"epoch": 0.08,
"learning_rate": 8.349900596421471e-06,
"loss": 0.3253,
"step": 420
},
{
"epoch": 0.09,
"learning_rate": 8.548707753479126e-06,
"loss": 0.4011,
"step": 430
},
{
"epoch": 0.09,
"learning_rate": 8.74751491053678e-06,
"loss": 0.3972,
"step": 440
},
{
"epoch": 0.09,
"learning_rate": 8.946322067594433e-06,
"loss": 0.4415,
"step": 450
},
{
"epoch": 0.09,
"learning_rate": 9.145129224652088e-06,
"loss": 0.3809,
"step": 460
},
{
"epoch": 0.09,
"learning_rate": 9.343936381709743e-06,
"loss": 0.4083,
"step": 470
},
{
"epoch": 0.1,
"learning_rate": 9.542743538767397e-06,
"loss": 0.3875,
"step": 480
},
{
"epoch": 0.1,
"learning_rate": 9.74155069582505e-06,
"loss": 0.4106,
"step": 490
},
{
"epoch": 0.1,
"learning_rate": 9.940357852882705e-06,
"loss": 0.4624,
"step": 500
},
{
"epoch": 0.1,
"learning_rate": 9.984509847311354e-06,
"loss": 0.4058,
"step": 510
},
{
"epoch": 0.1,
"learning_rate": 9.962381057756142e-06,
"loss": 0.4908,
"step": 520
},
{
"epoch": 0.11,
"learning_rate": 9.94025226820093e-06,
"loss": 0.3859,
"step": 530
},
{
"epoch": 0.11,
"learning_rate": 9.918123478645718e-06,
"loss": 0.4327,
"step": 540
},
{
"epoch": 0.11,
"learning_rate": 9.895994689090508e-06,
"loss": 0.4058,
"step": 550
},
{
"epoch": 0.11,
"learning_rate": 9.873865899535297e-06,
"loss": 0.4407,
"step": 560
},
{
"epoch": 0.11,
"learning_rate": 9.851737109980085e-06,
"loss": 0.34,
"step": 570
},
{
"epoch": 0.12,
"learning_rate": 9.829608320424873e-06,
"loss": 0.3745,
"step": 580
},
{
"epoch": 0.12,
"learning_rate": 9.807479530869663e-06,
"loss": 0.3693,
"step": 590
},
{
"epoch": 0.12,
"learning_rate": 9.785350741314451e-06,
"loss": 0.3525,
"step": 600
},
{
"epoch": 0.12,
"learning_rate": 9.76322195175924e-06,
"loss": 0.3235,
"step": 610
},
{
"epoch": 0.12,
"learning_rate": 9.741093162204028e-06,
"loss": 0.2694,
"step": 620
},
{
"epoch": 0.13,
"learning_rate": 9.718964372648817e-06,
"loss": 0.3649,
"step": 630
},
{
"epoch": 0.13,
"learning_rate": 9.696835583093606e-06,
"loss": 0.3023,
"step": 640
},
{
"epoch": 0.13,
"learning_rate": 9.674706793538394e-06,
"loss": 0.3799,
"step": 650
},
{
"epoch": 0.13,
"learning_rate": 9.652578003983182e-06,
"loss": 0.4931,
"step": 660
},
{
"epoch": 0.13,
"learning_rate": 9.630449214427972e-06,
"loss": 0.3699,
"step": 670
},
{
"epoch": 0.14,
"learning_rate": 9.60832042487276e-06,
"loss": 0.3177,
"step": 680
},
{
"epoch": 0.14,
"learning_rate": 9.586191635317548e-06,
"loss": 0.3574,
"step": 690
},
{
"epoch": 0.14,
"learning_rate": 9.564062845762337e-06,
"loss": 0.3777,
"step": 700
},
{
"epoch": 0.14,
"learning_rate": 9.541934056207127e-06,
"loss": 0.2595,
"step": 710
},
{
"epoch": 0.14,
"learning_rate": 9.519805266651915e-06,
"loss": 0.2837,
"step": 720
},
{
"epoch": 0.15,
"learning_rate": 9.497676477096703e-06,
"loss": 0.2935,
"step": 730
},
{
"epoch": 0.15,
"learning_rate": 9.475547687541491e-06,
"loss": 0.2305,
"step": 740
},
{
"epoch": 0.15,
"learning_rate": 9.453418897986281e-06,
"loss": 0.3272,
"step": 750
},
{
"epoch": 0.15,
"learning_rate": 9.43129010843107e-06,
"loss": 0.2583,
"step": 760
},
{
"epoch": 0.15,
"learning_rate": 9.409161318875858e-06,
"loss": 0.3476,
"step": 770
},
{
"epoch": 0.16,
"learning_rate": 9.387032529320646e-06,
"loss": 0.2818,
"step": 780
},
{
"epoch": 0.16,
"learning_rate": 9.364903739765436e-06,
"loss": 0.3278,
"step": 790
},
{
"epoch": 0.16,
"learning_rate": 9.342774950210224e-06,
"loss": 0.306,
"step": 800
},
{
"epoch": 0.16,
"learning_rate": 9.320646160655012e-06,
"loss": 0.406,
"step": 810
},
{
"epoch": 0.16,
"learning_rate": 9.2985173710998e-06,
"loss": 0.2104,
"step": 820
},
{
"epoch": 0.17,
"learning_rate": 9.27638858154459e-06,
"loss": 0.232,
"step": 830
},
{
"epoch": 0.17,
"eval_accuracy": 0.7812622500980008,
"eval_loss": 0.5134419202804565,
"eval_runtime": 577.7526,
"eval_samples_per_second": 4.415,
"eval_steps_per_second": 0.369,
"step": 838
},
{
"epoch": 1.0,
"learning_rate": 9.254259791989379e-06,
"loss": 0.3927,
"step": 840
},
{
"epoch": 1.0,
"learning_rate": 9.232131002434168e-06,
"loss": 0.3709,
"step": 850
},
{
"epoch": 1.0,
"learning_rate": 9.210002212878957e-06,
"loss": 0.3054,
"step": 860
},
{
"epoch": 1.01,
"learning_rate": 9.187873423323745e-06,
"loss": 0.2988,
"step": 870
},
{
"epoch": 1.01,
"learning_rate": 9.165744633768535e-06,
"loss": 0.2673,
"step": 880
},
{
"epoch": 1.01,
"learning_rate": 9.143615844213323e-06,
"loss": 0.3064,
"step": 890
},
{
"epoch": 1.01,
"learning_rate": 9.121487054658111e-06,
"loss": 0.3245,
"step": 900
},
{
"epoch": 1.01,
"learning_rate": 9.0993582651029e-06,
"loss": 0.2463,
"step": 910
},
{
"epoch": 1.02,
"learning_rate": 9.07722947554769e-06,
"loss": 0.2583,
"step": 920
},
{
"epoch": 1.02,
"learning_rate": 9.055100685992478e-06,
"loss": 0.302,
"step": 930
},
{
"epoch": 1.02,
"learning_rate": 9.032971896437266e-06,
"loss": 0.2733,
"step": 940
},
{
"epoch": 1.02,
"learning_rate": 9.010843106882054e-06,
"loss": 0.2683,
"step": 950
},
{
"epoch": 1.02,
"learning_rate": 8.988714317326844e-06,
"loss": 0.2354,
"step": 960
},
{
"epoch": 1.03,
"learning_rate": 8.966585527771632e-06,
"loss": 0.1559,
"step": 970
},
{
"epoch": 1.03,
"learning_rate": 8.94445673821642e-06,
"loss": 0.4634,
"step": 980
},
{
"epoch": 1.03,
"learning_rate": 8.922327948661209e-06,
"loss": 0.2703,
"step": 990
},
{
"epoch": 1.03,
"learning_rate": 8.900199159105998e-06,
"loss": 0.3276,
"step": 1000
},
{
"epoch": 1.03,
"learning_rate": 8.878070369550787e-06,
"loss": 0.3241,
"step": 1010
},
{
"epoch": 1.04,
"learning_rate": 8.855941579995575e-06,
"loss": 0.339,
"step": 1020
},
{
"epoch": 1.04,
"learning_rate": 8.833812790440363e-06,
"loss": 0.2769,
"step": 1030
},
{
"epoch": 1.04,
"learning_rate": 8.811684000885153e-06,
"loss": 0.3192,
"step": 1040
},
{
"epoch": 1.04,
"learning_rate": 8.789555211329941e-06,
"loss": 0.2547,
"step": 1050
},
{
"epoch": 1.04,
"learning_rate": 8.76742642177473e-06,
"loss": 0.3132,
"step": 1060
},
{
"epoch": 1.05,
"learning_rate": 8.745297632219518e-06,
"loss": 0.2438,
"step": 1070
},
{
"epoch": 1.05,
"learning_rate": 8.723168842664308e-06,
"loss": 0.2702,
"step": 1080
},
{
"epoch": 1.05,
"learning_rate": 8.701040053109096e-06,
"loss": 0.2292,
"step": 1090
},
{
"epoch": 1.05,
"learning_rate": 8.678911263553884e-06,
"loss": 0.1375,
"step": 1100
},
{
"epoch": 1.05,
"learning_rate": 8.656782473998672e-06,
"loss": 0.2329,
"step": 1110
},
{
"epoch": 1.06,
"learning_rate": 8.634653684443462e-06,
"loss": 0.2198,
"step": 1120
},
{
"epoch": 1.06,
"learning_rate": 8.61252489488825e-06,
"loss": 0.2714,
"step": 1130
},
{
"epoch": 1.06,
"learning_rate": 8.590396105333039e-06,
"loss": 0.2835,
"step": 1140
},
{
"epoch": 1.06,
"learning_rate": 8.568267315777827e-06,
"loss": 0.2106,
"step": 1150
},
{
"epoch": 1.06,
"learning_rate": 8.546138526222617e-06,
"loss": 0.225,
"step": 1160
},
{
"epoch": 1.07,
"learning_rate": 8.524009736667405e-06,
"loss": 0.3149,
"step": 1170
},
{
"epoch": 1.07,
"learning_rate": 8.501880947112193e-06,
"loss": 0.2412,
"step": 1180
},
{
"epoch": 1.07,
"learning_rate": 8.479752157556981e-06,
"loss": 0.2139,
"step": 1190
},
{
"epoch": 1.07,
"learning_rate": 8.457623368001771e-06,
"loss": 0.2128,
"step": 1200
},
{
"epoch": 1.07,
"learning_rate": 8.43549457844656e-06,
"loss": 0.1696,
"step": 1210
},
{
"epoch": 1.08,
"learning_rate": 8.413365788891348e-06,
"loss": 0.196,
"step": 1220
},
{
"epoch": 1.08,
"learning_rate": 8.391236999336136e-06,
"loss": 0.3033,
"step": 1230
},
{
"epoch": 1.08,
"learning_rate": 8.369108209780926e-06,
"loss": 0.2646,
"step": 1240
},
{
"epoch": 1.08,
"learning_rate": 8.346979420225714e-06,
"loss": 0.344,
"step": 1250
},
{
"epoch": 1.08,
"learning_rate": 8.324850630670504e-06,
"loss": 0.1595,
"step": 1260
},
{
"epoch": 1.09,
"learning_rate": 8.302721841115292e-06,
"loss": 0.3235,
"step": 1270
},
{
"epoch": 1.09,
"learning_rate": 8.28059305156008e-06,
"loss": 0.1695,
"step": 1280
},
{
"epoch": 1.09,
"learning_rate": 8.258464262004869e-06,
"loss": 0.2382,
"step": 1290
},
{
"epoch": 1.09,
"learning_rate": 8.236335472449659e-06,
"loss": 0.3283,
"step": 1300
},
{
"epoch": 1.09,
"learning_rate": 8.214206682894447e-06,
"loss": 0.3028,
"step": 1310
},
{
"epoch": 1.1,
"learning_rate": 8.192077893339235e-06,
"loss": 0.231,
"step": 1320
},
{
"epoch": 1.1,
"learning_rate": 8.169949103784023e-06,
"loss": 0.319,
"step": 1330
},
{
"epoch": 1.1,
"learning_rate": 8.147820314228813e-06,
"loss": 0.2392,
"step": 1340
},
{
"epoch": 1.1,
"learning_rate": 8.125691524673601e-06,
"loss": 0.2225,
"step": 1350
},
{
"epoch": 1.1,
"learning_rate": 8.10356273511839e-06,
"loss": 0.449,
"step": 1360
},
{
"epoch": 1.11,
"learning_rate": 8.081433945563178e-06,
"loss": 0.3471,
"step": 1370
},
{
"epoch": 1.11,
"learning_rate": 8.059305156007968e-06,
"loss": 0.3111,
"step": 1380
},
{
"epoch": 1.11,
"learning_rate": 8.037176366452756e-06,
"loss": 0.2577,
"step": 1390
},
{
"epoch": 1.11,
"learning_rate": 8.015047576897544e-06,
"loss": 0.205,
"step": 1400
},
{
"epoch": 1.11,
"learning_rate": 7.992918787342332e-06,
"loss": 0.2821,
"step": 1410
},
{
"epoch": 1.12,
"learning_rate": 7.970789997787122e-06,
"loss": 0.2689,
"step": 1420
},
{
"epoch": 1.12,
"learning_rate": 7.94866120823191e-06,
"loss": 0.2167,
"step": 1430
},
{
"epoch": 1.12,
"learning_rate": 7.926532418676699e-06,
"loss": 0.2621,
"step": 1440
},
{
"epoch": 1.12,
"learning_rate": 7.904403629121487e-06,
"loss": 0.3638,
"step": 1450
},
{
"epoch": 1.12,
"learning_rate": 7.882274839566277e-06,
"loss": 0.2401,
"step": 1460
},
{
"epoch": 1.13,
"learning_rate": 7.860146050011065e-06,
"loss": 0.2378,
"step": 1470
},
{
"epoch": 1.13,
"learning_rate": 7.838017260455853e-06,
"loss": 0.2154,
"step": 1480
},
{
"epoch": 1.13,
"learning_rate": 7.815888470900642e-06,
"loss": 0.2149,
"step": 1490
},
{
"epoch": 1.13,
"learning_rate": 7.793759681345431e-06,
"loss": 0.2284,
"step": 1500
},
{
"epoch": 1.13,
"learning_rate": 7.77163089179022e-06,
"loss": 0.1492,
"step": 1510
},
{
"epoch": 1.14,
"learning_rate": 7.749502102235008e-06,
"loss": 0.3516,
"step": 1520
},
{
"epoch": 1.14,
"learning_rate": 7.727373312679796e-06,
"loss": 0.2962,
"step": 1530
},
{
"epoch": 1.14,
"learning_rate": 7.705244523124586e-06,
"loss": 0.1978,
"step": 1540
},
{
"epoch": 1.14,
"learning_rate": 7.683115733569374e-06,
"loss": 0.2855,
"step": 1550
},
{
"epoch": 1.14,
"learning_rate": 7.660986944014162e-06,
"loss": 0.2107,
"step": 1560
},
{
"epoch": 1.15,
"learning_rate": 7.63885815445895e-06,
"loss": 0.1989,
"step": 1570
},
{
"epoch": 1.15,
"learning_rate": 7.61672936490374e-06,
"loss": 0.1848,
"step": 1580
},
{
"epoch": 1.15,
"learning_rate": 7.594600575348529e-06,
"loss": 0.2676,
"step": 1590
},
{
"epoch": 1.15,
"learning_rate": 7.572471785793317e-06,
"loss": 0.2119,
"step": 1600
},
{
"epoch": 1.15,
"learning_rate": 7.550342996238106e-06,
"loss": 0.3076,
"step": 1610
},
{
"epoch": 1.16,
"learning_rate": 7.528214206682896e-06,
"loss": 0.2303,
"step": 1620
},
{
"epoch": 1.16,
"learning_rate": 7.506085417127684e-06,
"loss": 0.1646,
"step": 1630
},
{
"epoch": 1.16,
"learning_rate": 7.4839566275724725e-06,
"loss": 0.2384,
"step": 1640
},
{
"epoch": 1.16,
"learning_rate": 7.4618278380172615e-06,
"loss": 0.1908,
"step": 1650
},
{
"epoch": 1.16,
"learning_rate": 7.43969904846205e-06,
"loss": 0.3155,
"step": 1660
},
{
"epoch": 1.17,
"learning_rate": 7.417570258906839e-06,
"loss": 0.1099,
"step": 1670
},
{
"epoch": 1.17,
"eval_accuracy": 0.782438259506076,
"eval_loss": 0.7191832661628723,
"eval_runtime": 556.1282,
"eval_samples_per_second": 4.587,
"eval_steps_per_second": 0.383,
"step": 1676
},
{
"epoch": 2.0,
"learning_rate": 7.395441469351627e-06,
"loss": 0.1875,
"step": 1680
},
{
"epoch": 2.0,
"learning_rate": 7.373312679796416e-06,
"loss": 0.2107,
"step": 1690
},
{
"epoch": 2.0,
"learning_rate": 7.351183890241204e-06,
"loss": 0.374,
"step": 1700
},
{
"epoch": 2.01,
"learning_rate": 7.329055100685993e-06,
"loss": 0.2293,
"step": 1710
},
{
"epoch": 2.01,
"learning_rate": 7.306926311130782e-06,
"loss": 0.2004,
"step": 1720
},
{
"epoch": 2.01,
"learning_rate": 7.284797521575571e-06,
"loss": 0.2731,
"step": 1730
},
{
"epoch": 2.01,
"learning_rate": 7.262668732020359e-06,
"loss": 0.2156,
"step": 1740
},
{
"epoch": 2.01,
"learning_rate": 7.240539942465148e-06,
"loss": 0.3421,
"step": 1750
},
{
"epoch": 2.02,
"learning_rate": 7.218411152909936e-06,
"loss": 0.1165,
"step": 1760
},
{
"epoch": 2.02,
"learning_rate": 7.196282363354725e-06,
"loss": 0.1781,
"step": 1770
},
{
"epoch": 2.02,
"learning_rate": 7.1741535737995134e-06,
"loss": 0.1665,
"step": 1780
},
{
"epoch": 2.02,
"learning_rate": 7.1520247842443025e-06,
"loss": 0.1456,
"step": 1790
},
{
"epoch": 2.02,
"learning_rate": 7.129895994689091e-06,
"loss": 0.2544,
"step": 1800
},
{
"epoch": 2.03,
"learning_rate": 7.10776720513388e-06,
"loss": 0.2747,
"step": 1810
},
{
"epoch": 2.03,
"learning_rate": 7.085638415578668e-06,
"loss": 0.2413,
"step": 1820
},
{
"epoch": 2.03,
"learning_rate": 7.063509626023457e-06,
"loss": 0.2187,
"step": 1830
},
{
"epoch": 2.03,
"learning_rate": 7.041380836468245e-06,
"loss": 0.2101,
"step": 1840
},
{
"epoch": 2.03,
"learning_rate": 7.019252046913034e-06,
"loss": 0.2719,
"step": 1850
},
{
"epoch": 2.04,
"learning_rate": 6.997123257357823e-06,
"loss": 0.2055,
"step": 1860
},
{
"epoch": 2.04,
"learning_rate": 6.974994467802612e-06,
"loss": 0.1508,
"step": 1870
},
{
"epoch": 2.04,
"learning_rate": 6.9528656782474e-06,
"loss": 0.2017,
"step": 1880
},
{
"epoch": 2.04,
"learning_rate": 6.930736888692189e-06,
"loss": 0.1292,
"step": 1890
},
{
"epoch": 2.04,
"learning_rate": 6.908608099136977e-06,
"loss": 0.1796,
"step": 1900
},
{
"epoch": 2.05,
"learning_rate": 6.886479309581766e-06,
"loss": 0.2117,
"step": 1910
},
{
"epoch": 2.05,
"learning_rate": 6.8643505200265544e-06,
"loss": 0.2878,
"step": 1920
},
{
"epoch": 2.05,
"learning_rate": 6.8422217304713435e-06,
"loss": 0.1418,
"step": 1930
},
{
"epoch": 2.05,
"learning_rate": 6.820092940916132e-06,
"loss": 0.0803,
"step": 1940
},
{
"epoch": 2.05,
"learning_rate": 6.797964151360921e-06,
"loss": 0.3428,
"step": 1950
},
{
"epoch": 2.06,
"learning_rate": 6.775835361805709e-06,
"loss": 0.2008,
"step": 1960
},
{
"epoch": 2.06,
"learning_rate": 6.753706572250498e-06,
"loss": 0.2444,
"step": 1970
},
{
"epoch": 2.06,
"learning_rate": 6.731577782695286e-06,
"loss": 0.2775,
"step": 1980
},
{
"epoch": 2.06,
"learning_rate": 6.709448993140075e-06,
"loss": 0.1756,
"step": 1990
},
{
"epoch": 2.06,
"learning_rate": 6.687320203584865e-06,
"loss": 0.2606,
"step": 2000
},
{
"epoch": 2.07,
"learning_rate": 6.6651914140296535e-06,
"loss": 0.2213,
"step": 2010
},
{
"epoch": 2.07,
"learning_rate": 6.6430626244744426e-06,
"loss": 0.1869,
"step": 2020
},
{
"epoch": 2.07,
"learning_rate": 6.620933834919231e-06,
"loss": 0.2149,
"step": 2030
},
{
"epoch": 2.07,
"learning_rate": 6.59880504536402e-06,
"loss": 0.2346,
"step": 2040
},
{
"epoch": 2.07,
"learning_rate": 6.576676255808808e-06,
"loss": 0.1693,
"step": 2050
},
{
"epoch": 2.08,
"learning_rate": 6.554547466253597e-06,
"loss": 0.2225,
"step": 2060
},
{
"epoch": 2.08,
"learning_rate": 6.532418676698385e-06,
"loss": 0.2497,
"step": 2070
},
{
"epoch": 2.08,
"learning_rate": 6.510289887143174e-06,
"loss": 0.2825,
"step": 2080
},
{
"epoch": 2.08,
"learning_rate": 6.488161097587963e-06,
"loss": 0.2909,
"step": 2090
},
{
"epoch": 2.08,
"learning_rate": 6.466032308032752e-06,
"loss": 0.3198,
"step": 2100
},
{
"epoch": 2.09,
"learning_rate": 6.44390351847754e-06,
"loss": 0.1954,
"step": 2110
},
{
"epoch": 2.09,
"learning_rate": 6.421774728922329e-06,
"loss": 0.2166,
"step": 2120
},
{
"epoch": 2.09,
"learning_rate": 6.399645939367117e-06,
"loss": 0.3095,
"step": 2130
},
{
"epoch": 2.09,
"learning_rate": 6.377517149811906e-06,
"loss": 0.0666,
"step": 2140
},
{
"epoch": 2.09,
"learning_rate": 6.3553883602566945e-06,
"loss": 0.2469,
"step": 2150
},
{
"epoch": 2.1,
"learning_rate": 6.3332595707014835e-06,
"loss": 0.1981,
"step": 2160
},
{
"epoch": 2.1,
"learning_rate": 6.311130781146272e-06,
"loss": 0.0946,
"step": 2170
},
{
"epoch": 2.1,
"learning_rate": 6.289001991591061e-06,
"loss": 0.209,
"step": 2180
},
{
"epoch": 2.1,
"learning_rate": 6.266873202035849e-06,
"loss": 0.1831,
"step": 2190
},
{
"epoch": 2.1,
"learning_rate": 6.244744412480638e-06,
"loss": 0.226,
"step": 2200
},
{
"epoch": 2.11,
"learning_rate": 6.222615622925426e-06,
"loss": 0.1695,
"step": 2210
},
{
"epoch": 2.11,
"learning_rate": 6.200486833370215e-06,
"loss": 0.2599,
"step": 2220
},
{
"epoch": 2.11,
"learning_rate": 6.178358043815004e-06,
"loss": 0.0863,
"step": 2230
},
{
"epoch": 2.11,
"learning_rate": 6.156229254259793e-06,
"loss": 0.2787,
"step": 2240
},
{
"epoch": 2.11,
"learning_rate": 6.134100464704581e-06,
"loss": 0.3413,
"step": 2250
},
{
"epoch": 2.12,
"learning_rate": 6.11197167514937e-06,
"loss": 0.2457,
"step": 2260
},
{
"epoch": 2.12,
"learning_rate": 6.089842885594158e-06,
"loss": 0.1474,
"step": 2270
},
{
"epoch": 2.12,
"learning_rate": 6.067714096038947e-06,
"loss": 0.1682,
"step": 2280
},
{
"epoch": 2.12,
"learning_rate": 6.0455853064837355e-06,
"loss": 0.236,
"step": 2290
},
{
"epoch": 2.12,
"learning_rate": 6.0234565169285245e-06,
"loss": 0.1295,
"step": 2300
},
{
"epoch": 2.13,
"learning_rate": 6.001327727373313e-06,
"loss": 0.1731,
"step": 2310
},
{
"epoch": 2.13,
"learning_rate": 5.979198937818102e-06,
"loss": 0.1829,
"step": 2320
},
{
"epoch": 2.13,
"learning_rate": 5.95707014826289e-06,
"loss": 0.2921,
"step": 2330
},
{
"epoch": 2.13,
"learning_rate": 5.934941358707679e-06,
"loss": 0.2537,
"step": 2340
},
{
"epoch": 2.13,
"learning_rate": 5.912812569152467e-06,
"loss": 0.2838,
"step": 2350
},
{
"epoch": 2.14,
"learning_rate": 5.890683779597256e-06,
"loss": 0.2342,
"step": 2360
},
{
"epoch": 2.14,
"learning_rate": 5.868554990042045e-06,
"loss": 0.1875,
"step": 2370
},
{
"epoch": 2.14,
"learning_rate": 5.8464262004868345e-06,
"loss": 0.2775,
"step": 2380
},
{
"epoch": 2.14,
"learning_rate": 5.824297410931623e-06,
"loss": 0.2561,
"step": 2390
},
{
"epoch": 2.14,
"learning_rate": 5.802168621376412e-06,
"loss": 0.292,
"step": 2400
},
{
"epoch": 2.15,
"learning_rate": 5.7800398318212e-06,
"loss": 0.2448,
"step": 2410
},
{
"epoch": 2.15,
"learning_rate": 5.757911042265989e-06,
"loss": 0.2695,
"step": 2420
},
{
"epoch": 2.15,
"learning_rate": 5.735782252710777e-06,
"loss": 0.1965,
"step": 2430
},
{
"epoch": 2.15,
"learning_rate": 5.713653463155566e-06,
"loss": 0.203,
"step": 2440
},
{
"epoch": 2.15,
"learning_rate": 5.691524673600355e-06,
"loss": 0.239,
"step": 2450
},
{
"epoch": 2.16,
"learning_rate": 5.669395884045144e-06,
"loss": 0.2028,
"step": 2460
},
{
"epoch": 2.16,
"learning_rate": 5.647267094489932e-06,
"loss": 0.1763,
"step": 2470
},
{
"epoch": 2.16,
"learning_rate": 5.625138304934721e-06,
"loss": 0.2413,
"step": 2480
},
{
"epoch": 2.16,
"learning_rate": 5.603009515379509e-06,
"loss": 0.2726,
"step": 2490
},
{
"epoch": 2.16,
"learning_rate": 5.580880725824298e-06,
"loss": 0.2998,
"step": 2500
},
{
"epoch": 2.17,
"learning_rate": 5.5587519362690865e-06,
"loss": 0.2178,
"step": 2510
},
{
"epoch": 2.17,
"eval_accuracy": 0.8298706389651117,
"eval_loss": 0.4418592154979706,
"eval_runtime": 455.5394,
"eval_samples_per_second": 5.6,
"eval_steps_per_second": 0.468,
"step": 2514
},
{
"epoch": 3.0,
"learning_rate": 5.5366231467138755e-06,
"loss": 0.1391,
"step": 2520
},
{
"epoch": 3.0,
"learning_rate": 5.514494357158664e-06,
"loss": 0.1202,
"step": 2530
},
{
"epoch": 3.01,
"learning_rate": 5.492365567603453e-06,
"loss": 0.1799,
"step": 2540
},
{
"epoch": 3.01,
"learning_rate": 5.470236778048241e-06,
"loss": 0.1843,
"step": 2550
},
{
"epoch": 3.01,
"learning_rate": 5.44810798849303e-06,
"loss": 0.1504,
"step": 2560
},
{
"epoch": 3.01,
"learning_rate": 5.425979198937818e-06,
"loss": 0.1064,
"step": 2570
},
{
"epoch": 3.01,
"learning_rate": 5.403850409382607e-06,
"loss": 0.397,
"step": 2580
},
{
"epoch": 3.02,
"learning_rate": 5.381721619827396e-06,
"loss": 0.1298,
"step": 2590
},
{
"epoch": 3.02,
"learning_rate": 5.359592830272185e-06,
"loss": 0.1984,
"step": 2600
},
{
"epoch": 3.02,
"learning_rate": 5.337464040716973e-06,
"loss": 0.3349,
"step": 2610
},
{
"epoch": 3.02,
"learning_rate": 5.315335251161762e-06,
"loss": 0.1957,
"step": 2620
},
{
"epoch": 3.02,
"learning_rate": 5.29320646160655e-06,
"loss": 0.2681,
"step": 2630
},
{
"epoch": 3.03,
"learning_rate": 5.271077672051339e-06,
"loss": 0.0977,
"step": 2640
},
{
"epoch": 3.03,
"learning_rate": 5.2489488824961274e-06,
"loss": 0.3088,
"step": 2650
},
{
"epoch": 3.03,
"learning_rate": 5.2268200929409165e-06,
"loss": 0.2288,
"step": 2660
},
{
"epoch": 3.03,
"learning_rate": 5.204691303385705e-06,
"loss": 0.2878,
"step": 2670
},
{
"epoch": 3.03,
"learning_rate": 5.182562513830494e-06,
"loss": 0.2257,
"step": 2680
},
{
"epoch": 3.04,
"learning_rate": 5.160433724275282e-06,
"loss": 0.15,
"step": 2690
},
{
"epoch": 3.04,
"learning_rate": 5.138304934720071e-06,
"loss": 0.2424,
"step": 2700
},
{
"epoch": 3.04,
"learning_rate": 5.116176145164859e-06,
"loss": 0.2213,
"step": 2710
},
{
"epoch": 3.04,
"learning_rate": 5.094047355609648e-06,
"loss": 0.1488,
"step": 2720
},
{
"epoch": 3.04,
"learning_rate": 5.071918566054437e-06,
"loss": 0.1004,
"step": 2730
},
{
"epoch": 3.05,
"learning_rate": 5.049789776499226e-06,
"loss": 0.2068,
"step": 2740
},
{
"epoch": 3.05,
"learning_rate": 5.027660986944014e-06,
"loss": 0.3556,
"step": 2750
},
{
"epoch": 3.05,
"learning_rate": 5.005532197388803e-06,
"loss": 0.2188,
"step": 2760
},
{
"epoch": 3.05,
"learning_rate": 4.983403407833592e-06,
"loss": 0.2531,
"step": 2770
},
{
"epoch": 3.05,
"learning_rate": 4.96127461827838e-06,
"loss": 0.2659,
"step": 2780
},
{
"epoch": 3.05,
"learning_rate": 4.939145828723169e-06,
"loss": 0.2333,
"step": 2790
},
{
"epoch": 3.06,
"learning_rate": 4.9170170391679575e-06,
"loss": 0.2603,
"step": 2800
},
{
"epoch": 3.06,
"learning_rate": 4.8948882496127466e-06,
"loss": 0.1662,
"step": 2810
},
{
"epoch": 3.06,
"learning_rate": 4.872759460057535e-06,
"loss": 0.1891,
"step": 2820
},
{
"epoch": 3.06,
"learning_rate": 4.850630670502324e-06,
"loss": 0.1719,
"step": 2830
},
{
"epoch": 3.06,
"learning_rate": 4.828501880947112e-06,
"loss": 0.1205,
"step": 2840
},
{
"epoch": 3.07,
"learning_rate": 4.806373091391901e-06,
"loss": 0.1061,
"step": 2850
},
{
"epoch": 3.07,
"learning_rate": 4.78424430183669e-06,
"loss": 0.2271,
"step": 2860
},
{
"epoch": 3.07,
"learning_rate": 4.762115512281478e-06,
"loss": 0.2392,
"step": 2870
},
{
"epoch": 3.07,
"learning_rate": 4.7399867227262675e-06,
"loss": 0.2029,
"step": 2880
},
{
"epoch": 3.07,
"learning_rate": 4.717857933171056e-06,
"loss": 0.2938,
"step": 2890
},
{
"epoch": 3.08,
"learning_rate": 4.695729143615845e-06,
"loss": 0.2804,
"step": 2900
},
{
"epoch": 3.08,
"learning_rate": 4.673600354060633e-06,
"loss": 0.2101,
"step": 2910
},
{
"epoch": 3.08,
"learning_rate": 4.651471564505422e-06,
"loss": 0.2806,
"step": 2920
},
{
"epoch": 3.08,
"learning_rate": 4.62934277495021e-06,
"loss": 0.1367,
"step": 2930
},
{
"epoch": 3.08,
"learning_rate": 4.607213985394999e-06,
"loss": 0.2148,
"step": 2940
},
{
"epoch": 3.09,
"learning_rate": 4.5850851958397876e-06,
"loss": 0.0975,
"step": 2950
},
{
"epoch": 3.09,
"learning_rate": 4.562956406284577e-06,
"loss": 0.1914,
"step": 2960
},
{
"epoch": 3.09,
"learning_rate": 4.540827616729365e-06,
"loss": 0.1859,
"step": 2970
},
{
"epoch": 3.09,
"learning_rate": 4.518698827174154e-06,
"loss": 0.1547,
"step": 2980
},
{
"epoch": 3.09,
"learning_rate": 4.496570037618942e-06,
"loss": 0.1743,
"step": 2990
},
{
"epoch": 3.1,
"learning_rate": 4.474441248063731e-06,
"loss": 0.2247,
"step": 3000
},
{
"epoch": 3.1,
"learning_rate": 4.452312458508519e-06,
"loss": 0.2033,
"step": 3010
},
{
"epoch": 3.1,
"learning_rate": 4.4301836689533085e-06,
"loss": 0.1757,
"step": 3020
},
{
"epoch": 3.1,
"learning_rate": 4.408054879398097e-06,
"loss": 0.212,
"step": 3030
},
{
"epoch": 3.1,
"learning_rate": 4.385926089842886e-06,
"loss": 0.2183,
"step": 3040
},
{
"epoch": 3.11,
"learning_rate": 4.363797300287675e-06,
"loss": 0.2498,
"step": 3050
},
{
"epoch": 3.11,
"learning_rate": 4.341668510732464e-06,
"loss": 0.0668,
"step": 3060
},
{
"epoch": 3.11,
"learning_rate": 4.319539721177252e-06,
"loss": 0.1867,
"step": 3070
},
{
"epoch": 3.11,
"learning_rate": 4.297410931622041e-06,
"loss": 0.1403,
"step": 3080
},
{
"epoch": 3.11,
"learning_rate": 4.275282142066829e-06,
"loss": 0.2343,
"step": 3090
},
{
"epoch": 3.12,
"learning_rate": 4.2531533525116185e-06,
"loss": 0.1536,
"step": 3100
},
{
"epoch": 3.12,
"learning_rate": 4.231024562956407e-06,
"loss": 0.1656,
"step": 3110
},
{
"epoch": 3.12,
"learning_rate": 4.208895773401196e-06,
"loss": 0.2044,
"step": 3120
},
{
"epoch": 3.12,
"learning_rate": 4.186766983845984e-06,
"loss": 0.1283,
"step": 3130
},
{
"epoch": 3.12,
"learning_rate": 4.164638194290773e-06,
"loss": 0.0793,
"step": 3140
},
{
"epoch": 3.13,
"learning_rate": 4.142509404735561e-06,
"loss": 0.2157,
"step": 3150
},
{
"epoch": 3.13,
"learning_rate": 4.12038061518035e-06,
"loss": 0.2458,
"step": 3160
},
{
"epoch": 3.13,
"learning_rate": 4.0982518256251385e-06,
"loss": 0.1386,
"step": 3170
},
{
"epoch": 3.13,
"learning_rate": 4.076123036069928e-06,
"loss": 0.1956,
"step": 3180
},
{
"epoch": 3.13,
"learning_rate": 4.053994246514716e-06,
"loss": 0.1353,
"step": 3190
},
{
"epoch": 3.14,
"learning_rate": 4.031865456959505e-06,
"loss": 0.1785,
"step": 3200
},
{
"epoch": 3.14,
"learning_rate": 4.009736667404293e-06,
"loss": 0.2949,
"step": 3210
},
{
"epoch": 3.14,
"learning_rate": 3.987607877849082e-06,
"loss": 0.2012,
"step": 3220
},
{
"epoch": 3.14,
"learning_rate": 3.96547908829387e-06,
"loss": 0.1827,
"step": 3230
},
{
"epoch": 3.14,
"learning_rate": 3.9433502987386595e-06,
"loss": 0.2234,
"step": 3240
},
{
"epoch": 3.15,
"learning_rate": 3.9212215091834485e-06,
"loss": 0.1809,
"step": 3250
},
{
"epoch": 3.15,
"learning_rate": 3.899092719628237e-06,
"loss": 0.1309,
"step": 3260
},
{
"epoch": 3.15,
"learning_rate": 3.876963930073026e-06,
"loss": 0.1616,
"step": 3270
},
{
"epoch": 3.15,
"learning_rate": 3.854835140517814e-06,
"loss": 0.2984,
"step": 3280
},
{
"epoch": 3.15,
"learning_rate": 3.832706350962603e-06,
"loss": 0.0939,
"step": 3290
},
{
"epoch": 3.16,
"learning_rate": 3.8105775614073913e-06,
"loss": 0.1858,
"step": 3300
},
{
"epoch": 3.16,
"learning_rate": 3.78844877185218e-06,
"loss": 0.2291,
"step": 3310
},
{
"epoch": 3.16,
"learning_rate": 3.7663199822969686e-06,
"loss": 0.2996,
"step": 3320
},
{
"epoch": 3.16,
"learning_rate": 3.7441911927417572e-06,
"loss": 0.1286,
"step": 3330
},
{
"epoch": 3.16,
"learning_rate": 3.722062403186546e-06,
"loss": 0.0735,
"step": 3340
},
{
"epoch": 3.17,
"learning_rate": 3.6999336136313345e-06,
"loss": 0.0697,
"step": 3350
},
{
"epoch": 3.17,
"eval_accuracy": 0.8447667581340651,
"eval_loss": 0.6156530976295471,
"eval_runtime": 478.3766,
"eval_samples_per_second": 5.333,
"eval_steps_per_second": 0.445,
"step": 3352
},
{
"epoch": 4.0,
"learning_rate": 3.677804824076123e-06,
"loss": 0.0433,
"step": 3360
},
{
"epoch": 4.0,
"learning_rate": 3.655676034520912e-06,
"loss": 0.3206,
"step": 3370
},
{
"epoch": 4.01,
"learning_rate": 3.6335472449657004e-06,
"loss": 0.228,
"step": 3380
},
{
"epoch": 4.01,
"learning_rate": 3.611418455410489e-06,
"loss": 0.2148,
"step": 3390
},
{
"epoch": 4.01,
"learning_rate": 3.5892896658552777e-06,
"loss": 0.1528,
"step": 3400
},
{
"epoch": 4.01,
"learning_rate": 3.5671608763000664e-06,
"loss": 0.2124,
"step": 3410
},
{
"epoch": 4.01,
"learning_rate": 3.545032086744855e-06,
"loss": 0.0536,
"step": 3420
},
{
"epoch": 4.02,
"learning_rate": 3.5229032971896437e-06,
"loss": 0.2198,
"step": 3430
},
{
"epoch": 4.02,
"learning_rate": 3.500774507634433e-06,
"loss": 0.1849,
"step": 3440
},
{
"epoch": 4.02,
"learning_rate": 3.478645718079222e-06,
"loss": 0.2319,
"step": 3450
},
{
"epoch": 4.02,
"learning_rate": 3.45651692852401e-06,
"loss": 0.0702,
"step": 3460
},
{
"epoch": 4.02,
"learning_rate": 3.4343881389687987e-06,
"loss": 0.1208,
"step": 3470
},
{
"epoch": 4.03,
"learning_rate": 3.4122593494135873e-06,
"loss": 0.108,
"step": 3480
},
{
"epoch": 4.03,
"learning_rate": 3.390130559858376e-06,
"loss": 0.2169,
"step": 3490
},
{
"epoch": 4.03,
"learning_rate": 3.3680017703031646e-06,
"loss": 0.1446,
"step": 3500
},
{
"epoch": 4.03,
"learning_rate": 3.3458729807479532e-06,
"loss": 0.1869,
"step": 3510
},
{
"epoch": 4.03,
"learning_rate": 3.323744191192742e-06,
"loss": 0.3277,
"step": 3520
},
{
"epoch": 4.04,
"learning_rate": 3.3016154016375305e-06,
"loss": 0.2244,
"step": 3530
},
{
"epoch": 4.04,
"learning_rate": 3.279486612082319e-06,
"loss": 0.1608,
"step": 3540
},
{
"epoch": 4.04,
"learning_rate": 3.257357822527108e-06,
"loss": 0.1947,
"step": 3550
},
{
"epoch": 4.04,
"learning_rate": 3.2352290329718964e-06,
"loss": 0.2295,
"step": 3560
},
{
"epoch": 4.04,
"learning_rate": 3.213100243416685e-06,
"loss": 0.1933,
"step": 3570
},
{
"epoch": 4.05,
"learning_rate": 3.1909714538614737e-06,
"loss": 0.1222,
"step": 3580
},
{
"epoch": 4.05,
"learning_rate": 3.1688426643062624e-06,
"loss": 0.1036,
"step": 3590
},
{
"epoch": 4.05,
"learning_rate": 3.146713874751051e-06,
"loss": 0.1496,
"step": 3600
},
{
"epoch": 4.05,
"learning_rate": 3.1245850851958396e-06,
"loss": 0.0575,
"step": 3610
},
{
"epoch": 4.05,
"learning_rate": 3.1024562956406283e-06,
"loss": 0.0818,
"step": 3620
},
{
"epoch": 4.06,
"learning_rate": 3.0803275060854178e-06,
"loss": 0.216,
"step": 3630
},
{
"epoch": 4.06,
"learning_rate": 3.0581987165302064e-06,
"loss": 0.097,
"step": 3640
},
{
"epoch": 4.06,
"learning_rate": 3.036069926974995e-06,
"loss": 0.0846,
"step": 3650
},
{
"epoch": 4.06,
"learning_rate": 3.0139411374197837e-06,
"loss": 0.178,
"step": 3660
},
{
"epoch": 4.06,
"learning_rate": 2.9918123478645723e-06,
"loss": 0.3417,
"step": 3670
},
{
"epoch": 4.07,
"learning_rate": 2.969683558309361e-06,
"loss": 0.1691,
"step": 3680
},
{
"epoch": 4.07,
"learning_rate": 2.9475547687541496e-06,
"loss": 0.2325,
"step": 3690
},
{
"epoch": 4.07,
"learning_rate": 2.9254259791989383e-06,
"loss": 0.0148,
"step": 3700
},
{
"epoch": 4.07,
"learning_rate": 2.903297189643727e-06,
"loss": 0.3002,
"step": 3710
},
{
"epoch": 4.07,
"learning_rate": 2.8811684000885156e-06,
"loss": 0.1167,
"step": 3720
},
{
"epoch": 4.08,
"learning_rate": 2.859039610533304e-06,
"loss": 0.3281,
"step": 3730
},
{
"epoch": 4.08,
"learning_rate": 2.836910820978093e-06,
"loss": 0.1938,
"step": 3740
},
{
"epoch": 4.08,
"learning_rate": 2.8147820314228815e-06,
"loss": 0.1634,
"step": 3750
},
{
"epoch": 4.08,
"learning_rate": 2.79265324186767e-06,
"loss": 0.1627,
"step": 3760
},
{
"epoch": 4.08,
"learning_rate": 2.7705244523124588e-06,
"loss": 0.1519,
"step": 3770
},
{
"epoch": 4.09,
"learning_rate": 2.7483956627572474e-06,
"loss": 0.0827,
"step": 3780
},
{
"epoch": 4.09,
"learning_rate": 2.726266873202036e-06,
"loss": 0.181,
"step": 3790
},
{
"epoch": 4.09,
"learning_rate": 2.7041380836468247e-06,
"loss": 0.1907,
"step": 3800
},
{
"epoch": 4.09,
"learning_rate": 2.6820092940916133e-06,
"loss": 0.1519,
"step": 3810
},
{
"epoch": 4.09,
"learning_rate": 2.6598805045364024e-06,
"loss": 0.2911,
"step": 3820
},
{
"epoch": 4.1,
"learning_rate": 2.637751714981191e-06,
"loss": 0.2609,
"step": 3830
},
{
"epoch": 4.1,
"learning_rate": 2.6156229254259797e-06,
"loss": 0.1138,
"step": 3840
},
{
"epoch": 4.1,
"learning_rate": 2.5934941358707683e-06,
"loss": 0.1796,
"step": 3850
},
{
"epoch": 4.1,
"learning_rate": 2.571365346315557e-06,
"loss": 0.1406,
"step": 3860
},
{
"epoch": 4.1,
"learning_rate": 2.5492365567603456e-06,
"loss": 0.3324,
"step": 3870
},
{
"epoch": 4.11,
"learning_rate": 2.5271077672051343e-06,
"loss": 0.1073,
"step": 3880
},
{
"epoch": 4.11,
"learning_rate": 2.504978977649923e-06,
"loss": 0.1545,
"step": 3890
},
{
"epoch": 4.11,
"learning_rate": 2.4828501880947115e-06,
"loss": 0.0673,
"step": 3900
},
{
"epoch": 4.11,
"learning_rate": 2.4607213985395e-06,
"loss": 0.1819,
"step": 3910
},
{
"epoch": 4.11,
"learning_rate": 2.438592608984289e-06,
"loss": 0.0896,
"step": 3920
},
{
"epoch": 4.12,
"learning_rate": 2.4164638194290775e-06,
"loss": 0.1293,
"step": 3930
},
{
"epoch": 4.12,
"learning_rate": 2.394335029873866e-06,
"loss": 0.2196,
"step": 3940
},
{
"epoch": 4.12,
"learning_rate": 2.3722062403186548e-06,
"loss": 0.2235,
"step": 3950
},
{
"epoch": 4.12,
"learning_rate": 2.3500774507634434e-06,
"loss": 0.1067,
"step": 3960
},
{
"epoch": 4.12,
"learning_rate": 2.327948661208232e-06,
"loss": 0.2652,
"step": 3970
},
{
"epoch": 4.13,
"learning_rate": 2.3058198716530207e-06,
"loss": 0.2255,
"step": 3980
},
{
"epoch": 4.13,
"learning_rate": 2.2836910820978093e-06,
"loss": 0.2405,
"step": 3990
},
{
"epoch": 4.13,
"learning_rate": 2.261562292542598e-06,
"loss": 0.137,
"step": 4000
},
{
"epoch": 4.13,
"learning_rate": 2.2394335029873866e-06,
"loss": 0.133,
"step": 4010
},
{
"epoch": 4.13,
"learning_rate": 2.2173047134321752e-06,
"loss": 0.1682,
"step": 4020
},
{
"epoch": 4.14,
"learning_rate": 2.195175923876964e-06,
"loss": 0.2465,
"step": 4030
},
{
"epoch": 4.14,
"learning_rate": 2.1730471343217525e-06,
"loss": 0.1982,
"step": 4040
},
{
"epoch": 4.14,
"learning_rate": 2.150918344766541e-06,
"loss": 0.1348,
"step": 4050
},
{
"epoch": 4.14,
"learning_rate": 2.1287895552113302e-06,
"loss": 0.1349,
"step": 4060
},
{
"epoch": 4.14,
"learning_rate": 2.106660765656119e-06,
"loss": 0.1451,
"step": 4070
},
{
"epoch": 4.14,
"learning_rate": 2.0845319761009075e-06,
"loss": 0.2781,
"step": 4080
},
{
"epoch": 4.15,
"learning_rate": 2.062403186545696e-06,
"loss": 0.1213,
"step": 4090
},
{
"epoch": 4.15,
"learning_rate": 2.040274396990485e-06,
"loss": 0.1548,
"step": 4100
},
{
"epoch": 4.15,
"learning_rate": 2.0181456074352735e-06,
"loss": 0.1743,
"step": 4110
},
{
"epoch": 4.15,
"learning_rate": 1.996016817880062e-06,
"loss": 0.163,
"step": 4120
},
{
"epoch": 4.15,
"learning_rate": 1.9738880283248507e-06,
"loss": 0.1655,
"step": 4130
},
{
"epoch": 4.16,
"learning_rate": 1.9517592387696394e-06,
"loss": 0.1672,
"step": 4140
},
{
"epoch": 4.16,
"learning_rate": 1.929630449214428e-06,
"loss": 0.0966,
"step": 4150
},
{
"epoch": 4.16,
"learning_rate": 1.907501659659217e-06,
"loss": 0.1906,
"step": 4160
},
{
"epoch": 4.16,
"learning_rate": 1.8853728701040055e-06,
"loss": 0.2156,
"step": 4170
},
{
"epoch": 4.16,
"learning_rate": 1.8632440805487942e-06,
"loss": 0.1546,
"step": 4180
},
{
"epoch": 4.17,
"learning_rate": 1.8411152909935828e-06,
"loss": 0.2116,
"step": 4190
},
{
"epoch": 4.17,
"eval_accuracy": 0.8643669149353195,
"eval_loss": 0.4997994899749756,
"eval_runtime": 440.6246,
"eval_samples_per_second": 5.79,
"eval_steps_per_second": 0.483,
"step": 4190
},
{
"epoch": 5.0,
"learning_rate": 1.8189865014383714e-06,
"loss": 0.1667,
"step": 4200
},
{
"epoch": 5.0,
"learning_rate": 1.79685771188316e-06,
"loss": 0.396,
"step": 4210
},
{
"epoch": 5.01,
"learning_rate": 1.7747289223279487e-06,
"loss": 0.12,
"step": 4220
},
{
"epoch": 5.01,
"learning_rate": 1.7526001327727374e-06,
"loss": 0.1906,
"step": 4230
},
{
"epoch": 5.01,
"learning_rate": 1.730471343217526e-06,
"loss": 0.3117,
"step": 4240
},
{
"epoch": 5.01,
"learning_rate": 1.7083425536623149e-06,
"loss": 0.1781,
"step": 4250
},
{
"epoch": 5.01,
"learning_rate": 1.6862137641071035e-06,
"loss": 0.0814,
"step": 4260
},
{
"epoch": 5.02,
"learning_rate": 1.6640849745518922e-06,
"loss": 0.1559,
"step": 4270
},
{
"epoch": 5.02,
"learning_rate": 1.6419561849966808e-06,
"loss": 0.2145,
"step": 4280
},
{
"epoch": 5.02,
"learning_rate": 1.6198273954414694e-06,
"loss": 0.1389,
"step": 4290
},
{
"epoch": 5.02,
"learning_rate": 1.597698605886258e-06,
"loss": 0.0715,
"step": 4300
},
{
"epoch": 5.02,
"learning_rate": 1.5755698163310467e-06,
"loss": 0.2595,
"step": 4310
},
{
"epoch": 5.03,
"learning_rate": 1.5534410267758354e-06,
"loss": 0.119,
"step": 4320
},
{
"epoch": 5.03,
"learning_rate": 1.531312237220624e-06,
"loss": 0.0869,
"step": 4330
},
{
"epoch": 5.03,
"learning_rate": 1.5091834476654126e-06,
"loss": 0.1115,
"step": 4340
},
{
"epoch": 5.03,
"learning_rate": 1.4870546581102017e-06,
"loss": 0.249,
"step": 4350
},
{
"epoch": 5.03,
"learning_rate": 1.4649258685549901e-06,
"loss": 0.1659,
"step": 4360
},
{
"epoch": 5.04,
"learning_rate": 1.4427970789997788e-06,
"loss": 0.1521,
"step": 4370
},
{
"epoch": 5.04,
"learning_rate": 1.4206682894445674e-06,
"loss": 0.1376,
"step": 4380
},
{
"epoch": 5.04,
"learning_rate": 1.398539499889356e-06,
"loss": 0.1015,
"step": 4390
},
{
"epoch": 5.04,
"learning_rate": 1.3764107103341447e-06,
"loss": 0.2383,
"step": 4400
},
{
"epoch": 5.04,
"learning_rate": 1.3542819207789334e-06,
"loss": 0.1839,
"step": 4410
},
{
"epoch": 5.05,
"learning_rate": 1.332153131223722e-06,
"loss": 0.0749,
"step": 4420
},
{
"epoch": 5.05,
"learning_rate": 1.3100243416685106e-06,
"loss": 0.2221,
"step": 4430
},
{
"epoch": 5.05,
"learning_rate": 1.2878955521132997e-06,
"loss": 0.1028,
"step": 4440
},
{
"epoch": 5.05,
"learning_rate": 1.2657667625580883e-06,
"loss": 0.1556,
"step": 4450
},
{
"epoch": 5.05,
"learning_rate": 1.2436379730028768e-06,
"loss": 0.3584,
"step": 4460
},
{
"epoch": 5.06,
"learning_rate": 1.2215091834476656e-06,
"loss": 0.1174,
"step": 4470
},
{
"epoch": 5.06,
"learning_rate": 1.1993803938924543e-06,
"loss": 0.2224,
"step": 4480
},
{
"epoch": 5.06,
"learning_rate": 1.177251604337243e-06,
"loss": 0.0997,
"step": 4490
},
{
"epoch": 5.06,
"learning_rate": 1.1551228147820316e-06,
"loss": 0.1691,
"step": 4500
},
{
"epoch": 5.06,
"learning_rate": 1.1329940252268202e-06,
"loss": 0.1221,
"step": 4510
},
{
"epoch": 5.07,
"learning_rate": 1.1108652356716088e-06,
"loss": 0.1768,
"step": 4520
},
{
"epoch": 5.07,
"learning_rate": 1.0887364461163975e-06,
"loss": 0.1979,
"step": 4530
},
{
"epoch": 5.07,
"learning_rate": 1.0666076565611861e-06,
"loss": 0.2115,
"step": 4540
},
{
"epoch": 5.07,
"learning_rate": 1.0444788670059748e-06,
"loss": 0.1275,
"step": 4550
},
{
"epoch": 5.07,
"learning_rate": 1.0223500774507636e-06,
"loss": 0.1195,
"step": 4560
},
{
"epoch": 5.08,
"learning_rate": 1.0002212878955523e-06,
"loss": 0.0997,
"step": 4570
},
{
"epoch": 5.08,
"learning_rate": 9.78092498340341e-07,
"loss": 0.1918,
"step": 4580
},
{
"epoch": 5.08,
"learning_rate": 9.559637087851296e-07,
"loss": 0.1582,
"step": 4590
},
{
"epoch": 5.08,
"learning_rate": 9.338349192299182e-07,
"loss": 0.2074,
"step": 4600
},
{
"epoch": 5.08,
"learning_rate": 9.117061296747069e-07,
"loss": 0.3362,
"step": 4610
},
{
"epoch": 5.09,
"learning_rate": 8.895773401194956e-07,
"loss": 0.2022,
"step": 4620
},
{
"epoch": 5.09,
"learning_rate": 8.674485505642842e-07,
"loss": 0.1409,
"step": 4630
},
{
"epoch": 5.09,
"learning_rate": 8.453197610090729e-07,
"loss": 0.1028,
"step": 4640
},
{
"epoch": 5.09,
"learning_rate": 8.231909714538615e-07,
"loss": 0.0855,
"step": 4650
},
{
"epoch": 5.09,
"learning_rate": 8.010621818986503e-07,
"loss": 0.1453,
"step": 4660
},
{
"epoch": 5.1,
"learning_rate": 7.789333923434389e-07,
"loss": 0.1247,
"step": 4670
},
{
"epoch": 5.1,
"learning_rate": 7.568046027882275e-07,
"loss": 0.0749,
"step": 4680
},
{
"epoch": 5.1,
"learning_rate": 7.346758132330162e-07,
"loss": 0.1832,
"step": 4690
},
{
"epoch": 5.1,
"learning_rate": 7.125470236778048e-07,
"loss": 0.0402,
"step": 4700
},
{
"epoch": 5.1,
"learning_rate": 6.904182341225936e-07,
"loss": 0.1689,
"step": 4710
},
{
"epoch": 5.11,
"learning_rate": 6.682894445673822e-07,
"loss": 0.1668,
"step": 4720
},
{
"epoch": 5.11,
"learning_rate": 6.461606550121709e-07,
"loss": 0.0635,
"step": 4730
},
{
"epoch": 5.11,
"learning_rate": 6.240318654569596e-07,
"loss": 0.3186,
"step": 4740
},
{
"epoch": 5.11,
"learning_rate": 6.019030759017483e-07,
"loss": 0.0793,
"step": 4750
},
{
"epoch": 5.11,
"learning_rate": 5.797742863465369e-07,
"loss": 0.1578,
"step": 4760
},
{
"epoch": 5.12,
"learning_rate": 5.576454967913255e-07,
"loss": 0.0247,
"step": 4770
},
{
"epoch": 5.12,
"learning_rate": 5.355167072361142e-07,
"loss": 0.1533,
"step": 4780
},
{
"epoch": 5.12,
"learning_rate": 5.133879176809029e-07,
"loss": 0.089,
"step": 4790
},
{
"epoch": 5.12,
"learning_rate": 4.912591281256916e-07,
"loss": 0.1192,
"step": 4800
},
{
"epoch": 5.12,
"learning_rate": 4.6913033857048026e-07,
"loss": 0.1591,
"step": 4810
},
{
"epoch": 5.13,
"learning_rate": 4.470015490152689e-07,
"loss": 0.169,
"step": 4820
},
{
"epoch": 5.13,
"learning_rate": 4.2487275946005755e-07,
"loss": 0.0843,
"step": 4830
},
{
"epoch": 5.13,
"learning_rate": 4.0274396990484624e-07,
"loss": 0.1278,
"step": 4840
},
{
"epoch": 5.13,
"learning_rate": 3.806151803496349e-07,
"loss": 0.1355,
"step": 4850
},
{
"epoch": 5.13,
"learning_rate": 3.584863907944236e-07,
"loss": 0.1263,
"step": 4860
},
{
"epoch": 5.14,
"learning_rate": 3.363576012392122e-07,
"loss": 0.1327,
"step": 4870
},
{
"epoch": 5.14,
"learning_rate": 3.1422881168400087e-07,
"loss": 0.1344,
"step": 4880
},
{
"epoch": 5.14,
"learning_rate": 2.9210002212878956e-07,
"loss": 0.0959,
"step": 4890
},
{
"epoch": 5.14,
"learning_rate": 2.6997123257357826e-07,
"loss": 0.1882,
"step": 4900
},
{
"epoch": 5.14,
"learning_rate": 2.478424430183669e-07,
"loss": 0.2254,
"step": 4910
},
{
"epoch": 5.15,
"learning_rate": 2.2571365346315557e-07,
"loss": 0.0817,
"step": 4920
},
{
"epoch": 5.15,
"learning_rate": 2.0358486390794424e-07,
"loss": 0.1513,
"step": 4930
},
{
"epoch": 5.15,
"learning_rate": 1.8145607435273293e-07,
"loss": 0.0656,
"step": 4940
},
{
"epoch": 5.15,
"learning_rate": 1.593272847975216e-07,
"loss": 0.1388,
"step": 4950
},
{
"epoch": 5.15,
"learning_rate": 1.3719849524231024e-07,
"loss": 0.2887,
"step": 4960
},
{
"epoch": 5.16,
"learning_rate": 1.1506970568709893e-07,
"loss": 0.2351,
"step": 4970
},
{
"epoch": 5.16,
"learning_rate": 9.294091613188761e-08,
"loss": 0.1093,
"step": 4980
},
{
"epoch": 5.16,
"learning_rate": 7.081212657667626e-08,
"loss": 0.231,
"step": 4990
},
{
"epoch": 5.16,
"learning_rate": 4.868333702146493e-08,
"loss": 0.0945,
"step": 5000
},
{
"epoch": 5.16,
"learning_rate": 2.6554547466253597e-08,
"loss": 0.119,
"step": 5010
},
{
"epoch": 5.17,
"learning_rate": 4.4257579110422664e-09,
"loss": 0.1655,
"step": 5020
},
{
"epoch": 5.17,
"eval_accuracy": 0.8541748333986672,
"eval_loss": 0.5614621043205261,
"eval_runtime": 458.2183,
"eval_samples_per_second": 5.567,
"eval_steps_per_second": 0.465,
"step": 5022
},
{
"epoch": 5.17,
"step": 5022,
"total_flos": 7.508660609697782e+19,
"train_loss": 0.24504626986812852,
"train_runtime": 16648.2238,
"train_samples_per_second": 3.62,
"train_steps_per_second": 0.302
},
{
"epoch": 5.17,
"eval_accuracy": 0.8643669149353195,
"eval_loss": 0.4997994899749756,
"eval_runtime": 558.362,
"eval_samples_per_second": 4.569,
"eval_steps_per_second": 0.381,
"step": 5022
},
{
"epoch": 5.17,
"eval_accuracy": 0.5961538461538461,
"eval_loss": 1.6456845998764038,
"eval_runtime": 40.3879,
"eval_samples_per_second": 2.575,
"eval_steps_per_second": 0.223,
"step": 5022
}
],
"logging_steps": 10,
"max_steps": 5022,
"num_train_epochs": 9223372036854775807,
"save_steps": 500,
"total_flos": 7.508660609697782e+19,
"trial_name": null,
"trial_params": null
}