Box_Packing_0320_20k_tae_model / trainer_state.json
Dongkkka's picture
Upload folder using huggingface_hub
430ad53 verified
Raw
History Blame Contribute Delete
278 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.4,
"eval_steps": 500,
"global_step": 20000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"grad_norm": 1.6401886940002441,
"learning_rate": 3.6e-07,
"loss": 1.3149,
"step": 10
},
{
"grad_norm": 1.8945657014846802,
"learning_rate": 7.6e-07,
"loss": 1.3163,
"step": 20
},
{
"grad_norm": 1.5859267711639404,
"learning_rate": 1.16e-06,
"loss": 1.2994,
"step": 30
},
{
"grad_norm": 1.2642822265625,
"learning_rate": 1.56e-06,
"loss": 1.297,
"step": 40
},
{
"grad_norm": 0.8344032168388367,
"learning_rate": 1.96e-06,
"loss": 1.2569,
"step": 50
},
{
"grad_norm": 0.49314084649086,
"learning_rate": 2.36e-06,
"loss": 1.2307,
"step": 60
},
{
"grad_norm": 0.32446572184562683,
"learning_rate": 2.7600000000000003e-06,
"loss": 1.203,
"step": 70
},
{
"grad_norm": 0.3321860134601593,
"learning_rate": 3.1600000000000007e-06,
"loss": 1.1971,
"step": 80
},
{
"grad_norm": 0.4447484612464905,
"learning_rate": 3.5600000000000002e-06,
"loss": 1.1866,
"step": 90
},
{
"grad_norm": 0.6310877799987793,
"learning_rate": 3.96e-06,
"loss": 1.1729,
"step": 100
},
{
"grad_norm": 0.47680121660232544,
"learning_rate": 4.360000000000001e-06,
"loss": 1.1282,
"step": 110
},
{
"grad_norm": 0.5493061542510986,
"learning_rate": 4.76e-06,
"loss": 1.0984,
"step": 120
},
{
"grad_norm": 0.5502547025680542,
"learning_rate": 5.1600000000000006e-06,
"loss": 1.0866,
"step": 130
},
{
"grad_norm": 0.3978302776813507,
"learning_rate": 5.56e-06,
"loss": 1.0817,
"step": 140
},
{
"grad_norm": 0.4087013304233551,
"learning_rate": 5.9600000000000005e-06,
"loss": 1.0658,
"step": 150
},
{
"grad_norm": 0.5459769368171692,
"learning_rate": 6.360000000000001e-06,
"loss": 1.0608,
"step": 160
},
{
"grad_norm": 0.519679069519043,
"learning_rate": 6.76e-06,
"loss": 1.0519,
"step": 170
},
{
"grad_norm": 0.5130072236061096,
"learning_rate": 7.16e-06,
"loss": 1.0519,
"step": 180
},
{
"grad_norm": 0.4664212167263031,
"learning_rate": 7.5600000000000005e-06,
"loss": 1.0414,
"step": 190
},
{
"grad_norm": 0.48712092638015747,
"learning_rate": 7.96e-06,
"loss": 1.0413,
"step": 200
},
{
"grad_norm": 0.47439441084861755,
"learning_rate": 8.36e-06,
"loss": 1.0382,
"step": 210
},
{
"grad_norm": 0.46334874629974365,
"learning_rate": 8.76e-06,
"loss": 1.0401,
"step": 220
},
{
"grad_norm": 0.4873521327972412,
"learning_rate": 9.16e-06,
"loss": 1.0294,
"step": 230
},
{
"grad_norm": 0.47923916578292847,
"learning_rate": 9.560000000000002e-06,
"loss": 1.0318,
"step": 240
},
{
"grad_norm": 0.42912131547927856,
"learning_rate": 9.96e-06,
"loss": 1.0239,
"step": 250
},
{
"grad_norm": 0.4370642602443695,
"learning_rate": 1.036e-05,
"loss": 1.028,
"step": 260
},
{
"grad_norm": 0.6046174168586731,
"learning_rate": 1.076e-05,
"loss": 1.0146,
"step": 270
},
{
"grad_norm": 0.8675687909126282,
"learning_rate": 1.1160000000000002e-05,
"loss": 0.9965,
"step": 280
},
{
"grad_norm": 1.0801470279693604,
"learning_rate": 1.156e-05,
"loss": 0.9631,
"step": 290
},
{
"grad_norm": 1.6936357021331787,
"learning_rate": 1.196e-05,
"loss": 0.9276,
"step": 300
},
{
"grad_norm": 1.015356421470642,
"learning_rate": 1.236e-05,
"loss": 0.9121,
"step": 310
},
{
"grad_norm": 1.3639075756072998,
"learning_rate": 1.276e-05,
"loss": 0.882,
"step": 320
},
{
"grad_norm": 1.4270645380020142,
"learning_rate": 1.316e-05,
"loss": 0.8411,
"step": 330
},
{
"grad_norm": 1.2292282581329346,
"learning_rate": 1.356e-05,
"loss": 0.8054,
"step": 340
},
{
"grad_norm": 1.5038810968399048,
"learning_rate": 1.396e-05,
"loss": 0.7666,
"step": 350
},
{
"grad_norm": 1.6285403966903687,
"learning_rate": 1.4360000000000001e-05,
"loss": 0.7208,
"step": 360
},
{
"grad_norm": 1.6673287153244019,
"learning_rate": 1.4760000000000001e-05,
"loss": 0.6717,
"step": 370
},
{
"grad_norm": 2.0559983253479004,
"learning_rate": 1.5160000000000002e-05,
"loss": 0.6369,
"step": 380
},
{
"grad_norm": 1.7887955904006958,
"learning_rate": 1.556e-05,
"loss": 0.6178,
"step": 390
},
{
"grad_norm": 1.8226420879364014,
"learning_rate": 1.596e-05,
"loss": 0.5882,
"step": 400
},
{
"grad_norm": 1.6779323816299438,
"learning_rate": 1.636e-05,
"loss": 0.5664,
"step": 410
},
{
"grad_norm": 1.654363751411438,
"learning_rate": 1.6760000000000002e-05,
"loss": 0.5242,
"step": 420
},
{
"grad_norm": 1.8008185625076294,
"learning_rate": 1.7160000000000002e-05,
"loss": 0.503,
"step": 430
},
{
"grad_norm": 2.010864496231079,
"learning_rate": 1.756e-05,
"loss": 0.4767,
"step": 440
},
{
"grad_norm": 1.4857263565063477,
"learning_rate": 1.796e-05,
"loss": 0.461,
"step": 450
},
{
"grad_norm": 1.8379981517791748,
"learning_rate": 1.8360000000000004e-05,
"loss": 0.4264,
"step": 460
},
{
"grad_norm": 2.315039873123169,
"learning_rate": 1.876e-05,
"loss": 0.4101,
"step": 470
},
{
"grad_norm": 2.186854362487793,
"learning_rate": 1.916e-05,
"loss": 0.3689,
"step": 480
},
{
"grad_norm": 2.3028440475463867,
"learning_rate": 1.956e-05,
"loss": 0.3323,
"step": 490
},
{
"grad_norm": 1.9360477924346924,
"learning_rate": 1.9960000000000002e-05,
"loss": 0.3115,
"step": 500
},
{
"grad_norm": 1.9707876443862915,
"learning_rate": 2.036e-05,
"loss": 0.2932,
"step": 510
},
{
"grad_norm": 1.7735759019851685,
"learning_rate": 2.076e-05,
"loss": 0.283,
"step": 520
},
{
"grad_norm": 1.597814917564392,
"learning_rate": 2.116e-05,
"loss": 0.273,
"step": 530
},
{
"grad_norm": 1.711505651473999,
"learning_rate": 2.1560000000000004e-05,
"loss": 0.2731,
"step": 540
},
{
"grad_norm": 1.8393617868423462,
"learning_rate": 2.196e-05,
"loss": 0.25,
"step": 550
},
{
"grad_norm": 1.7389594316482544,
"learning_rate": 2.236e-05,
"loss": 0.2398,
"step": 560
},
{
"grad_norm": 1.752402663230896,
"learning_rate": 2.2760000000000002e-05,
"loss": 0.2307,
"step": 570
},
{
"grad_norm": 2.040132522583008,
"learning_rate": 2.3160000000000002e-05,
"loss": 0.2114,
"step": 580
},
{
"grad_norm": 1.9711309671401978,
"learning_rate": 2.356e-05,
"loss": 0.1986,
"step": 590
},
{
"grad_norm": 2.056579113006592,
"learning_rate": 2.396e-05,
"loss": 0.1997,
"step": 600
},
{
"grad_norm": 1.6812186241149902,
"learning_rate": 2.4360000000000004e-05,
"loss": 0.1862,
"step": 610
},
{
"grad_norm": 1.5239371061325073,
"learning_rate": 2.476e-05,
"loss": 0.1885,
"step": 620
},
{
"grad_norm": 1.9391425848007202,
"learning_rate": 2.516e-05,
"loss": 0.176,
"step": 630
},
{
"grad_norm": 2.1032874584198,
"learning_rate": 2.556e-05,
"loss": 0.1607,
"step": 640
},
{
"grad_norm": 1.9331555366516113,
"learning_rate": 2.5960000000000002e-05,
"loss": 0.1512,
"step": 650
},
{
"grad_norm": 1.6696467399597168,
"learning_rate": 2.6360000000000002e-05,
"loss": 0.1471,
"step": 660
},
{
"grad_norm": 1.8185946941375732,
"learning_rate": 2.676e-05,
"loss": 0.1313,
"step": 670
},
{
"grad_norm": 1.7012594938278198,
"learning_rate": 2.716e-05,
"loss": 0.1212,
"step": 680
},
{
"grad_norm": 1.7694181203842163,
"learning_rate": 2.7560000000000004e-05,
"loss": 0.1189,
"step": 690
},
{
"grad_norm": 1.5570451021194458,
"learning_rate": 2.7960000000000003e-05,
"loss": 0.1159,
"step": 700
},
{
"grad_norm": 2.040879011154175,
"learning_rate": 2.8360000000000003e-05,
"loss": 0.1267,
"step": 710
},
{
"grad_norm": 2.0064566135406494,
"learning_rate": 2.8760000000000002e-05,
"loss": 0.1156,
"step": 720
},
{
"grad_norm": 1.5553596019744873,
"learning_rate": 2.9160000000000005e-05,
"loss": 0.0993,
"step": 730
},
{
"grad_norm": 1.4947905540466309,
"learning_rate": 2.9559999999999998e-05,
"loss": 0.1109,
"step": 740
},
{
"grad_norm": 2.195685386657715,
"learning_rate": 2.9959999999999998e-05,
"loss": 0.1163,
"step": 750
},
{
"grad_norm": 1.4413272142410278,
"learning_rate": 3.036e-05,
"loss": 0.1173,
"step": 760
},
{
"grad_norm": 1.55973482131958,
"learning_rate": 3.076e-05,
"loss": 0.1087,
"step": 770
},
{
"grad_norm": 1.8807109594345093,
"learning_rate": 3.116e-05,
"loss": 0.1133,
"step": 780
},
{
"grad_norm": 1.6292966604232788,
"learning_rate": 3.156e-05,
"loss": 0.1103,
"step": 790
},
{
"grad_norm": 1.3404572010040283,
"learning_rate": 3.196e-05,
"loss": 0.1092,
"step": 800
},
{
"grad_norm": 1.244386076927185,
"learning_rate": 3.236e-05,
"loss": 0.1037,
"step": 810
},
{
"grad_norm": 1.2594668865203857,
"learning_rate": 3.2760000000000005e-05,
"loss": 0.0993,
"step": 820
},
{
"grad_norm": 1.5505069494247437,
"learning_rate": 3.316e-05,
"loss": 0.1032,
"step": 830
},
{
"grad_norm": 1.2635992765426636,
"learning_rate": 3.3560000000000004e-05,
"loss": 0.0968,
"step": 840
},
{
"grad_norm": 1.6217236518859863,
"learning_rate": 3.396e-05,
"loss": 0.1031,
"step": 850
},
{
"grad_norm": 2.1571388244628906,
"learning_rate": 3.436e-05,
"loss": 0.1082,
"step": 860
},
{
"grad_norm": 1.3784488439559937,
"learning_rate": 3.4760000000000006e-05,
"loss": 0.0983,
"step": 870
},
{
"grad_norm": 1.2693129777908325,
"learning_rate": 3.516e-05,
"loss": 0.1013,
"step": 880
},
{
"grad_norm": 1.463412880897522,
"learning_rate": 3.5560000000000005e-05,
"loss": 0.0936,
"step": 890
},
{
"grad_norm": 1.6243302822113037,
"learning_rate": 3.596e-05,
"loss": 0.0878,
"step": 900
},
{
"grad_norm": 1.4497721195220947,
"learning_rate": 3.636e-05,
"loss": 0.0871,
"step": 910
},
{
"grad_norm": 0.9397141933441162,
"learning_rate": 3.676e-05,
"loss": 0.0863,
"step": 920
},
{
"grad_norm": 1.5587925910949707,
"learning_rate": 3.716e-05,
"loss": 0.0889,
"step": 930
},
{
"grad_norm": 1.3005576133728027,
"learning_rate": 3.756e-05,
"loss": 0.084,
"step": 940
},
{
"grad_norm": 1.2193269729614258,
"learning_rate": 3.796e-05,
"loss": 0.0909,
"step": 950
},
{
"grad_norm": 0.9646158814430237,
"learning_rate": 3.836e-05,
"loss": 0.0895,
"step": 960
},
{
"grad_norm": 1.4135878086090088,
"learning_rate": 3.876e-05,
"loss": 0.0896,
"step": 970
},
{
"grad_norm": 1.3432016372680664,
"learning_rate": 3.9160000000000005e-05,
"loss": 0.0922,
"step": 980
},
{
"grad_norm": 1.2035871744155884,
"learning_rate": 3.956e-05,
"loss": 0.0923,
"step": 990
},
{
"grad_norm": 1.2520123720169067,
"learning_rate": 3.9960000000000004e-05,
"loss": 0.0958,
"step": 1000
},
{
"grad_norm": 1.2242250442504883,
"learning_rate": 4.0360000000000007e-05,
"loss": 0.0867,
"step": 1010
},
{
"grad_norm": 1.252663016319275,
"learning_rate": 4.076e-05,
"loss": 0.0909,
"step": 1020
},
{
"grad_norm": 1.4396473169326782,
"learning_rate": 4.1160000000000006e-05,
"loss": 0.0929,
"step": 1030
},
{
"grad_norm": 1.2132742404937744,
"learning_rate": 4.156e-05,
"loss": 0.0862,
"step": 1040
},
{
"grad_norm": 1.1309123039245605,
"learning_rate": 4.196e-05,
"loss": 0.0796,
"step": 1050
},
{
"grad_norm": 0.9484611749649048,
"learning_rate": 4.236e-05,
"loss": 0.0781,
"step": 1060
},
{
"grad_norm": 0.9082660675048828,
"learning_rate": 4.276e-05,
"loss": 0.0806,
"step": 1070
},
{
"grad_norm": 1.2793047428131104,
"learning_rate": 4.316e-05,
"loss": 0.076,
"step": 1080
},
{
"grad_norm": 1.2996289730072021,
"learning_rate": 4.356e-05,
"loss": 0.0812,
"step": 1090
},
{
"grad_norm": 1.446408987045288,
"learning_rate": 4.396e-05,
"loss": 0.0863,
"step": 1100
},
{
"grad_norm": 1.1310473680496216,
"learning_rate": 4.436e-05,
"loss": 0.0732,
"step": 1110
},
{
"grad_norm": 1.1836631298065186,
"learning_rate": 4.4760000000000005e-05,
"loss": 0.0839,
"step": 1120
},
{
"grad_norm": 1.1174145936965942,
"learning_rate": 4.516e-05,
"loss": 0.0831,
"step": 1130
},
{
"grad_norm": 1.3114209175109863,
"learning_rate": 4.5560000000000004e-05,
"loss": 0.0809,
"step": 1140
},
{
"grad_norm": 1.1305618286132812,
"learning_rate": 4.596e-05,
"loss": 0.0873,
"step": 1150
},
{
"grad_norm": 1.0533171892166138,
"learning_rate": 4.636e-05,
"loss": 0.0781,
"step": 1160
},
{
"grad_norm": 1.1742881536483765,
"learning_rate": 4.6760000000000006e-05,
"loss": 0.0822,
"step": 1170
},
{
"grad_norm": 1.1944715976715088,
"learning_rate": 4.716e-05,
"loss": 0.0839,
"step": 1180
},
{
"grad_norm": 1.0838624238967896,
"learning_rate": 4.7560000000000005e-05,
"loss": 0.0768,
"step": 1190
},
{
"grad_norm": 1.0316380262374878,
"learning_rate": 4.796e-05,
"loss": 0.0823,
"step": 1200
},
{
"grad_norm": 0.9976275563240051,
"learning_rate": 4.836e-05,
"loss": 0.0751,
"step": 1210
},
{
"grad_norm": 0.7807214260101318,
"learning_rate": 4.876e-05,
"loss": 0.0668,
"step": 1220
},
{
"grad_norm": 1.175624132156372,
"learning_rate": 4.9160000000000004e-05,
"loss": 0.0707,
"step": 1230
},
{
"grad_norm": 1.2602707147598267,
"learning_rate": 4.956e-05,
"loss": 0.0791,
"step": 1240
},
{
"grad_norm": 1.0604500770568848,
"learning_rate": 4.996e-05,
"loss": 0.0773,
"step": 1250
},
{
"grad_norm": 0.9894817471504211,
"learning_rate": 5.0360000000000006e-05,
"loss": 0.0628,
"step": 1260
},
{
"grad_norm": 1.2252613306045532,
"learning_rate": 5.076000000000001e-05,
"loss": 0.0809,
"step": 1270
},
{
"grad_norm": 1.3312374353408813,
"learning_rate": 5.1160000000000005e-05,
"loss": 0.0725,
"step": 1280
},
{
"grad_norm": 1.1644082069396973,
"learning_rate": 5.1559999999999994e-05,
"loss": 0.0793,
"step": 1290
},
{
"grad_norm": 1.075032353401184,
"learning_rate": 5.196e-05,
"loss": 0.0771,
"step": 1300
},
{
"grad_norm": 0.9398416876792908,
"learning_rate": 5.236e-05,
"loss": 0.0659,
"step": 1310
},
{
"grad_norm": 0.9868608117103577,
"learning_rate": 5.2759999999999996e-05,
"loss": 0.082,
"step": 1320
},
{
"grad_norm": 0.8943616151809692,
"learning_rate": 5.316e-05,
"loss": 0.0716,
"step": 1330
},
{
"grad_norm": 0.9181155562400818,
"learning_rate": 5.356e-05,
"loss": 0.0651,
"step": 1340
},
{
"grad_norm": 0.9491348266601562,
"learning_rate": 5.396e-05,
"loss": 0.0745,
"step": 1350
},
{
"grad_norm": 1.0354893207550049,
"learning_rate": 5.436e-05,
"loss": 0.0764,
"step": 1360
},
{
"grad_norm": 0.8962678909301758,
"learning_rate": 5.476e-05,
"loss": 0.0692,
"step": 1370
},
{
"grad_norm": 1.0790551900863647,
"learning_rate": 5.516e-05,
"loss": 0.0664,
"step": 1380
},
{
"grad_norm": 0.828025758266449,
"learning_rate": 5.556e-05,
"loss": 0.0625,
"step": 1390
},
{
"grad_norm": 0.9519407749176025,
"learning_rate": 5.596e-05,
"loss": 0.0623,
"step": 1400
},
{
"grad_norm": 0.8859499096870422,
"learning_rate": 5.636e-05,
"loss": 0.0651,
"step": 1410
},
{
"grad_norm": 0.8652734160423279,
"learning_rate": 5.6760000000000005e-05,
"loss": 0.0629,
"step": 1420
},
{
"grad_norm": 1.0568864345550537,
"learning_rate": 5.716e-05,
"loss": 0.0775,
"step": 1430
},
{
"grad_norm": 0.8096494078636169,
"learning_rate": 5.7560000000000005e-05,
"loss": 0.0649,
"step": 1440
},
{
"grad_norm": 0.9572166800498962,
"learning_rate": 5.796e-05,
"loss": 0.0637,
"step": 1450
},
{
"grad_norm": 0.8617724180221558,
"learning_rate": 5.8360000000000004e-05,
"loss": 0.0683,
"step": 1460
},
{
"grad_norm": 0.9124924540519714,
"learning_rate": 5.876000000000001e-05,
"loss": 0.0627,
"step": 1470
},
{
"grad_norm": 0.869451642036438,
"learning_rate": 5.916e-05,
"loss": 0.0624,
"step": 1480
},
{
"grad_norm": 0.8673956394195557,
"learning_rate": 5.9560000000000006e-05,
"loss": 0.0671,
"step": 1490
},
{
"grad_norm": 0.6734641790390015,
"learning_rate": 5.996e-05,
"loss": 0.0631,
"step": 1500
},
{
"grad_norm": 0.7621887922286987,
"learning_rate": 6.0360000000000005e-05,
"loss": 0.0663,
"step": 1510
},
{
"grad_norm": 0.8125986456871033,
"learning_rate": 6.076000000000001e-05,
"loss": 0.0655,
"step": 1520
},
{
"grad_norm": 0.7236118316650391,
"learning_rate": 6.116e-05,
"loss": 0.0629,
"step": 1530
},
{
"grad_norm": 0.6856288313865662,
"learning_rate": 6.156e-05,
"loss": 0.0679,
"step": 1540
},
{
"grad_norm": 0.9942567348480225,
"learning_rate": 6.196000000000001e-05,
"loss": 0.0637,
"step": 1550
},
{
"grad_norm": 0.9096291661262512,
"learning_rate": 6.236e-05,
"loss": 0.064,
"step": 1560
},
{
"grad_norm": 0.9005893468856812,
"learning_rate": 6.276e-05,
"loss": 0.0636,
"step": 1570
},
{
"grad_norm": 0.9303935766220093,
"learning_rate": 6.316000000000001e-05,
"loss": 0.0614,
"step": 1580
},
{
"grad_norm": 0.6894645094871521,
"learning_rate": 6.356000000000001e-05,
"loss": 0.0621,
"step": 1590
},
{
"grad_norm": 0.6815313696861267,
"learning_rate": 6.396e-05,
"loss": 0.0671,
"step": 1600
},
{
"grad_norm": 0.6604280471801758,
"learning_rate": 6.436e-05,
"loss": 0.0585,
"step": 1610
},
{
"grad_norm": 0.7586976885795593,
"learning_rate": 6.476e-05,
"loss": 0.0557,
"step": 1620
},
{
"grad_norm": 0.9669461846351624,
"learning_rate": 6.515999999999999e-05,
"loss": 0.0622,
"step": 1630
},
{
"grad_norm": 0.8043057918548584,
"learning_rate": 6.556e-05,
"loss": 0.0646,
"step": 1640
},
{
"grad_norm": 0.7622464299201965,
"learning_rate": 6.596e-05,
"loss": 0.0591,
"step": 1650
},
{
"grad_norm": 0.752494215965271,
"learning_rate": 6.636e-05,
"loss": 0.0575,
"step": 1660
},
{
"grad_norm": 0.8137586712837219,
"learning_rate": 6.676e-05,
"loss": 0.0612,
"step": 1670
},
{
"grad_norm": 0.7831684947013855,
"learning_rate": 6.716e-05,
"loss": 0.0572,
"step": 1680
},
{
"grad_norm": 0.7652895450592041,
"learning_rate": 6.756e-05,
"loss": 0.0575,
"step": 1690
},
{
"grad_norm": 0.8497827053070068,
"learning_rate": 6.796e-05,
"loss": 0.0681,
"step": 1700
},
{
"grad_norm": 0.6377294659614563,
"learning_rate": 6.836e-05,
"loss": 0.0589,
"step": 1710
},
{
"grad_norm": 0.891244113445282,
"learning_rate": 6.876e-05,
"loss": 0.0639,
"step": 1720
},
{
"grad_norm": 0.7680009603500366,
"learning_rate": 6.916000000000001e-05,
"loss": 0.0616,
"step": 1730
},
{
"grad_norm": 0.7107235193252563,
"learning_rate": 6.956e-05,
"loss": 0.0546,
"step": 1740
},
{
"grad_norm": 0.6824011206626892,
"learning_rate": 6.996e-05,
"loss": 0.0577,
"step": 1750
},
{
"grad_norm": 0.9730836153030396,
"learning_rate": 7.036e-05,
"loss": 0.0534,
"step": 1760
},
{
"grad_norm": 0.9074521660804749,
"learning_rate": 7.076000000000001e-05,
"loss": 0.0671,
"step": 1770
},
{
"grad_norm": 0.644514799118042,
"learning_rate": 7.116e-05,
"loss": 0.0607,
"step": 1780
},
{
"grad_norm": 0.9121113419532776,
"learning_rate": 7.156e-05,
"loss": 0.0589,
"step": 1790
},
{
"grad_norm": 0.8029610514640808,
"learning_rate": 7.196000000000001e-05,
"loss": 0.0664,
"step": 1800
},
{
"grad_norm": 0.8757071495056152,
"learning_rate": 7.236e-05,
"loss": 0.0579,
"step": 1810
},
{
"grad_norm": 0.8301620483398438,
"learning_rate": 7.276e-05,
"loss": 0.0607,
"step": 1820
},
{
"grad_norm": 0.6021782755851746,
"learning_rate": 7.316000000000001e-05,
"loss": 0.0562,
"step": 1830
},
{
"grad_norm": 0.9442379474639893,
"learning_rate": 7.356000000000001e-05,
"loss": 0.0531,
"step": 1840
},
{
"grad_norm": 0.744971513748169,
"learning_rate": 7.396e-05,
"loss": 0.0584,
"step": 1850
},
{
"grad_norm": 0.7188789248466492,
"learning_rate": 7.436000000000001e-05,
"loss": 0.0561,
"step": 1860
},
{
"grad_norm": 0.6340389847755432,
"learning_rate": 7.476000000000001e-05,
"loss": 0.0566,
"step": 1870
},
{
"grad_norm": 0.7155025005340576,
"learning_rate": 7.516e-05,
"loss": 0.0563,
"step": 1880
},
{
"grad_norm": 0.662661612033844,
"learning_rate": 7.556000000000002e-05,
"loss": 0.0514,
"step": 1890
},
{
"grad_norm": 0.6533991694450378,
"learning_rate": 7.596000000000001e-05,
"loss": 0.0567,
"step": 1900
},
{
"grad_norm": 0.7804444432258606,
"learning_rate": 7.636e-05,
"loss": 0.0544,
"step": 1910
},
{
"grad_norm": 0.9438216090202332,
"learning_rate": 7.676e-05,
"loss": 0.0513,
"step": 1920
},
{
"grad_norm": 0.7270596623420715,
"learning_rate": 7.716e-05,
"loss": 0.0557,
"step": 1930
},
{
"grad_norm": 0.6200523972511292,
"learning_rate": 7.756e-05,
"loss": 0.0658,
"step": 1940
},
{
"grad_norm": 0.5124297738075256,
"learning_rate": 7.796e-05,
"loss": 0.0529,
"step": 1950
},
{
"grad_norm": 0.5669113993644714,
"learning_rate": 7.836e-05,
"loss": 0.0507,
"step": 1960
},
{
"grad_norm": 0.7467186450958252,
"learning_rate": 7.876e-05,
"loss": 0.054,
"step": 1970
},
{
"grad_norm": 0.6200578212738037,
"learning_rate": 7.916e-05,
"loss": 0.0599,
"step": 1980
},
{
"grad_norm": 0.691246509552002,
"learning_rate": 7.956e-05,
"loss": 0.0527,
"step": 1990
},
{
"grad_norm": 0.7206867337226868,
"learning_rate": 7.996e-05,
"loss": 0.0546,
"step": 2000
},
{
"grad_norm": 0.6038048267364502,
"learning_rate": 8.036e-05,
"loss": 0.0548,
"step": 2010
},
{
"grad_norm": 0.5746732950210571,
"learning_rate": 8.076e-05,
"loss": 0.0567,
"step": 2020
},
{
"grad_norm": 0.6559705138206482,
"learning_rate": 8.116e-05,
"loss": 0.0509,
"step": 2030
},
{
"grad_norm": 0.6057372689247131,
"learning_rate": 8.156e-05,
"loss": 0.0561,
"step": 2040
},
{
"grad_norm": 0.7387638092041016,
"learning_rate": 8.196000000000001e-05,
"loss": 0.0564,
"step": 2050
},
{
"grad_norm": 0.7291343808174133,
"learning_rate": 8.236e-05,
"loss": 0.0547,
"step": 2060
},
{
"grad_norm": 0.6363125443458557,
"learning_rate": 8.276e-05,
"loss": 0.0552,
"step": 2070
},
{
"grad_norm": 0.6508821845054626,
"learning_rate": 8.316000000000001e-05,
"loss": 0.0573,
"step": 2080
},
{
"grad_norm": 0.6537685990333557,
"learning_rate": 8.356e-05,
"loss": 0.0551,
"step": 2090
},
{
"grad_norm": 0.6614780426025391,
"learning_rate": 8.396e-05,
"loss": 0.0528,
"step": 2100
},
{
"grad_norm": 0.6560401320457458,
"learning_rate": 8.436000000000001e-05,
"loss": 0.0545,
"step": 2110
},
{
"grad_norm": 0.5808461904525757,
"learning_rate": 8.476000000000001e-05,
"loss": 0.0533,
"step": 2120
},
{
"grad_norm": 0.5264371037483215,
"learning_rate": 8.516e-05,
"loss": 0.0582,
"step": 2130
},
{
"grad_norm": 0.5954933762550354,
"learning_rate": 8.556e-05,
"loss": 0.0523,
"step": 2140
},
{
"grad_norm": 0.7801850438117981,
"learning_rate": 8.596000000000001e-05,
"loss": 0.0537,
"step": 2150
},
{
"grad_norm": 0.5850076079368591,
"learning_rate": 8.636e-05,
"loss": 0.0454,
"step": 2160
},
{
"grad_norm": 0.5922155380249023,
"learning_rate": 8.676e-05,
"loss": 0.0509,
"step": 2170
},
{
"grad_norm": 0.8190131187438965,
"learning_rate": 8.716000000000001e-05,
"loss": 0.0542,
"step": 2180
},
{
"grad_norm": 0.587446928024292,
"learning_rate": 8.756000000000001e-05,
"loss": 0.0553,
"step": 2190
},
{
"grad_norm": 0.6227629780769348,
"learning_rate": 8.796e-05,
"loss": 0.0487,
"step": 2200
},
{
"grad_norm": 0.6692546010017395,
"learning_rate": 8.836000000000001e-05,
"loss": 0.0476,
"step": 2210
},
{
"grad_norm": 0.7408078908920288,
"learning_rate": 8.876e-05,
"loss": 0.0489,
"step": 2220
},
{
"grad_norm": 0.586682140827179,
"learning_rate": 8.916e-05,
"loss": 0.0509,
"step": 2230
},
{
"grad_norm": 0.6249874234199524,
"learning_rate": 8.956e-05,
"loss": 0.0526,
"step": 2240
},
{
"grad_norm": 0.7161738872528076,
"learning_rate": 8.996e-05,
"loss": 0.0596,
"step": 2250
},
{
"grad_norm": 0.6018122434616089,
"learning_rate": 9.036e-05,
"loss": 0.0584,
"step": 2260
},
{
"grad_norm": 0.4742252826690674,
"learning_rate": 9.076e-05,
"loss": 0.0574,
"step": 2270
},
{
"grad_norm": 0.4201463758945465,
"learning_rate": 9.116e-05,
"loss": 0.0496,
"step": 2280
},
{
"grad_norm": 0.4542571008205414,
"learning_rate": 9.156e-05,
"loss": 0.0446,
"step": 2290
},
{
"grad_norm": 0.5460566282272339,
"learning_rate": 9.196000000000001e-05,
"loss": 0.0512,
"step": 2300
},
{
"grad_norm": 0.38791629672050476,
"learning_rate": 9.236e-05,
"loss": 0.052,
"step": 2310
},
{
"grad_norm": 0.5282217264175415,
"learning_rate": 9.276e-05,
"loss": 0.0476,
"step": 2320
},
{
"grad_norm": 0.5786757469177246,
"learning_rate": 9.316000000000001e-05,
"loss": 0.054,
"step": 2330
},
{
"grad_norm": 0.5746905207633972,
"learning_rate": 9.356e-05,
"loss": 0.0531,
"step": 2340
},
{
"grad_norm": 0.6251006722450256,
"learning_rate": 9.396e-05,
"loss": 0.0492,
"step": 2350
},
{
"grad_norm": 0.7246794700622559,
"learning_rate": 9.436e-05,
"loss": 0.0553,
"step": 2360
},
{
"grad_norm": 0.5014413595199585,
"learning_rate": 9.476000000000001e-05,
"loss": 0.0527,
"step": 2370
},
{
"grad_norm": 0.5953740477561951,
"learning_rate": 9.516e-05,
"loss": 0.0535,
"step": 2380
},
{
"grad_norm": 0.5419349074363708,
"learning_rate": 9.556e-05,
"loss": 0.0508,
"step": 2390
},
{
"grad_norm": 0.5652883052825928,
"learning_rate": 9.596000000000001e-05,
"loss": 0.0514,
"step": 2400
},
{
"grad_norm": 0.4460015594959259,
"learning_rate": 9.636e-05,
"loss": 0.0481,
"step": 2410
},
{
"grad_norm": 0.6789939999580383,
"learning_rate": 9.676e-05,
"loss": 0.0491,
"step": 2420
},
{
"grad_norm": 0.5071524381637573,
"learning_rate": 9.716000000000001e-05,
"loss": 0.0491,
"step": 2430
},
{
"grad_norm": 0.6390159726142883,
"learning_rate": 9.756000000000001e-05,
"loss": 0.0485,
"step": 2440
},
{
"grad_norm": 0.6449014544487,
"learning_rate": 9.796e-05,
"loss": 0.05,
"step": 2450
},
{
"grad_norm": 0.506130576133728,
"learning_rate": 9.836000000000001e-05,
"loss": 0.0529,
"step": 2460
},
{
"grad_norm": 0.5097265243530273,
"learning_rate": 9.876000000000001e-05,
"loss": 0.0463,
"step": 2470
},
{
"grad_norm": 0.4680483937263489,
"learning_rate": 9.916e-05,
"loss": 0.0554,
"step": 2480
},
{
"grad_norm": 0.7338577508926392,
"learning_rate": 9.956e-05,
"loss": 0.0483,
"step": 2490
},
{
"grad_norm": 0.46772661805152893,
"learning_rate": 9.996000000000001e-05,
"loss": 0.0513,
"step": 2500
},
{
"grad_norm": 0.5334926843643188,
"learning_rate": 9.999999114196196e-05,
"loss": 0.0502,
"step": 2510
},
{
"grad_norm": 0.5277791619300842,
"learning_rate": 9.99999605215876e-05,
"loss": 0.0487,
"step": 2520
},
{
"grad_norm": 0.683641254901886,
"learning_rate": 9.999990802953179e-05,
"loss": 0.0472,
"step": 2530
},
{
"grad_norm": 0.5659726858139038,
"learning_rate": 9.99998336658175e-05,
"loss": 0.0479,
"step": 2540
},
{
"grad_norm": 0.6926059126853943,
"learning_rate": 9.999973743047727e-05,
"loss": 0.0563,
"step": 2550
},
{
"grad_norm": 0.4505929946899414,
"learning_rate": 9.999961932355319e-05,
"loss": 0.0498,
"step": 2560
},
{
"grad_norm": 0.6377073526382446,
"learning_rate": 9.999947934509693e-05,
"loss": 0.0481,
"step": 2570
},
{
"grad_norm": 0.4898621737957001,
"learning_rate": 9.999931749516971e-05,
"loss": 0.052,
"step": 2580
},
{
"grad_norm": 0.5683681964874268,
"learning_rate": 9.999913377384233e-05,
"loss": 0.0467,
"step": 2590
},
{
"grad_norm": 0.5124620795249939,
"learning_rate": 9.999892818119517e-05,
"loss": 0.0499,
"step": 2600
},
{
"grad_norm": 0.645530641078949,
"learning_rate": 9.999870071731814e-05,
"loss": 0.0497,
"step": 2610
},
{
"grad_norm": 0.6242179870605469,
"learning_rate": 9.999845138231076e-05,
"loss": 0.0477,
"step": 2620
},
{
"grad_norm": 0.5192215442657471,
"learning_rate": 9.999818017628208e-05,
"loss": 0.0539,
"step": 2630
},
{
"grad_norm": 0.4066641628742218,
"learning_rate": 9.999788709935078e-05,
"loss": 0.049,
"step": 2640
},
{
"grad_norm": 0.4780401587486267,
"learning_rate": 9.9997572151645e-05,
"loss": 0.049,
"step": 2650
},
{
"grad_norm": 0.460101842880249,
"learning_rate": 9.999723533330254e-05,
"loss": 0.0483,
"step": 2660
},
{
"grad_norm": 0.5579531192779541,
"learning_rate": 9.999687664447074e-05,
"loss": 0.0505,
"step": 2670
},
{
"grad_norm": 0.45036232471466064,
"learning_rate": 9.99964960853065e-05,
"loss": 0.0462,
"step": 2680
},
{
"grad_norm": 0.6312862634658813,
"learning_rate": 9.999609365597627e-05,
"loss": 0.0491,
"step": 2690
},
{
"grad_norm": 0.6391765475273132,
"learning_rate": 9.99956693566561e-05,
"loss": 0.0495,
"step": 2700
},
{
"grad_norm": 0.5391504168510437,
"learning_rate": 9.99952231875316e-05,
"loss": 0.0532,
"step": 2710
},
{
"grad_norm": 0.39547762274742126,
"learning_rate": 9.999475514879795e-05,
"loss": 0.0457,
"step": 2720
},
{
"grad_norm": 0.5656508207321167,
"learning_rate": 9.999426524065984e-05,
"loss": 0.0486,
"step": 2730
},
{
"grad_norm": 0.45238953828811646,
"learning_rate": 9.999375346333162e-05,
"loss": 0.0493,
"step": 2740
},
{
"grad_norm": 0.586006760597229,
"learning_rate": 9.999321981703715e-05,
"loss": 0.0488,
"step": 2750
},
{
"grad_norm": 0.5166579484939575,
"learning_rate": 9.999266430200985e-05,
"loss": 0.0494,
"step": 2760
},
{
"grad_norm": 0.6533814072608948,
"learning_rate": 9.999208691849271e-05,
"loss": 0.0418,
"step": 2770
},
{
"grad_norm": 0.4532153904438019,
"learning_rate": 9.999148766673832e-05,
"loss": 0.0484,
"step": 2780
},
{
"grad_norm": 0.43949344754219055,
"learning_rate": 9.999086654700881e-05,
"loss": 0.0478,
"step": 2790
},
{
"grad_norm": 0.6996206045150757,
"learning_rate": 9.999022355957588e-05,
"loss": 0.0445,
"step": 2800
},
{
"grad_norm": 0.31124061346054077,
"learning_rate": 9.998955870472079e-05,
"loss": 0.0429,
"step": 2810
},
{
"grad_norm": 0.3255634605884552,
"learning_rate": 9.998887198273437e-05,
"loss": 0.0405,
"step": 2820
},
{
"grad_norm": 0.48686057329177856,
"learning_rate": 9.998816339391701e-05,
"loss": 0.0446,
"step": 2830
},
{
"grad_norm": 0.4439123272895813,
"learning_rate": 9.998743293857868e-05,
"loss": 0.0492,
"step": 2840
},
{
"grad_norm": 0.448848158121109,
"learning_rate": 9.998668061703891e-05,
"loss": 0.0417,
"step": 2850
},
{
"grad_norm": 0.5039438009262085,
"learning_rate": 9.998590642962679e-05,
"loss": 0.0414,
"step": 2860
},
{
"grad_norm": 0.46082308888435364,
"learning_rate": 9.998511037668095e-05,
"loss": 0.0431,
"step": 2870
},
{
"grad_norm": 0.38872405886650085,
"learning_rate": 9.998429245854964e-05,
"loss": 0.0422,
"step": 2880
},
{
"grad_norm": 0.4246666729450226,
"learning_rate": 9.998345267559064e-05,
"loss": 0.041,
"step": 2890
},
{
"grad_norm": 0.6673298478126526,
"learning_rate": 9.998259102817129e-05,
"loss": 0.0443,
"step": 2900
},
{
"grad_norm": 0.5126121044158936,
"learning_rate": 9.99817075166685e-05,
"loss": 0.0425,
"step": 2910
},
{
"grad_norm": 0.48576462268829346,
"learning_rate": 9.998080214146878e-05,
"loss": 0.0424,
"step": 2920
},
{
"grad_norm": 0.5307922959327698,
"learning_rate": 9.997987490296813e-05,
"loss": 0.0476,
"step": 2930
},
{
"grad_norm": 0.5131188631057739,
"learning_rate": 9.99789258015722e-05,
"loss": 0.0489,
"step": 2940
},
{
"grad_norm": 0.4211786687374115,
"learning_rate": 9.997795483769611e-05,
"loss": 0.0458,
"step": 2950
},
{
"grad_norm": 0.43335118889808655,
"learning_rate": 9.997696201176462e-05,
"loss": 0.0506,
"step": 2960
},
{
"grad_norm": 0.41095250844955444,
"learning_rate": 9.997594732421203e-05,
"loss": 0.0437,
"step": 2970
},
{
"grad_norm": 0.5414423942565918,
"learning_rate": 9.997491077548217e-05,
"loss": 0.0426,
"step": 2980
},
{
"grad_norm": 0.36229678988456726,
"learning_rate": 9.997385236602851e-05,
"loss": 0.0419,
"step": 2990
},
{
"grad_norm": 0.34936150908470154,
"learning_rate": 9.997277209631399e-05,
"loss": 0.0427,
"step": 3000
},
{
"grad_norm": 0.6067139506340027,
"learning_rate": 9.997166996681118e-05,
"loss": 0.0499,
"step": 3010
},
{
"grad_norm": 0.49866238236427307,
"learning_rate": 9.997054597800218e-05,
"loss": 0.0442,
"step": 3020
},
{
"grad_norm": 0.3345346748828888,
"learning_rate": 9.996940013037866e-05,
"loss": 0.0411,
"step": 3030
},
{
"grad_norm": 0.45981162786483765,
"learning_rate": 9.996823242444186e-05,
"loss": 0.0396,
"step": 3040
},
{
"grad_norm": 0.4103314280509949,
"learning_rate": 9.996704286070258e-05,
"loss": 0.0417,
"step": 3050
},
{
"grad_norm": 0.5575783848762512,
"learning_rate": 9.996583143968115e-05,
"loss": 0.0397,
"step": 3060
},
{
"grad_norm": 0.404458612203598,
"learning_rate": 9.99645981619075e-05,
"loss": 0.0414,
"step": 3070
},
{
"grad_norm": 0.36830517649650574,
"learning_rate": 9.996334302792114e-05,
"loss": 0.042,
"step": 3080
},
{
"grad_norm": 0.41416144371032715,
"learning_rate": 9.996206603827105e-05,
"loss": 0.0443,
"step": 3090
},
{
"grad_norm": 0.42167651653289795,
"learning_rate": 9.996076719351587e-05,
"loss": 0.0466,
"step": 3100
},
{
"grad_norm": 0.4996941387653351,
"learning_rate": 9.995944649422374e-05,
"loss": 0.0483,
"step": 3110
},
{
"grad_norm": 0.3907119035720825,
"learning_rate": 9.995810394097239e-05,
"loss": 0.0393,
"step": 3120
},
{
"grad_norm": 0.540632426738739,
"learning_rate": 9.995673953434909e-05,
"loss": 0.0495,
"step": 3130
},
{
"grad_norm": 0.4541913568973541,
"learning_rate": 9.995535327495068e-05,
"loss": 0.0401,
"step": 3140
},
{
"grad_norm": 0.5103195309638977,
"learning_rate": 9.995394516338355e-05,
"loss": 0.0454,
"step": 3150
},
{
"grad_norm": 0.4949314296245575,
"learning_rate": 9.995251520026367e-05,
"loss": 0.0367,
"step": 3160
},
{
"grad_norm": 0.4270646870136261,
"learning_rate": 9.995106338621656e-05,
"loss": 0.0413,
"step": 3170
},
{
"grad_norm": 0.3582088351249695,
"learning_rate": 9.994958972187726e-05,
"loss": 0.0422,
"step": 3180
},
{
"grad_norm": 0.5443354845046997,
"learning_rate": 9.994809420789044e-05,
"loss": 0.0464,
"step": 3190
},
{
"grad_norm": 0.4865322709083557,
"learning_rate": 9.994657684491027e-05,
"loss": 0.0416,
"step": 3200
},
{
"grad_norm": 0.44792330265045166,
"learning_rate": 9.994503763360048e-05,
"loss": 0.0384,
"step": 3210
},
{
"grad_norm": 0.5200015306472778,
"learning_rate": 9.99434765746344e-05,
"loss": 0.0448,
"step": 3220
},
{
"grad_norm": 0.3982260525226593,
"learning_rate": 9.994189366869488e-05,
"loss": 0.0484,
"step": 3230
},
{
"grad_norm": 0.35745036602020264,
"learning_rate": 9.994028891647433e-05,
"loss": 0.0415,
"step": 3240
},
{
"grad_norm": 0.38287830352783203,
"learning_rate": 9.993866231867475e-05,
"loss": 0.0425,
"step": 3250
},
{
"grad_norm": 0.4272738993167877,
"learning_rate": 9.993701387600762e-05,
"loss": 0.0436,
"step": 3260
},
{
"grad_norm": 0.560173749923706,
"learning_rate": 9.993534358919408e-05,
"loss": 0.0432,
"step": 3270
},
{
"grad_norm": 0.49206626415252686,
"learning_rate": 9.993365145896473e-05,
"loss": 0.0386,
"step": 3280
},
{
"grad_norm": 0.40479663014411926,
"learning_rate": 9.993193748605977e-05,
"loss": 0.0418,
"step": 3290
},
{
"grad_norm": 0.3908839821815491,
"learning_rate": 9.993020167122898e-05,
"loss": 0.0448,
"step": 3300
},
{
"grad_norm": 0.33099737763404846,
"learning_rate": 9.992844401523164e-05,
"loss": 0.0433,
"step": 3310
},
{
"grad_norm": 0.39015039801597595,
"learning_rate": 9.992666451883661e-05,
"loss": 0.0447,
"step": 3320
},
{
"grad_norm": 0.38739025592803955,
"learning_rate": 9.99248631828223e-05,
"loss": 0.0422,
"step": 3330
},
{
"grad_norm": 0.3889550566673279,
"learning_rate": 9.99230400079767e-05,
"loss": 0.0432,
"step": 3340
},
{
"grad_norm": 0.5984971523284912,
"learning_rate": 9.992119499509728e-05,
"loss": 0.0449,
"step": 3350
},
{
"grad_norm": 0.4535196125507355,
"learning_rate": 9.991932814499114e-05,
"loss": 0.0396,
"step": 3360
},
{
"grad_norm": 0.3221626281738281,
"learning_rate": 9.991743945847493e-05,
"loss": 0.0402,
"step": 3370
},
{
"grad_norm": 0.523133397102356,
"learning_rate": 9.991552893637478e-05,
"loss": 0.0394,
"step": 3380
},
{
"grad_norm": 0.5344473719596863,
"learning_rate": 9.991359657952644e-05,
"loss": 0.044,
"step": 3390
},
{
"grad_norm": 0.4005912244319916,
"learning_rate": 9.991164238877519e-05,
"loss": 0.0448,
"step": 3400
},
{
"grad_norm": 0.535717248916626,
"learning_rate": 9.990966636497585e-05,
"loss": 0.0462,
"step": 3410
},
{
"grad_norm": 0.4947056174278259,
"learning_rate": 9.99076685089928e-05,
"loss": 0.041,
"step": 3420
},
{
"grad_norm": 0.5003987550735474,
"learning_rate": 9.990564882169998e-05,
"loss": 0.0395,
"step": 3430
},
{
"grad_norm": 0.6668880581855774,
"learning_rate": 9.990360730398088e-05,
"loss": 0.0447,
"step": 3440
},
{
"grad_norm": 0.480677992105484,
"learning_rate": 9.990154395672849e-05,
"loss": 0.0368,
"step": 3450
},
{
"grad_norm": 0.36002954840660095,
"learning_rate": 9.989945878084541e-05,
"loss": 0.0424,
"step": 3460
},
{
"grad_norm": 0.48320573568344116,
"learning_rate": 9.989735177724378e-05,
"loss": 0.0459,
"step": 3470
},
{
"grad_norm": 0.3653718829154968,
"learning_rate": 9.989522294684526e-05,
"loss": 0.0411,
"step": 3480
},
{
"grad_norm": 0.3104921579360962,
"learning_rate": 9.989307229058107e-05,
"loss": 0.04,
"step": 3490
},
{
"grad_norm": 0.4608096182346344,
"learning_rate": 9.989089980939202e-05,
"loss": 0.0367,
"step": 3500
},
{
"grad_norm": 0.44114038348197937,
"learning_rate": 9.988870550422835e-05,
"loss": 0.045,
"step": 3510
},
{
"grad_norm": 0.3840514123439789,
"learning_rate": 9.988648937604999e-05,
"loss": 0.0401,
"step": 3520
},
{
"grad_norm": 0.3829437792301178,
"learning_rate": 9.988425142582632e-05,
"loss": 0.0412,
"step": 3530
},
{
"grad_norm": 0.5176998972892761,
"learning_rate": 9.98819916545363e-05,
"loss": 0.0405,
"step": 3540
},
{
"grad_norm": 0.38363850116729736,
"learning_rate": 9.987971006316844e-05,
"loss": 0.0398,
"step": 3550
},
{
"grad_norm": 0.5735336542129517,
"learning_rate": 9.987740665272077e-05,
"loss": 0.0412,
"step": 3560
},
{
"grad_norm": 0.41206949949264526,
"learning_rate": 9.98750814242009e-05,
"loss": 0.0407,
"step": 3570
},
{
"grad_norm": 0.45664188265800476,
"learning_rate": 9.987273437862594e-05,
"loss": 0.0414,
"step": 3580
},
{
"grad_norm": 0.45491260290145874,
"learning_rate": 9.987036551702259e-05,
"loss": 0.0394,
"step": 3590
},
{
"grad_norm": 0.3253578841686249,
"learning_rate": 9.986797484042706e-05,
"loss": 0.0388,
"step": 3600
},
{
"grad_norm": 0.42229610681533813,
"learning_rate": 9.986556234988512e-05,
"loss": 0.0403,
"step": 3610
},
{
"grad_norm": 0.332193523645401,
"learning_rate": 9.986312804645205e-05,
"loss": 0.039,
"step": 3620
},
{
"grad_norm": 0.3959410488605499,
"learning_rate": 9.986067193119273e-05,
"loss": 0.0423,
"step": 3630
},
{
"grad_norm": 0.3983510136604309,
"learning_rate": 9.985819400518153e-05,
"loss": 0.0375,
"step": 3640
},
{
"grad_norm": 0.4366046190261841,
"learning_rate": 9.985569426950239e-05,
"loss": 0.0368,
"step": 3650
},
{
"grad_norm": 0.4879259169101715,
"learning_rate": 9.985317272524876e-05,
"loss": 0.0358,
"step": 3660
},
{
"grad_norm": 0.40022969245910645,
"learning_rate": 9.985062937352366e-05,
"loss": 0.0365,
"step": 3670
},
{
"grad_norm": 0.33493074774742126,
"learning_rate": 9.984806421543966e-05,
"loss": 0.0374,
"step": 3680
},
{
"grad_norm": 0.3884679079055786,
"learning_rate": 9.984547725211881e-05,
"loss": 0.0442,
"step": 3690
},
{
"grad_norm": 0.4603472650051117,
"learning_rate": 9.984286848469276e-05,
"loss": 0.035,
"step": 3700
},
{
"grad_norm": 0.31187203526496887,
"learning_rate": 9.984023791430266e-05,
"loss": 0.0363,
"step": 3710
},
{
"grad_norm": 0.35833173990249634,
"learning_rate": 9.983758554209924e-05,
"loss": 0.0376,
"step": 3720
},
{
"grad_norm": 0.3557681441307068,
"learning_rate": 9.983491136924268e-05,
"loss": 0.0403,
"step": 3730
},
{
"grad_norm": 0.4058561325073242,
"learning_rate": 9.983221539690282e-05,
"loss": 0.0409,
"step": 3740
},
{
"grad_norm": 0.31761378049850464,
"learning_rate": 9.982949762625892e-05,
"loss": 0.0374,
"step": 3750
},
{
"grad_norm": 0.4844707250595093,
"learning_rate": 9.982675805849986e-05,
"loss": 0.0415,
"step": 3760
},
{
"grad_norm": 0.40428078174591064,
"learning_rate": 9.982399669482399e-05,
"loss": 0.0322,
"step": 3770
},
{
"grad_norm": 0.485557496547699,
"learning_rate": 9.982121353643924e-05,
"loss": 0.0394,
"step": 3780
},
{
"grad_norm": 0.4449644684791565,
"learning_rate": 9.981840858456306e-05,
"loss": 0.0372,
"step": 3790
},
{
"grad_norm": 0.40097731351852417,
"learning_rate": 9.981558184042243e-05,
"loss": 0.0413,
"step": 3800
},
{
"grad_norm": 0.42122742533683777,
"learning_rate": 9.981273330525387e-05,
"loss": 0.0423,
"step": 3810
},
{
"grad_norm": 0.3714742958545685,
"learning_rate": 9.980986298030341e-05,
"loss": 0.0464,
"step": 3820
},
{
"grad_norm": 0.29959240555763245,
"learning_rate": 9.980697086682662e-05,
"loss": 0.0357,
"step": 3830
},
{
"grad_norm": 0.3232460916042328,
"learning_rate": 9.980405696608866e-05,
"loss": 0.0359,
"step": 3840
},
{
"grad_norm": 0.3844173550605774,
"learning_rate": 9.980112127936409e-05,
"loss": 0.039,
"step": 3850
},
{
"grad_norm": 0.3285090923309326,
"learning_rate": 9.979816380793717e-05,
"loss": 0.0392,
"step": 3860
},
{
"grad_norm": 0.38894811272621155,
"learning_rate": 9.979518455310151e-05,
"loss": 0.0336,
"step": 3870
},
{
"grad_norm": 0.40085750818252563,
"learning_rate": 9.97921835161604e-05,
"loss": 0.0377,
"step": 3880
},
{
"grad_norm": 0.4129883348941803,
"learning_rate": 9.978916069842656e-05,
"loss": 0.0421,
"step": 3890
},
{
"grad_norm": 0.3532140851020813,
"learning_rate": 9.97861161012223e-05,
"loss": 0.0428,
"step": 3900
},
{
"grad_norm": 0.40982306003570557,
"learning_rate": 9.978304972587942e-05,
"loss": 0.0381,
"step": 3910
},
{
"grad_norm": 0.4418235421180725,
"learning_rate": 9.977996157373925e-05,
"loss": 0.0447,
"step": 3920
},
{
"grad_norm": 0.35203543305397034,
"learning_rate": 9.977685164615265e-05,
"loss": 0.0343,
"step": 3930
},
{
"grad_norm": 0.3843795359134674,
"learning_rate": 9.977371994448002e-05,
"loss": 0.0393,
"step": 3940
},
{
"grad_norm": 0.35057252645492554,
"learning_rate": 9.977056647009127e-05,
"loss": 0.0394,
"step": 3950
},
{
"grad_norm": 0.4007676839828491,
"learning_rate": 9.976739122436582e-05,
"loss": 0.0374,
"step": 3960
},
{
"grad_norm": 0.4060984253883362,
"learning_rate": 9.976419420869265e-05,
"loss": 0.0346,
"step": 3970
},
{
"grad_norm": 0.2825969457626343,
"learning_rate": 9.976097542447025e-05,
"loss": 0.0343,
"step": 3980
},
{
"grad_norm": 0.4013255834579468,
"learning_rate": 9.97577348731066e-05,
"loss": 0.0362,
"step": 3990
},
{
"grad_norm": 0.480281263589859,
"learning_rate": 9.975447255601927e-05,
"loss": 0.0397,
"step": 4000
},
{
"grad_norm": 0.4256203770637512,
"learning_rate": 9.975118847463525e-05,
"loss": 0.0354,
"step": 4010
},
{
"grad_norm": 0.36489900946617126,
"learning_rate": 9.974788263039114e-05,
"loss": 0.0347,
"step": 4020
},
{
"grad_norm": 0.3175728917121887,
"learning_rate": 9.974455502473303e-05,
"loss": 0.0378,
"step": 4030
},
{
"grad_norm": 0.32769936323165894,
"learning_rate": 9.974120565911652e-05,
"loss": 0.0345,
"step": 4040
},
{
"grad_norm": 0.38217347860336304,
"learning_rate": 9.973783453500674e-05,
"loss": 0.0366,
"step": 4050
},
{
"grad_norm": 0.4321262538433075,
"learning_rate": 9.973444165387835e-05,
"loss": 0.0372,
"step": 4060
},
{
"grad_norm": 0.4731610417366028,
"learning_rate": 9.973102701721549e-05,
"loss": 0.0313,
"step": 4070
},
{
"grad_norm": 0.31932616233825684,
"learning_rate": 9.972759062651184e-05,
"loss": 0.0367,
"step": 4080
},
{
"grad_norm": 0.3345484733581543,
"learning_rate": 9.972413248327059e-05,
"loss": 0.0346,
"step": 4090
},
{
"grad_norm": 0.3954044282436371,
"learning_rate": 9.972065258900447e-05,
"loss": 0.0327,
"step": 4100
},
{
"grad_norm": 0.4238516688346863,
"learning_rate": 9.971715094523569e-05,
"loss": 0.0374,
"step": 4110
},
{
"grad_norm": 0.371136337518692,
"learning_rate": 9.971362755349598e-05,
"loss": 0.0378,
"step": 4120
},
{
"grad_norm": 0.365295946598053,
"learning_rate": 9.971008241532662e-05,
"loss": 0.0394,
"step": 4130
},
{
"grad_norm": 0.41665971279144287,
"learning_rate": 9.970651553227835e-05,
"loss": 0.035,
"step": 4140
},
{
"grad_norm": 0.3277624845504761,
"learning_rate": 9.970292690591143e-05,
"loss": 0.0375,
"step": 4150
},
{
"grad_norm": 0.40566933155059814,
"learning_rate": 9.969931653779569e-05,
"loss": 0.0345,
"step": 4160
},
{
"grad_norm": 0.3218092620372772,
"learning_rate": 9.969568442951038e-05,
"loss": 0.0379,
"step": 4170
},
{
"grad_norm": 0.38037654757499695,
"learning_rate": 9.969203058264436e-05,
"loss": 0.0379,
"step": 4180
},
{
"grad_norm": 0.34302985668182373,
"learning_rate": 9.96883549987959e-05,
"loss": 0.0402,
"step": 4190
},
{
"grad_norm": 0.40920278429985046,
"learning_rate": 9.968465767957287e-05,
"loss": 0.0348,
"step": 4200
},
{
"grad_norm": 0.36454278230667114,
"learning_rate": 9.968093862659256e-05,
"loss": 0.0342,
"step": 4210
},
{
"grad_norm": 0.38657626509666443,
"learning_rate": 9.967719784148182e-05,
"loss": 0.034,
"step": 4220
},
{
"grad_norm": 0.29038262367248535,
"learning_rate": 9.967343532587702e-05,
"loss": 0.0346,
"step": 4230
},
{
"grad_norm": 0.4502968490123749,
"learning_rate": 9.966965108142399e-05,
"loss": 0.035,
"step": 4240
},
{
"grad_norm": 0.4211561381816864,
"learning_rate": 9.96658451097781e-05,
"loss": 0.036,
"step": 4250
},
{
"grad_norm": 0.370222806930542,
"learning_rate": 9.966201741260419e-05,
"loss": 0.0366,
"step": 4260
},
{
"grad_norm": 0.4462031424045563,
"learning_rate": 9.965816799157665e-05,
"loss": 0.0376,
"step": 4270
},
{
"grad_norm": 0.43808743357658386,
"learning_rate": 9.965429684837935e-05,
"loss": 0.0362,
"step": 4280
},
{
"grad_norm": 0.3189237415790558,
"learning_rate": 9.965040398470562e-05,
"loss": 0.0321,
"step": 4290
},
{
"grad_norm": 0.5428202152252197,
"learning_rate": 9.964648940225838e-05,
"loss": 0.0324,
"step": 4300
},
{
"grad_norm": 0.37513720989227295,
"learning_rate": 9.964255310274997e-05,
"loss": 0.0345,
"step": 4310
},
{
"grad_norm": 0.37615296244621277,
"learning_rate": 9.963859508790228e-05,
"loss": 0.0389,
"step": 4320
},
{
"grad_norm": 0.33625322580337524,
"learning_rate": 9.963461535944664e-05,
"loss": 0.034,
"step": 4330
},
{
"grad_norm": 0.3464892506599426,
"learning_rate": 9.963061391912399e-05,
"loss": 0.0375,
"step": 4340
},
{
"grad_norm": 0.3498099744319916,
"learning_rate": 9.962659076868463e-05,
"loss": 0.0401,
"step": 4350
},
{
"grad_norm": 0.3511611223220825,
"learning_rate": 9.962254590988846e-05,
"loss": 0.0337,
"step": 4360
},
{
"grad_norm": 0.37777942419052124,
"learning_rate": 9.961847934450481e-05,
"loss": 0.0365,
"step": 4370
},
{
"grad_norm": 0.37865445017814636,
"learning_rate": 9.961439107431257e-05,
"loss": 0.039,
"step": 4380
},
{
"grad_norm": 0.32945024967193604,
"learning_rate": 9.961028110110006e-05,
"loss": 0.0421,
"step": 4390
},
{
"grad_norm": 0.5437269806861877,
"learning_rate": 9.960614942666513e-05,
"loss": 0.0356,
"step": 4400
},
{
"grad_norm": 0.39982321858406067,
"learning_rate": 9.960199605281511e-05,
"loss": 0.0386,
"step": 4410
},
{
"grad_norm": 0.3944191634654999,
"learning_rate": 9.959782098136683e-05,
"loss": 0.0365,
"step": 4420
},
{
"grad_norm": 0.34251973032951355,
"learning_rate": 9.959362421414662e-05,
"loss": 0.0338,
"step": 4430
},
{
"grad_norm": 0.4052746593952179,
"learning_rate": 9.958940575299027e-05,
"loss": 0.0359,
"step": 4440
},
{
"grad_norm": 0.35605090856552124,
"learning_rate": 9.95851655997431e-05,
"loss": 0.0363,
"step": 4450
},
{
"grad_norm": 0.42771291732788086,
"learning_rate": 9.958090375625986e-05,
"loss": 0.035,
"step": 4460
},
{
"grad_norm": 0.4660288989543915,
"learning_rate": 9.957662022440486e-05,
"loss": 0.0333,
"step": 4470
},
{
"grad_norm": 0.3814123868942261,
"learning_rate": 9.957231500605187e-05,
"loss": 0.0353,
"step": 4480
},
{
"grad_norm": 0.4554194211959839,
"learning_rate": 9.95679881030841e-05,
"loss": 0.0406,
"step": 4490
},
{
"grad_norm": 0.3063037395477295,
"learning_rate": 9.95636395173943e-05,
"loss": 0.0347,
"step": 4500
},
{
"grad_norm": 0.30606895685195923,
"learning_rate": 9.95592692508847e-05,
"loss": 0.0318,
"step": 4510
},
{
"grad_norm": 0.45677411556243896,
"learning_rate": 9.9554877305467e-05,
"loss": 0.0335,
"step": 4520
},
{
"grad_norm": 0.38935282826423645,
"learning_rate": 9.955046368306237e-05,
"loss": 0.0365,
"step": 4530
},
{
"grad_norm": 0.3629598617553711,
"learning_rate": 9.954602838560153e-05,
"loss": 0.0355,
"step": 4540
},
{
"grad_norm": 0.3336105942726135,
"learning_rate": 9.954157141502456e-05,
"loss": 0.0314,
"step": 4550
},
{
"grad_norm": 0.35257044434547424,
"learning_rate": 9.953709277328112e-05,
"loss": 0.0334,
"step": 4560
},
{
"grad_norm": 0.3599191904067993,
"learning_rate": 9.953259246233032e-05,
"loss": 0.0329,
"step": 4570
},
{
"grad_norm": 0.3616459369659424,
"learning_rate": 9.952807048414077e-05,
"loss": 0.038,
"step": 4580
},
{
"grad_norm": 0.434116005897522,
"learning_rate": 9.95235268406905e-05,
"loss": 0.0387,
"step": 4590
},
{
"grad_norm": 0.38972005248069763,
"learning_rate": 9.951896153396708e-05,
"loss": 0.0331,
"step": 4600
},
{
"grad_norm": 0.5667542815208435,
"learning_rate": 9.95143745659675e-05,
"loss": 0.0381,
"step": 4610
},
{
"grad_norm": 0.27837735414505005,
"learning_rate": 9.95097659386983e-05,
"loss": 0.0336,
"step": 4620
},
{
"grad_norm": 0.4152267277240753,
"learning_rate": 9.950513565417542e-05,
"loss": 0.0353,
"step": 4630
},
{
"grad_norm": 0.3352091312408447,
"learning_rate": 9.95004837144243e-05,
"loss": 0.036,
"step": 4640
},
{
"grad_norm": 0.29839518666267395,
"learning_rate": 9.949581012147988e-05,
"loss": 0.0337,
"step": 4650
},
{
"grad_norm": 0.2854501008987427,
"learning_rate": 9.949111487738653e-05,
"loss": 0.0349,
"step": 4660
},
{
"grad_norm": 0.4064194858074188,
"learning_rate": 9.948639798419813e-05,
"loss": 0.0339,
"step": 4670
},
{
"grad_norm": 0.35676681995391846,
"learning_rate": 9.948165944397799e-05,
"loss": 0.0299,
"step": 4680
},
{
"grad_norm": 0.4107678532600403,
"learning_rate": 9.94768992587989e-05,
"loss": 0.0338,
"step": 4690
},
{
"grad_norm": 0.4234924912452698,
"learning_rate": 9.947211743074313e-05,
"loss": 0.0429,
"step": 4700
},
{
"grad_norm": 0.3785814344882965,
"learning_rate": 9.946731396190246e-05,
"loss": 0.0365,
"step": 4710
},
{
"grad_norm": 0.30872663855552673,
"learning_rate": 9.946248885437803e-05,
"loss": 0.0307,
"step": 4720
},
{
"grad_norm": 0.29361608624458313,
"learning_rate": 9.945764211028053e-05,
"loss": 0.0325,
"step": 4730
},
{
"grad_norm": 0.44983023405075073,
"learning_rate": 9.94527737317301e-05,
"loss": 0.0341,
"step": 4740
},
{
"grad_norm": 0.3165139853954315,
"learning_rate": 9.944788372085631e-05,
"loss": 0.0342,
"step": 4750
},
{
"grad_norm": 0.3707144856452942,
"learning_rate": 9.944297207979825e-05,
"loss": 0.0355,
"step": 4760
},
{
"grad_norm": 0.32467520236968994,
"learning_rate": 9.943803881070441e-05,
"loss": 0.0382,
"step": 4770
},
{
"grad_norm": 0.2798452079296112,
"learning_rate": 9.943308391573278e-05,
"loss": 0.033,
"step": 4780
},
{
"grad_norm": 0.4505842924118042,
"learning_rate": 9.942810739705079e-05,
"loss": 0.0366,
"step": 4790
},
{
"grad_norm": 0.30525127053260803,
"learning_rate": 9.942310925683538e-05,
"loss": 0.0355,
"step": 4800
},
{
"grad_norm": 0.39801350235939026,
"learning_rate": 9.941808949727285e-05,
"loss": 0.031,
"step": 4810
},
{
"grad_norm": 0.38724657893180847,
"learning_rate": 9.941304812055903e-05,
"loss": 0.0338,
"step": 4820
},
{
"grad_norm": 0.35385754704475403,
"learning_rate": 9.940798512889921e-05,
"loss": 0.0325,
"step": 4830
},
{
"grad_norm": 0.2838450074195862,
"learning_rate": 9.94029005245081e-05,
"loss": 0.0338,
"step": 4840
},
{
"grad_norm": 0.3314788341522217,
"learning_rate": 9.939779430960988e-05,
"loss": 0.0337,
"step": 4850
},
{
"grad_norm": 0.366000235080719,
"learning_rate": 9.939266648643817e-05,
"loss": 0.0315,
"step": 4860
},
{
"grad_norm": 0.3744954466819763,
"learning_rate": 9.938751705723607e-05,
"loss": 0.0388,
"step": 4870
},
{
"grad_norm": 0.355292946100235,
"learning_rate": 9.938234602425613e-05,
"loss": 0.033,
"step": 4880
},
{
"grad_norm": 0.3489528298377991,
"learning_rate": 9.93771533897603e-05,
"loss": 0.0334,
"step": 4890
},
{
"grad_norm": 0.3268028199672699,
"learning_rate": 9.937193915602004e-05,
"loss": 0.0335,
"step": 4900
},
{
"grad_norm": 0.21146169304847717,
"learning_rate": 9.936670332531621e-05,
"loss": 0.0306,
"step": 4910
},
{
"grad_norm": 0.3154968023300171,
"learning_rate": 9.936144589993916e-05,
"loss": 0.035,
"step": 4920
},
{
"grad_norm": 0.45724233984947205,
"learning_rate": 9.935616688218867e-05,
"loss": 0.0356,
"step": 4930
},
{
"grad_norm": 0.3169020414352417,
"learning_rate": 9.935086627437395e-05,
"loss": 0.0306,
"step": 4940
},
{
"grad_norm": 0.28987300395965576,
"learning_rate": 9.934554407881366e-05,
"loss": 0.0354,
"step": 4950
},
{
"grad_norm": 0.2962575852870941,
"learning_rate": 9.934020029783593e-05,
"loss": 0.0353,
"step": 4960
},
{
"grad_norm": 0.2583666145801544,
"learning_rate": 9.933483493377829e-05,
"loss": 0.0296,
"step": 4970
},
{
"grad_norm": 0.304384708404541,
"learning_rate": 9.932944798898774e-05,
"loss": 0.0366,
"step": 4980
},
{
"grad_norm": 0.3262456953525543,
"learning_rate": 9.932403946582072e-05,
"loss": 0.0338,
"step": 4990
},
{
"grad_norm": 0.29961907863616943,
"learning_rate": 9.93186093666431e-05,
"loss": 0.0315,
"step": 5000
},
{
"grad_norm": 0.27681347727775574,
"learning_rate": 9.931315769383018e-05,
"loss": 0.0325,
"step": 5010
},
{
"grad_norm": 0.35694533586502075,
"learning_rate": 9.930768444976672e-05,
"loss": 0.0361,
"step": 5020
},
{
"grad_norm": 0.31411150097846985,
"learning_rate": 9.93021896368469e-05,
"loss": 0.0312,
"step": 5030
},
{
"grad_norm": 0.3105534613132477,
"learning_rate": 9.929667325747431e-05,
"loss": 0.0299,
"step": 5040
},
{
"grad_norm": 0.22749543190002441,
"learning_rate": 9.929113531406205e-05,
"loss": 0.0344,
"step": 5050
},
{
"grad_norm": 0.3893921673297882,
"learning_rate": 9.928557580903257e-05,
"loss": 0.0348,
"step": 5060
},
{
"grad_norm": 0.3666151165962219,
"learning_rate": 9.927999474481779e-05,
"loss": 0.0337,
"step": 5070
},
{
"grad_norm": 0.26170024275779724,
"learning_rate": 9.927439212385907e-05,
"loss": 0.0305,
"step": 5080
},
{
"grad_norm": 0.324009507894516,
"learning_rate": 9.926876794860718e-05,
"loss": 0.0336,
"step": 5090
},
{
"grad_norm": 0.3852055072784424,
"learning_rate": 9.926312222152235e-05,
"loss": 0.0345,
"step": 5100
},
{
"grad_norm": 0.309367835521698,
"learning_rate": 9.925745494507414e-05,
"loss": 0.0329,
"step": 5110
},
{
"grad_norm": 0.3883343040943146,
"learning_rate": 9.925176612174169e-05,
"loss": 0.0291,
"step": 5120
},
{
"grad_norm": 0.3420883119106293,
"learning_rate": 9.924605575401346e-05,
"loss": 0.0311,
"step": 5130
},
{
"grad_norm": 0.388572096824646,
"learning_rate": 9.924032384438733e-05,
"loss": 0.0336,
"step": 5140
},
{
"grad_norm": 0.35020217299461365,
"learning_rate": 9.923457039537066e-05,
"loss": 0.0326,
"step": 5150
},
{
"grad_norm": 0.3022792637348175,
"learning_rate": 9.92287954094802e-05,
"loss": 0.0335,
"step": 5160
},
{
"grad_norm": 0.33203238248825073,
"learning_rate": 9.922299888924212e-05,
"loss": 0.0294,
"step": 5170
},
{
"grad_norm": 0.3135257661342621,
"learning_rate": 9.921718083719203e-05,
"loss": 0.0281,
"step": 5180
},
{
"grad_norm": 0.3928318917751312,
"learning_rate": 9.921134125587491e-05,
"loss": 0.0345,
"step": 5190
},
{
"grad_norm": 0.3113667070865631,
"learning_rate": 9.920548014784523e-05,
"loss": 0.0325,
"step": 5200
},
{
"grad_norm": 0.3952057659626007,
"learning_rate": 9.919959751566681e-05,
"loss": 0.0376,
"step": 5210
},
{
"grad_norm": 0.3353860676288605,
"learning_rate": 9.919369336191291e-05,
"loss": 0.0383,
"step": 5220
},
{
"grad_norm": 0.3621787428855896,
"learning_rate": 9.918776768916625e-05,
"loss": 0.0348,
"step": 5230
},
{
"grad_norm": 0.39558669924736023,
"learning_rate": 9.918182050001888e-05,
"loss": 0.0291,
"step": 5240
},
{
"grad_norm": 0.3138406276702881,
"learning_rate": 9.91758517970723e-05,
"loss": 0.0348,
"step": 5250
},
{
"grad_norm": 0.3236713409423828,
"learning_rate": 9.916986158293744e-05,
"loss": 0.035,
"step": 5260
},
{
"grad_norm": 0.2943304479122162,
"learning_rate": 9.916384986023463e-05,
"loss": 0.0331,
"step": 5270
},
{
"grad_norm": 0.38190874457359314,
"learning_rate": 9.915781663159359e-05,
"loss": 0.0349,
"step": 5280
},
{
"grad_norm": 0.3932427763938904,
"learning_rate": 9.915176189965346e-05,
"loss": 0.0333,
"step": 5290
},
{
"grad_norm": 0.366428405046463,
"learning_rate": 9.914568566706279e-05,
"loss": 0.0368,
"step": 5300
},
{
"grad_norm": 0.4889920651912689,
"learning_rate": 9.913958793647953e-05,
"loss": 0.0406,
"step": 5310
},
{
"grad_norm": 0.3270202875137329,
"learning_rate": 9.913346871057102e-05,
"loss": 0.0407,
"step": 5320
},
{
"grad_norm": 0.36705857515335083,
"learning_rate": 9.912732799201403e-05,
"loss": 0.0319,
"step": 5330
},
{
"grad_norm": 0.37916940450668335,
"learning_rate": 9.912116578349474e-05,
"loss": 0.0345,
"step": 5340
},
{
"grad_norm": 0.316608726978302,
"learning_rate": 9.911498208770867e-05,
"loss": 0.0345,
"step": 5350
},
{
"grad_norm": 0.34297117590904236,
"learning_rate": 9.910877690736078e-05,
"loss": 0.029,
"step": 5360
},
{
"grad_norm": 0.22223787009716034,
"learning_rate": 9.910255024516546e-05,
"loss": 0.0304,
"step": 5370
},
{
"grad_norm": 0.31005755066871643,
"learning_rate": 9.909630210384644e-05,
"loss": 0.0322,
"step": 5380
},
{
"grad_norm": 0.24371714890003204,
"learning_rate": 9.909003248613688e-05,
"loss": 0.0302,
"step": 5390
},
{
"grad_norm": 0.28578463196754456,
"learning_rate": 9.908374139477934e-05,
"loss": 0.0287,
"step": 5400
},
{
"grad_norm": 0.2964169979095459,
"learning_rate": 9.907742883252571e-05,
"loss": 0.0336,
"step": 5410
},
{
"grad_norm": 0.2196938693523407,
"learning_rate": 9.907109480213736e-05,
"loss": 0.0319,
"step": 5420
},
{
"grad_norm": 0.3562096357345581,
"learning_rate": 9.906473930638498e-05,
"loss": 0.0351,
"step": 5430
},
{
"grad_norm": 0.4180792272090912,
"learning_rate": 9.90583623480487e-05,
"loss": 0.0326,
"step": 5440
},
{
"grad_norm": 0.3615745007991791,
"learning_rate": 9.905196392991802e-05,
"loss": 0.0341,
"step": 5450
},
{
"grad_norm": 0.23729492723941803,
"learning_rate": 9.904554405479183e-05,
"loss": 0.0331,
"step": 5460
},
{
"grad_norm": 0.31525325775146484,
"learning_rate": 9.903910272547838e-05,
"loss": 0.0318,
"step": 5470
},
{
"grad_norm": 0.30022549629211426,
"learning_rate": 9.903263994479533e-05,
"loss": 0.0329,
"step": 5480
},
{
"grad_norm": 0.3107682764530182,
"learning_rate": 9.902615571556972e-05,
"loss": 0.033,
"step": 5490
},
{
"grad_norm": 0.3192632794380188,
"learning_rate": 9.9019650040638e-05,
"loss": 0.0293,
"step": 5500
},
{
"grad_norm": 0.3539043962955475,
"learning_rate": 9.901312292284594e-05,
"loss": 0.0335,
"step": 5510
},
{
"grad_norm": 0.2970781922340393,
"learning_rate": 9.900657436504875e-05,
"loss": 0.0341,
"step": 5520
},
{
"grad_norm": 0.38731932640075684,
"learning_rate": 9.900000437011097e-05,
"loss": 0.0383,
"step": 5530
},
{
"grad_norm": 0.43598106503486633,
"learning_rate": 9.899341294090653e-05,
"loss": 0.0395,
"step": 5540
},
{
"grad_norm": 0.3633699119091034,
"learning_rate": 9.898680008031877e-05,
"loss": 0.0386,
"step": 5550
},
{
"grad_norm": 0.346344530582428,
"learning_rate": 9.898016579124037e-05,
"loss": 0.0361,
"step": 5560
},
{
"grad_norm": 0.3388567566871643,
"learning_rate": 9.89735100765734e-05,
"loss": 0.0357,
"step": 5570
},
{
"grad_norm": 0.36766669154167175,
"learning_rate": 9.89668329392293e-05,
"loss": 0.0309,
"step": 5580
},
{
"grad_norm": 0.2749458849430084,
"learning_rate": 9.896013438212885e-05,
"loss": 0.0325,
"step": 5590
},
{
"grad_norm": 0.25573599338531494,
"learning_rate": 9.895341440820225e-05,
"loss": 0.0315,
"step": 5600
},
{
"grad_norm": 0.35110175609588623,
"learning_rate": 9.894667302038902e-05,
"loss": 0.0314,
"step": 5610
},
{
"grad_norm": 0.42693403363227844,
"learning_rate": 9.893991022163812e-05,
"loss": 0.0325,
"step": 5620
},
{
"grad_norm": 0.2904145419597626,
"learning_rate": 9.893312601490779e-05,
"loss": 0.0291,
"step": 5630
},
{
"grad_norm": 0.3274777829647064,
"learning_rate": 9.892632040316568e-05,
"loss": 0.0289,
"step": 5640
},
{
"grad_norm": 0.2478848248720169,
"learning_rate": 9.891949338938877e-05,
"loss": 0.0287,
"step": 5650
},
{
"grad_norm": 0.291238009929657,
"learning_rate": 9.89126449765635e-05,
"loss": 0.0373,
"step": 5660
},
{
"grad_norm": 0.43003249168395996,
"learning_rate": 9.890577516768551e-05,
"loss": 0.029,
"step": 5670
},
{
"grad_norm": 0.32455873489379883,
"learning_rate": 9.889888396575995e-05,
"loss": 0.032,
"step": 5680
},
{
"grad_norm": 0.2745262384414673,
"learning_rate": 9.889197137380125e-05,
"loss": 0.0281,
"step": 5690
},
{
"grad_norm": 0.3585900664329529,
"learning_rate": 9.888503739483321e-05,
"loss": 0.0317,
"step": 5700
},
{
"grad_norm": 0.4106243848800659,
"learning_rate": 9.887808203188897e-05,
"loss": 0.0297,
"step": 5710
},
{
"grad_norm": 0.3652435839176178,
"learning_rate": 9.887110528801108e-05,
"loss": 0.0335,
"step": 5720
},
{
"grad_norm": 0.28336411714553833,
"learning_rate": 9.886410716625137e-05,
"loss": 0.0337,
"step": 5730
},
{
"grad_norm": 0.33693593740463257,
"learning_rate": 9.885708766967108e-05,
"loss": 0.0354,
"step": 5740
},
{
"grad_norm": 0.32107090950012207,
"learning_rate": 9.885004680134076e-05,
"loss": 0.0333,
"step": 5750
},
{
"grad_norm": 0.26424849033355713,
"learning_rate": 9.884298456434033e-05,
"loss": 0.0291,
"step": 5760
},
{
"grad_norm": 0.2907412648200989,
"learning_rate": 9.883590096175905e-05,
"loss": 0.0321,
"step": 5770
},
{
"grad_norm": 0.31279417872428894,
"learning_rate": 9.882879599669554e-05,
"loss": 0.0316,
"step": 5780
},
{
"grad_norm": 0.33282291889190674,
"learning_rate": 9.882166967225775e-05,
"loss": 0.034,
"step": 5790
},
{
"grad_norm": 0.358163058757782,
"learning_rate": 9.881452199156296e-05,
"loss": 0.031,
"step": 5800
},
{
"grad_norm": 0.25389009714126587,
"learning_rate": 9.880735295773782e-05,
"loss": 0.0287,
"step": 5810
},
{
"grad_norm": 0.36743617057800293,
"learning_rate": 9.88001625739183e-05,
"loss": 0.0324,
"step": 5820
},
{
"grad_norm": 0.3328872323036194,
"learning_rate": 9.879295084324976e-05,
"loss": 0.0334,
"step": 5830
},
{
"grad_norm": 0.34347450733184814,
"learning_rate": 9.87857177688868e-05,
"loss": 0.0324,
"step": 5840
},
{
"grad_norm": 0.28508445620536804,
"learning_rate": 9.877846335399343e-05,
"loss": 0.0322,
"step": 5850
},
{
"grad_norm": 0.2935417890548706,
"learning_rate": 9.8771187601743e-05,
"loss": 0.0334,
"step": 5860
},
{
"grad_norm": 0.29842033982276917,
"learning_rate": 9.876389051531813e-05,
"loss": 0.0347,
"step": 5870
},
{
"grad_norm": 0.385525107383728,
"learning_rate": 9.875657209791088e-05,
"loss": 0.0326,
"step": 5880
},
{
"grad_norm": 0.3795137107372284,
"learning_rate": 9.874923235272248e-05,
"loss": 0.03,
"step": 5890
},
{
"grad_norm": 0.4169817864894867,
"learning_rate": 9.874187128296369e-05,
"loss": 0.0315,
"step": 5900
},
{
"grad_norm": 0.22202163934707642,
"learning_rate": 9.873448889185439e-05,
"loss": 0.0313,
"step": 5910
},
{
"grad_norm": 0.4209665358066559,
"learning_rate": 9.872708518262397e-05,
"loss": 0.0299,
"step": 5920
},
{
"grad_norm": 0.3770769238471985,
"learning_rate": 9.871966015851101e-05,
"loss": 0.0261,
"step": 5930
},
{
"grad_norm": 0.3174689710140228,
"learning_rate": 9.87122138227635e-05,
"loss": 0.0324,
"step": 5940
},
{
"grad_norm": 0.26743605732917786,
"learning_rate": 9.87047461786387e-05,
"loss": 0.0266,
"step": 5950
},
{
"grad_norm": 0.4305480420589447,
"learning_rate": 9.869725722940323e-05,
"loss": 0.0285,
"step": 5960
},
{
"grad_norm": 0.2827478051185608,
"learning_rate": 9.868974697833299e-05,
"loss": 0.0311,
"step": 5970
},
{
"grad_norm": 0.2934468686580658,
"learning_rate": 9.868221542871324e-05,
"loss": 0.0282,
"step": 5980
},
{
"grad_norm": 0.2977944016456604,
"learning_rate": 9.867466258383853e-05,
"loss": 0.0309,
"step": 5990
},
{
"grad_norm": 0.3600994050502777,
"learning_rate": 9.866708844701272e-05,
"loss": 0.033,
"step": 6000
},
{
"grad_norm": 0.3719973564147949,
"learning_rate": 9.865949302154899e-05,
"loss": 0.0276,
"step": 6010
},
{
"grad_norm": 0.28464677929878235,
"learning_rate": 9.865187631076987e-05,
"loss": 0.0276,
"step": 6020
},
{
"grad_norm": 0.3651869595050812,
"learning_rate": 9.864423831800712e-05,
"loss": 0.0267,
"step": 6030
},
{
"grad_norm": 0.3451078534126282,
"learning_rate": 9.863657904660191e-05,
"loss": 0.0314,
"step": 6040
},
{
"grad_norm": 0.2570732533931732,
"learning_rate": 9.862889849990462e-05,
"loss": 0.0284,
"step": 6050
},
{
"grad_norm": 0.3290814757347107,
"learning_rate": 9.8621196681275e-05,
"loss": 0.0264,
"step": 6060
},
{
"grad_norm": 0.39716318249702454,
"learning_rate": 9.861347359408211e-05,
"loss": 0.0312,
"step": 6070
},
{
"grad_norm": 0.30969712138175964,
"learning_rate": 9.860572924170426e-05,
"loss": 0.0335,
"step": 6080
},
{
"grad_norm": 0.3009736239910126,
"learning_rate": 9.85979636275291e-05,
"loss": 0.0258,
"step": 6090
},
{
"grad_norm": 0.3390500843524933,
"learning_rate": 9.859017675495357e-05,
"loss": 0.0339,
"step": 6100
},
{
"grad_norm": 0.31303486227989197,
"learning_rate": 9.858236862738392e-05,
"loss": 0.0288,
"step": 6110
},
{
"grad_norm": 0.29130563139915466,
"learning_rate": 9.85745392482357e-05,
"loss": 0.0343,
"step": 6120
},
{
"grad_norm": 0.3054701089859009,
"learning_rate": 9.856668862093372e-05,
"loss": 0.0281,
"step": 6130
},
{
"grad_norm": 0.32072675228118896,
"learning_rate": 9.855881674891214e-05,
"loss": 0.0276,
"step": 6140
},
{
"grad_norm": 0.3291424512863159,
"learning_rate": 9.855092363561437e-05,
"loss": 0.0327,
"step": 6150
},
{
"grad_norm": 0.3477396070957184,
"learning_rate": 9.854300928449312e-05,
"loss": 0.0286,
"step": 6160
},
{
"grad_norm": 0.2765977382659912,
"learning_rate": 9.85350736990104e-05,
"loss": 0.0306,
"step": 6170
},
{
"grad_norm": 0.3779289722442627,
"learning_rate": 9.852711688263752e-05,
"loss": 0.029,
"step": 6180
},
{
"grad_norm": 0.2800227105617523,
"learning_rate": 9.851913883885503e-05,
"loss": 0.027,
"step": 6190
},
{
"grad_norm": 0.31865155696868896,
"learning_rate": 9.851113957115283e-05,
"loss": 0.0288,
"step": 6200
},
{
"grad_norm": 0.22526144981384277,
"learning_rate": 9.850311908303005e-05,
"loss": 0.0294,
"step": 6210
},
{
"grad_norm": 0.32080957293510437,
"learning_rate": 9.849507737799514e-05,
"loss": 0.032,
"step": 6220
},
{
"grad_norm": 0.2765636146068573,
"learning_rate": 9.84870144595658e-05,
"loss": 0.0298,
"step": 6230
},
{
"grad_norm": 0.2436542510986328,
"learning_rate": 9.847893033126903e-05,
"loss": 0.0285,
"step": 6240
},
{
"grad_norm": 0.2776239514350891,
"learning_rate": 9.847082499664111e-05,
"loss": 0.0334,
"step": 6250
},
{
"grad_norm": 0.342976838350296,
"learning_rate": 9.846269845922758e-05,
"loss": 0.0333,
"step": 6260
},
{
"grad_norm": 0.3209812641143799,
"learning_rate": 9.845455072258326e-05,
"loss": 0.0287,
"step": 6270
},
{
"grad_norm": 0.4500429928302765,
"learning_rate": 9.844638179027226e-05,
"loss": 0.032,
"step": 6280
},
{
"grad_norm": 0.33174213767051697,
"learning_rate": 9.843819166586795e-05,
"loss": 0.028,
"step": 6290
},
{
"grad_norm": 0.3553401529788971,
"learning_rate": 9.842998035295294e-05,
"loss": 0.0353,
"step": 6300
},
{
"grad_norm": 0.3887661397457123,
"learning_rate": 9.842174785511918e-05,
"loss": 0.0315,
"step": 6310
},
{
"grad_norm": 0.33735039830207825,
"learning_rate": 9.841349417596779e-05,
"loss": 0.0307,
"step": 6320
},
{
"grad_norm": 0.24815963208675385,
"learning_rate": 9.840521931910926e-05,
"loss": 0.031,
"step": 6330
},
{
"grad_norm": 0.34863215684890747,
"learning_rate": 9.839692328816327e-05,
"loss": 0.0313,
"step": 6340
},
{
"grad_norm": 0.3684467375278473,
"learning_rate": 9.838860608675879e-05,
"loss": 0.0329,
"step": 6350
},
{
"grad_norm": 0.3091985583305359,
"learning_rate": 9.838026771853406e-05,
"loss": 0.0351,
"step": 6360
},
{
"grad_norm": 0.342666894197464,
"learning_rate": 9.837190818713655e-05,
"loss": 0.0319,
"step": 6370
},
{
"grad_norm": 0.32199227809906006,
"learning_rate": 9.836352749622299e-05,
"loss": 0.0324,
"step": 6380
},
{
"grad_norm": 0.2459018975496292,
"learning_rate": 9.835512564945941e-05,
"loss": 0.0313,
"step": 6390
},
{
"grad_norm": 0.30317413806915283,
"learning_rate": 9.834670265052104e-05,
"loss": 0.0275,
"step": 6400
},
{
"grad_norm": 0.2592995762825012,
"learning_rate": 9.833825850309239e-05,
"loss": 0.0331,
"step": 6410
},
{
"grad_norm": 0.33231961727142334,
"learning_rate": 9.832979321086723e-05,
"loss": 0.0334,
"step": 6420
},
{
"grad_norm": 0.3258611857891083,
"learning_rate": 9.832130677754854e-05,
"loss": 0.0296,
"step": 6430
},
{
"grad_norm": 0.3048709034919739,
"learning_rate": 9.831279920684861e-05,
"loss": 0.0305,
"step": 6440
},
{
"grad_norm": 0.31239446997642517,
"learning_rate": 9.830427050248891e-05,
"loss": 0.03,
"step": 6450
},
{
"grad_norm": 0.3263850510120392,
"learning_rate": 9.829572066820021e-05,
"loss": 0.029,
"step": 6460
},
{
"grad_norm": 0.2974740266799927,
"learning_rate": 9.828714970772247e-05,
"loss": 0.0295,
"step": 6470
},
{
"grad_norm": 0.3160155713558197,
"learning_rate": 9.827855762480493e-05,
"loss": 0.033,
"step": 6480
},
{
"grad_norm": 0.33003395795822144,
"learning_rate": 9.826994442320608e-05,
"loss": 0.0308,
"step": 6490
},
{
"grad_norm": 0.27782559394836426,
"learning_rate": 9.82613101066936e-05,
"loss": 0.0268,
"step": 6500
},
{
"grad_norm": 0.3297159671783447,
"learning_rate": 9.825265467904445e-05,
"loss": 0.0289,
"step": 6510
},
{
"grad_norm": 0.2415226399898529,
"learning_rate": 9.82439781440448e-05,
"loss": 0.0277,
"step": 6520
},
{
"grad_norm": 0.2889556288719177,
"learning_rate": 9.823528050549006e-05,
"loss": 0.0285,
"step": 6530
},
{
"grad_norm": 0.3066122829914093,
"learning_rate": 9.822656176718487e-05,
"loss": 0.028,
"step": 6540
},
{
"grad_norm": 0.3318462669849396,
"learning_rate": 9.82178219329431e-05,
"loss": 0.0292,
"step": 6550
},
{
"grad_norm": 0.4174264669418335,
"learning_rate": 9.820906100658789e-05,
"loss": 0.0313,
"step": 6560
},
{
"grad_norm": 0.3039760887622833,
"learning_rate": 9.82002789919515e-05,
"loss": 0.0348,
"step": 6570
},
{
"grad_norm": 0.2670954465866089,
"learning_rate": 9.819147589287554e-05,
"loss": 0.0292,
"step": 6580
},
{
"grad_norm": 0.32622256875038147,
"learning_rate": 9.818265171321074e-05,
"loss": 0.0299,
"step": 6590
},
{
"grad_norm": 0.2653975486755371,
"learning_rate": 9.817380645681714e-05,
"loss": 0.0328,
"step": 6600
},
{
"grad_norm": 0.33552253246307373,
"learning_rate": 9.816494012756392e-05,
"loss": 0.0325,
"step": 6610
},
{
"grad_norm": 0.3273544907569885,
"learning_rate": 9.815605272932953e-05,
"loss": 0.0253,
"step": 6620
},
{
"grad_norm": 0.29956215620040894,
"learning_rate": 9.814714426600162e-05,
"loss": 0.0336,
"step": 6630
},
{
"grad_norm": 0.29610592126846313,
"learning_rate": 9.813821474147704e-05,
"loss": 0.0314,
"step": 6640
},
{
"grad_norm": 0.2388060986995697,
"learning_rate": 9.812926415966189e-05,
"loss": 0.0274,
"step": 6650
},
{
"grad_norm": 0.25543105602264404,
"learning_rate": 9.812029252447143e-05,
"loss": 0.0321,
"step": 6660
},
{
"grad_norm": 0.24750353395938873,
"learning_rate": 9.811129983983018e-05,
"loss": 0.03,
"step": 6670
},
{
"grad_norm": 0.36393025517463684,
"learning_rate": 9.810228610967185e-05,
"loss": 0.0311,
"step": 6680
},
{
"grad_norm": 0.3045990765094757,
"learning_rate": 9.809325133793934e-05,
"loss": 0.03,
"step": 6690
},
{
"grad_norm": 0.2678467333316803,
"learning_rate": 9.808419552858477e-05,
"loss": 0.0309,
"step": 6700
},
{
"grad_norm": 0.30948904156684875,
"learning_rate": 9.807511868556944e-05,
"loss": 0.0325,
"step": 6710
},
{
"grad_norm": 0.2672676146030426,
"learning_rate": 9.806602081286392e-05,
"loss": 0.0297,
"step": 6720
},
{
"grad_norm": 0.2727344334125519,
"learning_rate": 9.805690191444785e-05,
"loss": 0.0268,
"step": 6730
},
{
"grad_norm": 0.2761107385158539,
"learning_rate": 9.804776199431022e-05,
"loss": 0.0299,
"step": 6740
},
{
"grad_norm": 0.2995108366012573,
"learning_rate": 9.80386010564491e-05,
"loss": 0.0275,
"step": 6750
},
{
"grad_norm": 0.32105541229248047,
"learning_rate": 9.802941910487183e-05,
"loss": 0.0278,
"step": 6760
},
{
"grad_norm": 0.23851805925369263,
"learning_rate": 9.802021614359487e-05,
"loss": 0.0272,
"step": 6770
},
{
"grad_norm": 0.27570438385009766,
"learning_rate": 9.801099217664394e-05,
"loss": 0.0278,
"step": 6780
},
{
"grad_norm": 0.2814277708530426,
"learning_rate": 9.800174720805388e-05,
"loss": 0.0267,
"step": 6790
},
{
"grad_norm": 0.24848848581314087,
"learning_rate": 9.799248124186878e-05,
"loss": 0.0277,
"step": 6800
},
{
"grad_norm": 0.29091838002204895,
"learning_rate": 9.798319428214187e-05,
"loss": 0.0298,
"step": 6810
},
{
"grad_norm": 0.3515194058418274,
"learning_rate": 9.79738863329356e-05,
"loss": 0.0313,
"step": 6820
},
{
"grad_norm": 0.2511124908924103,
"learning_rate": 9.79645573983216e-05,
"loss": 0.0272,
"step": 6830
},
{
"grad_norm": 0.2849966287612915,
"learning_rate": 9.795520748238062e-05,
"loss": 0.0285,
"step": 6840
},
{
"grad_norm": 0.27315571904182434,
"learning_rate": 9.794583658920264e-05,
"loss": 0.0308,
"step": 6850
},
{
"grad_norm": 0.29518911242485046,
"learning_rate": 9.793644472288683e-05,
"loss": 0.033,
"step": 6860
},
{
"grad_norm": 0.2919248044490814,
"learning_rate": 9.79270318875415e-05,
"loss": 0.0253,
"step": 6870
},
{
"grad_norm": 0.3057709336280823,
"learning_rate": 9.791759808728416e-05,
"loss": 0.0307,
"step": 6880
},
{
"grad_norm": 0.24380561709403992,
"learning_rate": 9.790814332624143e-05,
"loss": 0.025,
"step": 6890
},
{
"grad_norm": 0.24087966978549957,
"learning_rate": 9.789866760854919e-05,
"loss": 0.0254,
"step": 6900
},
{
"grad_norm": 0.29995715618133545,
"learning_rate": 9.788917093835242e-05,
"loss": 0.0245,
"step": 6910
},
{
"grad_norm": 0.4264205992221832,
"learning_rate": 9.787965331980528e-05,
"loss": 0.0288,
"step": 6920
},
{
"grad_norm": 0.29289960861206055,
"learning_rate": 9.787011475707111e-05,
"loss": 0.0294,
"step": 6930
},
{
"grad_norm": 0.30965739488601685,
"learning_rate": 9.786055525432241e-05,
"loss": 0.0273,
"step": 6940
},
{
"grad_norm": 0.3240816593170166,
"learning_rate": 9.78509748157408e-05,
"loss": 0.0257,
"step": 6950
},
{
"grad_norm": 0.29325491189956665,
"learning_rate": 9.784137344551713e-05,
"loss": 0.0292,
"step": 6960
},
{
"grad_norm": 0.2223806232213974,
"learning_rate": 9.783175114785134e-05,
"loss": 0.0252,
"step": 6970
},
{
"grad_norm": 0.283572793006897,
"learning_rate": 9.782210792695254e-05,
"loss": 0.0274,
"step": 6980
},
{
"grad_norm": 0.2902943789958954,
"learning_rate": 9.781244378703904e-05,
"loss": 0.0262,
"step": 6990
},
{
"grad_norm": 0.30948299169540405,
"learning_rate": 9.780275873233824e-05,
"loss": 0.0275,
"step": 7000
},
{
"grad_norm": 0.24978826940059662,
"learning_rate": 9.77930527670867e-05,
"loss": 0.0299,
"step": 7010
},
{
"grad_norm": 0.269963800907135,
"learning_rate": 9.778332589553016e-05,
"loss": 0.0299,
"step": 7020
},
{
"grad_norm": 0.264514297246933,
"learning_rate": 9.777357812192349e-05,
"loss": 0.0252,
"step": 7030
},
{
"grad_norm": 0.3384951949119568,
"learning_rate": 9.776380945053066e-05,
"loss": 0.0277,
"step": 7040
},
{
"grad_norm": 0.24846996366977692,
"learning_rate": 9.775401988562486e-05,
"loss": 0.0258,
"step": 7050
},
{
"grad_norm": 0.2733401656150818,
"learning_rate": 9.774420943148835e-05,
"loss": 0.0281,
"step": 7060
},
{
"grad_norm": 0.31899774074554443,
"learning_rate": 9.773437809241257e-05,
"loss": 0.0254,
"step": 7070
},
{
"grad_norm": 0.3077262043952942,
"learning_rate": 9.772452587269808e-05,
"loss": 0.0273,
"step": 7080
},
{
"grad_norm": 0.2500836253166199,
"learning_rate": 9.771465277665457e-05,
"loss": 0.0278,
"step": 7090
},
{
"grad_norm": 0.2937280535697937,
"learning_rate": 9.770475880860089e-05,
"loss": 0.0249,
"step": 7100
},
{
"grad_norm": 0.2563726007938385,
"learning_rate": 9.769484397286495e-05,
"loss": 0.0269,
"step": 7110
},
{
"grad_norm": 0.2707694470882416,
"learning_rate": 9.768490827378388e-05,
"loss": 0.0274,
"step": 7120
},
{
"grad_norm": 0.2777843475341797,
"learning_rate": 9.767495171570387e-05,
"loss": 0.0236,
"step": 7130
},
{
"grad_norm": 0.3606092035770416,
"learning_rate": 9.766497430298027e-05,
"loss": 0.0288,
"step": 7140
},
{
"grad_norm": 0.2568415701389313,
"learning_rate": 9.765497603997754e-05,
"loss": 0.0248,
"step": 7150
},
{
"grad_norm": 0.38418450951576233,
"learning_rate": 9.764495693106926e-05,
"loss": 0.0276,
"step": 7160
},
{
"grad_norm": 0.31452423334121704,
"learning_rate": 9.76349169806381e-05,
"loss": 0.0262,
"step": 7170
},
{
"grad_norm": 0.36276009678840637,
"learning_rate": 9.762485619307591e-05,
"loss": 0.0286,
"step": 7180
},
{
"grad_norm": 0.32332664728164673,
"learning_rate": 9.76147745727836e-05,
"loss": 0.0265,
"step": 7190
},
{
"grad_norm": 0.3577694892883301,
"learning_rate": 9.760467212417124e-05,
"loss": 0.028,
"step": 7200
},
{
"grad_norm": 0.3289055824279785,
"learning_rate": 9.759454885165796e-05,
"loss": 0.0293,
"step": 7210
},
{
"grad_norm": 0.3744608163833618,
"learning_rate": 9.7584404759672e-05,
"loss": 0.0282,
"step": 7220
},
{
"grad_norm": 0.2586668133735657,
"learning_rate": 9.75742398526508e-05,
"loss": 0.0272,
"step": 7230
},
{
"grad_norm": 0.29433903098106384,
"learning_rate": 9.756405413504077e-05,
"loss": 0.0245,
"step": 7240
},
{
"grad_norm": 0.3648439943790436,
"learning_rate": 9.755384761129752e-05,
"loss": 0.0246,
"step": 7250
},
{
"grad_norm": 0.320087730884552,
"learning_rate": 9.754362028588572e-05,
"loss": 0.0254,
"step": 7260
},
{
"grad_norm": 0.3326079249382019,
"learning_rate": 9.753337216327917e-05,
"loss": 0.0256,
"step": 7270
},
{
"grad_norm": 0.2626769542694092,
"learning_rate": 9.752310324796073e-05,
"loss": 0.0241,
"step": 7280
},
{
"grad_norm": 0.2519007921218872,
"learning_rate": 9.751281354442235e-05,
"loss": 0.0273,
"step": 7290
},
{
"grad_norm": 0.2648239731788635,
"learning_rate": 9.750250305716513e-05,
"loss": 0.0269,
"step": 7300
},
{
"grad_norm": 0.34812217950820923,
"learning_rate": 9.749217179069923e-05,
"loss": 0.0254,
"step": 7310
},
{
"grad_norm": 0.2994368374347687,
"learning_rate": 9.748181974954386e-05,
"loss": 0.0269,
"step": 7320
},
{
"grad_norm": 0.2660892903804779,
"learning_rate": 9.74714469382274e-05,
"loss": 0.0301,
"step": 7330
},
{
"grad_norm": 0.2892247438430786,
"learning_rate": 9.746105336128724e-05,
"loss": 0.0279,
"step": 7340
},
{
"grad_norm": 0.28384554386138916,
"learning_rate": 9.745063902326988e-05,
"loss": 0.0283,
"step": 7350
},
{
"grad_norm": 0.19578608870506287,
"learning_rate": 9.744020392873093e-05,
"loss": 0.0253,
"step": 7360
},
{
"grad_norm": 0.2858792841434479,
"learning_rate": 9.742974808223504e-05,
"loss": 0.0262,
"step": 7370
},
{
"grad_norm": 0.3058668076992035,
"learning_rate": 9.741927148835596e-05,
"loss": 0.0255,
"step": 7380
},
{
"grad_norm": 0.3022083342075348,
"learning_rate": 9.740877415167648e-05,
"loss": 0.0275,
"step": 7390
},
{
"grad_norm": 0.29682692885398865,
"learning_rate": 9.739825607678854e-05,
"loss": 0.0283,
"step": 7400
},
{
"grad_norm": 0.3016035854816437,
"learning_rate": 9.738771726829308e-05,
"loss": 0.0315,
"step": 7410
},
{
"grad_norm": 0.34295621514320374,
"learning_rate": 9.737715773080011e-05,
"loss": 0.0282,
"step": 7420
},
{
"grad_norm": 0.3245771825313568,
"learning_rate": 9.736657746892877e-05,
"loss": 0.0357,
"step": 7430
},
{
"grad_norm": 0.26664912700653076,
"learning_rate": 9.735597648730718e-05,
"loss": 0.0293,
"step": 7440
},
{
"grad_norm": 0.28579097986221313,
"learning_rate": 9.734535479057262e-05,
"loss": 0.0278,
"step": 7450
},
{
"grad_norm": 0.2451818287372589,
"learning_rate": 9.733471238337136e-05,
"loss": 0.028,
"step": 7460
},
{
"grad_norm": 0.33957910537719727,
"learning_rate": 9.732404927035871e-05,
"loss": 0.0325,
"step": 7470
},
{
"grad_norm": 0.3972800374031067,
"learning_rate": 9.731336545619915e-05,
"loss": 0.0262,
"step": 7480
},
{
"grad_norm": 0.329716295003891,
"learning_rate": 9.73026609455661e-05,
"loss": 0.029,
"step": 7490
},
{
"grad_norm": 0.2676087021827698,
"learning_rate": 9.72919357431421e-05,
"loss": 0.0289,
"step": 7500
},
{
"grad_norm": 0.35349008440971375,
"learning_rate": 9.72811898536187e-05,
"loss": 0.0251,
"step": 7510
},
{
"grad_norm": 0.2381967306137085,
"learning_rate": 9.72704232816965e-05,
"loss": 0.023,
"step": 7520
},
{
"grad_norm": 0.22329097986221313,
"learning_rate": 9.725963603208519e-05,
"loss": 0.0268,
"step": 7530
},
{
"grad_norm": 0.23530860245227814,
"learning_rate": 9.724882810950348e-05,
"loss": 0.0283,
"step": 7540
},
{
"grad_norm": 0.24802492558956146,
"learning_rate": 9.723799951867913e-05,
"loss": 0.0277,
"step": 7550
},
{
"grad_norm": 0.25947779417037964,
"learning_rate": 9.722715026434889e-05,
"loss": 0.0284,
"step": 7560
},
{
"grad_norm": 0.26385605335235596,
"learning_rate": 9.721628035125866e-05,
"loss": 0.0238,
"step": 7570
},
{
"grad_norm": 0.22414451837539673,
"learning_rate": 9.720538978416325e-05,
"loss": 0.027,
"step": 7580
},
{
"grad_norm": 0.29326093196868896,
"learning_rate": 9.71944785678266e-05,
"loss": 0.0222,
"step": 7590
},
{
"grad_norm": 0.27058354020118713,
"learning_rate": 9.718354670702161e-05,
"loss": 0.0224,
"step": 7600
},
{
"grad_norm": 0.31051915884017944,
"learning_rate": 9.717259420653029e-05,
"loss": 0.026,
"step": 7610
},
{
"grad_norm": 0.3452931046485901,
"learning_rate": 9.716162107114361e-05,
"loss": 0.0264,
"step": 7620
},
{
"grad_norm": 0.28981807827949524,
"learning_rate": 9.71506273056616e-05,
"loss": 0.0256,
"step": 7630
},
{
"grad_norm": 0.28817668557167053,
"learning_rate": 9.713961291489331e-05,
"loss": 0.0311,
"step": 7640
},
{
"grad_norm": 0.2584552764892578,
"learning_rate": 9.71285779036568e-05,
"loss": 0.028,
"step": 7650
},
{
"grad_norm": 0.3475687801837921,
"learning_rate": 9.711752227677916e-05,
"loss": 0.0217,
"step": 7660
},
{
"grad_norm": 0.4387996792793274,
"learning_rate": 9.710644603909649e-05,
"loss": 0.0261,
"step": 7670
},
{
"grad_norm": 0.3155224919319153,
"learning_rate": 9.709534919545393e-05,
"loss": 0.0264,
"step": 7680
},
{
"grad_norm": 0.24571619927883148,
"learning_rate": 9.708423175070563e-05,
"loss": 0.0278,
"step": 7690
},
{
"grad_norm": 0.3032851815223694,
"learning_rate": 9.707309370971468e-05,
"loss": 0.0257,
"step": 7700
},
{
"grad_norm": 0.2854657769203186,
"learning_rate": 9.70619350773533e-05,
"loss": 0.0269,
"step": 7710
},
{
"grad_norm": 0.3204635977745056,
"learning_rate": 9.70507558585026e-05,
"loss": 0.0225,
"step": 7720
},
{
"grad_norm": 0.1806948035955429,
"learning_rate": 9.703955605805281e-05,
"loss": 0.0254,
"step": 7730
},
{
"grad_norm": 0.28693827986717224,
"learning_rate": 9.702833568090306e-05,
"loss": 0.0248,
"step": 7740
},
{
"grad_norm": 0.29066547751426697,
"learning_rate": 9.701709473196154e-05,
"loss": 0.0263,
"step": 7750
},
{
"grad_norm": 0.26554927229881287,
"learning_rate": 9.700583321614541e-05,
"loss": 0.0263,
"step": 7760
},
{
"grad_norm": 0.3386361300945282,
"learning_rate": 9.699455113838085e-05,
"loss": 0.0238,
"step": 7770
},
{
"grad_norm": 0.31081080436706543,
"learning_rate": 9.698324850360303e-05,
"loss": 0.0256,
"step": 7780
},
{
"grad_norm": 0.27737879753112793,
"learning_rate": 9.69719253167561e-05,
"loss": 0.0224,
"step": 7790
},
{
"grad_norm": 0.2834523916244507,
"learning_rate": 9.69605815827932e-05,
"loss": 0.0293,
"step": 7800
},
{
"grad_norm": 0.2754724323749542,
"learning_rate": 9.694921730667647e-05,
"loss": 0.0282,
"step": 7810
},
{
"grad_norm": 0.2540271580219269,
"learning_rate": 9.693783249337705e-05,
"loss": 0.0252,
"step": 7820
},
{
"grad_norm": 0.21979637444019318,
"learning_rate": 9.692642714787501e-05,
"loss": 0.0238,
"step": 7830
},
{
"grad_norm": 0.3163398802280426,
"learning_rate": 9.691500127515945e-05,
"loss": 0.0246,
"step": 7840
},
{
"grad_norm": 0.27172935009002686,
"learning_rate": 9.690355488022844e-05,
"loss": 0.0213,
"step": 7850
},
{
"grad_norm": 0.2636539340019226,
"learning_rate": 9.689208796808902e-05,
"loss": 0.0231,
"step": 7860
},
{
"grad_norm": 0.25851958990097046,
"learning_rate": 9.68806005437572e-05,
"loss": 0.022,
"step": 7870
},
{
"grad_norm": 0.31418976187705994,
"learning_rate": 9.686909261225796e-05,
"loss": 0.0287,
"step": 7880
},
{
"grad_norm": 0.2511616051197052,
"learning_rate": 9.685756417862531e-05,
"loss": 0.0231,
"step": 7890
},
{
"grad_norm": 0.215915709733963,
"learning_rate": 9.684601524790212e-05,
"loss": 0.0241,
"step": 7900
},
{
"grad_norm": 0.24998539686203003,
"learning_rate": 9.68344458251403e-05,
"loss": 0.0288,
"step": 7910
},
{
"grad_norm": 0.2416553944349289,
"learning_rate": 9.682285591540072e-05,
"loss": 0.0253,
"step": 7920
},
{
"grad_norm": 0.30611878633499146,
"learning_rate": 9.681124552375322e-05,
"loss": 0.0268,
"step": 7930
},
{
"grad_norm": 0.23860104382038116,
"learning_rate": 9.679961465527654e-05,
"loss": 0.0269,
"step": 7940
},
{
"grad_norm": 0.3028927445411682,
"learning_rate": 9.678796331505843e-05,
"loss": 0.0217,
"step": 7950
},
{
"grad_norm": 0.2753969728946686,
"learning_rate": 9.677629150819558e-05,
"loss": 0.0253,
"step": 7960
},
{
"grad_norm": 0.23730877041816711,
"learning_rate": 9.676459923979366e-05,
"loss": 0.0248,
"step": 7970
},
{
"grad_norm": 0.2380107194185257,
"learning_rate": 9.675288651496723e-05,
"loss": 0.0257,
"step": 7980
},
{
"grad_norm": 0.31520572304725647,
"learning_rate": 9.674115333883986e-05,
"loss": 0.0223,
"step": 7990
},
{
"grad_norm": 0.28640514612197876,
"learning_rate": 9.672939971654402e-05,
"loss": 0.0221,
"step": 8000
},
{
"grad_norm": 0.23724617063999176,
"learning_rate": 9.671762565322117e-05,
"loss": 0.0237,
"step": 8010
},
{
"grad_norm": 0.2490135133266449,
"learning_rate": 9.670583115402166e-05,
"loss": 0.027,
"step": 8020
},
{
"grad_norm": 0.25580787658691406,
"learning_rate": 9.669401622410482e-05,
"loss": 0.0221,
"step": 8030
},
{
"grad_norm": 0.3217068910598755,
"learning_rate": 9.668218086863887e-05,
"loss": 0.0264,
"step": 8040
},
{
"grad_norm": 0.24185623228549957,
"learning_rate": 9.667032509280103e-05,
"loss": 0.0226,
"step": 8050
},
{
"grad_norm": 0.2768763303756714,
"learning_rate": 9.665844890177743e-05,
"loss": 0.0226,
"step": 8060
},
{
"grad_norm": 0.22353193163871765,
"learning_rate": 9.664655230076309e-05,
"loss": 0.0256,
"step": 8070
},
{
"grad_norm": 0.2809293568134308,
"learning_rate": 9.663463529496199e-05,
"loss": 0.0257,
"step": 8080
},
{
"grad_norm": 0.31753069162368774,
"learning_rate": 9.662269788958705e-05,
"loss": 0.0236,
"step": 8090
},
{
"grad_norm": 0.2565624415874481,
"learning_rate": 9.661074008986008e-05,
"loss": 0.0243,
"step": 8100
},
{
"grad_norm": 0.22407996654510498,
"learning_rate": 9.659876190101184e-05,
"loss": 0.0281,
"step": 8110
},
{
"grad_norm": 0.3168320655822754,
"learning_rate": 9.6586763328282e-05,
"loss": 0.0259,
"step": 8120
},
{
"grad_norm": 0.27397286891937256,
"learning_rate": 9.657474437691913e-05,
"loss": 0.0245,
"step": 8130
},
{
"grad_norm": 0.34911566972732544,
"learning_rate": 9.656270505218073e-05,
"loss": 0.0265,
"step": 8140
},
{
"grad_norm": 0.2502925992012024,
"learning_rate": 9.655064535933321e-05,
"loss": 0.0262,
"step": 8150
},
{
"grad_norm": 0.25083398818969727,
"learning_rate": 9.653856530365189e-05,
"loss": 0.0244,
"step": 8160
},
{
"grad_norm": 0.29719749093055725,
"learning_rate": 9.6526464890421e-05,
"loss": 0.0235,
"step": 8170
},
{
"grad_norm": 0.3182894289493561,
"learning_rate": 9.651434412493367e-05,
"loss": 0.0251,
"step": 8180
},
{
"grad_norm": 0.22344397008419037,
"learning_rate": 9.650220301249195e-05,
"loss": 0.0242,
"step": 8190
},
{
"grad_norm": 0.27554601430892944,
"learning_rate": 9.649004155840675e-05,
"loss": 0.0202,
"step": 8200
},
{
"grad_norm": 0.2753584682941437,
"learning_rate": 9.647785976799792e-05,
"loss": 0.0224,
"step": 8210
},
{
"grad_norm": 0.2775469720363617,
"learning_rate": 9.646565764659417e-05,
"loss": 0.0217,
"step": 8220
},
{
"grad_norm": 0.3759711682796478,
"learning_rate": 9.645343519953314e-05,
"loss": 0.0231,
"step": 8230
},
{
"grad_norm": 0.3116992115974426,
"learning_rate": 9.644119243216135e-05,
"loss": 0.0214,
"step": 8240
},
{
"grad_norm": 0.22607944905757904,
"learning_rate": 9.642892934983417e-05,
"loss": 0.0221,
"step": 8250
},
{
"grad_norm": 0.2502546012401581,
"learning_rate": 9.641664595791593e-05,
"loss": 0.0245,
"step": 8260
},
{
"grad_norm": 0.209568589925766,
"learning_rate": 9.640434226177977e-05,
"loss": 0.0259,
"step": 8270
},
{
"grad_norm": 0.27021053433418274,
"learning_rate": 9.639201826680777e-05,
"loss": 0.0244,
"step": 8280
},
{
"grad_norm": 0.2539534568786621,
"learning_rate": 9.637967397839083e-05,
"loss": 0.0241,
"step": 8290
},
{
"grad_norm": 0.297733336687088,
"learning_rate": 9.63673094019288e-05,
"loss": 0.0214,
"step": 8300
},
{
"grad_norm": 0.23181241750717163,
"learning_rate": 9.635492454283035e-05,
"loss": 0.0248,
"step": 8310
},
{
"grad_norm": 0.31577733159065247,
"learning_rate": 9.634251940651302e-05,
"loss": 0.0248,
"step": 8320
},
{
"grad_norm": 0.2532350420951843,
"learning_rate": 9.633009399840327e-05,
"loss": 0.0219,
"step": 8330
},
{
"grad_norm": 0.3019818067550659,
"learning_rate": 9.631764832393639e-05,
"loss": 0.0237,
"step": 8340
},
{
"grad_norm": 0.19652318954467773,
"learning_rate": 9.630518238855651e-05,
"loss": 0.0209,
"step": 8350
},
{
"grad_norm": 0.3000640273094177,
"learning_rate": 9.629269619771668e-05,
"loss": 0.0206,
"step": 8360
},
{
"grad_norm": 0.2535010874271393,
"learning_rate": 9.628018975687878e-05,
"loss": 0.0213,
"step": 8370
},
{
"grad_norm": 0.25915199518203735,
"learning_rate": 9.626766307151355e-05,
"loss": 0.0229,
"step": 8380
},
{
"grad_norm": 0.31963661313056946,
"learning_rate": 9.62551161471006e-05,
"loss": 0.0235,
"step": 8390
},
{
"grad_norm": 0.2695634663105011,
"learning_rate": 9.624254898912837e-05,
"loss": 0.0235,
"step": 8400
},
{
"grad_norm": 0.3161536753177643,
"learning_rate": 9.622996160309414e-05,
"loss": 0.022,
"step": 8410
},
{
"grad_norm": 0.2878867983818054,
"learning_rate": 9.62173539945041e-05,
"loss": 0.0222,
"step": 8420
},
{
"grad_norm": 0.2850261926651001,
"learning_rate": 9.620472616887323e-05,
"loss": 0.0203,
"step": 8430
},
{
"grad_norm": 0.2819199562072754,
"learning_rate": 9.619207813172536e-05,
"loss": 0.0239,
"step": 8440
},
{
"grad_norm": 0.2379349172115326,
"learning_rate": 9.617940988859318e-05,
"loss": 0.0239,
"step": 8450
},
{
"grad_norm": 0.31933924555778503,
"learning_rate": 9.616672144501821e-05,
"loss": 0.027,
"step": 8460
},
{
"grad_norm": 0.2723536789417267,
"learning_rate": 9.615401280655081e-05,
"loss": 0.0256,
"step": 8470
},
{
"grad_norm": 0.29725322127342224,
"learning_rate": 9.614128397875017e-05,
"loss": 0.0227,
"step": 8480
},
{
"grad_norm": 0.2414250522851944,
"learning_rate": 9.612853496718429e-05,
"loss": 0.0216,
"step": 8490
},
{
"grad_norm": 0.25188010931015015,
"learning_rate": 9.611576577743004e-05,
"loss": 0.0227,
"step": 8500
},
{
"grad_norm": 0.3515135943889618,
"learning_rate": 9.610297641507313e-05,
"loss": 0.0244,
"step": 8510
},
{
"grad_norm": 0.294953852891922,
"learning_rate": 9.6090166885708e-05,
"loss": 0.0263,
"step": 8520
},
{
"grad_norm": 0.24369871616363525,
"learning_rate": 9.607733719493798e-05,
"loss": 0.0278,
"step": 8530
},
{
"grad_norm": 0.28182312846183777,
"learning_rate": 9.606448734837527e-05,
"loss": 0.0259,
"step": 8540
},
{
"grad_norm": 0.29961687326431274,
"learning_rate": 9.605161735164079e-05,
"loss": 0.0242,
"step": 8550
},
{
"grad_norm": 0.30143076181411743,
"learning_rate": 9.60387272103643e-05,
"loss": 0.0254,
"step": 8560
},
{
"grad_norm": 0.24056176841259003,
"learning_rate": 9.60258169301844e-05,
"loss": 0.0206,
"step": 8570
},
{
"grad_norm": 0.224228173494339,
"learning_rate": 9.601288651674851e-05,
"loss": 0.023,
"step": 8580
},
{
"grad_norm": 0.24074408411979675,
"learning_rate": 9.59999359757128e-05,
"loss": 0.023,
"step": 8590
},
{
"grad_norm": 0.23999366164207458,
"learning_rate": 9.598696531274227e-05,
"loss": 0.0211,
"step": 8600
},
{
"grad_norm": 0.357965350151062,
"learning_rate": 9.597397453351076e-05,
"loss": 0.0225,
"step": 8610
},
{
"grad_norm": 0.2597416341304779,
"learning_rate": 9.596096364370087e-05,
"loss": 0.0233,
"step": 8620
},
{
"grad_norm": 0.29217636585235596,
"learning_rate": 9.594793264900399e-05,
"loss": 0.0206,
"step": 8630
},
{
"grad_norm": 0.2660157084465027,
"learning_rate": 9.593488155512032e-05,
"loss": 0.0231,
"step": 8640
},
{
"grad_norm": 0.18962566554546356,
"learning_rate": 9.592181036775886e-05,
"loss": 0.0234,
"step": 8650
},
{
"grad_norm": 0.2694273293018341,
"learning_rate": 9.590871909263741e-05,
"loss": 0.0246,
"step": 8660
},
{
"grad_norm": 0.32581937313079834,
"learning_rate": 9.589560773548252e-05,
"loss": 0.0253,
"step": 8670
},
{
"grad_norm": 0.2393707036972046,
"learning_rate": 9.588247630202952e-05,
"loss": 0.0252,
"step": 8680
},
{
"grad_norm": 0.2798861265182495,
"learning_rate": 9.586932479802258e-05,
"loss": 0.023,
"step": 8690
},
{
"grad_norm": 0.309002161026001,
"learning_rate": 9.585615322921462e-05,
"loss": 0.0218,
"step": 8700
},
{
"grad_norm": 0.19604802131652832,
"learning_rate": 9.58429616013673e-05,
"loss": 0.0215,
"step": 8710
},
{
"grad_norm": 0.3414742648601532,
"learning_rate": 9.58297499202511e-05,
"loss": 0.0193,
"step": 8720
},
{
"grad_norm": 0.33937686681747437,
"learning_rate": 9.581651819164525e-05,
"loss": 0.0246,
"step": 8730
},
{
"grad_norm": 0.26246705651283264,
"learning_rate": 9.580326642133777e-05,
"loss": 0.0237,
"step": 8740
},
{
"grad_norm": 0.2491609752178192,
"learning_rate": 9.578999461512544e-05,
"loss": 0.0273,
"step": 8750
},
{
"grad_norm": 0.24505281448364258,
"learning_rate": 9.577670277881378e-05,
"loss": 0.0221,
"step": 8760
},
{
"grad_norm": 0.25111615657806396,
"learning_rate": 9.57633909182171e-05,
"loss": 0.019,
"step": 8770
},
{
"grad_norm": 0.20893190801143646,
"learning_rate": 9.575005903915847e-05,
"loss": 0.02,
"step": 8780
},
{
"grad_norm": 0.24406443536281586,
"learning_rate": 9.573670714746972e-05,
"loss": 0.0198,
"step": 8790
},
{
"grad_norm": 0.23990103602409363,
"learning_rate": 9.572333524899136e-05,
"loss": 0.0246,
"step": 8800
},
{
"grad_norm": 0.22216957807540894,
"learning_rate": 9.570994334957278e-05,
"loss": 0.0229,
"step": 8810
},
{
"grad_norm": 0.23326730728149414,
"learning_rate": 9.569653145507201e-05,
"loss": 0.0236,
"step": 8820
},
{
"grad_norm": 0.2843077480792999,
"learning_rate": 9.568309957135587e-05,
"loss": 0.028,
"step": 8830
},
{
"grad_norm": 0.2572084963321686,
"learning_rate": 9.566964770429995e-05,
"loss": 0.0234,
"step": 8840
},
{
"grad_norm": 0.27594682574272156,
"learning_rate": 9.565617585978853e-05,
"loss": 0.0257,
"step": 8850
},
{
"grad_norm": 0.239472433924675,
"learning_rate": 9.564268404371466e-05,
"loss": 0.0244,
"step": 8860
},
{
"grad_norm": 0.22838908433914185,
"learning_rate": 9.562917226198013e-05,
"loss": 0.0251,
"step": 8870
},
{
"grad_norm": 0.31908294558525085,
"learning_rate": 9.561564052049543e-05,
"loss": 0.0223,
"step": 8880
},
{
"grad_norm": 0.24166066944599152,
"learning_rate": 9.560208882517982e-05,
"loss": 0.0257,
"step": 8890
},
{
"grad_norm": 0.304926335811615,
"learning_rate": 9.558851718196127e-05,
"loss": 0.0192,
"step": 8900
},
{
"grad_norm": 0.2896885871887207,
"learning_rate": 9.557492559677646e-05,
"loss": 0.0213,
"step": 8910
},
{
"grad_norm": 0.2170281857252121,
"learning_rate": 9.556131407557082e-05,
"loss": 0.0204,
"step": 8920
},
{
"grad_norm": 0.25211814045906067,
"learning_rate": 9.554768262429853e-05,
"loss": 0.0209,
"step": 8930
},
{
"grad_norm": 0.28438377380371094,
"learning_rate": 9.553403124892239e-05,
"loss": 0.02,
"step": 8940
},
{
"grad_norm": 0.38985973596572876,
"learning_rate": 9.552035995541402e-05,
"loss": 0.0232,
"step": 8950
},
{
"grad_norm": 0.3205428421497345,
"learning_rate": 9.550666874975368e-05,
"loss": 0.0259,
"step": 8960
},
{
"grad_norm": 0.25402259826660156,
"learning_rate": 9.549295763793038e-05,
"loss": 0.0226,
"step": 8970
},
{
"grad_norm": 0.16712293028831482,
"learning_rate": 9.547922662594183e-05,
"loss": 0.0242,
"step": 8980
},
{
"grad_norm": 0.3072149455547333,
"learning_rate": 9.546547571979443e-05,
"loss": 0.0222,
"step": 8990
},
{
"grad_norm": 0.206142395734787,
"learning_rate": 9.545170492550331e-05,
"loss": 0.0258,
"step": 9000
},
{
"grad_norm": 0.32440656423568726,
"learning_rate": 9.543791424909226e-05,
"loss": 0.0259,
"step": 9010
},
{
"grad_norm": 0.24773822724819183,
"learning_rate": 9.542410369659382e-05,
"loss": 0.0207,
"step": 9020
},
{
"grad_norm": 0.20210446417331696,
"learning_rate": 9.541027327404916e-05,
"loss": 0.024,
"step": 9030
},
{
"grad_norm": 0.28600460290908813,
"learning_rate": 9.539642298750821e-05,
"loss": 0.023,
"step": 9040
},
{
"grad_norm": 0.23598986864089966,
"learning_rate": 9.538255284302954e-05,
"loss": 0.0198,
"step": 9050
},
{
"grad_norm": 0.20501859486103058,
"learning_rate": 9.536866284668043e-05,
"loss": 0.0204,
"step": 9060
},
{
"grad_norm": 0.2832529544830322,
"learning_rate": 9.53547530045368e-05,
"loss": 0.0222,
"step": 9070
},
{
"grad_norm": 0.3008792996406555,
"learning_rate": 9.534082332268335e-05,
"loss": 0.0237,
"step": 9080
},
{
"grad_norm": 0.2605428695678711,
"learning_rate": 9.532687380721334e-05,
"loss": 0.0221,
"step": 9090
},
{
"grad_norm": 0.2775779962539673,
"learning_rate": 9.53129044642288e-05,
"loss": 0.0224,
"step": 9100
},
{
"grad_norm": 0.2296871393918991,
"learning_rate": 9.529891529984039e-05,
"loss": 0.0224,
"step": 9110
},
{
"grad_norm": 0.24996177852153778,
"learning_rate": 9.528490632016743e-05,
"loss": 0.0226,
"step": 9120
},
{
"grad_norm": 0.30968156456947327,
"learning_rate": 9.52708775313379e-05,
"loss": 0.0223,
"step": 9130
},
{
"grad_norm": 0.29486387968063354,
"learning_rate": 9.525682893948853e-05,
"loss": 0.0224,
"step": 9140
},
{
"grad_norm": 0.2608632445335388,
"learning_rate": 9.524276055076463e-05,
"loss": 0.0204,
"step": 9150
},
{
"grad_norm": 0.25088000297546387,
"learning_rate": 9.522867237132016e-05,
"loss": 0.0208,
"step": 9160
},
{
"grad_norm": 0.29771706461906433,
"learning_rate": 9.52145644073178e-05,
"loss": 0.0239,
"step": 9170
},
{
"grad_norm": 0.29451388120651245,
"learning_rate": 9.520043666492884e-05,
"loss": 0.0203,
"step": 9180
},
{
"grad_norm": 0.23304559290409088,
"learning_rate": 9.518628915033323e-05,
"loss": 0.0221,
"step": 9190
},
{
"grad_norm": 0.2494295835494995,
"learning_rate": 9.517212186971957e-05,
"loss": 0.022,
"step": 9200
},
{
"grad_norm": 0.23576436936855316,
"learning_rate": 9.515793482928515e-05,
"loss": 0.0232,
"step": 9210
},
{
"grad_norm": 0.2726961672306061,
"learning_rate": 9.514372803523582e-05,
"loss": 0.0246,
"step": 9220
},
{
"grad_norm": 0.2853924334049225,
"learning_rate": 9.512950149378613e-05,
"loss": 0.0205,
"step": 9230
},
{
"grad_norm": 0.3077450096607208,
"learning_rate": 9.511525521115925e-05,
"loss": 0.0235,
"step": 9240
},
{
"grad_norm": 0.23647278547286987,
"learning_rate": 9.510098919358698e-05,
"loss": 0.0204,
"step": 9250
},
{
"grad_norm": 0.2955886423587799,
"learning_rate": 9.508670344730979e-05,
"loss": 0.0204,
"step": 9260
},
{
"grad_norm": 0.21963225305080414,
"learning_rate": 9.507239797857672e-05,
"loss": 0.0216,
"step": 9270
},
{
"grad_norm": 0.2760219871997833,
"learning_rate": 9.505807279364548e-05,
"loss": 0.0219,
"step": 9280
},
{
"grad_norm": 0.26910877227783203,
"learning_rate": 9.504372789878239e-05,
"loss": 0.0219,
"step": 9290
},
{
"grad_norm": 0.2729431986808777,
"learning_rate": 9.502936330026239e-05,
"loss": 0.0224,
"step": 9300
},
{
"grad_norm": 0.2641892731189728,
"learning_rate": 9.501497900436906e-05,
"loss": 0.0198,
"step": 9310
},
{
"grad_norm": 0.27855247259140015,
"learning_rate": 9.500057501739457e-05,
"loss": 0.0195,
"step": 9320
},
{
"grad_norm": 0.2845800220966339,
"learning_rate": 9.49861513456397e-05,
"loss": 0.0208,
"step": 9330
},
{
"grad_norm": 0.2375309318304062,
"learning_rate": 9.497170799541388e-05,
"loss": 0.0192,
"step": 9340
},
{
"grad_norm": 0.30093836784362793,
"learning_rate": 9.495724497303508e-05,
"loss": 0.0224,
"step": 9350
},
{
"grad_norm": 0.24772000312805176,
"learning_rate": 9.494276228482998e-05,
"loss": 0.0208,
"step": 9360
},
{
"grad_norm": 0.232768252491951,
"learning_rate": 9.492825993713374e-05,
"loss": 0.0178,
"step": 9370
},
{
"grad_norm": 0.23551689088344574,
"learning_rate": 9.491373793629023e-05,
"loss": 0.021,
"step": 9380
},
{
"grad_norm": 0.2408761978149414,
"learning_rate": 9.489919628865182e-05,
"loss": 0.0181,
"step": 9390
},
{
"grad_norm": 0.28038832545280457,
"learning_rate": 9.488463500057955e-05,
"loss": 0.0213,
"step": 9400
},
{
"grad_norm": 0.3298349678516388,
"learning_rate": 9.487005407844302e-05,
"loss": 0.0189,
"step": 9410
},
{
"grad_norm": 0.2700117230415344,
"learning_rate": 9.485545352862039e-05,
"loss": 0.0222,
"step": 9420
},
{
"grad_norm": 0.1923956722021103,
"learning_rate": 9.48408333574985e-05,
"loss": 0.023,
"step": 9430
},
{
"grad_norm": 0.31371650099754333,
"learning_rate": 9.482619357147264e-05,
"loss": 0.0206,
"step": 9440
},
{
"grad_norm": 0.25857922434806824,
"learning_rate": 9.481153417694679e-05,
"loss": 0.0228,
"step": 9450
},
{
"grad_norm": 0.18859760463237762,
"learning_rate": 9.479685518033347e-05,
"loss": 0.019,
"step": 9460
},
{
"grad_norm": 0.2824561297893524,
"learning_rate": 9.478215658805375e-05,
"loss": 0.0192,
"step": 9470
},
{
"grad_norm": 0.34904444217681885,
"learning_rate": 9.47674384065373e-05,
"loss": 0.0228,
"step": 9480
},
{
"grad_norm": 0.2556479573249817,
"learning_rate": 9.475270064222237e-05,
"loss": 0.0216,
"step": 9490
},
{
"grad_norm": 0.23711754381656647,
"learning_rate": 9.473794330155572e-05,
"loss": 0.0219,
"step": 9500
},
{
"grad_norm": 0.22615624964237213,
"learning_rate": 9.472316639099276e-05,
"loss": 0.0217,
"step": 9510
},
{
"grad_norm": 0.22031806409358978,
"learning_rate": 9.470836991699739e-05,
"loss": 0.0188,
"step": 9520
},
{
"grad_norm": 0.19253870844841003,
"learning_rate": 9.469355388604208e-05,
"loss": 0.0161,
"step": 9530
},
{
"grad_norm": 0.18653284013271332,
"learning_rate": 9.467871830460787e-05,
"loss": 0.0192,
"step": 9540
},
{
"grad_norm": 0.25115537643432617,
"learning_rate": 9.466386317918436e-05,
"loss": 0.0249,
"step": 9550
},
{
"grad_norm": 0.37245863676071167,
"learning_rate": 9.464898851626969e-05,
"loss": 0.0224,
"step": 9560
},
{
"grad_norm": 0.27260255813598633,
"learning_rate": 9.463409432237051e-05,
"loss": 0.0208,
"step": 9570
},
{
"grad_norm": 0.2518012523651123,
"learning_rate": 9.461918060400209e-05,
"loss": 0.0232,
"step": 9580
},
{
"grad_norm": 0.31925898790359497,
"learning_rate": 9.460424736768816e-05,
"loss": 0.0247,
"step": 9590
},
{
"grad_norm": 0.2036992609500885,
"learning_rate": 9.458929461996105e-05,
"loss": 0.0208,
"step": 9600
},
{
"grad_norm": 0.2276897132396698,
"learning_rate": 9.457432236736158e-05,
"loss": 0.0237,
"step": 9610
},
{
"grad_norm": 0.23287692666053772,
"learning_rate": 9.455933061643916e-05,
"loss": 0.018,
"step": 9620
},
{
"grad_norm": 0.2919159531593323,
"learning_rate": 9.454431937375164e-05,
"loss": 0.0202,
"step": 9630
},
{
"grad_norm": 0.2592717111110687,
"learning_rate": 9.452928864586547e-05,
"loss": 0.0238,
"step": 9640
},
{
"grad_norm": 0.33278924226760864,
"learning_rate": 9.451423843935563e-05,
"loss": 0.0235,
"step": 9650
},
{
"grad_norm": 0.23154206573963165,
"learning_rate": 9.449916876080553e-05,
"loss": 0.0225,
"step": 9660
},
{
"grad_norm": 0.26563605666160583,
"learning_rate": 9.44840796168072e-05,
"loss": 0.0176,
"step": 9670
},
{
"grad_norm": 0.20860138535499573,
"learning_rate": 9.446897101396115e-05,
"loss": 0.0177,
"step": 9680
},
{
"grad_norm": 0.271651029586792,
"learning_rate": 9.445384295887638e-05,
"loss": 0.0225,
"step": 9690
},
{
"grad_norm": 0.26481950283050537,
"learning_rate": 9.443869545817043e-05,
"loss": 0.0222,
"step": 9700
},
{
"grad_norm": 0.24151715636253357,
"learning_rate": 9.442352851846929e-05,
"loss": 0.0225,
"step": 9710
},
{
"grad_norm": 0.2635901868343353,
"learning_rate": 9.440834214640755e-05,
"loss": 0.0212,
"step": 9720
},
{
"grad_norm": 0.24533984065055847,
"learning_rate": 9.439313634862823e-05,
"loss": 0.0215,
"step": 9730
},
{
"grad_norm": 0.17376433312892914,
"learning_rate": 9.437791113178282e-05,
"loss": 0.0218,
"step": 9740
},
{
"grad_norm": 0.2715037167072296,
"learning_rate": 9.43626665025314e-05,
"loss": 0.0234,
"step": 9750
},
{
"grad_norm": 0.30662694573402405,
"learning_rate": 9.434740246754248e-05,
"loss": 0.0205,
"step": 9760
},
{
"grad_norm": 0.27739012241363525,
"learning_rate": 9.433211903349304e-05,
"loss": 0.0186,
"step": 9770
},
{
"grad_norm": 0.23447513580322266,
"learning_rate": 9.431681620706858e-05,
"loss": 0.0179,
"step": 9780
},
{
"grad_norm": 0.30972781777381897,
"learning_rate": 9.43014939949631e-05,
"loss": 0.0198,
"step": 9790
},
{
"grad_norm": 0.23031923174858093,
"learning_rate": 9.428615240387904e-05,
"loss": 0.0194,
"step": 9800
},
{
"grad_norm": 0.29500728845596313,
"learning_rate": 9.427079144052732e-05,
"loss": 0.0272,
"step": 9810
},
{
"grad_norm": 0.3305860161781311,
"learning_rate": 9.425541111162739e-05,
"loss": 0.0229,
"step": 9820
},
{
"grad_norm": 0.2628893256187439,
"learning_rate": 9.424001142390709e-05,
"loss": 0.0248,
"step": 9830
},
{
"grad_norm": 0.1825234591960907,
"learning_rate": 9.422459238410277e-05,
"loss": 0.0226,
"step": 9840
},
{
"grad_norm": 0.27743107080459595,
"learning_rate": 9.420915399895926e-05,
"loss": 0.0233,
"step": 9850
},
{
"grad_norm": 0.3016844093799591,
"learning_rate": 9.419369627522981e-05,
"loss": 0.0208,
"step": 9860
},
{
"grad_norm": 0.23935569822788239,
"learning_rate": 9.417821921967618e-05,
"loss": 0.0179,
"step": 9870
},
{
"grad_norm": 0.294528067111969,
"learning_rate": 9.416272283906855e-05,
"loss": 0.0191,
"step": 9880
},
{
"grad_norm": 0.3194102346897125,
"learning_rate": 9.414720714018554e-05,
"loss": 0.0211,
"step": 9890
},
{
"grad_norm": 0.24373705685138702,
"learning_rate": 9.413167212981427e-05,
"loss": 0.0202,
"step": 9900
},
{
"grad_norm": 0.19963878393173218,
"learning_rate": 9.41161178147503e-05,
"loss": 0.0205,
"step": 9910
},
{
"grad_norm": 0.25904932618141174,
"learning_rate": 9.410054420179755e-05,
"loss": 0.0185,
"step": 9920
},
{
"grad_norm": 0.25355279445648193,
"learning_rate": 9.408495129776851e-05,
"loss": 0.0189,
"step": 9930
},
{
"grad_norm": 0.2866595983505249,
"learning_rate": 9.406933910948403e-05,
"loss": 0.0171,
"step": 9940
},
{
"grad_norm": 0.2677626609802246,
"learning_rate": 9.40537076437734e-05,
"loss": 0.0249,
"step": 9950
},
{
"grad_norm": 0.24794548749923706,
"learning_rate": 9.403805690747436e-05,
"loss": 0.0196,
"step": 9960
},
{
"grad_norm": 0.2662081718444824,
"learning_rate": 9.402238690743308e-05,
"loss": 0.0188,
"step": 9970
},
{
"grad_norm": 0.2326134592294693,
"learning_rate": 9.400669765050413e-05,
"loss": 0.0159,
"step": 9980
},
{
"grad_norm": 0.2483377754688263,
"learning_rate": 9.399098914355055e-05,
"loss": 0.0181,
"step": 9990
},
{
"grad_norm": 0.28269460797309875,
"learning_rate": 9.397526139344378e-05,
"loss": 0.0201,
"step": 10000
},
{
"grad_norm": 0.21608909964561462,
"learning_rate": 9.395951440706362e-05,
"loss": 0.0169,
"step": 10010
},
{
"grad_norm": 0.2812051475048065,
"learning_rate": 9.394374819129839e-05,
"loss": 0.0205,
"step": 10020
},
{
"grad_norm": 0.25992244482040405,
"learning_rate": 9.392796275304474e-05,
"loss": 0.0217,
"step": 10030
},
{
"grad_norm": 0.2358676940202713,
"learning_rate": 9.391215809920778e-05,
"loss": 0.0174,
"step": 10040
},
{
"grad_norm": 0.25701212882995605,
"learning_rate": 9.389633423670099e-05,
"loss": 0.019,
"step": 10050
},
{
"grad_norm": 0.16488133370876312,
"learning_rate": 9.388049117244626e-05,
"loss": 0.0177,
"step": 10060
},
{
"grad_norm": 0.2120034098625183,
"learning_rate": 9.386462891337389e-05,
"loss": 0.0175,
"step": 10070
},
{
"grad_norm": 0.22943192720413208,
"learning_rate": 9.384874746642257e-05,
"loss": 0.0258,
"step": 10080
},
{
"grad_norm": 0.3567817211151123,
"learning_rate": 9.383284683853937e-05,
"loss": 0.0184,
"step": 10090
},
{
"grad_norm": 0.19279062747955322,
"learning_rate": 9.381692703667981e-05,
"loss": 0.0177,
"step": 10100
},
{
"grad_norm": 0.21944560110569,
"learning_rate": 9.380098806780771e-05,
"loss": 0.0161,
"step": 10110
},
{
"grad_norm": 0.26716071367263794,
"learning_rate": 9.378502993889533e-05,
"loss": 0.019,
"step": 10120
},
{
"grad_norm": 0.23674671351909637,
"learning_rate": 9.376905265692329e-05,
"loss": 0.0217,
"step": 10130
},
{
"grad_norm": 0.21083150804042816,
"learning_rate": 9.37530562288806e-05,
"loss": 0.0182,
"step": 10140
},
{
"grad_norm": 0.28007814288139343,
"learning_rate": 9.373704066176465e-05,
"loss": 0.0179,
"step": 10150
},
{
"grad_norm": 0.25921449065208435,
"learning_rate": 9.372100596258118e-05,
"loss": 0.0201,
"step": 10160
},
{
"grad_norm": 0.1855979561805725,
"learning_rate": 9.370495213834433e-05,
"loss": 0.0204,
"step": 10170
},
{
"grad_norm": 0.2711697816848755,
"learning_rate": 9.368887919607657e-05,
"loss": 0.0207,
"step": 10180
},
{
"grad_norm": 0.2301127165555954,
"learning_rate": 9.367278714280876e-05,
"loss": 0.0195,
"step": 10190
},
{
"grad_norm": 0.17540240287780762,
"learning_rate": 9.36566759855801e-05,
"loss": 0.0192,
"step": 10200
},
{
"grad_norm": 0.25268253684043884,
"learning_rate": 9.36405457314382e-05,
"loss": 0.0215,
"step": 10210
},
{
"grad_norm": 0.24855005741119385,
"learning_rate": 9.36243963874389e-05,
"loss": 0.0234,
"step": 10220
},
{
"grad_norm": 0.2017088681459427,
"learning_rate": 9.360822796064655e-05,
"loss": 0.0193,
"step": 10230
},
{
"grad_norm": 0.2046915888786316,
"learning_rate": 9.359204045813372e-05,
"loss": 0.0179,
"step": 10240
},
{
"grad_norm": 0.21998535096645355,
"learning_rate": 9.357583388698141e-05,
"loss": 0.0166,
"step": 10250
},
{
"grad_norm": 0.2539973855018616,
"learning_rate": 9.35596082542789e-05,
"loss": 0.0219,
"step": 10260
},
{
"grad_norm": 0.18894517421722412,
"learning_rate": 9.354336356712383e-05,
"loss": 0.0212,
"step": 10270
},
{
"grad_norm": 0.25493210554122925,
"learning_rate": 9.35270998326222e-05,
"loss": 0.0206,
"step": 10280
},
{
"grad_norm": 0.24909581243991852,
"learning_rate": 9.351081705788831e-05,
"loss": 0.0169,
"step": 10290
},
{
"grad_norm": 0.22944337129592896,
"learning_rate": 9.349451525004477e-05,
"loss": 0.0191,
"step": 10300
},
{
"grad_norm": 0.20833955705165863,
"learning_rate": 9.347819441622261e-05,
"loss": 0.0185,
"step": 10310
},
{
"grad_norm": 0.3538637161254883,
"learning_rate": 9.346185456356105e-05,
"loss": 0.0172,
"step": 10320
},
{
"grad_norm": 0.26469504833221436,
"learning_rate": 9.344549569920774e-05,
"loss": 0.0206,
"step": 10330
},
{
"grad_norm": 0.2403581738471985,
"learning_rate": 9.342911783031858e-05,
"loss": 0.0185,
"step": 10340
},
{
"grad_norm": 0.2301088571548462,
"learning_rate": 9.341272096405782e-05,
"loss": 0.0238,
"step": 10350
},
{
"grad_norm": 0.2804258167743683,
"learning_rate": 9.3396305107598e-05,
"loss": 0.0212,
"step": 10360
},
{
"grad_norm": 0.21539585292339325,
"learning_rate": 9.337987026811998e-05,
"loss": 0.0188,
"step": 10370
},
{
"grad_norm": 0.22304101288318634,
"learning_rate": 9.33634164528129e-05,
"loss": 0.0175,
"step": 10380
},
{
"grad_norm": 0.28763920068740845,
"learning_rate": 9.334694366887424e-05,
"loss": 0.0198,
"step": 10390
},
{
"grad_norm": 0.2419157475233078,
"learning_rate": 9.333045192350973e-05,
"loss": 0.0204,
"step": 10400
},
{
"grad_norm": 0.1978996992111206,
"learning_rate": 9.331394122393345e-05,
"loss": 0.0184,
"step": 10410
},
{
"grad_norm": 0.24739816784858704,
"learning_rate": 9.329741157736771e-05,
"loss": 0.0152,
"step": 10420
},
{
"grad_norm": 0.23371540009975433,
"learning_rate": 9.328086299104317e-05,
"loss": 0.02,
"step": 10430
},
{
"grad_norm": 0.2428014874458313,
"learning_rate": 9.326429547219872e-05,
"loss": 0.0192,
"step": 10440
},
{
"grad_norm": 0.3076070249080658,
"learning_rate": 9.324770902808155e-05,
"loss": 0.0181,
"step": 10450
},
{
"grad_norm": 0.24483181536197662,
"learning_rate": 9.323110366594717e-05,
"loss": 0.0189,
"step": 10460
},
{
"grad_norm": 0.19933010637760162,
"learning_rate": 9.32144793930593e-05,
"loss": 0.0182,
"step": 10470
},
{
"grad_norm": 0.2416263222694397,
"learning_rate": 9.319783621668996e-05,
"loss": 0.0177,
"step": 10480
},
{
"grad_norm": 0.24554041028022766,
"learning_rate": 9.318117414411947e-05,
"loss": 0.0176,
"step": 10490
},
{
"grad_norm": 0.26497769355773926,
"learning_rate": 9.316449318263635e-05,
"loss": 0.0178,
"step": 10500
},
{
"grad_norm": 0.22803758084774017,
"learning_rate": 9.314779333953744e-05,
"loss": 0.0202,
"step": 10510
},
{
"grad_norm": 0.2899167239665985,
"learning_rate": 9.313107462212781e-05,
"loss": 0.021,
"step": 10520
},
{
"grad_norm": 0.2137484848499298,
"learning_rate": 9.311433703772082e-05,
"loss": 0.0203,
"step": 10530
},
{
"grad_norm": 0.23829394578933716,
"learning_rate": 9.3097580593638e-05,
"loss": 0.0191,
"step": 10540
},
{
"grad_norm": 0.22926506400108337,
"learning_rate": 9.308080529720926e-05,
"loss": 0.0209,
"step": 10550
},
{
"grad_norm": 0.2344760000705719,
"learning_rate": 9.306401115577264e-05,
"loss": 0.018,
"step": 10560
},
{
"grad_norm": 0.28586333990097046,
"learning_rate": 9.304719817667447e-05,
"loss": 0.0207,
"step": 10570
},
{
"grad_norm": 0.1966448724269867,
"learning_rate": 9.303036636726934e-05,
"loss": 0.0173,
"step": 10580
},
{
"grad_norm": 0.2780885696411133,
"learning_rate": 9.301351573492003e-05,
"loss": 0.023,
"step": 10590
},
{
"grad_norm": 0.20525626838207245,
"learning_rate": 9.299664628699758e-05,
"loss": 0.0209,
"step": 10600
},
{
"grad_norm": 0.27389705181121826,
"learning_rate": 9.297975803088129e-05,
"loss": 0.0229,
"step": 10610
},
{
"grad_norm": 0.23744262754917145,
"learning_rate": 9.296285097395864e-05,
"loss": 0.0241,
"step": 10620
},
{
"grad_norm": 0.3115321397781372,
"learning_rate": 9.294592512362533e-05,
"loss": 0.0179,
"step": 10630
},
{
"grad_norm": 0.2978227138519287,
"learning_rate": 9.292898048728533e-05,
"loss": 0.0235,
"step": 10640
},
{
"grad_norm": 0.17475973069667816,
"learning_rate": 9.29120170723508e-05,
"loss": 0.0194,
"step": 10650
},
{
"grad_norm": 0.28474026918411255,
"learning_rate": 9.28950348862421e-05,
"loss": 0.0203,
"step": 10660
},
{
"grad_norm": 0.25000232458114624,
"learning_rate": 9.287803393638781e-05,
"loss": 0.0181,
"step": 10670
},
{
"grad_norm": 0.23203131556510925,
"learning_rate": 9.286101423022474e-05,
"loss": 0.0161,
"step": 10680
},
{
"grad_norm": 0.2135542780160904,
"learning_rate": 9.284397577519788e-05,
"loss": 0.0173,
"step": 10690
},
{
"grad_norm": 0.14061014354228973,
"learning_rate": 9.282691857876043e-05,
"loss": 0.0189,
"step": 10700
},
{
"grad_norm": 0.24653032422065735,
"learning_rate": 9.280984264837377e-05,
"loss": 0.0171,
"step": 10710
},
{
"grad_norm": 0.32819023728370667,
"learning_rate": 9.279274799150752e-05,
"loss": 0.0209,
"step": 10720
},
{
"grad_norm": 0.21283037960529327,
"learning_rate": 9.277563461563945e-05,
"loss": 0.0205,
"step": 10730
},
{
"grad_norm": 0.2716512978076935,
"learning_rate": 9.275850252825555e-05,
"loss": 0.0184,
"step": 10740
},
{
"grad_norm": 0.2231690138578415,
"learning_rate": 9.274135173684994e-05,
"loss": 0.017,
"step": 10750
},
{
"grad_norm": 0.2893235385417938,
"learning_rate": 9.272418224892498e-05,
"loss": 0.0183,
"step": 10760
},
{
"grad_norm": 0.2521395981311798,
"learning_rate": 9.27069940719912e-05,
"loss": 0.0196,
"step": 10770
},
{
"grad_norm": 0.27969545125961304,
"learning_rate": 9.268978721356727e-05,
"loss": 0.0183,
"step": 10780
},
{
"grad_norm": 0.257140576839447,
"learning_rate": 9.267256168118008e-05,
"loss": 0.0223,
"step": 10790
},
{
"grad_norm": 0.24515564739704132,
"learning_rate": 9.265531748236463e-05,
"loss": 0.0202,
"step": 10800
},
{
"grad_norm": 0.22425413131713867,
"learning_rate": 9.263805462466416e-05,
"loss": 0.0153,
"step": 10810
},
{
"grad_norm": 0.25628969073295593,
"learning_rate": 9.262077311562998e-05,
"loss": 0.0205,
"step": 10820
},
{
"grad_norm": 0.2316131889820099,
"learning_rate": 9.260347296282165e-05,
"loss": 0.021,
"step": 10830
},
{
"grad_norm": 0.2972199618816376,
"learning_rate": 9.258615417380683e-05,
"loss": 0.0197,
"step": 10840
},
{
"grad_norm": 0.25416114926338196,
"learning_rate": 9.256881675616133e-05,
"loss": 0.0165,
"step": 10850
},
{
"grad_norm": 0.23889869451522827,
"learning_rate": 9.255146071746917e-05,
"loss": 0.0208,
"step": 10860
},
{
"grad_norm": 0.19452854990959167,
"learning_rate": 9.253408606532241e-05,
"loss": 0.019,
"step": 10870
},
{
"grad_norm": 0.19842718541622162,
"learning_rate": 9.251669280732137e-05,
"loss": 0.0179,
"step": 10880
},
{
"grad_norm": 0.5598425269126892,
"learning_rate": 9.249928095107441e-05,
"loss": 0.0193,
"step": 10890
},
{
"grad_norm": 0.2181270718574524,
"learning_rate": 9.248185050419811e-05,
"loss": 0.0162,
"step": 10900
},
{
"grad_norm": 0.20888836681842804,
"learning_rate": 9.24644014743171e-05,
"loss": 0.0194,
"step": 10910
},
{
"grad_norm": 0.16374655067920685,
"learning_rate": 9.24469338690642e-05,
"loss": 0.019,
"step": 10920
},
{
"grad_norm": 0.24447907507419586,
"learning_rate": 9.242944769608033e-05,
"loss": 0.0177,
"step": 10930
},
{
"grad_norm": 0.21491360664367676,
"learning_rate": 9.241194296301454e-05,
"loss": 0.0179,
"step": 10940
},
{
"grad_norm": 0.13870009779930115,
"learning_rate": 9.239441967752397e-05,
"loss": 0.0152,
"step": 10950
},
{
"grad_norm": 0.22660881280899048,
"learning_rate": 9.237687784727393e-05,
"loss": 0.0148,
"step": 10960
},
{
"grad_norm": 0.21305285394191742,
"learning_rate": 9.235931747993781e-05,
"loss": 0.0148,
"step": 10970
},
{
"grad_norm": 0.1817772090435028,
"learning_rate": 9.234173858319707e-05,
"loss": 0.0185,
"step": 10980
},
{
"grad_norm": 0.2243172824382782,
"learning_rate": 9.23241411647414e-05,
"loss": 0.0196,
"step": 10990
},
{
"grad_norm": 0.24859297275543213,
"learning_rate": 9.230652523226841e-05,
"loss": 0.0187,
"step": 11000
},
{
"grad_norm": 0.26410484313964844,
"learning_rate": 9.2288890793484e-05,
"loss": 0.0173,
"step": 11010
},
{
"grad_norm": 0.2054034322500229,
"learning_rate": 9.227123785610199e-05,
"loss": 0.0175,
"step": 11020
},
{
"grad_norm": 0.21280667185783386,
"learning_rate": 9.225356642784443e-05,
"loss": 0.0191,
"step": 11030
},
{
"grad_norm": 0.19988057017326355,
"learning_rate": 9.223587651644139e-05,
"loss": 0.0181,
"step": 11040
},
{
"grad_norm": 0.26031792163848877,
"learning_rate": 9.221816812963104e-05,
"loss": 0.0205,
"step": 11050
},
{
"grad_norm": 0.2584247887134552,
"learning_rate": 9.22004412751596e-05,
"loss": 0.0179,
"step": 11060
},
{
"grad_norm": 0.26613327860832214,
"learning_rate": 9.218269596078146e-05,
"loss": 0.0187,
"step": 11070
},
{
"grad_norm": 0.23883669078350067,
"learning_rate": 9.216493219425895e-05,
"loss": 0.02,
"step": 11080
},
{
"grad_norm": 0.3305536210536957,
"learning_rate": 9.21471499833626e-05,
"loss": 0.0173,
"step": 11090
},
{
"grad_norm": 0.23506855964660645,
"learning_rate": 9.212934933587094e-05,
"loss": 0.0185,
"step": 11100
},
{
"grad_norm": 0.25789836049079895,
"learning_rate": 9.211153025957056e-05,
"loss": 0.0179,
"step": 11110
},
{
"grad_norm": 0.2348192185163498,
"learning_rate": 9.209369276225614e-05,
"loss": 0.0172,
"step": 11120
},
{
"grad_norm": 0.20443031191825867,
"learning_rate": 9.207583685173042e-05,
"loss": 0.0186,
"step": 11130
},
{
"grad_norm": 0.23772911727428436,
"learning_rate": 9.205796253580417e-05,
"loss": 0.0197,
"step": 11140
},
{
"grad_norm": 0.18360383808612823,
"learning_rate": 9.204006982229621e-05,
"loss": 0.0192,
"step": 11150
},
{
"grad_norm": 0.23849916458129883,
"learning_rate": 9.202215871903346e-05,
"loss": 0.0147,
"step": 11160
},
{
"grad_norm": 0.16656629741191864,
"learning_rate": 9.20042292338508e-05,
"loss": 0.0166,
"step": 11170
},
{
"grad_norm": 0.1823406219482422,
"learning_rate": 9.198628137459123e-05,
"loss": 0.0194,
"step": 11180
},
{
"grad_norm": 0.2135552167892456,
"learning_rate": 9.196831514910572e-05,
"loss": 0.0167,
"step": 11190
},
{
"grad_norm": 0.18353556096553802,
"learning_rate": 9.195033056525332e-05,
"loss": 0.0176,
"step": 11200
},
{
"grad_norm": 0.2681991159915924,
"learning_rate": 9.193232763090114e-05,
"loss": 0.021,
"step": 11210
},
{
"grad_norm": 0.2665809392929077,
"learning_rate": 9.191430635392422e-05,
"loss": 0.0164,
"step": 11220
},
{
"grad_norm": 0.1974109262228012,
"learning_rate": 9.18962667422057e-05,
"loss": 0.0167,
"step": 11230
},
{
"grad_norm": 0.1896234005689621,
"learning_rate": 9.187820880363671e-05,
"loss": 0.0168,
"step": 11240
},
{
"grad_norm": 0.22469575703144073,
"learning_rate": 9.186013254611643e-05,
"loss": 0.0177,
"step": 11250
},
{
"grad_norm": 0.1996164321899414,
"learning_rate": 9.1842037977552e-05,
"loss": 0.019,
"step": 11260
},
{
"grad_norm": 0.2855274975299835,
"learning_rate": 9.182392510585862e-05,
"loss": 0.0234,
"step": 11270
},
{
"grad_norm": 0.1962725669145584,
"learning_rate": 9.180579393895946e-05,
"loss": 0.0167,
"step": 11280
},
{
"grad_norm": 0.1875135451555252,
"learning_rate": 9.178764448478572e-05,
"loss": 0.0158,
"step": 11290
},
{
"grad_norm": 0.2569526433944702,
"learning_rate": 9.176947675127658e-05,
"loss": 0.017,
"step": 11300
},
{
"grad_norm": 0.22981354594230652,
"learning_rate": 9.175129074637924e-05,
"loss": 0.0207,
"step": 11310
},
{
"grad_norm": 0.19379164278507233,
"learning_rate": 9.173308647804884e-05,
"loss": 0.0149,
"step": 11320
},
{
"grad_norm": 0.23140884935855865,
"learning_rate": 9.171486395424859e-05,
"loss": 0.0184,
"step": 11330
},
{
"grad_norm": 0.245038703083992,
"learning_rate": 9.16966231829496e-05,
"loss": 0.0198,
"step": 11340
},
{
"grad_norm": 0.24129366874694824,
"learning_rate": 9.167836417213105e-05,
"loss": 0.0195,
"step": 11350
},
{
"grad_norm": 0.1795470118522644,
"learning_rate": 9.166008692978001e-05,
"loss": 0.0158,
"step": 11360
},
{
"grad_norm": 0.28362950682640076,
"learning_rate": 9.164179146389158e-05,
"loss": 0.0179,
"step": 11370
},
{
"grad_norm": 0.25175175070762634,
"learning_rate": 9.162347778246882e-05,
"loss": 0.0204,
"step": 11380
},
{
"grad_norm": 0.27670517563819885,
"learning_rate": 9.160514589352276e-05,
"loss": 0.0185,
"step": 11390
},
{
"grad_norm": 0.23314358294010162,
"learning_rate": 9.15867958050724e-05,
"loss": 0.0181,
"step": 11400
},
{
"grad_norm": 0.20065483450889587,
"learning_rate": 9.156842752514466e-05,
"loss": 0.0198,
"step": 11410
},
{
"grad_norm": 0.21583005785942078,
"learning_rate": 9.155004106177447e-05,
"loss": 0.0215,
"step": 11420
},
{
"grad_norm": 0.26928848028182983,
"learning_rate": 9.153163642300471e-05,
"loss": 0.0205,
"step": 11430
},
{
"grad_norm": 0.26945260167121887,
"learning_rate": 9.151321361688616e-05,
"loss": 0.0189,
"step": 11440
},
{
"grad_norm": 0.25080248713493347,
"learning_rate": 9.149477265147762e-05,
"loss": 0.0163,
"step": 11450
},
{
"grad_norm": 0.19466321170330048,
"learning_rate": 9.147631353484574e-05,
"loss": 0.0182,
"step": 11460
},
{
"grad_norm": 0.23434223234653473,
"learning_rate": 9.145783627506522e-05,
"loss": 0.0192,
"step": 11470
},
{
"grad_norm": 0.21764545142650604,
"learning_rate": 9.143934088021861e-05,
"loss": 0.0178,
"step": 11480
},
{
"grad_norm": 0.16071482002735138,
"learning_rate": 9.142082735839645e-05,
"loss": 0.0159,
"step": 11490
},
{
"grad_norm": 0.134405255317688,
"learning_rate": 9.140229571769715e-05,
"loss": 0.0199,
"step": 11500
},
{
"grad_norm": 0.22431392967700958,
"learning_rate": 9.138374596622709e-05,
"loss": 0.0176,
"step": 11510
},
{
"grad_norm": 0.34809592366218567,
"learning_rate": 9.136517811210059e-05,
"loss": 0.0183,
"step": 11520
},
{
"grad_norm": 0.2679460048675537,
"learning_rate": 9.134659216343984e-05,
"loss": 0.0182,
"step": 11530
},
{
"grad_norm": 0.28422755002975464,
"learning_rate": 9.132798812837494e-05,
"loss": 0.02,
"step": 11540
},
{
"grad_norm": 0.2886165976524353,
"learning_rate": 9.130936601504396e-05,
"loss": 0.0157,
"step": 11550
},
{
"grad_norm": 0.22985675930976868,
"learning_rate": 9.129072583159284e-05,
"loss": 0.0205,
"step": 11560
},
{
"grad_norm": 0.2613084018230438,
"learning_rate": 9.127206758617542e-05,
"loss": 0.0192,
"step": 11570
},
{
"grad_norm": 0.21359777450561523,
"learning_rate": 9.125339128695346e-05,
"loss": 0.0164,
"step": 11580
},
{
"grad_norm": 0.18640129268169403,
"learning_rate": 9.123469694209659e-05,
"loss": 0.0154,
"step": 11590
},
{
"grad_norm": 0.255043625831604,
"learning_rate": 9.121598455978239e-05,
"loss": 0.0152,
"step": 11600
},
{
"grad_norm": 0.24625079333782196,
"learning_rate": 9.119725414819624e-05,
"loss": 0.0182,
"step": 11610
},
{
"grad_norm": 0.21849319338798523,
"learning_rate": 9.117850571553149e-05,
"loss": 0.02,
"step": 11620
},
{
"grad_norm": 0.24009265005588531,
"learning_rate": 9.115973926998935e-05,
"loss": 0.0153,
"step": 11630
},
{
"grad_norm": 0.20689012110233307,
"learning_rate": 9.114095481977888e-05,
"loss": 0.0164,
"step": 11640
},
{
"grad_norm": 0.23804868757724762,
"learning_rate": 9.112215237311703e-05,
"loss": 0.0168,
"step": 11650
},
{
"grad_norm": 0.253255695104599,
"learning_rate": 9.110333193822867e-05,
"loss": 0.0158,
"step": 11660
},
{
"grad_norm": 0.2433309108018875,
"learning_rate": 9.108449352334645e-05,
"loss": 0.0138,
"step": 11670
},
{
"grad_norm": 0.19066566228866577,
"learning_rate": 9.106563713671094e-05,
"loss": 0.0154,
"step": 11680
},
{
"grad_norm": 0.2008248269557953,
"learning_rate": 9.104676278657061e-05,
"loss": 0.0204,
"step": 11690
},
{
"grad_norm": 0.22830387949943542,
"learning_rate": 9.102787048118169e-05,
"loss": 0.0191,
"step": 11700
},
{
"grad_norm": 0.25851309299468994,
"learning_rate": 9.100896022880834e-05,
"loss": 0.0175,
"step": 11710
},
{
"grad_norm": 0.25062787532806396,
"learning_rate": 9.099003203772254e-05,
"loss": 0.0185,
"step": 11720
},
{
"grad_norm": 0.19001345336437225,
"learning_rate": 9.097108591620413e-05,
"loss": 0.0146,
"step": 11730
},
{
"grad_norm": 0.14623180031776428,
"learning_rate": 9.095212187254078e-05,
"loss": 0.0149,
"step": 11740
},
{
"grad_norm": 0.2551042139530182,
"learning_rate": 9.093313991502801e-05,
"loss": 0.0158,
"step": 11750
},
{
"grad_norm": 0.245066300034523,
"learning_rate": 9.091414005196917e-05,
"loss": 0.0175,
"step": 11760
},
{
"grad_norm": 0.3145027458667755,
"learning_rate": 9.089512229167545e-05,
"loss": 0.0162,
"step": 11770
},
{
"grad_norm": 0.2046465128660202,
"learning_rate": 9.087608664246587e-05,
"loss": 0.0209,
"step": 11780
},
{
"grad_norm": 0.1765555590391159,
"learning_rate": 9.085703311266727e-05,
"loss": 0.0151,
"step": 11790
},
{
"grad_norm": 0.2854391038417816,
"learning_rate": 9.083796171061429e-05,
"loss": 0.0216,
"step": 11800
},
{
"grad_norm": 0.3248160183429718,
"learning_rate": 9.081887244464941e-05,
"loss": 0.0185,
"step": 11810
},
{
"grad_norm": 0.24133481085300446,
"learning_rate": 9.079976532312297e-05,
"loss": 0.0217,
"step": 11820
},
{
"grad_norm": 0.20805026590824127,
"learning_rate": 9.078064035439301e-05,
"loss": 0.0178,
"step": 11830
},
{
"grad_norm": 0.25035861134529114,
"learning_rate": 9.07614975468255e-05,
"loss": 0.0174,
"step": 11840
},
{
"grad_norm": 0.24027961492538452,
"learning_rate": 9.074233690879412e-05,
"loss": 0.0185,
"step": 11850
},
{
"grad_norm": 0.22835983335971832,
"learning_rate": 9.072315844868038e-05,
"loss": 0.0189,
"step": 11860
},
{
"grad_norm": 0.21079373359680176,
"learning_rate": 9.07039621748736e-05,
"loss": 0.0198,
"step": 11870
},
{
"grad_norm": 0.24112087488174438,
"learning_rate": 9.06847480957709e-05,
"loss": 0.0173,
"step": 11880
},
{
"grad_norm": 0.17388151586055756,
"learning_rate": 9.066551621977713e-05,
"loss": 0.0177,
"step": 11890
},
{
"grad_norm": 0.3177785873413086,
"learning_rate": 9.064626655530501e-05,
"loss": 0.0266,
"step": 11900
},
{
"grad_norm": 0.2170921266078949,
"learning_rate": 9.062699911077497e-05,
"loss": 0.0173,
"step": 11910
},
{
"grad_norm": 0.314103364944458,
"learning_rate": 9.060771389461524e-05,
"loss": 0.0182,
"step": 11920
},
{
"grad_norm": 0.2900501787662506,
"learning_rate": 9.058841091526187e-05,
"loss": 0.0175,
"step": 11930
},
{
"grad_norm": 0.25115472078323364,
"learning_rate": 9.056909018115858e-05,
"loss": 0.0163,
"step": 11940
},
{
"grad_norm": 0.2428686022758484,
"learning_rate": 9.054975170075697e-05,
"loss": 0.0201,
"step": 11950
},
{
"grad_norm": 0.22682823240756989,
"learning_rate": 9.053039548251631e-05,
"loss": 0.0197,
"step": 11960
},
{
"grad_norm": 0.1916792243719101,
"learning_rate": 9.051102153490368e-05,
"loss": 0.016,
"step": 11970
},
{
"grad_norm": 0.1472385972738266,
"learning_rate": 9.04916298663939e-05,
"loss": 0.0205,
"step": 11980
},
{
"grad_norm": 0.28521600365638733,
"learning_rate": 9.047222048546955e-05,
"loss": 0.0173,
"step": 11990
},
{
"grad_norm": 0.21148289740085602,
"learning_rate": 9.045279340062097e-05,
"loss": 0.0202,
"step": 12000
},
{
"grad_norm": 0.2704308331012726,
"learning_rate": 9.043334862034618e-05,
"loss": 0.0214,
"step": 12010
},
{
"grad_norm": 0.17110873758792877,
"learning_rate": 9.041388615315102e-05,
"loss": 0.0185,
"step": 12020
},
{
"grad_norm": 0.2233845591545105,
"learning_rate": 9.039440600754905e-05,
"loss": 0.0189,
"step": 12030
},
{
"grad_norm": 0.306430459022522,
"learning_rate": 9.037490819206151e-05,
"loss": 0.0201,
"step": 12040
},
{
"grad_norm": 0.229330375790596,
"learning_rate": 9.035539271521744e-05,
"loss": 0.017,
"step": 12050
},
{
"grad_norm": 0.19436146318912506,
"learning_rate": 9.033585958555356e-05,
"loss": 0.016,
"step": 12060
},
{
"grad_norm": 0.19887708127498627,
"learning_rate": 9.031630881161431e-05,
"loss": 0.0174,
"step": 12070
},
{
"grad_norm": 0.17754781246185303,
"learning_rate": 9.029674040195186e-05,
"loss": 0.0172,
"step": 12080
},
{
"grad_norm": 0.2013232558965683,
"learning_rate": 9.027715436512613e-05,
"loss": 0.0162,
"step": 12090
},
{
"grad_norm": 0.2513226866722107,
"learning_rate": 9.02575507097047e-05,
"loss": 0.0177,
"step": 12100
},
{
"grad_norm": 0.20022538304328918,
"learning_rate": 9.023792944426286e-05,
"loss": 0.0165,
"step": 12110
},
{
"grad_norm": 0.22361764311790466,
"learning_rate": 9.021829057738364e-05,
"loss": 0.0192,
"step": 12120
},
{
"grad_norm": 0.22778363525867462,
"learning_rate": 9.019863411765775e-05,
"loss": 0.0164,
"step": 12130
},
{
"grad_norm": 0.21175678074359894,
"learning_rate": 9.017896007368357e-05,
"loss": 0.0143,
"step": 12140
},
{
"grad_norm": 0.27405425906181335,
"learning_rate": 9.015926845406722e-05,
"loss": 0.0175,
"step": 12150
},
{
"grad_norm": 0.24207757413387299,
"learning_rate": 9.013955926742245e-05,
"loss": 0.0189,
"step": 12160
},
{
"grad_norm": 0.2595878541469574,
"learning_rate": 9.011983252237077e-05,
"loss": 0.0192,
"step": 12170
},
{
"grad_norm": 0.1792474389076233,
"learning_rate": 9.01000882275413e-05,
"loss": 0.0158,
"step": 12180
},
{
"grad_norm": 0.21619385480880737,
"learning_rate": 9.008032639157088e-05,
"loss": 0.0163,
"step": 12190
},
{
"grad_norm": 0.255868136882782,
"learning_rate": 9.006054702310401e-05,
"loss": 0.0192,
"step": 12200
},
{
"grad_norm": 0.19579842686653137,
"learning_rate": 9.004075013079283e-05,
"loss": 0.0154,
"step": 12210
},
{
"grad_norm": 0.2118227481842041,
"learning_rate": 9.00209357232972e-05,
"loss": 0.017,
"step": 12220
},
{
"grad_norm": 0.2707882523536682,
"learning_rate": 9.000110380928461e-05,
"loss": 0.0148,
"step": 12230
},
{
"grad_norm": 0.2244562953710556,
"learning_rate": 8.998125439743021e-05,
"loss": 0.016,
"step": 12240
},
{
"grad_norm": 0.2366371750831604,
"learning_rate": 8.996138749641682e-05,
"loss": 0.0183,
"step": 12250
},
{
"grad_norm": 0.20890265703201294,
"learning_rate": 8.994150311493488e-05,
"loss": 0.0171,
"step": 12260
},
{
"grad_norm": 0.21935853362083435,
"learning_rate": 8.992160126168247e-05,
"loss": 0.0198,
"step": 12270
},
{
"grad_norm": 0.1801203191280365,
"learning_rate": 8.99016819453654e-05,
"loss": 0.0174,
"step": 12280
},
{
"grad_norm": 0.21536654233932495,
"learning_rate": 8.988174517469702e-05,
"loss": 0.0165,
"step": 12290
},
{
"grad_norm": 0.24891389906406403,
"learning_rate": 8.986179095839835e-05,
"loss": 0.0182,
"step": 12300
},
{
"grad_norm": 0.3325229585170746,
"learning_rate": 8.984181930519804e-05,
"loss": 0.0195,
"step": 12310
},
{
"grad_norm": 0.23842647671699524,
"learning_rate": 8.982183022383237e-05,
"loss": 0.017,
"step": 12320
},
{
"grad_norm": 0.1979597508907318,
"learning_rate": 8.980182372304525e-05,
"loss": 0.0181,
"step": 12330
},
{
"grad_norm": 0.2200346738100052,
"learning_rate": 8.97817998115882e-05,
"loss": 0.0206,
"step": 12340
},
{
"grad_norm": 0.19707956910133362,
"learning_rate": 8.976175849822038e-05,
"loss": 0.0173,
"step": 12350
},
{
"grad_norm": 0.23864668607711792,
"learning_rate": 8.97416997917085e-05,
"loss": 0.0235,
"step": 12360
},
{
"grad_norm": 0.21148669719696045,
"learning_rate": 8.972162370082695e-05,
"loss": 0.0179,
"step": 12370
},
{
"grad_norm": 0.2258552759885788,
"learning_rate": 8.97015302343577e-05,
"loss": 0.0181,
"step": 12380
},
{
"grad_norm": 0.23386920988559723,
"learning_rate": 8.968141940109027e-05,
"loss": 0.0167,
"step": 12390
},
{
"grad_norm": 0.19266775250434875,
"learning_rate": 8.966129120982188e-05,
"loss": 0.0193,
"step": 12400
},
{
"grad_norm": 0.23731395602226257,
"learning_rate": 8.964114566935724e-05,
"loss": 0.0168,
"step": 12410
},
{
"grad_norm": 0.14120477437973022,
"learning_rate": 8.962098278850871e-05,
"loss": 0.0156,
"step": 12420
},
{
"grad_norm": 0.18466182053089142,
"learning_rate": 8.960080257609622e-05,
"loss": 0.0131,
"step": 12430
},
{
"grad_norm": 0.24158520996570587,
"learning_rate": 8.95806050409473e-05,
"loss": 0.0156,
"step": 12440
},
{
"grad_norm": 0.23820075392723083,
"learning_rate": 8.9560390191897e-05,
"loss": 0.0232,
"step": 12450
},
{
"grad_norm": 0.25189337134361267,
"learning_rate": 8.9540158037788e-05,
"loss": 0.0164,
"step": 12460
},
{
"grad_norm": 0.2547222673892975,
"learning_rate": 8.951990858747054e-05,
"loss": 0.0205,
"step": 12470
},
{
"grad_norm": 0.2187165915966034,
"learning_rate": 8.94996418498024e-05,
"loss": 0.0155,
"step": 12480
},
{
"grad_norm": 0.23191112279891968,
"learning_rate": 8.947935783364896e-05,
"loss": 0.0164,
"step": 12490
},
{
"grad_norm": 0.2584964632987976,
"learning_rate": 8.945905654788311e-05,
"loss": 0.0159,
"step": 12500
},
{
"grad_norm": 0.17940166592597961,
"learning_rate": 8.943873800138535e-05,
"loss": 0.0144,
"step": 12510
},
{
"grad_norm": 0.23682662844657898,
"learning_rate": 8.94184022030437e-05,
"loss": 0.0166,
"step": 12520
},
{
"grad_norm": 0.20393379032611847,
"learning_rate": 8.939804916175372e-05,
"loss": 0.0182,
"step": 12530
},
{
"grad_norm": 0.275746613740921,
"learning_rate": 8.93776788864185e-05,
"loss": 0.017,
"step": 12540
},
{
"grad_norm": 0.25151345133781433,
"learning_rate": 8.935729138594873e-05,
"loss": 0.0162,
"step": 12550
},
{
"grad_norm": 0.2015291005373001,
"learning_rate": 8.933688666926258e-05,
"loss": 0.0148,
"step": 12560
},
{
"grad_norm": 0.20738251507282257,
"learning_rate": 8.931646474528575e-05,
"loss": 0.0151,
"step": 12570
},
{
"grad_norm": 0.23546241223812103,
"learning_rate": 8.929602562295151e-05,
"loss": 0.0149,
"step": 12580
},
{
"grad_norm": 0.15644299983978271,
"learning_rate": 8.92755693112006e-05,
"loss": 0.015,
"step": 12590
},
{
"grad_norm": 0.2271387279033661,
"learning_rate": 8.925509581898136e-05,
"loss": 0.016,
"step": 12600
},
{
"grad_norm": 0.19471533596515656,
"learning_rate": 8.923460515524951e-05,
"loss": 0.0155,
"step": 12610
},
{
"grad_norm": 0.21621279418468475,
"learning_rate": 8.921409732896842e-05,
"loss": 0.018,
"step": 12620
},
{
"grad_norm": 0.18788500130176544,
"learning_rate": 8.919357234910887e-05,
"loss": 0.0149,
"step": 12630
},
{
"grad_norm": 0.25758692622184753,
"learning_rate": 8.917303022464923e-05,
"loss": 0.0156,
"step": 12640
},
{
"grad_norm": 0.24748460948467255,
"learning_rate": 8.915247096457531e-05,
"loss": 0.016,
"step": 12650
},
{
"grad_norm": 0.18178975582122803,
"learning_rate": 8.91318945778804e-05,
"loss": 0.0127,
"step": 12660
},
{
"grad_norm": 0.21342265605926514,
"learning_rate": 8.911130107356534e-05,
"loss": 0.0123,
"step": 12670
},
{
"grad_norm": 0.26807481050491333,
"learning_rate": 8.90906904606384e-05,
"loss": 0.0163,
"step": 12680
},
{
"grad_norm": 0.2555883824825287,
"learning_rate": 8.90700627481154e-05,
"loss": 0.017,
"step": 12690
},
{
"grad_norm": 0.22883762419223785,
"learning_rate": 8.904941794501957e-05,
"loss": 0.0145,
"step": 12700
},
{
"grad_norm": 0.18842226266860962,
"learning_rate": 8.902875606038166e-05,
"loss": 0.0151,
"step": 12710
},
{
"grad_norm": 0.23377849161624908,
"learning_rate": 8.900807710323989e-05,
"loss": 0.0165,
"step": 12720
},
{
"grad_norm": 0.23490281403064728,
"learning_rate": 8.898738108263993e-05,
"loss": 0.0153,
"step": 12730
},
{
"grad_norm": 0.21357516944408417,
"learning_rate": 8.896666800763491e-05,
"loss": 0.015,
"step": 12740
},
{
"grad_norm": 0.1871136575937271,
"learning_rate": 8.894593788728546e-05,
"loss": 0.0156,
"step": 12750
},
{
"grad_norm": 0.22777219116687775,
"learning_rate": 8.892519073065961e-05,
"loss": 0.0167,
"step": 12760
},
{
"grad_norm": 0.19241909682750702,
"learning_rate": 8.89044265468329e-05,
"loss": 0.0175,
"step": 12770
},
{
"grad_norm": 0.25077682733535767,
"learning_rate": 8.888364534488827e-05,
"loss": 0.019,
"step": 12780
},
{
"grad_norm": 0.20531544089317322,
"learning_rate": 8.886284713391613e-05,
"loss": 0.0161,
"step": 12790
},
{
"grad_norm": 0.2811039388179779,
"learning_rate": 8.884203192301431e-05,
"loss": 0.017,
"step": 12800
},
{
"grad_norm": 0.2918158173561096,
"learning_rate": 8.88211997212881e-05,
"loss": 0.0149,
"step": 12810
},
{
"grad_norm": 0.28741735219955444,
"learning_rate": 8.880035053785023e-05,
"loss": 0.0174,
"step": 12820
},
{
"grad_norm": 0.22104701399803162,
"learning_rate": 8.877948438182083e-05,
"loss": 0.0158,
"step": 12830
},
{
"grad_norm": 0.22504712641239166,
"learning_rate": 8.875860126232745e-05,
"loss": 0.0158,
"step": 12840
},
{
"grad_norm": 0.25644367933273315,
"learning_rate": 8.87377011885051e-05,
"loss": 0.016,
"step": 12850
},
{
"grad_norm": 0.2318536937236786,
"learning_rate": 8.871678416949617e-05,
"loss": 0.018,
"step": 12860
},
{
"grad_norm": 0.278922438621521,
"learning_rate": 8.869585021445046e-05,
"loss": 0.0156,
"step": 12870
},
{
"grad_norm": 0.17995630204677582,
"learning_rate": 8.867489933252521e-05,
"loss": 0.0147,
"step": 12880
},
{
"grad_norm": 0.21339863538742065,
"learning_rate": 8.865393153288504e-05,
"loss": 0.018,
"step": 12890
},
{
"grad_norm": 0.1941353678703308,
"learning_rate": 8.8632946824702e-05,
"loss": 0.0164,
"step": 12900
},
{
"grad_norm": 0.21743018925189972,
"learning_rate": 8.86119452171555e-05,
"loss": 0.0145,
"step": 12910
},
{
"grad_norm": 0.1750767081975937,
"learning_rate": 8.859092671943234e-05,
"loss": 0.016,
"step": 12920
},
{
"grad_norm": 0.21684449911117554,
"learning_rate": 8.856989134072676e-05,
"loss": 0.0157,
"step": 12930
},
{
"grad_norm": 0.2288552224636078,
"learning_rate": 8.85488390902403e-05,
"loss": 0.0175,
"step": 12940
},
{
"grad_norm": 0.19532908499240875,
"learning_rate": 8.852776997718199e-05,
"loss": 0.0153,
"step": 12950
},
{
"grad_norm": 0.23534490168094635,
"learning_rate": 8.850668401076812e-05,
"loss": 0.0169,
"step": 12960
},
{
"grad_norm": 0.19093267619609833,
"learning_rate": 8.848558120022246e-05,
"loss": 0.0144,
"step": 12970
},
{
"grad_norm": 0.1819402575492859,
"learning_rate": 8.846446155477603e-05,
"loss": 0.0167,
"step": 12980
},
{
"grad_norm": 0.268886536359787,
"learning_rate": 8.844332508366735e-05,
"loss": 0.0225,
"step": 12990
},
{
"grad_norm": 0.2621830999851227,
"learning_rate": 8.84221717961422e-05,
"loss": 0.0152,
"step": 13000
},
{
"grad_norm": 0.17516319453716278,
"learning_rate": 8.840100170145374e-05,
"loss": 0.0158,
"step": 13010
},
{
"grad_norm": 0.18356585502624512,
"learning_rate": 8.837981480886249e-05,
"loss": 0.0158,
"step": 13020
},
{
"grad_norm": 0.19135905802249908,
"learning_rate": 8.835861112763633e-05,
"loss": 0.0158,
"step": 13030
},
{
"grad_norm": 0.1786131113767624,
"learning_rate": 8.833739066705044e-05,
"loss": 0.0205,
"step": 13040
},
{
"grad_norm": 0.23772183060646057,
"learning_rate": 8.831615343638742e-05,
"loss": 0.0181,
"step": 13050
},
{
"grad_norm": 0.25261181592941284,
"learning_rate": 8.829489944493711e-05,
"loss": 0.0158,
"step": 13060
},
{
"grad_norm": 0.2413497418165207,
"learning_rate": 8.827362870199675e-05,
"loss": 0.0145,
"step": 13070
},
{
"grad_norm": 0.23130391538143158,
"learning_rate": 8.825234121687089e-05,
"loss": 0.0139,
"step": 13080
},
{
"grad_norm": 0.2268231213092804,
"learning_rate": 8.823103699887139e-05,
"loss": 0.0148,
"step": 13090
},
{
"grad_norm": 0.22658002376556396,
"learning_rate": 8.820971605731745e-05,
"loss": 0.0127,
"step": 13100
},
{
"grad_norm": 0.2857726514339447,
"learning_rate": 8.818837840153556e-05,
"loss": 0.0172,
"step": 13110
},
{
"grad_norm": 0.2606637179851532,
"learning_rate": 8.816702404085952e-05,
"loss": 0.0153,
"step": 13120
},
{
"grad_norm": 0.21300466358661652,
"learning_rate": 8.814565298463048e-05,
"loss": 0.018,
"step": 13130
},
{
"grad_norm": 0.20583316683769226,
"learning_rate": 8.812426524219688e-05,
"loss": 0.0178,
"step": 13140
},
{
"grad_norm": 0.21922001242637634,
"learning_rate": 8.81028608229144e-05,
"loss": 0.0195,
"step": 13150
},
{
"grad_norm": 0.253351092338562,
"learning_rate": 8.808143973614611e-05,
"loss": 0.0163,
"step": 13160
},
{
"grad_norm": 0.2331327199935913,
"learning_rate": 8.806000199126228e-05,
"loss": 0.0186,
"step": 13170
},
{
"grad_norm": 0.2720649540424347,
"learning_rate": 8.803854759764052e-05,
"loss": 0.0147,
"step": 13180
},
{
"grad_norm": 0.2434234917163849,
"learning_rate": 8.801707656466572e-05,
"loss": 0.0171,
"step": 13190
},
{
"grad_norm": 0.24383078515529633,
"learning_rate": 8.799558890173003e-05,
"loss": 0.0178,
"step": 13200
},
{
"grad_norm": 0.2232900708913803,
"learning_rate": 8.79740846182329e-05,
"loss": 0.0161,
"step": 13210
},
{
"grad_norm": 0.23591405153274536,
"learning_rate": 8.7952563723581e-05,
"loss": 0.0174,
"step": 13220
},
{
"grad_norm": 0.22698666155338287,
"learning_rate": 8.793102622718834e-05,
"loss": 0.0188,
"step": 13230
},
{
"grad_norm": 0.22234176099300385,
"learning_rate": 8.790947213847613e-05,
"loss": 0.0188,
"step": 13240
},
{
"grad_norm": 0.1794581264257431,
"learning_rate": 8.788790146687286e-05,
"loss": 0.0146,
"step": 13250
},
{
"grad_norm": 0.28435543179512024,
"learning_rate": 8.786631422181429e-05,
"loss": 0.0164,
"step": 13260
},
{
"grad_norm": 0.1640920341014862,
"learning_rate": 8.78447104127434e-05,
"loss": 0.0139,
"step": 13270
},
{
"grad_norm": 0.1934587061405182,
"learning_rate": 8.782309004911042e-05,
"loss": 0.0155,
"step": 13280
},
{
"grad_norm": 0.18053412437438965,
"learning_rate": 8.780145314037286e-05,
"loss": 0.0132,
"step": 13290
},
{
"grad_norm": 0.2604879140853882,
"learning_rate": 8.777979969599542e-05,
"loss": 0.0164,
"step": 13300
},
{
"grad_norm": 0.1752013862133026,
"learning_rate": 8.775812972545006e-05,
"loss": 0.0152,
"step": 13310
},
{
"grad_norm": 0.2255670130252838,
"learning_rate": 8.773644323821596e-05,
"loss": 0.0149,
"step": 13320
},
{
"grad_norm": 0.19538767635822296,
"learning_rate": 8.771474024377953e-05,
"loss": 0.0155,
"step": 13330
},
{
"grad_norm": 0.2076835185289383,
"learning_rate": 8.769302075163438e-05,
"loss": 0.0141,
"step": 13340
},
{
"grad_norm": 0.20142938196659088,
"learning_rate": 8.767128477128137e-05,
"loss": 0.0137,
"step": 13350
},
{
"grad_norm": 0.24773965775966644,
"learning_rate": 8.764953231222854e-05,
"loss": 0.0151,
"step": 13360
},
{
"grad_norm": 0.22969695925712585,
"learning_rate": 8.762776338399119e-05,
"loss": 0.0157,
"step": 13370
},
{
"grad_norm": 0.2214985191822052,
"learning_rate": 8.760597799609176e-05,
"loss": 0.0148,
"step": 13380
},
{
"grad_norm": 0.22393836081027985,
"learning_rate": 8.758417615805992e-05,
"loss": 0.0134,
"step": 13390
},
{
"grad_norm": 0.1673772782087326,
"learning_rate": 8.756235787943254e-05,
"loss": 0.0155,
"step": 13400
},
{
"grad_norm": 0.16354836523532867,
"learning_rate": 8.754052316975367e-05,
"loss": 0.0138,
"step": 13410
},
{
"grad_norm": 0.18902483582496643,
"learning_rate": 8.751867203857455e-05,
"loss": 0.0148,
"step": 13420
},
{
"grad_norm": 0.2009773999452591,
"learning_rate": 8.749680449545363e-05,
"loss": 0.0169,
"step": 13430
},
{
"grad_norm": 0.2766724228858948,
"learning_rate": 8.747492054995649e-05,
"loss": 0.0183,
"step": 13440
},
{
"grad_norm": 0.18602275848388672,
"learning_rate": 8.745302021165595e-05,
"loss": 0.017,
"step": 13450
},
{
"grad_norm": 0.24693943560123444,
"learning_rate": 8.743110349013192e-05,
"loss": 0.0177,
"step": 13460
},
{
"grad_norm": 0.1863868534564972,
"learning_rate": 8.740917039497153e-05,
"loss": 0.0201,
"step": 13470
},
{
"grad_norm": 0.210929736495018,
"learning_rate": 8.738722093576906e-05,
"loss": 0.0165,
"step": 13480
},
{
"grad_norm": 0.3053784966468811,
"learning_rate": 8.736525512212597e-05,
"loss": 0.0144,
"step": 13490
},
{
"grad_norm": 0.18047882616519928,
"learning_rate": 8.734327296365084e-05,
"loss": 0.0138,
"step": 13500
},
{
"grad_norm": 0.1734144389629364,
"learning_rate": 8.732127446995939e-05,
"loss": 0.016,
"step": 13510
},
{
"grad_norm": 0.2090451866388321,
"learning_rate": 8.729925965067454e-05,
"loss": 0.0132,
"step": 13520
},
{
"grad_norm": 0.18955454230308533,
"learning_rate": 8.72772285154263e-05,
"loss": 0.0132,
"step": 13530
},
{
"grad_norm": 0.1496249884366989,
"learning_rate": 8.725518107385187e-05,
"loss": 0.0138,
"step": 13540
},
{
"grad_norm": 0.21354801952838898,
"learning_rate": 8.72331173355955e-05,
"loss": 0.0143,
"step": 13550
},
{
"grad_norm": 0.19274447858333588,
"learning_rate": 8.721103731030867e-05,
"loss": 0.0186,
"step": 13560
},
{
"grad_norm": 0.30528634786605835,
"learning_rate": 8.718894100764989e-05,
"loss": 0.0153,
"step": 13570
},
{
"grad_norm": 0.29635849595069885,
"learning_rate": 8.716682843728485e-05,
"loss": 0.0144,
"step": 13580
},
{
"grad_norm": 0.2386879324913025,
"learning_rate": 8.714469960888634e-05,
"loss": 0.0161,
"step": 13590
},
{
"grad_norm": 0.20125404000282288,
"learning_rate": 8.712255453213427e-05,
"loss": 0.0133,
"step": 13600
},
{
"grad_norm": 0.2016288936138153,
"learning_rate": 8.710039321671563e-05,
"loss": 0.0164,
"step": 13610
},
{
"grad_norm": 0.24475125968456268,
"learning_rate": 8.707821567232456e-05,
"loss": 0.0152,
"step": 13620
},
{
"grad_norm": 0.14486651122570038,
"learning_rate": 8.705602190866225e-05,
"loss": 0.016,
"step": 13630
},
{
"grad_norm": 0.15197031199932098,
"learning_rate": 8.703381193543701e-05,
"loss": 0.0155,
"step": 13640
},
{
"grad_norm": 0.23353761434555054,
"learning_rate": 8.701158576236423e-05,
"loss": 0.0152,
"step": 13650
},
{
"grad_norm": 0.200966939330101,
"learning_rate": 8.69893433991664e-05,
"loss": 0.0129,
"step": 13660
},
{
"grad_norm": 0.20917513966560364,
"learning_rate": 8.69670848555731e-05,
"loss": 0.0157,
"step": 13670
},
{
"grad_norm": 0.22075428068637848,
"learning_rate": 8.694481014132096e-05,
"loss": 0.0155,
"step": 13680
},
{
"grad_norm": 0.2797356843948364,
"learning_rate": 8.69225192661537e-05,
"loss": 0.0172,
"step": 13690
},
{
"grad_norm": 0.21685753762722015,
"learning_rate": 8.690021223982208e-05,
"loss": 0.0151,
"step": 13700
},
{
"grad_norm": 0.18821197748184204,
"learning_rate": 8.687788907208398e-05,
"loss": 0.0165,
"step": 13710
},
{
"grad_norm": 0.21577297151088715,
"learning_rate": 8.685554977270431e-05,
"loss": 0.0174,
"step": 13720
},
{
"grad_norm": 0.23349544405937195,
"learning_rate": 8.683319435145503e-05,
"loss": 0.0151,
"step": 13730
},
{
"grad_norm": 0.2419474869966507,
"learning_rate": 8.681082281811517e-05,
"loss": 0.0174,
"step": 13740
},
{
"grad_norm": 0.2709105908870697,
"learning_rate": 8.67884351824708e-05,
"loss": 0.0129,
"step": 13750
},
{
"grad_norm": 0.21233297884464264,
"learning_rate": 8.676603145431501e-05,
"loss": 0.0147,
"step": 13760
},
{
"grad_norm": 0.17243273556232452,
"learning_rate": 8.674361164344799e-05,
"loss": 0.0153,
"step": 13770
},
{
"grad_norm": 0.2119709700345993,
"learning_rate": 8.672117575967688e-05,
"loss": 0.018,
"step": 13780
},
{
"grad_norm": 0.2274838089942932,
"learning_rate": 8.669872381281595e-05,
"loss": 0.0155,
"step": 13790
},
{
"grad_norm": 0.23185117542743683,
"learning_rate": 8.667625581268639e-05,
"loss": 0.0147,
"step": 13800
},
{
"grad_norm": 0.1754036247730255,
"learning_rate": 8.665377176911651e-05,
"loss": 0.0138,
"step": 13810
},
{
"grad_norm": 0.1926077902317047,
"learning_rate": 8.663127169194159e-05,
"loss": 0.0153,
"step": 13820
},
{
"grad_norm": 0.18572399020195007,
"learning_rate": 8.660875559100389e-05,
"loss": 0.0132,
"step": 13830
},
{
"grad_norm": 0.20062310993671417,
"learning_rate": 8.658622347615274e-05,
"loss": 0.0141,
"step": 13840
},
{
"grad_norm": 0.20048755407333374,
"learning_rate": 8.656367535724448e-05,
"loss": 0.016,
"step": 13850
},
{
"grad_norm": 0.25398576259613037,
"learning_rate": 8.65411112441424e-05,
"loss": 0.0177,
"step": 13860
},
{
"grad_norm": 0.15257631242275238,
"learning_rate": 8.651853114671679e-05,
"loss": 0.0157,
"step": 13870
},
{
"grad_norm": 0.20968742668628693,
"learning_rate": 8.649593507484499e-05,
"loss": 0.0157,
"step": 13880
},
{
"grad_norm": 0.16515901684761047,
"learning_rate": 8.647332303841126e-05,
"loss": 0.0145,
"step": 13890
},
{
"grad_norm": 0.16739679872989655,
"learning_rate": 8.645069504730689e-05,
"loss": 0.016,
"step": 13900
},
{
"grad_norm": 0.21535839140415192,
"learning_rate": 8.64280511114301e-05,
"loss": 0.0125,
"step": 13910
},
{
"grad_norm": 0.28381016850471497,
"learning_rate": 8.640539124068617e-05,
"loss": 0.0145,
"step": 13920
},
{
"grad_norm": 0.18303322792053223,
"learning_rate": 8.638271544498727e-05,
"loss": 0.015,
"step": 13930
},
{
"grad_norm": 0.22304949164390564,
"learning_rate": 8.636002373425257e-05,
"loss": 0.0142,
"step": 13940
},
{
"grad_norm": 0.19821374118328094,
"learning_rate": 8.633731611840817e-05,
"loss": 0.0163,
"step": 13950
},
{
"grad_norm": 0.24977527558803558,
"learning_rate": 8.631459260738717e-05,
"loss": 0.0143,
"step": 13960
},
{
"grad_norm": 0.19390562176704407,
"learning_rate": 8.62918532111296e-05,
"loss": 0.0148,
"step": 13970
},
{
"grad_norm": 0.21048428118228912,
"learning_rate": 8.626909793958248e-05,
"loss": 0.0169,
"step": 13980
},
{
"grad_norm": 0.2365146279335022,
"learning_rate": 8.624632680269969e-05,
"loss": 0.018,
"step": 13990
},
{
"grad_norm": 0.1944269835948944,
"learning_rate": 8.622353981044212e-05,
"loss": 0.0203,
"step": 14000
},
{
"grad_norm": 0.1678372472524643,
"learning_rate": 8.620073697277757e-05,
"loss": 0.0127,
"step": 14010
},
{
"grad_norm": 0.18243327736854553,
"learning_rate": 8.617791829968079e-05,
"loss": 0.0137,
"step": 14020
},
{
"grad_norm": 0.25175967812538147,
"learning_rate": 8.615508380113344e-05,
"loss": 0.0125,
"step": 14030
},
{
"grad_norm": 0.22590848803520203,
"learning_rate": 8.613223348712408e-05,
"loss": 0.0153,
"step": 14040
},
{
"grad_norm": 0.2392529845237732,
"learning_rate": 8.610936736764824e-05,
"loss": 0.0171,
"step": 14050
},
{
"grad_norm": 0.20959992706775665,
"learning_rate": 8.608648545270833e-05,
"loss": 0.0158,
"step": 14060
},
{
"grad_norm": 0.18744762241840363,
"learning_rate": 8.606358775231366e-05,
"loss": 0.0162,
"step": 14070
},
{
"grad_norm": 0.24289417266845703,
"learning_rate": 8.60406742764805e-05,
"loss": 0.014,
"step": 14080
},
{
"grad_norm": 0.2292727530002594,
"learning_rate": 8.601774503523195e-05,
"loss": 0.0162,
"step": 14090
},
{
"grad_norm": 0.15884733200073242,
"learning_rate": 8.599480003859805e-05,
"loss": 0.0148,
"step": 14100
},
{
"grad_norm": 0.17099329829216003,
"learning_rate": 8.597183929661573e-05,
"loss": 0.014,
"step": 14110
},
{
"grad_norm": 0.13494643568992615,
"learning_rate": 8.594886281932879e-05,
"loss": 0.0138,
"step": 14120
},
{
"grad_norm": 0.23635946214199066,
"learning_rate": 8.59258706167879e-05,
"loss": 0.0145,
"step": 14130
},
{
"grad_norm": 0.15357069671154022,
"learning_rate": 8.590286269905068e-05,
"loss": 0.0122,
"step": 14140
},
{
"grad_norm": 0.2180400788784027,
"learning_rate": 8.587983907618154e-05,
"loss": 0.0161,
"step": 14150
},
{
"grad_norm": 0.21714597940444946,
"learning_rate": 8.585679975825178e-05,
"loss": 0.0165,
"step": 14160
},
{
"grad_norm": 0.3220981955528259,
"learning_rate": 8.583374475533962e-05,
"loss": 0.0155,
"step": 14170
},
{
"grad_norm": 0.1914060115814209,
"learning_rate": 8.581067407753009e-05,
"loss": 0.018,
"step": 14180
},
{
"grad_norm": 0.20937086641788483,
"learning_rate": 8.578758773491507e-05,
"loss": 0.0153,
"step": 14190
},
{
"grad_norm": 0.19443756341934204,
"learning_rate": 8.576448573759332e-05,
"loss": 0.0163,
"step": 14200
},
{
"grad_norm": 0.16645048558712006,
"learning_rate": 8.574136809567044e-05,
"loss": 0.0107,
"step": 14210
},
{
"grad_norm": 0.23980863392353058,
"learning_rate": 8.57182348192589e-05,
"loss": 0.0131,
"step": 14220
},
{
"grad_norm": 0.25855135917663574,
"learning_rate": 8.569508591847792e-05,
"loss": 0.0141,
"step": 14230
},
{
"grad_norm": 0.21802735328674316,
"learning_rate": 8.567192140345367e-05,
"loss": 0.0171,
"step": 14240
},
{
"grad_norm": 0.21974194049835205,
"learning_rate": 8.564874128431906e-05,
"loss": 0.0162,
"step": 14250
},
{
"grad_norm": 0.25683102011680603,
"learning_rate": 8.562554557121389e-05,
"loss": 0.014,
"step": 14260
},
{
"grad_norm": 0.2032209038734436,
"learning_rate": 8.560233427428475e-05,
"loss": 0.0154,
"step": 14270
},
{
"grad_norm": 0.1902657151222229,
"learning_rate": 8.557910740368503e-05,
"loss": 0.0149,
"step": 14280
},
{
"grad_norm": 0.21987296640872955,
"learning_rate": 8.555586496957495e-05,
"loss": 0.015,
"step": 14290
},
{
"grad_norm": 0.18675370514392853,
"learning_rate": 8.553260698212155e-05,
"loss": 0.0186,
"step": 14300
},
{
"grad_norm": 0.1780502051115036,
"learning_rate": 8.550933345149868e-05,
"loss": 0.0137,
"step": 14310
},
{
"grad_norm": 0.19038142263889313,
"learning_rate": 8.548604438788696e-05,
"loss": 0.0157,
"step": 14320
},
{
"grad_norm": 0.15622252225875854,
"learning_rate": 8.546273980147383e-05,
"loss": 0.0154,
"step": 14330
},
{
"grad_norm": 0.21982739865779877,
"learning_rate": 8.543941970245348e-05,
"loss": 0.0129,
"step": 14340
},
{
"grad_norm": 0.180136576294899,
"learning_rate": 8.541608410102693e-05,
"loss": 0.0148,
"step": 14350
},
{
"grad_norm": 0.20178085565567017,
"learning_rate": 8.539273300740195e-05,
"loss": 0.0145,
"step": 14360
},
{
"grad_norm": 0.17430616915225983,
"learning_rate": 8.536936643179313e-05,
"loss": 0.012,
"step": 14370
},
{
"grad_norm": 0.1889248788356781,
"learning_rate": 8.534598438442179e-05,
"loss": 0.0123,
"step": 14380
},
{
"grad_norm": 0.22322812676429749,
"learning_rate": 8.532258687551603e-05,
"loss": 0.0167,
"step": 14390
},
{
"grad_norm": 0.24100732803344727,
"learning_rate": 8.529917391531071e-05,
"loss": 0.0144,
"step": 14400
},
{
"grad_norm": 0.24971657991409302,
"learning_rate": 8.527574551404747e-05,
"loss": 0.016,
"step": 14410
},
{
"grad_norm": 0.19462698698043823,
"learning_rate": 8.525230168197468e-05,
"loss": 0.015,
"step": 14420
},
{
"grad_norm": 0.20197591185569763,
"learning_rate": 8.522884242934745e-05,
"loss": 0.0127,
"step": 14430
},
{
"grad_norm": 0.21552525460720062,
"learning_rate": 8.520536776642768e-05,
"loss": 0.0145,
"step": 14440
},
{
"grad_norm": 0.23887455463409424,
"learning_rate": 8.5181877703484e-05,
"loss": 0.015,
"step": 14450
},
{
"grad_norm": 0.20110422372817993,
"learning_rate": 8.51583722507917e-05,
"loss": 0.0144,
"step": 14460
},
{
"grad_norm": 0.21691909432411194,
"learning_rate": 8.513485141863293e-05,
"loss": 0.0171,
"step": 14470
},
{
"grad_norm": 0.17417795956134796,
"learning_rate": 8.511131521729647e-05,
"loss": 0.0134,
"step": 14480
},
{
"grad_norm": 0.19722072780132294,
"learning_rate": 8.508776365707787e-05,
"loss": 0.0148,
"step": 14490
},
{
"grad_norm": 0.2367250919342041,
"learning_rate": 8.506419674827934e-05,
"loss": 0.0142,
"step": 14500
},
{
"grad_norm": 0.38434094190597534,
"learning_rate": 8.50406145012099e-05,
"loss": 0.0166,
"step": 14510
},
{
"grad_norm": 0.20360474288463593,
"learning_rate": 8.501701692618519e-05,
"loss": 0.0142,
"step": 14520
},
{
"grad_norm": 0.2321837991476059,
"learning_rate": 8.499340403352761e-05,
"loss": 0.0148,
"step": 14530
},
{
"grad_norm": 0.22018307447433472,
"learning_rate": 8.496977583356623e-05,
"loss": 0.0179,
"step": 14540
},
{
"grad_norm": 0.1777219921350479,
"learning_rate": 8.494613233663684e-05,
"loss": 0.0136,
"step": 14550
},
{
"grad_norm": 0.1491110771894455,
"learning_rate": 8.492247355308189e-05,
"loss": 0.014,
"step": 14560
},
{
"grad_norm": 0.18252411484718323,
"learning_rate": 8.489879949325056e-05,
"loss": 0.0162,
"step": 14570
},
{
"grad_norm": 0.24680772423744202,
"learning_rate": 8.487511016749868e-05,
"loss": 0.0151,
"step": 14580
},
{
"grad_norm": 0.23217321932315826,
"learning_rate": 8.485140558618874e-05,
"loss": 0.0142,
"step": 14590
},
{
"grad_norm": 0.17018473148345947,
"learning_rate": 8.482768575968995e-05,
"loss": 0.013,
"step": 14600
},
{
"grad_norm": 0.15557672083377838,
"learning_rate": 8.480395069837818e-05,
"loss": 0.0144,
"step": 14610
},
{
"grad_norm": 0.2567611336708069,
"learning_rate": 8.478020041263595e-05,
"loss": 0.0163,
"step": 14620
},
{
"grad_norm": 0.22482895851135254,
"learning_rate": 8.475643491285242e-05,
"loss": 0.0164,
"step": 14630
},
{
"grad_norm": 0.2013397067785263,
"learning_rate": 8.473265420942345e-05,
"loss": 0.0151,
"step": 14640
},
{
"grad_norm": 0.22968493402004242,
"learning_rate": 8.470885831275151e-05,
"loss": 0.0127,
"step": 14650
},
{
"grad_norm": 0.168733611702919,
"learning_rate": 8.468504723324574e-05,
"loss": 0.0174,
"step": 14660
},
{
"grad_norm": 0.2146560549736023,
"learning_rate": 8.466122098132193e-05,
"loss": 0.0137,
"step": 14670
},
{
"grad_norm": 0.14688140153884888,
"learning_rate": 8.463737956740245e-05,
"loss": 0.015,
"step": 14680
},
{
"grad_norm": 0.20747263729572296,
"learning_rate": 8.461352300191639e-05,
"loss": 0.0145,
"step": 14690
},
{
"grad_norm": 0.1828698068857193,
"learning_rate": 8.45896512952994e-05,
"loss": 0.0148,
"step": 14700
},
{
"grad_norm": 0.1762228161096573,
"learning_rate": 8.456576445799377e-05,
"loss": 0.0112,
"step": 14710
},
{
"grad_norm": 0.20817849040031433,
"learning_rate": 8.454186250044844e-05,
"loss": 0.0145,
"step": 14720
},
{
"grad_norm": 0.22076599299907684,
"learning_rate": 8.451794543311892e-05,
"loss": 0.0145,
"step": 14730
},
{
"grad_norm": 0.19213898479938507,
"learning_rate": 8.449401326646736e-05,
"loss": 0.017,
"step": 14740
},
{
"grad_norm": 0.15518803894519806,
"learning_rate": 8.447006601096248e-05,
"loss": 0.0141,
"step": 14750
},
{
"grad_norm": 0.21783936023712158,
"learning_rate": 8.444610367707964e-05,
"loss": 0.0122,
"step": 14760
},
{
"grad_norm": 0.14661931991577148,
"learning_rate": 8.442212627530078e-05,
"loss": 0.0143,
"step": 14770
},
{
"grad_norm": 0.1741727739572525,
"learning_rate": 8.439813381611441e-05,
"loss": 0.0127,
"step": 14780
},
{
"grad_norm": 0.2288089245557785,
"learning_rate": 8.437412631001567e-05,
"loss": 0.0146,
"step": 14790
},
{
"grad_norm": 0.20780333876609802,
"learning_rate": 8.435010376750626e-05,
"loss": 0.0127,
"step": 14800
},
{
"grad_norm": 0.22321802377700806,
"learning_rate": 8.432606619909442e-05,
"loss": 0.0119,
"step": 14810
},
{
"grad_norm": 0.21799057722091675,
"learning_rate": 8.430201361529506e-05,
"loss": 0.0145,
"step": 14820
},
{
"grad_norm": 0.1665000021457672,
"learning_rate": 8.427794602662954e-05,
"loss": 0.0141,
"step": 14830
},
{
"grad_norm": 0.15132303535938263,
"learning_rate": 8.425386344362586e-05,
"loss": 0.0139,
"step": 14840
},
{
"grad_norm": 0.20641767978668213,
"learning_rate": 8.422976587681859e-05,
"loss": 0.0137,
"step": 14850
},
{
"grad_norm": 0.15357741713523865,
"learning_rate": 8.42056533367488e-05,
"loss": 0.0112,
"step": 14860
},
{
"grad_norm": 0.17750313878059387,
"learning_rate": 8.41815258339641e-05,
"loss": 0.0114,
"step": 14870
},
{
"grad_norm": 0.23370198905467987,
"learning_rate": 8.415738337901874e-05,
"loss": 0.0148,
"step": 14880
},
{
"grad_norm": 0.17446325719356537,
"learning_rate": 8.413322598247342e-05,
"loss": 0.0145,
"step": 14890
},
{
"grad_norm": 0.2266729176044464,
"learning_rate": 8.41090536548954e-05,
"loss": 0.0131,
"step": 14900
},
{
"grad_norm": 0.2071094959974289,
"learning_rate": 8.408486640685849e-05,
"loss": 0.0129,
"step": 14910
},
{
"grad_norm": 0.19125568866729736,
"learning_rate": 8.4060664248943e-05,
"loss": 0.0177,
"step": 14920
},
{
"grad_norm": 0.21832269430160522,
"learning_rate": 8.40364471917358e-05,
"loss": 0.0157,
"step": 14930
},
{
"grad_norm": 0.17632268369197845,
"learning_rate": 8.401221524583024e-05,
"loss": 0.0167,
"step": 14940
},
{
"grad_norm": 0.2568971514701843,
"learning_rate": 8.398796842182619e-05,
"loss": 0.0132,
"step": 14950
},
{
"grad_norm": 0.24805079400539398,
"learning_rate": 8.396370673033006e-05,
"loss": 0.0138,
"step": 14960
},
{
"grad_norm": 0.14267311990261078,
"learning_rate": 8.39394301819547e-05,
"loss": 0.0193,
"step": 14970
},
{
"grad_norm": 0.21560396254062653,
"learning_rate": 8.391513878731949e-05,
"loss": 0.0149,
"step": 14980
},
{
"grad_norm": 0.1859932839870453,
"learning_rate": 8.389083255705037e-05,
"loss": 0.0133,
"step": 14990
},
{
"grad_norm": 0.23598697781562805,
"learning_rate": 8.386651150177968e-05,
"loss": 0.0161,
"step": 15000
},
{
"grad_norm": 0.13201259076595306,
"learning_rate": 8.384217563214627e-05,
"loss": 0.0132,
"step": 15010
},
{
"grad_norm": 0.2015417516231537,
"learning_rate": 8.381782495879546e-05,
"loss": 0.0144,
"step": 15020
},
{
"grad_norm": 0.22176051139831543,
"learning_rate": 8.37934594923791e-05,
"loss": 0.0126,
"step": 15030
},
{
"grad_norm": 0.2201116681098938,
"learning_rate": 8.376907924355546e-05,
"loss": 0.0127,
"step": 15040
},
{
"grad_norm": 0.2633454203605652,
"learning_rate": 8.374468422298926e-05,
"loss": 0.0141,
"step": 15050
},
{
"grad_norm": 0.22914880514144897,
"learning_rate": 8.372027444135176e-05,
"loss": 0.012,
"step": 15060
},
{
"grad_norm": 0.2098378837108612,
"learning_rate": 8.36958499093206e-05,
"loss": 0.0136,
"step": 15070
},
{
"grad_norm": 0.22794660925865173,
"learning_rate": 8.367141063757988e-05,
"loss": 0.0139,
"step": 15080
},
{
"grad_norm": 0.2820660471916199,
"learning_rate": 8.364695663682022e-05,
"loss": 0.0139,
"step": 15090
},
{
"grad_norm": 0.24393253028392792,
"learning_rate": 8.362248791773858e-05,
"loss": 0.0138,
"step": 15100
},
{
"grad_norm": 0.20994621515274048,
"learning_rate": 8.359800449103842e-05,
"loss": 0.0148,
"step": 15110
},
{
"grad_norm": 0.18136835098266602,
"learning_rate": 8.357350636742967e-05,
"loss": 0.0135,
"step": 15120
},
{
"grad_norm": 0.20085476338863373,
"learning_rate": 8.354899355762858e-05,
"loss": 0.0139,
"step": 15130
},
{
"grad_norm": 0.21443253755569458,
"learning_rate": 8.352446607235791e-05,
"loss": 0.0183,
"step": 15140
},
{
"grad_norm": 0.19582930207252502,
"learning_rate": 8.349992392234683e-05,
"loss": 0.0168,
"step": 15150
},
{
"grad_norm": 0.21877464652061462,
"learning_rate": 8.347536711833088e-05,
"loss": 0.0171,
"step": 15160
},
{
"grad_norm": 0.2102109044790268,
"learning_rate": 8.345079567105206e-05,
"loss": 0.0148,
"step": 15170
},
{
"grad_norm": 0.24954672157764435,
"learning_rate": 8.342620959125875e-05,
"loss": 0.0147,
"step": 15180
},
{
"grad_norm": 0.18947342038154602,
"learning_rate": 8.340160888970574e-05,
"loss": 0.0166,
"step": 15190
},
{
"grad_norm": 0.22434794902801514,
"learning_rate": 8.337699357715422e-05,
"loss": 0.0135,
"step": 15200
},
{
"grad_norm": 0.18232351541519165,
"learning_rate": 8.335236366437174e-05,
"loss": 0.015,
"step": 15210
},
{
"grad_norm": 0.20749615132808685,
"learning_rate": 8.332771916213227e-05,
"loss": 0.0135,
"step": 15220
},
{
"grad_norm": 0.20341046154499054,
"learning_rate": 8.330306008121618e-05,
"loss": 0.0129,
"step": 15230
},
{
"grad_norm": 0.22228018939495087,
"learning_rate": 8.327838643241014e-05,
"loss": 0.0132,
"step": 15240
},
{
"grad_norm": 0.2750200033187866,
"learning_rate": 8.32536982265073e-05,
"loss": 0.0134,
"step": 15250
},
{
"grad_norm": 0.1898784041404724,
"learning_rate": 8.32289954743071e-05,
"loss": 0.0138,
"step": 15260
},
{
"grad_norm": 0.19248391687870026,
"learning_rate": 8.320427818661532e-05,
"loss": 0.0132,
"step": 15270
},
{
"grad_norm": 0.18432924151420593,
"learning_rate": 8.317954637424419e-05,
"loss": 0.0151,
"step": 15280
},
{
"grad_norm": 0.18091033399105072,
"learning_rate": 8.315480004801221e-05,
"loss": 0.0133,
"step": 15290
},
{
"grad_norm": 0.22201557457447052,
"learning_rate": 8.31300392187443e-05,
"loss": 0.0133,
"step": 15300
},
{
"grad_norm": 0.20445020496845245,
"learning_rate": 8.31052638972717e-05,
"loss": 0.0123,
"step": 15310
},
{
"grad_norm": 0.22878718376159668,
"learning_rate": 8.30804740944319e-05,
"loss": 0.0153,
"step": 15320
},
{
"grad_norm": 0.19531208276748657,
"learning_rate": 8.305566982106887e-05,
"loss": 0.0131,
"step": 15330
},
{
"grad_norm": 0.3049273192882538,
"learning_rate": 8.303085108803282e-05,
"loss": 0.0151,
"step": 15340
},
{
"grad_norm": 0.2322346568107605,
"learning_rate": 8.300601790618032e-05,
"loss": 0.0148,
"step": 15350
},
{
"grad_norm": 0.23208405077457428,
"learning_rate": 8.298117028637423e-05,
"loss": 0.0142,
"step": 15360
},
{
"grad_norm": 0.21119150519371033,
"learning_rate": 8.295630823948375e-05,
"loss": 0.017,
"step": 15370
},
{
"grad_norm": 0.2214541733264923,
"learning_rate": 8.293143177638438e-05,
"loss": 0.0173,
"step": 15380
},
{
"grad_norm": 0.18076449632644653,
"learning_rate": 8.290654090795795e-05,
"loss": 0.0141,
"step": 15390
},
{
"grad_norm": 0.19351746141910553,
"learning_rate": 8.288163564509253e-05,
"loss": 0.0163,
"step": 15400
},
{
"grad_norm": 0.2070235162973404,
"learning_rate": 8.285671599868257e-05,
"loss": 0.0157,
"step": 15410
},
{
"grad_norm": 0.18620090186595917,
"learning_rate": 8.283178197962875e-05,
"loss": 0.0156,
"step": 15420
},
{
"grad_norm": 0.15074557065963745,
"learning_rate": 8.280683359883808e-05,
"loss": 0.0156,
"step": 15430
},
{
"grad_norm": 0.18230272829532623,
"learning_rate": 8.278187086722377e-05,
"loss": 0.0163,
"step": 15440
},
{
"grad_norm": 0.1810087114572525,
"learning_rate": 8.275689379570544e-05,
"loss": 0.0138,
"step": 15450
},
{
"grad_norm": 0.16751958429813385,
"learning_rate": 8.273190239520887e-05,
"loss": 0.0147,
"step": 15460
},
{
"grad_norm": 0.21276356279850006,
"learning_rate": 8.270689667666616e-05,
"loss": 0.016,
"step": 15470
},
{
"grad_norm": 0.19457021355628967,
"learning_rate": 8.268187665101563e-05,
"loss": 0.0161,
"step": 15480
},
{
"grad_norm": 0.17939449846744537,
"learning_rate": 8.265684232920194e-05,
"loss": 0.0216,
"step": 15490
},
{
"grad_norm": 0.22985893487930298,
"learning_rate": 8.263179372217592e-05,
"loss": 0.0179,
"step": 15500
},
{
"grad_norm": 0.19748789072036743,
"learning_rate": 8.260673084089468e-05,
"loss": 0.0168,
"step": 15510
},
{
"grad_norm": 0.2245890200138092,
"learning_rate": 8.25816536963216e-05,
"loss": 0.0153,
"step": 15520
},
{
"grad_norm": 0.18914680182933807,
"learning_rate": 8.255656229942626e-05,
"loss": 0.0141,
"step": 15530
},
{
"grad_norm": 0.1940603405237198,
"learning_rate": 8.253145666118447e-05,
"loss": 0.0127,
"step": 15540
},
{
"grad_norm": 0.20917953550815582,
"learning_rate": 8.250633679257834e-05,
"loss": 0.0157,
"step": 15550
},
{
"grad_norm": 0.23818765580654144,
"learning_rate": 8.24812027045961e-05,
"loss": 0.014,
"step": 15560
},
{
"grad_norm": 0.2166333943605423,
"learning_rate": 8.245605440823226e-05,
"loss": 0.0138,
"step": 15570
},
{
"grad_norm": 0.1811836063861847,
"learning_rate": 8.243089191448759e-05,
"loss": 0.0125,
"step": 15580
},
{
"grad_norm": 0.19130009412765503,
"learning_rate": 8.240571523436896e-05,
"loss": 0.0155,
"step": 15590
},
{
"grad_norm": 0.18153251707553864,
"learning_rate": 8.238052437888955e-05,
"loss": 0.0148,
"step": 15600
},
{
"grad_norm": 0.19900715351104736,
"learning_rate": 8.235531935906866e-05,
"loss": 0.0143,
"step": 15610
},
{
"grad_norm": 0.15515442192554474,
"learning_rate": 8.233010018593185e-05,
"loss": 0.0123,
"step": 15620
},
{
"grad_norm": 0.15772539377212524,
"learning_rate": 8.230486687051082e-05,
"loss": 0.0134,
"step": 15630
},
{
"grad_norm": 0.1765284538269043,
"learning_rate": 8.22796194238435e-05,
"loss": 0.0139,
"step": 15640
},
{
"grad_norm": 0.18094630539417267,
"learning_rate": 8.225435785697397e-05,
"loss": 0.0137,
"step": 15650
},
{
"grad_norm": 0.22996364533901215,
"learning_rate": 8.222908218095252e-05,
"loss": 0.0149,
"step": 15660
},
{
"grad_norm": 0.19521331787109375,
"learning_rate": 8.220379240683553e-05,
"loss": 0.0142,
"step": 15670
},
{
"grad_norm": 0.24973252415657043,
"learning_rate": 8.217848854568566e-05,
"loss": 0.014,
"step": 15680
},
{
"grad_norm": 0.15194422006607056,
"learning_rate": 8.215317060857167e-05,
"loss": 0.0129,
"step": 15690
},
{
"grad_norm": 0.18245375156402588,
"learning_rate": 8.212783860656847e-05,
"loss": 0.0112,
"step": 15700
},
{
"grad_norm": 0.17329463362693787,
"learning_rate": 8.210249255075717e-05,
"loss": 0.0121,
"step": 15710
},
{
"grad_norm": 0.2138417512178421,
"learning_rate": 8.207713245222496e-05,
"loss": 0.0133,
"step": 15720
},
{
"grad_norm": 0.25326699018478394,
"learning_rate": 8.205175832206522e-05,
"loss": 0.0152,
"step": 15730
},
{
"grad_norm": 0.22895698249340057,
"learning_rate": 8.202637017137747e-05,
"loss": 0.015,
"step": 15740
},
{
"grad_norm": 0.2192966789007187,
"learning_rate": 8.200096801126734e-05,
"loss": 0.0137,
"step": 15750
},
{
"grad_norm": 0.25613531470298767,
"learning_rate": 8.197555185284658e-05,
"loss": 0.0155,
"step": 15760
},
{
"grad_norm": 0.17924650013446808,
"learning_rate": 8.19501217072331e-05,
"loss": 0.0147,
"step": 15770
},
{
"grad_norm": 0.1897151917219162,
"learning_rate": 8.192467758555091e-05,
"loss": 0.0141,
"step": 15780
},
{
"grad_norm": 0.18003207445144653,
"learning_rate": 8.189921949893014e-05,
"loss": 0.0142,
"step": 15790
},
{
"grad_norm": 0.18742339313030243,
"learning_rate": 8.187374745850702e-05,
"loss": 0.0137,
"step": 15800
},
{
"grad_norm": 0.19231750071048737,
"learning_rate": 8.184826147542387e-05,
"loss": 0.0128,
"step": 15810
},
{
"grad_norm": 0.21481953561306,
"learning_rate": 8.18227615608291e-05,
"loss": 0.0182,
"step": 15820
},
{
"grad_norm": 0.18483859300613403,
"learning_rate": 8.179724772587732e-05,
"loss": 0.0126,
"step": 15830
},
{
"grad_norm": 0.17660047113895416,
"learning_rate": 8.177171998172908e-05,
"loss": 0.0178,
"step": 15840
},
{
"grad_norm": 0.1728576272726059,
"learning_rate": 8.174617833955112e-05,
"loss": 0.0138,
"step": 15850
},
{
"grad_norm": 0.21393314003944397,
"learning_rate": 8.172062281051618e-05,
"loss": 0.0147,
"step": 15860
},
{
"grad_norm": 0.16815169155597687,
"learning_rate": 8.169505340580316e-05,
"loss": 0.0132,
"step": 15870
},
{
"grad_norm": 0.25227484107017517,
"learning_rate": 8.166947013659696e-05,
"loss": 0.0148,
"step": 15880
},
{
"grad_norm": 0.19321474432945251,
"learning_rate": 8.164387301408858e-05,
"loss": 0.0151,
"step": 15890
},
{
"grad_norm": 0.29287952184677124,
"learning_rate": 8.161826204947507e-05,
"loss": 0.0188,
"step": 15900
},
{
"grad_norm": 0.20355576276779175,
"learning_rate": 8.159263725395954e-05,
"loss": 0.0182,
"step": 15910
},
{
"grad_norm": 0.2491297423839569,
"learning_rate": 8.156699863875115e-05,
"loss": 0.02,
"step": 15920
},
{
"grad_norm": 0.21967370808124542,
"learning_rate": 8.154134621506508e-05,
"loss": 0.0152,
"step": 15930
},
{
"grad_norm": 0.3160310983657837,
"learning_rate": 8.15156799941226e-05,
"loss": 0.0144,
"step": 15940
},
{
"grad_norm": 0.18960490822792053,
"learning_rate": 8.148999998715096e-05,
"loss": 0.0173,
"step": 15950
},
{
"grad_norm": 0.1715354025363922,
"learning_rate": 8.146430620538348e-05,
"loss": 0.0141,
"step": 15960
},
{
"grad_norm": 0.16950874030590057,
"learning_rate": 8.143859866005949e-05,
"loss": 0.014,
"step": 15970
},
{
"grad_norm": 0.20162555575370789,
"learning_rate": 8.141287736242438e-05,
"loss": 0.0136,
"step": 15980
},
{
"grad_norm": 0.1490805596113205,
"learning_rate": 8.138714232372945e-05,
"loss": 0.014,
"step": 15990
},
{
"grad_norm": 0.1909923553466797,
"learning_rate": 8.136139355523213e-05,
"loss": 0.0164,
"step": 16000
},
{
"grad_norm": 0.21604950726032257,
"learning_rate": 8.13356310681958e-05,
"loss": 0.0142,
"step": 16010
},
{
"grad_norm": 0.18406419456005096,
"learning_rate": 8.130985487388982e-05,
"loss": 0.0156,
"step": 16020
},
{
"grad_norm": 0.17565736174583435,
"learning_rate": 8.12840649835896e-05,
"loss": 0.0134,
"step": 16030
},
{
"grad_norm": 0.15658347308635712,
"learning_rate": 8.125826140857653e-05,
"loss": 0.0143,
"step": 16040
},
{
"grad_norm": 0.20856918394565582,
"learning_rate": 8.123244416013792e-05,
"loss": 0.0142,
"step": 16050
},
{
"grad_norm": 0.332287460565567,
"learning_rate": 8.120661324956716e-05,
"loss": 0.0159,
"step": 16060
},
{
"grad_norm": 0.19089867174625397,
"learning_rate": 8.118076868816353e-05,
"loss": 0.0128,
"step": 16070
},
{
"grad_norm": 0.18979990482330322,
"learning_rate": 8.115491048723234e-05,
"loss": 0.0145,
"step": 16080
},
{
"grad_norm": 0.19986410439014435,
"learning_rate": 8.112903865808486e-05,
"loss": 0.012,
"step": 16090
},
{
"grad_norm": 0.18363726139068604,
"learning_rate": 8.110315321203826e-05,
"loss": 0.013,
"step": 16100
},
{
"grad_norm": 0.2015133500099182,
"learning_rate": 8.107725416041575e-05,
"loss": 0.0125,
"step": 16110
},
{
"grad_norm": 0.20112544298171997,
"learning_rate": 8.105134151454645e-05,
"loss": 0.0128,
"step": 16120
},
{
"grad_norm": 0.1677011400461197,
"learning_rate": 8.102541528576544e-05,
"loss": 0.0143,
"step": 16130
},
{
"grad_norm": 0.22358575463294983,
"learning_rate": 8.099947548541371e-05,
"loss": 0.0133,
"step": 16140
},
{
"grad_norm": 0.20383265614509583,
"learning_rate": 8.097352212483824e-05,
"loss": 0.0138,
"step": 16150
},
{
"grad_norm": 0.2501258850097656,
"learning_rate": 8.094755521539187e-05,
"loss": 0.0147,
"step": 16160
},
{
"grad_norm": 0.1807689517736435,
"learning_rate": 8.092157476843346e-05,
"loss": 0.0127,
"step": 16170
},
{
"grad_norm": 0.1771225482225418,
"learning_rate": 8.089558079532769e-05,
"loss": 0.0146,
"step": 16180
},
{
"grad_norm": 0.2019912153482437,
"learning_rate": 8.086957330744522e-05,
"loss": 0.0134,
"step": 16190
},
{
"grad_norm": 0.26713356375694275,
"learning_rate": 8.084355231616265e-05,
"loss": 0.0132,
"step": 16200
},
{
"grad_norm": 0.2008967250585556,
"learning_rate": 8.08175178328624e-05,
"loss": 0.013,
"step": 16210
},
{
"grad_norm": 0.1896194964647293,
"learning_rate": 8.079146986893284e-05,
"loss": 0.0144,
"step": 16220
},
{
"grad_norm": 0.12849321961402893,
"learning_rate": 8.076540843576826e-05,
"loss": 0.0137,
"step": 16230
},
{
"grad_norm": 0.28719332814216614,
"learning_rate": 8.073933354476879e-05,
"loss": 0.0152,
"step": 16240
},
{
"grad_norm": 0.2001834660768509,
"learning_rate": 8.07132452073405e-05,
"loss": 0.0124,
"step": 16250
},
{
"grad_norm": 0.20847351849079132,
"learning_rate": 8.068714343489529e-05,
"loss": 0.0169,
"step": 16260
},
{
"grad_norm": 0.21027639508247375,
"learning_rate": 8.066102823885097e-05,
"loss": 0.0148,
"step": 16270
},
{
"grad_norm": 0.20712073147296906,
"learning_rate": 8.063489963063125e-05,
"loss": 0.017,
"step": 16280
},
{
"grad_norm": 0.18911238014698029,
"learning_rate": 8.06087576216656e-05,
"loss": 0.0117,
"step": 16290
},
{
"grad_norm": 0.20391049981117249,
"learning_rate": 8.058260222338946e-05,
"loss": 0.0136,
"step": 16300
},
{
"grad_norm": 0.20889630913734436,
"learning_rate": 8.05564334472441e-05,
"loss": 0.0152,
"step": 16310
},
{
"grad_norm": 0.21257750689983368,
"learning_rate": 8.053025130467664e-05,
"loss": 0.0134,
"step": 16320
},
{
"grad_norm": 0.21835851669311523,
"learning_rate": 8.050405580714e-05,
"loss": 0.0116,
"step": 16330
},
{
"grad_norm": 0.17834635078907013,
"learning_rate": 8.047784696609301e-05,
"loss": 0.0113,
"step": 16340
},
{
"grad_norm": 0.16305570304393768,
"learning_rate": 8.045162479300029e-05,
"loss": 0.014,
"step": 16350
},
{
"grad_norm": 0.22485600411891937,
"learning_rate": 8.042538929933232e-05,
"loss": 0.0164,
"step": 16360
},
{
"grad_norm": 0.18647557497024536,
"learning_rate": 8.039914049656539e-05,
"loss": 0.0115,
"step": 16370
},
{
"grad_norm": 0.1617908924818039,
"learning_rate": 8.037287839618163e-05,
"loss": 0.0109,
"step": 16380
},
{
"grad_norm": 0.18964210152626038,
"learning_rate": 8.034660300966898e-05,
"loss": 0.0126,
"step": 16390
},
{
"grad_norm": 0.17816808819770813,
"learning_rate": 8.032031434852115e-05,
"loss": 0.0161,
"step": 16400
},
{
"grad_norm": 0.18561047315597534,
"learning_rate": 8.029401242423773e-05,
"loss": 0.0145,
"step": 16410
},
{
"grad_norm": 0.21064558625221252,
"learning_rate": 8.026769724832406e-05,
"loss": 0.0145,
"step": 16420
},
{
"grad_norm": 0.23553688824176788,
"learning_rate": 8.02413688322913e-05,
"loss": 0.0114,
"step": 16430
},
{
"grad_norm": 0.21091671288013458,
"learning_rate": 8.021502718765639e-05,
"loss": 0.0117,
"step": 16440
},
{
"grad_norm": 0.23021844029426575,
"learning_rate": 8.018867232594204e-05,
"loss": 0.015,
"step": 16450
},
{
"grad_norm": 0.24437488615512848,
"learning_rate": 8.016230425867679e-05,
"loss": 0.0171,
"step": 16460
},
{
"grad_norm": 0.16909712553024292,
"learning_rate": 8.013592299739494e-05,
"loss": 0.0126,
"step": 16470
},
{
"grad_norm": 0.14658832550048828,
"learning_rate": 8.01095285536365e-05,
"loss": 0.0137,
"step": 16480
},
{
"grad_norm": 0.14560791850090027,
"learning_rate": 8.008312093894735e-05,
"loss": 0.0119,
"step": 16490
},
{
"grad_norm": 0.16564151644706726,
"learning_rate": 8.005670016487903e-05,
"loss": 0.012,
"step": 16500
},
{
"grad_norm": 0.23175086081027985,
"learning_rate": 8.003026624298892e-05,
"loss": 0.0102,
"step": 16510
},
{
"grad_norm": 0.16667188704013824,
"learning_rate": 8.00038191848401e-05,
"loss": 0.0134,
"step": 16520
},
{
"grad_norm": 0.20373383164405823,
"learning_rate": 7.997735900200142e-05,
"loss": 0.0144,
"step": 16530
},
{
"grad_norm": 0.1875503957271576,
"learning_rate": 7.995088570604745e-05,
"loss": 0.013,
"step": 16540
},
{
"grad_norm": 0.14034579694271088,
"learning_rate": 7.992439930855852e-05,
"loss": 0.0109,
"step": 16550
},
{
"grad_norm": 0.22447502613067627,
"learning_rate": 7.989789982112067e-05,
"loss": 0.0143,
"step": 16560
},
{
"grad_norm": 0.1913711130619049,
"learning_rate": 7.987138725532572e-05,
"loss": 0.0147,
"step": 16570
},
{
"grad_norm": 0.17126096785068512,
"learning_rate": 7.98448616227711e-05,
"loss": 0.0123,
"step": 16580
},
{
"grad_norm": 0.15150243043899536,
"learning_rate": 7.981832293506004e-05,
"loss": 0.013,
"step": 16590
},
{
"grad_norm": 0.14691191911697388,
"learning_rate": 7.97917712038015e-05,
"loss": 0.0122,
"step": 16600
},
{
"grad_norm": 0.15040814876556396,
"learning_rate": 7.976520644061007e-05,
"loss": 0.0101,
"step": 16610
},
{
"grad_norm": 0.22900432348251343,
"learning_rate": 7.973862865710611e-05,
"loss": 0.0125,
"step": 16620
},
{
"grad_norm": 0.14741674065589905,
"learning_rate": 7.971203786491563e-05,
"loss": 0.0156,
"step": 16630
},
{
"grad_norm": 0.23117634654045105,
"learning_rate": 7.968543407567034e-05,
"loss": 0.0131,
"step": 16640
},
{
"grad_norm": 0.1545952558517456,
"learning_rate": 7.965881730100765e-05,
"loss": 0.0168,
"step": 16650
},
{
"grad_norm": 0.20839174091815948,
"learning_rate": 7.963218755257064e-05,
"loss": 0.012,
"step": 16660
},
{
"grad_norm": 0.20110207796096802,
"learning_rate": 7.960554484200807e-05,
"loss": 0.0152,
"step": 16670
},
{
"grad_norm": 0.14714448153972626,
"learning_rate": 7.957888918097436e-05,
"loss": 0.0119,
"step": 16680
},
{
"grad_norm": 0.19018179178237915,
"learning_rate": 7.955222058112962e-05,
"loss": 0.0108,
"step": 16690
},
{
"grad_norm": 0.18157163262367249,
"learning_rate": 7.952553905413955e-05,
"loss": 0.0113,
"step": 16700
},
{
"grad_norm": 0.16330525279045105,
"learning_rate": 7.949884461167563e-05,
"loss": 0.0137,
"step": 16710
},
{
"grad_norm": 0.13949407637119293,
"learning_rate": 7.947213726541484e-05,
"loss": 0.0113,
"step": 16720
},
{
"grad_norm": 0.26493871212005615,
"learning_rate": 7.944541702703996e-05,
"loss": 0.0126,
"step": 16730
},
{
"grad_norm": 0.18622152507305145,
"learning_rate": 7.941868390823928e-05,
"loss": 0.0163,
"step": 16740
},
{
"grad_norm": 0.20533102750778198,
"learning_rate": 7.939193792070677e-05,
"loss": 0.0134,
"step": 16750
},
{
"grad_norm": 0.18917490541934967,
"learning_rate": 7.936517907614207e-05,
"loss": 0.0136,
"step": 16760
},
{
"grad_norm": 0.2670118510723114,
"learning_rate": 7.933840738625036e-05,
"loss": 0.0144,
"step": 16770
},
{
"grad_norm": 0.23025615513324738,
"learning_rate": 7.931162286274252e-05,
"loss": 0.0144,
"step": 16780
},
{
"grad_norm": 0.1937585324048996,
"learning_rate": 7.928482551733503e-05,
"loss": 0.0126,
"step": 16790
},
{
"grad_norm": 0.17140038311481476,
"learning_rate": 7.925801536174991e-05,
"loss": 0.0126,
"step": 16800
},
{
"grad_norm": 0.15472757816314697,
"learning_rate": 7.923119240771484e-05,
"loss": 0.0095,
"step": 16810
},
{
"grad_norm": 0.2222132533788681,
"learning_rate": 7.920435666696313e-05,
"loss": 0.0124,
"step": 16820
},
{
"grad_norm": 0.21090425550937653,
"learning_rate": 7.917750815123362e-05,
"loss": 0.0128,
"step": 16830
},
{
"grad_norm": 0.16505002975463867,
"learning_rate": 7.915064687227075e-05,
"loss": 0.0119,
"step": 16840
},
{
"grad_norm": 0.16034339368343353,
"learning_rate": 7.912377284182458e-05,
"loss": 0.0143,
"step": 16850
},
{
"grad_norm": 0.2239542007446289,
"learning_rate": 7.90968860716507e-05,
"loss": 0.012,
"step": 16860
},
{
"grad_norm": 0.19641442596912384,
"learning_rate": 7.906998657351033e-05,
"loss": 0.0123,
"step": 16870
},
{
"grad_norm": 0.17314167320728302,
"learning_rate": 7.90430743591702e-05,
"loss": 0.0131,
"step": 16880
},
{
"grad_norm": 0.19985292851924896,
"learning_rate": 7.901614944040263e-05,
"loss": 0.0121,
"step": 16890
},
{
"grad_norm": 0.15974561870098114,
"learning_rate": 7.89892118289855e-05,
"loss": 0.0132,
"step": 16900
},
{
"grad_norm": 0.17452605068683624,
"learning_rate": 7.896226153670223e-05,
"loss": 0.0114,
"step": 16910
},
{
"grad_norm": 0.2433462291955948,
"learning_rate": 7.89352985753418e-05,
"loss": 0.0168,
"step": 16920
},
{
"grad_norm": 0.17155568301677704,
"learning_rate": 7.890832295669871e-05,
"loss": 0.0116,
"step": 16930
},
{
"grad_norm": 0.15582984685897827,
"learning_rate": 7.888133469257303e-05,
"loss": 0.0126,
"step": 16940
},
{
"grad_norm": 0.1882326751947403,
"learning_rate": 7.885433379477034e-05,
"loss": 0.0145,
"step": 16950
},
{
"grad_norm": 0.16054409742355347,
"learning_rate": 7.882732027510174e-05,
"loss": 0.0147,
"step": 16960
},
{
"grad_norm": 0.17706015706062317,
"learning_rate": 7.880029414538386e-05,
"loss": 0.0147,
"step": 16970
},
{
"grad_norm": 0.17251335084438324,
"learning_rate": 7.877325541743888e-05,
"loss": 0.0109,
"step": 16980
},
{
"grad_norm": 0.22736628353595734,
"learning_rate": 7.874620410309443e-05,
"loss": 0.0117,
"step": 16990
},
{
"grad_norm": 0.1712624579668045,
"learning_rate": 7.871914021418368e-05,
"loss": 0.0141,
"step": 17000
},
{
"grad_norm": 0.2757171094417572,
"learning_rate": 7.869206376254528e-05,
"loss": 0.0186,
"step": 17010
},
{
"grad_norm": 0.18450172245502472,
"learning_rate": 7.86649747600234e-05,
"loss": 0.0152,
"step": 17020
},
{
"grad_norm": 0.19964875280857086,
"learning_rate": 7.86378732184677e-05,
"loss": 0.0124,
"step": 17030
},
{
"grad_norm": 0.21817098557949066,
"learning_rate": 7.86107591497333e-05,
"loss": 0.0144,
"step": 17040
},
{
"grad_norm": 0.182329922914505,
"learning_rate": 7.858363256568083e-05,
"loss": 0.0139,
"step": 17050
},
{
"grad_norm": 0.1959627866744995,
"learning_rate": 7.855649347817637e-05,
"loss": 0.0147,
"step": 17060
},
{
"grad_norm": 0.1999945044517517,
"learning_rate": 7.85293418990915e-05,
"loss": 0.0138,
"step": 17070
},
{
"grad_norm": 0.16402937471866608,
"learning_rate": 7.85021778403032e-05,
"loss": 0.0111,
"step": 17080
},
{
"grad_norm": 0.19569021463394165,
"learning_rate": 7.847500131369397e-05,
"loss": 0.0142,
"step": 17090
},
{
"grad_norm": 0.155470609664917,
"learning_rate": 7.844781233115176e-05,
"loss": 0.0106,
"step": 17100
},
{
"grad_norm": 0.2566491961479187,
"learning_rate": 7.842061090456996e-05,
"loss": 0.0105,
"step": 17110
},
{
"grad_norm": 0.2471739500761032,
"learning_rate": 7.839339704584736e-05,
"loss": 0.0132,
"step": 17120
},
{
"grad_norm": 0.19571788609027863,
"learning_rate": 7.836617076688828e-05,
"loss": 0.0123,
"step": 17130
},
{
"grad_norm": 0.2076190710067749,
"learning_rate": 7.833893207960236e-05,
"loss": 0.0144,
"step": 17140
},
{
"grad_norm": 0.1908150315284729,
"learning_rate": 7.831168099590478e-05,
"loss": 0.0133,
"step": 17150
},
{
"grad_norm": 0.17001797258853912,
"learning_rate": 7.828441752771607e-05,
"loss": 0.0141,
"step": 17160
},
{
"grad_norm": 0.149556964635849,
"learning_rate": 7.825714168696219e-05,
"loss": 0.0109,
"step": 17170
},
{
"grad_norm": 0.1783851683139801,
"learning_rate": 7.82298534855745e-05,
"loss": 0.0124,
"step": 17180
},
{
"grad_norm": 0.20724186301231384,
"learning_rate": 7.820255293548984e-05,
"loss": 0.0164,
"step": 17190
},
{
"grad_norm": 0.19871698319911957,
"learning_rate": 7.817524004865035e-05,
"loss": 0.0113,
"step": 17200
},
{
"grad_norm": 0.13494695723056793,
"learning_rate": 7.814791483700366e-05,
"loss": 0.0098,
"step": 17210
},
{
"grad_norm": 0.20947898924350739,
"learning_rate": 7.81205773125027e-05,
"loss": 0.0129,
"step": 17220
},
{
"grad_norm": 0.1666402667760849,
"learning_rate": 7.809322748710585e-05,
"loss": 0.0144,
"step": 17230
},
{
"grad_norm": 0.1581578403711319,
"learning_rate": 7.806586537277685e-05,
"loss": 0.0146,
"step": 17240
},
{
"grad_norm": 0.14638042449951172,
"learning_rate": 7.803849098148484e-05,
"loss": 0.0119,
"step": 17250
},
{
"grad_norm": 0.18746882677078247,
"learning_rate": 7.801110432520427e-05,
"loss": 0.0131,
"step": 17260
},
{
"grad_norm": 0.15351933240890503,
"learning_rate": 7.798370541591504e-05,
"loss": 0.0136,
"step": 17270
},
{
"grad_norm": 0.17540200054645538,
"learning_rate": 7.795629426560232e-05,
"loss": 0.0124,
"step": 17280
},
{
"grad_norm": 0.1643502414226532,
"learning_rate": 7.792887088625671e-05,
"loss": 0.0108,
"step": 17290
},
{
"grad_norm": 0.20626762509346008,
"learning_rate": 7.790143528987412e-05,
"loss": 0.0144,
"step": 17300
},
{
"grad_norm": 0.18912595510482788,
"learning_rate": 7.787398748845578e-05,
"loss": 0.0109,
"step": 17310
},
{
"grad_norm": 0.1651771068572998,
"learning_rate": 7.784652749400834e-05,
"loss": 0.012,
"step": 17320
},
{
"grad_norm": 0.278839647769928,
"learning_rate": 7.78190553185437e-05,
"loss": 0.0151,
"step": 17330
},
{
"grad_norm": 0.2048417180776596,
"learning_rate": 7.779157097407915e-05,
"loss": 0.0129,
"step": 17340
},
{
"grad_norm": 0.21829654276371002,
"learning_rate": 7.776407447263725e-05,
"loss": 0.0152,
"step": 17350
},
{
"grad_norm": 0.168159618973732,
"learning_rate": 7.773656582624593e-05,
"loss": 0.0124,
"step": 17360
},
{
"grad_norm": 0.18229460716247559,
"learning_rate": 7.770904504693837e-05,
"loss": 0.0121,
"step": 17370
},
{
"grad_norm": 0.15205134451389313,
"learning_rate": 7.768151214675314e-05,
"loss": 0.013,
"step": 17380
},
{
"grad_norm": 0.18675142526626587,
"learning_rate": 7.765396713773402e-05,
"loss": 0.0129,
"step": 17390
},
{
"grad_norm": 0.2199348658323288,
"learning_rate": 7.762641003193017e-05,
"loss": 0.0156,
"step": 17400
},
{
"grad_norm": 0.19740426540374756,
"learning_rate": 7.759884084139597e-05,
"loss": 0.0128,
"step": 17410
},
{
"grad_norm": 0.23451554775238037,
"learning_rate": 7.757125957819116e-05,
"loss": 0.0118,
"step": 17420
},
{
"grad_norm": 0.2191203534603119,
"learning_rate": 7.754366625438069e-05,
"loss": 0.0118,
"step": 17430
},
{
"grad_norm": 0.2032255083322525,
"learning_rate": 7.751606088203483e-05,
"loss": 0.0131,
"step": 17440
},
{
"grad_norm": 0.2743101418018341,
"learning_rate": 7.74884434732291e-05,
"loss": 0.0138,
"step": 17450
},
{
"grad_norm": 0.1422770917415619,
"learning_rate": 7.746081404004431e-05,
"loss": 0.0114,
"step": 17460
},
{
"grad_norm": 0.15598182380199432,
"learning_rate": 7.743317259456649e-05,
"loss": 0.0132,
"step": 17470
},
{
"grad_norm": 0.24664343893527985,
"learning_rate": 7.740551914888696e-05,
"loss": 0.0133,
"step": 17480
},
{
"grad_norm": 0.13909658789634705,
"learning_rate": 7.737785371510229e-05,
"loss": 0.0124,
"step": 17490
},
{
"grad_norm": 0.1832265853881836,
"learning_rate": 7.735017630531426e-05,
"loss": 0.0132,
"step": 17500
},
{
"grad_norm": 0.24958227574825287,
"learning_rate": 7.732248693162991e-05,
"loss": 0.0141,
"step": 17510
},
{
"grad_norm": 0.18425600230693817,
"learning_rate": 7.729478560616153e-05,
"loss": 0.01,
"step": 17520
},
{
"grad_norm": 0.21296411752700806,
"learning_rate": 7.726707234102659e-05,
"loss": 0.0134,
"step": 17530
},
{
"grad_norm": 0.2786808907985687,
"learning_rate": 7.723934714834785e-05,
"loss": 0.0129,
"step": 17540
},
{
"grad_norm": 0.16840443015098572,
"learning_rate": 7.721161004025323e-05,
"loss": 0.0118,
"step": 17550
},
{
"grad_norm": 0.2019394338130951,
"learning_rate": 7.718386102887588e-05,
"loss": 0.013,
"step": 17560
},
{
"grad_norm": 0.1984574794769287,
"learning_rate": 7.715610012635418e-05,
"loss": 0.0144,
"step": 17570
},
{
"grad_norm": 0.19724710285663605,
"learning_rate": 7.712832734483166e-05,
"loss": 0.0134,
"step": 17580
},
{
"grad_norm": 0.17074300348758698,
"learning_rate": 7.710054269645713e-05,
"loss": 0.0133,
"step": 17590
},
{
"grad_norm": 0.19136108458042145,
"learning_rate": 7.707274619338449e-05,
"loss": 0.0111,
"step": 17600
},
{
"grad_norm": 0.22624105215072632,
"learning_rate": 7.704493784777289e-05,
"loss": 0.0123,
"step": 17610
},
{
"grad_norm": 0.14387543499469757,
"learning_rate": 7.701711767178665e-05,
"loss": 0.0115,
"step": 17620
},
{
"grad_norm": 0.20454372465610504,
"learning_rate": 7.698928567759525e-05,
"loss": 0.0137,
"step": 17630
},
{
"grad_norm": 0.182696133852005,
"learning_rate": 7.696144187737335e-05,
"loss": 0.0125,
"step": 17640
},
{
"grad_norm": 0.20369169116020203,
"learning_rate": 7.69335862833008e-05,
"loss": 0.0093,
"step": 17650
},
{
"grad_norm": 0.15515895187854767,
"learning_rate": 7.690571890756256e-05,
"loss": 0.0097,
"step": 17660
},
{
"grad_norm": 0.2032262086868286,
"learning_rate": 7.687783976234877e-05,
"loss": 0.012,
"step": 17670
},
{
"grad_norm": 0.18972934782505035,
"learning_rate": 7.684994885985472e-05,
"loss": 0.0155,
"step": 17680
},
{
"grad_norm": 0.1714819371700287,
"learning_rate": 7.682204621228083e-05,
"loss": 0.0119,
"step": 17690
},
{
"grad_norm": 0.18884821236133575,
"learning_rate": 7.679413183183267e-05,
"loss": 0.0139,
"step": 17700
},
{
"grad_norm": 0.14690712094306946,
"learning_rate": 7.676620573072093e-05,
"loss": 0.0117,
"step": 17710
},
{
"grad_norm": 0.15074053406715393,
"learning_rate": 7.673826792116145e-05,
"loss": 0.0144,
"step": 17720
},
{
"grad_norm": 0.15411476790905,
"learning_rate": 7.671031841537519e-05,
"loss": 0.0132,
"step": 17730
},
{
"grad_norm": 0.17561569809913635,
"learning_rate": 7.668235722558819e-05,
"loss": 0.0164,
"step": 17740
},
{
"grad_norm": 0.2420274317264557,
"learning_rate": 7.665438436403161e-05,
"loss": 0.0118,
"step": 17750
},
{
"grad_norm": 0.17214705049991608,
"learning_rate": 7.662639984294178e-05,
"loss": 0.018,
"step": 17760
},
{
"grad_norm": 0.16211435198783875,
"learning_rate": 7.659840367456002e-05,
"loss": 0.0129,
"step": 17770
},
{
"grad_norm": 0.12783071398735046,
"learning_rate": 7.657039587113287e-05,
"loss": 0.012,
"step": 17780
},
{
"grad_norm": 0.15509523451328278,
"learning_rate": 7.654237644491182e-05,
"loss": 0.0137,
"step": 17790
},
{
"grad_norm": 0.17439597845077515,
"learning_rate": 7.651434540815358e-05,
"loss": 0.0117,
"step": 17800
},
{
"grad_norm": 0.16544301807880402,
"learning_rate": 7.648630277311986e-05,
"loss": 0.0148,
"step": 17810
},
{
"grad_norm": 0.15848210453987122,
"learning_rate": 7.645824855207744e-05,
"loss": 0.0131,
"step": 17820
},
{
"grad_norm": 0.17910870909690857,
"learning_rate": 7.643018275729821e-05,
"loss": 0.0099,
"step": 17830
},
{
"grad_norm": 0.17907653748989105,
"learning_rate": 7.640210540105911e-05,
"loss": 0.0131,
"step": 17840
},
{
"grad_norm": 0.16509029269218445,
"learning_rate": 7.637401649564213e-05,
"loss": 0.0118,
"step": 17850
},
{
"grad_norm": 0.21781806647777557,
"learning_rate": 7.63459160533343e-05,
"loss": 0.0133,
"step": 17860
},
{
"grad_norm": 0.241594135761261,
"learning_rate": 7.631780408642769e-05,
"loss": 0.0153,
"step": 17870
},
{
"grad_norm": 0.2747977375984192,
"learning_rate": 7.628968060721946e-05,
"loss": 0.0118,
"step": 17880
},
{
"grad_norm": 0.27169549465179443,
"learning_rate": 7.626154562801176e-05,
"loss": 0.0153,
"step": 17890
},
{
"grad_norm": 0.17441566288471222,
"learning_rate": 7.623339916111177e-05,
"loss": 0.0137,
"step": 17900
},
{
"grad_norm": 0.19402961432933807,
"learning_rate": 7.620524121883174e-05,
"loss": 0.0152,
"step": 17910
},
{
"grad_norm": 0.1799282729625702,
"learning_rate": 7.617707181348892e-05,
"loss": 0.0155,
"step": 17920
},
{
"grad_norm": 0.18157483637332916,
"learning_rate": 7.614889095740554e-05,
"loss": 0.0116,
"step": 17930
},
{
"grad_norm": 0.18750829994678497,
"learning_rate": 7.612069866290885e-05,
"loss": 0.0138,
"step": 17940
},
{
"grad_norm": 0.20470339059829712,
"learning_rate": 7.609249494233115e-05,
"loss": 0.0103,
"step": 17950
},
{
"grad_norm": 0.2344280332326889,
"learning_rate": 7.60642798080097e-05,
"loss": 0.0152,
"step": 17960
},
{
"grad_norm": 0.19096671044826508,
"learning_rate": 7.603605327228673e-05,
"loss": 0.0133,
"step": 17970
},
{
"grad_norm": 0.2365921437740326,
"learning_rate": 7.600781534750953e-05,
"loss": 0.0124,
"step": 17980
},
{
"grad_norm": 0.12231196463108063,
"learning_rate": 7.597956604603029e-05,
"loss": 0.0138,
"step": 17990
},
{
"grad_norm": 0.2049366980791092,
"learning_rate": 7.595130538020622e-05,
"loss": 0.0115,
"step": 18000
},
{
"grad_norm": 0.19563253223896027,
"learning_rate": 7.592303336239952e-05,
"loss": 0.0133,
"step": 18010
},
{
"grad_norm": 0.1886538416147232,
"learning_rate": 7.589475000497733e-05,
"loss": 0.0112,
"step": 18020
},
{
"grad_norm": 0.1753348559141159,
"learning_rate": 7.586645532031171e-05,
"loss": 0.0109,
"step": 18030
},
{
"grad_norm": 0.1693079024553299,
"learning_rate": 7.583814932077975e-05,
"loss": 0.0112,
"step": 18040
},
{
"grad_norm": 0.14030446112155914,
"learning_rate": 7.580983201876343e-05,
"loss": 0.0124,
"step": 18050
},
{
"grad_norm": 0.20239360630512238,
"learning_rate": 7.578150342664973e-05,
"loss": 0.0139,
"step": 18060
},
{
"grad_norm": 0.24271516501903534,
"learning_rate": 7.575316355683055e-05,
"loss": 0.0145,
"step": 18070
},
{
"grad_norm": 0.20339468121528625,
"learning_rate": 7.572481242170266e-05,
"loss": 0.0125,
"step": 18080
},
{
"grad_norm": 0.18271459639072418,
"learning_rate": 7.569645003366786e-05,
"loss": 0.0137,
"step": 18090
},
{
"grad_norm": 0.19255180656909943,
"learning_rate": 7.566807640513278e-05,
"loss": 0.0143,
"step": 18100
},
{
"grad_norm": 0.17189113795757294,
"learning_rate": 7.563969154850905e-05,
"loss": 0.0112,
"step": 18110
},
{
"grad_norm": 0.22535055875778198,
"learning_rate": 7.561129547621314e-05,
"loss": 0.0176,
"step": 18120
},
{
"grad_norm": 0.18858055770397186,
"learning_rate": 7.558288820066648e-05,
"loss": 0.0141,
"step": 18130
},
{
"grad_norm": 0.16816578805446625,
"learning_rate": 7.555446973429534e-05,
"loss": 0.0111,
"step": 18140
},
{
"grad_norm": 0.15724124014377594,
"learning_rate": 7.552604008953097e-05,
"loss": 0.0096,
"step": 18150
},
{
"grad_norm": 0.21814119815826416,
"learning_rate": 7.549759927880942e-05,
"loss": 0.0123,
"step": 18160
},
{
"grad_norm": 0.1996580958366394,
"learning_rate": 7.54691473145717e-05,
"loss": 0.0122,
"step": 18170
},
{
"grad_norm": 0.2246442437171936,
"learning_rate": 7.544068420926365e-05,
"loss": 0.012,
"step": 18180
},
{
"grad_norm": 0.14493007957935333,
"learning_rate": 7.5412209975336e-05,
"loss": 0.0128,
"step": 18190
},
{
"grad_norm": 0.22834785282611847,
"learning_rate": 7.538372462524435e-05,
"loss": 0.0123,
"step": 18200
},
{
"grad_norm": 0.16393938660621643,
"learning_rate": 7.535522817144915e-05,
"loss": 0.0106,
"step": 18210
},
{
"grad_norm": 0.1759633868932724,
"learning_rate": 7.532672062641575e-05,
"loss": 0.0121,
"step": 18220
},
{
"grad_norm": 0.24654430150985718,
"learning_rate": 7.529820200261427e-05,
"loss": 0.0138,
"step": 18230
},
{
"grad_norm": 0.16910642385482788,
"learning_rate": 7.526967231251977e-05,
"loss": 0.013,
"step": 18240
},
{
"grad_norm": 0.15669231116771698,
"learning_rate": 7.524113156861206e-05,
"loss": 0.0124,
"step": 18250
},
{
"grad_norm": 0.15529100596904755,
"learning_rate": 7.521257978337587e-05,
"loss": 0.0116,
"step": 18260
},
{
"grad_norm": 0.16490167379379272,
"learning_rate": 7.518401696930072e-05,
"loss": 0.013,
"step": 18270
},
{
"grad_norm": 0.16230081021785736,
"learning_rate": 7.515544313888092e-05,
"loss": 0.0126,
"step": 18280
},
{
"grad_norm": 0.15758302807807922,
"learning_rate": 7.512685830461568e-05,
"loss": 0.0105,
"step": 18290
},
{
"grad_norm": 0.17590612173080444,
"learning_rate": 7.509826247900892e-05,
"loss": 0.0107,
"step": 18300
},
{
"grad_norm": 0.21104857325553894,
"learning_rate": 7.50696556745695e-05,
"loss": 0.0167,
"step": 18310
},
{
"grad_norm": 0.19481642544269562,
"learning_rate": 7.504103790381095e-05,
"loss": 0.0109,
"step": 18320
},
{
"grad_norm": 0.17435726523399353,
"learning_rate": 7.501240917925167e-05,
"loss": 0.0157,
"step": 18330
},
{
"grad_norm": 0.1838396042585373,
"learning_rate": 7.498376951341483e-05,
"loss": 0.0128,
"step": 18340
},
{
"grad_norm": 0.18683254718780518,
"learning_rate": 7.495511891882844e-05,
"loss": 0.0126,
"step": 18350
},
{
"grad_norm": 0.20800252258777618,
"learning_rate": 7.492645740802517e-05,
"loss": 0.0117,
"step": 18360
},
{
"grad_norm": 0.21343742311000824,
"learning_rate": 7.489778499354259e-05,
"loss": 0.0111,
"step": 18370
},
{
"grad_norm": 0.1595340371131897,
"learning_rate": 7.486910168792298e-05,
"loss": 0.0125,
"step": 18380
},
{
"grad_norm": 0.19294749200344086,
"learning_rate": 7.484040750371337e-05,
"loss": 0.01,
"step": 18390
},
{
"grad_norm": 0.17441585659980774,
"learning_rate": 7.48117024534656e-05,
"loss": 0.011,
"step": 18400
},
{
"grad_norm": 0.18727931380271912,
"learning_rate": 7.478298654973621e-05,
"loss": 0.0116,
"step": 18410
},
{
"grad_norm": 0.19633245468139648,
"learning_rate": 7.475425980508654e-05,
"loss": 0.0114,
"step": 18420
},
{
"grad_norm": 0.1921997219324112,
"learning_rate": 7.47255222320826e-05,
"loss": 0.0131,
"step": 18430
},
{
"grad_norm": 0.2119644582271576,
"learning_rate": 7.469677384329522e-05,
"loss": 0.0143,
"step": 18440
},
{
"grad_norm": 0.1979852169752121,
"learning_rate": 7.46680146512999e-05,
"loss": 0.0136,
"step": 18450
},
{
"grad_norm": 0.24530205130577087,
"learning_rate": 7.463924466867692e-05,
"loss": 0.011,
"step": 18460
},
{
"grad_norm": 0.1790321320295334,
"learning_rate": 7.461046390801121e-05,
"loss": 0.0114,
"step": 18470
},
{
"grad_norm": 0.18571199476718903,
"learning_rate": 7.458167238189248e-05,
"loss": 0.0108,
"step": 18480
},
{
"grad_norm": 0.2013978362083435,
"learning_rate": 7.455287010291509e-05,
"loss": 0.0149,
"step": 18490
},
{
"grad_norm": 0.19386424124240875,
"learning_rate": 7.452405708367818e-05,
"loss": 0.0121,
"step": 18500
},
{
"grad_norm": 0.21166321635246277,
"learning_rate": 7.449523333678549e-05,
"loss": 0.0123,
"step": 18510
},
{
"grad_norm": 0.14023329317569733,
"learning_rate": 7.446639887484555e-05,
"loss": 0.0125,
"step": 18520
},
{
"grad_norm": 0.15196268260478973,
"learning_rate": 7.443755371047152e-05,
"loss": 0.0111,
"step": 18530
},
{
"grad_norm": 0.2037297934293747,
"learning_rate": 7.440869785628124e-05,
"loss": 0.013,
"step": 18540
},
{
"grad_norm": 0.19182361662387848,
"learning_rate": 7.437983132489727e-05,
"loss": 0.0141,
"step": 18550
},
{
"grad_norm": 0.1650853306055069,
"learning_rate": 7.435095412894678e-05,
"loss": 0.0122,
"step": 18560
},
{
"grad_norm": 0.1313398778438568,
"learning_rate": 7.432206628106165e-05,
"loss": 0.0109,
"step": 18570
},
{
"grad_norm": 0.14341309666633606,
"learning_rate": 7.429316779387842e-05,
"loss": 0.0094,
"step": 18580
},
{
"grad_norm": 0.15422917902469635,
"learning_rate": 7.426425868003825e-05,
"loss": 0.0113,
"step": 18590
},
{
"grad_norm": 0.16208846867084503,
"learning_rate": 7.423533895218698e-05,
"loss": 0.0107,
"step": 18600
},
{
"grad_norm": 0.16285823285579681,
"learning_rate": 7.420640862297507e-05,
"loss": 0.0131,
"step": 18610
},
{
"grad_norm": 0.16573621332645416,
"learning_rate": 7.417746770505766e-05,
"loss": 0.0139,
"step": 18620
},
{
"grad_norm": 0.21183575689792633,
"learning_rate": 7.414851621109444e-05,
"loss": 0.0138,
"step": 18630
},
{
"grad_norm": 0.1331263929605484,
"learning_rate": 7.411955415374982e-05,
"loss": 0.011,
"step": 18640
},
{
"grad_norm": 0.19629089534282684,
"learning_rate": 7.409058154569278e-05,
"loss": 0.0114,
"step": 18650
},
{
"grad_norm": 0.15443626046180725,
"learning_rate": 7.406159839959689e-05,
"loss": 0.0112,
"step": 18660
},
{
"grad_norm": 0.19833770394325256,
"learning_rate": 7.403260472814039e-05,
"loss": 0.0104,
"step": 18670
},
{
"grad_norm": 0.2044007033109665,
"learning_rate": 7.400360054400611e-05,
"loss": 0.0137,
"step": 18680
},
{
"grad_norm": 0.1870245635509491,
"learning_rate": 7.397458585988143e-05,
"loss": 0.0126,
"step": 18690
},
{
"grad_norm": 0.19683417677879333,
"learning_rate": 7.394556068845841e-05,
"loss": 0.0113,
"step": 18700
},
{
"grad_norm": 0.21472954750061035,
"learning_rate": 7.39165250424336e-05,
"loss": 0.0113,
"step": 18710
},
{
"grad_norm": 0.1248503252863884,
"learning_rate": 7.38874789345082e-05,
"loss": 0.01,
"step": 18720
},
{
"grad_norm": 0.18112768232822418,
"learning_rate": 7.385842237738794e-05,
"loss": 0.0137,
"step": 18730
},
{
"grad_norm": 0.21326950192451477,
"learning_rate": 7.382935538378318e-05,
"loss": 0.0108,
"step": 18740
},
{
"grad_norm": 0.1938982456922531,
"learning_rate": 7.38002779664088e-05,
"loss": 0.0112,
"step": 18750
},
{
"grad_norm": 0.15143512189388275,
"learning_rate": 7.377119013798422e-05,
"loss": 0.0086,
"step": 18760
},
{
"grad_norm": 0.21318207681179047,
"learning_rate": 7.374209191123349e-05,
"loss": 0.011,
"step": 18770
},
{
"grad_norm": 0.20381887257099152,
"learning_rate": 7.371298329888512e-05,
"loss": 0.0114,
"step": 18780
},
{
"grad_norm": 0.14822277426719666,
"learning_rate": 7.368386431367223e-05,
"loss": 0.0102,
"step": 18790
},
{
"grad_norm": 0.2085358202457428,
"learning_rate": 7.365473496833246e-05,
"loss": 0.0119,
"step": 18800
},
{
"grad_norm": 0.12606675922870636,
"learning_rate": 7.362559527560797e-05,
"loss": 0.0106,
"step": 18810
},
{
"grad_norm": 0.18155249953269958,
"learning_rate": 7.359644524824545e-05,
"loss": 0.0142,
"step": 18820
},
{
"grad_norm": 0.21405895054340363,
"learning_rate": 7.35672848989961e-05,
"loss": 0.0115,
"step": 18830
},
{
"grad_norm": 0.26266056299209595,
"learning_rate": 7.353811424061565e-05,
"loss": 0.0122,
"step": 18840
},
{
"grad_norm": 0.17924107611179352,
"learning_rate": 7.350893328586435e-05,
"loss": 0.0107,
"step": 18850
},
{
"grad_norm": 0.208436518907547,
"learning_rate": 7.347974204750696e-05,
"loss": 0.0106,
"step": 18860
},
{
"grad_norm": 0.14565600454807281,
"learning_rate": 7.345054053831266e-05,
"loss": 0.0097,
"step": 18870
},
{
"grad_norm": 0.18950042128562927,
"learning_rate": 7.342132877105523e-05,
"loss": 0.0121,
"step": 18880
},
{
"grad_norm": 0.165813609957695,
"learning_rate": 7.339210675851288e-05,
"loss": 0.0132,
"step": 18890
},
{
"grad_norm": 0.2116619497537613,
"learning_rate": 7.336287451346831e-05,
"loss": 0.0103,
"step": 18900
},
{
"grad_norm": 0.17606577277183533,
"learning_rate": 7.333363204870868e-05,
"loss": 0.0121,
"step": 18910
},
{
"grad_norm": 0.16686955094337463,
"learning_rate": 7.330437937702566e-05,
"loss": 0.0131,
"step": 18920
},
{
"grad_norm": 0.21528185904026031,
"learning_rate": 7.327511651121534e-05,
"loss": 0.0129,
"step": 18930
},
{
"grad_norm": 0.17814429104328156,
"learning_rate": 7.324584346407833e-05,
"loss": 0.0112,
"step": 18940
},
{
"grad_norm": 0.1568351686000824,
"learning_rate": 7.321656024841962e-05,
"loss": 0.0125,
"step": 18950
},
{
"grad_norm": 0.1489446759223938,
"learning_rate": 7.31872668770487e-05,
"loss": 0.0127,
"step": 18960
},
{
"grad_norm": 0.17591091990470886,
"learning_rate": 7.315796336277949e-05,
"loss": 0.0108,
"step": 18970
},
{
"grad_norm": 0.17621080577373505,
"learning_rate": 7.312864971843031e-05,
"loss": 0.0114,
"step": 18980
},
{
"grad_norm": 0.14916348457336426,
"learning_rate": 7.309932595682398e-05,
"loss": 0.0117,
"step": 18990
},
{
"grad_norm": 0.19298994541168213,
"learning_rate": 7.30699920907877e-05,
"loss": 0.0132,
"step": 19000
},
{
"grad_norm": 0.16541431844234467,
"learning_rate": 7.304064813315308e-05,
"loss": 0.0115,
"step": 19010
},
{
"grad_norm": 0.1469961255788803,
"learning_rate": 7.30112940967562e-05,
"loss": 0.0109,
"step": 19020
},
{
"grad_norm": 0.14893992245197296,
"learning_rate": 7.298192999443747e-05,
"loss": 0.0131,
"step": 19030
},
{
"grad_norm": 0.1689481884241104,
"learning_rate": 7.295255583904179e-05,
"loss": 0.0104,
"step": 19040
},
{
"grad_norm": 0.19867761433124542,
"learning_rate": 7.29231716434184e-05,
"loss": 0.0116,
"step": 19050
},
{
"grad_norm": 0.16729967296123505,
"learning_rate": 7.289377742042092e-05,
"loss": 0.0107,
"step": 19060
},
{
"grad_norm": 0.12557503581047058,
"learning_rate": 7.286437318290741e-05,
"loss": 0.0136,
"step": 19070
},
{
"grad_norm": 0.18714050948619843,
"learning_rate": 7.283495894374027e-05,
"loss": 0.0154,
"step": 19080
},
{
"grad_norm": 0.13715846836566925,
"learning_rate": 7.280553471578629e-05,
"loss": 0.0114,
"step": 19090
},
{
"grad_norm": 0.19969865679740906,
"learning_rate": 7.277610051191663e-05,
"loss": 0.0117,
"step": 19100
},
{
"grad_norm": 0.1655474454164505,
"learning_rate": 7.27466563450068e-05,
"loss": 0.011,
"step": 19110
},
{
"grad_norm": 0.18666283786296844,
"learning_rate": 7.271720222793668e-05,
"loss": 0.0129,
"step": 19120
},
{
"grad_norm": 0.18705947697162628,
"learning_rate": 7.268773817359054e-05,
"loss": 0.0107,
"step": 19130
},
{
"grad_norm": 0.19990399479866028,
"learning_rate": 7.265826419485689e-05,
"loss": 0.0111,
"step": 19140
},
{
"grad_norm": 0.1833992600440979,
"learning_rate": 7.26287803046287e-05,
"loss": 0.0125,
"step": 19150
},
{
"grad_norm": 0.2192019522190094,
"learning_rate": 7.259928651580322e-05,
"loss": 0.0106,
"step": 19160
},
{
"grad_norm": 0.2461443841457367,
"learning_rate": 7.256978284128202e-05,
"loss": 0.0119,
"step": 19170
},
{
"grad_norm": 0.22517727315425873,
"learning_rate": 7.254026929397101e-05,
"loss": 0.0137,
"step": 19180
},
{
"grad_norm": 0.15226878225803375,
"learning_rate": 7.251074588678044e-05,
"loss": 0.0124,
"step": 19190
},
{
"grad_norm": 0.14343887567520142,
"learning_rate": 7.248121263262482e-05,
"loss": 0.011,
"step": 19200
},
{
"grad_norm": 0.12526266276836395,
"learning_rate": 7.245166954442304e-05,
"loss": 0.0093,
"step": 19210
},
{
"grad_norm": 0.13337427377700806,
"learning_rate": 7.24221166350982e-05,
"loss": 0.0092,
"step": 19220
},
{
"grad_norm": 0.1772834062576294,
"learning_rate": 7.239255391757777e-05,
"loss": 0.0111,
"step": 19230
},
{
"grad_norm": 0.19824360311031342,
"learning_rate": 7.236298140479351e-05,
"loss": 0.0103,
"step": 19240
},
{
"grad_norm": 0.17644356191158295,
"learning_rate": 7.233339910968141e-05,
"loss": 0.0139,
"step": 19250
},
{
"grad_norm": 0.18103212118148804,
"learning_rate": 7.230380704518178e-05,
"loss": 0.0102,
"step": 19260
},
{
"grad_norm": 0.21373401582241058,
"learning_rate": 7.227420522423919e-05,
"loss": 0.0117,
"step": 19270
},
{
"grad_norm": 0.21635258197784424,
"learning_rate": 7.224459365980249e-05,
"loss": 0.0133,
"step": 19280
},
{
"grad_norm": 0.16442619264125824,
"learning_rate": 7.221497236482479e-05,
"loss": 0.0126,
"step": 19290
},
{
"grad_norm": 0.14004424214363098,
"learning_rate": 7.218534135226345e-05,
"loss": 0.0097,
"step": 19300
},
{
"grad_norm": 0.14560888707637787,
"learning_rate": 7.215570063508005e-05,
"loss": 0.0099,
"step": 19310
},
{
"grad_norm": 0.15902572870254517,
"learning_rate": 7.21260502262405e-05,
"loss": 0.0122,
"step": 19320
},
{
"grad_norm": 0.13771773874759674,
"learning_rate": 7.209639013871487e-05,
"loss": 0.0133,
"step": 19330
},
{
"grad_norm": 0.15242278575897217,
"learning_rate": 7.206672038547746e-05,
"loss": 0.0119,
"step": 19340
},
{
"grad_norm": 0.15112042427062988,
"learning_rate": 7.203704097950687e-05,
"loss": 0.0102,
"step": 19350
},
{
"grad_norm": 0.18977764248847961,
"learning_rate": 7.200735193378587e-05,
"loss": 0.0112,
"step": 19360
},
{
"grad_norm": 0.1985086053609848,
"learning_rate": 7.197765326130145e-05,
"loss": 0.0103,
"step": 19370
},
{
"grad_norm": 0.17211055755615234,
"learning_rate": 7.194794497504481e-05,
"loss": 0.0108,
"step": 19380
},
{
"grad_norm": 0.16687899827957153,
"learning_rate": 7.19182270880114e-05,
"loss": 0.0126,
"step": 19390
},
{
"grad_norm": 0.18941353261470795,
"learning_rate": 7.188849961320079e-05,
"loss": 0.0116,
"step": 19400
},
{
"grad_norm": 0.18274323642253876,
"learning_rate": 7.185876256361681e-05,
"loss": 0.013,
"step": 19410
},
{
"grad_norm": 0.14908260107040405,
"learning_rate": 7.182901595226745e-05,
"loss": 0.0099,
"step": 19420
},
{
"grad_norm": 0.1887970268726349,
"learning_rate": 7.179925979216491e-05,
"loss": 0.0104,
"step": 19430
},
{
"grad_norm": 0.18476919829845428,
"learning_rate": 7.17694940963255e-05,
"loss": 0.0107,
"step": 19440
},
{
"grad_norm": 0.19148029386997223,
"learning_rate": 7.17397188777698e-05,
"loss": 0.012,
"step": 19450
},
{
"grad_norm": 0.23468837141990662,
"learning_rate": 7.170993414952245e-05,
"loss": 0.0102,
"step": 19460
},
{
"grad_norm": 0.22746720910072327,
"learning_rate": 7.168013992461233e-05,
"loss": 0.0131,
"step": 19470
},
{
"grad_norm": 0.21804428100585938,
"learning_rate": 7.165033621607245e-05,
"loss": 0.0138,
"step": 19480
},
{
"grad_norm": 0.20864491164684296,
"learning_rate": 7.162052303693995e-05,
"loss": 0.0107,
"step": 19490
},
{
"grad_norm": 0.1613885462284088,
"learning_rate": 7.159070040025614e-05,
"loss": 0.0098,
"step": 19500
},
{
"grad_norm": 0.21238669753074646,
"learning_rate": 7.156086831906646e-05,
"loss": 0.0137,
"step": 19510
},
{
"grad_norm": 0.19478407502174377,
"learning_rate": 7.153102680642044e-05,
"loss": 0.0103,
"step": 19520
},
{
"grad_norm": 0.17616073787212372,
"learning_rate": 7.150117587537182e-05,
"loss": 0.0117,
"step": 19530
},
{
"grad_norm": 0.13292518258094788,
"learning_rate": 7.147131553897838e-05,
"loss": 0.0105,
"step": 19540
},
{
"grad_norm": 0.26289835572242737,
"learning_rate": 7.144144581030205e-05,
"loss": 0.0129,
"step": 19550
},
{
"grad_norm": 0.2064499855041504,
"learning_rate": 7.141156670240889e-05,
"loss": 0.0115,
"step": 19560
},
{
"grad_norm": 0.17051909863948822,
"learning_rate": 7.138167822836899e-05,
"loss": 0.0133,
"step": 19570
},
{
"grad_norm": 0.20623430609703064,
"learning_rate": 7.135178040125661e-05,
"loss": 0.0095,
"step": 19580
},
{
"grad_norm": 0.21441036462783813,
"learning_rate": 7.13218732341501e-05,
"loss": 0.0123,
"step": 19590
},
{
"grad_norm": 0.22051948308944702,
"learning_rate": 7.129195674013181e-05,
"loss": 0.0136,
"step": 19600
},
{
"grad_norm": 0.1402568817138672,
"learning_rate": 7.126203093228829e-05,
"loss": 0.0103,
"step": 19610
},
{
"grad_norm": 0.17667120695114136,
"learning_rate": 7.123209582371006e-05,
"loss": 0.0118,
"step": 19620
},
{
"grad_norm": 0.1691339612007141,
"learning_rate": 7.120215142749179e-05,
"loss": 0.0098,
"step": 19630
},
{
"grad_norm": 0.16157539188861847,
"learning_rate": 7.117219775673218e-05,
"loss": 0.0127,
"step": 19640
},
{
"grad_norm": 0.17221374809741974,
"learning_rate": 7.114223482453394e-05,
"loss": 0.0104,
"step": 19650
},
{
"grad_norm": 0.2151339054107666,
"learning_rate": 7.11122626440039e-05,
"loss": 0.0107,
"step": 19660
},
{
"grad_norm": 0.20373307168483734,
"learning_rate": 7.10822812282529e-05,
"loss": 0.0109,
"step": 19670
},
{
"grad_norm": 0.18632051348686218,
"learning_rate": 7.105229059039586e-05,
"loss": 0.0121,
"step": 19680
},
{
"grad_norm": 0.1845545917749405,
"learning_rate": 7.102229074355168e-05,
"loss": 0.01,
"step": 19690
},
{
"grad_norm": 0.15541477501392365,
"learning_rate": 7.099228170084332e-05,
"loss": 0.0088,
"step": 19700
},
{
"grad_norm": 0.18808230757713318,
"learning_rate": 7.096226347539771e-05,
"loss": 0.0123,
"step": 19710
},
{
"grad_norm": 0.1887749433517456,
"learning_rate": 7.093223608034592e-05,
"loss": 0.0097,
"step": 19720
},
{
"grad_norm": 0.20691046118736267,
"learning_rate": 7.090219952882291e-05,
"loss": 0.0146,
"step": 19730
},
{
"grad_norm": 0.19972388446331024,
"learning_rate": 7.087215383396768e-05,
"loss": 0.0112,
"step": 19740
},
{
"grad_norm": 0.20749326050281525,
"learning_rate": 7.084209900892325e-05,
"loss": 0.0108,
"step": 19750
},
{
"grad_norm": 0.2301788628101349,
"learning_rate": 7.081203506683663e-05,
"loss": 0.0102,
"step": 19760
},
{
"grad_norm": 0.221941739320755,
"learning_rate": 7.078196202085879e-05,
"loss": 0.0139,
"step": 19770
},
{
"grad_norm": 0.19017480313777924,
"learning_rate": 7.075187988414473e-05,
"loss": 0.0106,
"step": 19780
},
{
"grad_norm": 0.17279358208179474,
"learning_rate": 7.072178866985334e-05,
"loss": 0.0112,
"step": 19790
},
{
"grad_norm": 0.15703575313091278,
"learning_rate": 7.069168839114764e-05,
"loss": 0.009,
"step": 19800
},
{
"grad_norm": 0.16718357801437378,
"learning_rate": 7.066157906119441e-05,
"loss": 0.011,
"step": 19810
},
{
"grad_norm": 0.18201936781406403,
"learning_rate": 7.063146069316456e-05,
"loss": 0.0121,
"step": 19820
},
{
"grad_norm": 0.2503196597099304,
"learning_rate": 7.060133330023287e-05,
"loss": 0.0133,
"step": 19830
},
{
"grad_norm": 0.18651476502418518,
"learning_rate": 7.057119689557807e-05,
"loss": 0.0136,
"step": 19840
},
{
"grad_norm": 0.1670781672000885,
"learning_rate": 7.054105149238287e-05,
"loss": 0.011,
"step": 19850
},
{
"grad_norm": 0.25149157643318176,
"learning_rate": 7.051089710383387e-05,
"loss": 0.0144,
"step": 19860
},
{
"grad_norm": 0.17136549949645996,
"learning_rate": 7.048073374312166e-05,
"loss": 0.0142,
"step": 19870
},
{
"grad_norm": 0.18975761532783508,
"learning_rate": 7.045056142344067e-05,
"loss": 0.0131,
"step": 19880
},
{
"grad_norm": 0.20178429782390594,
"learning_rate": 7.042038015798934e-05,
"loss": 0.0117,
"step": 19890
},
{
"grad_norm": 0.18146274983882904,
"learning_rate": 7.039018995996995e-05,
"loss": 0.0122,
"step": 19900
},
{
"grad_norm": 0.18776217103004456,
"learning_rate": 7.035999084258876e-05,
"loss": 0.0144,
"step": 19910
},
{
"grad_norm": 0.22703418135643005,
"learning_rate": 7.032978281905583e-05,
"loss": 0.0107,
"step": 19920
},
{
"grad_norm": 0.17703606188297272,
"learning_rate": 7.029956590258522e-05,
"loss": 0.0113,
"step": 19930
},
{
"grad_norm": 0.1451052576303482,
"learning_rate": 7.026934010639483e-05,
"loss": 0.0122,
"step": 19940
},
{
"grad_norm": 0.17386291921138763,
"learning_rate": 7.023910544370645e-05,
"loss": 0.0105,
"step": 19950
},
{
"grad_norm": 0.23453392088413239,
"learning_rate": 7.020886192774573e-05,
"loss": 0.0126,
"step": 19960
},
{
"grad_norm": 0.2455233484506607,
"learning_rate": 7.017860957174222e-05,
"loss": 0.014,
"step": 19970
},
{
"grad_norm": 0.15349385142326355,
"learning_rate": 7.014834838892934e-05,
"loss": 0.0094,
"step": 19980
},
{
"grad_norm": 0.13492955267429352,
"learning_rate": 7.011807839254435e-05,
"loss": 0.0097,
"step": 19990
},
{
"grad_norm": 0.2587132155895233,
"learning_rate": 7.008779959582837e-05,
"loss": 0.0127,
"step": 20000
}
],
"logging_steps": 10,
"max_steps": 50000,
"num_input_tokens_seen": 0,
"num_train_epochs": 9223372036854775807,
"save_steps": 10000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 48,
"trial_name": null,
"trial_params": null
}