PnPCubeLineChecks / checkpoint-22500 /trainer_state.json
LucaFrat's picture
Upload folder using huggingface_hub
fe20604 verified
Raw
History Blame Contribute Delete
315 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.75,
"eval_steps": 500,
"global_step": 22500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"grad_norm": 1.4126290082931519,
"learning_rate": 6.000000000000001e-07,
"loss": 1.4568,
"step": 10
},
{
"grad_norm": 1.3735394477844238,
"learning_rate": 1.2666666666666667e-06,
"loss": 1.4539,
"step": 20
},
{
"grad_norm": 1.1533366441726685,
"learning_rate": 1.9333333333333336e-06,
"loss": 1.4465,
"step": 30
},
{
"grad_norm": 0.8541414737701416,
"learning_rate": 2.6e-06,
"loss": 1.4158,
"step": 40
},
{
"grad_norm": 0.6532411575317383,
"learning_rate": 3.2666666666666666e-06,
"loss": 1.366,
"step": 50
},
{
"grad_norm": 0.43932828307151794,
"learning_rate": 3.9333333333333335e-06,
"loss": 1.3258,
"step": 60
},
{
"grad_norm": 0.4106330871582031,
"learning_rate": 4.6e-06,
"loss": 1.2879,
"step": 70
},
{
"grad_norm": 0.46380355954170227,
"learning_rate": 5.266666666666667e-06,
"loss": 1.2307,
"step": 80
},
{
"grad_norm": 0.39837679266929626,
"learning_rate": 5.933333333333334e-06,
"loss": 1.1701,
"step": 90
},
{
"grad_norm": 0.5506791472434998,
"learning_rate": 6.6e-06,
"loss": 1.1316,
"step": 100
},
{
"grad_norm": 0.4063026010990143,
"learning_rate": 7.266666666666668e-06,
"loss": 1.0984,
"step": 110
},
{
"grad_norm": 0.5370113849639893,
"learning_rate": 7.933333333333334e-06,
"loss": 1.0844,
"step": 120
},
{
"grad_norm": 0.517879068851471,
"learning_rate": 8.599999999999999e-06,
"loss": 1.0775,
"step": 130
},
{
"grad_norm": 0.4200724959373474,
"learning_rate": 9.266666666666667e-06,
"loss": 1.0754,
"step": 140
},
{
"grad_norm": 0.6313542127609253,
"learning_rate": 9.933333333333334e-06,
"loss": 1.0645,
"step": 150
},
{
"grad_norm": 0.5384342074394226,
"learning_rate": 1.06e-05,
"loss": 1.0539,
"step": 160
},
{
"grad_norm": 0.47652921080589294,
"learning_rate": 1.1266666666666667e-05,
"loss": 1.0473,
"step": 170
},
{
"grad_norm": 0.6397068500518799,
"learning_rate": 1.1933333333333333e-05,
"loss": 1.0471,
"step": 180
},
{
"grad_norm": 0.33853602409362793,
"learning_rate": 1.2600000000000001e-05,
"loss": 1.0428,
"step": 190
},
{
"grad_norm": 0.5751429200172424,
"learning_rate": 1.3266666666666666e-05,
"loss": 1.0328,
"step": 200
},
{
"grad_norm": 0.5356873273849487,
"learning_rate": 1.3933333333333334e-05,
"loss": 1.0211,
"step": 210
},
{
"grad_norm": 0.5052016973495483,
"learning_rate": 1.4599999999999999e-05,
"loss": 1.0104,
"step": 220
},
{
"grad_norm": 0.4714415967464447,
"learning_rate": 1.5266666666666667e-05,
"loss": 0.9949,
"step": 230
},
{
"grad_norm": 0.5818572640419006,
"learning_rate": 1.5933333333333332e-05,
"loss": 0.9771,
"step": 240
},
{
"grad_norm": 0.6291383504867554,
"learning_rate": 1.66e-05,
"loss": 0.9692,
"step": 250
},
{
"grad_norm": 0.43864965438842773,
"learning_rate": 1.726666666666667e-05,
"loss": 0.9473,
"step": 260
},
{
"grad_norm": 0.6750267744064331,
"learning_rate": 1.7933333333333337e-05,
"loss": 0.9305,
"step": 270
},
{
"grad_norm": 1.1174927949905396,
"learning_rate": 1.86e-05,
"loss": 0.9073,
"step": 280
},
{
"grad_norm": 0.7326529622077942,
"learning_rate": 1.926666666666667e-05,
"loss": 0.8804,
"step": 290
},
{
"grad_norm": 0.7365794777870178,
"learning_rate": 1.9933333333333334e-05,
"loss": 0.8393,
"step": 300
},
{
"grad_norm": 0.6675243377685547,
"learning_rate": 2.06e-05,
"loss": 0.8023,
"step": 310
},
{
"grad_norm": 0.926457405090332,
"learning_rate": 2.1266666666666667e-05,
"loss": 0.7797,
"step": 320
},
{
"grad_norm": 0.7966919541358948,
"learning_rate": 2.1933333333333332e-05,
"loss": 0.7535,
"step": 330
},
{
"grad_norm": 0.7804962992668152,
"learning_rate": 2.26e-05,
"loss": 0.7258,
"step": 340
},
{
"grad_norm": 0.9337165951728821,
"learning_rate": 2.326666666666667e-05,
"loss": 0.6942,
"step": 350
},
{
"grad_norm": 0.7417464852333069,
"learning_rate": 2.3933333333333337e-05,
"loss": 0.6635,
"step": 360
},
{
"grad_norm": 1.0197713375091553,
"learning_rate": 2.46e-05,
"loss": 0.6353,
"step": 370
},
{
"grad_norm": 1.0989516973495483,
"learning_rate": 2.5266666666666666e-05,
"loss": 0.6127,
"step": 380
},
{
"grad_norm": 1.1899223327636719,
"learning_rate": 2.5933333333333338e-05,
"loss": 0.5922,
"step": 390
},
{
"grad_norm": 0.9237262010574341,
"learning_rate": 2.6600000000000003e-05,
"loss": 0.5758,
"step": 400
},
{
"grad_norm": 1.4765948057174683,
"learning_rate": 2.7266666666666668e-05,
"loss": 0.5479,
"step": 410
},
{
"grad_norm": 0.9667250514030457,
"learning_rate": 2.7933333333333332e-05,
"loss": 0.5333,
"step": 420
},
{
"grad_norm": 0.9509170651435852,
"learning_rate": 2.86e-05,
"loss": 0.5143,
"step": 430
},
{
"grad_norm": 1.2793920040130615,
"learning_rate": 2.926666666666667e-05,
"loss": 0.4968,
"step": 440
},
{
"grad_norm": 0.9938004016876221,
"learning_rate": 2.9933333333333337e-05,
"loss": 0.4869,
"step": 450
},
{
"grad_norm": 0.9902573227882385,
"learning_rate": 3.06e-05,
"loss": 0.4659,
"step": 460
},
{
"grad_norm": 0.9396094679832458,
"learning_rate": 3.126666666666666e-05,
"loss": 0.4555,
"step": 470
},
{
"grad_norm": 1.2180593013763428,
"learning_rate": 3.1933333333333335e-05,
"loss": 0.4367,
"step": 480
},
{
"grad_norm": 1.2830370664596558,
"learning_rate": 3.26e-05,
"loss": 0.4122,
"step": 490
},
{
"grad_norm": 1.0384615659713745,
"learning_rate": 3.326666666666667e-05,
"loss": 0.3994,
"step": 500
},
{
"grad_norm": 1.0901414155960083,
"learning_rate": 3.3933333333333336e-05,
"loss": 0.3788,
"step": 510
},
{
"grad_norm": 1.0764013528823853,
"learning_rate": 3.46e-05,
"loss": 0.36,
"step": 520
},
{
"grad_norm": 1.283591628074646,
"learning_rate": 3.526666666666667e-05,
"loss": 0.3453,
"step": 530
},
{
"grad_norm": 1.4109214544296265,
"learning_rate": 3.593333333333334e-05,
"loss": 0.3376,
"step": 540
},
{
"grad_norm": 1.2783732414245605,
"learning_rate": 3.66e-05,
"loss": 0.3195,
"step": 550
},
{
"grad_norm": 1.2349673509597778,
"learning_rate": 3.726666666666667e-05,
"loss": 0.3045,
"step": 560
},
{
"grad_norm": 1.6351103782653809,
"learning_rate": 3.793333333333334e-05,
"loss": 0.297,
"step": 570
},
{
"grad_norm": 1.413041591644287,
"learning_rate": 3.86e-05,
"loss": 0.2832,
"step": 580
},
{
"grad_norm": 1.7158143520355225,
"learning_rate": 3.926666666666667e-05,
"loss": 0.2675,
"step": 590
},
{
"grad_norm": 1.5770541429519653,
"learning_rate": 3.993333333333333e-05,
"loss": 0.2511,
"step": 600
},
{
"grad_norm": 1.2448288202285767,
"learning_rate": 4.0600000000000004e-05,
"loss": 0.2385,
"step": 610
},
{
"grad_norm": 1.0363948345184326,
"learning_rate": 4.126666666666667e-05,
"loss": 0.2324,
"step": 620
},
{
"grad_norm": 1.3222676515579224,
"learning_rate": 4.1933333333333334e-05,
"loss": 0.2232,
"step": 630
},
{
"grad_norm": 1.592990756034851,
"learning_rate": 4.26e-05,
"loss": 0.2109,
"step": 640
},
{
"grad_norm": 1.5724996328353882,
"learning_rate": 4.3266666666666664e-05,
"loss": 0.1945,
"step": 650
},
{
"grad_norm": 1.4436169862747192,
"learning_rate": 4.3933333333333335e-05,
"loss": 0.1779,
"step": 660
},
{
"grad_norm": 1.4387675523757935,
"learning_rate": 4.46e-05,
"loss": 0.1764,
"step": 670
},
{
"grad_norm": 1.2351089715957642,
"learning_rate": 4.526666666666667e-05,
"loss": 0.1615,
"step": 680
},
{
"grad_norm": 1.2490671873092651,
"learning_rate": 4.5933333333333336e-05,
"loss": 0.1529,
"step": 690
},
{
"grad_norm": 1.2988910675048828,
"learning_rate": 4.660000000000001e-05,
"loss": 0.1524,
"step": 700
},
{
"grad_norm": 1.1097872257232666,
"learning_rate": 4.726666666666667e-05,
"loss": 0.1448,
"step": 710
},
{
"grad_norm": 1.7465487718582153,
"learning_rate": 4.793333333333334e-05,
"loss": 0.1437,
"step": 720
},
{
"grad_norm": 1.3112348318099976,
"learning_rate": 4.86e-05,
"loss": 0.143,
"step": 730
},
{
"grad_norm": 1.1917496919631958,
"learning_rate": 4.926666666666667e-05,
"loss": 0.1295,
"step": 740
},
{
"grad_norm": 1.729387640953064,
"learning_rate": 4.993333333333334e-05,
"loss": 0.123,
"step": 750
},
{
"grad_norm": 1.7255092859268188,
"learning_rate": 5.0600000000000003e-05,
"loss": 0.1219,
"step": 760
},
{
"grad_norm": 1.6423131227493286,
"learning_rate": 5.1266666666666675e-05,
"loss": 0.1212,
"step": 770
},
{
"grad_norm": 1.2457637786865234,
"learning_rate": 5.193333333333333e-05,
"loss": 0.1153,
"step": 780
},
{
"grad_norm": 1.7827743291854858,
"learning_rate": 5.2600000000000005e-05,
"loss": 0.1132,
"step": 790
},
{
"grad_norm": 1.400822639465332,
"learning_rate": 5.326666666666666e-05,
"loss": 0.114,
"step": 800
},
{
"grad_norm": 1.4029743671417236,
"learning_rate": 5.3933333333333334e-05,
"loss": 0.1042,
"step": 810
},
{
"grad_norm": 1.3635342121124268,
"learning_rate": 5.4600000000000006e-05,
"loss": 0.1042,
"step": 820
},
{
"grad_norm": 2.1806066036224365,
"learning_rate": 5.5266666666666664e-05,
"loss": 0.0922,
"step": 830
},
{
"grad_norm": 1.3689796924591064,
"learning_rate": 5.5933333333333335e-05,
"loss": 0.0904,
"step": 840
},
{
"grad_norm": 1.4890871047973633,
"learning_rate": 5.66e-05,
"loss": 0.0928,
"step": 850
},
{
"grad_norm": 1.2499778270721436,
"learning_rate": 5.726666666666667e-05,
"loss": 0.0846,
"step": 860
},
{
"grad_norm": 1.0627532005310059,
"learning_rate": 5.7933333333333337e-05,
"loss": 0.081,
"step": 870
},
{
"grad_norm": 1.0695182085037231,
"learning_rate": 5.86e-05,
"loss": 0.0824,
"step": 880
},
{
"grad_norm": 1.68373441696167,
"learning_rate": 5.926666666666667e-05,
"loss": 0.0821,
"step": 890
},
{
"grad_norm": 1.0833736658096313,
"learning_rate": 5.9933333333333345e-05,
"loss": 0.0802,
"step": 900
},
{
"grad_norm": 1.1082783937454224,
"learning_rate": 6.06e-05,
"loss": 0.0766,
"step": 910
},
{
"grad_norm": 1.22420072555542,
"learning_rate": 6.126666666666667e-05,
"loss": 0.0772,
"step": 920
},
{
"grad_norm": 1.7994489669799805,
"learning_rate": 6.193333333333333e-05,
"loss": 0.0785,
"step": 930
},
{
"grad_norm": 0.9922900795936584,
"learning_rate": 6.26e-05,
"loss": 0.0775,
"step": 940
},
{
"grad_norm": 1.074314832687378,
"learning_rate": 6.326666666666667e-05,
"loss": 0.0761,
"step": 950
},
{
"grad_norm": 1.3164290189743042,
"learning_rate": 6.393333333333333e-05,
"loss": 0.0739,
"step": 960
},
{
"grad_norm": 1.3246431350708008,
"learning_rate": 6.460000000000001e-05,
"loss": 0.0764,
"step": 970
},
{
"grad_norm": 1.1399774551391602,
"learning_rate": 6.526666666666666e-05,
"loss": 0.0757,
"step": 980
},
{
"grad_norm": 0.9545037746429443,
"learning_rate": 6.593333333333334e-05,
"loss": 0.0769,
"step": 990
},
{
"grad_norm": 0.9759553670883179,
"learning_rate": 6.66e-05,
"loss": 0.0745,
"step": 1000
},
{
"grad_norm": 1.342103123664856,
"learning_rate": 6.726666666666667e-05,
"loss": 0.073,
"step": 1010
},
{
"grad_norm": 1.022963523864746,
"learning_rate": 6.793333333333334e-05,
"loss": 0.0711,
"step": 1020
},
{
"grad_norm": 1.0277061462402344,
"learning_rate": 6.860000000000001e-05,
"loss": 0.0714,
"step": 1030
},
{
"grad_norm": 1.4367462396621704,
"learning_rate": 6.926666666666667e-05,
"loss": 0.0708,
"step": 1040
},
{
"grad_norm": 1.2014946937561035,
"learning_rate": 6.993333333333334e-05,
"loss": 0.0705,
"step": 1050
},
{
"grad_norm": 1.111018419265747,
"learning_rate": 7.06e-05,
"loss": 0.0647,
"step": 1060
},
{
"grad_norm": 1.1501191854476929,
"learning_rate": 7.126666666666667e-05,
"loss": 0.0669,
"step": 1070
},
{
"grad_norm": 0.975878119468689,
"learning_rate": 7.193333333333334e-05,
"loss": 0.0648,
"step": 1080
},
{
"grad_norm": 1.1001148223876953,
"learning_rate": 7.26e-05,
"loss": 0.0662,
"step": 1090
},
{
"grad_norm": 0.9814893007278442,
"learning_rate": 7.326666666666667e-05,
"loss": 0.0679,
"step": 1100
},
{
"grad_norm": 1.2277650833129883,
"learning_rate": 7.393333333333333e-05,
"loss": 0.0688,
"step": 1110
},
{
"grad_norm": 0.9856839776039124,
"learning_rate": 7.46e-05,
"loss": 0.0648,
"step": 1120
},
{
"grad_norm": 1.1703674793243408,
"learning_rate": 7.526666666666668e-05,
"loss": 0.0687,
"step": 1130
},
{
"grad_norm": 1.1658049821853638,
"learning_rate": 7.593333333333334e-05,
"loss": 0.0646,
"step": 1140
},
{
"grad_norm": 1.1635832786560059,
"learning_rate": 7.66e-05,
"loss": 0.0657,
"step": 1150
},
{
"grad_norm": 1.1898596286773682,
"learning_rate": 7.726666666666667e-05,
"loss": 0.0679,
"step": 1160
},
{
"grad_norm": 1.04132878780365,
"learning_rate": 7.793333333333333e-05,
"loss": 0.0674,
"step": 1170
},
{
"grad_norm": 0.9637924432754517,
"learning_rate": 7.860000000000001e-05,
"loss": 0.0666,
"step": 1180
},
{
"grad_norm": 1.0488718748092651,
"learning_rate": 7.926666666666666e-05,
"loss": 0.0638,
"step": 1190
},
{
"grad_norm": 0.9946198463439941,
"learning_rate": 7.993333333333334e-05,
"loss": 0.0639,
"step": 1200
},
{
"grad_norm": 0.9626931548118591,
"learning_rate": 8.060000000000001e-05,
"loss": 0.0638,
"step": 1210
},
{
"grad_norm": 0.8904481530189514,
"learning_rate": 8.126666666666667e-05,
"loss": 0.0614,
"step": 1220
},
{
"grad_norm": 1.0349605083465576,
"learning_rate": 8.193333333333334e-05,
"loss": 0.0606,
"step": 1230
},
{
"grad_norm": 1.2032581567764282,
"learning_rate": 8.26e-05,
"loss": 0.06,
"step": 1240
},
{
"grad_norm": 1.0489283800125122,
"learning_rate": 8.326666666666667e-05,
"loss": 0.0629,
"step": 1250
},
{
"grad_norm": 0.8637824058532715,
"learning_rate": 8.393333333333335e-05,
"loss": 0.0606,
"step": 1260
},
{
"grad_norm": 0.9523459672927856,
"learning_rate": 8.46e-05,
"loss": 0.0608,
"step": 1270
},
{
"grad_norm": 0.910315990447998,
"learning_rate": 8.526666666666667e-05,
"loss": 0.0629,
"step": 1280
},
{
"grad_norm": 1.4048068523406982,
"learning_rate": 8.593333333333333e-05,
"loss": 0.0598,
"step": 1290
},
{
"grad_norm": 0.8386713862419128,
"learning_rate": 8.66e-05,
"loss": 0.0622,
"step": 1300
},
{
"grad_norm": 0.864817202091217,
"learning_rate": 8.726666666666667e-05,
"loss": 0.0598,
"step": 1310
},
{
"grad_norm": 0.9372347593307495,
"learning_rate": 8.793333333333333e-05,
"loss": 0.0594,
"step": 1320
},
{
"grad_norm": 0.9286304116249084,
"learning_rate": 8.86e-05,
"loss": 0.0621,
"step": 1330
},
{
"grad_norm": 0.8566858172416687,
"learning_rate": 8.926666666666668e-05,
"loss": 0.0587,
"step": 1340
},
{
"grad_norm": 1.0118904113769531,
"learning_rate": 8.993333333333334e-05,
"loss": 0.058,
"step": 1350
},
{
"grad_norm": 0.9429856538772583,
"learning_rate": 9.06e-05,
"loss": 0.0559,
"step": 1360
},
{
"grad_norm": 0.8768374919891357,
"learning_rate": 9.126666666666667e-05,
"loss": 0.0567,
"step": 1370
},
{
"grad_norm": 0.8839522004127502,
"learning_rate": 9.193333333333334e-05,
"loss": 0.0544,
"step": 1380
},
{
"grad_norm": 0.9084612131118774,
"learning_rate": 9.260000000000001e-05,
"loss": 0.0565,
"step": 1390
},
{
"grad_norm": 0.9119452238082886,
"learning_rate": 9.326666666666667e-05,
"loss": 0.0558,
"step": 1400
},
{
"grad_norm": 0.9000705480575562,
"learning_rate": 9.393333333333334e-05,
"loss": 0.0579,
"step": 1410
},
{
"grad_norm": 0.8316766023635864,
"learning_rate": 9.46e-05,
"loss": 0.0527,
"step": 1420
},
{
"grad_norm": 1.1867926120758057,
"learning_rate": 9.526666666666667e-05,
"loss": 0.0576,
"step": 1430
},
{
"grad_norm": 0.8146836161613464,
"learning_rate": 9.593333333333334e-05,
"loss": 0.054,
"step": 1440
},
{
"grad_norm": 1.0247355699539185,
"learning_rate": 9.66e-05,
"loss": 0.0543,
"step": 1450
},
{
"grad_norm": 0.990510880947113,
"learning_rate": 9.726666666666667e-05,
"loss": 0.0543,
"step": 1460
},
{
"grad_norm": 0.8735817074775696,
"learning_rate": 9.793333333333333e-05,
"loss": 0.0589,
"step": 1470
},
{
"grad_norm": 0.981984555721283,
"learning_rate": 9.86e-05,
"loss": 0.055,
"step": 1480
},
{
"grad_norm": 0.7862304449081421,
"learning_rate": 9.926666666666668e-05,
"loss": 0.051,
"step": 1490
},
{
"grad_norm": 1.1150578260421753,
"learning_rate": 9.993333333333334e-05,
"loss": 0.0527,
"step": 1500
},
{
"grad_norm": 0.7199082970619202,
"learning_rate": 9.999997539434007e-05,
"loss": 0.0551,
"step": 1510
},
{
"grad_norm": 0.9026912450790405,
"learning_rate": 9.999989033776898e-05,
"loss": 0.0533,
"step": 1520
},
{
"grad_norm": 0.6898888349533081,
"learning_rate": 9.999974452661641e-05,
"loss": 0.0532,
"step": 1530
},
{
"grad_norm": 0.8844292759895325,
"learning_rate": 9.999953796105959e-05,
"loss": 0.0521,
"step": 1540
},
{
"grad_norm": 0.9320904612541199,
"learning_rate": 9.999927064134949e-05,
"loss": 0.0556,
"step": 1550
},
{
"grad_norm": 0.9313002228736877,
"learning_rate": 9.999894256781095e-05,
"loss": 0.0541,
"step": 1560
},
{
"grad_norm": 0.8291693329811096,
"learning_rate": 9.99985537408426e-05,
"loss": 0.0536,
"step": 1570
},
{
"grad_norm": 0.5824195742607117,
"learning_rate": 9.999810416091688e-05,
"loss": 0.0501,
"step": 1580
},
{
"grad_norm": 0.7193527221679688,
"learning_rate": 9.99975938285801e-05,
"loss": 0.0488,
"step": 1590
},
{
"grad_norm": 0.6793032884597778,
"learning_rate": 9.999702274445236e-05,
"loss": 0.0504,
"step": 1600
},
{
"grad_norm": 0.8828063011169434,
"learning_rate": 9.999639090922756e-05,
"loss": 0.0491,
"step": 1610
},
{
"grad_norm": 0.9317867159843445,
"learning_rate": 9.999569832367346e-05,
"loss": 0.0483,
"step": 1620
},
{
"grad_norm": 0.7257526516914368,
"learning_rate": 9.999494498863162e-05,
"loss": 0.0501,
"step": 1630
},
{
"grad_norm": 1.0141751766204834,
"learning_rate": 9.99941309050174e-05,
"loss": 0.0502,
"step": 1640
},
{
"grad_norm": 0.7642090320587158,
"learning_rate": 9.999325607381999e-05,
"loss": 0.0539,
"step": 1650
},
{
"grad_norm": 0.6591696739196777,
"learning_rate": 9.999232049610238e-05,
"loss": 0.0492,
"step": 1660
},
{
"grad_norm": 0.8247687220573425,
"learning_rate": 9.999132417300142e-05,
"loss": 0.0459,
"step": 1670
},
{
"grad_norm": 0.912160336971283,
"learning_rate": 9.99902671057277e-05,
"loss": 0.0475,
"step": 1680
},
{
"grad_norm": 0.7522016763687134,
"learning_rate": 9.998914929556569e-05,
"loss": 0.0493,
"step": 1690
},
{
"grad_norm": 0.7177233099937439,
"learning_rate": 9.998797074387361e-05,
"loss": 0.0475,
"step": 1700
},
{
"grad_norm": 0.8339974880218506,
"learning_rate": 9.99867314520835e-05,
"loss": 0.0446,
"step": 1710
},
{
"grad_norm": 0.8249689340591431,
"learning_rate": 9.998543142170126e-05,
"loss": 0.0495,
"step": 1720
},
{
"grad_norm": 0.7558274865150452,
"learning_rate": 9.99840706543065e-05,
"loss": 0.0493,
"step": 1730
},
{
"grad_norm": 0.6998085379600525,
"learning_rate": 9.998264915155274e-05,
"loss": 0.0452,
"step": 1740
},
{
"grad_norm": 0.7514432072639465,
"learning_rate": 9.998116691516718e-05,
"loss": 0.0455,
"step": 1750
},
{
"grad_norm": 0.9076456427574158,
"learning_rate": 9.997962394695091e-05,
"loss": 0.0459,
"step": 1760
},
{
"grad_norm": 0.5168872475624084,
"learning_rate": 9.997802024877875e-05,
"loss": 0.0465,
"step": 1770
},
{
"grad_norm": 0.6847232580184937,
"learning_rate": 9.99763558225994e-05,
"loss": 0.046,
"step": 1780
},
{
"grad_norm": 0.6557778120040894,
"learning_rate": 9.997463067043526e-05,
"loss": 0.0485,
"step": 1790
},
{
"grad_norm": 0.7632912993431091,
"learning_rate": 9.997284479438253e-05,
"loss": 0.0466,
"step": 1800
},
{
"grad_norm": 0.8196837902069092,
"learning_rate": 9.997099819661127e-05,
"loss": 0.0445,
"step": 1810
},
{
"grad_norm": 0.7271484732627869,
"learning_rate": 9.996909087936524e-05,
"loss": 0.0438,
"step": 1820
},
{
"grad_norm": 0.7890754342079163,
"learning_rate": 9.996712284496201e-05,
"loss": 0.0461,
"step": 1830
},
{
"grad_norm": 0.6526637077331543,
"learning_rate": 9.996509409579293e-05,
"loss": 0.0444,
"step": 1840
},
{
"grad_norm": 0.8112354874610901,
"learning_rate": 9.996300463432312e-05,
"loss": 0.0452,
"step": 1850
},
{
"grad_norm": 0.6816654205322266,
"learning_rate": 9.996085446309148e-05,
"loss": 0.0457,
"step": 1860
},
{
"grad_norm": 0.7028149962425232,
"learning_rate": 9.995864358471066e-05,
"loss": 0.0449,
"step": 1870
},
{
"grad_norm": 0.7199757695198059,
"learning_rate": 9.99563720018671e-05,
"loss": 0.0441,
"step": 1880
},
{
"grad_norm": 0.6490606069564819,
"learning_rate": 9.995403971732098e-05,
"loss": 0.043,
"step": 1890
},
{
"grad_norm": 0.8152973651885986,
"learning_rate": 9.995164673390625e-05,
"loss": 0.0457,
"step": 1900
},
{
"grad_norm": 0.769864022731781,
"learning_rate": 9.994919305453059e-05,
"loss": 0.0425,
"step": 1910
},
{
"grad_norm": 0.6499236226081848,
"learning_rate": 9.994667868217548e-05,
"loss": 0.0435,
"step": 1920
},
{
"grad_norm": 0.7776778936386108,
"learning_rate": 9.99441036198961e-05,
"loss": 0.0429,
"step": 1930
},
{
"grad_norm": 0.7239458560943604,
"learning_rate": 9.99414678708214e-05,
"loss": 0.0435,
"step": 1940
},
{
"grad_norm": 0.601598858833313,
"learning_rate": 9.993877143815407e-05,
"loss": 0.042,
"step": 1950
},
{
"grad_norm": 0.7758349180221558,
"learning_rate": 9.993601432517053e-05,
"loss": 0.0476,
"step": 1960
},
{
"grad_norm": 0.6978219151496887,
"learning_rate": 9.993319653522091e-05,
"loss": 0.0418,
"step": 1970
},
{
"grad_norm": 0.7210837006568909,
"learning_rate": 9.993031807172911e-05,
"loss": 0.0428,
"step": 1980
},
{
"grad_norm": 0.7555508017539978,
"learning_rate": 9.992737893819273e-05,
"loss": 0.0468,
"step": 1990
},
{
"grad_norm": 0.6775459051132202,
"learning_rate": 9.992437913818312e-05,
"loss": 0.0447,
"step": 2000
},
{
"grad_norm": 0.7015236020088196,
"learning_rate": 9.992131867534526e-05,
"loss": 0.0426,
"step": 2010
},
{
"grad_norm": 0.7684049010276794,
"learning_rate": 9.991819755339796e-05,
"loss": 0.0425,
"step": 2020
},
{
"grad_norm": 0.6014548540115356,
"learning_rate": 9.991501577613365e-05,
"loss": 0.0428,
"step": 2030
},
{
"grad_norm": 0.6015975475311279,
"learning_rate": 9.99117733474185e-05,
"loss": 0.0412,
"step": 2040
},
{
"grad_norm": 0.6926084756851196,
"learning_rate": 9.990847027119234e-05,
"loss": 0.0409,
"step": 2050
},
{
"grad_norm": 0.5292817950248718,
"learning_rate": 9.990510655146877e-05,
"loss": 0.0404,
"step": 2060
},
{
"grad_norm": 0.842787504196167,
"learning_rate": 9.990168219233496e-05,
"loss": 0.0425,
"step": 2070
},
{
"grad_norm": 0.7378549575805664,
"learning_rate": 9.989819719795188e-05,
"loss": 0.0427,
"step": 2080
},
{
"grad_norm": 0.6256470680236816,
"learning_rate": 9.989465157255412e-05,
"loss": 0.0409,
"step": 2090
},
{
"grad_norm": 0.6263523697853088,
"learning_rate": 9.989104532044994e-05,
"loss": 0.0408,
"step": 2100
},
{
"grad_norm": 0.7206677794456482,
"learning_rate": 9.988737844602128e-05,
"loss": 0.0404,
"step": 2110
},
{
"grad_norm": 0.7646312713623047,
"learning_rate": 9.988365095372372e-05,
"loss": 0.0411,
"step": 2120
},
{
"grad_norm": 0.6586979031562805,
"learning_rate": 9.987986284808654e-05,
"loss": 0.0409,
"step": 2130
},
{
"grad_norm": 0.6952325701713562,
"learning_rate": 9.987601413371264e-05,
"loss": 0.0387,
"step": 2140
},
{
"grad_norm": 0.6108909249305725,
"learning_rate": 9.987210481527855e-05,
"loss": 0.0401,
"step": 2150
},
{
"grad_norm": 0.618565022945404,
"learning_rate": 9.98681348975345e-05,
"loss": 0.041,
"step": 2160
},
{
"grad_norm": 0.6037882566452026,
"learning_rate": 9.986410438530427e-05,
"loss": 0.0414,
"step": 2170
},
{
"grad_norm": 0.685737669467926,
"learning_rate": 9.986001328348534e-05,
"loss": 0.0379,
"step": 2180
},
{
"grad_norm": 0.6301174163818359,
"learning_rate": 9.985586159704878e-05,
"loss": 0.0415,
"step": 2190
},
{
"grad_norm": 0.7248358726501465,
"learning_rate": 9.985164933103929e-05,
"loss": 0.0407,
"step": 2200
},
{
"grad_norm": 0.6090236306190491,
"learning_rate": 9.984737649057513e-05,
"loss": 0.0404,
"step": 2210
},
{
"grad_norm": 0.6727856993675232,
"learning_rate": 9.984304308084827e-05,
"loss": 0.0407,
"step": 2220
},
{
"grad_norm": 0.5758669376373291,
"learning_rate": 9.983864910712416e-05,
"loss": 0.0392,
"step": 2230
},
{
"grad_norm": 0.6936368942260742,
"learning_rate": 9.98341945747419e-05,
"loss": 0.0396,
"step": 2240
},
{
"grad_norm": 0.7542185187339783,
"learning_rate": 9.98296794891142e-05,
"loss": 0.0392,
"step": 2250
},
{
"grad_norm": 0.7693107724189758,
"learning_rate": 9.982510385572725e-05,
"loss": 0.0371,
"step": 2260
},
{
"grad_norm": 0.6100176572799683,
"learning_rate": 9.982046768014094e-05,
"loss": 0.0383,
"step": 2270
},
{
"grad_norm": 0.8347607254981995,
"learning_rate": 9.981577096798863e-05,
"loss": 0.0398,
"step": 2280
},
{
"grad_norm": 0.7694611549377441,
"learning_rate": 9.981101372497727e-05,
"loss": 0.0393,
"step": 2290
},
{
"grad_norm": 0.7548644542694092,
"learning_rate": 9.980619595688737e-05,
"loss": 0.04,
"step": 2300
},
{
"grad_norm": 0.5724036693572998,
"learning_rate": 9.980131766957295e-05,
"loss": 0.0394,
"step": 2310
},
{
"grad_norm": 0.6039924621582031,
"learning_rate": 9.979637886896163e-05,
"loss": 0.038,
"step": 2320
},
{
"grad_norm": 0.5768024921417236,
"learning_rate": 9.979137956105447e-05,
"loss": 0.0392,
"step": 2330
},
{
"grad_norm": 0.6355780959129333,
"learning_rate": 9.978631975192613e-05,
"loss": 0.0384,
"step": 2340
},
{
"grad_norm": 0.7645574808120728,
"learning_rate": 9.978119944772475e-05,
"loss": 0.0409,
"step": 2350
},
{
"grad_norm": 0.6496811509132385,
"learning_rate": 9.977601865467197e-05,
"loss": 0.0393,
"step": 2360
},
{
"grad_norm": 0.5277372002601624,
"learning_rate": 9.977077737906297e-05,
"loss": 0.038,
"step": 2370
},
{
"grad_norm": 0.7123770117759705,
"learning_rate": 9.976547562726636e-05,
"loss": 0.0411,
"step": 2380
},
{
"grad_norm": 0.6734181642532349,
"learning_rate": 9.976011340572429e-05,
"loss": 0.0416,
"step": 2390
},
{
"grad_norm": 0.5722440481185913,
"learning_rate": 9.975469072095237e-05,
"loss": 0.039,
"step": 2400
},
{
"grad_norm": 0.7019569873809814,
"learning_rate": 9.974920757953965e-05,
"loss": 0.0403,
"step": 2410
},
{
"grad_norm": 0.6497634649276733,
"learning_rate": 9.97436639881487e-05,
"loss": 0.039,
"step": 2420
},
{
"grad_norm": 0.44458356499671936,
"learning_rate": 9.973805995351545e-05,
"loss": 0.0377,
"step": 2430
},
{
"grad_norm": 0.6675490140914917,
"learning_rate": 9.973239548244939e-05,
"loss": 0.0369,
"step": 2440
},
{
"grad_norm": 0.5237282514572144,
"learning_rate": 9.972667058183333e-05,
"loss": 0.0367,
"step": 2450
},
{
"grad_norm": 0.6908047795295715,
"learning_rate": 9.972088525862362e-05,
"loss": 0.0388,
"step": 2460
},
{
"grad_norm": 0.5526571273803711,
"learning_rate": 9.971503951984995e-05,
"loss": 0.0386,
"step": 2470
},
{
"grad_norm": 0.8267020583152771,
"learning_rate": 9.970913337261543e-05,
"loss": 0.0409,
"step": 2480
},
{
"grad_norm": 0.6126002073287964,
"learning_rate": 9.97031668240966e-05,
"loss": 0.0385,
"step": 2490
},
{
"grad_norm": 0.6993176341056824,
"learning_rate": 9.969713988154339e-05,
"loss": 0.0392,
"step": 2500
},
{
"grad_norm": 0.711223304271698,
"learning_rate": 9.969105255227906e-05,
"loss": 0.0394,
"step": 2510
},
{
"grad_norm": 0.5168538093566895,
"learning_rate": 9.968490484370035e-05,
"loss": 0.0372,
"step": 2520
},
{
"grad_norm": 0.6115143299102783,
"learning_rate": 9.967869676327726e-05,
"loss": 0.036,
"step": 2530
},
{
"grad_norm": 0.57094407081604,
"learning_rate": 9.96724283185532e-05,
"loss": 0.0352,
"step": 2540
},
{
"grad_norm": 0.5119908452033997,
"learning_rate": 9.966609951714494e-05,
"loss": 0.0362,
"step": 2550
},
{
"grad_norm": 0.7033934593200684,
"learning_rate": 9.965971036674255e-05,
"loss": 0.0373,
"step": 2560
},
{
"grad_norm": 0.6613783240318298,
"learning_rate": 9.965326087510947e-05,
"loss": 0.0391,
"step": 2570
},
{
"grad_norm": 0.6085628867149353,
"learning_rate": 9.964675105008244e-05,
"loss": 0.0376,
"step": 2580
},
{
"grad_norm": 0.5930835604667664,
"learning_rate": 9.964018089957147e-05,
"loss": 0.0344,
"step": 2590
},
{
"grad_norm": 0.5690402388572693,
"learning_rate": 9.963355043155996e-05,
"loss": 0.0368,
"step": 2600
},
{
"grad_norm": 0.6252157092094421,
"learning_rate": 9.962685965410455e-05,
"loss": 0.0383,
"step": 2610
},
{
"grad_norm": 0.690443754196167,
"learning_rate": 9.962010857533514e-05,
"loss": 0.0369,
"step": 2620
},
{
"grad_norm": 0.6155582070350647,
"learning_rate": 9.961329720345493e-05,
"loss": 0.0371,
"step": 2630
},
{
"grad_norm": 0.6473574042320251,
"learning_rate": 9.96064255467404e-05,
"loss": 0.0375,
"step": 2640
},
{
"grad_norm": 0.6162963509559631,
"learning_rate": 9.959949361354126e-05,
"loss": 0.0364,
"step": 2650
},
{
"grad_norm": 0.6339418888092041,
"learning_rate": 9.959250141228045e-05,
"loss": 0.0356,
"step": 2660
},
{
"grad_norm": 0.4604366719722748,
"learning_rate": 9.958544895145414e-05,
"loss": 0.0374,
"step": 2670
},
{
"grad_norm": 0.5806917548179626,
"learning_rate": 9.957833623963177e-05,
"loss": 0.0353,
"step": 2680
},
{
"grad_norm": 0.653438925743103,
"learning_rate": 9.957116328545593e-05,
"loss": 0.0351,
"step": 2690
},
{
"grad_norm": 0.6917750835418701,
"learning_rate": 9.956393009764244e-05,
"loss": 0.0353,
"step": 2700
},
{
"grad_norm": 0.626438558101654,
"learning_rate": 9.955663668498032e-05,
"loss": 0.0355,
"step": 2710
},
{
"grad_norm": 0.478832870721817,
"learning_rate": 9.954928305633173e-05,
"loss": 0.034,
"step": 2720
},
{
"grad_norm": 0.5281680226325989,
"learning_rate": 9.954186922063204e-05,
"loss": 0.0337,
"step": 2730
},
{
"grad_norm": 0.6203818917274475,
"learning_rate": 9.953439518688974e-05,
"loss": 0.0349,
"step": 2740
},
{
"grad_norm": 0.5640031695365906,
"learning_rate": 9.952686096418652e-05,
"loss": 0.0365,
"step": 2750
},
{
"grad_norm": 0.5402900576591492,
"learning_rate": 9.951926656167714e-05,
"loss": 0.0346,
"step": 2760
},
{
"grad_norm": 0.6839205026626587,
"learning_rate": 9.951161198858953e-05,
"loss": 0.0352,
"step": 2770
},
{
"grad_norm": 0.5779417753219604,
"learning_rate": 9.950389725422471e-05,
"loss": 0.0373,
"step": 2780
},
{
"grad_norm": 0.6267558336257935,
"learning_rate": 9.949612236795682e-05,
"loss": 0.0384,
"step": 2790
},
{
"grad_norm": 0.5585941076278687,
"learning_rate": 9.948828733923305e-05,
"loss": 0.0375,
"step": 2800
},
{
"grad_norm": 0.7183846831321716,
"learning_rate": 9.948039217757374e-05,
"loss": 0.0369,
"step": 2810
},
{
"grad_norm": 0.6528863906860352,
"learning_rate": 9.947243689257225e-05,
"loss": 0.0355,
"step": 2820
},
{
"grad_norm": 0.43302688002586365,
"learning_rate": 9.946442149389497e-05,
"loss": 0.0346,
"step": 2830
},
{
"grad_norm": 0.4747207462787628,
"learning_rate": 9.945634599128139e-05,
"loss": 0.0353,
"step": 2840
},
{
"grad_norm": 0.516400158405304,
"learning_rate": 9.944821039454402e-05,
"loss": 0.034,
"step": 2850
},
{
"grad_norm": 0.5410118699073792,
"learning_rate": 9.944001471356835e-05,
"loss": 0.034,
"step": 2860
},
{
"grad_norm": 0.5934143662452698,
"learning_rate": 9.94317589583129e-05,
"loss": 0.0376,
"step": 2870
},
{
"grad_norm": 0.5651362538337708,
"learning_rate": 9.942344313880922e-05,
"loss": 0.0345,
"step": 2880
},
{
"grad_norm": 0.7414537668228149,
"learning_rate": 9.941506726516179e-05,
"loss": 0.0355,
"step": 2890
},
{
"grad_norm": 0.4679614007472992,
"learning_rate": 9.94066313475481e-05,
"loss": 0.0349,
"step": 2900
},
{
"grad_norm": 0.5023335218429565,
"learning_rate": 9.939813539621857e-05,
"loss": 0.0352,
"step": 2910
},
{
"grad_norm": 0.5964629054069519,
"learning_rate": 9.93895794214966e-05,
"loss": 0.0334,
"step": 2920
},
{
"grad_norm": 0.5581484436988831,
"learning_rate": 9.938096343377852e-05,
"loss": 0.0352,
"step": 2930
},
{
"grad_norm": 0.5380269289016724,
"learning_rate": 9.937228744353353e-05,
"loss": 0.0349,
"step": 2940
},
{
"grad_norm": 0.4668833017349243,
"learning_rate": 9.936355146130379e-05,
"loss": 0.0328,
"step": 2950
},
{
"grad_norm": 0.43839725852012634,
"learning_rate": 9.935475549770435e-05,
"loss": 0.0335,
"step": 2960
},
{
"grad_norm": 0.5909776091575623,
"learning_rate": 9.934589956342315e-05,
"loss": 0.0356,
"step": 2970
},
{
"grad_norm": 0.46072277426719666,
"learning_rate": 9.933698366922093e-05,
"loss": 0.035,
"step": 2980
},
{
"grad_norm": 0.6207431554794312,
"learning_rate": 9.93280078259314e-05,
"loss": 0.0341,
"step": 2990
},
{
"grad_norm": 0.5444117188453674,
"learning_rate": 9.931897204446104e-05,
"loss": 0.033,
"step": 3000
},
{
"grad_norm": 0.5664342641830444,
"learning_rate": 9.930987633578915e-05,
"loss": 0.0344,
"step": 3010
},
{
"grad_norm": 0.7013835310935974,
"learning_rate": 9.93007207109679e-05,
"loss": 0.0342,
"step": 3020
},
{
"grad_norm": 0.41322097182273865,
"learning_rate": 9.929150518112224e-05,
"loss": 0.0338,
"step": 3030
},
{
"grad_norm": 0.4599977135658264,
"learning_rate": 9.928222975744991e-05,
"loss": 0.0349,
"step": 3040
},
{
"grad_norm": 0.4189639091491699,
"learning_rate": 9.92728944512214e-05,
"loss": 0.0334,
"step": 3050
},
{
"grad_norm": 0.6589005589485168,
"learning_rate": 9.926349927378001e-05,
"loss": 0.0348,
"step": 3060
},
{
"grad_norm": 0.4410434365272522,
"learning_rate": 9.925404423654174e-05,
"loss": 0.0349,
"step": 3070
},
{
"grad_norm": 0.5676413774490356,
"learning_rate": 9.924452935099537e-05,
"loss": 0.0323,
"step": 3080
},
{
"grad_norm": 0.5551483631134033,
"learning_rate": 9.92349546287024e-05,
"loss": 0.0316,
"step": 3090
},
{
"grad_norm": 0.6876076459884644,
"learning_rate": 9.9225320081297e-05,
"loss": 0.0339,
"step": 3100
},
{
"grad_norm": 0.5772173404693604,
"learning_rate": 9.921562572048606e-05,
"loss": 0.0347,
"step": 3110
},
{
"grad_norm": 0.5337933897972107,
"learning_rate": 9.920587155804913e-05,
"loss": 0.0329,
"step": 3120
},
{
"grad_norm": 0.5200109481811523,
"learning_rate": 9.919605760583845e-05,
"loss": 0.0314,
"step": 3130
},
{
"grad_norm": 0.4010808765888214,
"learning_rate": 9.91861838757789e-05,
"loss": 0.0312,
"step": 3140
},
{
"grad_norm": 0.602595329284668,
"learning_rate": 9.917625037986798e-05,
"loss": 0.0329,
"step": 3150
},
{
"grad_norm": 0.42525333166122437,
"learning_rate": 9.916625713017583e-05,
"loss": 0.0324,
"step": 3160
},
{
"grad_norm": 0.645604133605957,
"learning_rate": 9.915620413884519e-05,
"loss": 0.0328,
"step": 3170
},
{
"grad_norm": 0.5573173761367798,
"learning_rate": 9.914609141809139e-05,
"loss": 0.0347,
"step": 3180
},
{
"grad_norm": 0.5239332318305969,
"learning_rate": 9.913591898020235e-05,
"loss": 0.0326,
"step": 3190
},
{
"grad_norm": 0.6686249375343323,
"learning_rate": 9.912568683753853e-05,
"loss": 0.0354,
"step": 3200
},
{
"grad_norm": 0.5051627159118652,
"learning_rate": 9.911539500253295e-05,
"loss": 0.0341,
"step": 3210
},
{
"grad_norm": 0.49810031056404114,
"learning_rate": 9.910504348769118e-05,
"loss": 0.0315,
"step": 3220
},
{
"grad_norm": 0.6298574209213257,
"learning_rate": 9.909463230559127e-05,
"loss": 0.0319,
"step": 3230
},
{
"grad_norm": 0.4464428424835205,
"learning_rate": 9.908416146888376e-05,
"loss": 0.0307,
"step": 3240
},
{
"grad_norm": 0.5502170324325562,
"learning_rate": 9.907363099029175e-05,
"loss": 0.0336,
"step": 3250
},
{
"grad_norm": 0.5409323573112488,
"learning_rate": 9.906304088261073e-05,
"loss": 0.0333,
"step": 3260
},
{
"grad_norm": 0.5345064997673035,
"learning_rate": 9.905239115870872e-05,
"loss": 0.0306,
"step": 3270
},
{
"grad_norm": 0.710795521736145,
"learning_rate": 9.90416818315261e-05,
"loss": 0.0327,
"step": 3280
},
{
"grad_norm": 0.6401283144950867,
"learning_rate": 9.903091291407573e-05,
"loss": 0.0315,
"step": 3290
},
{
"grad_norm": 0.4734502136707306,
"learning_rate": 9.902008441944286e-05,
"loss": 0.0319,
"step": 3300
},
{
"grad_norm": 0.6390936970710754,
"learning_rate": 9.900919636078512e-05,
"loss": 0.0343,
"step": 3310
},
{
"grad_norm": 0.6288327574729919,
"learning_rate": 9.899824875133255e-05,
"loss": 0.0357,
"step": 3320
},
{
"grad_norm": 0.41811782121658325,
"learning_rate": 9.898724160438749e-05,
"loss": 0.0331,
"step": 3330
},
{
"grad_norm": 0.4712206721305847,
"learning_rate": 9.89761749333247e-05,
"loss": 0.0321,
"step": 3340
},
{
"grad_norm": 0.541106104850769,
"learning_rate": 9.896504875159122e-05,
"loss": 0.0317,
"step": 3350
},
{
"grad_norm": 0.6272668242454529,
"learning_rate": 9.89538630727064e-05,
"loss": 0.0337,
"step": 3360
},
{
"grad_norm": 0.4778730869293213,
"learning_rate": 9.894261791026189e-05,
"loss": 0.0329,
"step": 3370
},
{
"grad_norm": 0.5161771178245544,
"learning_rate": 9.893131327792165e-05,
"loss": 0.0358,
"step": 3380
},
{
"grad_norm": 0.5454795956611633,
"learning_rate": 9.891994918942182e-05,
"loss": 0.0337,
"step": 3390
},
{
"grad_norm": 0.5579949021339417,
"learning_rate": 9.890852565857092e-05,
"loss": 0.0325,
"step": 3400
},
{
"grad_norm": 0.43336305022239685,
"learning_rate": 9.889704269924954e-05,
"loss": 0.033,
"step": 3410
},
{
"grad_norm": 0.5180725455284119,
"learning_rate": 9.888550032541059e-05,
"loss": 0.0324,
"step": 3420
},
{
"grad_norm": 0.4674279987812042,
"learning_rate": 9.887389855107916e-05,
"loss": 0.0308,
"step": 3430
},
{
"grad_norm": 0.5220392346382141,
"learning_rate": 9.886223739035248e-05,
"loss": 0.032,
"step": 3440
},
{
"grad_norm": 0.5359625816345215,
"learning_rate": 9.885051685739997e-05,
"loss": 0.0312,
"step": 3450
},
{
"grad_norm": 0.5261549949645996,
"learning_rate": 9.883873696646316e-05,
"loss": 0.0327,
"step": 3460
},
{
"grad_norm": 0.43261486291885376,
"learning_rate": 9.882689773185575e-05,
"loss": 0.0313,
"step": 3470
},
{
"grad_norm": 0.5521197319030762,
"learning_rate": 9.881499916796353e-05,
"loss": 0.0314,
"step": 3480
},
{
"grad_norm": 0.6331434845924377,
"learning_rate": 9.880304128924434e-05,
"loss": 0.0313,
"step": 3490
},
{
"grad_norm": 0.49536824226379395,
"learning_rate": 9.879102411022817e-05,
"loss": 0.0292,
"step": 3500
},
{
"grad_norm": 0.5600954294204712,
"learning_rate": 9.877894764551703e-05,
"loss": 0.0308,
"step": 3510
},
{
"grad_norm": 0.46873748302459717,
"learning_rate": 9.876681190978494e-05,
"loss": 0.0305,
"step": 3520
},
{
"grad_norm": 0.5754407644271851,
"learning_rate": 9.875461691777797e-05,
"loss": 0.0296,
"step": 3530
},
{
"grad_norm": 0.552854597568512,
"learning_rate": 9.874236268431417e-05,
"loss": 0.0324,
"step": 3540
},
{
"grad_norm": 0.4065485894680023,
"learning_rate": 9.873004922428361e-05,
"loss": 0.0304,
"step": 3550
},
{
"grad_norm": 0.44858065247535706,
"learning_rate": 9.871767655264829e-05,
"loss": 0.0304,
"step": 3560
},
{
"grad_norm": 0.5338795781135559,
"learning_rate": 9.87052446844422e-05,
"loss": 0.0307,
"step": 3570
},
{
"grad_norm": 0.6342114210128784,
"learning_rate": 9.869275363477122e-05,
"loss": 0.0303,
"step": 3580
},
{
"grad_norm": 0.6753395199775696,
"learning_rate": 9.868020341881312e-05,
"loss": 0.0322,
"step": 3590
},
{
"grad_norm": 0.46530067920684814,
"learning_rate": 9.866759405181765e-05,
"loss": 0.032,
"step": 3600
},
{
"grad_norm": 0.5572906732559204,
"learning_rate": 9.865492554910633e-05,
"loss": 0.0299,
"step": 3610
},
{
"grad_norm": 0.5319712162017822,
"learning_rate": 9.864219792607262e-05,
"loss": 0.0335,
"step": 3620
},
{
"grad_norm": 0.5416569709777832,
"learning_rate": 9.862941119818177e-05,
"loss": 0.029,
"step": 3630
},
{
"grad_norm": 0.4529987871646881,
"learning_rate": 9.861656538097086e-05,
"loss": 0.0312,
"step": 3640
},
{
"grad_norm": 0.49536070227622986,
"learning_rate": 9.860366049004877e-05,
"loss": 0.0306,
"step": 3650
},
{
"grad_norm": 0.5354276895523071,
"learning_rate": 9.859069654109615e-05,
"loss": 0.0317,
"step": 3660
},
{
"grad_norm": 0.5388392806053162,
"learning_rate": 9.857767354986545e-05,
"loss": 0.029,
"step": 3670
},
{
"grad_norm": 0.4168851971626282,
"learning_rate": 9.856459153218078e-05,
"loss": 0.0331,
"step": 3680
},
{
"grad_norm": 0.5648592114448547,
"learning_rate": 9.855145050393808e-05,
"loss": 0.0326,
"step": 3690
},
{
"grad_norm": 0.51558917760849,
"learning_rate": 9.85382504811049e-05,
"loss": 0.0308,
"step": 3700
},
{
"grad_norm": 0.46784237027168274,
"learning_rate": 9.852499147972054e-05,
"loss": 0.0293,
"step": 3710
},
{
"grad_norm": 0.4863705337047577,
"learning_rate": 9.851167351589592e-05,
"loss": 0.0293,
"step": 3720
},
{
"grad_norm": 0.5770860910415649,
"learning_rate": 9.849829660581363e-05,
"loss": 0.0315,
"step": 3730
},
{
"grad_norm": 0.537215530872345,
"learning_rate": 9.848486076572787e-05,
"loss": 0.0294,
"step": 3740
},
{
"grad_norm": 0.5241726040840149,
"learning_rate": 9.847136601196446e-05,
"loss": 0.0294,
"step": 3750
},
{
"grad_norm": 0.5277009010314941,
"learning_rate": 9.845781236092078e-05,
"loss": 0.0297,
"step": 3760
},
{
"grad_norm": 0.6135159134864807,
"learning_rate": 9.844419982906583e-05,
"loss": 0.0293,
"step": 3770
},
{
"grad_norm": 0.5374872088432312,
"learning_rate": 9.843052843294008e-05,
"loss": 0.0314,
"step": 3780
},
{
"grad_norm": 0.561579704284668,
"learning_rate": 9.841679818915559e-05,
"loss": 0.0293,
"step": 3790
},
{
"grad_norm": 0.48467305302619934,
"learning_rate": 9.840300911439591e-05,
"loss": 0.0296,
"step": 3800
},
{
"grad_norm": 0.6467391848564148,
"learning_rate": 9.838916122541603e-05,
"loss": 0.029,
"step": 3810
},
{
"grad_norm": 0.5941223502159119,
"learning_rate": 9.837525453904246e-05,
"loss": 0.0308,
"step": 3820
},
{
"grad_norm": 0.6276981234550476,
"learning_rate": 9.836128907217314e-05,
"loss": 0.0309,
"step": 3830
},
{
"grad_norm": 0.576546847820282,
"learning_rate": 9.834726484177743e-05,
"loss": 0.0299,
"step": 3840
},
{
"grad_norm": 0.539531409740448,
"learning_rate": 9.833318186489609e-05,
"loss": 0.0308,
"step": 3850
},
{
"grad_norm": 0.5105360150337219,
"learning_rate": 9.831904015864126e-05,
"loss": 0.0302,
"step": 3860
},
{
"grad_norm": 0.4079248011112213,
"learning_rate": 9.830483974019645e-05,
"loss": 0.0294,
"step": 3870
},
{
"grad_norm": 0.5691694021224976,
"learning_rate": 9.82905806268165e-05,
"loss": 0.029,
"step": 3880
},
{
"grad_norm": 0.3930612802505493,
"learning_rate": 9.82762628358276e-05,
"loss": 0.0317,
"step": 3890
},
{
"grad_norm": 0.48195546865463257,
"learning_rate": 9.826188638462718e-05,
"loss": 0.0275,
"step": 3900
},
{
"grad_norm": 0.44009217619895935,
"learning_rate": 9.824745129068402e-05,
"loss": 0.0269,
"step": 3910
},
{
"grad_norm": 0.46501123905181885,
"learning_rate": 9.82329575715381e-05,
"loss": 0.0289,
"step": 3920
},
{
"grad_norm": 0.48853984475135803,
"learning_rate": 9.821840524480066e-05,
"loss": 0.029,
"step": 3930
},
{
"grad_norm": 0.4357215166091919,
"learning_rate": 9.820379432815414e-05,
"loss": 0.0283,
"step": 3940
},
{
"grad_norm": 0.5993157029151917,
"learning_rate": 9.81891248393522e-05,
"loss": 0.0297,
"step": 3950
},
{
"grad_norm": 0.48249173164367676,
"learning_rate": 9.817439679621963e-05,
"loss": 0.0295,
"step": 3960
},
{
"grad_norm": 0.5071114301681519,
"learning_rate": 9.815961021665243e-05,
"loss": 0.0297,
"step": 3970
},
{
"grad_norm": 0.5427010655403137,
"learning_rate": 9.814476511861763e-05,
"loss": 0.03,
"step": 3980
},
{
"grad_norm": 0.5940617322921753,
"learning_rate": 9.812986152015348e-05,
"loss": 0.0319,
"step": 3990
},
{
"grad_norm": 0.6007356643676758,
"learning_rate": 9.811489943936922e-05,
"loss": 0.0283,
"step": 4000
},
{
"grad_norm": 0.45132508873939514,
"learning_rate": 9.809987889444522e-05,
"loss": 0.0273,
"step": 4010
},
{
"grad_norm": 0.4298675060272217,
"learning_rate": 9.808479990363282e-05,
"loss": 0.0284,
"step": 4020
},
{
"grad_norm": 0.39303600788116455,
"learning_rate": 9.806966248525445e-05,
"loss": 0.028,
"step": 4030
},
{
"grad_norm": 0.5677952766418457,
"learning_rate": 9.805446665770348e-05,
"loss": 0.0284,
"step": 4040
},
{
"grad_norm": 0.41741976141929626,
"learning_rate": 9.803921243944429e-05,
"loss": 0.0288,
"step": 4050
},
{
"grad_norm": 0.4958172142505646,
"learning_rate": 9.802389984901218e-05,
"loss": 0.0279,
"step": 4060
},
{
"grad_norm": 0.5058831572532654,
"learning_rate": 9.80085289050134e-05,
"loss": 0.0277,
"step": 4070
},
{
"grad_norm": 0.4799667298793793,
"learning_rate": 9.799309962612508e-05,
"loss": 0.0293,
"step": 4080
},
{
"grad_norm": 0.4996417164802551,
"learning_rate": 9.797761203109527e-05,
"loss": 0.029,
"step": 4090
},
{
"grad_norm": 0.46496379375457764,
"learning_rate": 9.796206613874283e-05,
"loss": 0.0289,
"step": 4100
},
{
"grad_norm": 0.44552916288375854,
"learning_rate": 9.794646196795754e-05,
"loss": 0.0277,
"step": 4110
},
{
"grad_norm": 0.4660540819168091,
"learning_rate": 9.793079953769987e-05,
"loss": 0.0277,
"step": 4120
},
{
"grad_norm": 0.5758588910102844,
"learning_rate": 9.79150788670012e-05,
"loss": 0.0286,
"step": 4130
},
{
"grad_norm": 0.47274714708328247,
"learning_rate": 9.78992999749636e-05,
"loss": 0.028,
"step": 4140
},
{
"grad_norm": 0.5877339839935303,
"learning_rate": 9.788346288075994e-05,
"loss": 0.0281,
"step": 4150
},
{
"grad_norm": 0.6938865780830383,
"learning_rate": 9.786756760363373e-05,
"loss": 0.0278,
"step": 4160
},
{
"grad_norm": 0.4909265339374542,
"learning_rate": 9.78516141628993e-05,
"loss": 0.0275,
"step": 4170
},
{
"grad_norm": 0.4589878022670746,
"learning_rate": 9.783560257794154e-05,
"loss": 0.0284,
"step": 4180
},
{
"grad_norm": 0.6123479604721069,
"learning_rate": 9.781953286821603e-05,
"loss": 0.0257,
"step": 4190
},
{
"grad_norm": 0.46670061349868774,
"learning_rate": 9.780340505324901e-05,
"loss": 0.0272,
"step": 4200
},
{
"grad_norm": 0.6732054352760315,
"learning_rate": 9.778721915263727e-05,
"loss": 0.0281,
"step": 4210
},
{
"grad_norm": 0.5753781795501709,
"learning_rate": 9.777097518604824e-05,
"loss": 0.0274,
"step": 4220
},
{
"grad_norm": 0.6165180802345276,
"learning_rate": 9.775467317321984e-05,
"loss": 0.0303,
"step": 4230
},
{
"grad_norm": 0.5160939693450928,
"learning_rate": 9.773831313396055e-05,
"loss": 0.0273,
"step": 4240
},
{
"grad_norm": 0.5413023829460144,
"learning_rate": 9.77218950881494e-05,
"loss": 0.0286,
"step": 4250
},
{
"grad_norm": 0.5136318802833557,
"learning_rate": 9.770541905573583e-05,
"loss": 0.0271,
"step": 4260
},
{
"grad_norm": 0.662514865398407,
"learning_rate": 9.768888505673976e-05,
"loss": 0.0276,
"step": 4270
},
{
"grad_norm": 0.5162094235420227,
"learning_rate": 9.767229311125162e-05,
"loss": 0.0278,
"step": 4280
},
{
"grad_norm": 0.5230931639671326,
"learning_rate": 9.765564323943211e-05,
"loss": 0.0298,
"step": 4290
},
{
"grad_norm": 0.5208863615989685,
"learning_rate": 9.763893546151244e-05,
"loss": 0.0265,
"step": 4300
},
{
"grad_norm": 0.44033685326576233,
"learning_rate": 9.762216979779412e-05,
"loss": 0.0273,
"step": 4310
},
{
"grad_norm": 0.5224273204803467,
"learning_rate": 9.760534626864902e-05,
"loss": 0.0278,
"step": 4320
},
{
"grad_norm": 0.503278374671936,
"learning_rate": 9.758846489451931e-05,
"loss": 0.0264,
"step": 4330
},
{
"grad_norm": 0.43135321140289307,
"learning_rate": 9.757152569591748e-05,
"loss": 0.0278,
"step": 4340
},
{
"grad_norm": 0.41968685388565063,
"learning_rate": 9.75545286934262e-05,
"loss": 0.0247,
"step": 4350
},
{
"grad_norm": 0.5300660729408264,
"learning_rate": 9.753747390769847e-05,
"loss": 0.0265,
"step": 4360
},
{
"grad_norm": 0.5508757829666138,
"learning_rate": 9.752036135945744e-05,
"loss": 0.026,
"step": 4370
},
{
"grad_norm": 0.48966965079307556,
"learning_rate": 9.750319106949649e-05,
"loss": 0.024,
"step": 4380
},
{
"grad_norm": 0.4271833002567291,
"learning_rate": 9.748596305867913e-05,
"loss": 0.0281,
"step": 4390
},
{
"grad_norm": 0.4649449586868286,
"learning_rate": 9.746867734793903e-05,
"loss": 0.0249,
"step": 4400
},
{
"grad_norm": 0.6460062265396118,
"learning_rate": 9.745133395827993e-05,
"loss": 0.0256,
"step": 4410
},
{
"grad_norm": 0.5220698714256287,
"learning_rate": 9.743393291077572e-05,
"loss": 0.0273,
"step": 4420
},
{
"grad_norm": 0.4801582396030426,
"learning_rate": 9.741647422657028e-05,
"loss": 0.0275,
"step": 4430
},
{
"grad_norm": 0.4439697563648224,
"learning_rate": 9.739895792687758e-05,
"loss": 0.0271,
"step": 4440
},
{
"grad_norm": 0.5416855812072754,
"learning_rate": 9.738138403298157e-05,
"loss": 0.0288,
"step": 4450
},
{
"grad_norm": 0.47156187891960144,
"learning_rate": 9.736375256623619e-05,
"loss": 0.0267,
"step": 4460
},
{
"grad_norm": 0.48665690422058105,
"learning_rate": 9.734606354806533e-05,
"loss": 0.0281,
"step": 4470
},
{
"grad_norm": 0.4060969650745392,
"learning_rate": 9.73283169999628e-05,
"loss": 0.0272,
"step": 4480
},
{
"grad_norm": 0.4355490505695343,
"learning_rate": 9.731051294349238e-05,
"loss": 0.0272,
"step": 4490
},
{
"grad_norm": 0.5726194381713867,
"learning_rate": 9.729265140028762e-05,
"loss": 0.0281,
"step": 4500
},
{
"grad_norm": 0.49670320749282837,
"learning_rate": 9.727473239205201e-05,
"loss": 0.0285,
"step": 4510
},
{
"grad_norm": 0.5777426958084106,
"learning_rate": 9.725675594055883e-05,
"loss": 0.0273,
"step": 4520
},
{
"grad_norm": 0.36380887031555176,
"learning_rate": 9.723872206765116e-05,
"loss": 0.0253,
"step": 4530
},
{
"grad_norm": 0.5655593872070312,
"learning_rate": 9.722063079524185e-05,
"loss": 0.0257,
"step": 4540
},
{
"grad_norm": 0.5297718644142151,
"learning_rate": 9.720248214531351e-05,
"loss": 0.0248,
"step": 4550
},
{
"grad_norm": 0.6011912226676941,
"learning_rate": 9.718427613991848e-05,
"loss": 0.0247,
"step": 4560
},
{
"grad_norm": 0.46993371844291687,
"learning_rate": 9.716601280117873e-05,
"loss": 0.0243,
"step": 4570
},
{
"grad_norm": 0.6736533045768738,
"learning_rate": 9.714769215128596e-05,
"loss": 0.0257,
"step": 4580
},
{
"grad_norm": 0.4797600209712982,
"learning_rate": 9.712931421250152e-05,
"loss": 0.0248,
"step": 4590
},
{
"grad_norm": 0.5109550356864929,
"learning_rate": 9.711087900715627e-05,
"loss": 0.0251,
"step": 4600
},
{
"grad_norm": 0.5943213701248169,
"learning_rate": 9.709238655765078e-05,
"loss": 0.0257,
"step": 4610
},
{
"grad_norm": 0.4454824924468994,
"learning_rate": 9.707383688645511e-05,
"loss": 0.0249,
"step": 4620
},
{
"grad_norm": 0.5341625213623047,
"learning_rate": 9.705523001610883e-05,
"loss": 0.0266,
"step": 4630
},
{
"grad_norm": 0.7142083048820496,
"learning_rate": 9.703656596922107e-05,
"loss": 0.0279,
"step": 4640
},
{
"grad_norm": 0.5267425775527954,
"learning_rate": 9.70178447684704e-05,
"loss": 0.0254,
"step": 4650
},
{
"grad_norm": 0.4902753531932831,
"learning_rate": 9.699906643660483e-05,
"loss": 0.0236,
"step": 4660
},
{
"grad_norm": 0.456157922744751,
"learning_rate": 9.698023099644185e-05,
"loss": 0.0262,
"step": 4670
},
{
"grad_norm": 0.5741419196128845,
"learning_rate": 9.696133847086823e-05,
"loss": 0.0256,
"step": 4680
},
{
"grad_norm": 0.5706208348274231,
"learning_rate": 9.694238888284022e-05,
"loss": 0.0237,
"step": 4690
},
{
"grad_norm": 0.5595664381980896,
"learning_rate": 9.692338225538333e-05,
"loss": 0.0243,
"step": 4700
},
{
"grad_norm": 0.6362490653991699,
"learning_rate": 9.690431861159241e-05,
"loss": 0.0243,
"step": 4710
},
{
"grad_norm": 0.49317440390586853,
"learning_rate": 9.688519797463161e-05,
"loss": 0.0241,
"step": 4720
},
{
"grad_norm": 0.5495986938476562,
"learning_rate": 9.686602036773426e-05,
"loss": 0.0238,
"step": 4730
},
{
"grad_norm": 0.5672884583473206,
"learning_rate": 9.684678581420302e-05,
"loss": 0.0232,
"step": 4740
},
{
"grad_norm": 0.3791441321372986,
"learning_rate": 9.682749433740962e-05,
"loss": 0.0214,
"step": 4750
},
{
"grad_norm": 0.635084867477417,
"learning_rate": 9.680814596079507e-05,
"loss": 0.024,
"step": 4760
},
{
"grad_norm": 0.44687172770500183,
"learning_rate": 9.678874070786945e-05,
"loss": 0.0222,
"step": 4770
},
{
"grad_norm": 0.5880815982818604,
"learning_rate": 9.676927860221199e-05,
"loss": 0.0231,
"step": 4780
},
{
"grad_norm": 0.4695528745651245,
"learning_rate": 9.674975966747097e-05,
"loss": 0.0243,
"step": 4790
},
{
"grad_norm": 0.5622825026512146,
"learning_rate": 9.673018392736374e-05,
"loss": 0.0229,
"step": 4800
},
{
"grad_norm": 0.48140478134155273,
"learning_rate": 9.671055140567667e-05,
"loss": 0.0217,
"step": 4810
},
{
"grad_norm": 0.4679363965988159,
"learning_rate": 9.669086212626511e-05,
"loss": 0.023,
"step": 4820
},
{
"grad_norm": 0.6131110787391663,
"learning_rate": 9.667111611305341e-05,
"loss": 0.0225,
"step": 4830
},
{
"grad_norm": 0.4789270758628845,
"learning_rate": 9.665131339003486e-05,
"loss": 0.0232,
"step": 4840
},
{
"grad_norm": 0.6402637362480164,
"learning_rate": 9.663145398127158e-05,
"loss": 0.0219,
"step": 4850
},
{
"grad_norm": 0.47417306900024414,
"learning_rate": 9.661153791089467e-05,
"loss": 0.0226,
"step": 4860
},
{
"grad_norm": 0.5275768637657166,
"learning_rate": 9.659156520310402e-05,
"loss": 0.0232,
"step": 4870
},
{
"grad_norm": 0.5174130797386169,
"learning_rate": 9.657153588216835e-05,
"loss": 0.024,
"step": 4880
},
{
"grad_norm": 0.5008310675621033,
"learning_rate": 9.655144997242516e-05,
"loss": 0.0239,
"step": 4890
},
{
"grad_norm": 0.508402943611145,
"learning_rate": 9.653130749828075e-05,
"loss": 0.023,
"step": 4900
},
{
"grad_norm": 0.4857371747493744,
"learning_rate": 9.65111084842101e-05,
"loss": 0.0207,
"step": 4910
},
{
"grad_norm": 0.5261239409446716,
"learning_rate": 9.649085295475695e-05,
"loss": 0.0222,
"step": 4920
},
{
"grad_norm": 0.4184090197086334,
"learning_rate": 9.647054093453365e-05,
"loss": 0.023,
"step": 4930
},
{
"grad_norm": 0.6007316708564758,
"learning_rate": 9.645017244822123e-05,
"loss": 0.0231,
"step": 4940
},
{
"grad_norm": 0.44578850269317627,
"learning_rate": 9.642974752056931e-05,
"loss": 0.0221,
"step": 4950
},
{
"grad_norm": 0.5718969106674194,
"learning_rate": 9.640926617639613e-05,
"loss": 0.0228,
"step": 4960
},
{
"grad_norm": 0.47065165638923645,
"learning_rate": 9.638872844058843e-05,
"loss": 0.0226,
"step": 4970
},
{
"grad_norm": 0.6126935482025146,
"learning_rate": 9.63681343381015e-05,
"loss": 0.0225,
"step": 4980
},
{
"grad_norm": 0.424108624458313,
"learning_rate": 9.634748389395914e-05,
"loss": 0.0213,
"step": 4990
},
{
"grad_norm": 0.5950810313224792,
"learning_rate": 9.632677713325353e-05,
"loss": 0.0207,
"step": 5000
},
{
"grad_norm": 0.4594224691390991,
"learning_rate": 9.63060140811454e-05,
"loss": 0.021,
"step": 5010
},
{
"grad_norm": 0.46782585978507996,
"learning_rate": 9.628519476286379e-05,
"loss": 0.0214,
"step": 5020
},
{
"grad_norm": 0.4983988404273987,
"learning_rate": 9.626431920370612e-05,
"loss": 0.0201,
"step": 5030
},
{
"grad_norm": 0.6005197167396545,
"learning_rate": 9.624338742903819e-05,
"loss": 0.0213,
"step": 5040
},
{
"grad_norm": 0.6117371320724487,
"learning_rate": 9.622239946429406e-05,
"loss": 0.0231,
"step": 5050
},
{
"grad_norm": 0.5228967070579529,
"learning_rate": 9.620135533497609e-05,
"loss": 0.0207,
"step": 5060
},
{
"grad_norm": 0.6595703959465027,
"learning_rate": 9.61802550666549e-05,
"loss": 0.0194,
"step": 5070
},
{
"grad_norm": 0.4813227653503418,
"learning_rate": 9.615909868496928e-05,
"loss": 0.0197,
"step": 5080
},
{
"grad_norm": 0.41596779227256775,
"learning_rate": 9.613788621562622e-05,
"loss": 0.0224,
"step": 5090
},
{
"grad_norm": 0.45717138051986694,
"learning_rate": 9.611661768440093e-05,
"loss": 0.0195,
"step": 5100
},
{
"grad_norm": 0.5034703016281128,
"learning_rate": 9.609529311713661e-05,
"loss": 0.0202,
"step": 5110
},
{
"grad_norm": 0.40311121940612793,
"learning_rate": 9.607391253974466e-05,
"loss": 0.0198,
"step": 5120
},
{
"grad_norm": 0.5923986434936523,
"learning_rate": 9.605247597820448e-05,
"loss": 0.0199,
"step": 5130
},
{
"grad_norm": 0.4329419732093811,
"learning_rate": 9.603098345856354e-05,
"loss": 0.0204,
"step": 5140
},
{
"grad_norm": 0.4753156006336212,
"learning_rate": 9.600943500693725e-05,
"loss": 0.0213,
"step": 5150
},
{
"grad_norm": 0.42589521408081055,
"learning_rate": 9.598783064950902e-05,
"loss": 0.0194,
"step": 5160
},
{
"grad_norm": 0.5389040112495422,
"learning_rate": 9.596617041253018e-05,
"loss": 0.0202,
"step": 5170
},
{
"grad_norm": 0.491046279668808,
"learning_rate": 9.594445432231996e-05,
"loss": 0.0187,
"step": 5180
},
{
"grad_norm": 0.429523229598999,
"learning_rate": 9.592268240526547e-05,
"loss": 0.0177,
"step": 5190
},
{
"grad_norm": 0.498996376991272,
"learning_rate": 9.590085468782162e-05,
"loss": 0.0212,
"step": 5200
},
{
"grad_norm": 0.4649933874607086,
"learning_rate": 9.587897119651116e-05,
"loss": 0.0188,
"step": 5210
},
{
"grad_norm": 0.38946202397346497,
"learning_rate": 9.585703195792459e-05,
"loss": 0.0172,
"step": 5220
},
{
"grad_norm": 0.5122786164283752,
"learning_rate": 9.583503699872016e-05,
"loss": 0.0183,
"step": 5230
},
{
"grad_norm": 0.37925073504447937,
"learning_rate": 9.581298634562381e-05,
"loss": 0.0189,
"step": 5240
},
{
"grad_norm": 0.3267979025840759,
"learning_rate": 9.579088002542917e-05,
"loss": 0.0214,
"step": 5250
},
{
"grad_norm": 0.5075067281723022,
"learning_rate": 9.57687180649975e-05,
"loss": 0.0173,
"step": 5260
},
{
"grad_norm": 0.5464387536048889,
"learning_rate": 9.574650049125768e-05,
"loss": 0.0203,
"step": 5270
},
{
"grad_norm": 0.49496564269065857,
"learning_rate": 9.572422733120614e-05,
"loss": 0.0191,
"step": 5280
},
{
"grad_norm": 0.4231927990913391,
"learning_rate": 9.570189861190689e-05,
"loss": 0.0182,
"step": 5290
},
{
"grad_norm": 0.469819039106369,
"learning_rate": 9.56795143604914e-05,
"loss": 0.0181,
"step": 5300
},
{
"grad_norm": 0.5372419953346252,
"learning_rate": 9.565707460415869e-05,
"loss": 0.0186,
"step": 5310
},
{
"grad_norm": 0.46158352494239807,
"learning_rate": 9.563457937017515e-05,
"loss": 0.0197,
"step": 5320
},
{
"grad_norm": 0.43332675099372864,
"learning_rate": 9.56120286858746e-05,
"loss": 0.0176,
"step": 5330
},
{
"grad_norm": 0.605982780456543,
"learning_rate": 9.558942257865829e-05,
"loss": 0.0178,
"step": 5340
},
{
"grad_norm": 0.555808424949646,
"learning_rate": 9.556676107599472e-05,
"loss": 0.0182,
"step": 5350
},
{
"grad_norm": 0.44589316844940186,
"learning_rate": 9.554404420541978e-05,
"loss": 0.0185,
"step": 5360
},
{
"grad_norm": 0.44258564710617065,
"learning_rate": 9.55212719945366e-05,
"loss": 0.0187,
"step": 5370
},
{
"grad_norm": 0.5140361785888672,
"learning_rate": 9.549844447101559e-05,
"loss": 0.0182,
"step": 5380
},
{
"grad_norm": 0.6253278255462646,
"learning_rate": 9.547556166259433e-05,
"loss": 0.0166,
"step": 5390
},
{
"grad_norm": 0.46718496084213257,
"learning_rate": 9.545262359707756e-05,
"loss": 0.0188,
"step": 5400
},
{
"grad_norm": 0.383916974067688,
"learning_rate": 9.542963030233724e-05,
"loss": 0.0184,
"step": 5410
},
{
"grad_norm": 0.4682512581348419,
"learning_rate": 9.540658180631237e-05,
"loss": 0.0171,
"step": 5420
},
{
"grad_norm": 0.4831123948097229,
"learning_rate": 9.538347813700904e-05,
"loss": 0.0178,
"step": 5430
},
{
"grad_norm": 0.4267755150794983,
"learning_rate": 9.536031932250036e-05,
"loss": 0.0177,
"step": 5440
},
{
"grad_norm": 0.47811415791511536,
"learning_rate": 9.533710539092653e-05,
"loss": 0.017,
"step": 5450
},
{
"grad_norm": 0.5367674231529236,
"learning_rate": 9.531383637049464e-05,
"loss": 0.0173,
"step": 5460
},
{
"grad_norm": 0.45475828647613525,
"learning_rate": 9.529051228947875e-05,
"loss": 0.017,
"step": 5470
},
{
"grad_norm": 0.5281142592430115,
"learning_rate": 9.52671331762198e-05,
"loss": 0.0165,
"step": 5480
},
{
"grad_norm": 0.45434918999671936,
"learning_rate": 9.524369905912565e-05,
"loss": 0.0183,
"step": 5490
},
{
"grad_norm": 0.5526027083396912,
"learning_rate": 9.522020996667092e-05,
"loss": 0.0181,
"step": 5500
},
{
"grad_norm": 0.42288097739219666,
"learning_rate": 9.519666592739709e-05,
"loss": 0.0171,
"step": 5510
},
{
"grad_norm": 0.502322256565094,
"learning_rate": 9.517306696991241e-05,
"loss": 0.0163,
"step": 5520
},
{
"grad_norm": 0.4991615116596222,
"learning_rate": 9.51494131228918e-05,
"loss": 0.0183,
"step": 5530
},
{
"grad_norm": 0.46957167983055115,
"learning_rate": 9.512570441507695e-05,
"loss": 0.016,
"step": 5540
},
{
"grad_norm": 0.5327632427215576,
"learning_rate": 9.510194087527615e-05,
"loss": 0.0166,
"step": 5550
},
{
"grad_norm": 0.4170159697532654,
"learning_rate": 9.507812253236435e-05,
"loss": 0.017,
"step": 5560
},
{
"grad_norm": 0.4968635141849518,
"learning_rate": 9.505424941528309e-05,
"loss": 0.0164,
"step": 5570
},
{
"grad_norm": 0.4826819896697998,
"learning_rate": 9.503032155304046e-05,
"loss": 0.0162,
"step": 5580
},
{
"grad_norm": 0.3677280843257904,
"learning_rate": 9.500633897471106e-05,
"loss": 0.0164,
"step": 5590
},
{
"grad_norm": 0.4160604178905487,
"learning_rate": 9.498230170943596e-05,
"loss": 0.018,
"step": 5600
},
{
"grad_norm": 0.45811036229133606,
"learning_rate": 9.495820978642275e-05,
"loss": 0.0163,
"step": 5610
},
{
"grad_norm": 0.42683902382850647,
"learning_rate": 9.493406323494535e-05,
"loss": 0.0178,
"step": 5620
},
{
"grad_norm": 0.4382336735725403,
"learning_rate": 9.490986208434413e-05,
"loss": 0.0156,
"step": 5630
},
{
"grad_norm": 0.3592792749404907,
"learning_rate": 9.488560636402577e-05,
"loss": 0.0145,
"step": 5640
},
{
"grad_norm": 0.34286990761756897,
"learning_rate": 9.486129610346321e-05,
"loss": 0.0146,
"step": 5650
},
{
"grad_norm": 0.49023711681365967,
"learning_rate": 9.483693133219576e-05,
"loss": 0.0153,
"step": 5660
},
{
"grad_norm": 0.3683493435382843,
"learning_rate": 9.481251207982888e-05,
"loss": 0.0162,
"step": 5670
},
{
"grad_norm": 0.4639711380004883,
"learning_rate": 9.47880383760343e-05,
"loss": 0.0179,
"step": 5680
},
{
"grad_norm": 0.3855854868888855,
"learning_rate": 9.476351025054983e-05,
"loss": 0.0173,
"step": 5690
},
{
"grad_norm": 0.46952077746391296,
"learning_rate": 9.473892773317952e-05,
"loss": 0.0167,
"step": 5700
},
{
"grad_norm": 0.46543654799461365,
"learning_rate": 9.471429085379338e-05,
"loss": 0.0162,
"step": 5710
},
{
"grad_norm": 0.43368780612945557,
"learning_rate": 9.468959964232757e-05,
"loss": 0.015,
"step": 5720
},
{
"grad_norm": 0.6242442727088928,
"learning_rate": 9.466485412878425e-05,
"loss": 0.0163,
"step": 5730
},
{
"grad_norm": 0.46993181109428406,
"learning_rate": 9.464005434323154e-05,
"loss": 0.0147,
"step": 5740
},
{
"grad_norm": 0.4298034608364105,
"learning_rate": 9.461520031580352e-05,
"loss": 0.0149,
"step": 5750
},
{
"grad_norm": 0.679725170135498,
"learning_rate": 9.459029207670019e-05,
"loss": 0.0161,
"step": 5760
},
{
"grad_norm": 0.3906918466091156,
"learning_rate": 9.456532965618737e-05,
"loss": 0.0185,
"step": 5770
},
{
"grad_norm": 0.4709409177303314,
"learning_rate": 9.454031308459681e-05,
"loss": 0.017,
"step": 5780
},
{
"grad_norm": 0.4399077594280243,
"learning_rate": 9.451524239232595e-05,
"loss": 0.0165,
"step": 5790
},
{
"grad_norm": 0.4794357717037201,
"learning_rate": 9.449011760983809e-05,
"loss": 0.0163,
"step": 5800
},
{
"grad_norm": 0.37406089901924133,
"learning_rate": 9.446493876766218e-05,
"loss": 0.0192,
"step": 5810
},
{
"grad_norm": 0.4109271764755249,
"learning_rate": 9.44397058963929e-05,
"loss": 0.0185,
"step": 5820
},
{
"grad_norm": 0.42586711049079895,
"learning_rate": 9.441441902669056e-05,
"loss": 0.0161,
"step": 5830
},
{
"grad_norm": 0.35982322692871094,
"learning_rate": 9.43890781892811e-05,
"loss": 0.0152,
"step": 5840
},
{
"grad_norm": 0.302796870470047,
"learning_rate": 9.436368341495603e-05,
"loss": 0.0159,
"step": 5850
},
{
"grad_norm": 0.4041033685207367,
"learning_rate": 9.43382347345724e-05,
"loss": 0.0155,
"step": 5860
},
{
"grad_norm": 0.5660633444786072,
"learning_rate": 9.431273217905273e-05,
"loss": 0.0167,
"step": 5870
},
{
"grad_norm": 0.40607672929763794,
"learning_rate": 9.428717577938504e-05,
"loss": 0.0156,
"step": 5880
},
{
"grad_norm": 0.4061409831047058,
"learning_rate": 9.426156556662276e-05,
"loss": 0.014,
"step": 5890
},
{
"grad_norm": 0.4410693347454071,
"learning_rate": 9.423590157188474e-05,
"loss": 0.016,
"step": 5900
},
{
"grad_norm": 0.4097030758857727,
"learning_rate": 9.421018382635513e-05,
"loss": 0.0161,
"step": 5910
},
{
"grad_norm": 0.4834526479244232,
"learning_rate": 9.418441236128343e-05,
"loss": 0.0159,
"step": 5920
},
{
"grad_norm": 0.36734217405319214,
"learning_rate": 9.41585872079844e-05,
"loss": 0.0145,
"step": 5930
},
{
"grad_norm": 0.4003128707408905,
"learning_rate": 9.413270839783802e-05,
"loss": 0.0164,
"step": 5940
},
{
"grad_norm": 0.4849574863910675,
"learning_rate": 9.41067759622895e-05,
"loss": 0.0151,
"step": 5950
},
{
"grad_norm": 0.40424999594688416,
"learning_rate": 9.408078993284917e-05,
"loss": 0.0159,
"step": 5960
},
{
"grad_norm": 0.49761006236076355,
"learning_rate": 9.405475034109254e-05,
"loss": 0.0189,
"step": 5970
},
{
"grad_norm": 0.36773961782455444,
"learning_rate": 9.402865721866015e-05,
"loss": 0.0169,
"step": 5980
},
{
"grad_norm": 0.3295658826828003,
"learning_rate": 9.400251059725762e-05,
"loss": 0.0153,
"step": 5990
},
{
"grad_norm": 0.45937928557395935,
"learning_rate": 9.397631050865554e-05,
"loss": 0.0157,
"step": 6000
},
{
"grad_norm": 0.4183763861656189,
"learning_rate": 9.395005698468949e-05,
"loss": 0.0164,
"step": 6010
},
{
"grad_norm": 0.33879679441452026,
"learning_rate": 9.392375005726e-05,
"loss": 0.0137,
"step": 6020
},
{
"grad_norm": 0.37298041582107544,
"learning_rate": 9.389738975833242e-05,
"loss": 0.0147,
"step": 6030
},
{
"grad_norm": 0.4308323264122009,
"learning_rate": 9.387097611993707e-05,
"loss": 0.0144,
"step": 6040
},
{
"grad_norm": 0.437456339597702,
"learning_rate": 9.384450917416894e-05,
"loss": 0.0148,
"step": 6050
},
{
"grad_norm": 0.4685937166213989,
"learning_rate": 9.381798895318792e-05,
"loss": 0.0145,
"step": 6060
},
{
"grad_norm": 0.37430495023727417,
"learning_rate": 9.379141548921853e-05,
"loss": 0.0147,
"step": 6070
},
{
"grad_norm": 0.45154932141304016,
"learning_rate": 9.376478881455009e-05,
"loss": 0.0151,
"step": 6080
},
{
"grad_norm": 0.4732506573200226,
"learning_rate": 9.373810896153647e-05,
"loss": 0.0146,
"step": 6090
},
{
"grad_norm": 0.46154171228408813,
"learning_rate": 9.371137596259623e-05,
"loss": 0.0144,
"step": 6100
},
{
"grad_norm": 0.387769877910614,
"learning_rate": 9.368458985021248e-05,
"loss": 0.0144,
"step": 6110
},
{
"grad_norm": 0.3920918107032776,
"learning_rate": 9.365775065693287e-05,
"loss": 0.0164,
"step": 6120
},
{
"grad_norm": 0.43867599964141846,
"learning_rate": 9.363085841536957e-05,
"loss": 0.0135,
"step": 6130
},
{
"grad_norm": 0.44755738973617554,
"learning_rate": 9.360391315819917e-05,
"loss": 0.0149,
"step": 6140
},
{
"grad_norm": 0.48710331320762634,
"learning_rate": 9.357691491816269e-05,
"loss": 0.0141,
"step": 6150
},
{
"grad_norm": 0.41324517130851746,
"learning_rate": 9.354986372806557e-05,
"loss": 0.0149,
"step": 6160
},
{
"grad_norm": 0.3770032525062561,
"learning_rate": 9.352275962077752e-05,
"loss": 0.0138,
"step": 6170
},
{
"grad_norm": 0.44077855348587036,
"learning_rate": 9.349560262923262e-05,
"loss": 0.0152,
"step": 6180
},
{
"grad_norm": 0.40747708082199097,
"learning_rate": 9.346839278642913e-05,
"loss": 0.0133,
"step": 6190
},
{
"grad_norm": 0.4391169548034668,
"learning_rate": 9.344113012542963e-05,
"loss": 0.0145,
"step": 6200
},
{
"grad_norm": 0.4199869930744171,
"learning_rate": 9.341381467936079e-05,
"loss": 0.0133,
"step": 6210
},
{
"grad_norm": 0.4460112750530243,
"learning_rate": 9.338644648141346e-05,
"loss": 0.0152,
"step": 6220
},
{
"grad_norm": 0.30460286140441895,
"learning_rate": 9.335902556484257e-05,
"loss": 0.0131,
"step": 6230
},
{
"grad_norm": 0.37233051657676697,
"learning_rate": 9.333155196296712e-05,
"loss": 0.0141,
"step": 6240
},
{
"grad_norm": 0.3617931306362152,
"learning_rate": 9.330402570917016e-05,
"loss": 0.0146,
"step": 6250
},
{
"grad_norm": 0.3526821732521057,
"learning_rate": 9.327644683689865e-05,
"loss": 0.014,
"step": 6260
},
{
"grad_norm": 0.45122718811035156,
"learning_rate": 9.324881537966354e-05,
"loss": 0.0135,
"step": 6270
},
{
"grad_norm": 0.5205270648002625,
"learning_rate": 9.322113137103964e-05,
"loss": 0.0154,
"step": 6280
},
{
"grad_norm": 0.3603147566318512,
"learning_rate": 9.319339484466564e-05,
"loss": 0.0166,
"step": 6290
},
{
"grad_norm": 0.41952037811279297,
"learning_rate": 9.316560583424403e-05,
"loss": 0.0134,
"step": 6300
},
{
"grad_norm": 0.36976832151412964,
"learning_rate": 9.313776437354108e-05,
"loss": 0.0125,
"step": 6310
},
{
"grad_norm": 0.4265200197696686,
"learning_rate": 9.31098704963868e-05,
"loss": 0.0147,
"step": 6320
},
{
"grad_norm": 0.34672072529792786,
"learning_rate": 9.308192423667485e-05,
"loss": 0.0133,
"step": 6330
},
{
"grad_norm": 0.3036075532436371,
"learning_rate": 9.305392562836262e-05,
"loss": 0.0134,
"step": 6340
},
{
"grad_norm": 0.3667495846748352,
"learning_rate": 9.3025874705471e-05,
"loss": 0.0139,
"step": 6350
},
{
"grad_norm": 0.41114041209220886,
"learning_rate": 9.299777150208457e-05,
"loss": 0.0147,
"step": 6360
},
{
"grad_norm": 0.42483291029930115,
"learning_rate": 9.296961605235133e-05,
"loss": 0.0143,
"step": 6370
},
{
"grad_norm": 0.4150006175041199,
"learning_rate": 9.29414083904828e-05,
"loss": 0.0154,
"step": 6380
},
{
"grad_norm": 0.4819064736366272,
"learning_rate": 9.291314855075397e-05,
"loss": 0.0145,
"step": 6390
},
{
"grad_norm": 0.41249290108680725,
"learning_rate": 9.288483656750322e-05,
"loss": 0.0155,
"step": 6400
},
{
"grad_norm": 0.39605265855789185,
"learning_rate": 9.285647247513225e-05,
"loss": 0.0134,
"step": 6410
},
{
"grad_norm": 0.35166269540786743,
"learning_rate": 9.282805630810614e-05,
"loss": 0.0135,
"step": 6420
},
{
"grad_norm": 0.4163077771663666,
"learning_rate": 9.279958810095317e-05,
"loss": 0.0145,
"step": 6430
},
{
"grad_norm": 0.3122580349445343,
"learning_rate": 9.277106788826494e-05,
"loss": 0.0135,
"step": 6440
},
{
"grad_norm": 0.5408252477645874,
"learning_rate": 9.274249570469618e-05,
"loss": 0.0136,
"step": 6450
},
{
"grad_norm": 0.3665207624435425,
"learning_rate": 9.271387158496476e-05,
"loss": 0.0136,
"step": 6460
},
{
"grad_norm": 0.38548943400382996,
"learning_rate": 9.268519556385173e-05,
"loss": 0.0148,
"step": 6470
},
{
"grad_norm": 0.4338033199310303,
"learning_rate": 9.265646767620113e-05,
"loss": 0.0138,
"step": 6480
},
{
"grad_norm": 0.3697988986968994,
"learning_rate": 9.262768795692006e-05,
"loss": 0.0134,
"step": 6490
},
{
"grad_norm": 0.43436676263809204,
"learning_rate": 9.259885644097861e-05,
"loss": 0.0145,
"step": 6500
},
{
"grad_norm": 0.35602453351020813,
"learning_rate": 9.256997316340976e-05,
"loss": 0.0139,
"step": 6510
},
{
"grad_norm": 0.3094992935657501,
"learning_rate": 9.254103815930943e-05,
"loss": 0.0131,
"step": 6520
},
{
"grad_norm": 0.39769303798675537,
"learning_rate": 9.251205146383636e-05,
"loss": 0.0139,
"step": 6530
},
{
"grad_norm": 0.3360579013824463,
"learning_rate": 9.248301311221216e-05,
"loss": 0.0131,
"step": 6540
},
{
"grad_norm": 0.43902263045310974,
"learning_rate": 9.245392313972115e-05,
"loss": 0.0141,
"step": 6550
},
{
"grad_norm": 0.3751843273639679,
"learning_rate": 9.242478158171037e-05,
"loss": 0.0139,
"step": 6560
},
{
"grad_norm": 0.34897536039352417,
"learning_rate": 9.239558847358957e-05,
"loss": 0.0128,
"step": 6570
},
{
"grad_norm": 0.3229854106903076,
"learning_rate": 9.236634385083114e-05,
"loss": 0.013,
"step": 6580
},
{
"grad_norm": 0.33645179867744446,
"learning_rate": 9.233704774897005e-05,
"loss": 0.013,
"step": 6590
},
{
"grad_norm": 0.36346444487571716,
"learning_rate": 9.230770020360382e-05,
"loss": 0.013,
"step": 6600
},
{
"grad_norm": 0.38480672240257263,
"learning_rate": 9.227830125039247e-05,
"loss": 0.0129,
"step": 6610
},
{
"grad_norm": 0.42673030495643616,
"learning_rate": 9.224885092505853e-05,
"loss": 0.0133,
"step": 6620
},
{
"grad_norm": 0.4402414560317993,
"learning_rate": 9.221934926338691e-05,
"loss": 0.0139,
"step": 6630
},
{
"grad_norm": 0.39116808772087097,
"learning_rate": 9.21897963012249e-05,
"loss": 0.0133,
"step": 6640
},
{
"grad_norm": 0.28943920135498047,
"learning_rate": 9.216019207448217e-05,
"loss": 0.0147,
"step": 6650
},
{
"grad_norm": 0.3455876410007477,
"learning_rate": 9.21305366191306e-05,
"loss": 0.0147,
"step": 6660
},
{
"grad_norm": 0.44500064849853516,
"learning_rate": 9.21008299712044e-05,
"loss": 0.0137,
"step": 6670
},
{
"grad_norm": 0.5418192744255066,
"learning_rate": 9.207107216679993e-05,
"loss": 0.0141,
"step": 6680
},
{
"grad_norm": 0.44647619128227234,
"learning_rate": 9.204126324207574e-05,
"loss": 0.0134,
"step": 6690
},
{
"grad_norm": 0.33688196539878845,
"learning_rate": 9.201140323325248e-05,
"loss": 0.0137,
"step": 6700
},
{
"grad_norm": 0.4310382008552551,
"learning_rate": 9.198149217661287e-05,
"loss": 0.0142,
"step": 6710
},
{
"grad_norm": 0.39126962423324585,
"learning_rate": 9.195153010850167e-05,
"loss": 0.0151,
"step": 6720
},
{
"grad_norm": 0.46993929147720337,
"learning_rate": 9.192151706532562e-05,
"loss": 0.0133,
"step": 6730
},
{
"grad_norm": 0.4224424362182617,
"learning_rate": 9.189145308355337e-05,
"loss": 0.0146,
"step": 6740
},
{
"grad_norm": 0.3198283910751343,
"learning_rate": 9.186133819971555e-05,
"loss": 0.0138,
"step": 6750
},
{
"grad_norm": 0.45175471901893616,
"learning_rate": 9.183117245040455e-05,
"loss": 0.0133,
"step": 6760
},
{
"grad_norm": 0.3550010323524475,
"learning_rate": 9.180095587227458e-05,
"loss": 0.0133,
"step": 6770
},
{
"grad_norm": 0.36675015091896057,
"learning_rate": 9.177068850204167e-05,
"loss": 0.0144,
"step": 6780
},
{
"grad_norm": 0.322945773601532,
"learning_rate": 9.174037037648351e-05,
"loss": 0.0134,
"step": 6790
},
{
"grad_norm": 0.4099406898021698,
"learning_rate": 9.171000153243947e-05,
"loss": 0.0139,
"step": 6800
},
{
"grad_norm": 0.32047194242477417,
"learning_rate": 9.167958200681057e-05,
"loss": 0.012,
"step": 6810
},
{
"grad_norm": 0.3311690390110016,
"learning_rate": 9.164911183655943e-05,
"loss": 0.0132,
"step": 6820
},
{
"grad_norm": 0.34048816561698914,
"learning_rate": 9.161859105871013e-05,
"loss": 0.0135,
"step": 6830
},
{
"grad_norm": 0.3637918531894684,
"learning_rate": 9.158801971034832e-05,
"loss": 0.0138,
"step": 6840
},
{
"grad_norm": 0.3191274404525757,
"learning_rate": 9.155739782862107e-05,
"loss": 0.0132,
"step": 6850
},
{
"grad_norm": 0.30677056312561035,
"learning_rate": 9.152672545073687e-05,
"loss": 0.0121,
"step": 6860
},
{
"grad_norm": 0.327228844165802,
"learning_rate": 9.149600261396551e-05,
"loss": 0.0139,
"step": 6870
},
{
"grad_norm": 0.4942983090877533,
"learning_rate": 9.146522935563817e-05,
"loss": 0.0145,
"step": 6880
},
{
"grad_norm": 0.43315163254737854,
"learning_rate": 9.143440571314723e-05,
"loss": 0.0119,
"step": 6890
},
{
"grad_norm": 0.27714836597442627,
"learning_rate": 9.140353172394636e-05,
"loss": 0.0119,
"step": 6900
},
{
"grad_norm": 0.2607634663581848,
"learning_rate": 9.137260742555032e-05,
"loss": 0.0123,
"step": 6910
},
{
"grad_norm": 0.40268388390541077,
"learning_rate": 9.13416328555351e-05,
"loss": 0.0133,
"step": 6920
},
{
"grad_norm": 0.35649073123931885,
"learning_rate": 9.131060805153769e-05,
"loss": 0.0118,
"step": 6930
},
{
"grad_norm": 0.389036625623703,
"learning_rate": 9.127953305125618e-05,
"loss": 0.0136,
"step": 6940
},
{
"grad_norm": 0.40989917516708374,
"learning_rate": 9.124840789244958e-05,
"loss": 0.0122,
"step": 6950
},
{
"grad_norm": 0.40351611375808716,
"learning_rate": 9.121723261293794e-05,
"loss": 0.0135,
"step": 6960
},
{
"grad_norm": 0.3241773843765259,
"learning_rate": 9.118600725060214e-05,
"loss": 0.0117,
"step": 6970
},
{
"grad_norm": 0.3243855834007263,
"learning_rate": 9.115473184338392e-05,
"loss": 0.0128,
"step": 6980
},
{
"grad_norm": 0.4551032781600952,
"learning_rate": 9.11234064292859e-05,
"loss": 0.0135,
"step": 6990
},
{
"grad_norm": 0.3283938765525818,
"learning_rate": 9.109203104637137e-05,
"loss": 0.0118,
"step": 7000
},
{
"grad_norm": 0.5484637022018433,
"learning_rate": 9.10606057327644e-05,
"loss": 0.0136,
"step": 7010
},
{
"grad_norm": 0.39262712001800537,
"learning_rate": 9.10291305266497e-05,
"loss": 0.0123,
"step": 7020
},
{
"grad_norm": 0.39485543966293335,
"learning_rate": 9.099760546627261e-05,
"loss": 0.0133,
"step": 7030
},
{
"grad_norm": 0.4103463590145111,
"learning_rate": 9.096603058993907e-05,
"loss": 0.011,
"step": 7040
},
{
"grad_norm": 0.445277601480484,
"learning_rate": 9.093440593601552e-05,
"loss": 0.0117,
"step": 7050
},
{
"grad_norm": 0.48510605096817017,
"learning_rate": 9.090273154292889e-05,
"loss": 0.013,
"step": 7060
},
{
"grad_norm": 0.3229779899120331,
"learning_rate": 9.087100744916656e-05,
"loss": 0.0121,
"step": 7070
},
{
"grad_norm": 0.34501615166664124,
"learning_rate": 9.08392336932763e-05,
"loss": 0.012,
"step": 7080
},
{
"grad_norm": 0.2764420211315155,
"learning_rate": 9.080741031386619e-05,
"loss": 0.0133,
"step": 7090
},
{
"grad_norm": 0.3332461416721344,
"learning_rate": 9.077553734960468e-05,
"loss": 0.0128,
"step": 7100
},
{
"grad_norm": 0.3700748383998871,
"learning_rate": 9.074361483922041e-05,
"loss": 0.0129,
"step": 7110
},
{
"grad_norm": 0.39643654227256775,
"learning_rate": 9.071164282150224e-05,
"loss": 0.0125,
"step": 7120
},
{
"grad_norm": 0.2998582720756531,
"learning_rate": 9.067962133529919e-05,
"loss": 0.0122,
"step": 7130
},
{
"grad_norm": 0.29886800050735474,
"learning_rate": 9.064755041952036e-05,
"loss": 0.0118,
"step": 7140
},
{
"grad_norm": 0.3528709411621094,
"learning_rate": 9.061543011313497e-05,
"loss": 0.0118,
"step": 7150
},
{
"grad_norm": 0.3907919228076935,
"learning_rate": 9.05832604551722e-05,
"loss": 0.0135,
"step": 7160
},
{
"grad_norm": 0.276755690574646,
"learning_rate": 9.055104148472124e-05,
"loss": 0.0119,
"step": 7170
},
{
"grad_norm": 0.37680584192276,
"learning_rate": 9.051877324093114e-05,
"loss": 0.0114,
"step": 7180
},
{
"grad_norm": 0.36407947540283203,
"learning_rate": 9.04864557630109e-05,
"loss": 0.0116,
"step": 7190
},
{
"grad_norm": 0.4268858730792999,
"learning_rate": 9.045408909022928e-05,
"loss": 0.0125,
"step": 7200
},
{
"grad_norm": 0.34331417083740234,
"learning_rate": 9.042167326191484e-05,
"loss": 0.0108,
"step": 7210
},
{
"grad_norm": 0.42327192425727844,
"learning_rate": 9.038920831745587e-05,
"loss": 0.0121,
"step": 7220
},
{
"grad_norm": 0.2950005531311035,
"learning_rate": 9.035669429630036e-05,
"loss": 0.0122,
"step": 7230
},
{
"grad_norm": 0.31393739581108093,
"learning_rate": 9.032413123795588e-05,
"loss": 0.0113,
"step": 7240
},
{
"grad_norm": 0.290615051984787,
"learning_rate": 9.029151918198962e-05,
"loss": 0.0124,
"step": 7250
},
{
"grad_norm": 0.4486510753631592,
"learning_rate": 9.025885816802833e-05,
"loss": 0.0117,
"step": 7260
},
{
"grad_norm": 0.3639298379421234,
"learning_rate": 9.02261482357582e-05,
"loss": 0.0114,
"step": 7270
},
{
"grad_norm": 0.3103725016117096,
"learning_rate": 9.019338942492485e-05,
"loss": 0.0136,
"step": 7280
},
{
"grad_norm": 0.39008089900016785,
"learning_rate": 9.01605817753334e-05,
"loss": 0.0116,
"step": 7290
},
{
"grad_norm": 0.37852030992507935,
"learning_rate": 9.012772532684818e-05,
"loss": 0.0124,
"step": 7300
},
{
"grad_norm": 0.29308199882507324,
"learning_rate": 9.00948201193929e-05,
"loss": 0.0118,
"step": 7310
},
{
"grad_norm": 0.42500293254852295,
"learning_rate": 9.006186619295047e-05,
"loss": 0.012,
"step": 7320
},
{
"grad_norm": 0.34131696820259094,
"learning_rate": 9.002886358756305e-05,
"loss": 0.0115,
"step": 7330
},
{
"grad_norm": 0.36960646510124207,
"learning_rate": 8.999581234333188e-05,
"loss": 0.0125,
"step": 7340
},
{
"grad_norm": 0.34096142649650574,
"learning_rate": 8.996271250041734e-05,
"loss": 0.0116,
"step": 7350
},
{
"grad_norm": 0.33445802330970764,
"learning_rate": 8.99295640990389e-05,
"loss": 0.011,
"step": 7360
},
{
"grad_norm": 0.28105318546295166,
"learning_rate": 8.989636717947494e-05,
"loss": 0.0115,
"step": 7370
},
{
"grad_norm": 0.3778381943702698,
"learning_rate": 8.986312178206291e-05,
"loss": 0.0123,
"step": 7380
},
{
"grad_norm": 0.3436017334461212,
"learning_rate": 8.982982794719904e-05,
"loss": 0.0114,
"step": 7390
},
{
"grad_norm": 0.33777639269828796,
"learning_rate": 8.979648571533852e-05,
"loss": 0.0124,
"step": 7400
},
{
"grad_norm": 0.39780232310295105,
"learning_rate": 8.976309512699528e-05,
"loss": 0.0105,
"step": 7410
},
{
"grad_norm": 0.336264044046402,
"learning_rate": 8.972965622274205e-05,
"loss": 0.0113,
"step": 7420
},
{
"grad_norm": 0.4956374168395996,
"learning_rate": 8.969616904321026e-05,
"loss": 0.011,
"step": 7430
},
{
"grad_norm": 0.4496384561061859,
"learning_rate": 8.966263362908998e-05,
"loss": 0.0118,
"step": 7440
},
{
"grad_norm": 0.3216903805732727,
"learning_rate": 8.962905002112989e-05,
"loss": 0.012,
"step": 7450
},
{
"grad_norm": 0.3218890130519867,
"learning_rate": 8.959541826013724e-05,
"loss": 0.0124,
"step": 7460
},
{
"grad_norm": 0.35675048828125,
"learning_rate": 8.956173838697781e-05,
"loss": 0.0128,
"step": 7470
},
{
"grad_norm": 0.32960838079452515,
"learning_rate": 8.952801044257581e-05,
"loss": 0.0111,
"step": 7480
},
{
"grad_norm": 0.35388556122779846,
"learning_rate": 8.949423446791387e-05,
"loss": 0.0119,
"step": 7490
},
{
"grad_norm": 0.32837608456611633,
"learning_rate": 8.9460410504033e-05,
"loss": 0.0115,
"step": 7500
},
{
"grad_norm": 0.31394892930984497,
"learning_rate": 8.942653859203248e-05,
"loss": 0.0136,
"step": 7510
},
{
"grad_norm": 0.28792908787727356,
"learning_rate": 8.93926187730699e-05,
"loss": 0.0107,
"step": 7520
},
{
"grad_norm": 0.2549435794353485,
"learning_rate": 8.935865108836102e-05,
"loss": 0.0124,
"step": 7530
},
{
"grad_norm": 0.3519458770751953,
"learning_rate": 8.932463557917981e-05,
"loss": 0.0142,
"step": 7540
},
{
"grad_norm": 0.34155184030532837,
"learning_rate": 8.929057228685829e-05,
"loss": 0.0135,
"step": 7550
},
{
"grad_norm": 0.37917962670326233,
"learning_rate": 8.925646125278658e-05,
"loss": 0.0124,
"step": 7560
},
{
"grad_norm": 0.3055376410484314,
"learning_rate": 8.92223025184128e-05,
"loss": 0.0111,
"step": 7570
},
{
"grad_norm": 0.2570594549179077,
"learning_rate": 8.918809612524303e-05,
"loss": 0.0111,
"step": 7580
},
{
"grad_norm": 0.37878426909446716,
"learning_rate": 8.91538421148413e-05,
"loss": 0.0112,
"step": 7590
},
{
"grad_norm": 0.36314377188682556,
"learning_rate": 8.91195405288294e-05,
"loss": 0.0131,
"step": 7600
},
{
"grad_norm": 0.3628675639629364,
"learning_rate": 8.908519140888705e-05,
"loss": 0.0108,
"step": 7610
},
{
"grad_norm": 0.35584089159965515,
"learning_rate": 8.905079479675164e-05,
"loss": 0.0121,
"step": 7620
},
{
"grad_norm": 0.40685468912124634,
"learning_rate": 8.901635073421831e-05,
"loss": 0.0119,
"step": 7630
},
{
"grad_norm": 0.344491183757782,
"learning_rate": 8.898185926313983e-05,
"loss": 0.0127,
"step": 7640
},
{
"grad_norm": 0.34751275181770325,
"learning_rate": 8.894732042542659e-05,
"loss": 0.013,
"step": 7650
},
{
"grad_norm": 0.2718348801136017,
"learning_rate": 8.891273426304654e-05,
"loss": 0.0098,
"step": 7660
},
{
"grad_norm": 0.35238078236579895,
"learning_rate": 8.887810081802515e-05,
"loss": 0.0113,
"step": 7670
},
{
"grad_norm": 0.3028378486633301,
"learning_rate": 8.884342013244529e-05,
"loss": 0.0119,
"step": 7680
},
{
"grad_norm": 0.3731449544429779,
"learning_rate": 8.880869224844725e-05,
"loss": 0.0123,
"step": 7690
},
{
"grad_norm": 0.3167056739330292,
"learning_rate": 8.877391720822875e-05,
"loss": 0.0132,
"step": 7700
},
{
"grad_norm": 0.3493010699748993,
"learning_rate": 8.873909505404468e-05,
"loss": 0.012,
"step": 7710
},
{
"grad_norm": 0.34445878863334656,
"learning_rate": 8.870422582820726e-05,
"loss": 0.0126,
"step": 7720
},
{
"grad_norm": 0.33221665024757385,
"learning_rate": 8.866930957308588e-05,
"loss": 0.0106,
"step": 7730
},
{
"grad_norm": 0.334201842546463,
"learning_rate": 8.86343463311071e-05,
"loss": 0.0114,
"step": 7740
},
{
"grad_norm": 0.3063510060310364,
"learning_rate": 8.859933614475452e-05,
"loss": 0.0131,
"step": 7750
},
{
"grad_norm": 0.5343878865242004,
"learning_rate": 8.85642790565689e-05,
"loss": 0.0123,
"step": 7760
},
{
"grad_norm": 0.32829737663269043,
"learning_rate": 8.852917510914782e-05,
"loss": 0.0112,
"step": 7770
},
{
"grad_norm": 0.31713294982910156,
"learning_rate": 8.84940243451459e-05,
"loss": 0.0129,
"step": 7780
},
{
"grad_norm": 0.44927090406417847,
"learning_rate": 8.845882680727469e-05,
"loss": 0.0127,
"step": 7790
},
{
"grad_norm": 0.29228851199150085,
"learning_rate": 8.842358253830246e-05,
"loss": 0.0115,
"step": 7800
},
{
"grad_norm": 0.3498315215110779,
"learning_rate": 8.838829158105434e-05,
"loss": 0.0121,
"step": 7810
},
{
"grad_norm": 0.3420019745826721,
"learning_rate": 8.835295397841217e-05,
"loss": 0.0115,
"step": 7820
},
{
"grad_norm": 0.30869531631469727,
"learning_rate": 8.831756977331447e-05,
"loss": 0.0117,
"step": 7830
},
{
"grad_norm": 0.37823766469955444,
"learning_rate": 8.828213900875638e-05,
"loss": 0.0108,
"step": 7840
},
{
"grad_norm": 0.3189360499382019,
"learning_rate": 8.824666172778964e-05,
"loss": 0.0104,
"step": 7850
},
{
"grad_norm": 0.476632297039032,
"learning_rate": 8.821113797352245e-05,
"loss": 0.0118,
"step": 7860
},
{
"grad_norm": 0.2909100353717804,
"learning_rate": 8.817556778911956e-05,
"loss": 0.0112,
"step": 7870
},
{
"grad_norm": 0.3038470149040222,
"learning_rate": 8.81399512178021e-05,
"loss": 0.012,
"step": 7880
},
{
"grad_norm": 0.3964444696903229,
"learning_rate": 8.810428830284752e-05,
"loss": 0.0116,
"step": 7890
},
{
"grad_norm": 0.6055150032043457,
"learning_rate": 8.806857908758967e-05,
"loss": 0.0129,
"step": 7900
},
{
"grad_norm": 0.3260117173194885,
"learning_rate": 8.80328236154186e-05,
"loss": 0.0117,
"step": 7910
},
{
"grad_norm": 0.19797098636627197,
"learning_rate": 8.799702192978056e-05,
"loss": 0.0117,
"step": 7920
},
{
"grad_norm": 0.3774504065513611,
"learning_rate": 8.7961174074178e-05,
"loss": 0.0119,
"step": 7930
},
{
"grad_norm": 0.40719759464263916,
"learning_rate": 8.792528009216942e-05,
"loss": 0.0117,
"step": 7940
},
{
"grad_norm": 0.41024884581565857,
"learning_rate": 8.788934002736943e-05,
"loss": 0.012,
"step": 7950
},
{
"grad_norm": 0.34125572443008423,
"learning_rate": 8.785335392344857e-05,
"loss": 0.0108,
"step": 7960
},
{
"grad_norm": 0.20444460213184357,
"learning_rate": 8.781732182413336e-05,
"loss": 0.0109,
"step": 7970
},
{
"grad_norm": 0.38777855038642883,
"learning_rate": 8.778124377320618e-05,
"loss": 0.0112,
"step": 7980
},
{
"grad_norm": 0.3357694745063782,
"learning_rate": 8.774511981450529e-05,
"loss": 0.0116,
"step": 7990
},
{
"grad_norm": 0.3549976348876953,
"learning_rate": 8.770894999192467e-05,
"loss": 0.0109,
"step": 8000
},
{
"grad_norm": 0.3552756607532501,
"learning_rate": 8.767273434941412e-05,
"loss": 0.0111,
"step": 8010
},
{
"grad_norm": 0.3434692621231079,
"learning_rate": 8.7636472930979e-05,
"loss": 0.0108,
"step": 8020
},
{
"grad_norm": 0.3151755630970001,
"learning_rate": 8.76001657806804e-05,
"loss": 0.0112,
"step": 8030
},
{
"grad_norm": 0.40102118253707886,
"learning_rate": 8.756381294263491e-05,
"loss": 0.0108,
"step": 8040
},
{
"grad_norm": 0.31904926896095276,
"learning_rate": 8.752741446101464e-05,
"loss": 0.0108,
"step": 8050
},
{
"grad_norm": 0.3375629484653473,
"learning_rate": 8.749097038004722e-05,
"loss": 0.0105,
"step": 8060
},
{
"grad_norm": 0.36504364013671875,
"learning_rate": 8.745448074401561e-05,
"loss": 0.0115,
"step": 8070
},
{
"grad_norm": 0.275889128446579,
"learning_rate": 8.741794559725818e-05,
"loss": 0.0106,
"step": 8080
},
{
"grad_norm": 0.3825940191745758,
"learning_rate": 8.738136498416856e-05,
"loss": 0.0105,
"step": 8090
},
{
"grad_norm": 0.4263371229171753,
"learning_rate": 8.734473894919564e-05,
"loss": 0.0125,
"step": 8100
},
{
"grad_norm": 0.3766123950481415,
"learning_rate": 8.730806753684353e-05,
"loss": 0.0112,
"step": 8110
},
{
"grad_norm": 0.27596545219421387,
"learning_rate": 8.727135079167144e-05,
"loss": 0.0098,
"step": 8120
},
{
"grad_norm": 0.26507529616355896,
"learning_rate": 8.723458875829368e-05,
"loss": 0.0112,
"step": 8130
},
{
"grad_norm": 0.32793647050857544,
"learning_rate": 8.719778148137958e-05,
"loss": 0.0117,
"step": 8140
},
{
"grad_norm": 0.5006524920463562,
"learning_rate": 8.716092900565346e-05,
"loss": 0.0141,
"step": 8150
},
{
"grad_norm": 0.29135891795158386,
"learning_rate": 8.712403137589456e-05,
"loss": 0.0117,
"step": 8160
},
{
"grad_norm": 0.3051094114780426,
"learning_rate": 8.708708863693697e-05,
"loss": 0.0125,
"step": 8170
},
{
"grad_norm": 0.33508554100990295,
"learning_rate": 8.70501008336696e-05,
"loss": 0.0107,
"step": 8180
},
{
"grad_norm": 0.33325785398483276,
"learning_rate": 8.701306801103611e-05,
"loss": 0.0101,
"step": 8190
},
{
"grad_norm": 0.3076310455799103,
"learning_rate": 8.697599021403489e-05,
"loss": 0.0119,
"step": 8200
},
{
"grad_norm": 0.2822284996509552,
"learning_rate": 8.693886748771896e-05,
"loss": 0.0102,
"step": 8210
},
{
"grad_norm": 0.29608970880508423,
"learning_rate": 8.690169987719593e-05,
"loss": 0.011,
"step": 8220
},
{
"grad_norm": 0.36461561918258667,
"learning_rate": 8.686448742762791e-05,
"loss": 0.0107,
"step": 8230
},
{
"grad_norm": 0.3555935323238373,
"learning_rate": 8.682723018423161e-05,
"loss": 0.0106,
"step": 8240
},
{
"grad_norm": 0.2973068356513977,
"learning_rate": 8.678992819227804e-05,
"loss": 0.011,
"step": 8250
},
{
"grad_norm": 0.3376704156398773,
"learning_rate": 8.675258149709265e-05,
"loss": 0.0104,
"step": 8260
},
{
"grad_norm": 0.340263694524765,
"learning_rate": 8.67151901440552e-05,
"loss": 0.0101,
"step": 8270
},
{
"grad_norm": 0.293972373008728,
"learning_rate": 8.667775417859971e-05,
"loss": 0.0097,
"step": 8280
},
{
"grad_norm": 0.27219337224960327,
"learning_rate": 8.664027364621441e-05,
"loss": 0.0105,
"step": 8290
},
{
"grad_norm": 0.27255040407180786,
"learning_rate": 8.660274859244168e-05,
"loss": 0.0123,
"step": 8300
},
{
"grad_norm": 0.3804474472999573,
"learning_rate": 8.656517906287799e-05,
"loss": 0.0103,
"step": 8310
},
{
"grad_norm": 0.4029916226863861,
"learning_rate": 8.652756510317387e-05,
"loss": 0.0116,
"step": 8320
},
{
"grad_norm": 0.3002796173095703,
"learning_rate": 8.648990675903382e-05,
"loss": 0.0106,
"step": 8330
},
{
"grad_norm": 0.2673972249031067,
"learning_rate": 8.64522040762163e-05,
"loss": 0.0105,
"step": 8340
},
{
"grad_norm": 0.31608739495277405,
"learning_rate": 8.64144571005336e-05,
"loss": 0.0127,
"step": 8350
},
{
"grad_norm": 0.2921021580696106,
"learning_rate": 8.637666587785184e-05,
"loss": 0.011,
"step": 8360
},
{
"grad_norm": 0.2442844808101654,
"learning_rate": 8.633883045409096e-05,
"loss": 0.0101,
"step": 8370
},
{
"grad_norm": 0.2928016483783722,
"learning_rate": 8.630095087522457e-05,
"loss": 0.0109,
"step": 8380
},
{
"grad_norm": 0.3342704474925995,
"learning_rate": 8.62630271872799e-05,
"loss": 0.0104,
"step": 8390
},
{
"grad_norm": 0.30834606289863586,
"learning_rate": 8.62250594363378e-05,
"loss": 0.0114,
"step": 8400
},
{
"grad_norm": 0.2779328227043152,
"learning_rate": 8.61870476685327e-05,
"loss": 0.0122,
"step": 8410
},
{
"grad_norm": 0.37507641315460205,
"learning_rate": 8.614899193005248e-05,
"loss": 0.0105,
"step": 8420
},
{
"grad_norm": 0.3060291111469269,
"learning_rate": 8.611089226713842e-05,
"loss": 0.0097,
"step": 8430
},
{
"grad_norm": 0.3532843291759491,
"learning_rate": 8.607274872608521e-05,
"loss": 0.0101,
"step": 8440
},
{
"grad_norm": 0.3082657754421234,
"learning_rate": 8.603456135324088e-05,
"loss": 0.0108,
"step": 8450
},
{
"grad_norm": 0.27097657322883606,
"learning_rate": 8.599633019500665e-05,
"loss": 0.0097,
"step": 8460
},
{
"grad_norm": 0.4173792600631714,
"learning_rate": 8.595805529783702e-05,
"loss": 0.0097,
"step": 8470
},
{
"grad_norm": 0.48021912574768066,
"learning_rate": 8.591973670823958e-05,
"loss": 0.0107,
"step": 8480
},
{
"grad_norm": 0.2856282889842987,
"learning_rate": 8.588137447277503e-05,
"loss": 0.0104,
"step": 8490
},
{
"grad_norm": 0.3201524317264557,
"learning_rate": 8.584296863805709e-05,
"loss": 0.0119,
"step": 8500
},
{
"grad_norm": 0.3024746775627136,
"learning_rate": 8.580451925075248e-05,
"loss": 0.0113,
"step": 8510
},
{
"grad_norm": 0.37119582295417786,
"learning_rate": 8.576602635758085e-05,
"loss": 0.0107,
"step": 8520
},
{
"grad_norm": 0.3784583806991577,
"learning_rate": 8.572749000531469e-05,
"loss": 0.0108,
"step": 8530
},
{
"grad_norm": 0.2842702567577362,
"learning_rate": 8.568891024077926e-05,
"loss": 0.0105,
"step": 8540
},
{
"grad_norm": 0.382940411567688,
"learning_rate": 8.565028711085265e-05,
"loss": 0.0105,
"step": 8550
},
{
"grad_norm": 0.30965447425842285,
"learning_rate": 8.561162066246561e-05,
"loss": 0.0105,
"step": 8560
},
{
"grad_norm": 0.32315099239349365,
"learning_rate": 8.557291094260152e-05,
"loss": 0.0108,
"step": 8570
},
{
"grad_norm": 0.3546385169029236,
"learning_rate": 8.553415799829629e-05,
"loss": 0.0107,
"step": 8580
},
{
"grad_norm": 0.4725189208984375,
"learning_rate": 8.549536187663847e-05,
"loss": 0.011,
"step": 8590
},
{
"grad_norm": 0.37511712312698364,
"learning_rate": 8.545652262476897e-05,
"loss": 0.0104,
"step": 8600
},
{
"grad_norm": 0.38934120535850525,
"learning_rate": 8.541764028988115e-05,
"loss": 0.0115,
"step": 8610
},
{
"grad_norm": 0.3524245023727417,
"learning_rate": 8.537871491922071e-05,
"loss": 0.0099,
"step": 8620
},
{
"grad_norm": 0.30628371238708496,
"learning_rate": 8.533974656008566e-05,
"loss": 0.0103,
"step": 8630
},
{
"grad_norm": 0.31296461820602417,
"learning_rate": 8.530073525982621e-05,
"loss": 0.0095,
"step": 8640
},
{
"grad_norm": 0.35006681084632874,
"learning_rate": 8.526168106584476e-05,
"loss": 0.0106,
"step": 8650
},
{
"grad_norm": 0.3150104880332947,
"learning_rate": 8.522258402559586e-05,
"loss": 0.011,
"step": 8660
},
{
"grad_norm": 0.34715455770492554,
"learning_rate": 8.518344418658613e-05,
"loss": 0.0102,
"step": 8670
},
{
"grad_norm": 0.2967567443847656,
"learning_rate": 8.51442615963741e-05,
"loss": 0.0103,
"step": 8680
},
{
"grad_norm": 0.32458803057670593,
"learning_rate": 8.510503630257034e-05,
"loss": 0.0102,
"step": 8690
},
{
"grad_norm": 0.3020278215408325,
"learning_rate": 8.50657683528373e-05,
"loss": 0.0121,
"step": 8700
},
{
"grad_norm": 0.243942990899086,
"learning_rate": 8.502645779488922e-05,
"loss": 0.0095,
"step": 8710
},
{
"grad_norm": 0.2725197374820709,
"learning_rate": 8.498710467649213e-05,
"loss": 0.0093,
"step": 8720
},
{
"grad_norm": 0.341840922832489,
"learning_rate": 8.494770904546381e-05,
"loss": 0.0098,
"step": 8730
},
{
"grad_norm": 0.34156742691993713,
"learning_rate": 8.490827094967363e-05,
"loss": 0.0099,
"step": 8740
},
{
"grad_norm": 0.2555910050868988,
"learning_rate": 8.486879043704263e-05,
"loss": 0.0098,
"step": 8750
},
{
"grad_norm": 0.2553873658180237,
"learning_rate": 8.482926755554332e-05,
"loss": 0.0093,
"step": 8760
},
{
"grad_norm": 0.3187851011753082,
"learning_rate": 8.478970235319975e-05,
"loss": 0.011,
"step": 8770
},
{
"grad_norm": 0.4050900936126709,
"learning_rate": 8.475009487808737e-05,
"loss": 0.0105,
"step": 8780
},
{
"grad_norm": 0.36939358711242676,
"learning_rate": 8.471044517833298e-05,
"loss": 0.0115,
"step": 8790
},
{
"grad_norm": 0.3504938781261444,
"learning_rate": 8.467075330211474e-05,
"loss": 0.0117,
"step": 8800
},
{
"grad_norm": 0.3362886905670166,
"learning_rate": 8.463101929766198e-05,
"loss": 0.0094,
"step": 8810
},
{
"grad_norm": 0.3649793267250061,
"learning_rate": 8.459124321325528e-05,
"loss": 0.0111,
"step": 8820
},
{
"grad_norm": 0.24022740125656128,
"learning_rate": 8.455142509722634e-05,
"loss": 0.0109,
"step": 8830
},
{
"grad_norm": 0.33446770906448364,
"learning_rate": 8.451156499795792e-05,
"loss": 0.0094,
"step": 8840
},
{
"grad_norm": 0.3062475621700287,
"learning_rate": 8.447166296388381e-05,
"loss": 0.0111,
"step": 8850
},
{
"grad_norm": 0.36087968945503235,
"learning_rate": 8.443171904348872e-05,
"loss": 0.0089,
"step": 8860
},
{
"grad_norm": 0.282789021730423,
"learning_rate": 8.43917332853083e-05,
"loss": 0.0097,
"step": 8870
},
{
"grad_norm": 0.2516290545463562,
"learning_rate": 8.435170573792901e-05,
"loss": 0.0089,
"step": 8880
},
{
"grad_norm": 0.29854705929756165,
"learning_rate": 8.431163644998808e-05,
"loss": 0.009,
"step": 8890
},
{
"grad_norm": 0.3754189908504486,
"learning_rate": 8.42715254701735e-05,
"loss": 0.0096,
"step": 8900
},
{
"grad_norm": 0.33703139424324036,
"learning_rate": 8.42313728472239e-05,
"loss": 0.011,
"step": 8910
},
{
"grad_norm": 0.3538967967033386,
"learning_rate": 8.419117862992844e-05,
"loss": 0.0096,
"step": 8920
},
{
"grad_norm": 0.3214809000492096,
"learning_rate": 8.415094286712694e-05,
"loss": 0.0098,
"step": 8930
},
{
"grad_norm": 0.38746607303619385,
"learning_rate": 8.411066560770965e-05,
"loss": 0.0099,
"step": 8940
},
{
"grad_norm": 0.23421555757522583,
"learning_rate": 8.407034690061721e-05,
"loss": 0.0093,
"step": 8950
},
{
"grad_norm": 0.25281527638435364,
"learning_rate": 8.402998679484067e-05,
"loss": 0.01,
"step": 8960
},
{
"grad_norm": 0.3989478647708893,
"learning_rate": 8.398958533942135e-05,
"loss": 0.0099,
"step": 8970
},
{
"grad_norm": 0.24676960706710815,
"learning_rate": 8.394914258345083e-05,
"loss": 0.0091,
"step": 8980
},
{
"grad_norm": 0.28471508622169495,
"learning_rate": 8.390865857607089e-05,
"loss": 0.0096,
"step": 8990
},
{
"grad_norm": 0.3082621991634369,
"learning_rate": 8.386813336647339e-05,
"loss": 0.0095,
"step": 9000
},
{
"grad_norm": 0.2667902410030365,
"learning_rate": 8.38275670039003e-05,
"loss": 0.0098,
"step": 9010
},
{
"grad_norm": 0.2579623758792877,
"learning_rate": 8.378695953764356e-05,
"loss": 0.0098,
"step": 9020
},
{
"grad_norm": 0.42782002687454224,
"learning_rate": 8.374631101704509e-05,
"loss": 0.0099,
"step": 9030
},
{
"grad_norm": 0.2860717475414276,
"learning_rate": 8.370562149149665e-05,
"loss": 0.0102,
"step": 9040
},
{
"grad_norm": 0.25799623131752014,
"learning_rate": 8.366489101043989e-05,
"loss": 0.0099,
"step": 9050
},
{
"grad_norm": 0.31252026557922363,
"learning_rate": 8.362411962336614e-05,
"loss": 0.0096,
"step": 9060
},
{
"grad_norm": 0.23370763659477234,
"learning_rate": 8.358330737981651e-05,
"loss": 0.01,
"step": 9070
},
{
"grad_norm": 0.3088170886039734,
"learning_rate": 8.35424543293817e-05,
"loss": 0.009,
"step": 9080
},
{
"grad_norm": 0.29783859848976135,
"learning_rate": 8.350156052170206e-05,
"loss": 0.0115,
"step": 9090
},
{
"grad_norm": 0.3804084360599518,
"learning_rate": 8.346062600646738e-05,
"loss": 0.0096,
"step": 9100
},
{
"grad_norm": 0.30282413959503174,
"learning_rate": 8.341965083341697e-05,
"loss": 0.0103,
"step": 9110
},
{
"grad_norm": 0.38241881132125854,
"learning_rate": 8.337863505233953e-05,
"loss": 0.0102,
"step": 9120
},
{
"grad_norm": 0.3249412178993225,
"learning_rate": 8.33375787130731e-05,
"loss": 0.009,
"step": 9130
},
{
"grad_norm": 0.27458664774894714,
"learning_rate": 8.3296481865505e-05,
"loss": 0.0099,
"step": 9140
},
{
"grad_norm": 0.4108836352825165,
"learning_rate": 8.325534455957181e-05,
"loss": 0.0101,
"step": 9150
},
{
"grad_norm": 0.4223904311656952,
"learning_rate": 8.321416684525917e-05,
"loss": 0.0105,
"step": 9160
},
{
"grad_norm": 0.32968321442604065,
"learning_rate": 8.317294877260193e-05,
"loss": 0.0108,
"step": 9170
},
{
"grad_norm": 0.26523229479789734,
"learning_rate": 8.313169039168394e-05,
"loss": 0.0092,
"step": 9180
},
{
"grad_norm": 0.2565534710884094,
"learning_rate": 8.3090391752638e-05,
"loss": 0.0099,
"step": 9190
},
{
"grad_norm": 0.308163583278656,
"learning_rate": 8.304905290564586e-05,
"loss": 0.01,
"step": 9200
},
{
"grad_norm": 0.3425454795360565,
"learning_rate": 8.300767390093813e-05,
"loss": 0.0113,
"step": 9210
},
{
"grad_norm": 0.19908463954925537,
"learning_rate": 8.296625478879417e-05,
"loss": 0.0103,
"step": 9220
},
{
"grad_norm": 0.316671758890152,
"learning_rate": 8.292479561954214e-05,
"loss": 0.0102,
"step": 9230
},
{
"grad_norm": 0.3078480660915375,
"learning_rate": 8.288329644355883e-05,
"loss": 0.0096,
"step": 9240
},
{
"grad_norm": 0.29191339015960693,
"learning_rate": 8.284175731126965e-05,
"loss": 0.0096,
"step": 9250
},
{
"grad_norm": 0.35205793380737305,
"learning_rate": 8.280017827314854e-05,
"loss": 0.0109,
"step": 9260
},
{
"grad_norm": 0.36453571915626526,
"learning_rate": 8.2758559379718e-05,
"loss": 0.0104,
"step": 9270
},
{
"grad_norm": 0.2580679655075073,
"learning_rate": 8.271690068154887e-05,
"loss": 0.0098,
"step": 9280
},
{
"grad_norm": 0.2611638605594635,
"learning_rate": 8.26752022292604e-05,
"loss": 0.0101,
"step": 9290
},
{
"grad_norm": 0.29041630029678345,
"learning_rate": 8.263346407352018e-05,
"loss": 0.0097,
"step": 9300
},
{
"grad_norm": 0.2506728172302246,
"learning_rate": 8.259168626504395e-05,
"loss": 0.0095,
"step": 9310
},
{
"grad_norm": 0.2942103445529938,
"learning_rate": 8.254986885459572e-05,
"loss": 0.0093,
"step": 9320
},
{
"grad_norm": 0.30901575088500977,
"learning_rate": 8.250801189298757e-05,
"loss": 0.0092,
"step": 9330
},
{
"grad_norm": 0.381786584854126,
"learning_rate": 8.246611543107967e-05,
"loss": 0.0087,
"step": 9340
},
{
"grad_norm": 0.3429117500782013,
"learning_rate": 8.242417951978018e-05,
"loss": 0.0098,
"step": 9350
},
{
"grad_norm": 0.30365362763404846,
"learning_rate": 8.238220421004517e-05,
"loss": 0.0093,
"step": 9360
},
{
"grad_norm": 0.3773025572299957,
"learning_rate": 8.234018955287859e-05,
"loss": 0.0098,
"step": 9370
},
{
"grad_norm": 0.2867434620857239,
"learning_rate": 8.229813559933224e-05,
"loss": 0.0096,
"step": 9380
},
{
"grad_norm": 0.36871495842933655,
"learning_rate": 8.22560424005056e-05,
"loss": 0.0097,
"step": 9390
},
{
"grad_norm": 0.29946839809417725,
"learning_rate": 8.22139100075459e-05,
"loss": 0.0101,
"step": 9400
},
{
"grad_norm": 0.33686700463294983,
"learning_rate": 8.217173847164799e-05,
"loss": 0.0104,
"step": 9410
},
{
"grad_norm": 0.3452436923980713,
"learning_rate": 8.212952784405422e-05,
"loss": 0.0095,
"step": 9420
},
{
"grad_norm": 0.3390505313873291,
"learning_rate": 8.208727817605451e-05,
"loss": 0.0103,
"step": 9430
},
{
"grad_norm": 0.2770306468009949,
"learning_rate": 8.204498951898618e-05,
"loss": 0.0089,
"step": 9440
},
{
"grad_norm": 0.26577332615852356,
"learning_rate": 8.200266192423395e-05,
"loss": 0.0091,
"step": 9450
},
{
"grad_norm": 0.31393006443977356,
"learning_rate": 8.196029544322981e-05,
"loss": 0.0098,
"step": 9460
},
{
"grad_norm": 0.2694907784461975,
"learning_rate": 8.191789012745305e-05,
"loss": 0.0088,
"step": 9470
},
{
"grad_norm": 0.28339341282844543,
"learning_rate": 8.187544602843014e-05,
"loss": 0.0091,
"step": 9480
},
{
"grad_norm": 0.28915339708328247,
"learning_rate": 8.183296319773466e-05,
"loss": 0.0104,
"step": 9490
},
{
"grad_norm": 0.2694634199142456,
"learning_rate": 8.179044168698721e-05,
"loss": 0.0096,
"step": 9500
},
{
"grad_norm": 0.31086036562919617,
"learning_rate": 8.174788154785548e-05,
"loss": 0.0088,
"step": 9510
},
{
"grad_norm": 0.31560900807380676,
"learning_rate": 8.170528283205403e-05,
"loss": 0.0103,
"step": 9520
},
{
"grad_norm": 0.2754019498825073,
"learning_rate": 8.166264559134435e-05,
"loss": 0.0113,
"step": 9530
},
{
"grad_norm": 0.2802848815917969,
"learning_rate": 8.161996987753467e-05,
"loss": 0.0104,
"step": 9540
},
{
"grad_norm": 0.2925932705402374,
"learning_rate": 8.157725574248e-05,
"loss": 0.0103,
"step": 9550
},
{
"grad_norm": 0.3291204869747162,
"learning_rate": 8.153450323808209e-05,
"loss": 0.0115,
"step": 9560
},
{
"grad_norm": 0.3283328711986542,
"learning_rate": 8.149171241628923e-05,
"loss": 0.0106,
"step": 9570
},
{
"grad_norm": 0.3074251413345337,
"learning_rate": 8.144888332909631e-05,
"loss": 0.0098,
"step": 9580
},
{
"grad_norm": 0.27979621291160583,
"learning_rate": 8.14060160285447e-05,
"loss": 0.0089,
"step": 9590
},
{
"grad_norm": 0.26148489117622375,
"learning_rate": 8.136311056672223e-05,
"loss": 0.0116,
"step": 9600
},
{
"grad_norm": 0.3204076886177063,
"learning_rate": 8.132016699576308e-05,
"loss": 0.0097,
"step": 9610
},
{
"grad_norm": 0.3098124861717224,
"learning_rate": 8.127718536784771e-05,
"loss": 0.0094,
"step": 9620
},
{
"grad_norm": 0.3159780204296112,
"learning_rate": 8.12341657352029e-05,
"loss": 0.0109,
"step": 9630
},
{
"grad_norm": 0.3584921956062317,
"learning_rate": 8.119110815010151e-05,
"loss": 0.0099,
"step": 9640
},
{
"grad_norm": 0.2659761905670166,
"learning_rate": 8.114801266486261e-05,
"loss": 0.0096,
"step": 9650
},
{
"grad_norm": 0.3183245360851288,
"learning_rate": 8.110487933185122e-05,
"loss": 0.0104,
"step": 9660
},
{
"grad_norm": 0.3073471486568451,
"learning_rate": 8.106170820347847e-05,
"loss": 0.0094,
"step": 9670
},
{
"grad_norm": 0.2520548105239868,
"learning_rate": 8.101849933220132e-05,
"loss": 0.0085,
"step": 9680
},
{
"grad_norm": 0.270577609539032,
"learning_rate": 8.097525277052264e-05,
"loss": 0.0109,
"step": 9690
},
{
"grad_norm": 0.3279707133769989,
"learning_rate": 8.093196857099106e-05,
"loss": 0.0105,
"step": 9700
},
{
"grad_norm": 0.41190341114997864,
"learning_rate": 8.088864678620096e-05,
"loss": 0.0102,
"step": 9710
},
{
"grad_norm": 0.32842499017715454,
"learning_rate": 8.084528746879243e-05,
"loss": 0.01,
"step": 9720
},
{
"grad_norm": 0.3014456331729889,
"learning_rate": 8.080189067145108e-05,
"loss": 0.0083,
"step": 9730
},
{
"grad_norm": 0.259885311126709,
"learning_rate": 8.075845644690813e-05,
"loss": 0.0082,
"step": 9740
},
{
"grad_norm": 0.2711023986339569,
"learning_rate": 8.07149848479403e-05,
"loss": 0.0079,
"step": 9750
},
{
"grad_norm": 0.29589828848838806,
"learning_rate": 8.067147592736962e-05,
"loss": 0.0095,
"step": 9760
},
{
"grad_norm": 0.24040688574314117,
"learning_rate": 8.062792973806357e-05,
"loss": 0.0087,
"step": 9770
},
{
"grad_norm": 0.3271734118461609,
"learning_rate": 8.058434633293485e-05,
"loss": 0.0097,
"step": 9780
},
{
"grad_norm": 0.2870268225669861,
"learning_rate": 8.054072576494143e-05,
"loss": 0.0092,
"step": 9790
},
{
"grad_norm": 0.23761416971683502,
"learning_rate": 8.04970680870864e-05,
"loss": 0.0084,
"step": 9800
},
{
"grad_norm": 0.31880688667297363,
"learning_rate": 8.045337335241793e-05,
"loss": 0.0089,
"step": 9810
},
{
"grad_norm": 0.2461557537317276,
"learning_rate": 8.040964161402932e-05,
"loss": 0.0088,
"step": 9820
},
{
"grad_norm": 0.34059616923332214,
"learning_rate": 8.036587292505869e-05,
"loss": 0.0104,
"step": 9830
},
{
"grad_norm": 0.2754742503166199,
"learning_rate": 8.032206733868912e-05,
"loss": 0.0094,
"step": 9840
},
{
"grad_norm": 0.33969369530677795,
"learning_rate": 8.027822490814858e-05,
"loss": 0.0089,
"step": 9850
},
{
"grad_norm": 0.28036972880363464,
"learning_rate": 8.023434568670972e-05,
"loss": 0.0087,
"step": 9860
},
{
"grad_norm": 0.3701011538505554,
"learning_rate": 8.019042972768992e-05,
"loss": 0.0092,
"step": 9870
},
{
"grad_norm": 0.27006396651268005,
"learning_rate": 8.014647708445124e-05,
"loss": 0.011,
"step": 9880
},
{
"grad_norm": 0.31688278913497925,
"learning_rate": 8.010248781040027e-05,
"loss": 0.0088,
"step": 9890
},
{
"grad_norm": 0.2984282374382019,
"learning_rate": 8.005846195898814e-05,
"loss": 0.0098,
"step": 9900
},
{
"grad_norm": 0.34147998690605164,
"learning_rate": 8.00143995837104e-05,
"loss": 0.0095,
"step": 9910
},
{
"grad_norm": 0.30249854922294617,
"learning_rate": 7.997030073810698e-05,
"loss": 0.0096,
"step": 9920
},
{
"grad_norm": 0.3344995081424713,
"learning_rate": 7.992616547576217e-05,
"loss": 0.0088,
"step": 9930
},
{
"grad_norm": 0.2232092171907425,
"learning_rate": 7.988199385030445e-05,
"loss": 0.0093,
"step": 9940
},
{
"grad_norm": 0.28129565715789795,
"learning_rate": 7.98377859154065e-05,
"loss": 0.0088,
"step": 9950
},
{
"grad_norm": 0.3453581631183624,
"learning_rate": 7.979354172478516e-05,
"loss": 0.0106,
"step": 9960
},
{
"grad_norm": 0.3263741731643677,
"learning_rate": 7.974926133220127e-05,
"loss": 0.0105,
"step": 9970
},
{
"grad_norm": 0.32977229356765747,
"learning_rate": 7.970494479145967e-05,
"loss": 0.011,
"step": 9980
},
{
"grad_norm": 0.3182139992713928,
"learning_rate": 7.966059215640919e-05,
"loss": 0.009,
"step": 9990
},
{
"grad_norm": 0.25233185291290283,
"learning_rate": 7.961620348094241e-05,
"loss": 0.0088,
"step": 10000
},
{
"grad_norm": 0.28312796354293823,
"learning_rate": 7.957177881899577e-05,
"loss": 0.0082,
"step": 10010
},
{
"grad_norm": 0.2667032778263092,
"learning_rate": 7.952731822454944e-05,
"loss": 0.0092,
"step": 10020
},
{
"grad_norm": 0.2328304648399353,
"learning_rate": 7.948282175162722e-05,
"loss": 0.008,
"step": 10030
},
{
"grad_norm": 0.26578933000564575,
"learning_rate": 7.943828945429652e-05,
"loss": 0.0086,
"step": 10040
},
{
"grad_norm": 0.3218497335910797,
"learning_rate": 7.939372138666827e-05,
"loss": 0.0087,
"step": 10050
},
{
"grad_norm": 0.2721211016178131,
"learning_rate": 7.934911760289692e-05,
"loss": 0.0093,
"step": 10060
},
{
"grad_norm": 0.2776382565498352,
"learning_rate": 7.930447815718022e-05,
"loss": 0.0089,
"step": 10070
},
{
"grad_norm": 0.3087117075920105,
"learning_rate": 7.925980310375934e-05,
"loss": 0.0094,
"step": 10080
},
{
"grad_norm": 0.3460996150970459,
"learning_rate": 7.921509249691865e-05,
"loss": 0.0097,
"step": 10090
},
{
"grad_norm": 0.345557302236557,
"learning_rate": 7.917034639098579e-05,
"loss": 0.0091,
"step": 10100
},
{
"grad_norm": 0.32438960671424866,
"learning_rate": 7.912556484033146e-05,
"loss": 0.0098,
"step": 10110
},
{
"grad_norm": 0.2700251340866089,
"learning_rate": 7.908074789936952e-05,
"loss": 0.0088,
"step": 10120
},
{
"grad_norm": 0.259725958108902,
"learning_rate": 7.903589562255673e-05,
"loss": 0.0086,
"step": 10130
},
{
"grad_norm": 0.2990911304950714,
"learning_rate": 7.899100806439285e-05,
"loss": 0.0087,
"step": 10140
},
{
"grad_norm": 0.32164642214775085,
"learning_rate": 7.894608527942049e-05,
"loss": 0.0092,
"step": 10150
},
{
"grad_norm": 0.2375648319721222,
"learning_rate": 7.89011273222251e-05,
"loss": 0.0081,
"step": 10160
},
{
"grad_norm": 0.2520039677619934,
"learning_rate": 7.885613424743482e-05,
"loss": 0.0082,
"step": 10170
},
{
"grad_norm": 0.28493061661720276,
"learning_rate": 7.881110610972044e-05,
"loss": 0.0091,
"step": 10180
},
{
"grad_norm": 0.2752293050289154,
"learning_rate": 7.876604296379544e-05,
"loss": 0.0101,
"step": 10190
},
{
"grad_norm": 0.2828473448753357,
"learning_rate": 7.872094486441579e-05,
"loss": 0.0087,
"step": 10200
},
{
"grad_norm": 0.2889810800552368,
"learning_rate": 7.867581186637991e-05,
"loss": 0.0082,
"step": 10210
},
{
"grad_norm": 0.2776321768760681,
"learning_rate": 7.863064402452866e-05,
"loss": 0.0088,
"step": 10220
},
{
"grad_norm": 0.2540366053581238,
"learning_rate": 7.858544139374524e-05,
"loss": 0.0091,
"step": 10230
},
{
"grad_norm": 0.31785818934440613,
"learning_rate": 7.854020402895508e-05,
"loss": 0.0109,
"step": 10240
},
{
"grad_norm": 0.24002966284751892,
"learning_rate": 7.849493198512587e-05,
"loss": 0.0083,
"step": 10250
},
{
"grad_norm": 0.249736025929451,
"learning_rate": 7.84496253172674e-05,
"loss": 0.0076,
"step": 10260
},
{
"grad_norm": 0.3291330337524414,
"learning_rate": 7.840428408043155e-05,
"loss": 0.0095,
"step": 10270
},
{
"grad_norm": 0.22596527636051178,
"learning_rate": 7.835890832971218e-05,
"loss": 0.0086,
"step": 10280
},
{
"grad_norm": 0.3245750963687897,
"learning_rate": 7.831349812024512e-05,
"loss": 0.0095,
"step": 10290
},
{
"grad_norm": 0.23606570065021515,
"learning_rate": 7.826805350720807e-05,
"loss": 0.0091,
"step": 10300
},
{
"grad_norm": 0.2839299142360687,
"learning_rate": 7.822257454582049e-05,
"loss": 0.0085,
"step": 10310
},
{
"grad_norm": 0.22629347443580627,
"learning_rate": 7.817706129134363e-05,
"loss": 0.0087,
"step": 10320
},
{
"grad_norm": 0.375497967004776,
"learning_rate": 7.813151379908036e-05,
"loss": 0.0079,
"step": 10330
},
{
"grad_norm": 0.2748655378818512,
"learning_rate": 7.808593212437523e-05,
"loss": 0.0082,
"step": 10340
},
{
"grad_norm": 0.3295239806175232,
"learning_rate": 7.804031632261421e-05,
"loss": 0.0096,
"step": 10350
},
{
"grad_norm": 0.33085861802101135,
"learning_rate": 7.799466644922484e-05,
"loss": 0.011,
"step": 10360
},
{
"grad_norm": 0.2350122630596161,
"learning_rate": 7.794898255967602e-05,
"loss": 0.0086,
"step": 10370
},
{
"grad_norm": 0.27413979172706604,
"learning_rate": 7.790326470947795e-05,
"loss": 0.0084,
"step": 10380
},
{
"grad_norm": 0.2807300090789795,
"learning_rate": 7.785751295418217e-05,
"loss": 0.0091,
"step": 10390
},
{
"grad_norm": 0.24402906000614166,
"learning_rate": 7.781172734938136e-05,
"loss": 0.0078,
"step": 10400
},
{
"grad_norm": 0.1861390620470047,
"learning_rate": 7.776590795070933e-05,
"loss": 0.0086,
"step": 10410
},
{
"grad_norm": 0.2955159842967987,
"learning_rate": 7.772005481384099e-05,
"loss": 0.009,
"step": 10420
},
{
"grad_norm": 0.3031066954135895,
"learning_rate": 7.767416799449223e-05,
"loss": 0.0088,
"step": 10430
},
{
"grad_norm": 0.30942216515541077,
"learning_rate": 7.762824754841984e-05,
"loss": 0.0084,
"step": 10440
},
{
"grad_norm": 0.27772602438926697,
"learning_rate": 7.758229353142152e-05,
"loss": 0.0088,
"step": 10450
},
{
"grad_norm": 0.18874435126781464,
"learning_rate": 7.753630599933572e-05,
"loss": 0.009,
"step": 10460
},
{
"grad_norm": 0.20754432678222656,
"learning_rate": 7.749028500804161e-05,
"loss": 0.0093,
"step": 10470
},
{
"grad_norm": 0.3341434895992279,
"learning_rate": 7.744423061345906e-05,
"loss": 0.0092,
"step": 10480
},
{
"grad_norm": 0.24594378471374512,
"learning_rate": 7.73981428715485e-05,
"loss": 0.0087,
"step": 10490
},
{
"grad_norm": 0.2970474362373352,
"learning_rate": 7.735202183831085e-05,
"loss": 0.0087,
"step": 10500
},
{
"grad_norm": 0.2976945638656616,
"learning_rate": 7.730586756978757e-05,
"loss": 0.0088,
"step": 10510
},
{
"grad_norm": 0.2520144581794739,
"learning_rate": 7.72596801220604e-05,
"loss": 0.009,
"step": 10520
},
{
"grad_norm": 0.21115921437740326,
"learning_rate": 7.721345955125147e-05,
"loss": 0.0076,
"step": 10530
},
{
"grad_norm": 0.2453879415988922,
"learning_rate": 7.71672059135231e-05,
"loss": 0.0082,
"step": 10540
},
{
"grad_norm": 0.333455890417099,
"learning_rate": 7.712091926507787e-05,
"loss": 0.0082,
"step": 10550
},
{
"grad_norm": 0.20065860450267792,
"learning_rate": 7.70745996621584e-05,
"loss": 0.0082,
"step": 10560
},
{
"grad_norm": 0.3492734134197235,
"learning_rate": 7.702824716104735e-05,
"loss": 0.0091,
"step": 10570
},
{
"grad_norm": 0.2791425585746765,
"learning_rate": 7.698186181806743e-05,
"loss": 0.0087,
"step": 10580
},
{
"grad_norm": 0.2732599079608917,
"learning_rate": 7.693544368958116e-05,
"loss": 0.0083,
"step": 10590
},
{
"grad_norm": 0.30734577775001526,
"learning_rate": 7.688899283199096e-05,
"loss": 0.0087,
"step": 10600
},
{
"grad_norm": 0.29557982087135315,
"learning_rate": 7.684250930173901e-05,
"loss": 0.0087,
"step": 10610
},
{
"grad_norm": 0.27425116300582886,
"learning_rate": 7.679599315530718e-05,
"loss": 0.0099,
"step": 10620
},
{
"grad_norm": 0.2503387928009033,
"learning_rate": 7.674944444921695e-05,
"loss": 0.0089,
"step": 10630
},
{
"grad_norm": 0.22014181315898895,
"learning_rate": 7.670286324002944e-05,
"loss": 0.0083,
"step": 10640
},
{
"grad_norm": 0.2801256477832794,
"learning_rate": 7.665624958434514e-05,
"loss": 0.0085,
"step": 10650
},
{
"grad_norm": 0.2603195905685425,
"learning_rate": 7.66096035388041e-05,
"loss": 0.0101,
"step": 10660
},
{
"grad_norm": 0.32898423075675964,
"learning_rate": 7.656292516008563e-05,
"loss": 0.0091,
"step": 10670
},
{
"grad_norm": 0.2944939136505127,
"learning_rate": 7.651621450490837e-05,
"loss": 0.0093,
"step": 10680
},
{
"grad_norm": 0.3123464584350586,
"learning_rate": 7.646947163003017e-05,
"loss": 0.0086,
"step": 10690
},
{
"grad_norm": 0.29304465651512146,
"learning_rate": 7.642269659224803e-05,
"loss": 0.0087,
"step": 10700
},
{
"grad_norm": 0.3500259220600128,
"learning_rate": 7.637588944839803e-05,
"loss": 0.009,
"step": 10710
},
{
"grad_norm": 0.26453834772109985,
"learning_rate": 7.632905025535528e-05,
"loss": 0.0089,
"step": 10720
},
{
"grad_norm": 0.25538867712020874,
"learning_rate": 7.628217907003378e-05,
"loss": 0.0082,
"step": 10730
},
{
"grad_norm": 0.21325016021728516,
"learning_rate": 7.623527594938649e-05,
"loss": 0.0084,
"step": 10740
},
{
"grad_norm": 0.2767277657985687,
"learning_rate": 7.618834095040509e-05,
"loss": 0.008,
"step": 10750
},
{
"grad_norm": 0.2665347158908844,
"learning_rate": 7.614137413012e-05,
"loss": 0.0091,
"step": 10760
},
{
"grad_norm": 0.2695922255516052,
"learning_rate": 7.609437554560042e-05,
"loss": 0.0078,
"step": 10770
},
{
"grad_norm": 0.23686271905899048,
"learning_rate": 7.604734525395398e-05,
"loss": 0.0083,
"step": 10780
},
{
"grad_norm": 0.23653416335582733,
"learning_rate": 7.600028331232699e-05,
"loss": 0.0085,
"step": 10790
},
{
"grad_norm": 0.20893239974975586,
"learning_rate": 7.595318977790408e-05,
"loss": 0.0081,
"step": 10800
},
{
"grad_norm": 0.348706990480423,
"learning_rate": 7.590606470790836e-05,
"loss": 0.0086,
"step": 10810
},
{
"grad_norm": 0.27015960216522217,
"learning_rate": 7.585890815960125e-05,
"loss": 0.0082,
"step": 10820
},
{
"grad_norm": 0.22292320430278778,
"learning_rate": 7.581172019028238e-05,
"loss": 0.0077,
"step": 10830
},
{
"grad_norm": 0.3161787688732147,
"learning_rate": 7.576450085728959e-05,
"loss": 0.0079,
"step": 10840
},
{
"grad_norm": 0.4077858626842499,
"learning_rate": 7.571725021799885e-05,
"loss": 0.0084,
"step": 10850
},
{
"grad_norm": 0.3302555978298187,
"learning_rate": 7.566996832982409e-05,
"loss": 0.0094,
"step": 10860
},
{
"grad_norm": 0.2723492980003357,
"learning_rate": 7.56226552502173e-05,
"loss": 0.0082,
"step": 10870
},
{
"grad_norm": 0.2489624172449112,
"learning_rate": 7.557531103666833e-05,
"loss": 0.0096,
"step": 10880
},
{
"grad_norm": 0.2531261146068573,
"learning_rate": 7.552793574670485e-05,
"loss": 0.0087,
"step": 10890
},
{
"grad_norm": 0.3141566812992096,
"learning_rate": 7.54805294378923e-05,
"loss": 0.0078,
"step": 10900
},
{
"grad_norm": 0.2886313498020172,
"learning_rate": 7.543309216783384e-05,
"loss": 0.0084,
"step": 10910
},
{
"grad_norm": 0.33143842220306396,
"learning_rate": 7.53856239941702e-05,
"loss": 0.0084,
"step": 10920
},
{
"grad_norm": 0.3636597692966461,
"learning_rate": 7.533812497457972e-05,
"loss": 0.009,
"step": 10930
},
{
"grad_norm": 0.2903405725955963,
"learning_rate": 7.529059516677814e-05,
"loss": 0.0088,
"step": 10940
},
{
"grad_norm": 0.2362307757139206,
"learning_rate": 7.524303462851872e-05,
"loss": 0.0078,
"step": 10950
},
{
"grad_norm": 0.2994009256362915,
"learning_rate": 7.519544341759192e-05,
"loss": 0.0084,
"step": 10960
},
{
"grad_norm": 0.22063317894935608,
"learning_rate": 7.514782159182562e-05,
"loss": 0.0096,
"step": 10970
},
{
"grad_norm": 0.22048303484916687,
"learning_rate": 7.510016920908481e-05,
"loss": 0.0078,
"step": 10980
},
{
"grad_norm": 0.3061755895614624,
"learning_rate": 7.505248632727158e-05,
"loss": 0.0086,
"step": 10990
},
{
"grad_norm": 0.2510005235671997,
"learning_rate": 7.50047730043252e-05,
"loss": 0.0085,
"step": 11000
},
{
"grad_norm": 0.2653082609176636,
"learning_rate": 7.495702929822183e-05,
"loss": 0.0078,
"step": 11010
},
{
"grad_norm": 0.24982401728630066,
"learning_rate": 7.490925526697455e-05,
"loss": 0.008,
"step": 11020
},
{
"grad_norm": 0.33277788758277893,
"learning_rate": 7.486145096863334e-05,
"loss": 0.0075,
"step": 11030
},
{
"grad_norm": 0.2504388689994812,
"learning_rate": 7.481361646128491e-05,
"loss": 0.0085,
"step": 11040
},
{
"grad_norm": 0.264525443315506,
"learning_rate": 7.476575180305271e-05,
"loss": 0.0081,
"step": 11050
},
{
"grad_norm": 0.25063708424568176,
"learning_rate": 7.471785705209682e-05,
"loss": 0.0076,
"step": 11060
},
{
"grad_norm": 0.3602023124694824,
"learning_rate": 7.466993226661387e-05,
"loss": 0.0089,
"step": 11070
},
{
"grad_norm": 0.3457334339618683,
"learning_rate": 7.462197750483699e-05,
"loss": 0.0084,
"step": 11080
},
{
"grad_norm": 0.3209291398525238,
"learning_rate": 7.457399282503574e-05,
"loss": 0.0084,
"step": 11090
},
{
"grad_norm": 0.2748512923717499,
"learning_rate": 7.452597828551604e-05,
"loss": 0.0079,
"step": 11100
},
{
"grad_norm": 0.29751718044281006,
"learning_rate": 7.447793394462006e-05,
"loss": 0.0089,
"step": 11110
},
{
"grad_norm": 0.25582262873649597,
"learning_rate": 7.442985986072624e-05,
"loss": 0.008,
"step": 11120
},
{
"grad_norm": 0.2571868300437927,
"learning_rate": 7.438175609224908e-05,
"loss": 0.0077,
"step": 11130
},
{
"grad_norm": 0.24162840843200684,
"learning_rate": 7.433362269763924e-05,
"loss": 0.0083,
"step": 11140
},
{
"grad_norm": 0.2867334187030792,
"learning_rate": 7.428545973538329e-05,
"loss": 0.0084,
"step": 11150
},
{
"grad_norm": 0.2729656994342804,
"learning_rate": 7.42372672640038e-05,
"loss": 0.0082,
"step": 11160
},
{
"grad_norm": 0.26211899518966675,
"learning_rate": 7.418904534205917e-05,
"loss": 0.0086,
"step": 11170
},
{
"grad_norm": 0.2512105405330658,
"learning_rate": 7.414079402814356e-05,
"loss": 0.0079,
"step": 11180
},
{
"grad_norm": 0.28634539246559143,
"learning_rate": 7.409251338088687e-05,
"loss": 0.0079,
"step": 11190
},
{
"grad_norm": 0.28483590483665466,
"learning_rate": 7.404420345895467e-05,
"loss": 0.009,
"step": 11200
},
{
"grad_norm": 0.27629390358924866,
"learning_rate": 7.399586432104804e-05,
"loss": 0.0093,
"step": 11210
},
{
"grad_norm": 0.32245320081710815,
"learning_rate": 7.394749602590358e-05,
"loss": 0.0085,
"step": 11220
},
{
"grad_norm": 0.37746337056159973,
"learning_rate": 7.389909863229336e-05,
"loss": 0.0104,
"step": 11230
},
{
"grad_norm": 0.2625359892845154,
"learning_rate": 7.385067219902477e-05,
"loss": 0.0074,
"step": 11240
},
{
"grad_norm": 0.3064190149307251,
"learning_rate": 7.380221678494049e-05,
"loss": 0.0078,
"step": 11250
},
{
"grad_norm": 0.3272659480571747,
"learning_rate": 7.37537324489184e-05,
"loss": 0.0092,
"step": 11260
},
{
"grad_norm": 0.23744618892669678,
"learning_rate": 7.370521924987155e-05,
"loss": 0.0078,
"step": 11270
},
{
"grad_norm": 0.2976019084453583,
"learning_rate": 7.365667724674806e-05,
"loss": 0.0084,
"step": 11280
},
{
"grad_norm": 0.2628666162490845,
"learning_rate": 7.360810649853105e-05,
"loss": 0.0089,
"step": 11290
},
{
"grad_norm": 0.249125674366951,
"learning_rate": 7.355950706423853e-05,
"loss": 0.0076,
"step": 11300
},
{
"grad_norm": 0.22084388136863708,
"learning_rate": 7.351087900292342e-05,
"loss": 0.0095,
"step": 11310
},
{
"grad_norm": 0.2580353021621704,
"learning_rate": 7.346222237367339e-05,
"loss": 0.0082,
"step": 11320
},
{
"grad_norm": 0.2719237506389618,
"learning_rate": 7.341353723561081e-05,
"loss": 0.0085,
"step": 11330
},
{
"grad_norm": 0.3120487630367279,
"learning_rate": 7.336482364789277e-05,
"loss": 0.0088,
"step": 11340
},
{
"grad_norm": 0.2787841558456421,
"learning_rate": 7.331608166971082e-05,
"loss": 0.0097,
"step": 11350
},
{
"grad_norm": 0.3463439643383026,
"learning_rate": 7.326731136029108e-05,
"loss": 0.0092,
"step": 11360
},
{
"grad_norm": 0.38997554779052734,
"learning_rate": 7.321851277889408e-05,
"loss": 0.0082,
"step": 11370
},
{
"grad_norm": 0.239556223154068,
"learning_rate": 7.316968598481469e-05,
"loss": 0.0087,
"step": 11380
},
{
"grad_norm": 0.3022553622722626,
"learning_rate": 7.312083103738206e-05,
"loss": 0.009,
"step": 11390
},
{
"grad_norm": 0.2007443904876709,
"learning_rate": 7.307194799595958e-05,
"loss": 0.0075,
"step": 11400
},
{
"grad_norm": 0.2427217811346054,
"learning_rate": 7.302303691994473e-05,
"loss": 0.0074,
"step": 11410
},
{
"grad_norm": 0.2285958081483841,
"learning_rate": 7.29740978687691e-05,
"loss": 0.0066,
"step": 11420
},
{
"grad_norm": 0.30119019746780396,
"learning_rate": 7.292513090189825e-05,
"loss": 0.0071,
"step": 11430
},
{
"grad_norm": 0.26628202199935913,
"learning_rate": 7.287613607883163e-05,
"loss": 0.0076,
"step": 11440
},
{
"grad_norm": 0.21646924316883087,
"learning_rate": 7.282711345910263e-05,
"loss": 0.0077,
"step": 11450
},
{
"grad_norm": 0.2584187090396881,
"learning_rate": 7.27780631022783e-05,
"loss": 0.0081,
"step": 11460
},
{
"grad_norm": 0.25080376863479614,
"learning_rate": 7.272898506795948e-05,
"loss": 0.0093,
"step": 11470
},
{
"grad_norm": 0.24241487681865692,
"learning_rate": 7.267987941578058e-05,
"loss": 0.008,
"step": 11480
},
{
"grad_norm": 0.2837870419025421,
"learning_rate": 7.263074620540962e-05,
"loss": 0.0086,
"step": 11490
},
{
"grad_norm": 0.23800799250602722,
"learning_rate": 7.25815854965481e-05,
"loss": 0.0077,
"step": 11500
},
{
"grad_norm": 0.20950715243816376,
"learning_rate": 7.253239734893089e-05,
"loss": 0.0079,
"step": 11510
},
{
"grad_norm": 0.27476486563682556,
"learning_rate": 7.248318182232623e-05,
"loss": 0.0081,
"step": 11520
},
{
"grad_norm": 0.22000481188297272,
"learning_rate": 7.243393897653565e-05,
"loss": 0.0078,
"step": 11530
},
{
"grad_norm": 0.22252796590328217,
"learning_rate": 7.23846688713938e-05,
"loss": 0.0078,
"step": 11540
},
{
"grad_norm": 0.24442344903945923,
"learning_rate": 7.233537156676854e-05,
"loss": 0.0082,
"step": 11550
},
{
"grad_norm": 0.28191429376602173,
"learning_rate": 7.228604712256076e-05,
"loss": 0.0078,
"step": 11560
},
{
"grad_norm": 0.24343600869178772,
"learning_rate": 7.223669559870426e-05,
"loss": 0.0075,
"step": 11570
},
{
"grad_norm": 0.2522268295288086,
"learning_rate": 7.218731705516585e-05,
"loss": 0.0094,
"step": 11580
},
{
"grad_norm": 0.25070708990097046,
"learning_rate": 7.21379115519451e-05,
"loss": 0.0078,
"step": 11590
},
{
"grad_norm": 0.21620993316173553,
"learning_rate": 7.20884791490743e-05,
"loss": 0.0079,
"step": 11600
},
{
"grad_norm": 0.2193872481584549,
"learning_rate": 7.203901990661857e-05,
"loss": 0.0077,
"step": 11610
},
{
"grad_norm": 0.2501574158668518,
"learning_rate": 7.198953388467549e-05,
"loss": 0.0076,
"step": 11620
},
{
"grad_norm": 0.3814297914505005,
"learning_rate": 7.194002114337526e-05,
"loss": 0.0085,
"step": 11630
},
{
"grad_norm": 0.3631609380245209,
"learning_rate": 7.189048174288054e-05,
"loss": 0.009,
"step": 11640
},
{
"grad_norm": 0.35405850410461426,
"learning_rate": 7.184091574338636e-05,
"loss": 0.0085,
"step": 11650
},
{
"grad_norm": 0.26702335476875305,
"learning_rate": 7.179132320512009e-05,
"loss": 0.0083,
"step": 11660
},
{
"grad_norm": 0.24685519933700562,
"learning_rate": 7.174170418834134e-05,
"loss": 0.0081,
"step": 11670
},
{
"grad_norm": 0.24434106051921844,
"learning_rate": 7.169205875334189e-05,
"loss": 0.0074,
"step": 11680
},
{
"grad_norm": 0.3144436776638031,
"learning_rate": 7.164238696044562e-05,
"loss": 0.0081,
"step": 11690
},
{
"grad_norm": 0.19701911509037018,
"learning_rate": 7.159268887000846e-05,
"loss": 0.0078,
"step": 11700
},
{
"grad_norm": 0.25164148211479187,
"learning_rate": 7.154296454241827e-05,
"loss": 0.0084,
"step": 11710
},
{
"grad_norm": 0.28079986572265625,
"learning_rate": 7.149321403809479e-05,
"loss": 0.0088,
"step": 11720
},
{
"grad_norm": 0.2987631857395172,
"learning_rate": 7.144343741748956e-05,
"loss": 0.0088,
"step": 11730
},
{
"grad_norm": 0.25953763723373413,
"learning_rate": 7.139363474108589e-05,
"loss": 0.0086,
"step": 11740
},
{
"grad_norm": 0.2453896552324295,
"learning_rate": 7.134380606939874e-05,
"loss": 0.0074,
"step": 11750
},
{
"grad_norm": 0.36023178696632385,
"learning_rate": 7.12939514629746e-05,
"loss": 0.0084,
"step": 11760
},
{
"grad_norm": 0.24952717125415802,
"learning_rate": 7.124407098239155e-05,
"loss": 0.0088,
"step": 11770
},
{
"grad_norm": 0.2500772774219513,
"learning_rate": 7.119416468825908e-05,
"loss": 0.0084,
"step": 11780
},
{
"grad_norm": 0.29303258657455444,
"learning_rate": 7.114423264121804e-05,
"loss": 0.0086,
"step": 11790
},
{
"grad_norm": 0.25186672806739807,
"learning_rate": 7.109427490194056e-05,
"loss": 0.0076,
"step": 11800
},
{
"grad_norm": 0.2728545367717743,
"learning_rate": 7.104429153113001e-05,
"loss": 0.0076,
"step": 11810
},
{
"grad_norm": 0.23277583718299866,
"learning_rate": 7.099428258952092e-05,
"loss": 0.0073,
"step": 11820
},
{
"grad_norm": 0.3232274055480957,
"learning_rate": 7.094424813787883e-05,
"loss": 0.0085,
"step": 11830
},
{
"grad_norm": 0.24028784036636353,
"learning_rate": 7.089418823700034e-05,
"loss": 0.0074,
"step": 11840
},
{
"grad_norm": 0.17265081405639648,
"learning_rate": 7.084410294771298e-05,
"loss": 0.0071,
"step": 11850
},
{
"grad_norm": 0.2584075331687927,
"learning_rate": 7.079399233087504e-05,
"loss": 0.0076,
"step": 11860
},
{
"grad_norm": 0.2501102387905121,
"learning_rate": 7.074385644737568e-05,
"loss": 0.0078,
"step": 11870
},
{
"grad_norm": 0.25903502106666565,
"learning_rate": 7.069369535813473e-05,
"loss": 0.0086,
"step": 11880
},
{
"grad_norm": 0.1961066722869873,
"learning_rate": 7.06435091241026e-05,
"loss": 0.0072,
"step": 11890
},
{
"grad_norm": 0.3219556212425232,
"learning_rate": 7.059329780626034e-05,
"loss": 0.0069,
"step": 11900
},
{
"grad_norm": 0.29326993227005005,
"learning_rate": 7.054306146561944e-05,
"loss": 0.0078,
"step": 11910
},
{
"grad_norm": 0.3069089353084564,
"learning_rate": 7.049280016322178e-05,
"loss": 0.0078,
"step": 11920
},
{
"grad_norm": 0.2479066103696823,
"learning_rate": 7.044251396013957e-05,
"loss": 0.0077,
"step": 11930
},
{
"grad_norm": 0.2422267347574234,
"learning_rate": 7.039220291747528e-05,
"loss": 0.0071,
"step": 11940
},
{
"grad_norm": 0.21664093434810638,
"learning_rate": 7.034186709636159e-05,
"loss": 0.007,
"step": 11950
},
{
"grad_norm": 0.300942599773407,
"learning_rate": 7.029150655796129e-05,
"loss": 0.0063,
"step": 11960
},
{
"grad_norm": 0.2523380219936371,
"learning_rate": 7.024112136346712e-05,
"loss": 0.007,
"step": 11970
},
{
"grad_norm": 0.25909993052482605,
"learning_rate": 7.01907115741019e-05,
"loss": 0.0078,
"step": 11980
},
{
"grad_norm": 0.3041617274284363,
"learning_rate": 7.014027725111825e-05,
"loss": 0.0085,
"step": 11990
},
{
"grad_norm": 0.2782539129257202,
"learning_rate": 7.008981845579864e-05,
"loss": 0.0075,
"step": 12000
},
{
"grad_norm": 0.3118305802345276,
"learning_rate": 7.003933524945528e-05,
"loss": 0.0081,
"step": 12010
},
{
"grad_norm": 0.26557084918022156,
"learning_rate": 6.998882769342998e-05,
"loss": 0.0072,
"step": 12020
},
{
"grad_norm": 0.23537853360176086,
"learning_rate": 6.993829584909423e-05,
"loss": 0.0078,
"step": 12030
},
{
"grad_norm": 0.22218097746372223,
"learning_rate": 6.988773977784895e-05,
"loss": 0.0073,
"step": 12040
},
{
"grad_norm": 0.22848103940486908,
"learning_rate": 6.983715954112454e-05,
"loss": 0.0077,
"step": 12050
},
{
"grad_norm": 0.21472135186195374,
"learning_rate": 6.978655520038079e-05,
"loss": 0.0077,
"step": 12060
},
{
"grad_norm": 0.26741930842399597,
"learning_rate": 6.97359268171067e-05,
"loss": 0.007,
"step": 12070
},
{
"grad_norm": 0.20105968415737152,
"learning_rate": 6.968527445282056e-05,
"loss": 0.0083,
"step": 12080
},
{
"grad_norm": 0.3267427682876587,
"learning_rate": 6.963459816906977e-05,
"loss": 0.0076,
"step": 12090
},
{
"grad_norm": 0.19545020163059235,
"learning_rate": 6.958389802743077e-05,
"loss": 0.0083,
"step": 12100
},
{
"grad_norm": 0.2549017369747162,
"learning_rate": 6.953317408950903e-05,
"loss": 0.0083,
"step": 12110
},
{
"grad_norm": 0.23277691006660461,
"learning_rate": 6.94824264169389e-05,
"loss": 0.0071,
"step": 12120
},
{
"grad_norm": 0.2652171850204468,
"learning_rate": 6.94316550713836e-05,
"loss": 0.0072,
"step": 12130
},
{
"grad_norm": 0.23182570934295654,
"learning_rate": 6.938086011453513e-05,
"loss": 0.0071,
"step": 12140
},
{
"grad_norm": 0.23776176571846008,
"learning_rate": 6.93300416081141e-05,
"loss": 0.0083,
"step": 12150
},
{
"grad_norm": 0.2618558406829834,
"learning_rate": 6.927919961386984e-05,
"loss": 0.0085,
"step": 12160
},
{
"grad_norm": 0.28387129306793213,
"learning_rate": 6.922833419358013e-05,
"loss": 0.0076,
"step": 12170
},
{
"grad_norm": 0.257773220539093,
"learning_rate": 6.917744540905124e-05,
"loss": 0.0083,
"step": 12180
},
{
"grad_norm": 0.23209062218666077,
"learning_rate": 6.912653332211787e-05,
"loss": 0.0067,
"step": 12190
},
{
"grad_norm": 0.24082745611667633,
"learning_rate": 6.9075597994643e-05,
"loss": 0.0079,
"step": 12200
},
{
"grad_norm": 0.2725173532962799,
"learning_rate": 6.902463948851787e-05,
"loss": 0.0077,
"step": 12210
},
{
"grad_norm": 0.2495144158601761,
"learning_rate": 6.897365786566184e-05,
"loss": 0.0083,
"step": 12220
},
{
"grad_norm": 0.270396888256073,
"learning_rate": 6.892265318802242e-05,
"loss": 0.0083,
"step": 12230
},
{
"grad_norm": 0.3428340554237366,
"learning_rate": 6.887162551757507e-05,
"loss": 0.0075,
"step": 12240
},
{
"grad_norm": 0.28904852271080017,
"learning_rate": 6.882057491632326e-05,
"loss": 0.0075,
"step": 12250
},
{
"grad_norm": 0.23016634583473206,
"learning_rate": 6.876950144629824e-05,
"loss": 0.0079,
"step": 12260
},
{
"grad_norm": 0.24582698941230774,
"learning_rate": 6.87184051695591e-05,
"loss": 0.0071,
"step": 12270
},
{
"grad_norm": 0.2511565685272217,
"learning_rate": 6.866728614819268e-05,
"loss": 0.0081,
"step": 12280
},
{
"grad_norm": 0.21527400612831116,
"learning_rate": 6.861614444431337e-05,
"loss": 0.0079,
"step": 12290
},
{
"grad_norm": 0.21463048458099365,
"learning_rate": 6.856498012006318e-05,
"loss": 0.0067,
"step": 12300
},
{
"grad_norm": 0.24686744809150696,
"learning_rate": 6.851379323761157e-05,
"loss": 0.0076,
"step": 12310
},
{
"grad_norm": 0.2256661057472229,
"learning_rate": 6.846258385915545e-05,
"loss": 0.0071,
"step": 12320
},
{
"grad_norm": 0.2230372130870819,
"learning_rate": 6.841135204691902e-05,
"loss": 0.0074,
"step": 12330
},
{
"grad_norm": 0.2623993754386902,
"learning_rate": 6.836009786315377e-05,
"loss": 0.0066,
"step": 12340
},
{
"grad_norm": 0.28395700454711914,
"learning_rate": 6.830882137013839e-05,
"loss": 0.0067,
"step": 12350
},
{
"grad_norm": 0.3418019115924835,
"learning_rate": 6.825752263017863e-05,
"loss": 0.0081,
"step": 12360
},
{
"grad_norm": 0.2549242675304413,
"learning_rate": 6.820620170560731e-05,
"loss": 0.0073,
"step": 12370
},
{
"grad_norm": 0.2908019423484802,
"learning_rate": 6.815485865878418e-05,
"loss": 0.0075,
"step": 12380
},
{
"grad_norm": 0.1923939436674118,
"learning_rate": 6.810349355209588e-05,
"loss": 0.0082,
"step": 12390
},
{
"grad_norm": 0.28272947669029236,
"learning_rate": 6.805210644795588e-05,
"loss": 0.0077,
"step": 12400
},
{
"grad_norm": 0.2060726433992386,
"learning_rate": 6.800069740880433e-05,
"loss": 0.0083,
"step": 12410
},
{
"grad_norm": 0.22574064135551453,
"learning_rate": 6.794926649710807e-05,
"loss": 0.0072,
"step": 12420
},
{
"grad_norm": 0.2950822114944458,
"learning_rate": 6.789781377536052e-05,
"loss": 0.008,
"step": 12430
},
{
"grad_norm": 0.25280317664146423,
"learning_rate": 6.784633930608158e-05,
"loss": 0.0078,
"step": 12440
},
{
"grad_norm": 0.26652100682258606,
"learning_rate": 6.779484315181759e-05,
"loss": 0.0072,
"step": 12450
},
{
"grad_norm": 0.3308437764644623,
"learning_rate": 6.774332537514122e-05,
"loss": 0.0077,
"step": 12460
},
{
"grad_norm": 0.21044404804706573,
"learning_rate": 6.769178603865143e-05,
"loss": 0.0084,
"step": 12470
},
{
"grad_norm": 0.24346868693828583,
"learning_rate": 6.764022520497337e-05,
"loss": 0.0078,
"step": 12480
},
{
"grad_norm": 0.25575554370880127,
"learning_rate": 6.758864293675833e-05,
"loss": 0.007,
"step": 12490
},
{
"grad_norm": 0.26012754440307617,
"learning_rate": 6.753703929668362e-05,
"loss": 0.008,
"step": 12500
},
{
"grad_norm": 0.27795130014419556,
"learning_rate": 6.748541434745253e-05,
"loss": 0.0067,
"step": 12510
},
{
"grad_norm": 0.22350628674030304,
"learning_rate": 6.743376815179424e-05,
"loss": 0.0077,
"step": 12520
},
{
"grad_norm": 0.24714377522468567,
"learning_rate": 6.738210077246376e-05,
"loss": 0.0071,
"step": 12530
},
{
"grad_norm": 0.2536865472793579,
"learning_rate": 6.733041227224181e-05,
"loss": 0.0082,
"step": 12540
},
{
"grad_norm": 0.28324517607688904,
"learning_rate": 6.72787027139348e-05,
"loss": 0.0071,
"step": 12550
},
{
"grad_norm": 0.25217628479003906,
"learning_rate": 6.72269721603747e-05,
"loss": 0.0073,
"step": 12560
},
{
"grad_norm": 0.2609931528568268,
"learning_rate": 6.717522067441904e-05,
"loss": 0.0076,
"step": 12570
},
{
"grad_norm": 0.22231438755989075,
"learning_rate": 6.712344831895074e-05,
"loss": 0.0077,
"step": 12580
},
{
"grad_norm": 0.2796629071235657,
"learning_rate": 6.707165515687811e-05,
"loss": 0.007,
"step": 12590
},
{
"grad_norm": 0.20269301533699036,
"learning_rate": 6.70198412511347e-05,
"loss": 0.0067,
"step": 12600
},
{
"grad_norm": 0.2400275319814682,
"learning_rate": 6.69680066646793e-05,
"loss": 0.0077,
"step": 12610
},
{
"grad_norm": 0.2624037563800812,
"learning_rate": 6.691615146049583e-05,
"loss": 0.0069,
"step": 12620
},
{
"grad_norm": 0.2606382966041565,
"learning_rate": 6.686427570159324e-05,
"loss": 0.0073,
"step": 12630
},
{
"grad_norm": 0.23165848851203918,
"learning_rate": 6.681237945100548e-05,
"loss": 0.0075,
"step": 12640
},
{
"grad_norm": 0.24344083666801453,
"learning_rate": 6.676046277179139e-05,
"loss": 0.007,
"step": 12650
},
{
"grad_norm": 0.21247272193431854,
"learning_rate": 6.670852572703462e-05,
"loss": 0.0068,
"step": 12660
},
{
"grad_norm": 0.2795293629169464,
"learning_rate": 6.665656837984359e-05,
"loss": 0.0066,
"step": 12670
},
{
"grad_norm": 0.21867285668849945,
"learning_rate": 6.660459079335135e-05,
"loss": 0.0073,
"step": 12680
},
{
"grad_norm": 0.23383796215057373,
"learning_rate": 6.655259303071558e-05,
"loss": 0.0063,
"step": 12690
},
{
"grad_norm": 0.1923694759607315,
"learning_rate": 6.650057515511848e-05,
"loss": 0.0064,
"step": 12700
},
{
"grad_norm": 0.23878122866153717,
"learning_rate": 6.644853722976667e-05,
"loss": 0.0068,
"step": 12710
},
{
"grad_norm": 0.19844236969947815,
"learning_rate": 6.639647931789113e-05,
"loss": 0.0068,
"step": 12720
},
{
"grad_norm": 0.3030112385749817,
"learning_rate": 6.634440148274713e-05,
"loss": 0.0072,
"step": 12730
},
{
"grad_norm": 0.29436954855918884,
"learning_rate": 6.629230378761415e-05,
"loss": 0.0085,
"step": 12740
},
{
"grad_norm": 0.251598984003067,
"learning_rate": 6.624018629579581e-05,
"loss": 0.0078,
"step": 12750
},
{
"grad_norm": 0.22182928025722504,
"learning_rate": 6.618804907061976e-05,
"loss": 0.0069,
"step": 12760
},
{
"grad_norm": 0.2940364181995392,
"learning_rate": 6.613589217543766e-05,
"loss": 0.0076,
"step": 12770
},
{
"grad_norm": 0.20308679342269897,
"learning_rate": 6.608371567362504e-05,
"loss": 0.0077,
"step": 12780
},
{
"grad_norm": 0.3306617736816406,
"learning_rate": 6.60315196285813e-05,
"loss": 0.0069,
"step": 12790
},
{
"grad_norm": 0.27817803621292114,
"learning_rate": 6.597930410372954e-05,
"loss": 0.008,
"step": 12800
},
{
"grad_norm": 0.22359612584114075,
"learning_rate": 6.592706916251653e-05,
"loss": 0.0067,
"step": 12810
},
{
"grad_norm": 0.2547834813594818,
"learning_rate": 6.587481486841267e-05,
"loss": 0.0069,
"step": 12820
},
{
"grad_norm": 0.23497673869132996,
"learning_rate": 6.582254128491184e-05,
"loss": 0.006,
"step": 12830
},
{
"grad_norm": 0.2852087616920471,
"learning_rate": 6.577024847553139e-05,
"loss": 0.0073,
"step": 12840
},
{
"grad_norm": 0.3237002491950989,
"learning_rate": 6.571793650381201e-05,
"loss": 0.0077,
"step": 12850
},
{
"grad_norm": 0.2525921165943146,
"learning_rate": 6.566560543331766e-05,
"loss": 0.0078,
"step": 12860
},
{
"grad_norm": 0.29381003975868225,
"learning_rate": 6.561325532763554e-05,
"loss": 0.0074,
"step": 12870
},
{
"grad_norm": 0.2689189016819,
"learning_rate": 6.556088625037597e-05,
"loss": 0.0072,
"step": 12880
},
{
"grad_norm": 0.26632052659988403,
"learning_rate": 6.55084982651723e-05,
"loss": 0.0067,
"step": 12890
},
{
"grad_norm": 0.19162796437740326,
"learning_rate": 6.54560914356809e-05,
"loss": 0.0077,
"step": 12900
},
{
"grad_norm": 0.2577548027038574,
"learning_rate": 6.540366582558097e-05,
"loss": 0.0066,
"step": 12910
},
{
"grad_norm": 0.2573550045490265,
"learning_rate": 6.53512214985746e-05,
"loss": 0.0064,
"step": 12920
},
{
"grad_norm": 0.27676379680633545,
"learning_rate": 6.529875851838658e-05,
"loss": 0.0064,
"step": 12930
},
{
"grad_norm": 0.23883642256259918,
"learning_rate": 6.52462769487644e-05,
"loss": 0.007,
"step": 12940
},
{
"grad_norm": 0.2487030029296875,
"learning_rate": 6.519377685347807e-05,
"loss": 0.0074,
"step": 12950
},
{
"grad_norm": 0.2392069399356842,
"learning_rate": 6.514125829632021e-05,
"loss": 0.007,
"step": 12960
},
{
"grad_norm": 0.2278149276971817,
"learning_rate": 6.508872134110577e-05,
"loss": 0.0077,
"step": 12970
},
{
"grad_norm": 0.26694855093955994,
"learning_rate": 6.503616605167213e-05,
"loss": 0.0072,
"step": 12980
},
{
"grad_norm": 0.2206481695175171,
"learning_rate": 6.498359249187893e-05,
"loss": 0.0079,
"step": 12990
},
{
"grad_norm": 0.2661944627761841,
"learning_rate": 6.493100072560799e-05,
"loss": 0.0075,
"step": 13000
},
{
"grad_norm": 0.2793487012386322,
"learning_rate": 6.487839081676327e-05,
"loss": 0.0069,
"step": 13010
},
{
"grad_norm": 0.28858742117881775,
"learning_rate": 6.482576282927076e-05,
"loss": 0.0076,
"step": 13020
},
{
"grad_norm": 0.3059164583683014,
"learning_rate": 6.477311682707845e-05,
"loss": 0.0079,
"step": 13030
},
{
"grad_norm": 0.2655072510242462,
"learning_rate": 6.472045287415616e-05,
"loss": 0.0067,
"step": 13040
},
{
"grad_norm": 0.20142611861228943,
"learning_rate": 6.466777103449558e-05,
"loss": 0.006,
"step": 13050
},
{
"grad_norm": 0.24640636146068573,
"learning_rate": 6.461507137211012e-05,
"loss": 0.0069,
"step": 13060
},
{
"grad_norm": 0.26599934697151184,
"learning_rate": 6.456235395103483e-05,
"loss": 0.0062,
"step": 13070
},
{
"grad_norm": 0.2124205380678177,
"learning_rate": 6.450961883532634e-05,
"loss": 0.0066,
"step": 13080
},
{
"grad_norm": 0.2221207469701767,
"learning_rate": 6.445686608906283e-05,
"loss": 0.0064,
"step": 13090
},
{
"grad_norm": 0.22872713208198547,
"learning_rate": 6.44040957763438e-05,
"loss": 0.0068,
"step": 13100
},
{
"grad_norm": 0.24420705437660217,
"learning_rate": 6.435130796129018e-05,
"loss": 0.0062,
"step": 13110
},
{
"grad_norm": 0.21613144874572754,
"learning_rate": 6.429850270804416e-05,
"loss": 0.0072,
"step": 13120
},
{
"grad_norm": 0.22340647876262665,
"learning_rate": 6.424568008076909e-05,
"loss": 0.0084,
"step": 13130
},
{
"grad_norm": 0.26508018374443054,
"learning_rate": 6.419284014364943e-05,
"loss": 0.007,
"step": 13140
},
{
"grad_norm": 0.2540311813354492,
"learning_rate": 6.41399829608907e-05,
"loss": 0.0063,
"step": 13150
},
{
"grad_norm": 0.24151162803173065,
"learning_rate": 6.408710859671938e-05,
"loss": 0.0085,
"step": 13160
},
{
"grad_norm": 0.23258985579013824,
"learning_rate": 6.403421711538278e-05,
"loss": 0.008,
"step": 13170
},
{
"grad_norm": 0.21937763690948486,
"learning_rate": 6.398130858114903e-05,
"loss": 0.0061,
"step": 13180
},
{
"grad_norm": 0.2620888948440552,
"learning_rate": 6.392838305830701e-05,
"loss": 0.0067,
"step": 13190
},
{
"grad_norm": 0.25849398970603943,
"learning_rate": 6.387544061116621e-05,
"loss": 0.0083,
"step": 13200
},
{
"grad_norm": 0.19696420431137085,
"learning_rate": 6.382248130405671e-05,
"loss": 0.0067,
"step": 13210
},
{
"grad_norm": 0.20594185590744019,
"learning_rate": 6.376950520132905e-05,
"loss": 0.0073,
"step": 13220
},
{
"grad_norm": 0.2561691701412201,
"learning_rate": 6.371651236735418e-05,
"loss": 0.0062,
"step": 13230
},
{
"grad_norm": 0.2856631875038147,
"learning_rate": 6.366350286652341e-05,
"loss": 0.0074,
"step": 13240
},
{
"grad_norm": 0.22115884721279144,
"learning_rate": 6.361047676324827e-05,
"loss": 0.0072,
"step": 13250
},
{
"grad_norm": 0.21611319482326508,
"learning_rate": 6.355743412196047e-05,
"loss": 0.007,
"step": 13260
},
{
"grad_norm": 0.25783073902130127,
"learning_rate": 6.350437500711184e-05,
"loss": 0.008,
"step": 13270
},
{
"grad_norm": 0.2509796619415283,
"learning_rate": 6.345129948317418e-05,
"loss": 0.0066,
"step": 13280
},
{
"grad_norm": 0.20480823516845703,
"learning_rate": 6.33982076146393e-05,
"loss": 0.0078,
"step": 13290
},
{
"grad_norm": 0.22611509263515472,
"learning_rate": 6.334509946601879e-05,
"loss": 0.0077,
"step": 13300
},
{
"grad_norm": 0.2350149154663086,
"learning_rate": 6.329197510184406e-05,
"loss": 0.0073,
"step": 13310
},
{
"grad_norm": 0.1989920735359192,
"learning_rate": 6.323883458666624e-05,
"loss": 0.0075,
"step": 13320
},
{
"grad_norm": 0.24860800802707672,
"learning_rate": 6.318567798505605e-05,
"loss": 0.0069,
"step": 13330
},
{
"grad_norm": 0.14864976704120636,
"learning_rate": 6.313250536160378e-05,
"loss": 0.0067,
"step": 13340
},
{
"grad_norm": 0.2616286873817444,
"learning_rate": 6.307931678091917e-05,
"loss": 0.0065,
"step": 13350
},
{
"grad_norm": 0.27229225635528564,
"learning_rate": 6.302611230763138e-05,
"loss": 0.0071,
"step": 13360
},
{
"grad_norm": 0.25116783380508423,
"learning_rate": 6.297289200638887e-05,
"loss": 0.0083,
"step": 13370
},
{
"grad_norm": 0.24302475154399872,
"learning_rate": 6.29196559418593e-05,
"loss": 0.0068,
"step": 13380
},
{
"grad_norm": 0.18964216113090515,
"learning_rate": 6.286640417872951e-05,
"loss": 0.0072,
"step": 13390
},
{
"grad_norm": 0.21581992506980896,
"learning_rate": 6.281313678170542e-05,
"loss": 0.0083,
"step": 13400
},
{
"grad_norm": 0.270751416683197,
"learning_rate": 6.275985381551194e-05,
"loss": 0.0066,
"step": 13410
},
{
"grad_norm": 0.24866518378257751,
"learning_rate": 6.270655534489292e-05,
"loss": 0.0065,
"step": 13420
},
{
"grad_norm": 0.2527863383293152,
"learning_rate": 6.265324143461098e-05,
"loss": 0.0071,
"step": 13430
},
{
"grad_norm": 0.20228345692157745,
"learning_rate": 6.259991214944758e-05,
"loss": 0.0065,
"step": 13440
},
{
"grad_norm": 0.23330335319042206,
"learning_rate": 6.254656755420283e-05,
"loss": 0.0064,
"step": 13450
},
{
"grad_norm": 0.23438553512096405,
"learning_rate": 6.249320771369545e-05,
"loss": 0.0067,
"step": 13460
},
{
"grad_norm": 0.20001573860645294,
"learning_rate": 6.243983269276263e-05,
"loss": 0.0068,
"step": 13470
},
{
"grad_norm": 0.23964233696460724,
"learning_rate": 6.238644255626012e-05,
"loss": 0.0061,
"step": 13480
},
{
"grad_norm": 0.21075086295604706,
"learning_rate": 6.233303736906193e-05,
"loss": 0.0061,
"step": 13490
},
{
"grad_norm": 0.2517348527908325,
"learning_rate": 6.227961719606039e-05,
"loss": 0.0075,
"step": 13500
},
{
"grad_norm": 0.19695648550987244,
"learning_rate": 6.222618210216608e-05,
"loss": 0.0066,
"step": 13510
},
{
"grad_norm": 0.2171541452407837,
"learning_rate": 6.217273215230767e-05,
"loss": 0.008,
"step": 13520
},
{
"grad_norm": 0.23161739110946655,
"learning_rate": 6.211926741143188e-05,
"loss": 0.0065,
"step": 13530
},
{
"grad_norm": 0.2714325487613678,
"learning_rate": 6.20657879445034e-05,
"loss": 0.0075,
"step": 13540
},
{
"grad_norm": 0.30451586842536926,
"learning_rate": 6.201229381650485e-05,
"loss": 0.0064,
"step": 13550
},
{
"grad_norm": 0.2556527256965637,
"learning_rate": 6.195878509243661e-05,
"loss": 0.0061,
"step": 13560
},
{
"grad_norm": 0.253057599067688,
"learning_rate": 6.190526183731685e-05,
"loss": 0.007,
"step": 13570
},
{
"grad_norm": 0.19131217896938324,
"learning_rate": 6.185172411618138e-05,
"loss": 0.0061,
"step": 13580
},
{
"grad_norm": 0.30017247796058655,
"learning_rate": 6.179817199408355e-05,
"loss": 0.0071,
"step": 13590
},
{
"grad_norm": 0.21754956245422363,
"learning_rate": 6.174460553609426e-05,
"loss": 0.0065,
"step": 13600
},
{
"grad_norm": 0.2323118895292282,
"learning_rate": 6.16910248073018e-05,
"loss": 0.0072,
"step": 13610
},
{
"grad_norm": 0.273316890001297,
"learning_rate": 6.16374298728118e-05,
"loss": 0.0061,
"step": 13620
},
{
"grad_norm": 0.21758437156677246,
"learning_rate": 6.158382079774716e-05,
"loss": 0.0072,
"step": 13630
},
{
"grad_norm": 0.21978215873241425,
"learning_rate": 6.153019764724799e-05,
"loss": 0.0063,
"step": 13640
},
{
"grad_norm": 0.27762216329574585,
"learning_rate": 6.147656048647144e-05,
"loss": 0.0071,
"step": 13650
},
{
"grad_norm": 0.26004552841186523,
"learning_rate": 6.142290938059173e-05,
"loss": 0.0067,
"step": 13660
},
{
"grad_norm": 0.26268357038497925,
"learning_rate": 6.136924439480001e-05,
"loss": 0.0071,
"step": 13670
},
{
"grad_norm": 0.2489512413740158,
"learning_rate": 6.13155655943043e-05,
"loss": 0.0061,
"step": 13680
},
{
"grad_norm": 0.19315683841705322,
"learning_rate": 6.126187304432941e-05,
"loss": 0.0056,
"step": 13690
},
{
"grad_norm": 0.19838185608386993,
"learning_rate": 6.120816681011684e-05,
"loss": 0.0058,
"step": 13700
},
{
"grad_norm": 0.22212015092372894,
"learning_rate": 6.115444695692474e-05,
"loss": 0.0057,
"step": 13710
},
{
"grad_norm": 0.27834391593933105,
"learning_rate": 6.110071355002779e-05,
"loss": 0.007,
"step": 13720
},
{
"grad_norm": 0.21555571258068085,
"learning_rate": 6.104696665471714e-05,
"loss": 0.0073,
"step": 13730
},
{
"grad_norm": 0.25469234585762024,
"learning_rate": 6.099320633630036e-05,
"loss": 0.0062,
"step": 13740
},
{
"grad_norm": 0.24184435606002808,
"learning_rate": 6.093943266010128e-05,
"loss": 0.0062,
"step": 13750
},
{
"grad_norm": 0.20043589174747467,
"learning_rate": 6.088564569146e-05,
"loss": 0.0066,
"step": 13760
},
{
"grad_norm": 0.19360867142677307,
"learning_rate": 6.083184549573274e-05,
"loss": 0.006,
"step": 13770
},
{
"grad_norm": 0.257904052734375,
"learning_rate": 6.077803213829184e-05,
"loss": 0.0068,
"step": 13780
},
{
"grad_norm": 0.228912353515625,
"learning_rate": 6.0724205684525594e-05,
"loss": 0.0061,
"step": 13790
},
{
"grad_norm": 0.19492770731449127,
"learning_rate": 6.0670366199838215e-05,
"loss": 0.0068,
"step": 13800
},
{
"grad_norm": 0.2720438539981842,
"learning_rate": 6.061651374964974e-05,
"loss": 0.0068,
"step": 13810
},
{
"grad_norm": 0.26898762583732605,
"learning_rate": 6.0562648399396004e-05,
"loss": 0.0067,
"step": 13820
},
{
"grad_norm": 0.24783796072006226,
"learning_rate": 6.050877021452845e-05,
"loss": 0.0064,
"step": 13830
},
{
"grad_norm": 0.22201155126094818,
"learning_rate": 6.04548792605142e-05,
"loss": 0.0061,
"step": 13840
},
{
"grad_norm": 0.1546681523323059,
"learning_rate": 6.04009756028358e-05,
"loss": 0.0068,
"step": 13850
},
{
"grad_norm": 0.25796008110046387,
"learning_rate": 6.0347059306991294e-05,
"loss": 0.0083,
"step": 13860
},
{
"grad_norm": 0.2468590885400772,
"learning_rate": 6.029313043849407e-05,
"loss": 0.0066,
"step": 13870
},
{
"grad_norm": 0.29722049832344055,
"learning_rate": 6.023918906287279e-05,
"loss": 0.0069,
"step": 13880
},
{
"grad_norm": 0.23768748342990875,
"learning_rate": 6.01852352456713e-05,
"loss": 0.0067,
"step": 13890
},
{
"grad_norm": 0.23087532818317413,
"learning_rate": 6.013126905244858e-05,
"loss": 0.0074,
"step": 13900
},
{
"grad_norm": 0.2194361388683319,
"learning_rate": 6.007729054877864e-05,
"loss": 0.0067,
"step": 13910
},
{
"grad_norm": 0.25105589628219604,
"learning_rate": 6.0023299800250466e-05,
"loss": 0.0062,
"step": 13920
},
{
"grad_norm": 0.2088538259267807,
"learning_rate": 5.9969296872467894e-05,
"loss": 0.0058,
"step": 13930
},
{
"grad_norm": 0.19388312101364136,
"learning_rate": 5.9915281831049585e-05,
"loss": 0.0074,
"step": 13940
},
{
"grad_norm": 0.25578510761260986,
"learning_rate": 5.98612547416289e-05,
"loss": 0.0073,
"step": 13950
},
{
"grad_norm": 0.2675463557243347,
"learning_rate": 5.9807215669853855e-05,
"loss": 0.0084,
"step": 13960
},
{
"grad_norm": 0.23419392108917236,
"learning_rate": 5.9753164681387e-05,
"loss": 0.0062,
"step": 13970
},
{
"grad_norm": 0.21404705941677094,
"learning_rate": 5.969910184190539e-05,
"loss": 0.0062,
"step": 13980
},
{
"grad_norm": 0.20408913493156433,
"learning_rate": 5.9645027217100477e-05,
"loss": 0.0056,
"step": 13990
},
{
"grad_norm": 0.28686222434043884,
"learning_rate": 5.959094087267804e-05,
"loss": 0.0076,
"step": 14000
},
{
"grad_norm": 0.20522059500217438,
"learning_rate": 5.9536842874358065e-05,
"loss": 0.0072,
"step": 14010
},
{
"grad_norm": 0.2946614921092987,
"learning_rate": 5.9482733287874734e-05,
"loss": 0.0068,
"step": 14020
},
{
"grad_norm": 0.3324415981769562,
"learning_rate": 5.942861217897631e-05,
"loss": 0.0074,
"step": 14030
},
{
"grad_norm": 0.27357664704322815,
"learning_rate": 5.9374479613425e-05,
"loss": 0.0071,
"step": 14040
},
{
"grad_norm": 0.21822060644626617,
"learning_rate": 5.932033565699704e-05,
"loss": 0.0066,
"step": 14050
},
{
"grad_norm": 0.22324790060520172,
"learning_rate": 5.926618037548237e-05,
"loss": 0.0072,
"step": 14060
},
{
"grad_norm": 0.22239309549331665,
"learning_rate": 5.9212013834684824e-05,
"loss": 0.0076,
"step": 14070
},
{
"grad_norm": 0.20068594813346863,
"learning_rate": 5.9157836100421824e-05,
"loss": 0.007,
"step": 14080
},
{
"grad_norm": 0.18320992588996887,
"learning_rate": 5.910364723852444e-05,
"loss": 0.0069,
"step": 14090
},
{
"grad_norm": 0.2011718899011612,
"learning_rate": 5.904944731483723e-05,
"loss": 0.0057,
"step": 14100
},
{
"grad_norm": 0.2642921805381775,
"learning_rate": 5.899523639521825e-05,
"loss": 0.006,
"step": 14110
},
{
"grad_norm": 0.2612369954586029,
"learning_rate": 5.894101454553883e-05,
"loss": 0.0063,
"step": 14120
},
{
"grad_norm": 0.2539650499820709,
"learning_rate": 5.888678183168368e-05,
"loss": 0.0078,
"step": 14130
},
{
"grad_norm": 0.24627746641635895,
"learning_rate": 5.883253831955061e-05,
"loss": 0.0066,
"step": 14140
},
{
"grad_norm": 0.2440510392189026,
"learning_rate": 5.8778284075050625e-05,
"loss": 0.0058,
"step": 14150
},
{
"grad_norm": 0.20609524846076965,
"learning_rate": 5.872401916410777e-05,
"loss": 0.0064,
"step": 14160
},
{
"grad_norm": 0.265025794506073,
"learning_rate": 5.8669743652659014e-05,
"loss": 0.0062,
"step": 14170
},
{
"grad_norm": 0.1870139241218567,
"learning_rate": 5.86154576066542e-05,
"loss": 0.0062,
"step": 14180
},
{
"grad_norm": 0.25761324167251587,
"learning_rate": 5.856116109205602e-05,
"loss": 0.0062,
"step": 14190
},
{
"grad_norm": 0.20956628024578094,
"learning_rate": 5.850685417483983e-05,
"loss": 0.0063,
"step": 14200
},
{
"grad_norm": 0.20504550635814667,
"learning_rate": 5.8452536920993684e-05,
"loss": 0.0056,
"step": 14210
},
{
"grad_norm": 0.20799843966960907,
"learning_rate": 5.839820939651817e-05,
"loss": 0.0054,
"step": 14220
},
{
"grad_norm": 0.1858857274055481,
"learning_rate": 5.834387166742632e-05,
"loss": 0.0056,
"step": 14230
},
{
"grad_norm": 0.19511808454990387,
"learning_rate": 5.828952379974364e-05,
"loss": 0.0066,
"step": 14240
},
{
"grad_norm": 0.2534532845020294,
"learning_rate": 5.8235165859507864e-05,
"loss": 0.0064,
"step": 14250
},
{
"grad_norm": 0.2107435017824173,
"learning_rate": 5.818079791276907e-05,
"loss": 0.0057,
"step": 14260
},
{
"grad_norm": 0.23028282821178436,
"learning_rate": 5.8126420025589415e-05,
"loss": 0.006,
"step": 14270
},
{
"grad_norm": 0.1975833773612976,
"learning_rate": 5.807203226404313e-05,
"loss": 0.0057,
"step": 14280
},
{
"grad_norm": 0.2542743682861328,
"learning_rate": 5.801763469421652e-05,
"loss": 0.0078,
"step": 14290
},
{
"grad_norm": 0.23853014409542084,
"learning_rate": 5.796322738220774e-05,
"loss": 0.007,
"step": 14300
},
{
"grad_norm": 0.2695060670375824,
"learning_rate": 5.79088103941268e-05,
"loss": 0.0063,
"step": 14310
},
{
"grad_norm": 0.23394541442394257,
"learning_rate": 5.785438379609549e-05,
"loss": 0.0063,
"step": 14320
},
{
"grad_norm": 0.19033122062683105,
"learning_rate": 5.779994765424724e-05,
"loss": 0.0056,
"step": 14330
},
{
"grad_norm": 0.19316981732845306,
"learning_rate": 5.7745502034727114e-05,
"loss": 0.0061,
"step": 14340
},
{
"grad_norm": 0.24229130148887634,
"learning_rate": 5.769104700369164e-05,
"loss": 0.0063,
"step": 14350
},
{
"grad_norm": 0.23179645836353302,
"learning_rate": 5.7636582627308854e-05,
"loss": 0.0059,
"step": 14360
},
{
"grad_norm": 0.24487490952014923,
"learning_rate": 5.7582108971758095e-05,
"loss": 0.0061,
"step": 14370
},
{
"grad_norm": 0.2611483633518219,
"learning_rate": 5.752762610322999e-05,
"loss": 0.0075,
"step": 14380
},
{
"grad_norm": 0.247283935546875,
"learning_rate": 5.747313408792636e-05,
"loss": 0.0063,
"step": 14390
},
{
"grad_norm": 0.2301962524652481,
"learning_rate": 5.741863299206014e-05,
"loss": 0.0058,
"step": 14400
},
{
"grad_norm": 0.25670161843299866,
"learning_rate": 5.73641228818553e-05,
"loss": 0.0072,
"step": 14410
},
{
"grad_norm": 0.20070618391036987,
"learning_rate": 5.730960382354676e-05,
"loss": 0.0061,
"step": 14420
},
{
"grad_norm": 0.2672232687473297,
"learning_rate": 5.725507588338035e-05,
"loss": 0.0066,
"step": 14430
},
{
"grad_norm": 0.3072107136249542,
"learning_rate": 5.7200539127612604e-05,
"loss": 0.0076,
"step": 14440
},
{
"grad_norm": 0.2517894506454468,
"learning_rate": 5.714599362251087e-05,
"loss": 0.0061,
"step": 14450
},
{
"grad_norm": 0.13219939172267914,
"learning_rate": 5.709143943435307e-05,
"loss": 0.007,
"step": 14460
},
{
"grad_norm": 0.19510792195796967,
"learning_rate": 5.7036876629427646e-05,
"loss": 0.006,
"step": 14470
},
{
"grad_norm": 0.22808505594730377,
"learning_rate": 5.698230527403361e-05,
"loss": 0.0073,
"step": 14480
},
{
"grad_norm": 0.22400452196598053,
"learning_rate": 5.6927725434480274e-05,
"loss": 0.006,
"step": 14490
},
{
"grad_norm": 0.23427478969097137,
"learning_rate": 5.687313717708728e-05,
"loss": 0.0061,
"step": 14500
},
{
"grad_norm": 0.17823611199855804,
"learning_rate": 5.681854056818453e-05,
"loss": 0.0062,
"step": 14510
},
{
"grad_norm": 0.21296104788780212,
"learning_rate": 5.676393567411205e-05,
"loss": 0.0062,
"step": 14520
},
{
"grad_norm": 0.2652239501476288,
"learning_rate": 5.670932256121991e-05,
"loss": 0.007,
"step": 14530
},
{
"grad_norm": 0.2570689022541046,
"learning_rate": 5.6654701295868216e-05,
"loss": 0.0068,
"step": 14540
},
{
"grad_norm": 0.22959604859352112,
"learning_rate": 5.660007194442697e-05,
"loss": 0.0057,
"step": 14550
},
{
"grad_norm": 0.2390802949666977,
"learning_rate": 5.6545434573275946e-05,
"loss": 0.0063,
"step": 14560
},
{
"grad_norm": 0.23341608047485352,
"learning_rate": 5.649078924880472e-05,
"loss": 0.0066,
"step": 14570
},
{
"grad_norm": 0.213889017701149,
"learning_rate": 5.643613603741252e-05,
"loss": 0.0072,
"step": 14580
},
{
"grad_norm": 0.29240983724594116,
"learning_rate": 5.6381475005508154e-05,
"loss": 0.0067,
"step": 14590
},
{
"grad_norm": 0.2720557451248169,
"learning_rate": 5.632680621950992e-05,
"loss": 0.0058,
"step": 14600
},
{
"grad_norm": 0.21651709079742432,
"learning_rate": 5.627212974584555e-05,
"loss": 0.0067,
"step": 14610
},
{
"grad_norm": 0.2748400568962097,
"learning_rate": 5.62174456509521e-05,
"loss": 0.0063,
"step": 14620
},
{
"grad_norm": 0.22031934559345245,
"learning_rate": 5.616275400127594e-05,
"loss": 0.0065,
"step": 14630
},
{
"grad_norm": 0.2475021928548813,
"learning_rate": 5.610805486327254e-05,
"loss": 0.0068,
"step": 14640
},
{
"grad_norm": 0.2812642753124237,
"learning_rate": 5.6053348303406536e-05,
"loss": 0.0068,
"step": 14650
},
{
"grad_norm": 0.2446913719177246,
"learning_rate": 5.599863438815156e-05,
"loss": 0.0065,
"step": 14660
},
{
"grad_norm": 0.24713528156280518,
"learning_rate": 5.594391318399017e-05,
"loss": 0.0064,
"step": 14670
},
{
"grad_norm": 0.223225399851799,
"learning_rate": 5.588918475741377e-05,
"loss": 0.0068,
"step": 14680
},
{
"grad_norm": 0.21757501363754272,
"learning_rate": 5.5834449174922586e-05,
"loss": 0.0056,
"step": 14690
},
{
"grad_norm": 0.2678114175796509,
"learning_rate": 5.57797065030255e-05,
"loss": 0.0059,
"step": 14700
},
{
"grad_norm": 0.26760223507881165,
"learning_rate": 5.5724956808240016e-05,
"loss": 0.0061,
"step": 14710
},
{
"grad_norm": 0.1902293711900711,
"learning_rate": 5.567020015709219e-05,
"loss": 0.0064,
"step": 14720
},
{
"grad_norm": 0.2306230366230011,
"learning_rate": 5.561543661611649e-05,
"loss": 0.007,
"step": 14730
},
{
"grad_norm": 0.2748568058013916,
"learning_rate": 5.556066625185583e-05,
"loss": 0.007,
"step": 14740
},
{
"grad_norm": 0.17141860723495483,
"learning_rate": 5.550588913086131e-05,
"loss": 0.0072,
"step": 14750
},
{
"grad_norm": 0.22994983196258545,
"learning_rate": 5.545110531969234e-05,
"loss": 0.0064,
"step": 14760
},
{
"grad_norm": 0.24939128756523132,
"learning_rate": 5.539631488491641e-05,
"loss": 0.0059,
"step": 14770
},
{
"grad_norm": 0.22121156752109528,
"learning_rate": 5.534151789310904e-05,
"loss": 0.006,
"step": 14780
},
{
"grad_norm": 0.3021535277366638,
"learning_rate": 5.528671441085376e-05,
"loss": 0.0068,
"step": 14790
},
{
"grad_norm": 0.21206553280353546,
"learning_rate": 5.523190450474197e-05,
"loss": 0.0056,
"step": 14800
},
{
"grad_norm": 0.20038598775863647,
"learning_rate": 5.5177088241372865e-05,
"loss": 0.0055,
"step": 14810
},
{
"grad_norm": 0.24848155677318573,
"learning_rate": 5.512226568735338e-05,
"loss": 0.0074,
"step": 14820
},
{
"grad_norm": 0.20866146683692932,
"learning_rate": 5.506743690929809e-05,
"loss": 0.0055,
"step": 14830
},
{
"grad_norm": 0.207877978682518,
"learning_rate": 5.501260197382912e-05,
"loss": 0.0054,
"step": 14840
},
{
"grad_norm": 0.20717130601406097,
"learning_rate": 5.49577609475761e-05,
"loss": 0.0067,
"step": 14850
},
{
"grad_norm": 0.24062608182430267,
"learning_rate": 5.490291389717603e-05,
"loss": 0.0059,
"step": 14860
},
{
"grad_norm": 0.21328893303871155,
"learning_rate": 5.484806088927329e-05,
"loss": 0.0053,
"step": 14870
},
{
"grad_norm": 0.2662334442138672,
"learning_rate": 5.479320199051942e-05,
"loss": 0.0062,
"step": 14880
},
{
"grad_norm": 0.1990499496459961,
"learning_rate": 5.473833726757314e-05,
"loss": 0.0064,
"step": 14890
},
{
"grad_norm": 0.19696538150310516,
"learning_rate": 5.4683466787100304e-05,
"loss": 0.0058,
"step": 14900
},
{
"grad_norm": 0.1558971107006073,
"learning_rate": 5.462859061577369e-05,
"loss": 0.0064,
"step": 14910
},
{
"grad_norm": 0.19675497710704803,
"learning_rate": 5.4573708820273026e-05,
"loss": 0.0059,
"step": 14920
},
{
"grad_norm": 0.20575101673603058,
"learning_rate": 5.451882146728488e-05,
"loss": 0.0066,
"step": 14930
},
{
"grad_norm": 0.1856343150138855,
"learning_rate": 5.446392862350255e-05,
"loss": 0.0058,
"step": 14940
},
{
"grad_norm": 0.22339941561222076,
"learning_rate": 5.440903035562603e-05,
"loss": 0.0061,
"step": 14950
},
{
"grad_norm": 0.19470913708209991,
"learning_rate": 5.435412673036188e-05,
"loss": 0.0062,
"step": 14960
},
{
"grad_norm": 0.21660403907299042,
"learning_rate": 5.429921781442318e-05,
"loss": 0.0064,
"step": 14970
},
{
"grad_norm": 0.1847912073135376,
"learning_rate": 5.424430367452945e-05,
"loss": 0.006,
"step": 14980
},
{
"grad_norm": 0.19388660788536072,
"learning_rate": 5.418938437740655e-05,
"loss": 0.006,
"step": 14990
},
{
"grad_norm": 0.17676885426044464,
"learning_rate": 5.413445998978658e-05,
"loss": 0.0051,
"step": 15000
},
{
"grad_norm": 0.2650805115699768,
"learning_rate": 5.407953057840789e-05,
"loss": 0.0065,
"step": 15010
},
{
"grad_norm": 0.20982669293880463,
"learning_rate": 5.4024596210014853e-05,
"loss": 0.0065,
"step": 15020
},
{
"grad_norm": 0.19943812489509583,
"learning_rate": 5.3969656951357935e-05,
"loss": 0.0059,
"step": 15030
},
{
"grad_norm": 0.1904461681842804,
"learning_rate": 5.3914712869193504e-05,
"loss": 0.0071,
"step": 15040
},
{
"grad_norm": 0.19348599016666412,
"learning_rate": 5.385976403028381e-05,
"loss": 0.0076,
"step": 15050
},
{
"grad_norm": 0.19900880753993988,
"learning_rate": 5.3804810501396875e-05,
"loss": 0.0057,
"step": 15060
},
{
"grad_norm": 0.2089642435312271,
"learning_rate": 5.374985234930639e-05,
"loss": 0.006,
"step": 15070
},
{
"grad_norm": 0.20423781871795654,
"learning_rate": 5.3694889640791724e-05,
"loss": 0.0056,
"step": 15080
},
{
"grad_norm": 0.18859027326107025,
"learning_rate": 5.3639922442637736e-05,
"loss": 0.0063,
"step": 15090
},
{
"grad_norm": 0.2387283593416214,
"learning_rate": 5.358495082163476e-05,
"loss": 0.0055,
"step": 15100
},
{
"grad_norm": 0.23824751377105713,
"learning_rate": 5.35299748445785e-05,
"loss": 0.0058,
"step": 15110
},
{
"grad_norm": 0.3053557574748993,
"learning_rate": 5.3474994578269944e-05,
"loss": 0.0063,
"step": 15120
},
{
"grad_norm": 0.25392332673072815,
"learning_rate": 5.34200100895153e-05,
"loss": 0.0068,
"step": 15130
},
{
"grad_norm": 0.17796850204467773,
"learning_rate": 5.3365021445125916e-05,
"loss": 0.0064,
"step": 15140
},
{
"grad_norm": 0.18557415902614594,
"learning_rate": 5.331002871191817e-05,
"loss": 0.0059,
"step": 15150
},
{
"grad_norm": 0.17247475683689117,
"learning_rate": 5.3255031956713443e-05,
"loss": 0.0053,
"step": 15160
},
{
"grad_norm": 0.20370087027549744,
"learning_rate": 5.320003124633795e-05,
"loss": 0.0072,
"step": 15170
},
{
"grad_norm": 0.2208911031484604,
"learning_rate": 5.314502664762274e-05,
"loss": 0.0062,
"step": 15180
},
{
"grad_norm": 0.18324925005435944,
"learning_rate": 5.3090018227403604e-05,
"loss": 0.0056,
"step": 15190
},
{
"grad_norm": 0.2513402998447418,
"learning_rate": 5.3035006052520955e-05,
"loss": 0.0069,
"step": 15200
},
{
"grad_norm": 0.24593324959278107,
"learning_rate": 5.297999018981976e-05,
"loss": 0.0075,
"step": 15210
},
{
"grad_norm": 0.20086830854415894,
"learning_rate": 5.2924970706149505e-05,
"loss": 0.005,
"step": 15220
},
{
"grad_norm": 0.219964399933815,
"learning_rate": 5.2869947668364014e-05,
"loss": 0.0057,
"step": 15230
},
{
"grad_norm": 0.16223347187042236,
"learning_rate": 5.2814921143321504e-05,
"loss": 0.0051,
"step": 15240
},
{
"grad_norm": 0.19967032968997955,
"learning_rate": 5.275989119788436e-05,
"loss": 0.0051,
"step": 15250
},
{
"grad_norm": 0.19502277672290802,
"learning_rate": 5.270485789891919e-05,
"loss": 0.0058,
"step": 15260
},
{
"grad_norm": 0.190113365650177,
"learning_rate": 5.264982131329661e-05,
"loss": 0.0057,
"step": 15270
},
{
"grad_norm": 0.17518578469753265,
"learning_rate": 5.259478150789128e-05,
"loss": 0.0057,
"step": 15280
},
{
"grad_norm": 0.1997334212064743,
"learning_rate": 5.253973854958173e-05,
"loss": 0.0055,
"step": 15290
},
{
"grad_norm": 0.17797137796878815,
"learning_rate": 5.2484692505250376e-05,
"loss": 0.0062,
"step": 15300
},
{
"grad_norm": 0.20721536874771118,
"learning_rate": 5.242964344178333e-05,
"loss": 0.0056,
"step": 15310
},
{
"grad_norm": 0.21633847057819366,
"learning_rate": 5.237459142607041e-05,
"loss": 0.0051,
"step": 15320
},
{
"grad_norm": 0.2559736669063568,
"learning_rate": 5.231953652500498e-05,
"loss": 0.0053,
"step": 15330
},
{
"grad_norm": 0.333212286233902,
"learning_rate": 5.2264478805483976e-05,
"loss": 0.0065,
"step": 15340
},
{
"grad_norm": 0.2521781027317047,
"learning_rate": 5.220941833440768e-05,
"loss": 0.0057,
"step": 15350
},
{
"grad_norm": 0.20170144736766815,
"learning_rate": 5.215435517867977e-05,
"loss": 0.0059,
"step": 15360
},
{
"grad_norm": 0.2324669510126114,
"learning_rate": 5.209928940520719e-05,
"loss": 0.0065,
"step": 15370
},
{
"grad_norm": 0.20224812626838684,
"learning_rate": 5.204422108090004e-05,
"loss": 0.0062,
"step": 15380
},
{
"grad_norm": 0.2948107421398163,
"learning_rate": 5.19891502726715e-05,
"loss": 0.0054,
"step": 15390
},
{
"grad_norm": 0.22204329073429108,
"learning_rate": 5.193407704743782e-05,
"loss": 0.0054,
"step": 15400
},
{
"grad_norm": 0.3043134808540344,
"learning_rate": 5.187900147211815e-05,
"loss": 0.0056,
"step": 15410
},
{
"grad_norm": 0.24707256257534027,
"learning_rate": 5.182392361363453e-05,
"loss": 0.0053,
"step": 15420
},
{
"grad_norm": 0.25414401292800903,
"learning_rate": 5.176884353891172e-05,
"loss": 0.0051,
"step": 15430
},
{
"grad_norm": 0.22148506343364716,
"learning_rate": 5.171376131487722e-05,
"loss": 0.0051,
"step": 15440
},
{
"grad_norm": 0.18523675203323364,
"learning_rate": 5.165867700846113e-05,
"loss": 0.007,
"step": 15450
},
{
"grad_norm": 0.18334928154945374,
"learning_rate": 5.160359068659606e-05,
"loss": 0.0053,
"step": 15460
},
{
"grad_norm": 0.17495884001255035,
"learning_rate": 5.154850241621712e-05,
"loss": 0.006,
"step": 15470
},
{
"grad_norm": 0.2466358244419098,
"learning_rate": 5.149341226426172e-05,
"loss": 0.0064,
"step": 15480
},
{
"grad_norm": 0.2567436397075653,
"learning_rate": 5.14383202976696e-05,
"loss": 0.0047,
"step": 15490
},
{
"grad_norm": 0.22839559614658356,
"learning_rate": 5.1383226583382684e-05,
"loss": 0.0059,
"step": 15500
},
{
"grad_norm": 0.2122078835964203,
"learning_rate": 5.132813118834504e-05,
"loss": 0.0061,
"step": 15510
},
{
"grad_norm": 0.18810637295246124,
"learning_rate": 5.1273034179502777e-05,
"loss": 0.0056,
"step": 15520
},
{
"grad_norm": 0.2104889154434204,
"learning_rate": 5.1217935623803945e-05,
"loss": 0.0064,
"step": 15530
},
{
"grad_norm": 0.2014603614807129,
"learning_rate": 5.1162835588198476e-05,
"loss": 0.0052,
"step": 15540
},
{
"grad_norm": 0.20973831415176392,
"learning_rate": 5.110773413963813e-05,
"loss": 0.0057,
"step": 15550
},
{
"grad_norm": 0.24219033122062683,
"learning_rate": 5.105263134507636e-05,
"loss": 0.0061,
"step": 15560
},
{
"grad_norm": 0.22000108659267426,
"learning_rate": 5.099752727146824e-05,
"loss": 0.0054,
"step": 15570
},
{
"grad_norm": 0.20554766058921814,
"learning_rate": 5.094242198577042e-05,
"loss": 0.0058,
"step": 15580
},
{
"grad_norm": 0.21714714169502258,
"learning_rate": 5.088731555494102e-05,
"loss": 0.0062,
"step": 15590
},
{
"grad_norm": 0.22331134974956512,
"learning_rate": 5.0832208045939556e-05,
"loss": 0.0063,
"step": 15600
},
{
"grad_norm": 0.19590726494789124,
"learning_rate": 5.0777099525726844e-05,
"loss": 0.0059,
"step": 15610
},
{
"grad_norm": 0.20743784308433533,
"learning_rate": 5.072199006126494e-05,
"loss": 0.0048,
"step": 15620
},
{
"grad_norm": 0.2564847767353058,
"learning_rate": 5.0666879719517026e-05,
"loss": 0.0055,
"step": 15630
},
{
"grad_norm": 0.20698420703411102,
"learning_rate": 5.061176856744737e-05,
"loss": 0.0055,
"step": 15640
},
{
"grad_norm": 0.1851045787334442,
"learning_rate": 5.0556656672021205e-05,
"loss": 0.0053,
"step": 15650
},
{
"grad_norm": 0.210643008351326,
"learning_rate": 5.050154410020472e-05,
"loss": 0.0051,
"step": 15660
},
{
"grad_norm": 0.19705039262771606,
"learning_rate": 5.044643091896485e-05,
"loss": 0.0067,
"step": 15670
},
{
"grad_norm": 0.21425563097000122,
"learning_rate": 5.039131719526932e-05,
"loss": 0.0052,
"step": 15680
},
{
"grad_norm": 0.17841550707817078,
"learning_rate": 5.03362029960865e-05,
"loss": 0.0056,
"step": 15690
},
{
"grad_norm": 0.22135190665721893,
"learning_rate": 5.028108838838533e-05,
"loss": 0.0061,
"step": 15700
},
{
"grad_norm": 0.20895838737487793,
"learning_rate": 5.0225973439135276e-05,
"loss": 0.0062,
"step": 15710
},
{
"grad_norm": 0.2432011067867279,
"learning_rate": 5.017085821530617e-05,
"loss": 0.0058,
"step": 15720
},
{
"grad_norm": 0.15518110990524292,
"learning_rate": 5.011574278386822e-05,
"loss": 0.0053,
"step": 15730
},
{
"grad_norm": 0.22483888268470764,
"learning_rate": 5.006062721179189e-05,
"loss": 0.0054,
"step": 15740
},
{
"grad_norm": 0.17014619708061218,
"learning_rate": 5.000551156604777e-05,
"loss": 0.0051,
"step": 15750
},
{
"grad_norm": 0.17364399135112762,
"learning_rate": 4.9950395913606594e-05,
"loss": 0.0052,
"step": 15760
},
{
"grad_norm": 0.19253990054130554,
"learning_rate": 4.989528032143903e-05,
"loss": 0.0055,
"step": 15770
},
{
"grad_norm": 0.20240378379821777,
"learning_rate": 4.984016485651578e-05,
"loss": 0.0059,
"step": 15780
},
{
"grad_norm": 0.21755445003509521,
"learning_rate": 4.9785049585807274e-05,
"loss": 0.0054,
"step": 15790
},
{
"grad_norm": 0.150907963514328,
"learning_rate": 4.972993457628381e-05,
"loss": 0.0051,
"step": 15800
},
{
"grad_norm": 0.26881611347198486,
"learning_rate": 4.9674819894915306e-05,
"loss": 0.0055,
"step": 15810
},
{
"grad_norm": 0.2065955251455307,
"learning_rate": 4.9619705608671265e-05,
"loss": 0.0057,
"step": 15820
},
{
"grad_norm": 0.20155812799930573,
"learning_rate": 4.956459178452079e-05,
"loss": 0.0053,
"step": 15830
},
{
"grad_norm": 0.1860569268465042,
"learning_rate": 4.950947848943235e-05,
"loss": 0.0058,
"step": 15840
},
{
"grad_norm": 0.2278584986925125,
"learning_rate": 4.9454365790373805e-05,
"loss": 0.0051,
"step": 15850
},
{
"grad_norm": 0.17520834505558014,
"learning_rate": 4.939925375431226e-05,
"loss": 0.0054,
"step": 15860
},
{
"grad_norm": 0.19751957058906555,
"learning_rate": 4.9344142448214046e-05,
"loss": 0.0066,
"step": 15870
},
{
"grad_norm": 0.1812175214290619,
"learning_rate": 4.928903193904461e-05,
"loss": 0.0056,
"step": 15880
},
{
"grad_norm": 0.18880075216293335,
"learning_rate": 4.923392229376842e-05,
"loss": 0.0059,
"step": 15890
},
{
"grad_norm": 0.22860367596149445,
"learning_rate": 4.9178813579348865e-05,
"loss": 0.006,
"step": 15900
},
{
"grad_norm": 0.23990696668624878,
"learning_rate": 4.9123705862748253e-05,
"loss": 0.0057,
"step": 15910
},
{
"grad_norm": 0.20536799728870392,
"learning_rate": 4.9068599210927627e-05,
"loss": 0.0065,
"step": 15920
},
{
"grad_norm": 0.20585836470127106,
"learning_rate": 4.901349369084681e-05,
"loss": 0.0058,
"step": 15930
},
{
"grad_norm": 0.17564554512500763,
"learning_rate": 4.8958389369464156e-05,
"loss": 0.006,
"step": 15940
},
{
"grad_norm": 0.2334105223417282,
"learning_rate": 4.890328631373666e-05,
"loss": 0.0065,
"step": 15950
},
{
"grad_norm": 0.21485723555088043,
"learning_rate": 4.88481845906197e-05,
"loss": 0.0051,
"step": 15960
},
{
"grad_norm": 0.20629790425300598,
"learning_rate": 4.879308426706707e-05,
"loss": 0.0055,
"step": 15970
},
{
"grad_norm": 0.3343374729156494,
"learning_rate": 4.873798541003084e-05,
"loss": 0.0072,
"step": 15980
},
{
"grad_norm": 0.2559071481227875,
"learning_rate": 4.868288808646136e-05,
"loss": 0.0065,
"step": 15990
},
{
"grad_norm": 0.253345251083374,
"learning_rate": 4.862779236330705e-05,
"loss": 0.0054,
"step": 16000
},
{
"grad_norm": 0.1967095583677292,
"learning_rate": 4.8572698307514395e-05,
"loss": 0.0064,
"step": 16010
},
{
"grad_norm": 0.21190856397151947,
"learning_rate": 4.85176059860279e-05,
"loss": 0.0063,
"step": 16020
},
{
"grad_norm": 0.17401458323001862,
"learning_rate": 4.846251546578989e-05,
"loss": 0.0054,
"step": 16030
},
{
"grad_norm": 0.21087060868740082,
"learning_rate": 4.840742681374058e-05,
"loss": 0.0053,
"step": 16040
},
{
"grad_norm": 0.19508184492588043,
"learning_rate": 4.835234009681787e-05,
"loss": 0.0055,
"step": 16050
},
{
"grad_norm": 0.2123178094625473,
"learning_rate": 4.8297255381957296e-05,
"loss": 0.0057,
"step": 16060
},
{
"grad_norm": 0.21818460524082184,
"learning_rate": 4.824217273609198e-05,
"loss": 0.0054,
"step": 16070
},
{
"grad_norm": 0.23035231232643127,
"learning_rate": 4.8187092226152555e-05,
"loss": 0.0056,
"step": 16080
},
{
"grad_norm": 0.2384929656982422,
"learning_rate": 4.8132013919067016e-05,
"loss": 0.005,
"step": 16090
},
{
"grad_norm": 0.17770709097385406,
"learning_rate": 4.807693788176071e-05,
"loss": 0.0053,
"step": 16100
},
{
"grad_norm": 0.21176333725452423,
"learning_rate": 4.8021864181156214e-05,
"loss": 0.0049,
"step": 16110
},
{
"grad_norm": 0.20433728396892548,
"learning_rate": 4.796679288417326e-05,
"loss": 0.0056,
"step": 16120
},
{
"grad_norm": 0.22551023960113525,
"learning_rate": 4.791172405772866e-05,
"loss": 0.0052,
"step": 16130
},
{
"grad_norm": 0.1886228621006012,
"learning_rate": 4.785665776873625e-05,
"loss": 0.005,
"step": 16140
},
{
"grad_norm": 0.25276118516921997,
"learning_rate": 4.7801594084106763e-05,
"loss": 0.006,
"step": 16150
},
{
"grad_norm": 0.23134861886501312,
"learning_rate": 4.774653307074775e-05,
"loss": 0.0057,
"step": 16160
},
{
"grad_norm": 0.208319291472435,
"learning_rate": 4.769147479556355e-05,
"loss": 0.0056,
"step": 16170
},
{
"grad_norm": 0.21682918071746826,
"learning_rate": 4.763641932545515e-05,
"loss": 0.0058,
"step": 16180
},
{
"grad_norm": 0.21697677671909332,
"learning_rate": 4.758136672732013e-05,
"loss": 0.0054,
"step": 16190
},
{
"grad_norm": 0.20813141763210297,
"learning_rate": 4.752631706805261e-05,
"loss": 0.0054,
"step": 16200
},
{
"grad_norm": 0.2454807311296463,
"learning_rate": 4.7471270414543114e-05,
"loss": 0.0068,
"step": 16210
},
{
"grad_norm": 0.21605922281742096,
"learning_rate": 4.741622683367849e-05,
"loss": 0.0061,
"step": 16220
},
{
"grad_norm": 0.2152635008096695,
"learning_rate": 4.736118639234191e-05,
"loss": 0.0051,
"step": 16230
},
{
"grad_norm": 0.22276683151721954,
"learning_rate": 4.730614915741267e-05,
"loss": 0.0053,
"step": 16240
},
{
"grad_norm": 0.24436958134174347,
"learning_rate": 4.7251115195766234e-05,
"loss": 0.0059,
"step": 16250
},
{
"grad_norm": 0.2095530927181244,
"learning_rate": 4.719608457427404e-05,
"loss": 0.0058,
"step": 16260
},
{
"grad_norm": 0.2204148918390274,
"learning_rate": 4.714105735980347e-05,
"loss": 0.0057,
"step": 16270
},
{
"grad_norm": 0.19364529848098755,
"learning_rate": 4.70860336192178e-05,
"loss": 0.0065,
"step": 16280
},
{
"grad_norm": 0.17767223715782166,
"learning_rate": 4.7031013419376034e-05,
"loss": 0.0064,
"step": 16290
},
{
"grad_norm": 0.18515127897262573,
"learning_rate": 4.6975996827132926e-05,
"loss": 0.0051,
"step": 16300
},
{
"grad_norm": 0.19271183013916016,
"learning_rate": 4.692098390933883e-05,
"loss": 0.0058,
"step": 16310
},
{
"grad_norm": 0.19758860766887665,
"learning_rate": 4.6865974732839615e-05,
"loss": 0.0051,
"step": 16320
},
{
"grad_norm": 0.19299399852752686,
"learning_rate": 4.6810969364476615e-05,
"loss": 0.005,
"step": 16330
},
{
"grad_norm": 0.21201811730861664,
"learning_rate": 4.675596787108653e-05,
"loss": 0.0053,
"step": 16340
},
{
"grad_norm": 0.1895458847284317,
"learning_rate": 4.670097031950139e-05,
"loss": 0.0051,
"step": 16350
},
{
"grad_norm": 0.19045665860176086,
"learning_rate": 4.664597677654838e-05,
"loss": 0.0049,
"step": 16360
},
{
"grad_norm": 0.16687598824501038,
"learning_rate": 4.659098730904986e-05,
"loss": 0.0054,
"step": 16370
},
{
"grad_norm": 0.20688343048095703,
"learning_rate": 4.6536001983823204e-05,
"loss": 0.0055,
"step": 16380
},
{
"grad_norm": 0.26292169094085693,
"learning_rate": 4.648102086768078e-05,
"loss": 0.0059,
"step": 16390
},
{
"grad_norm": 0.19072838127613068,
"learning_rate": 4.6426044027429786e-05,
"loss": 0.0053,
"step": 16400
},
{
"grad_norm": 0.15273305773735046,
"learning_rate": 4.637107152987234e-05,
"loss": 0.005,
"step": 16410
},
{
"grad_norm": 0.1860780268907547,
"learning_rate": 4.631610344180515e-05,
"loss": 0.005,
"step": 16420
},
{
"grad_norm": 0.20414364337921143,
"learning_rate": 4.626113983001965e-05,
"loss": 0.006,
"step": 16430
},
{
"grad_norm": 0.20100267231464386,
"learning_rate": 4.620618076130182e-05,
"loss": 0.0054,
"step": 16440
},
{
"grad_norm": 0.24512773752212524,
"learning_rate": 4.6151226302432074e-05,
"loss": 0.006,
"step": 16450
},
{
"grad_norm": 0.2385198175907135,
"learning_rate": 4.6096276520185304e-05,
"loss": 0.0057,
"step": 16460
},
{
"grad_norm": 0.23382139205932617,
"learning_rate": 4.604133148133066e-05,
"loss": 0.0051,
"step": 16470
},
{
"grad_norm": 0.1980346292257309,
"learning_rate": 4.598639125263155e-05,
"loss": 0.0062,
"step": 16480
},
{
"grad_norm": 0.24339403212070465,
"learning_rate": 4.593145590084553e-05,
"loss": 0.0053,
"step": 16490
},
{
"grad_norm": 0.2034509778022766,
"learning_rate": 4.58765254927242e-05,
"loss": 0.0059,
"step": 16500
},
{
"grad_norm": 0.19285395741462708,
"learning_rate": 4.582160009501323e-05,
"loss": 0.0048,
"step": 16510
},
{
"grad_norm": 0.15153950452804565,
"learning_rate": 4.5766679774452143e-05,
"loss": 0.005,
"step": 16520
},
{
"grad_norm": 0.18080031871795654,
"learning_rate": 4.571176459777431e-05,
"loss": 0.0047,
"step": 16530
},
{
"grad_norm": 0.23050755262374878,
"learning_rate": 4.5656854631706846e-05,
"loss": 0.0044,
"step": 16540
},
{
"grad_norm": 0.1557338535785675,
"learning_rate": 4.560194994297054e-05,
"loss": 0.0048,
"step": 16550
},
{
"grad_norm": 0.15073950588703156,
"learning_rate": 4.554705059827974e-05,
"loss": 0.0043,
"step": 16560
},
{
"grad_norm": 0.17491093277931213,
"learning_rate": 4.5492156664342365e-05,
"loss": 0.0048,
"step": 16570
},
{
"grad_norm": 0.2319764941930771,
"learning_rate": 4.5437268207859695e-05,
"loss": 0.0052,
"step": 16580
},
{
"grad_norm": 0.22784483432769775,
"learning_rate": 4.538238529552641e-05,
"loss": 0.0057,
"step": 16590
},
{
"grad_norm": 0.18912112712860107,
"learning_rate": 4.5327507994030394e-05,
"loss": 0.0054,
"step": 16600
},
{
"grad_norm": 0.1908450424671173,
"learning_rate": 4.527263637005273e-05,
"loss": 0.0062,
"step": 16610
},
{
"grad_norm": 0.22742292284965515,
"learning_rate": 4.521777049026766e-05,
"loss": 0.0057,
"step": 16620
},
{
"grad_norm": 0.263247549533844,
"learning_rate": 4.5162910421342386e-05,
"loss": 0.0058,
"step": 16630
},
{
"grad_norm": 0.19310393929481506,
"learning_rate": 4.510805622993706e-05,
"loss": 0.0049,
"step": 16640
},
{
"grad_norm": 0.20880667865276337,
"learning_rate": 4.505320798270467e-05,
"loss": 0.006,
"step": 16650
},
{
"grad_norm": 0.17734777927398682,
"learning_rate": 4.499836574629105e-05,
"loss": 0.0053,
"step": 16660
},
{
"grad_norm": 0.18983574211597443,
"learning_rate": 4.494352958733465e-05,
"loss": 0.0057,
"step": 16670
},
{
"grad_norm": 0.1658690720796585,
"learning_rate": 4.488869957246663e-05,
"loss": 0.006,
"step": 16680
},
{
"grad_norm": 0.15983198583126068,
"learning_rate": 4.4833875768310575e-05,
"loss": 0.0042,
"step": 16690
},
{
"grad_norm": 0.16793616116046906,
"learning_rate": 4.47790582414826e-05,
"loss": 0.0042,
"step": 16700
},
{
"grad_norm": 0.20175786316394806,
"learning_rate": 4.472424705859114e-05,
"loss": 0.0051,
"step": 16710
},
{
"grad_norm": 0.18174578249454498,
"learning_rate": 4.466944228623701e-05,
"loss": 0.0062,
"step": 16720
},
{
"grad_norm": 0.2416030466556549,
"learning_rate": 4.461464399101312e-05,
"loss": 0.0058,
"step": 16730
},
{
"grad_norm": 0.25312167406082153,
"learning_rate": 4.45598522395046e-05,
"loss": 0.0053,
"step": 16740
},
{
"grad_norm": 0.22538794577121735,
"learning_rate": 4.4505067098288577e-05,
"loss": 0.0057,
"step": 16750
},
{
"grad_norm": 0.23739664256572723,
"learning_rate": 4.445028863393417e-05,
"loss": 0.0057,
"step": 16760
},
{
"grad_norm": 0.15481190383434296,
"learning_rate": 4.439551691300236e-05,
"loss": 0.0048,
"step": 16770
},
{
"grad_norm": 0.21246597170829773,
"learning_rate": 4.434075200204599e-05,
"loss": 0.0055,
"step": 16780
},
{
"grad_norm": 0.1966785341501236,
"learning_rate": 4.428599396760956e-05,
"loss": 0.005,
"step": 16790
},
{
"grad_norm": 0.16649869084358215,
"learning_rate": 4.423124287622926e-05,
"loss": 0.0046,
"step": 16800
},
{
"grad_norm": 0.17013390362262726,
"learning_rate": 4.417649879443282e-05,
"loss": 0.0056,
"step": 16810
},
{
"grad_norm": 0.18564490973949432,
"learning_rate": 4.4121761788739444e-05,
"loss": 0.0058,
"step": 16820
},
{
"grad_norm": 0.23237548768520355,
"learning_rate": 4.4067031925659805e-05,
"loss": 0.0053,
"step": 16830
},
{
"grad_norm": 0.16918113827705383,
"learning_rate": 4.4012309271695816e-05,
"loss": 0.0049,
"step": 16840
},
{
"grad_norm": 0.21602590382099152,
"learning_rate": 4.395759389334067e-05,
"loss": 0.0065,
"step": 16850
},
{
"grad_norm": 0.18481950461864471,
"learning_rate": 4.3902885857078685e-05,
"loss": 0.005,
"step": 16860
},
{
"grad_norm": 0.2104102373123169,
"learning_rate": 4.384818522938531e-05,
"loss": 0.0055,
"step": 16870
},
{
"grad_norm": 0.20828987658023834,
"learning_rate": 4.379349207672695e-05,
"loss": 0.0056,
"step": 16880
},
{
"grad_norm": 0.22771115601062775,
"learning_rate": 4.373880646556098e-05,
"loss": 0.0055,
"step": 16890
},
{
"grad_norm": 0.1506836712360382,
"learning_rate": 4.3684128462335533e-05,
"loss": 0.0055,
"step": 16900
},
{
"grad_norm": 0.1899905949831009,
"learning_rate": 4.362945813348955e-05,
"loss": 0.0048,
"step": 16910
},
{
"grad_norm": 0.2529297471046448,
"learning_rate": 4.357479554545262e-05,
"loss": 0.0054,
"step": 16920
},
{
"grad_norm": 0.19107088446617126,
"learning_rate": 4.352014076464499e-05,
"loss": 0.0049,
"step": 16930
},
{
"grad_norm": 0.20431016385555267,
"learning_rate": 4.346549385747734e-05,
"loss": 0.0057,
"step": 16940
},
{
"grad_norm": 0.17568442225456238,
"learning_rate": 4.34108548903508e-05,
"loss": 0.0062,
"step": 16950
},
{
"grad_norm": 0.18666277825832367,
"learning_rate": 4.335622392965689e-05,
"loss": 0.0046,
"step": 16960
},
{
"grad_norm": 0.20157167315483093,
"learning_rate": 4.330160104177738e-05,
"loss": 0.006,
"step": 16970
},
{
"grad_norm": 0.22235330939292908,
"learning_rate": 4.32469862930842e-05,
"loss": 0.0052,
"step": 16980
},
{
"grad_norm": 0.21450001001358032,
"learning_rate": 4.319237974993946e-05,
"loss": 0.0055,
"step": 16990
},
{
"grad_norm": 0.20912602543830872,
"learning_rate": 4.3137781478695236e-05,
"loss": 0.0046,
"step": 17000
},
{
"grad_norm": 0.18810540437698364,
"learning_rate": 4.308319154569358e-05,
"loss": 0.0047,
"step": 17010
},
{
"grad_norm": 0.17878614366054535,
"learning_rate": 4.302861001726641e-05,
"loss": 0.0052,
"step": 17020
},
{
"grad_norm": 0.17166592180728912,
"learning_rate": 4.2974036959735415e-05,
"loss": 0.0055,
"step": 17030
},
{
"grad_norm": 0.21694909036159515,
"learning_rate": 4.2919472439412034e-05,
"loss": 0.0044,
"step": 17040
},
{
"grad_norm": 0.1790798455476761,
"learning_rate": 4.286491652259729e-05,
"loss": 0.0056,
"step": 17050
},
{
"grad_norm": 0.15755033493041992,
"learning_rate": 4.281036927558178e-05,
"loss": 0.0049,
"step": 17060
},
{
"grad_norm": 0.19428835809230804,
"learning_rate": 4.275583076464552e-05,
"loss": 0.0054,
"step": 17070
},
{
"grad_norm": 0.2802768647670746,
"learning_rate": 4.270130105605794e-05,
"loss": 0.0052,
"step": 17080
},
{
"grad_norm": 0.2341921031475067,
"learning_rate": 4.264678021607782e-05,
"loss": 0.0054,
"step": 17090
},
{
"grad_norm": 0.19393925368785858,
"learning_rate": 4.25922683109531e-05,
"loss": 0.0063,
"step": 17100
},
{
"grad_norm": 0.16245296597480774,
"learning_rate": 4.25377654069209e-05,
"loss": 0.0053,
"step": 17110
},
{
"grad_norm": 0.20205679535865784,
"learning_rate": 4.248327157020737e-05,
"loss": 0.0054,
"step": 17120
},
{
"grad_norm": 0.18528905510902405,
"learning_rate": 4.242878686702763e-05,
"loss": 0.0049,
"step": 17130
},
{
"grad_norm": 0.1788809895515442,
"learning_rate": 4.23743113635858e-05,
"loss": 0.0056,
"step": 17140
},
{
"grad_norm": 0.17531035840511322,
"learning_rate": 4.2319845126074705e-05,
"loss": 0.0054,
"step": 17150
},
{
"grad_norm": 0.20132918655872345,
"learning_rate": 4.226538822067598e-05,
"loss": 0.0062,
"step": 17160
},
{
"grad_norm": 0.23592239618301392,
"learning_rate": 4.2210940713559896e-05,
"loss": 0.0063,
"step": 17170
},
{
"grad_norm": 0.17282961308956146,
"learning_rate": 4.21565026708853e-05,
"loss": 0.0058,
"step": 17180
},
{
"grad_norm": 0.22975854575634003,
"learning_rate": 4.2102074158799535e-05,
"loss": 0.0048,
"step": 17190
},
{
"grad_norm": 0.18393324315547943,
"learning_rate": 4.204765524343841e-05,
"loss": 0.0049,
"step": 17200
},
{
"grad_norm": 0.19748751819133759,
"learning_rate": 4.1993245990926036e-05,
"loss": 0.0052,
"step": 17210
},
{
"grad_norm": 0.15742751955986023,
"learning_rate": 4.193884646737474e-05,
"loss": 0.0049,
"step": 17220
},
{
"grad_norm": 0.20775699615478516,
"learning_rate": 4.1884456738885124e-05,
"loss": 0.0044,
"step": 17230
},
{
"grad_norm": 0.16741704940795898,
"learning_rate": 4.1830076871545795e-05,
"loss": 0.0047,
"step": 17240
},
{
"grad_norm": 0.23173849284648895,
"learning_rate": 4.177570693143346e-05,
"loss": 0.0052,
"step": 17250
},
{
"grad_norm": 0.1653534173965454,
"learning_rate": 4.172134698461271e-05,
"loss": 0.0045,
"step": 17260
},
{
"grad_norm": 0.2438388615846634,
"learning_rate": 4.166699709713599e-05,
"loss": 0.0061,
"step": 17270
},
{
"grad_norm": 0.22849225997924805,
"learning_rate": 4.1612657335043555e-05,
"loss": 0.0056,
"step": 17280
},
{
"grad_norm": 0.22238463163375854,
"learning_rate": 4.155832776436331e-05,
"loss": 0.0044,
"step": 17290
},
{
"grad_norm": 0.2011542022228241,
"learning_rate": 4.1504008451110845e-05,
"loss": 0.0061,
"step": 17300
},
{
"grad_norm": 0.1724104881286621,
"learning_rate": 4.1449699461289225e-05,
"loss": 0.0046,
"step": 17310
},
{
"grad_norm": 0.1807810515165329,
"learning_rate": 4.139540086088901e-05,
"loss": 0.0046,
"step": 17320
},
{
"grad_norm": 0.16764645278453827,
"learning_rate": 4.1341112715888106e-05,
"loss": 0.0047,
"step": 17330
},
{
"grad_norm": 0.17839482426643372,
"learning_rate": 4.128683509225172e-05,
"loss": 0.0044,
"step": 17340
},
{
"grad_norm": 0.20980997383594513,
"learning_rate": 4.123256805593231e-05,
"loss": 0.0039,
"step": 17350
},
{
"grad_norm": 0.18025140464305878,
"learning_rate": 4.117831167286943e-05,
"loss": 0.0043,
"step": 17360
},
{
"grad_norm": 0.24051155149936676,
"learning_rate": 4.1124066008989685e-05,
"loss": 0.0051,
"step": 17370
},
{
"grad_norm": 0.19812968373298645,
"learning_rate": 4.106983113020669e-05,
"loss": 0.0063,
"step": 17380
},
{
"grad_norm": 0.22067487239837646,
"learning_rate": 4.101560710242094e-05,
"loss": 0.0052,
"step": 17390
},
{
"grad_norm": 0.2652187943458557,
"learning_rate": 4.096139399151971e-05,
"loss": 0.0052,
"step": 17400
},
{
"grad_norm": 0.17559142410755157,
"learning_rate": 4.090719186337709e-05,
"loss": 0.0053,
"step": 17410
},
{
"grad_norm": 0.18371941149234772,
"learning_rate": 4.0853000783853746e-05,
"loss": 0.0048,
"step": 17420
},
{
"grad_norm": 0.16808755695819855,
"learning_rate": 4.0798820818796956e-05,
"loss": 0.0056,
"step": 17430
},
{
"grad_norm": 0.23052626848220825,
"learning_rate": 4.074465203404048e-05,
"loss": 0.0049,
"step": 17440
},
{
"grad_norm": 0.20733849704265594,
"learning_rate": 4.0690494495404495e-05,
"loss": 0.0044,
"step": 17450
},
{
"grad_norm": 0.18076413869857788,
"learning_rate": 4.0636348268695535e-05,
"loss": 0.0046,
"step": 17460
},
{
"grad_norm": 0.20382194221019745,
"learning_rate": 4.058221341970637e-05,
"loss": 0.0049,
"step": 17470
},
{
"grad_norm": 0.2146313637495041,
"learning_rate": 4.0528090014215945e-05,
"loss": 0.0046,
"step": 17480
},
{
"grad_norm": 0.21788740158081055,
"learning_rate": 4.047397811798929e-05,
"loss": 0.0051,
"step": 17490
},
{
"grad_norm": 0.195185586810112,
"learning_rate": 4.041987779677745e-05,
"loss": 0.0051,
"step": 17500
},
{
"grad_norm": 0.16076259315013885,
"learning_rate": 4.036578911631745e-05,
"loss": 0.0053,
"step": 17510
},
{
"grad_norm": 0.14859013259410858,
"learning_rate": 4.031171214233211e-05,
"loss": 0.0055,
"step": 17520
},
{
"grad_norm": 0.15592412650585175,
"learning_rate": 4.025764694053008e-05,
"loss": 0.0052,
"step": 17530
},
{
"grad_norm": 0.17401601374149323,
"learning_rate": 4.020359357660566e-05,
"loss": 0.0048,
"step": 17540
},
{
"grad_norm": 0.21960791945457458,
"learning_rate": 4.014955211623875e-05,
"loss": 0.0051,
"step": 17550
},
{
"grad_norm": 0.1800685077905655,
"learning_rate": 4.0095522625094895e-05,
"loss": 0.0047,
"step": 17560
},
{
"grad_norm": 0.16266880929470062,
"learning_rate": 4.004150516882497e-05,
"loss": 0.0042,
"step": 17570
},
{
"grad_norm": 0.1436653435230255,
"learning_rate": 3.99874998130653e-05,
"loss": 0.0048,
"step": 17580
},
{
"grad_norm": 0.16550645232200623,
"learning_rate": 3.9933506623437455e-05,
"loss": 0.0038,
"step": 17590
},
{
"grad_norm": 0.16834920644760132,
"learning_rate": 3.9879525665548276e-05,
"loss": 0.0042,
"step": 17600
},
{
"grad_norm": 0.15376010537147522,
"learning_rate": 3.9825557004989704e-05,
"loss": 0.0052,
"step": 17610
},
{
"grad_norm": 0.18857693672180176,
"learning_rate": 3.977160070733877e-05,
"loss": 0.0049,
"step": 17620
},
{
"grad_norm": 0.19286003708839417,
"learning_rate": 3.971765683815746e-05,
"loss": 0.0042,
"step": 17630
},
{
"grad_norm": 0.18401466310024261,
"learning_rate": 3.966372546299265e-05,
"loss": 0.0049,
"step": 17640
},
{
"grad_norm": 0.19646358489990234,
"learning_rate": 3.960980664737604e-05,
"loss": 0.0048,
"step": 17650
},
{
"grad_norm": 0.13006648421287537,
"learning_rate": 3.955590045682408e-05,
"loss": 0.0043,
"step": 17660
},
{
"grad_norm": 0.17503871023654938,
"learning_rate": 3.950200695683788e-05,
"loss": 0.0049,
"step": 17670
},
{
"grad_norm": 0.15699036419391632,
"learning_rate": 3.944812621290314e-05,
"loss": 0.0046,
"step": 17680
},
{
"grad_norm": 0.18417085707187653,
"learning_rate": 3.939425829049002e-05,
"loss": 0.0053,
"step": 17690
},
{
"grad_norm": 0.17070989310741425,
"learning_rate": 3.9340403255053135e-05,
"loss": 0.0049,
"step": 17700
},
{
"grad_norm": 0.2260528802871704,
"learning_rate": 3.9286561172031407e-05,
"loss": 0.0051,
"step": 17710
},
{
"grad_norm": 0.16860291361808777,
"learning_rate": 3.923273210684809e-05,
"loss": 0.005,
"step": 17720
},
{
"grad_norm": 0.17992687225341797,
"learning_rate": 3.9178916124910555e-05,
"loss": 0.0051,
"step": 17730
},
{
"grad_norm": 0.17968039214611053,
"learning_rate": 3.9125113291610274e-05,
"loss": 0.0047,
"step": 17740
},
{
"grad_norm": 0.19374966621398926,
"learning_rate": 3.907132367232278e-05,
"loss": 0.0045,
"step": 17750
},
{
"grad_norm": 0.23144283890724182,
"learning_rate": 3.901754733240752e-05,
"loss": 0.0056,
"step": 17760
},
{
"grad_norm": 0.18469102680683136,
"learning_rate": 3.896378433720785e-05,
"loss": 0.0049,
"step": 17770
},
{
"grad_norm": 0.16908034682273865,
"learning_rate": 3.891003475205086e-05,
"loss": 0.0047,
"step": 17780
},
{
"grad_norm": 0.18853403627872467,
"learning_rate": 3.885629864224736e-05,
"loss": 0.0047,
"step": 17790
},
{
"grad_norm": 0.1695813238620758,
"learning_rate": 3.880257607309178e-05,
"loss": 0.0047,
"step": 17800
},
{
"grad_norm": 0.19973872601985931,
"learning_rate": 3.874886710986213e-05,
"loss": 0.0047,
"step": 17810
},
{
"grad_norm": 0.19555094838142395,
"learning_rate": 3.8695171817819834e-05,
"loss": 0.0044,
"step": 17820
},
{
"grad_norm": 0.14206552505493164,
"learning_rate": 3.8641490262209774e-05,
"loss": 0.0046,
"step": 17830
},
{
"grad_norm": 0.26193562150001526,
"learning_rate": 3.8587822508260084e-05,
"loss": 0.0048,
"step": 17840
},
{
"grad_norm": 0.2157917469739914,
"learning_rate": 3.853416862118214e-05,
"loss": 0.005,
"step": 17850
},
{
"grad_norm": 0.17988722026348114,
"learning_rate": 3.848052866617049e-05,
"loss": 0.0059,
"step": 17860
},
{
"grad_norm": 0.19390439987182617,
"learning_rate": 3.84269027084027e-05,
"loss": 0.0046,
"step": 17870
},
{
"grad_norm": 0.22664739191532135,
"learning_rate": 3.83732908130394e-05,
"loss": 0.0044,
"step": 17880
},
{
"grad_norm": 0.17897790670394897,
"learning_rate": 3.831969304522409e-05,
"loss": 0.0046,
"step": 17890
},
{
"grad_norm": 0.18609483540058136,
"learning_rate": 3.826610947008312e-05,
"loss": 0.0046,
"step": 17900
},
{
"grad_norm": 0.20713898539543152,
"learning_rate": 3.821254015272559e-05,
"loss": 0.0055,
"step": 17910
},
{
"grad_norm": 0.1708584725856781,
"learning_rate": 3.8158985158243214e-05,
"loss": 0.0047,
"step": 17920
},
{
"grad_norm": 0.12369570136070251,
"learning_rate": 3.810544455171044e-05,
"loss": 0.0048,
"step": 17930
},
{
"grad_norm": 0.15960708260536194,
"learning_rate": 3.8051918398184114e-05,
"loss": 0.0046,
"step": 17940
},
{
"grad_norm": 0.17765159904956818,
"learning_rate": 3.799840676270356e-05,
"loss": 0.0043,
"step": 17950
},
{
"grad_norm": 0.20626385509967804,
"learning_rate": 3.794490971029048e-05,
"loss": 0.0056,
"step": 17960
},
{
"grad_norm": 0.1839437037706375,
"learning_rate": 3.789142730594881e-05,
"loss": 0.005,
"step": 17970
},
{
"grad_norm": 0.15623614192008972,
"learning_rate": 3.783795961466471e-05,
"loss": 0.0045,
"step": 17980
},
{
"grad_norm": 0.15824554860591888,
"learning_rate": 3.778450670140651e-05,
"loss": 0.0056,
"step": 17990
},
{
"grad_norm": 0.2152419239282608,
"learning_rate": 3.773106863112451e-05,
"loss": 0.0052,
"step": 18000
},
{
"grad_norm": 0.17177115380764008,
"learning_rate": 3.7677645468751e-05,
"loss": 0.0041,
"step": 18010
},
{
"grad_norm": 0.1628618687391281,
"learning_rate": 3.762423727920018e-05,
"loss": 0.0043,
"step": 18020
},
{
"grad_norm": 0.1526751071214676,
"learning_rate": 3.7570844127367995e-05,
"loss": 0.0041,
"step": 18030
},
{
"grad_norm": 0.1661258041858673,
"learning_rate": 3.751746607813221e-05,
"loss": 0.0051,
"step": 18040
},
{
"grad_norm": 0.13956239819526672,
"learning_rate": 3.746410319635217e-05,
"loss": 0.0047,
"step": 18050
},
{
"grad_norm": 0.2003924399614334,
"learning_rate": 3.7410755546868806e-05,
"loss": 0.0043,
"step": 18060
},
{
"grad_norm": 0.14374731481075287,
"learning_rate": 3.735742319450454e-05,
"loss": 0.0042,
"step": 18070
},
{
"grad_norm": 0.1980191320180893,
"learning_rate": 3.7304106204063184e-05,
"loss": 0.0043,
"step": 18080
},
{
"grad_norm": 0.1401374191045761,
"learning_rate": 3.725080464032996e-05,
"loss": 0.0047,
"step": 18090
},
{
"grad_norm": 0.17544469237327576,
"learning_rate": 3.719751856807125e-05,
"loss": 0.0046,
"step": 18100
},
{
"grad_norm": 0.18286515772342682,
"learning_rate": 3.71442480520347e-05,
"loss": 0.0044,
"step": 18110
},
{
"grad_norm": 0.22680023312568665,
"learning_rate": 3.709099315694897e-05,
"loss": 0.0051,
"step": 18120
},
{
"grad_norm": 0.24222224950790405,
"learning_rate": 3.703775394752378e-05,
"loss": 0.0045,
"step": 18130
},
{
"grad_norm": 0.1783822625875473,
"learning_rate": 3.698453048844983e-05,
"loss": 0.0056,
"step": 18140
},
{
"grad_norm": 0.21074515581130981,
"learning_rate": 3.693132284439861e-05,
"loss": 0.0049,
"step": 18150
},
{
"grad_norm": 0.17618469893932343,
"learning_rate": 3.6878131080022414e-05,
"loss": 0.0043,
"step": 18160
},
{
"grad_norm": 0.21250373125076294,
"learning_rate": 3.682495525995429e-05,
"loss": 0.0047,
"step": 18170
},
{
"grad_norm": 0.11130572855472565,
"learning_rate": 3.6771795448807846e-05,
"loss": 0.0043,
"step": 18180
},
{
"grad_norm": 0.22692075371742249,
"learning_rate": 3.671865171117724e-05,
"loss": 0.0056,
"step": 18190
},
{
"grad_norm": 0.19033771753311157,
"learning_rate": 3.666552411163718e-05,
"loss": 0.0058,
"step": 18200
},
{
"grad_norm": 0.18277636170387268,
"learning_rate": 3.661241271474269e-05,
"loss": 0.0045,
"step": 18210
},
{
"grad_norm": 0.18331150710582733,
"learning_rate": 3.6559317585029116e-05,
"loss": 0.0045,
"step": 18220
},
{
"grad_norm": 0.25718456506729126,
"learning_rate": 3.650623878701204e-05,
"loss": 0.0048,
"step": 18230
},
{
"grad_norm": 0.21202269196510315,
"learning_rate": 3.645317638518721e-05,
"loss": 0.0041,
"step": 18240
},
{
"grad_norm": 0.24504457414150238,
"learning_rate": 3.640013044403046e-05,
"loss": 0.0056,
"step": 18250
},
{
"grad_norm": 0.16236044466495514,
"learning_rate": 3.634710102799761e-05,
"loss": 0.0042,
"step": 18260
},
{
"grad_norm": 0.14248909056186676,
"learning_rate": 3.6294088201524394e-05,
"loss": 0.0048,
"step": 18270
},
{
"grad_norm": 0.15459634363651276,
"learning_rate": 3.62410920290264e-05,
"loss": 0.0043,
"step": 18280
},
{
"grad_norm": 0.19601555168628693,
"learning_rate": 3.6188112574898954e-05,
"loss": 0.0044,
"step": 18290
},
{
"grad_norm": 0.20809586346149445,
"learning_rate": 3.6135149903517115e-05,
"loss": 0.0046,
"step": 18300
},
{
"grad_norm": 0.21614643931388855,
"learning_rate": 3.608220407923552e-05,
"loss": 0.0045,
"step": 18310
},
{
"grad_norm": 0.18028244376182556,
"learning_rate": 3.602927516638833e-05,
"loss": 0.0062,
"step": 18320
},
{
"grad_norm": 0.1682916134595871,
"learning_rate": 3.5976363229289165e-05,
"loss": 0.0045,
"step": 18330
},
{
"grad_norm": 0.2125607281923294,
"learning_rate": 3.5923468332231e-05,
"loss": 0.0043,
"step": 18340
},
{
"grad_norm": 0.16382136940956116,
"learning_rate": 3.587059053948616e-05,
"loss": 0.0043,
"step": 18350
},
{
"grad_norm": 0.13429878652095795,
"learning_rate": 3.5817729915306134e-05,
"loss": 0.0046,
"step": 18360
},
{
"grad_norm": 0.14918677508831024,
"learning_rate": 3.5764886523921566e-05,
"loss": 0.0043,
"step": 18370
},
{
"grad_norm": 0.12077635526657104,
"learning_rate": 3.571206042954214e-05,
"loss": 0.0043,
"step": 18380
},
{
"grad_norm": 0.175746351480484,
"learning_rate": 3.5659251696356566e-05,
"loss": 0.0048,
"step": 18390
},
{
"grad_norm": 0.15519894659519196,
"learning_rate": 3.56064603885324e-05,
"loss": 0.0043,
"step": 18400
},
{
"grad_norm": 0.1530705988407135,
"learning_rate": 3.5553686570216115e-05,
"loss": 0.0053,
"step": 18410
},
{
"grad_norm": 0.19397637248039246,
"learning_rate": 3.550093030553284e-05,
"loss": 0.0049,
"step": 18420
},
{
"grad_norm": 0.14404350519180298,
"learning_rate": 3.544819165858642e-05,
"loss": 0.0048,
"step": 18430
},
{
"grad_norm": 0.13017989695072174,
"learning_rate": 3.539547069345926e-05,
"loss": 0.0046,
"step": 18440
},
{
"grad_norm": 0.1650015264749527,
"learning_rate": 3.5342767474212344e-05,
"loss": 0.0053,
"step": 18450
},
{
"grad_norm": 0.2442454695701599,
"learning_rate": 3.529008206488502e-05,
"loss": 0.0045,
"step": 18460
},
{
"grad_norm": 0.17530974745750427,
"learning_rate": 3.5237414529495055e-05,
"loss": 0.0049,
"step": 18470
},
{
"grad_norm": 0.17214961349964142,
"learning_rate": 3.5184764932038454e-05,
"loss": 0.0044,
"step": 18480
},
{
"grad_norm": 0.2167811393737793,
"learning_rate": 3.513213333648945e-05,
"loss": 0.0043,
"step": 18490
},
{
"grad_norm": 0.18678313493728638,
"learning_rate": 3.5079519806800374e-05,
"loss": 0.0043,
"step": 18500
},
{
"grad_norm": 0.1832263171672821,
"learning_rate": 3.502692440690165e-05,
"loss": 0.0041,
"step": 18510
},
{
"grad_norm": 0.17450891435146332,
"learning_rate": 3.497434720070165e-05,
"loss": 0.0051,
"step": 18520
},
{
"grad_norm": 0.1663646697998047,
"learning_rate": 3.4921788252086616e-05,
"loss": 0.0047,
"step": 18530
},
{
"grad_norm": 0.14724406599998474,
"learning_rate": 3.486924762492065e-05,
"loss": 0.0046,
"step": 18540
},
{
"grad_norm": 0.15757642686367035,
"learning_rate": 3.4816725383045534e-05,
"loss": 0.004,
"step": 18550
},
{
"grad_norm": 0.13360770046710968,
"learning_rate": 3.476422159028079e-05,
"loss": 0.0044,
"step": 18560
},
{
"grad_norm": 0.21180513501167297,
"learning_rate": 3.471173631042345e-05,
"loss": 0.0046,
"step": 18570
},
{
"grad_norm": 0.19546401500701904,
"learning_rate": 3.465926960724808e-05,
"loss": 0.0042,
"step": 18580
},
{
"grad_norm": 0.22103498876094818,
"learning_rate": 3.460682154450666e-05,
"loss": 0.0044,
"step": 18590
},
{
"grad_norm": 0.19766786694526672,
"learning_rate": 3.4554392185928564e-05,
"loss": 0.0062,
"step": 18600
},
{
"grad_norm": 0.1670973300933838,
"learning_rate": 3.450198159522037e-05,
"loss": 0.0042,
"step": 18610
},
{
"grad_norm": 0.20127828419208527,
"learning_rate": 3.444958983606592e-05,
"loss": 0.0044,
"step": 18620
},
{
"grad_norm": 0.22341984510421753,
"learning_rate": 3.439721697212612e-05,
"loss": 0.0048,
"step": 18630
},
{
"grad_norm": 0.199917271733284,
"learning_rate": 3.4344863067038954e-05,
"loss": 0.0052,
"step": 18640
},
{
"grad_norm": 0.18930475413799286,
"learning_rate": 3.429252818441935e-05,
"loss": 0.0056,
"step": 18650
},
{
"grad_norm": 0.23209547996520996,
"learning_rate": 3.4240212387859094e-05,
"loss": 0.0044,
"step": 18660
},
{
"grad_norm": 0.16475500166416168,
"learning_rate": 3.418791574092686e-05,
"loss": 0.0045,
"step": 18670
},
{
"grad_norm": 0.15590603649616241,
"learning_rate": 3.413563830716796e-05,
"loss": 0.0041,
"step": 18680
},
{
"grad_norm": 0.13032716512680054,
"learning_rate": 3.408338015010445e-05,
"loss": 0.0043,
"step": 18690
},
{
"grad_norm": 0.20032908022403717,
"learning_rate": 3.40311413332349e-05,
"loss": 0.0047,
"step": 18700
},
{
"grad_norm": 0.24781663715839386,
"learning_rate": 3.3978921920034365e-05,
"loss": 0.0047,
"step": 18710
},
{
"grad_norm": 0.20638228952884674,
"learning_rate": 3.392672197395441e-05,
"loss": 0.0041,
"step": 18720
},
{
"grad_norm": 0.19446144998073578,
"learning_rate": 3.387454155842287e-05,
"loss": 0.0052,
"step": 18730
},
{
"grad_norm": 0.20245669782161713,
"learning_rate": 3.382238073684386e-05,
"loss": 0.004,
"step": 18740
},
{
"grad_norm": 0.2096998691558838,
"learning_rate": 3.377023957259771e-05,
"loss": 0.0044,
"step": 18750
},
{
"grad_norm": 0.18447357416152954,
"learning_rate": 3.3718118129040835e-05,
"loss": 0.0061,
"step": 18760
},
{
"grad_norm": 0.1943117082118988,
"learning_rate": 3.3666016469505724e-05,
"loss": 0.0047,
"step": 18770
},
{
"grad_norm": 0.2152513861656189,
"learning_rate": 3.361393465730079e-05,
"loss": 0.0042,
"step": 18780
},
{
"grad_norm": 0.2068987339735031,
"learning_rate": 3.3561872755710366e-05,
"loss": 0.0046,
"step": 18790
},
{
"grad_norm": 0.1730862855911255,
"learning_rate": 3.350983082799456e-05,
"loss": 0.0059,
"step": 18800
},
{
"grad_norm": 0.14086872339248657,
"learning_rate": 3.34578089373892e-05,
"loss": 0.004,
"step": 18810
},
{
"grad_norm": 0.16468176245689392,
"learning_rate": 3.340580714710581e-05,
"loss": 0.0045,
"step": 18820
},
{
"grad_norm": 0.19593720138072968,
"learning_rate": 3.3353825520331466e-05,
"loss": 0.0041,
"step": 18830
},
{
"grad_norm": 0.19939513504505157,
"learning_rate": 3.330186412022876e-05,
"loss": 0.004,
"step": 18840
},
{
"grad_norm": 0.19535371661186218,
"learning_rate": 3.324992300993568e-05,
"loss": 0.0044,
"step": 18850
},
{
"grad_norm": 0.18535414338111877,
"learning_rate": 3.319800225256556e-05,
"loss": 0.0047,
"step": 18860
},
{
"grad_norm": 0.24224001169204712,
"learning_rate": 3.314610191120702e-05,
"loss": 0.0048,
"step": 18870
},
{
"grad_norm": 0.1814900040626526,
"learning_rate": 3.30942220489239e-05,
"loss": 0.004,
"step": 18880
},
{
"grad_norm": 0.1915854662656784,
"learning_rate": 3.3042362728755086e-05,
"loss": 0.0044,
"step": 18890
},
{
"grad_norm": 0.2150082290172577,
"learning_rate": 3.299052401371456e-05,
"loss": 0.0042,
"step": 18900
},
{
"grad_norm": 0.18461056053638458,
"learning_rate": 3.293870596679125e-05,
"loss": 0.0042,
"step": 18910
},
{
"grad_norm": 0.21303988993167877,
"learning_rate": 3.288690865094895e-05,
"loss": 0.0051,
"step": 18920
},
{
"grad_norm": 0.23073503375053406,
"learning_rate": 3.283513212912632e-05,
"loss": 0.0047,
"step": 18930
},
{
"grad_norm": 0.225327730178833,
"learning_rate": 3.2783376464236684e-05,
"loss": 0.0052,
"step": 18940
},
{
"grad_norm": 0.16180652379989624,
"learning_rate": 3.273164171916806e-05,
"loss": 0.0038,
"step": 18950
},
{
"grad_norm": 0.1904311627149582,
"learning_rate": 3.267992795678306e-05,
"loss": 0.0042,
"step": 18960
},
{
"grad_norm": 0.19278259575366974,
"learning_rate": 3.2628235239918744e-05,
"loss": 0.0046,
"step": 18970
},
{
"grad_norm": 0.2188052237033844,
"learning_rate": 3.2576563631386695e-05,
"loss": 0.0043,
"step": 18980
},
{
"grad_norm": 0.15587684512138367,
"learning_rate": 3.252491319397275e-05,
"loss": 0.0047,
"step": 18990
},
{
"grad_norm": 0.15427587926387787,
"learning_rate": 3.247328399043706e-05,
"loss": 0.0037,
"step": 19000
},
{
"grad_norm": 0.16822616755962372,
"learning_rate": 3.242167608351399e-05,
"loss": 0.0037,
"step": 19010
},
{
"grad_norm": 0.16888026893138885,
"learning_rate": 3.2370089535911986e-05,
"loss": 0.004,
"step": 19020
},
{
"grad_norm": 0.16207507252693176,
"learning_rate": 3.23185244103136e-05,
"loss": 0.0041,
"step": 19030
},
{
"grad_norm": 0.15801367163658142,
"learning_rate": 3.22669807693753e-05,
"loss": 0.0044,
"step": 19040
},
{
"grad_norm": 0.18546448647975922,
"learning_rate": 3.2215458675727495e-05,
"loss": 0.0039,
"step": 19050
},
{
"grad_norm": 0.1666295826435089,
"learning_rate": 3.216395819197438e-05,
"loss": 0.0039,
"step": 19060
},
{
"grad_norm": 0.1560066044330597,
"learning_rate": 3.211247938069387e-05,
"loss": 0.0048,
"step": 19070
},
{
"grad_norm": 0.26684293150901794,
"learning_rate": 3.206102230443759e-05,
"loss": 0.0049,
"step": 19080
},
{
"grad_norm": 0.21110324561595917,
"learning_rate": 3.2009587025730764e-05,
"loss": 0.0044,
"step": 19090
},
{
"grad_norm": 0.17278921604156494,
"learning_rate": 3.195817360707207e-05,
"loss": 0.0047,
"step": 19100
},
{
"grad_norm": 0.22686052322387695,
"learning_rate": 3.19067821109337e-05,
"loss": 0.0043,
"step": 19110
},
{
"grad_norm": 0.11207029223442078,
"learning_rate": 3.185541259976114e-05,
"loss": 0.0037,
"step": 19120
},
{
"grad_norm": 0.1977505087852478,
"learning_rate": 3.180406513597316e-05,
"loss": 0.0037,
"step": 19130
},
{
"grad_norm": 0.16978426277637482,
"learning_rate": 3.1752739781961835e-05,
"loss": 0.0038,
"step": 19140
},
{
"grad_norm": 0.19164127111434937,
"learning_rate": 3.170143660009228e-05,
"loss": 0.0046,
"step": 19150
},
{
"grad_norm": 0.20328111946582794,
"learning_rate": 3.16501556527027e-05,
"loss": 0.0046,
"step": 19160
},
{
"grad_norm": 0.19501109421253204,
"learning_rate": 3.1598897002104265e-05,
"loss": 0.0043,
"step": 19170
},
{
"grad_norm": 0.1605186015367508,
"learning_rate": 3.154766071058108e-05,
"loss": 0.004,
"step": 19180
},
{
"grad_norm": 0.17385664582252502,
"learning_rate": 3.149644684039008e-05,
"loss": 0.0035,
"step": 19190
},
{
"grad_norm": 0.20215871930122375,
"learning_rate": 3.144525545376095e-05,
"loss": 0.0043,
"step": 19200
},
{
"grad_norm": 0.15824094414710999,
"learning_rate": 3.139408661289603e-05,
"loss": 0.0039,
"step": 19210
},
{
"grad_norm": 0.15069256722927094,
"learning_rate": 3.134294037997032e-05,
"loss": 0.0044,
"step": 19220
},
{
"grad_norm": 0.13956040143966675,
"learning_rate": 3.129181681713127e-05,
"loss": 0.005,
"step": 19230
},
{
"grad_norm": 0.17186318337917328,
"learning_rate": 3.1240715986498855e-05,
"loss": 0.004,
"step": 19240
},
{
"grad_norm": 0.15590088069438934,
"learning_rate": 3.1189637950165394e-05,
"loss": 0.0042,
"step": 19250
},
{
"grad_norm": 0.18485751748085022,
"learning_rate": 3.1138582770195544e-05,
"loss": 0.0043,
"step": 19260
},
{
"grad_norm": 0.1822366565465927,
"learning_rate": 3.108755050862615e-05,
"loss": 0.0043,
"step": 19270
},
{
"grad_norm": 0.18143655359745026,
"learning_rate": 3.1036541227466204e-05,
"loss": 0.0043,
"step": 19280
},
{
"grad_norm": 0.14490589499473572,
"learning_rate": 3.0985554988696786e-05,
"loss": 0.0043,
"step": 19290
},
{
"grad_norm": 0.21443529427051544,
"learning_rate": 3.093459185427102e-05,
"loss": 0.0045,
"step": 19300
},
{
"grad_norm": 0.1526547223329544,
"learning_rate": 3.0883651886113905e-05,
"loss": 0.0044,
"step": 19310
},
{
"grad_norm": 0.22156177461147308,
"learning_rate": 3.0832735146122294e-05,
"loss": 0.0037,
"step": 19320
},
{
"grad_norm": 0.16703645884990692,
"learning_rate": 3.078184169616485e-05,
"loss": 0.0042,
"step": 19330
},
{
"grad_norm": 0.20904138684272766,
"learning_rate": 3.073097159808187e-05,
"loss": 0.0044,
"step": 19340
},
{
"grad_norm": 0.17865318059921265,
"learning_rate": 3.068012491368537e-05,
"loss": 0.0043,
"step": 19350
},
{
"grad_norm": 0.20035971701145172,
"learning_rate": 3.062930170475885e-05,
"loss": 0.0035,
"step": 19360
},
{
"grad_norm": 0.18480822443962097,
"learning_rate": 3.057850203305729e-05,
"loss": 0.0046,
"step": 19370
},
{
"grad_norm": 0.21740855276584625,
"learning_rate": 3.052772596030708e-05,
"loss": 0.0048,
"step": 19380
},
{
"grad_norm": 0.15248532593250275,
"learning_rate": 3.0476973548205943e-05,
"loss": 0.0037,
"step": 19390
},
{
"grad_norm": 0.16580456495285034,
"learning_rate": 3.0426244858422847e-05,
"loss": 0.0041,
"step": 19400
},
{
"grad_norm": 0.1699233055114746,
"learning_rate": 3.0375539952597943e-05,
"loss": 0.0044,
"step": 19410
},
{
"grad_norm": 0.17569361627101898,
"learning_rate": 3.0324858892342468e-05,
"loss": 0.0047,
"step": 19420
},
{
"grad_norm": 0.14564082026481628,
"learning_rate": 3.0274201739238672e-05,
"loss": 0.0043,
"step": 19430
},
{
"grad_norm": 0.17489130795001984,
"learning_rate": 3.0223568554839786e-05,
"loss": 0.0056,
"step": 19440
},
{
"grad_norm": 0.15464773774147034,
"learning_rate": 3.0172959400669886e-05,
"loss": 0.0043,
"step": 19450
},
{
"grad_norm": 0.18760403990745544,
"learning_rate": 3.0122374338223902e-05,
"loss": 0.0041,
"step": 19460
},
{
"grad_norm": 0.16187039017677307,
"learning_rate": 3.0071813428967427e-05,
"loss": 0.0043,
"step": 19470
},
{
"grad_norm": 0.21139109134674072,
"learning_rate": 3.0021276734336746e-05,
"loss": 0.0037,
"step": 19480
},
{
"grad_norm": 0.1583389788866043,
"learning_rate": 2.997076431573871e-05,
"loss": 0.0032,
"step": 19490
},
{
"grad_norm": 0.15640658140182495,
"learning_rate": 2.9920276234550636e-05,
"loss": 0.0051,
"step": 19500
},
{
"grad_norm": 0.2030418962240219,
"learning_rate": 2.986981255212035e-05,
"loss": 0.0039,
"step": 19510
},
{
"grad_norm": 0.2019844353199005,
"learning_rate": 2.9819373329765977e-05,
"loss": 0.0041,
"step": 19520
},
{
"grad_norm": 0.2277233600616455,
"learning_rate": 2.9768958628775902e-05,
"loss": 0.0039,
"step": 19530
},
{
"grad_norm": 0.20449942350387573,
"learning_rate": 2.9718568510408762e-05,
"loss": 0.0039,
"step": 19540
},
{
"grad_norm": 0.2035658210515976,
"learning_rate": 2.9668203035893272e-05,
"loss": 0.0035,
"step": 19550
},
{
"grad_norm": 0.1849784255027771,
"learning_rate": 2.9617862266428287e-05,
"loss": 0.0038,
"step": 19560
},
{
"grad_norm": 0.12337016314268112,
"learning_rate": 2.9567546263182556e-05,
"loss": 0.0037,
"step": 19570
},
{
"grad_norm": 0.15070250630378723,
"learning_rate": 2.951725508729476e-05,
"loss": 0.0042,
"step": 19580
},
{
"grad_norm": 0.1782853901386261,
"learning_rate": 2.946698879987344e-05,
"loss": 0.0043,
"step": 19590
},
{
"grad_norm": 0.1719118356704712,
"learning_rate": 2.9416747461996853e-05,
"loss": 0.0036,
"step": 19600
},
{
"grad_norm": 0.15068860352039337,
"learning_rate": 2.9366531134712973e-05,
"loss": 0.0052,
"step": 19610
},
{
"grad_norm": 0.14642858505249023,
"learning_rate": 2.9316339879039367e-05,
"loss": 0.0038,
"step": 19620
},
{
"grad_norm": 0.14940200746059418,
"learning_rate": 2.9266173755963167e-05,
"loss": 0.0043,
"step": 19630
},
{
"grad_norm": 0.13630560040473938,
"learning_rate": 2.9216032826440926e-05,
"loss": 0.004,
"step": 19640
},
{
"grad_norm": 0.1434028595685959,
"learning_rate": 2.9165917151398592e-05,
"loss": 0.0035,
"step": 19650
},
{
"grad_norm": 0.17234669625759125,
"learning_rate": 2.9115826791731428e-05,
"loss": 0.006,
"step": 19660
},
{
"grad_norm": 0.24688218533992767,
"learning_rate": 2.906576180830398e-05,
"loss": 0.0046,
"step": 19670
},
{
"grad_norm": 0.20949985086917877,
"learning_rate": 2.9015722261949917e-05,
"loss": 0.0051,
"step": 19680
},
{
"grad_norm": 0.1637730896472931,
"learning_rate": 2.8965708213471986e-05,
"loss": 0.0048,
"step": 19690
},
{
"grad_norm": 0.12858404219150543,
"learning_rate": 2.8915719723641975e-05,
"loss": 0.004,
"step": 19700
},
{
"grad_norm": 0.1770082265138626,
"learning_rate": 2.8865756853200604e-05,
"loss": 0.0042,
"step": 19710
},
{
"grad_norm": 0.2406347393989563,
"learning_rate": 2.8815819662857503e-05,
"loss": 0.0036,
"step": 19720
},
{
"grad_norm": 0.22423067688941956,
"learning_rate": 2.876590821329105e-05,
"loss": 0.0044,
"step": 19730
},
{
"grad_norm": 0.16921508312225342,
"learning_rate": 2.871602256514836e-05,
"loss": 0.0035,
"step": 19740
},
{
"grad_norm": 0.14694929122924805,
"learning_rate": 2.8666162779045203e-05,
"loss": 0.0032,
"step": 19750
},
{
"grad_norm": 0.12985700368881226,
"learning_rate": 2.8616328915565904e-05,
"loss": 0.0034,
"step": 19760
},
{
"grad_norm": 0.15689653158187866,
"learning_rate": 2.856652103526334e-05,
"loss": 0.0043,
"step": 19770
},
{
"grad_norm": 0.16950105130672455,
"learning_rate": 2.8516739198658755e-05,
"loss": 0.0041,
"step": 19780
},
{
"grad_norm": 0.22178786993026733,
"learning_rate": 2.846698346624177e-05,
"loss": 0.005,
"step": 19790
},
{
"grad_norm": 0.17227879166603088,
"learning_rate": 2.841725389847032e-05,
"loss": 0.0037,
"step": 19800
},
{
"grad_norm": 0.19812871515750885,
"learning_rate": 2.8367550555770506e-05,
"loss": 0.0047,
"step": 19810
},
{
"grad_norm": 0.14762230217456818,
"learning_rate": 2.831787349853655e-05,
"loss": 0.0035,
"step": 19820
},
{
"grad_norm": 0.16306933760643005,
"learning_rate": 2.8268222787130806e-05,
"loss": 0.0042,
"step": 19830
},
{
"grad_norm": 0.13166029751300812,
"learning_rate": 2.821859848188355e-05,
"loss": 0.0043,
"step": 19840
},
{
"grad_norm": 0.2033902257680893,
"learning_rate": 2.816900064309299e-05,
"loss": 0.0053,
"step": 19850
},
{
"grad_norm": 0.15772250294685364,
"learning_rate": 2.811942933102517e-05,
"loss": 0.0038,
"step": 19860
},
{
"grad_norm": 0.1778387576341629,
"learning_rate": 2.806988460591391e-05,
"loss": 0.004,
"step": 19870
},
{
"grad_norm": 0.14445197582244873,
"learning_rate": 2.802036652796074e-05,
"loss": 0.0035,
"step": 19880
},
{
"grad_norm": 0.15926769375801086,
"learning_rate": 2.797087515733478e-05,
"loss": 0.0042,
"step": 19890
},
{
"grad_norm": 0.1877395510673523,
"learning_rate": 2.7921410554172723e-05,
"loss": 0.0048,
"step": 19900
},
{
"grad_norm": 0.16779497265815735,
"learning_rate": 2.787197277857871e-05,
"loss": 0.0055,
"step": 19910
},
{
"grad_norm": 0.23836743831634521,
"learning_rate": 2.782256189062429e-05,
"loss": 0.0041,
"step": 19920
},
{
"grad_norm": 0.18422548472881317,
"learning_rate": 2.777317795034839e-05,
"loss": 0.0037,
"step": 19930
},
{
"grad_norm": 0.1820766031742096,
"learning_rate": 2.7723821017757112e-05,
"loss": 0.0039,
"step": 19940
},
{
"grad_norm": 0.2275509387254715,
"learning_rate": 2.7674491152823822e-05,
"loss": 0.0039,
"step": 19950
},
{
"grad_norm": 0.16690683364868164,
"learning_rate": 2.7625188415488944e-05,
"loss": 0.0036,
"step": 19960
},
{
"grad_norm": 0.17074181139469147,
"learning_rate": 2.7575912865659924e-05,
"loss": 0.0033,
"step": 19970
},
{
"grad_norm": 0.13822443783283234,
"learning_rate": 2.7526664563211245e-05,
"loss": 0.0035,
"step": 19980
},
{
"grad_norm": 0.21053001284599304,
"learning_rate": 2.7477443567984224e-05,
"loss": 0.0044,
"step": 19990
},
{
"grad_norm": 0.20478609204292297,
"learning_rate": 2.7428249939787e-05,
"loss": 0.0046,
"step": 20000
},
{
"grad_norm": 0.2003803253173828,
"learning_rate": 2.7379083738394484e-05,
"loss": 0.0039,
"step": 20010
},
{
"grad_norm": 0.1926230639219284,
"learning_rate": 2.732994502354823e-05,
"loss": 0.0042,
"step": 20020
},
{
"grad_norm": 0.17239689826965332,
"learning_rate": 2.72808338549564e-05,
"loss": 0.0042,
"step": 20030
},
{
"grad_norm": 0.1945382058620453,
"learning_rate": 2.723175029229374e-05,
"loss": 0.0056,
"step": 20040
},
{
"grad_norm": 0.29399025440216064,
"learning_rate": 2.718269439520138e-05,
"loss": 0.0045,
"step": 20050
},
{
"grad_norm": 0.1537390798330307,
"learning_rate": 2.7133666223286858e-05,
"loss": 0.004,
"step": 20060
},
{
"grad_norm": 0.16402965784072876,
"learning_rate": 2.7084665836124006e-05,
"loss": 0.0039,
"step": 20070
},
{
"grad_norm": 0.1838580220937729,
"learning_rate": 2.703569329325296e-05,
"loss": 0.0046,
"step": 20080
},
{
"grad_norm": 0.1634926199913025,
"learning_rate": 2.698674865417994e-05,
"loss": 0.0038,
"step": 20090
},
{
"grad_norm": 0.1596059948205948,
"learning_rate": 2.6937831978377325e-05,
"loss": 0.004,
"step": 20100
},
{
"grad_norm": 0.17000797390937805,
"learning_rate": 2.6888943325283485e-05,
"loss": 0.0037,
"step": 20110
},
{
"grad_norm": 0.1683209240436554,
"learning_rate": 2.6840082754302733e-05,
"loss": 0.0044,
"step": 20120
},
{
"grad_norm": 0.14748826622962952,
"learning_rate": 2.679125032480525e-05,
"loss": 0.0046,
"step": 20130
},
{
"grad_norm": 0.20720162987709045,
"learning_rate": 2.674244609612708e-05,
"loss": 0.0034,
"step": 20140
},
{
"grad_norm": 0.22020693123340607,
"learning_rate": 2.6693670127569958e-05,
"loss": 0.0047,
"step": 20150
},
{
"grad_norm": 0.15865576267242432,
"learning_rate": 2.6644922478401268e-05,
"loss": 0.0039,
"step": 20160
},
{
"grad_norm": 0.14532378315925598,
"learning_rate": 2.6596203207854004e-05,
"loss": 0.0037,
"step": 20170
},
{
"grad_norm": 0.1562606543302536,
"learning_rate": 2.654751237512666e-05,
"loss": 0.0041,
"step": 20180
},
{
"grad_norm": 0.1913229525089264,
"learning_rate": 2.649885003938323e-05,
"loss": 0.0039,
"step": 20190
},
{
"grad_norm": 0.14770890772342682,
"learning_rate": 2.6450216259753e-05,
"loss": 0.0038,
"step": 20200
},
{
"grad_norm": 0.1461491584777832,
"learning_rate": 2.640161109533063e-05,
"loss": 0.0043,
"step": 20210
},
{
"grad_norm": 0.15607817471027374,
"learning_rate": 2.6353034605175936e-05,
"loss": 0.004,
"step": 20220
},
{
"grad_norm": 0.17303478717803955,
"learning_rate": 2.630448684831398e-05,
"loss": 0.0031,
"step": 20230
},
{
"grad_norm": 0.16865776479244232,
"learning_rate": 2.625596788373482e-05,
"loss": 0.0036,
"step": 20240
},
{
"grad_norm": 0.1564178168773651,
"learning_rate": 2.620747777039363e-05,
"loss": 0.0036,
"step": 20250
},
{
"grad_norm": 0.14058387279510498,
"learning_rate": 2.6159016567210424e-05,
"loss": 0.0033,
"step": 20260
},
{
"grad_norm": 0.12073111534118652,
"learning_rate": 2.6110584333070153e-05,
"loss": 0.0038,
"step": 20270
},
{
"grad_norm": 0.18186058104038239,
"learning_rate": 2.6062181126822537e-05,
"loss": 0.0034,
"step": 20280
},
{
"grad_norm": 0.14595267176628113,
"learning_rate": 2.601380700728203e-05,
"loss": 0.0039,
"step": 20290
},
{
"grad_norm": 0.17465601861476898,
"learning_rate": 2.596546203322777e-05,
"loss": 0.0034,
"step": 20300
},
{
"grad_norm": 0.16456381976604462,
"learning_rate": 2.5917146263403457e-05,
"loss": 0.0044,
"step": 20310
},
{
"grad_norm": 0.12289533764123917,
"learning_rate": 2.5868859756517296e-05,
"loss": 0.0037,
"step": 20320
},
{
"grad_norm": 0.15113644301891327,
"learning_rate": 2.5820602571241947e-05,
"loss": 0.0036,
"step": 20330
},
{
"grad_norm": 0.14108332991600037,
"learning_rate": 2.577237476621442e-05,
"loss": 0.0034,
"step": 20340
},
{
"grad_norm": 0.2044163942337036,
"learning_rate": 2.572417640003609e-05,
"loss": 0.0036,
"step": 20350
},
{
"grad_norm": 0.2860066592693329,
"learning_rate": 2.5676007531272473e-05,
"loss": 0.0041,
"step": 20360
},
{
"grad_norm": 0.1858942210674286,
"learning_rate": 2.5627868218453332e-05,
"loss": 0.0047,
"step": 20370
},
{
"grad_norm": 0.19678841531276703,
"learning_rate": 2.5579758520072443e-05,
"loss": 0.004,
"step": 20380
},
{
"grad_norm": 0.2074602246284485,
"learning_rate": 2.5531678494587615e-05,
"loss": 0.0041,
"step": 20390
},
{
"grad_norm": 0.19061173498630524,
"learning_rate": 2.5483628200420644e-05,
"loss": 0.0038,
"step": 20400
},
{
"grad_norm": 0.24856042861938477,
"learning_rate": 2.5435607695957154e-05,
"loss": 0.0037,
"step": 20410
},
{
"grad_norm": 0.20185329020023346,
"learning_rate": 2.5387617039546586e-05,
"loss": 0.0039,
"step": 20420
},
{
"grad_norm": 0.17964529991149902,
"learning_rate": 2.5339656289502106e-05,
"loss": 0.0037,
"step": 20430
},
{
"grad_norm": 0.18105143308639526,
"learning_rate": 2.529172550410056e-05,
"loss": 0.0045,
"step": 20440
},
{
"grad_norm": 0.1333514153957367,
"learning_rate": 2.524382474158234e-05,
"loss": 0.0041,
"step": 20450
},
{
"grad_norm": 0.1410980373620987,
"learning_rate": 2.5195954060151433e-05,
"loss": 0.0035,
"step": 20460
},
{
"grad_norm": 0.17282365262508392,
"learning_rate": 2.5148113517975212e-05,
"loss": 0.0037,
"step": 20470
},
{
"grad_norm": 0.14284226298332214,
"learning_rate": 2.5100303173184447e-05,
"loss": 0.0043,
"step": 20480
},
{
"grad_norm": 0.15672485530376434,
"learning_rate": 2.5052523083873196e-05,
"loss": 0.0041,
"step": 20490
},
{
"grad_norm": 0.1434277892112732,
"learning_rate": 2.500477330809881e-05,
"loss": 0.0039,
"step": 20500
},
{
"grad_norm": 0.14215965569019318,
"learning_rate": 2.4957053903881733e-05,
"loss": 0.0035,
"step": 20510
},
{
"grad_norm": 0.19319464266300201,
"learning_rate": 2.4909364929205576e-05,
"loss": 0.0047,
"step": 20520
},
{
"grad_norm": 0.12266148626804352,
"learning_rate": 2.4861706442016923e-05,
"loss": 0.0042,
"step": 20530
},
{
"grad_norm": 0.13330376148223877,
"learning_rate": 2.481407850022533e-05,
"loss": 0.0031,
"step": 20540
},
{
"grad_norm": 0.14798171818256378,
"learning_rate": 2.4766481161703213e-05,
"loss": 0.0038,
"step": 20550
},
{
"grad_norm": 0.17796316742897034,
"learning_rate": 2.4718914484285876e-05,
"loss": 0.0055,
"step": 20560
},
{
"grad_norm": 0.14884668588638306,
"learning_rate": 2.467137852577129e-05,
"loss": 0.0037,
"step": 20570
},
{
"grad_norm": 0.18189206719398499,
"learning_rate": 2.4623873343920123e-05,
"loss": 0.0035,
"step": 20580
},
{
"grad_norm": 0.20921780169010162,
"learning_rate": 2.4576398996455656e-05,
"loss": 0.0044,
"step": 20590
},
{
"grad_norm": 0.1601589471101761,
"learning_rate": 2.452895554106368e-05,
"loss": 0.0038,
"step": 20600
},
{
"grad_norm": 0.17751874029636383,
"learning_rate": 2.4481543035392506e-05,
"loss": 0.0036,
"step": 20610
},
{
"grad_norm": 0.18711736798286438,
"learning_rate": 2.4434161537052774e-05,
"loss": 0.0037,
"step": 20620
},
{
"grad_norm": 0.17104773223400116,
"learning_rate": 2.4386811103617474e-05,
"loss": 0.004,
"step": 20630
},
{
"grad_norm": 0.1517324149608612,
"learning_rate": 2.4339491792621833e-05,
"loss": 0.0037,
"step": 20640
},
{
"grad_norm": 0.162267804145813,
"learning_rate": 2.4292203661563312e-05,
"loss": 0.004,
"step": 20650
},
{
"grad_norm": 0.1726798564195633,
"learning_rate": 2.424494676790141e-05,
"loss": 0.0038,
"step": 20660
},
{
"grad_norm": 0.1828690767288208,
"learning_rate": 2.4197721169057745e-05,
"loss": 0.0038,
"step": 20670
},
{
"grad_norm": 0.1538512259721756,
"learning_rate": 2.4150526922415857e-05,
"loss": 0.0038,
"step": 20680
},
{
"grad_norm": 0.190271258354187,
"learning_rate": 2.41033640853212e-05,
"loss": 0.0043,
"step": 20690
},
{
"grad_norm": 0.1895284801721573,
"learning_rate": 2.405623271508108e-05,
"loss": 0.0046,
"step": 20700
},
{
"grad_norm": 0.15800470113754272,
"learning_rate": 2.4009132868964522e-05,
"loss": 0.0032,
"step": 20710
},
{
"grad_norm": 0.14928565919399261,
"learning_rate": 2.3962064604202327e-05,
"loss": 0.0037,
"step": 20720
},
{
"grad_norm": 0.15557123720645905,
"learning_rate": 2.3915027977986858e-05,
"loss": 0.0034,
"step": 20730
},
{
"grad_norm": 0.16175757348537445,
"learning_rate": 2.386802304747205e-05,
"loss": 0.0039,
"step": 20740
},
{
"grad_norm": 0.12887556850910187,
"learning_rate": 2.382104986977332e-05,
"loss": 0.0036,
"step": 20750
},
{
"grad_norm": 0.13774782419204712,
"learning_rate": 2.377410850196749e-05,
"loss": 0.0035,
"step": 20760
},
{
"grad_norm": 0.20638243854045868,
"learning_rate": 2.3727199001092797e-05,
"loss": 0.0032,
"step": 20770
},
{
"grad_norm": 0.19087091088294983,
"learning_rate": 2.3680321424148677e-05,
"loss": 0.0039,
"step": 20780
},
{
"grad_norm": 0.1829594373703003,
"learning_rate": 2.3633475828095793e-05,
"loss": 0.0035,
"step": 20790
},
{
"grad_norm": 0.15062923729419708,
"learning_rate": 2.358666226985599e-05,
"loss": 0.0034,
"step": 20800
},
{
"grad_norm": 0.17616502940654755,
"learning_rate": 2.353988080631213e-05,
"loss": 0.0034,
"step": 20810
},
{
"grad_norm": 0.16354799270629883,
"learning_rate": 2.3493131494308142e-05,
"loss": 0.0039,
"step": 20820
},
{
"grad_norm": 0.15133219957351685,
"learning_rate": 2.344641439064881e-05,
"loss": 0.0038,
"step": 20830
},
{
"grad_norm": 0.15133504569530487,
"learning_rate": 2.339972955209984e-05,
"loss": 0.0035,
"step": 20840
},
{
"grad_norm": 0.17066644132137299,
"learning_rate": 2.335307703538771e-05,
"loss": 0.004,
"step": 20850
},
{
"grad_norm": 0.17749416828155518,
"learning_rate": 2.330645689719962e-05,
"loss": 0.0033,
"step": 20860
},
{
"grad_norm": 0.14155493676662445,
"learning_rate": 2.3259869194183414e-05,
"loss": 0.0039,
"step": 20870
},
{
"grad_norm": 0.13457655906677246,
"learning_rate": 2.321331398294759e-05,
"loss": 0.0037,
"step": 20880
},
{
"grad_norm": 0.18831995129585266,
"learning_rate": 2.3166791320061094e-05,
"loss": 0.0044,
"step": 20890
},
{
"grad_norm": 0.16765257716178894,
"learning_rate": 2.312030126205335e-05,
"loss": 0.0034,
"step": 20900
},
{
"grad_norm": 0.16341178119182587,
"learning_rate": 2.3073843865414162e-05,
"loss": 0.0034,
"step": 20910
},
{
"grad_norm": 0.14112359285354614,
"learning_rate": 2.302741918659363e-05,
"loss": 0.0045,
"step": 20920
},
{
"grad_norm": 0.1469755321741104,
"learning_rate": 2.2981027282002154e-05,
"loss": 0.0044,
"step": 20930
},
{
"grad_norm": 0.168148934841156,
"learning_rate": 2.2934668208010236e-05,
"loss": 0.0038,
"step": 20940
},
{
"grad_norm": 0.15941135585308075,
"learning_rate": 2.2888342020948555e-05,
"loss": 0.0041,
"step": 20950
},
{
"grad_norm": 0.12652379274368286,
"learning_rate": 2.2842048777107783e-05,
"loss": 0.0032,
"step": 20960
},
{
"grad_norm": 0.17613428831100464,
"learning_rate": 2.2795788532738553e-05,
"loss": 0.0037,
"step": 20970
},
{
"grad_norm": 0.20372003316879272,
"learning_rate": 2.2749561344051468e-05,
"loss": 0.0038,
"step": 20980
},
{
"grad_norm": 0.1964961439371109,
"learning_rate": 2.2703367267216896e-05,
"loss": 0.0041,
"step": 20990
},
{
"grad_norm": 0.15598396956920624,
"learning_rate": 2.2657206358364997e-05,
"loss": 0.0031,
"step": 21000
},
{
"grad_norm": 0.1497393250465393,
"learning_rate": 2.261107867358563e-05,
"loss": 0.0039,
"step": 21010
},
{
"grad_norm": 0.1271108090877533,
"learning_rate": 2.2564984268928263e-05,
"loss": 0.0032,
"step": 21020
},
{
"grad_norm": 0.11690191924571991,
"learning_rate": 2.251892320040198e-05,
"loss": 0.0043,
"step": 21030
},
{
"grad_norm": 0.13838931918144226,
"learning_rate": 2.2472895523975312e-05,
"loss": 0.0049,
"step": 21040
},
{
"grad_norm": 0.16570702195167542,
"learning_rate": 2.2426901295576212e-05,
"loss": 0.0029,
"step": 21050
},
{
"grad_norm": 0.1459100842475891,
"learning_rate": 2.2380940571092013e-05,
"loss": 0.0035,
"step": 21060
},
{
"grad_norm": 0.13789431750774384,
"learning_rate": 2.2335013406369302e-05,
"loss": 0.0035,
"step": 21070
},
{
"grad_norm": 0.15318845212459564,
"learning_rate": 2.228911985721397e-05,
"loss": 0.0038,
"step": 21080
},
{
"grad_norm": 0.16149044036865234,
"learning_rate": 2.224325997939095e-05,
"loss": 0.0039,
"step": 21090
},
{
"grad_norm": 0.12439160794019699,
"learning_rate": 2.219743382862437e-05,
"loss": 0.0033,
"step": 21100
},
{
"grad_norm": 0.14179007709026337,
"learning_rate": 2.2151641460597294e-05,
"loss": 0.0038,
"step": 21110
},
{
"grad_norm": 0.10861937701702118,
"learning_rate": 2.210588293095177e-05,
"loss": 0.0031,
"step": 21120
},
{
"grad_norm": 0.21190866827964783,
"learning_rate": 2.2060158295288714e-05,
"loss": 0.0035,
"step": 21130
},
{
"grad_norm": 0.14774608612060547,
"learning_rate": 2.2014467609167905e-05,
"loss": 0.0032,
"step": 21140
},
{
"grad_norm": 0.22291092574596405,
"learning_rate": 2.1968810928107806e-05,
"loss": 0.0032,
"step": 21150
},
{
"grad_norm": 0.1429012268781662,
"learning_rate": 2.1923188307585606e-05,
"loss": 0.0036,
"step": 21160
},
{
"grad_norm": 0.17996202409267426,
"learning_rate": 2.187759980303708e-05,
"loss": 0.0034,
"step": 21170
},
{
"grad_norm": 0.1727055311203003,
"learning_rate": 2.1832045469856545e-05,
"loss": 0.0034,
"step": 21180
},
{
"grad_norm": 0.18015415966510773,
"learning_rate": 2.1786525363396844e-05,
"loss": 0.0049,
"step": 21190
},
{
"grad_norm": 0.16095535457134247,
"learning_rate": 2.1741039538969183e-05,
"loss": 0.0035,
"step": 21200
},
{
"grad_norm": 0.16727764904499054,
"learning_rate": 2.169558805184313e-05,
"loss": 0.0037,
"step": 21210
},
{
"grad_norm": 0.16812165081501007,
"learning_rate": 2.165017095724651e-05,
"loss": 0.005,
"step": 21220
},
{
"grad_norm": 0.14964669942855835,
"learning_rate": 2.1604788310365405e-05,
"loss": 0.0039,
"step": 21230
},
{
"grad_norm": 0.15997344255447388,
"learning_rate": 2.155944016634398e-05,
"loss": 0.0036,
"step": 21240
},
{
"grad_norm": 0.1788637489080429,
"learning_rate": 2.1514126580284538e-05,
"loss": 0.0036,
"step": 21250
},
{
"grad_norm": 0.17881011962890625,
"learning_rate": 2.1468847607247344e-05,
"loss": 0.0037,
"step": 21260
},
{
"grad_norm": 0.1515558660030365,
"learning_rate": 2.1423603302250623e-05,
"loss": 0.004,
"step": 21270
},
{
"grad_norm": 0.15219047665596008,
"learning_rate": 2.1378393720270468e-05,
"loss": 0.0039,
"step": 21280
},
{
"grad_norm": 0.19061118364334106,
"learning_rate": 2.133321891624076e-05,
"loss": 0.0036,
"step": 21290
},
{
"grad_norm": 0.1355351358652115,
"learning_rate": 2.1288078945053193e-05,
"loss": 0.0037,
"step": 21300
},
{
"grad_norm": 0.13980156183242798,
"learning_rate": 2.1242973861557064e-05,
"loss": 0.0034,
"step": 21310
},
{
"grad_norm": 0.15740683674812317,
"learning_rate": 2.1197903720559304e-05,
"loss": 0.0042,
"step": 21320
},
{
"grad_norm": 0.14817576110363007,
"learning_rate": 2.115286857682438e-05,
"loss": 0.0039,
"step": 21330
},
{
"grad_norm": 0.165932759642601,
"learning_rate": 2.1107868485074228e-05,
"loss": 0.0035,
"step": 21340
},
{
"grad_norm": 0.12786421179771423,
"learning_rate": 2.1062903499988234e-05,
"loss": 0.0038,
"step": 21350
},
{
"grad_norm": 0.19997276365756989,
"learning_rate": 2.101797367620308e-05,
"loss": 0.0036,
"step": 21360
},
{
"grad_norm": 0.1627563238143921,
"learning_rate": 2.0973079068312713e-05,
"loss": 0.0035,
"step": 21370
},
{
"grad_norm": 0.14669613540172577,
"learning_rate": 2.0928219730868358e-05,
"loss": 0.0037,
"step": 21380
},
{
"grad_norm": 0.13211949169635773,
"learning_rate": 2.0883395718378302e-05,
"loss": 0.0042,
"step": 21390
},
{
"grad_norm": 0.17457225918769836,
"learning_rate": 2.083860708530798e-05,
"loss": 0.0032,
"step": 21400
},
{
"grad_norm": 0.14739926159381866,
"learning_rate": 2.0793853886079796e-05,
"loss": 0.0033,
"step": 21410
},
{
"grad_norm": 0.20395906269550323,
"learning_rate": 2.074913617507309e-05,
"loss": 0.0032,
"step": 21420
},
{
"grad_norm": 0.1531699150800705,
"learning_rate": 2.070445400662411e-05,
"loss": 0.0039,
"step": 21430
},
{
"grad_norm": 0.17874111235141754,
"learning_rate": 2.0659807435025907e-05,
"loss": 0.0029,
"step": 21440
},
{
"grad_norm": 0.12690158188343048,
"learning_rate": 2.061519651452825e-05,
"loss": 0.0036,
"step": 21450
},
{
"grad_norm": 0.11716289818286896,
"learning_rate": 2.0570621299337656e-05,
"loss": 0.0042,
"step": 21460
},
{
"grad_norm": 0.14322394132614136,
"learning_rate": 2.0526081843617183e-05,
"loss": 0.0029,
"step": 21470
},
{
"grad_norm": 0.13282859325408936,
"learning_rate": 2.0481578201486485e-05,
"loss": 0.0032,
"step": 21480
},
{
"grad_norm": 0.11797261983156204,
"learning_rate": 2.0437110427021677e-05,
"loss": 0.0034,
"step": 21490
},
{
"grad_norm": 0.17806091904640198,
"learning_rate": 2.0392678574255282e-05,
"loss": 0.0034,
"step": 21500
},
{
"grad_norm": 0.19595852494239807,
"learning_rate": 2.0348282697176217e-05,
"loss": 0.0035,
"step": 21510
},
{
"grad_norm": 0.1820956915616989,
"learning_rate": 2.030392284972964e-05,
"loss": 0.0038,
"step": 21520
},
{
"grad_norm": 0.22032277286052704,
"learning_rate": 2.0259599085816973e-05,
"loss": 0.0035,
"step": 21530
},
{
"grad_norm": 0.13953392207622528,
"learning_rate": 2.0215311459295755e-05,
"loss": 0.0038,
"step": 21540
},
{
"grad_norm": 0.16167394816875458,
"learning_rate": 2.0171060023979606e-05,
"loss": 0.0039,
"step": 21550
},
{
"grad_norm": 0.12433263659477234,
"learning_rate": 2.012684483363823e-05,
"loss": 0.0031,
"step": 21560
},
{
"grad_norm": 0.12760725617408752,
"learning_rate": 2.0082665941997237e-05,
"loss": 0.0033,
"step": 21570
},
{
"grad_norm": 0.18663541972637177,
"learning_rate": 2.0038523402738146e-05,
"loss": 0.0039,
"step": 21580
},
{
"grad_norm": 0.1703149825334549,
"learning_rate": 1.99944172694983e-05,
"loss": 0.0043,
"step": 21590
},
{
"grad_norm": 0.1486889123916626,
"learning_rate": 1.99503475958708e-05,
"loss": 0.0035,
"step": 21600
},
{
"grad_norm": 0.17407675087451935,
"learning_rate": 1.9906314435404484e-05,
"loss": 0.0038,
"step": 21610
},
{
"grad_norm": 0.2533443868160248,
"learning_rate": 1.986231784160378e-05,
"loss": 0.0045,
"step": 21620
},
{
"grad_norm": 0.15278710424900055,
"learning_rate": 1.9818357867928693e-05,
"loss": 0.0031,
"step": 21630
},
{
"grad_norm": 0.16832996904850006,
"learning_rate": 1.977443456779474e-05,
"loss": 0.0048,
"step": 21640
},
{
"grad_norm": 0.16247829794883728,
"learning_rate": 1.9730547994572858e-05,
"loss": 0.0032,
"step": 21650
},
{
"grad_norm": 0.16258040070533752,
"learning_rate": 1.9686698201589393e-05,
"loss": 0.0036,
"step": 21660
},
{
"grad_norm": 0.1769373118877411,
"learning_rate": 1.9642885242125964e-05,
"loss": 0.0035,
"step": 21670
},
{
"grad_norm": 0.18549641966819763,
"learning_rate": 1.9599109169419466e-05,
"loss": 0.0035,
"step": 21680
},
{
"grad_norm": 0.13956324756145477,
"learning_rate": 1.9555370036661948e-05,
"loss": 0.004,
"step": 21690
},
{
"grad_norm": 0.1261671930551529,
"learning_rate": 1.9511667897000574e-05,
"loss": 0.0028,
"step": 21700
},
{
"grad_norm": 0.0977945476770401,
"learning_rate": 1.946800280353755e-05,
"loss": 0.0034,
"step": 21710
},
{
"grad_norm": 0.15397758781909943,
"learning_rate": 1.9424374809330114e-05,
"loss": 0.0029,
"step": 21720
},
{
"grad_norm": 0.16497525572776794,
"learning_rate": 1.9380783967390382e-05,
"loss": 0.003,
"step": 21730
},
{
"grad_norm": 0.13930188119411469,
"learning_rate": 1.933723033068533e-05,
"loss": 0.0034,
"step": 21740
},
{
"grad_norm": 0.10556606203317642,
"learning_rate": 1.929371395213674e-05,
"loss": 0.0035,
"step": 21750
},
{
"grad_norm": 0.17041277885437012,
"learning_rate": 1.9250234884621092e-05,
"loss": 0.0032,
"step": 21760
},
{
"grad_norm": 0.15866322815418243,
"learning_rate": 1.920679318096959e-05,
"loss": 0.003,
"step": 21770
},
{
"grad_norm": 0.1705976128578186,
"learning_rate": 1.9163388893967982e-05,
"loss": 0.0045,
"step": 21780
},
{
"grad_norm": 0.13943716883659363,
"learning_rate": 1.9120022076356575e-05,
"loss": 0.0032,
"step": 21790
},
{
"grad_norm": 0.1533510833978653,
"learning_rate": 1.9076692780830114e-05,
"loss": 0.0043,
"step": 21800
},
{
"grad_norm": 0.1156059205532074,
"learning_rate": 1.903340106003782e-05,
"loss": 0.0038,
"step": 21810
},
{
"grad_norm": 0.14550359547138214,
"learning_rate": 1.8990146966583183e-05,
"loss": 0.0031,
"step": 21820
},
{
"grad_norm": 0.16395488381385803,
"learning_rate": 1.8946930553024035e-05,
"loss": 0.0034,
"step": 21830
},
{
"grad_norm": 0.15124045312404633,
"learning_rate": 1.8903751871872378e-05,
"loss": 0.0029,
"step": 21840
},
{
"grad_norm": 0.15901857614517212,
"learning_rate": 1.8860610975594382e-05,
"loss": 0.0033,
"step": 21850
},
{
"grad_norm": 0.12475143373012543,
"learning_rate": 1.8817507916610307e-05,
"loss": 0.0032,
"step": 21860
},
{
"grad_norm": 0.14250823855400085,
"learning_rate": 1.8774442747294407e-05,
"loss": 0.0028,
"step": 21870
},
{
"grad_norm": 0.15446875989437103,
"learning_rate": 1.8731415519974965e-05,
"loss": 0.0035,
"step": 21880
},
{
"grad_norm": 0.14338156580924988,
"learning_rate": 1.86884262869341e-05,
"loss": 0.003,
"step": 21890
},
{
"grad_norm": 0.13325582444667816,
"learning_rate": 1.8645475100407788e-05,
"loss": 0.0038,
"step": 21900
},
{
"grad_norm": 0.1351625770330429,
"learning_rate": 1.8602562012585768e-05,
"loss": 0.0031,
"step": 21910
},
{
"grad_norm": 0.13701239228248596,
"learning_rate": 1.8559687075611466e-05,
"loss": 0.0033,
"step": 21920
},
{
"grad_norm": 0.16110867261886597,
"learning_rate": 1.8516850341582017e-05,
"loss": 0.0034,
"step": 21930
},
{
"grad_norm": 0.1565161496400833,
"learning_rate": 1.847405186254807e-05,
"loss": 0.0032,
"step": 21940
},
{
"grad_norm": 0.20353558659553528,
"learning_rate": 1.843129169051379e-05,
"loss": 0.0036,
"step": 21950
},
{
"grad_norm": 0.17051005363464355,
"learning_rate": 1.8388569877436863e-05,
"loss": 0.0038,
"step": 21960
},
{
"grad_norm": 0.13585247099399567,
"learning_rate": 1.834588647522828e-05,
"loss": 0.0034,
"step": 21970
},
{
"grad_norm": 0.14147113263607025,
"learning_rate": 1.8303241535752435e-05,
"loss": 0.0035,
"step": 21980
},
{
"grad_norm": 0.18425710499286652,
"learning_rate": 1.8260635110826934e-05,
"loss": 0.0036,
"step": 21990
},
{
"grad_norm": 0.1377650946378708,
"learning_rate": 1.8218067252222597e-05,
"loss": 0.0033,
"step": 22000
},
{
"grad_norm": 0.1321236938238144,
"learning_rate": 1.817553801166339e-05,
"loss": 0.0028,
"step": 22010
},
{
"grad_norm": 0.10822811722755432,
"learning_rate": 1.8133047440826333e-05,
"loss": 0.0033,
"step": 22020
},
{
"grad_norm": 0.14122188091278076,
"learning_rate": 1.809059559134151e-05,
"loss": 0.0035,
"step": 22030
},
{
"grad_norm": 0.1690724641084671,
"learning_rate": 1.80481825147919e-05,
"loss": 0.0031,
"step": 22040
},
{
"grad_norm": 0.15640504658222198,
"learning_rate": 1.8005808262713396e-05,
"loss": 0.0038,
"step": 22050
},
{
"grad_norm": 0.14351242780685425,
"learning_rate": 1.7963472886594714e-05,
"loss": 0.0038,
"step": 22060
},
{
"grad_norm": 0.12544845044612885,
"learning_rate": 1.7921176437877302e-05,
"loss": 0.0033,
"step": 22070
},
{
"grad_norm": 0.1631358116865158,
"learning_rate": 1.7878918967955366e-05,
"loss": 0.0046,
"step": 22080
},
{
"grad_norm": 0.15666554868221283,
"learning_rate": 1.7836700528175693e-05,
"loss": 0.0037,
"step": 22090
},
{
"grad_norm": 0.12624230980873108,
"learning_rate": 1.779452116983769e-05,
"loss": 0.0031,
"step": 22100
},
{
"grad_norm": 0.1428545117378235,
"learning_rate": 1.775238094419325e-05,
"loss": 0.0043,
"step": 22110
},
{
"grad_norm": 0.15449422597885132,
"learning_rate": 1.7710279902446708e-05,
"loss": 0.0041,
"step": 22120
},
{
"grad_norm": 0.1390950232744217,
"learning_rate": 1.7668218095754795e-05,
"loss": 0.0036,
"step": 22130
},
{
"grad_norm": 0.15874336659908295,
"learning_rate": 1.7626195575226594e-05,
"loss": 0.0041,
"step": 22140
},
{
"grad_norm": 0.14355705678462982,
"learning_rate": 1.758421239192343e-05,
"loss": 0.003,
"step": 22150
},
{
"grad_norm": 0.19164258241653442,
"learning_rate": 1.7542268596858814e-05,
"loss": 0.0031,
"step": 22160
},
{
"grad_norm": 0.11927066743373871,
"learning_rate": 1.750036424099841e-05,
"loss": 0.0032,
"step": 22170
},
{
"grad_norm": 0.1440213918685913,
"learning_rate": 1.7458499375259956e-05,
"loss": 0.0032,
"step": 22180
},
{
"grad_norm": 0.12303514778614044,
"learning_rate": 1.7416674050513245e-05,
"loss": 0.0031,
"step": 22190
},
{
"grad_norm": 0.12144245207309723,
"learning_rate": 1.737488831757997e-05,
"loss": 0.0029,
"step": 22200
},
{
"grad_norm": 0.15181376039981842,
"learning_rate": 1.733314222723373e-05,
"loss": 0.0034,
"step": 22210
},
{
"grad_norm": 0.1315053105354309,
"learning_rate": 1.7291435830199954e-05,
"loss": 0.0032,
"step": 22220
},
{
"grad_norm": 0.12961259484291077,
"learning_rate": 1.7249769177155878e-05,
"loss": 0.0032,
"step": 22230
},
{
"grad_norm": 0.11451901495456696,
"learning_rate": 1.720814231873038e-05,
"loss": 0.0031,
"step": 22240
},
{
"grad_norm": 0.1466490477323532,
"learning_rate": 1.716655530550405e-05,
"loss": 0.003,
"step": 22250
},
{
"grad_norm": 0.15228396654129028,
"learning_rate": 1.7125008188009017e-05,
"loss": 0.0028,
"step": 22260
},
{
"grad_norm": 0.16324429214000702,
"learning_rate": 1.708350101672894e-05,
"loss": 0.004,
"step": 22270
},
{
"grad_norm": 0.16664114594459534,
"learning_rate": 1.7042033842098957e-05,
"loss": 0.0036,
"step": 22280
},
{
"grad_norm": 0.17449450492858887,
"learning_rate": 1.7000606714505567e-05,
"loss": 0.0042,
"step": 22290
},
{
"grad_norm": 0.12075686454772949,
"learning_rate": 1.695921968428668e-05,
"loss": 0.0025,
"step": 22300
},
{
"grad_norm": 0.12552675604820251,
"learning_rate": 1.6917872801731416e-05,
"loss": 0.0033,
"step": 22310
},
{
"grad_norm": 0.172083780169487,
"learning_rate": 1.687656611708014e-05,
"loss": 0.0027,
"step": 22320
},
{
"grad_norm": 0.1396370679140091,
"learning_rate": 1.6835299680524368e-05,
"loss": 0.0031,
"step": 22330
},
{
"grad_norm": 0.17838117480278015,
"learning_rate": 1.679407354220669e-05,
"loss": 0.0039,
"step": 22340
},
{
"grad_norm": 0.1274147480726242,
"learning_rate": 1.6752887752220793e-05,
"loss": 0.0037,
"step": 22350
},
{
"grad_norm": 0.15604834258556366,
"learning_rate": 1.6711742360611277e-05,
"loss": 0.0035,
"step": 22360
},
{
"grad_norm": 0.14894907176494598,
"learning_rate": 1.6670637417373652e-05,
"loss": 0.0032,
"step": 22370
},
{
"grad_norm": 0.12530560791492462,
"learning_rate": 1.6629572972454333e-05,
"loss": 0.0028,
"step": 22380
},
{
"grad_norm": 0.11448613554239273,
"learning_rate": 1.6588549075750464e-05,
"loss": 0.0028,
"step": 22390
},
{
"grad_norm": 0.12534135580062866,
"learning_rate": 1.6547565777109976e-05,
"loss": 0.0026,
"step": 22400
},
{
"grad_norm": 0.1331997662782669,
"learning_rate": 1.650662312633143e-05,
"loss": 0.0038,
"step": 22410
},
{
"grad_norm": 0.13440582156181335,
"learning_rate": 1.6465721173164002e-05,
"loss": 0.003,
"step": 22420
},
{
"grad_norm": 0.14853711426258087,
"learning_rate": 1.6424859967307428e-05,
"loss": 0.0031,
"step": 22430
},
{
"grad_norm": 0.11766880005598068,
"learning_rate": 1.6384039558411903e-05,
"loss": 0.0026,
"step": 22440
},
{
"grad_norm": 0.10507412254810333,
"learning_rate": 1.634325999607811e-05,
"loss": 0.0036,
"step": 22450
},
{
"grad_norm": 0.17563402652740479,
"learning_rate": 1.6302521329857045e-05,
"loss": 0.0038,
"step": 22460
},
{
"grad_norm": 0.1594231277704239,
"learning_rate": 1.6261823609250027e-05,
"loss": 0.0034,
"step": 22470
},
{
"grad_norm": 0.10568983107805252,
"learning_rate": 1.6221166883708627e-05,
"loss": 0.0029,
"step": 22480
},
{
"grad_norm": 0.11970749497413635,
"learning_rate": 1.6180551202634603e-05,
"loss": 0.0025,
"step": 22490
},
{
"grad_norm": 0.13270173966884613,
"learning_rate": 1.613997661537981e-05,
"loss": 0.004,
"step": 22500
}
],
"logging_steps": 10,
"max_steps": 30000,
"num_input_tokens_seen": 0,
"num_train_epochs": 9223372036854775807,
"save_steps": 7500,
"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": 128,
"trial_name": null,
"trial_params": null
}