depo_base_250K_lora / trainer_state.json
laoda's picture
Upload folder using huggingface_hub
5d8c551
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.9992321474276937,
"global_step": 3906,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 8.474576271186441e-07,
"loss": 10.048,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 8.47457627118644e-06,
"loss": 10.0462,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 1.694915254237288e-05,
"loss": 9.494,
"step": 20
},
{
"epoch": 0.02,
"learning_rate": 2.5423728813559322e-05,
"loss": 8.5624,
"step": 30
},
{
"epoch": 0.03,
"learning_rate": 3.389830508474576e-05,
"loss": 7.472,
"step": 40
},
{
"epoch": 0.04,
"learning_rate": 4.2372881355932206e-05,
"loss": 6.1188,
"step": 50
},
{
"epoch": 0.05,
"learning_rate": 5.0847457627118643e-05,
"loss": 5.1402,
"step": 60
},
{
"epoch": 0.05,
"learning_rate": 5.932203389830509e-05,
"loss": 4.7902,
"step": 70
},
{
"epoch": 0.06,
"learning_rate": 6.779661016949152e-05,
"loss": 4.4874,
"step": 80
},
{
"epoch": 0.07,
"learning_rate": 7.627118644067796e-05,
"loss": 4.1484,
"step": 90
},
{
"epoch": 0.08,
"learning_rate": 8.474576271186441e-05,
"loss": 3.8374,
"step": 100
},
{
"epoch": 0.08,
"learning_rate": 9.322033898305085e-05,
"loss": 3.6236,
"step": 110
},
{
"epoch": 0.09,
"learning_rate": 9.999993121722311e-05,
"loss": 3.4008,
"step": 120
},
{
"epoch": 0.1,
"learning_rate": 9.999752383990224e-05,
"loss": 3.2626,
"step": 130
},
{
"epoch": 0.11,
"learning_rate": 9.999167751297713e-05,
"loss": 3.1466,
"step": 140
},
{
"epoch": 0.12,
"learning_rate": 9.998239263857216e-05,
"loss": 3.0094,
"step": 150
},
{
"epoch": 0.12,
"learning_rate": 9.996966985532326e-05,
"loss": 2.9188,
"step": 160
},
{
"epoch": 0.13,
"learning_rate": 9.995351003833398e-05,
"loss": 2.8857,
"step": 170
},
{
"epoch": 0.14,
"learning_rate": 9.993391429911529e-05,
"loss": 2.8134,
"step": 180
},
{
"epoch": 0.15,
"learning_rate": 9.991088398550913e-05,
"loss": 2.7154,
"step": 190
},
{
"epoch": 0.15,
"learning_rate": 9.98844206815957e-05,
"loss": 2.6984,
"step": 200
},
{
"epoch": 0.16,
"learning_rate": 9.98607795006937e-05,
"loss": 2.6104,
"step": 210
},
{
"epoch": 0.17,
"learning_rate": 9.983155962119652e-05,
"loss": 2.5924,
"step": 220
},
{
"epoch": 0.18,
"learning_rate": 9.979583700334007e-05,
"loss": 2.5528,
"step": 230
},
{
"epoch": 0.18,
"learning_rate": 9.975668930838371e-05,
"loss": 2.5238,
"step": 240
},
{
"epoch": 0.19,
"learning_rate": 9.97141192289998e-05,
"loss": 2.4679,
"step": 250
},
{
"epoch": 0.2,
"learning_rate": 9.966812969326051e-05,
"loss": 2.4443,
"step": 260
},
{
"epoch": 0.21,
"learning_rate": 9.961872386443639e-05,
"loss": 2.4029,
"step": 270
},
{
"epoch": 0.21,
"learning_rate": 9.956590514077886e-05,
"loss": 2.341,
"step": 280
},
{
"epoch": 0.22,
"learning_rate": 9.95096771552864e-05,
"loss": 2.319,
"step": 290
},
{
"epoch": 0.23,
"learning_rate": 9.945004377545474e-05,
"loss": 2.2927,
"step": 300
},
{
"epoch": 0.24,
"learning_rate": 9.938700910301073e-05,
"loss": 2.2734,
"step": 310
},
{
"epoch": 0.25,
"learning_rate": 9.932057747363035e-05,
"loss": 2.2246,
"step": 320
},
{
"epoch": 0.25,
"learning_rate": 9.92507534566404e-05,
"loss": 2.237,
"step": 330
},
{
"epoch": 0.26,
"learning_rate": 9.91775418547042e-05,
"loss": 2.1853,
"step": 340
},
{
"epoch": 0.27,
"learning_rate": 9.910094770349136e-05,
"loss": 2.1479,
"step": 350
},
{
"epoch": 0.28,
"learning_rate": 9.902097627133131e-05,
"loss": 2.1409,
"step": 360
},
{
"epoch": 0.28,
"learning_rate": 9.893763305885094e-05,
"loss": 2.0847,
"step": 370
},
{
"epoch": 0.29,
"learning_rate": 9.885092379859627e-05,
"loss": 2.0763,
"step": 380
},
{
"epoch": 0.3,
"learning_rate": 9.876085445463823e-05,
"loss": 2.0605,
"step": 390
},
{
"epoch": 0.31,
"learning_rate": 9.866743122216229e-05,
"loss": 2.0667,
"step": 400
},
{
"epoch": 0.31,
"learning_rate": 9.857066052704247e-05,
"loss": 1.989,
"step": 410
},
{
"epoch": 0.32,
"learning_rate": 9.849083826303131e-05,
"loss": 1.978,
"step": 420
},
{
"epoch": 0.33,
"learning_rate": 9.838805906300996e-05,
"loss": 1.9588,
"step": 430
},
{
"epoch": 0.34,
"learning_rate": 9.828195161623185e-05,
"loss": 1.9445,
"step": 440
},
{
"epoch": 0.35,
"learning_rate": 9.817252322102168e-05,
"loss": 1.9124,
"step": 450
},
{
"epoch": 0.35,
"learning_rate": 9.805978140412695e-05,
"loss": 1.8489,
"step": 460
},
{
"epoch": 0.36,
"learning_rate": 9.795548719790117e-05,
"loss": 1.8561,
"step": 470
},
{
"epoch": 0.37,
"learning_rate": 9.783647143180754e-05,
"loss": 1.82,
"step": 480
},
{
"epoch": 0.38,
"learning_rate": 9.771416535847738e-05,
"loss": 1.8393,
"step": 490
},
{
"epoch": 0.38,
"learning_rate": 9.758857739041575e-05,
"loss": 1.7466,
"step": 500
},
{
"epoch": 0.39,
"learning_rate": 9.745971616586433e-05,
"loss": 1.7376,
"step": 510
},
{
"epoch": 0.4,
"learning_rate": 9.732759054820718e-05,
"loss": 1.713,
"step": 520
},
{
"epoch": 0.41,
"learning_rate": 9.719220962536115e-05,
"loss": 1.7354,
"step": 530
},
{
"epoch": 0.41,
"learning_rate": 9.705358270915086e-05,
"loss": 1.6935,
"step": 540
},
{
"epoch": 0.42,
"learning_rate": 9.691171933466809e-05,
"loss": 1.6771,
"step": 550
},
{
"epoch": 0.43,
"learning_rate": 9.676662925961595e-05,
"loss": 1.6194,
"step": 560
},
{
"epoch": 0.44,
"learning_rate": 9.663329760668263e-05,
"loss": 1.6364,
"step": 570
},
{
"epoch": 0.45,
"learning_rate": 9.648210447734683e-05,
"loss": 1.6002,
"step": 580
},
{
"epoch": 0.45,
"learning_rate": 9.632771419738195e-05,
"loss": 1.5933,
"step": 590
},
{
"epoch": 0.46,
"learning_rate": 9.617013738612175e-05,
"loss": 1.5974,
"step": 600
},
{
"epoch": 0.47,
"learning_rate": 9.602560272524263e-05,
"loss": 1.5245,
"step": 610
},
{
"epoch": 0.48,
"learning_rate": 9.586200154516864e-05,
"loss": 1.5327,
"step": 620
},
{
"epoch": 0.48,
"learning_rate": 9.569524586663253e-05,
"loss": 1.5931,
"step": 630
},
{
"epoch": 0.49,
"learning_rate": 9.552534715948977e-05,
"loss": 1.4526,
"step": 640
},
{
"epoch": 0.5,
"learning_rate": 9.535231710978097e-05,
"loss": 1.5182,
"step": 650
},
{
"epoch": 0.51,
"learning_rate": 9.517616761892793e-05,
"loss": 1.4276,
"step": 660
},
{
"epoch": 0.51,
"learning_rate": 9.499691080291511e-05,
"loss": 1.4381,
"step": 670
},
{
"epoch": 0.52,
"learning_rate": 9.481455899145624e-05,
"loss": 1.4444,
"step": 680
},
{
"epoch": 0.53,
"learning_rate": 9.462912472714627e-05,
"loss": 1.4319,
"step": 690
},
{
"epoch": 0.54,
"learning_rate": 9.444062076459867e-05,
"loss": 1.3555,
"step": 700
},
{
"epoch": 0.55,
"learning_rate": 9.424906006956805e-05,
"loss": 1.4182,
"step": 710
},
{
"epoch": 0.55,
"learning_rate": 9.407405282340516e-05,
"loss": 1.3665,
"step": 720
},
{
"epoch": 0.56,
"learning_rate": 9.389659004412879e-05,
"loss": 1.4031,
"step": 730
},
{
"epoch": 0.57,
"learning_rate": 9.369654126212313e-05,
"loss": 1.3698,
"step": 740
},
{
"epoch": 0.58,
"learning_rate": 9.349348692720728e-05,
"loss": 1.2911,
"step": 750
},
{
"epoch": 0.58,
"learning_rate": 9.328744100594535e-05,
"loss": 1.3292,
"step": 760
},
{
"epoch": 0.59,
"learning_rate": 9.307841767067004e-05,
"loss": 1.2707,
"step": 770
},
{
"epoch": 0.6,
"learning_rate": 9.286643129850765e-05,
"loss": 1.2704,
"step": 780
},
{
"epoch": 0.61,
"learning_rate": 9.26514964703893e-05,
"loss": 1.2859,
"step": 790
},
{
"epoch": 0.61,
"learning_rate": 9.243362797004795e-05,
"loss": 1.2781,
"step": 800
},
{
"epoch": 0.62,
"learning_rate": 9.22128407830016e-05,
"loss": 1.2209,
"step": 810
},
{
"epoch": 0.63,
"learning_rate": 9.198915009552253e-05,
"loss": 1.313,
"step": 820
},
{
"epoch": 0.64,
"learning_rate": 9.176257129359271e-05,
"loss": 1.2289,
"step": 830
},
{
"epoch": 0.64,
"learning_rate": 9.153311996184557e-05,
"loss": 1.2909,
"step": 840
},
{
"epoch": 0.65,
"learning_rate": 9.130081188249401e-05,
"loss": 1.2096,
"step": 850
},
{
"epoch": 0.66,
"learning_rate": 9.106566303424492e-05,
"loss": 1.2095,
"step": 860
},
{
"epoch": 0.67,
"learning_rate": 9.082768959120005e-05,
"loss": 1.2266,
"step": 870
},
{
"epoch": 0.68,
"learning_rate": 9.058690792174358e-05,
"loss": 1.2021,
"step": 880
},
{
"epoch": 0.68,
"learning_rate": 9.036781706979148e-05,
"loss": 1.1993,
"step": 890
},
{
"epoch": 0.69,
"learning_rate": 9.012174555592377e-05,
"loss": 1.1583,
"step": 900
},
{
"epoch": 0.7,
"learning_rate": 8.987291437216836e-05,
"loss": 1.1689,
"step": 910
},
{
"epoch": 0.71,
"learning_rate": 8.964662093084879e-05,
"loss": 1.201,
"step": 920
},
{
"epoch": 0.71,
"learning_rate": 8.939259368263329e-05,
"loss": 1.1071,
"step": 930
},
{
"epoch": 0.72,
"learning_rate": 8.913585691734469e-05,
"loss": 1.1394,
"step": 940
},
{
"epoch": 0.73,
"learning_rate": 8.887642829395353e-05,
"loss": 1.1179,
"step": 950
},
{
"epoch": 0.74,
"learning_rate": 8.86143256565827e-05,
"loss": 1.1696,
"step": 960
},
{
"epoch": 0.74,
"learning_rate": 8.834956703328026e-05,
"loss": 1.149,
"step": 970
},
{
"epoch": 0.75,
"learning_rate": 8.80821706347793e-05,
"loss": 1.1715,
"step": 980
},
{
"epoch": 0.76,
"learning_rate": 8.781215485324544e-05,
"loss": 1.1065,
"step": 990
},
{
"epoch": 0.77,
"learning_rate": 8.753953826101173e-05,
"loss": 1.1118,
"step": 1000
},
{
"epoch": 0.78,
"learning_rate": 8.726433960930126e-05,
"loss": 1.095,
"step": 1010
},
{
"epoch": 0.78,
"learning_rate": 8.698657782693731e-05,
"loss": 1.1179,
"step": 1020
},
{
"epoch": 0.79,
"learning_rate": 8.670627201904144e-05,
"loss": 1.084,
"step": 1030
},
{
"epoch": 0.8,
"learning_rate": 8.642344146571946e-05,
"loss": 1.1241,
"step": 1040
},
{
"epoch": 0.81,
"learning_rate": 8.613810562073512e-05,
"loss": 1.1366,
"step": 1050
},
{
"epoch": 0.81,
"learning_rate": 8.585028411017222e-05,
"loss": 1.0688,
"step": 1060
},
{
"epoch": 0.82,
"learning_rate": 8.555999673108455e-05,
"loss": 1.075,
"step": 1070
},
{
"epoch": 0.83,
"learning_rate": 8.526726345013429e-05,
"loss": 1.0886,
"step": 1080
},
{
"epoch": 0.84,
"learning_rate": 8.497210440221857e-05,
"loss": 1.1097,
"step": 1090
},
{
"epoch": 0.84,
"learning_rate": 8.467453988908465e-05,
"loss": 1.0666,
"step": 1100
},
{
"epoch": 0.85,
"learning_rate": 8.437459037793342e-05,
"loss": 1.0876,
"step": 1110
},
{
"epoch": 0.86,
"learning_rate": 8.413292742916203e-05,
"loss": 1.1324,
"step": 1120
},
{
"epoch": 0.87,
"learning_rate": 8.38287370216547e-05,
"loss": 1.0721,
"step": 1130
},
{
"epoch": 0.88,
"learning_rate": 8.352221979247983e-05,
"loss": 1.0975,
"step": 1140
},
{
"epoch": 0.88,
"learning_rate": 8.321339682462765e-05,
"loss": 1.0923,
"step": 1150
},
{
"epoch": 0.89,
"learning_rate": 8.290228935968252e-05,
"loss": 1.0432,
"step": 1160
},
{
"epoch": 0.9,
"learning_rate": 8.258891879636209e-05,
"loss": 1.0721,
"step": 1170
},
{
"epoch": 0.91,
"learning_rate": 8.227330668904527e-05,
"loss": 1.0608,
"step": 1180
},
{
"epoch": 0.91,
"learning_rate": 8.195547474628975e-05,
"loss": 1.0219,
"step": 1190
},
{
"epoch": 0.92,
"learning_rate": 8.163544482933888e-05,
"loss": 1.0834,
"step": 1200
},
{
"epoch": 0.93,
"learning_rate": 8.131323895061789e-05,
"loss": 1.0163,
"step": 1210
},
{
"epoch": 0.94,
"learning_rate": 8.105392244310324e-05,
"loss": 1.0814,
"step": 1220
},
{
"epoch": 0.94,
"learning_rate": 8.072785578127462e-05,
"loss": 1.051,
"step": 1230
},
{
"epoch": 0.95,
"learning_rate": 8.039967558382739e-05,
"loss": 1.0716,
"step": 1240
},
{
"epoch": 0.96,
"learning_rate": 8.006940442378264e-05,
"loss": 1.0664,
"step": 1250
},
{
"epoch": 0.97,
"learning_rate": 7.973706501798287e-05,
"loss": 1.0242,
"step": 1260
},
{
"epoch": 0.98,
"learning_rate": 7.94026802255295e-05,
"loss": 0.9976,
"step": 1270
},
{
"epoch": 0.98,
"learning_rate": 7.906627304621055e-05,
"loss": 1.0184,
"step": 1280
},
{
"epoch": 0.99,
"learning_rate": 7.872786661891866e-05,
"loss": 1.0965,
"step": 1290
},
{
"epoch": 1.0,
"learning_rate": 7.838748422005952e-05,
"loss": 0.9947,
"step": 1300
},
{
"epoch": 1.01,
"learning_rate": 7.804514926195093e-05,
"loss": 0.9582,
"step": 1310
},
{
"epoch": 1.01,
"learning_rate": 7.77698912716615e-05,
"loss": 0.9857,
"step": 1320
},
{
"epoch": 1.02,
"learning_rate": 7.742410113360196e-05,
"loss": 0.9274,
"step": 1330
},
{
"epoch": 1.03,
"learning_rate": 7.707642470009227e-05,
"loss": 1.0231,
"step": 1340
},
{
"epoch": 1.04,
"learning_rate": 7.672688588515131e-05,
"loss": 1.0032,
"step": 1350
},
{
"epoch": 1.04,
"learning_rate": 7.637550873089718e-05,
"loss": 0.9494,
"step": 1360
},
{
"epoch": 1.05,
"learning_rate": 7.602231740589323e-05,
"loss": 0.9802,
"step": 1370
},
{
"epoch": 1.06,
"learning_rate": 7.566733620348596e-05,
"loss": 0.9629,
"step": 1380
},
{
"epoch": 1.07,
"learning_rate": 7.531058954013384e-05,
"loss": 0.9837,
"step": 1390
},
{
"epoch": 1.07,
"learning_rate": 7.495210195372795e-05,
"loss": 0.9554,
"step": 1400
},
{
"epoch": 1.08,
"learning_rate": 7.459189810190437e-05,
"loss": 0.9801,
"step": 1410
},
{
"epoch": 1.09,
"learning_rate": 7.426626770302272e-05,
"loss": 0.9787,
"step": 1420
},
{
"epoch": 1.1,
"learning_rate": 7.393928533807334e-05,
"loss": 0.8972,
"step": 1430
},
{
"epoch": 1.11,
"learning_rate": 7.357440812000299e-05,
"loss": 1.0171,
"step": 1440
},
{
"epoch": 1.11,
"learning_rate": 7.320790939760105e-05,
"loss": 0.986,
"step": 1450
},
{
"epoch": 1.12,
"learning_rate": 7.283981437952866e-05,
"loss": 0.9629,
"step": 1460
},
{
"epoch": 1.13,
"learning_rate": 7.247014838424404e-05,
"loss": 1.007,
"step": 1470
},
{
"epoch": 1.14,
"learning_rate": 7.20989368382609e-05,
"loss": 0.9969,
"step": 1480
},
{
"epoch": 1.14,
"learning_rate": 7.172620527439976e-05,
"loss": 0.9416,
"step": 1490
},
{
"epoch": 1.15,
"learning_rate": 7.135197933003149e-05,
"loss": 0.9568,
"step": 1500
},
{
"epoch": 1.16,
"learning_rate": 7.09762847453141e-05,
"loss": 0.9682,
"step": 1510
},
{
"epoch": 1.17,
"learning_rate": 7.059914736142222e-05,
"loss": 0.93,
"step": 1520
},
{
"epoch": 1.17,
"learning_rate": 7.029641606757223e-05,
"loss": 0.983,
"step": 1530
},
{
"epoch": 1.18,
"learning_rate": 6.991674708055885e-05,
"loss": 0.9451,
"step": 1540
},
{
"epoch": 1.19,
"learning_rate": 6.953570817191274e-05,
"loss": 0.9834,
"step": 1550
},
{
"epoch": 1.2,
"learning_rate": 6.915332555040396e-05,
"loss": 0.9999,
"step": 1560
},
{
"epoch": 1.21,
"learning_rate": 6.87696255172263e-05,
"loss": 1.0262,
"step": 1570
},
{
"epoch": 1.21,
"learning_rate": 6.83846344641883e-05,
"loss": 0.9003,
"step": 1580
},
{
"epoch": 1.22,
"learning_rate": 6.799837887189797e-05,
"loss": 0.944,
"step": 1590
},
{
"epoch": 1.23,
"learning_rate": 6.761088530794136e-05,
"loss": 0.9244,
"step": 1600
},
{
"epoch": 1.24,
"learning_rate": 6.722218042505515e-05,
"loss": 0.9541,
"step": 1610
},
{
"epoch": 1.24,
"learning_rate": 6.683229095929347e-05,
"loss": 0.9495,
"step": 1620
},
{
"epoch": 1.25,
"learning_rate": 6.651954450570508e-05,
"loss": 0.9748,
"step": 1630
},
{
"epoch": 1.26,
"learning_rate": 6.612759042452568e-05,
"loss": 0.9887,
"step": 1640
},
{
"epoch": 1.27,
"learning_rate": 6.573452704899633e-05,
"loss": 0.9982,
"step": 1650
},
{
"epoch": 1.27,
"learning_rate": 6.53403814149587e-05,
"loss": 1.0086,
"step": 1660
},
{
"epoch": 1.28,
"learning_rate": 6.494518063269486e-05,
"loss": 0.9632,
"step": 1670
},
{
"epoch": 1.29,
"learning_rate": 6.454895188506242e-05,
"loss": 1.012,
"step": 1680
},
{
"epoch": 1.3,
"learning_rate": 6.419148962544263e-05,
"loss": 1.0013,
"step": 1690
},
{
"epoch": 1.31,
"learning_rate": 6.379338288410547e-05,
"loss": 0.9369,
"step": 1700
},
{
"epoch": 1.31,
"learning_rate": 6.339432740081137e-05,
"loss": 0.9435,
"step": 1710
},
{
"epoch": 1.32,
"learning_rate": 6.299435062355356e-05,
"loss": 0.9789,
"step": 1720
},
{
"epoch": 1.33,
"learning_rate": 6.25934800636941e-05,
"loss": 0.9898,
"step": 1730
},
{
"epoch": 1.34,
"learning_rate": 6.219174329407153e-05,
"loss": 0.9309,
"step": 1740
},
{
"epoch": 1.34,
"learning_rate": 6.17891679471044e-05,
"loss": 0.8868,
"step": 1750
},
{
"epoch": 1.35,
"learning_rate": 6.13857817128906e-05,
"loss": 0.9177,
"step": 1760
},
{
"epoch": 1.36,
"learning_rate": 6.098161233730284e-05,
"loss": 0.985,
"step": 1770
},
{
"epoch": 1.37,
"learning_rate": 6.057668762008011e-05,
"loss": 0.9585,
"step": 1780
},
{
"epoch": 1.37,
"learning_rate": 6.017103541291565e-05,
"loss": 0.893,
"step": 1790
},
{
"epoch": 1.38,
"learning_rate": 5.9764683617541175e-05,
"loss": 0.9716,
"step": 1800
},
{
"epoch": 1.39,
"learning_rate": 5.9357660183807776e-05,
"loss": 0.9447,
"step": 1810
},
{
"epoch": 1.4,
"learning_rate": 5.8949993107763425e-05,
"loss": 0.9263,
"step": 1820
},
{
"epoch": 1.41,
"learning_rate": 5.854171042972738e-05,
"loss": 0.921,
"step": 1830
},
{
"epoch": 1.41,
"learning_rate": 5.813284023236146e-05,
"loss": 0.9644,
"step": 1840
},
{
"epoch": 1.42,
"learning_rate": 5.7723410638738477e-05,
"loss": 0.931,
"step": 1850
},
{
"epoch": 1.43,
"learning_rate": 5.7313449810407846e-05,
"loss": 0.9491,
"step": 1860
},
{
"epoch": 1.44,
"learning_rate": 5.690298594545862e-05,
"loss": 0.9849,
"step": 1870
},
{
"epoch": 1.44,
"learning_rate": 5.649204727657988e-05,
"loss": 0.8915,
"step": 1880
},
{
"epoch": 1.45,
"learning_rate": 5.6162973475913305e-05,
"loss": 0.9784,
"step": 1890
},
{
"epoch": 1.46,
"learning_rate": 5.575125140934168e-05,
"loss": 0.9278,
"step": 1900
},
{
"epoch": 1.47,
"learning_rate": 5.533913375790425e-05,
"loss": 0.9365,
"step": 1910
},
{
"epoch": 1.47,
"learning_rate": 5.492664886804154e-05,
"loss": 0.9043,
"step": 1920
},
{
"epoch": 1.48,
"learning_rate": 5.451382511145355e-05,
"loss": 0.9631,
"step": 1930
},
{
"epoch": 1.49,
"learning_rate": 5.41006908831484e-05,
"loss": 0.9742,
"step": 1940
},
{
"epoch": 1.5,
"learning_rate": 5.372862811006992e-05,
"loss": 0.9232,
"step": 1950
},
{
"epoch": 1.5,
"learning_rate": 5.331498228865735e-05,
"loss": 0.8755,
"step": 1960
},
{
"epoch": 1.51,
"learning_rate": 5.290110845481224e-05,
"loss": 0.9385,
"step": 1970
},
{
"epoch": 1.52,
"learning_rate": 5.2487035075769594e-05,
"loss": 0.9194,
"step": 1980
},
{
"epoch": 1.53,
"learning_rate": 5.2072790632489497e-05,
"loss": 0.9309,
"step": 1990
},
{
"epoch": 1.54,
"learning_rate": 5.1658403617698304e-05,
"loss": 0.9154,
"step": 2000
},
{
"epoch": 1.54,
"learning_rate": 5.1243902533928754e-05,
"loss": 0.9985,
"step": 2010
},
{
"epoch": 1.55,
"learning_rate": 5.082931589155952e-05,
"loss": 0.9177,
"step": 2020
},
{
"epoch": 1.56,
"learning_rate": 5.041467220685424e-05,
"loss": 0.9511,
"step": 2030
},
{
"epoch": 1.57,
"learning_rate": 5e-05,
"loss": 0.9279,
"step": 2040
},
{
"epoch": 1.57,
"learning_rate": 4.958532779314578e-05,
"loss": 0.893,
"step": 2050
},
{
"epoch": 1.58,
"learning_rate": 4.9170684108440495e-05,
"loss": 0.8912,
"step": 2060
},
{
"epoch": 1.59,
"learning_rate": 4.875609746607127e-05,
"loss": 0.8978,
"step": 2070
},
{
"epoch": 1.6,
"learning_rate": 4.834159638230172e-05,
"loss": 1.0025,
"step": 2080
},
{
"epoch": 1.6,
"learning_rate": 4.79272093675105e-05,
"loss": 0.95,
"step": 2090
},
{
"epoch": 1.61,
"learning_rate": 4.751296492423041e-05,
"loss": 0.9162,
"step": 2100
},
{
"epoch": 1.62,
"learning_rate": 4.7098891545187755e-05,
"loss": 0.9056,
"step": 2110
},
{
"epoch": 1.63,
"learning_rate": 4.6685017711342664e-05,
"loss": 0.9234,
"step": 2120
},
{
"epoch": 1.64,
"learning_rate": 4.627137188993009e-05,
"loss": 0.9279,
"step": 2130
},
{
"epoch": 1.64,
"learning_rate": 4.585798253250169e-05,
"loss": 0.9338,
"step": 2140
},
{
"epoch": 1.65,
"learning_rate": 4.5444878072968963e-05,
"loss": 0.8987,
"step": 2150
},
{
"epoch": 1.66,
"learning_rate": 4.511461872695602e-05,
"loss": 0.9162,
"step": 2160
},
{
"epoch": 1.67,
"learning_rate": 4.470209867329438e-05,
"loss": 0.8368,
"step": 2170
},
{
"epoch": 1.67,
"learning_rate": 4.428994302199254e-05,
"loss": 0.9068,
"step": 2180
},
{
"epoch": 1.68,
"learning_rate": 4.387818012210471e-05,
"loss": 0.9102,
"step": 2190
},
{
"epoch": 1.69,
"learning_rate": 4.34668382956707e-05,
"loss": 0.9386,
"step": 2200
},
{
"epoch": 1.7,
"learning_rate": 4.3055945835767916e-05,
"loss": 0.9032,
"step": 2210
},
{
"epoch": 1.7,
"learning_rate": 4.264553100456523e-05,
"loss": 0.8746,
"step": 2220
},
{
"epoch": 1.71,
"learning_rate": 4.223562203137906e-05,
"loss": 0.9167,
"step": 2230
},
{
"epoch": 1.72,
"learning_rate": 4.1826247110731755e-05,
"loss": 0.8817,
"step": 2240
},
{
"epoch": 1.73,
"learning_rate": 4.141743440041218e-05,
"loss": 0.8855,
"step": 2250
},
{
"epoch": 1.74,
"learning_rate": 4.109080792098898e-05,
"loss": 0.9149,
"step": 2260
},
{
"epoch": 1.74,
"learning_rate": 4.068307802221734e-05,
"loss": 0.927,
"step": 2270
},
{
"epoch": 1.75,
"learning_rate": 4.027598896368508e-05,
"loss": 0.8904,
"step": 2280
},
{
"epoch": 1.76,
"learning_rate": 3.986956874595409e-05,
"loss": 0.8804,
"step": 2290
},
{
"epoch": 1.77,
"learning_rate": 3.946384532358167e-05,
"loss": 0.8714,
"step": 2300
},
{
"epoch": 1.77,
"learning_rate": 3.905884660319792e-05,
"loss": 0.9104,
"step": 2310
},
{
"epoch": 1.78,
"learning_rate": 3.8654600441586154e-05,
"loss": 0.9344,
"step": 2320
},
{
"epoch": 1.79,
"learning_rate": 3.825113464376693e-05,
"loss": 0.9403,
"step": 2330
},
{
"epoch": 1.8,
"learning_rate": 3.788870557439895e-05,
"loss": 0.9546,
"step": 2340
},
{
"epoch": 1.8,
"learning_rate": 3.7486798876840846e-05,
"loss": 0.9063,
"step": 2350
},
{
"epoch": 1.81,
"learning_rate": 3.7085752867267784e-05,
"loss": 0.9396,
"step": 2360
},
{
"epoch": 1.82,
"learning_rate": 3.668559513058615e-05,
"loss": 0.8966,
"step": 2370
},
{
"epoch": 1.83,
"learning_rate": 3.6286353190604846e-05,
"loss": 0.9468,
"step": 2380
},
{
"epoch": 1.84,
"learning_rate": 3.5888054508142025e-05,
"loss": 0.9004,
"step": 2390
},
{
"epoch": 1.84,
"learning_rate": 3.5490726479136336e-05,
"loss": 0.893,
"step": 2400
},
{
"epoch": 1.85,
"learning_rate": 3.5094396432762614e-05,
"loss": 0.9175,
"step": 2410
},
{
"epoch": 1.86,
"learning_rate": 3.469909162955198e-05,
"loss": 0.8904,
"step": 2420
},
{
"epoch": 1.87,
"learning_rate": 3.430483925951693e-05,
"loss": 0.9208,
"step": 2430
},
{
"epoch": 1.87,
"learning_rate": 3.3911666440281014e-05,
"loss": 0.9006,
"step": 2440
},
{
"epoch": 1.88,
"learning_rate": 3.351960021521376e-05,
"loss": 0.9156,
"step": 2450
},
{
"epoch": 1.89,
"learning_rate": 3.312866755157045e-05,
"loss": 0.8406,
"step": 2460
},
{
"epoch": 1.9,
"learning_rate": 3.27388953386373e-05,
"loss": 0.8823,
"step": 2470
},
{
"epoch": 1.9,
"learning_rate": 3.235031038588197e-05,
"loss": 0.9067,
"step": 2480
},
{
"epoch": 1.91,
"learning_rate": 3.196293942110945e-05,
"loss": 0.8937,
"step": 2490
},
{
"epoch": 1.92,
"learning_rate": 3.157680908862383e-05,
"loss": 0.9092,
"step": 2500
},
{
"epoch": 1.93,
"learning_rate": 3.119194594739544e-05,
"loss": 0.8983,
"step": 2510
},
{
"epoch": 1.93,
"learning_rate": 3.080837646923422e-05,
"loss": 0.9265,
"step": 2520
},
{
"epoch": 1.94,
"learning_rate": 3.0426127036968804e-05,
"loss": 0.8691,
"step": 2530
},
{
"epoch": 1.95,
"learning_rate": 3.012129559554442e-05,
"loss": 0.8581,
"step": 2540
},
{
"epoch": 1.96,
"learning_rate": 2.9741488439644472e-05,
"loss": 0.8933,
"step": 2550
},
{
"epoch": 1.97,
"learning_rate": 2.9363074712757848e-05,
"loss": 0.8743,
"step": 2560
},
{
"epoch": 1.97,
"learning_rate": 2.8986080443088294e-05,
"loss": 0.8545,
"step": 2570
},
{
"epoch": 1.98,
"learning_rate": 2.861053156120582e-05,
"loss": 0.831,
"step": 2580
},
{
"epoch": 1.99,
"learning_rate": 2.8236453898263293e-05,
"loss": 0.8708,
"step": 2590
},
{
"epoch": 2.0,
"learning_rate": 2.7863873184219548e-05,
"loss": 0.9156,
"step": 2600
},
{
"epoch": 2.0,
"learning_rate": 2.7492815046069658e-05,
"loss": 0.827,
"step": 2610
},
{
"epoch": 2.01,
"learning_rate": 2.7123305006082373e-05,
"loss": 0.8013,
"step": 2620
},
{
"epoch": 2.02,
"learning_rate": 2.675536848004443e-05,
"loss": 0.8654,
"step": 2630
},
{
"epoch": 2.03,
"learning_rate": 2.6389030775512595e-05,
"loss": 0.8614,
"step": 2640
},
{
"epoch": 2.03,
"learning_rate": 2.609712869988859e-05,
"loss": 0.8475,
"step": 2650
},
{
"epoch": 2.04,
"learning_rate": 2.5733732296977274e-05,
"loss": 0.8365,
"step": 2660
},
{
"epoch": 2.05,
"learning_rate": 2.537200498615905e-05,
"loss": 0.8436,
"step": 2670
},
{
"epoch": 2.06,
"learning_rate": 2.5011971647905863e-05,
"loss": 0.8814,
"step": 2680
},
{
"epoch": 2.07,
"learning_rate": 2.465365704617426e-05,
"loss": 0.884,
"step": 2690
},
{
"epoch": 2.07,
"learning_rate": 2.4297085826701955e-05,
"loss": 0.8301,
"step": 2700
},
{
"epoch": 2.08,
"learning_rate": 2.3942282515312602e-05,
"loss": 0.8606,
"step": 2710
},
{
"epoch": 2.09,
"learning_rate": 2.3589271516228928e-05,
"loss": 0.8943,
"step": 2720
},
{
"epoch": 2.1,
"learning_rate": 2.3238077110394086e-05,
"loss": 0.869,
"step": 2730
},
{
"epoch": 2.1,
"learning_rate": 2.2888723453801665e-05,
"loss": 0.8941,
"step": 2740
},
{
"epoch": 2.11,
"learning_rate": 2.2610582020023592e-05,
"loss": 0.8861,
"step": 2750
},
{
"epoch": 2.12,
"learning_rate": 2.2264602179973053e-05,
"loss": 0.8388,
"step": 2760
},
{
"epoch": 2.13,
"learning_rate": 2.1920530047105403e-05,
"loss": 0.8237,
"step": 2770
},
{
"epoch": 2.13,
"learning_rate": 2.1578389287527194e-05,
"loss": 0.8784,
"step": 2780
},
{
"epoch": 2.14,
"learning_rate": 2.1238203434500452e-05,
"loss": 0.8776,
"step": 2790
},
{
"epoch": 2.15,
"learning_rate": 2.0899995886824037e-05,
"loss": 0.8718,
"step": 2800
},
{
"epoch": 2.16,
"learning_rate": 2.056378990722427e-05,
"loss": 0.8362,
"step": 2810
},
{
"epoch": 2.17,
"learning_rate": 2.0229608620754752e-05,
"loss": 0.86,
"step": 2820
},
{
"epoch": 2.17,
"learning_rate": 1.9897475013205906e-05,
"loss": 0.8663,
"step": 2830
},
{
"epoch": 2.18,
"learning_rate": 1.9567411929523814e-05,
"loss": 0.8648,
"step": 2840
},
{
"epoch": 2.19,
"learning_rate": 1.9304867500687845e-05,
"loss": 0.8314,
"step": 2850
},
{
"epoch": 2.2,
"learning_rate": 1.897858847364245e-05,
"loss": 0.8936,
"step": 2860
},
{
"epoch": 2.2,
"learning_rate": 1.8654443173683906e-05,
"loss": 0.7923,
"step": 2870
},
{
"epoch": 2.21,
"learning_rate": 1.8332453896303352e-05,
"loss": 0.8587,
"step": 2880
},
{
"epoch": 2.22,
"learning_rate": 1.8012642788695538e-05,
"loss": 0.8422,
"step": 2890
},
{
"epoch": 2.23,
"learning_rate": 1.769503184823552e-05,
"loss": 0.8718,
"step": 2900
},
{
"epoch": 2.23,
"learning_rate": 1.737964292096552e-05,
"loss": 0.8353,
"step": 2910
},
{
"epoch": 2.24,
"learning_rate": 1.706649770009238e-05,
"loss": 0.851,
"step": 2920
},
{
"epoch": 2.25,
"learning_rate": 1.6755617724495465e-05,
"loss": 0.9074,
"step": 2930
},
{
"epoch": 2.26,
"learning_rate": 1.6447024377245106e-05,
"loss": 0.8933,
"step": 2940
},
{
"epoch": 2.27,
"learning_rate": 1.620181034090752e-05,
"loss": 0.8259,
"step": 2950
},
{
"epoch": 2.27,
"learning_rate": 1.5897386307048806e-05,
"loss": 0.8784,
"step": 2960
},
{
"epoch": 2.28,
"learning_rate": 1.5595307932751945e-05,
"loss": 0.9026,
"step": 2970
},
{
"epoch": 2.29,
"learning_rate": 1.5295595995692013e-05,
"loss": 0.8644,
"step": 2980
},
{
"epoch": 2.3,
"learning_rate": 1.4998271110774858e-05,
"loss": 0.8762,
"step": 2990
},
{
"epoch": 2.3,
"learning_rate": 1.4703353728719177e-05,
"loss": 0.8759,
"step": 3000
},
{
"epoch": 2.31,
"learning_rate": 1.4410864134649866e-05,
"loss": 0.8512,
"step": 3010
},
{
"epoch": 2.32,
"learning_rate": 1.4120822446702697e-05,
"loss": 0.8029,
"step": 3020
},
{
"epoch": 2.33,
"learning_rate": 1.3833248614640588e-05,
"loss": 0.8316,
"step": 3030
},
{
"epoch": 2.33,
"learning_rate": 1.3548162418481464e-05,
"loss": 0.8218,
"step": 3040
},
{
"epoch": 2.34,
"learning_rate": 1.3321897742377021e-05,
"loss": 0.814,
"step": 3050
},
{
"epoch": 2.35,
"learning_rate": 1.3041338589458008e-05,
"loss": 0.8535,
"step": 3060
},
{
"epoch": 2.36,
"learning_rate": 1.276332154191282e-05,
"loss": 0.8348,
"step": 3070
},
{
"epoch": 2.36,
"learning_rate": 1.2487865722420766e-05,
"loss": 0.8809,
"step": 3080
},
{
"epoch": 2.37,
"learning_rate": 1.2214990077493804e-05,
"loss": 0.8546,
"step": 3090
},
{
"epoch": 2.38,
"learning_rate": 1.1944713376173239e-05,
"loss": 0.8539,
"step": 3100
},
{
"epoch": 2.39,
"learning_rate": 1.1677054208738803e-05,
"loss": 0.8435,
"step": 3110
},
{
"epoch": 2.4,
"learning_rate": 1.1412030985430027e-05,
"loss": 0.8697,
"step": 3120
},
{
"epoch": 2.4,
"learning_rate": 1.1149661935179823e-05,
"loss": 0.7707,
"step": 3130
},
{
"epoch": 2.41,
"learning_rate": 1.0889965104360727e-05,
"loss": 0.8451,
"step": 3140
},
{
"epoch": 2.42,
"learning_rate": 1.0632958355543631e-05,
"loss": 0.8299,
"step": 3150
},
{
"epoch": 2.43,
"learning_rate": 1.0429301701086824e-05,
"loss": 0.8469,
"step": 3160
},
{
"epoch": 2.43,
"learning_rate": 1.0177181522158718e-05,
"loss": 0.8627,
"step": 3170
},
{
"epoch": 2.44,
"learning_rate": 9.927800452196933e-06,
"loss": 0.8862,
"step": 3180
},
{
"epoch": 2.45,
"learning_rate": 9.681175644229595e-06,
"loss": 0.88,
"step": 3190
},
{
"epoch": 2.46,
"learning_rate": 9.437324061702507e-06,
"loss": 0.8784,
"step": 3200
},
{
"epoch": 2.46,
"learning_rate": 9.196262477312368e-06,
"loss": 0.8466,
"step": 3210
},
{
"epoch": 2.47,
"learning_rate": 8.958007471853097e-06,
"loss": 0.8686,
"step": 3220
},
{
"epoch": 2.48,
"learning_rate": 8.72257543307543e-06,
"loss": 0.8486,
"step": 3230
},
{
"epoch": 2.49,
"learning_rate": 8.489982554559644e-06,
"loss": 0.8226,
"step": 3240
},
{
"epoch": 2.5,
"learning_rate": 8.260244834601776e-06,
"loss": 0.8904,
"step": 3250
},
{
"epoch": 2.5,
"learning_rate": 8.078520998307982e-06,
"loss": 0.8536,
"step": 3260
},
{
"epoch": 2.51,
"learning_rate": 7.853962251906332e-06,
"loss": 0.8472,
"step": 3270
},
{
"epoch": 2.52,
"learning_rate": 7.632302411064473e-06,
"loss": 0.8075,
"step": 3280
},
{
"epoch": 2.53,
"learning_rate": 7.413556722077869e-06,
"loss": 0.8093,
"step": 3290
},
{
"epoch": 2.53,
"learning_rate": 7.197740230799643e-06,
"loss": 0.8355,
"step": 3300
},
{
"epoch": 2.54,
"learning_rate": 6.984867781605698e-06,
"loss": 0.8723,
"step": 3310
},
{
"epoch": 2.55,
"learning_rate": 6.774954016373625e-06,
"loss": 0.813,
"step": 3320
},
{
"epoch": 2.56,
"learning_rate": 6.568013373475684e-06,
"loss": 0.8454,
"step": 3330
},
{
"epoch": 2.56,
"learning_rate": 6.364060086785589e-06,
"loss": 0.8601,
"step": 3340
},
{
"epoch": 2.57,
"learning_rate": 6.163108184699595e-06,
"loss": 0.8459,
"step": 3350
},
{
"epoch": 2.58,
"learning_rate": 6.004516955259492e-06,
"loss": 0.8269,
"step": 3360
},
{
"epoch": 2.59,
"learning_rate": 5.809002237447792e-06,
"loss": 0.8455,
"step": 3370
},
{
"epoch": 2.6,
"learning_rate": 5.6165270824503015e-06,
"loss": 0.8265,
"step": 3380
},
{
"epoch": 2.6,
"learning_rate": 5.42710472916983e-06,
"loss": 0.8267,
"step": 3390
},
{
"epoch": 2.61,
"learning_rate": 5.240748206530144e-06,
"loss": 0.8448,
"step": 3400
},
{
"epoch": 2.62,
"learning_rate": 5.057470332579817e-06,
"loss": 0.8593,
"step": 3410
},
{
"epoch": 2.63,
"learning_rate": 4.87728371361062e-06,
"loss": 0.8502,
"step": 3420
},
{
"epoch": 2.63,
"learning_rate": 4.70020074329035e-06,
"loss": 0.8452,
"step": 3430
},
{
"epoch": 2.64,
"learning_rate": 4.526233601810426e-06,
"loss": 0.8195,
"step": 3440
},
{
"epoch": 2.65,
"learning_rate": 4.355394255048078e-06,
"loss": 0.8991,
"step": 3450
},
{
"epoch": 2.66,
"learning_rate": 4.220982693581471e-06,
"loss": 0.8032,
"step": 3460
},
{
"epoch": 2.66,
"learning_rate": 4.055802844087037e-06,
"loss": 0.8793,
"step": 3470
},
{
"epoch": 2.67,
"learning_rate": 3.893783146665986e-06,
"loss": 0.831,
"step": 3480
},
{
"epoch": 2.68,
"learning_rate": 3.7349347454216866e-06,
"loss": 0.8563,
"step": 3490
},
{
"epoch": 2.69,
"learning_rate": 3.57926856632817e-06,
"loss": 0.8493,
"step": 3500
},
{
"epoch": 2.7,
"learning_rate": 3.426795316478598e-06,
"loss": 0.8212,
"step": 3510
},
{
"epoch": 2.7,
"learning_rate": 3.2775254833487702e-06,
"loss": 0.8156,
"step": 3520
},
{
"epoch": 2.71,
"learning_rate": 3.1314693340758204e-06,
"loss": 0.8268,
"step": 3530
},
{
"epoch": 2.72,
"learning_rate": 2.988636914752013e-06,
"loss": 0.8388,
"step": 3540
},
{
"epoch": 2.73,
"learning_rate": 2.84903804973371e-06,
"loss": 0.888,
"step": 3550
},
{
"epoch": 2.73,
"learning_rate": 2.7396935768158717e-06,
"loss": 0.8024,
"step": 3560
},
{
"epoch": 2.74,
"learning_rate": 2.6059391565656122e-06,
"loss": 0.8472,
"step": 3570
},
{
"epoch": 2.75,
"learning_rate": 2.4883471157567505e-06,
"loss": 0.8071,
"step": 3580
},
{
"epoch": 2.76,
"learning_rate": 2.360794142789635e-06,
"loss": 0.8513,
"step": 3590
},
{
"epoch": 2.76,
"learning_rate": 2.2365179086597076e-06,
"loss": 0.8273,
"step": 3600
},
{
"epoch": 2.77,
"learning_rate": 2.1155269613844066e-06,
"loss": 0.8456,
"step": 3610
},
{
"epoch": 2.78,
"learning_rate": 1.9978296230112826e-06,
"loss": 0.8651,
"step": 3620
},
{
"epoch": 2.79,
"learning_rate": 1.8834339890455466e-06,
"loss": 0.8572,
"step": 3630
},
{
"epoch": 2.79,
"learning_rate": 1.7723479278932753e-06,
"loss": 0.8304,
"step": 3640
},
{
"epoch": 2.8,
"learning_rate": 1.6645790803201889e-06,
"loss": 0.8025,
"step": 3650
},
{
"epoch": 2.81,
"learning_rate": 1.5601348589260822e-06,
"loss": 0.8305,
"step": 3660
},
{
"epoch": 2.82,
"learning_rate": 1.4590224476350167e-06,
"loss": 0.8974,
"step": 3670
},
{
"epoch": 2.83,
"learning_rate": 1.3612488012011482e-06,
"loss": 0.8565,
"step": 3680
},
{
"epoch": 2.83,
"learning_rate": 1.2668206447303766e-06,
"loss": 0.8744,
"step": 3690
},
{
"epoch": 2.84,
"learning_rate": 1.1757444732177824e-06,
"loss": 0.8676,
"step": 3700
},
{
"epoch": 2.85,
"learning_rate": 1.0880265511008736e-06,
"loss": 0.8538,
"step": 3710
},
{
"epoch": 2.86,
"learning_rate": 1.0036729118287346e-06,
"loss": 0.8466,
"step": 3720
},
{
"epoch": 2.86,
"learning_rate": 9.226893574469708e-07,
"loss": 0.8434,
"step": 3730
},
{
"epoch": 2.87,
"learning_rate": 8.450814581986932e-07,
"loss": 0.8087,
"step": 3740
},
{
"epoch": 2.88,
"learning_rate": 7.708545521413424e-07,
"loss": 0.8273,
"step": 3750
},
{
"epoch": 2.89,
"learning_rate": 7.000137447795429e-07,
"loss": 0.8292,
"step": 3760
},
{
"epoch": 2.89,
"learning_rate": 6.325639087139401e-07,
"loss": 0.898,
"step": 3770
},
{
"epoch": 2.9,
"learning_rate": 5.810486646907087e-07,
"loss": 0.8573,
"step": 3780
},
{
"epoch": 2.91,
"learning_rate": 5.197141112083914e-07,
"loss": 0.8482,
"step": 3790
},
{
"epoch": 2.92,
"learning_rate": 4.617829304634891e-07,
"loss": 0.8455,
"step": 3800
},
{
"epoch": 2.93,
"learning_rate": 4.072591071015708e-07,
"loss": 0.872,
"step": 3810
},
{
"epoch": 2.93,
"learning_rate": 3.561463914019625e-07,
"loss": 0.7806,
"step": 3820
},
{
"epoch": 2.94,
"learning_rate": 3.0844829901984784e-07,
"loss": 0.8572,
"step": 3830
},
{
"epoch": 2.95,
"learning_rate": 2.6416811074441717e-07,
"loss": 0.8283,
"step": 3840
},
{
"epoch": 2.96,
"learning_rate": 2.2330887227322018e-07,
"loss": 0.8094,
"step": 3850
},
{
"epoch": 2.96,
"learning_rate": 1.85873394002678e-07,
"loss": 0.8192,
"step": 3860
},
{
"epoch": 2.97,
"learning_rate": 1.5186425083477672e-07,
"loss": 0.8577,
"step": 3870
},
{
"epoch": 2.98,
"learning_rate": 1.2712547745995507e-07,
"loss": 0.8806,
"step": 3880
},
{
"epoch": 2.99,
"learning_rate": 9.928947386308452e-08,
"loss": 0.8377,
"step": 3890
},
{
"epoch": 2.99,
"learning_rate": 7.488576081853826e-08,
"loss": 0.8099,
"step": 3900
},
{
"epoch": 3.0,
"step": 3906,
"total_flos": 1.0401007701332066e+19,
"train_loss": 1.2600340369537557,
"train_runtime": 19958.7021,
"train_samples_per_second": 37.575,
"train_steps_per_second": 0.196
}
],
"max_steps": 3906,
"num_train_epochs": 3,
"total_flos": 1.0401007701332066e+19,
"trial_name": null,
"trial_params": null
}