| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.15600928255231186, | |
| "eval_steps": 500, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00015600928255231187, | |
| "grad_norm": 0.5937075614929199, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 3.7681, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00031201856510462375, | |
| "grad_norm": 0.5836828947067261, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 3.8601, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00046802784765693557, | |
| "grad_norm": 0.6111788749694824, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 3.643, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0006240371302092475, | |
| "grad_norm": 0.5807424783706665, | |
| "learning_rate": 1.0666666666666667e-05, | |
| "loss": 3.724, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0007800464127615594, | |
| "grad_norm": 0.5708947777748108, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 3.9727, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0009360556953138711, | |
| "grad_norm": 0.5662252902984619, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 3.6801, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.001092064977866183, | |
| "grad_norm": 0.5653729438781738, | |
| "learning_rate": 1.866666666666667e-05, | |
| "loss": 3.6898, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.001248074260418495, | |
| "grad_norm": 0.5451233983039856, | |
| "learning_rate": 2.1333333333333335e-05, | |
| "loss": 3.5484, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0014040835429708068, | |
| "grad_norm": 0.5682435035705566, | |
| "learning_rate": 2.4e-05, | |
| "loss": 3.538, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0015600928255231187, | |
| "grad_norm": 0.6180667877197266, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 3.6757, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0017161021080754305, | |
| "grad_norm": 0.6358373165130615, | |
| "learning_rate": 2.9333333333333336e-05, | |
| "loss": 3.6489, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0018721113906277423, | |
| "grad_norm": 0.6643233895301819, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 3.4313, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0020281206731800542, | |
| "grad_norm": 0.6591399908065796, | |
| "learning_rate": 3.466666666666667e-05, | |
| "loss": 3.472, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.002184129955732366, | |
| "grad_norm": 0.8929205536842346, | |
| "learning_rate": 3.733333333333334e-05, | |
| "loss": 3.3531, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0023401392382846778, | |
| "grad_norm": 1.2845464944839478, | |
| "learning_rate": 4e-05, | |
| "loss": 3.143, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.00249614852083699, | |
| "grad_norm": 1.216373085975647, | |
| "learning_rate": 4.266666666666667e-05, | |
| "loss": 3.1297, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0026521578033893017, | |
| "grad_norm": 0.9192391633987427, | |
| "learning_rate": 4.5333333333333335e-05, | |
| "loss": 2.9826, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0028081670859416135, | |
| "grad_norm": 0.8917486667633057, | |
| "learning_rate": 4.8e-05, | |
| "loss": 2.9068, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0029641763684939253, | |
| "grad_norm": 0.7141512632369995, | |
| "learning_rate": 5.0666666666666674e-05, | |
| "loss": 2.7797, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0031201856510462375, | |
| "grad_norm": 0.8795380592346191, | |
| "learning_rate": 5.333333333333333e-05, | |
| "loss": 2.2265, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0032761949335985492, | |
| "grad_norm": 1.047784447669983, | |
| "learning_rate": 5.6000000000000006e-05, | |
| "loss": 2.6089, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.003432204216150861, | |
| "grad_norm": 1.0959978103637695, | |
| "learning_rate": 5.866666666666667e-05, | |
| "loss": 2.3416, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0035882134987031728, | |
| "grad_norm": 1.283445954322815, | |
| "learning_rate": 6.133333333333334e-05, | |
| "loss": 2.0565, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0037442227812554845, | |
| "grad_norm": 1.655569314956665, | |
| "learning_rate": 6.400000000000001e-05, | |
| "loss": 1.6648, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0039002320638077967, | |
| "grad_norm": 1.4048818349838257, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 1.7566, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0040562413463601085, | |
| "grad_norm": 1.0755441188812256, | |
| "learning_rate": 6.933333333333334e-05, | |
| "loss": 1.6591, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.00421225062891242, | |
| "grad_norm": 0.7240940928459167, | |
| "learning_rate": 7.2e-05, | |
| "loss": 2.21, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.004368259911464732, | |
| "grad_norm": 0.48980680108070374, | |
| "learning_rate": 7.466666666666667e-05, | |
| "loss": 1.8157, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.004524269194017044, | |
| "grad_norm": 0.4145239591598511, | |
| "learning_rate": 7.733333333333333e-05, | |
| "loss": 1.8679, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0046802784765693555, | |
| "grad_norm": 0.3905705213546753, | |
| "learning_rate": 8e-05, | |
| "loss": 1.5733, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.004836287759121667, | |
| "grad_norm": 0.40969792008399963, | |
| "learning_rate": 8.266666666666667e-05, | |
| "loss": 1.531, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.00499229704167398, | |
| "grad_norm": 0.4269125759601593, | |
| "learning_rate": 8.533333333333334e-05, | |
| "loss": 1.3705, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.005148306324226292, | |
| "grad_norm": 0.5876020789146423, | |
| "learning_rate": 8.800000000000001e-05, | |
| "loss": 1.4055, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0053043156067786035, | |
| "grad_norm": 0.39753037691116333, | |
| "learning_rate": 9.066666666666667e-05, | |
| "loss": 1.5403, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.005460324889330915, | |
| "grad_norm": 0.4157419800758362, | |
| "learning_rate": 9.333333333333334e-05, | |
| "loss": 1.5698, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.005616334171883227, | |
| "grad_norm": 0.4430864155292511, | |
| "learning_rate": 9.6e-05, | |
| "loss": 1.5466, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.005772343454435539, | |
| "grad_norm": 0.5259338021278381, | |
| "learning_rate": 9.866666666666668e-05, | |
| "loss": 1.5878, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0059283527369878505, | |
| "grad_norm": 0.4409235417842865, | |
| "learning_rate": 0.00010133333333333335, | |
| "loss": 1.4203, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.006084362019540162, | |
| "grad_norm": 0.5432307124137878, | |
| "learning_rate": 0.00010400000000000001, | |
| "loss": 1.3843, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.006240371302092475, | |
| "grad_norm": 0.5789123177528381, | |
| "learning_rate": 0.00010666666666666667, | |
| "loss": 1.4649, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.006396380584644787, | |
| "grad_norm": 0.5596875548362732, | |
| "learning_rate": 0.00010933333333333333, | |
| "loss": 1.3569, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0065523898671970985, | |
| "grad_norm": 0.6517161726951599, | |
| "learning_rate": 0.00011200000000000001, | |
| "loss": 1.4306, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.00670839914974941, | |
| "grad_norm": 0.7665486335754395, | |
| "learning_rate": 0.00011466666666666667, | |
| "loss": 1.5369, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.006864408432301722, | |
| "grad_norm": 0.8421632647514343, | |
| "learning_rate": 0.00011733333333333334, | |
| "loss": 1.5651, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.007020417714854034, | |
| "grad_norm": 0.8437005877494812, | |
| "learning_rate": 0.00012, | |
| "loss": 1.33, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0071764269974063455, | |
| "grad_norm": 0.3544560670852661, | |
| "learning_rate": 0.00012266666666666668, | |
| "loss": 1.5436, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.007332436279958657, | |
| "grad_norm": 0.35725343227386475, | |
| "learning_rate": 0.00012533333333333334, | |
| "loss": 1.5792, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.007488445562510969, | |
| "grad_norm": 0.7203790545463562, | |
| "learning_rate": 0.00012800000000000002, | |
| "loss": 1.6707, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.007644454845063282, | |
| "grad_norm": 0.352791428565979, | |
| "learning_rate": 0.00013066666666666668, | |
| "loss": 1.5251, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0078004641276155934, | |
| "grad_norm": 0.49014368653297424, | |
| "learning_rate": 0.00013333333333333334, | |
| "loss": 1.3478, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.007956473410167905, | |
| "grad_norm": 0.29890525341033936, | |
| "learning_rate": 0.00013600000000000003, | |
| "loss": 1.3164, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.008112482692720217, | |
| "grad_norm": 0.34632885456085205, | |
| "learning_rate": 0.00013866666666666669, | |
| "loss": 1.3785, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.008268491975272529, | |
| "grad_norm": 0.3631187677383423, | |
| "learning_rate": 0.00014133333333333334, | |
| "loss": 1.7873, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.00842450125782484, | |
| "grad_norm": 0.290487140417099, | |
| "learning_rate": 0.000144, | |
| "loss": 1.1681, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.008580510540377152, | |
| "grad_norm": 0.3136501610279083, | |
| "learning_rate": 0.00014666666666666666, | |
| "loss": 1.1332, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.008736519822929464, | |
| "grad_norm": 0.3708946406841278, | |
| "learning_rate": 0.00014933333333333335, | |
| "loss": 1.4316, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.008892529105481776, | |
| "grad_norm": 0.3645316958427429, | |
| "learning_rate": 0.000152, | |
| "loss": 1.3522, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.009048538388034088, | |
| "grad_norm": 0.4074520170688629, | |
| "learning_rate": 0.00015466666666666667, | |
| "loss": 1.5344, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0092045476705864, | |
| "grad_norm": 0.3106740713119507, | |
| "learning_rate": 0.00015733333333333333, | |
| "loss": 1.2959, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.009360556953138711, | |
| "grad_norm": 0.32623976469039917, | |
| "learning_rate": 0.00016, | |
| "loss": 1.6472, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.009516566235691023, | |
| "grad_norm": 0.35396724939346313, | |
| "learning_rate": 0.00016266666666666667, | |
| "loss": 1.2655, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.009672575518243335, | |
| "grad_norm": 0.3455830216407776, | |
| "learning_rate": 0.00016533333333333333, | |
| "loss": 1.2153, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.009828584800795648, | |
| "grad_norm": 0.3116808235645294, | |
| "learning_rate": 0.000168, | |
| "loss": 1.0851, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.00998459408334796, | |
| "grad_norm": 0.3416989743709564, | |
| "learning_rate": 0.00017066666666666668, | |
| "loss": 1.5828, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.010140603365900272, | |
| "grad_norm": 0.3509654104709625, | |
| "learning_rate": 0.00017333333333333334, | |
| "loss": 1.4832, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.010296612648452583, | |
| "grad_norm": 0.3034147322177887, | |
| "learning_rate": 0.00017600000000000002, | |
| "loss": 1.4326, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.010452621931004895, | |
| "grad_norm": 0.3084355890750885, | |
| "learning_rate": 0.00017866666666666668, | |
| "loss": 1.2452, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.010608631213557207, | |
| "grad_norm": 0.3001956343650818, | |
| "learning_rate": 0.00018133333333333334, | |
| "loss": 1.2484, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.010764640496109519, | |
| "grad_norm": 0.30605360865592957, | |
| "learning_rate": 0.00018400000000000003, | |
| "loss": 1.2137, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.01092064977866183, | |
| "grad_norm": 0.32967764139175415, | |
| "learning_rate": 0.0001866666666666667, | |
| "loss": 1.4, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.011076659061214142, | |
| "grad_norm": 0.3161776661872864, | |
| "learning_rate": 0.00018933333333333335, | |
| "loss": 1.3203, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.011232668343766454, | |
| "grad_norm": 0.28808867931365967, | |
| "learning_rate": 0.000192, | |
| "loss": 1.3034, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.011388677626318766, | |
| "grad_norm": 0.2804367244243622, | |
| "learning_rate": 0.0001946666666666667, | |
| "loss": 1.2753, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.011544686908871078, | |
| "grad_norm": 0.30980467796325684, | |
| "learning_rate": 0.00019733333333333335, | |
| "loss": 1.3733, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.01170069619142339, | |
| "grad_norm": 0.31240588426589966, | |
| "learning_rate": 0.0002, | |
| "loss": 1.1602, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.011856705473975701, | |
| "grad_norm": 0.28906041383743286, | |
| "learning_rate": 0.00019999991608372393, | |
| "loss": 1.3243, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.012012714756528013, | |
| "grad_norm": 0.2740985155105591, | |
| "learning_rate": 0.00019999966433503652, | |
| "loss": 1.1853, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.012168724039080325, | |
| "grad_norm": 0.30425482988357544, | |
| "learning_rate": 0.0001999992447543603, | |
| "loss": 1.3282, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.012324733321632636, | |
| "grad_norm": 0.3216018080711365, | |
| "learning_rate": 0.00019999865734239946, | |
| "loss": 1.3696, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.01248074260418495, | |
| "grad_norm": 0.34770438075065613, | |
| "learning_rate": 0.00019999790210013988, | |
| "loss": 1.261, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.012636751886737262, | |
| "grad_norm": 0.3883892297744751, | |
| "learning_rate": 0.0001999969790288491, | |
| "loss": 1.5873, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.012792761169289573, | |
| "grad_norm": 0.3061410188674927, | |
| "learning_rate": 0.00019999588813007633, | |
| "loss": 1.5559, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.012948770451841885, | |
| "grad_norm": 0.3044775128364563, | |
| "learning_rate": 0.00019999462940565243, | |
| "loss": 1.2439, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.013104779734394197, | |
| "grad_norm": 0.3562803864479065, | |
| "learning_rate": 0.00019999320285769, | |
| "loss": 1.4121, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.013260789016946509, | |
| "grad_norm": 0.3367731273174286, | |
| "learning_rate": 0.0001999916084885832, | |
| "loss": 1.1937, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.01341679829949882, | |
| "grad_norm": 0.3613661527633667, | |
| "learning_rate": 0.00019998984630100792, | |
| "loss": 1.4267, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.013572807582051132, | |
| "grad_norm": 0.30924999713897705, | |
| "learning_rate": 0.0001999879162979217, | |
| "loss": 1.3358, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.013728816864603444, | |
| "grad_norm": 0.34925562143325806, | |
| "learning_rate": 0.0001999858184825637, | |
| "loss": 1.3758, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.013884826147155756, | |
| "grad_norm": 0.3182036280632019, | |
| "learning_rate": 0.00019998355285845475, | |
| "loss": 1.3151, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.014040835429708068, | |
| "grad_norm": 0.6028950810432434, | |
| "learning_rate": 0.0001999811194293973, | |
| "loss": 1.3797, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.01419684471226038, | |
| "grad_norm": 0.3221015930175781, | |
| "learning_rate": 0.00019997851819947537, | |
| "loss": 1.3293, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.014352853994812691, | |
| "grad_norm": 0.3003532290458679, | |
| "learning_rate": 0.00019997574917305478, | |
| "loss": 1.5671, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.014508863277365003, | |
| "grad_norm": 0.32144418358802795, | |
| "learning_rate": 0.00019997281235478278, | |
| "loss": 1.3733, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.014664872559917315, | |
| "grad_norm": 1.3427015542984009, | |
| "learning_rate": 0.00019996970774958836, | |
| "loss": 1.246, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.014820881842469626, | |
| "grad_norm": 0.3254302144050598, | |
| "learning_rate": 0.00019996643536268204, | |
| "loss": 1.3829, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.014976891125021938, | |
| "grad_norm": 0.2829325795173645, | |
| "learning_rate": 0.0001999629951995559, | |
| "loss": 1.2176, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.01513290040757425, | |
| "grad_norm": 0.2943004071712494, | |
| "learning_rate": 0.00019995938726598373, | |
| "loss": 1.4021, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.015288909690126563, | |
| "grad_norm": 0.2698727548122406, | |
| "learning_rate": 0.00019995561156802079, | |
| "loss": 1.2897, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.015444918972678875, | |
| "grad_norm": 0.32416194677352905, | |
| "learning_rate": 0.0001999516681120039, | |
| "loss": 1.218, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.015600928255231187, | |
| "grad_norm": 0.3309131860733032, | |
| "learning_rate": 0.00019994755690455152, | |
| "loss": 1.4658, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.015756937537783497, | |
| "grad_norm": 0.31126394867897034, | |
| "learning_rate": 0.0001999432779525635, | |
| "loss": 1.5518, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.01591294682033581, | |
| "grad_norm": 0.28427934646606445, | |
| "learning_rate": 0.0001999388312632214, | |
| "loss": 1.1435, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.01606895610288812, | |
| "grad_norm": 0.28065958619117737, | |
| "learning_rate": 0.00019993421684398824, | |
| "loss": 1.3537, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.016224965385440434, | |
| "grad_norm": 0.3787417411804199, | |
| "learning_rate": 0.00019992943470260844, | |
| "loss": 1.2151, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.016380974667992744, | |
| "grad_norm": 0.32704487442970276, | |
| "learning_rate": 0.00019992448484710797, | |
| "loss": 1.1383, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.016536983950545057, | |
| "grad_norm": 0.34436190128326416, | |
| "learning_rate": 0.00019991936728579437, | |
| "loss": 1.3949, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.01669299323309737, | |
| "grad_norm": 0.29938092827796936, | |
| "learning_rate": 0.00019991408202725655, | |
| "loss": 1.2821, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.01684900251564968, | |
| "grad_norm": 0.3192508816719055, | |
| "learning_rate": 0.0001999086290803649, | |
| "loss": 1.3655, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.017005011798201995, | |
| "grad_norm": 0.2626635730266571, | |
| "learning_rate": 0.00019990300845427125, | |
| "loss": 1.2366, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.017161021080754305, | |
| "grad_norm": 0.288725882768631, | |
| "learning_rate": 0.0001998972201584088, | |
| "loss": 1.0589, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.017317030363306618, | |
| "grad_norm": 0.3340204358100891, | |
| "learning_rate": 0.00019989126420249221, | |
| "loss": 1.7077, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.017473039645858928, | |
| "grad_norm": 0.27165043354034424, | |
| "learning_rate": 0.00019988514059651752, | |
| "loss": 1.3596, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.01762904892841124, | |
| "grad_norm": 0.2751217186450958, | |
| "learning_rate": 0.00019987884935076213, | |
| "loss": 1.281, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.01778505821096355, | |
| "grad_norm": 0.2712443172931671, | |
| "learning_rate": 0.00019987239047578482, | |
| "loss": 1.2686, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.017941067493515865, | |
| "grad_norm": 0.2898474931716919, | |
| "learning_rate": 0.00019986576398242566, | |
| "loss": 1.2425, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.018097076776068175, | |
| "grad_norm": 0.29883307218551636, | |
| "learning_rate": 0.00019985896988180605, | |
| "loss": 1.6326, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.01825308605862049, | |
| "grad_norm": 0.2548903524875641, | |
| "learning_rate": 0.00019985200818532875, | |
| "loss": 1.317, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.0184090953411728, | |
| "grad_norm": 0.260768860578537, | |
| "learning_rate": 0.0001998448789046777, | |
| "loss": 1.4137, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.018565104623725112, | |
| "grad_norm": 0.27813923358917236, | |
| "learning_rate": 0.00019983758205181822, | |
| "loss": 1.1758, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.018721113906277422, | |
| "grad_norm": 0.29539602994918823, | |
| "learning_rate": 0.00019983011763899673, | |
| "loss": 1.2805, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.018877123188829736, | |
| "grad_norm": 0.2691763937473297, | |
| "learning_rate": 0.00019982248567874098, | |
| "loss": 1.3098, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.019033132471382046, | |
| "grad_norm": 0.2895521819591522, | |
| "learning_rate": 0.00019981468618385988, | |
| "loss": 1.1475, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.01918914175393436, | |
| "grad_norm": 0.24555402994155884, | |
| "learning_rate": 0.00019980671916744352, | |
| "loss": 1.075, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.01934515103648667, | |
| "grad_norm": 0.29935726523399353, | |
| "learning_rate": 0.00019979858464286317, | |
| "loss": 1.278, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.019501160319038983, | |
| "grad_norm": 0.3469449579715729, | |
| "learning_rate": 0.00019979028262377118, | |
| "loss": 1.602, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.019657169601591296, | |
| "grad_norm": 0.2707567811012268, | |
| "learning_rate": 0.00019978181312410104, | |
| "loss": 1.3181, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.019813178884143606, | |
| "grad_norm": 0.32349273562431335, | |
| "learning_rate": 0.00019977317615806737, | |
| "loss": 1.4862, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.01996918816669592, | |
| "grad_norm": 0.24527911841869354, | |
| "learning_rate": 0.00019976437174016573, | |
| "loss": 1.169, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.02012519744924823, | |
| "grad_norm": 0.2882062494754791, | |
| "learning_rate": 0.00019975539988517288, | |
| "loss": 1.275, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.020281206731800543, | |
| "grad_norm": 0.3206437826156616, | |
| "learning_rate": 0.00019974626060814647, | |
| "loss": 1.682, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.020437216014352853, | |
| "grad_norm": 0.3423447012901306, | |
| "learning_rate": 0.0001997369539244252, | |
| "loss": 1.2018, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.020593225296905167, | |
| "grad_norm": 0.29081955552101135, | |
| "learning_rate": 0.0001997274798496287, | |
| "loss": 1.5849, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.020749234579457477, | |
| "grad_norm": 0.2659798860549927, | |
| "learning_rate": 0.00019971783839965756, | |
| "loss": 1.1371, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.02090524386200979, | |
| "grad_norm": 0.3395417034626007, | |
| "learning_rate": 0.00019970802959069328, | |
| "loss": 1.5046, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.0210612531445621, | |
| "grad_norm": 0.22527103126049042, | |
| "learning_rate": 0.00019969805343919821, | |
| "loss": 1.0543, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.021217262427114414, | |
| "grad_norm": 0.30680522322654724, | |
| "learning_rate": 0.0001996879099619156, | |
| "loss": 1.5067, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.021373271709666724, | |
| "grad_norm": 0.22828875482082367, | |
| "learning_rate": 0.00019967759917586953, | |
| "loss": 1.1201, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.021529280992219037, | |
| "grad_norm": 0.2578384280204773, | |
| "learning_rate": 0.00019966712109836476, | |
| "loss": 1.104, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.021685290274771347, | |
| "grad_norm": 0.23175813257694244, | |
| "learning_rate": 0.000199656475746987, | |
| "loss": 0.9706, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.02184129955732366, | |
| "grad_norm": 0.29308339953422546, | |
| "learning_rate": 0.00019964566313960264, | |
| "loss": 1.4769, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.02199730883987597, | |
| "grad_norm": 0.3059382438659668, | |
| "learning_rate": 0.0001996346832943587, | |
| "loss": 1.4555, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.022153318122428284, | |
| "grad_norm": 0.2929370701313019, | |
| "learning_rate": 0.00019962353622968295, | |
| "loss": 1.4051, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.022309327404980598, | |
| "grad_norm": 0.24365079402923584, | |
| "learning_rate": 0.00019961222196428378, | |
| "loss": 1.189, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.022465336687532908, | |
| "grad_norm": 0.27418485283851624, | |
| "learning_rate": 0.0001996007405171502, | |
| "loss": 1.206, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.02262134597008522, | |
| "grad_norm": 0.2554856836795807, | |
| "learning_rate": 0.00019958909190755187, | |
| "loss": 1.4053, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.02277735525263753, | |
| "grad_norm": 0.2674770951271057, | |
| "learning_rate": 0.00019957727615503888, | |
| "loss": 1.2412, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.022933364535189845, | |
| "grad_norm": 0.3177204728126526, | |
| "learning_rate": 0.00019956529327944198, | |
| "loss": 1.4231, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.023089373817742155, | |
| "grad_norm": 0.2678688168525696, | |
| "learning_rate": 0.00019955314330087225, | |
| "loss": 1.2494, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.02324538310029447, | |
| "grad_norm": 0.28164568543434143, | |
| "learning_rate": 0.00019954082623972142, | |
| "loss": 1.2008, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.02340139238284678, | |
| "grad_norm": 0.2897564172744751, | |
| "learning_rate": 0.0001995283421166614, | |
| "loss": 1.463, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.023557401665399092, | |
| "grad_norm": 0.276509165763855, | |
| "learning_rate": 0.00019951569095264473, | |
| "loss": 1.4891, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.023713410947951402, | |
| "grad_norm": 0.2585453689098358, | |
| "learning_rate": 0.0001995028727689041, | |
| "loss": 1.1551, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.023869420230503716, | |
| "grad_norm": 0.25659292936325073, | |
| "learning_rate": 0.00019948988758695263, | |
| "loss": 1.1622, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.024025429513056026, | |
| "grad_norm": 0.27132928371429443, | |
| "learning_rate": 0.00019947673542858367, | |
| "loss": 1.2015, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.02418143879560834, | |
| "grad_norm": 0.2951599955558777, | |
| "learning_rate": 0.00019946341631587087, | |
| "loss": 1.1842, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.02433744807816065, | |
| "grad_norm": 0.3114786148071289, | |
| "learning_rate": 0.00019944993027116797, | |
| "loss": 1.4509, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.024493457360712963, | |
| "grad_norm": 0.25183674693107605, | |
| "learning_rate": 0.00019943627731710897, | |
| "loss": 1.1474, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.024649466643265273, | |
| "grad_norm": 0.2717629075050354, | |
| "learning_rate": 0.00019942245747660796, | |
| "loss": 1.2899, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.024805475925817586, | |
| "grad_norm": 0.2532605826854706, | |
| "learning_rate": 0.00019940847077285916, | |
| "loss": 1.0811, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.0249614852083699, | |
| "grad_norm": 0.2951716482639313, | |
| "learning_rate": 0.0001993943172293368, | |
| "loss": 1.6252, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.02511749449092221, | |
| "grad_norm": 0.29894542694091797, | |
| "learning_rate": 0.0001993799968697951, | |
| "loss": 1.3754, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.025273503773474523, | |
| "grad_norm": 0.28648853302001953, | |
| "learning_rate": 0.00019936550971826834, | |
| "loss": 1.2769, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.025429513056026833, | |
| "grad_norm": 0.2540144920349121, | |
| "learning_rate": 0.00019935085579907063, | |
| "loss": 1.281, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.025585522338579147, | |
| "grad_norm": 0.30044910311698914, | |
| "learning_rate": 0.00019933603513679605, | |
| "loss": 1.1689, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.025741531621131457, | |
| "grad_norm": 0.31799909472465515, | |
| "learning_rate": 0.00019932104775631846, | |
| "loss": 1.287, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.02589754090368377, | |
| "grad_norm": 0.290565550327301, | |
| "learning_rate": 0.0001993058936827916, | |
| "loss": 1.4751, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.02605355018623608, | |
| "grad_norm": 0.28967443108558655, | |
| "learning_rate": 0.00019929057294164893, | |
| "loss": 1.2459, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.026209559468788394, | |
| "grad_norm": 0.25141966342926025, | |
| "learning_rate": 0.0001992750855586036, | |
| "loss": 1.1215, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.026365568751340704, | |
| "grad_norm": 0.2819644808769226, | |
| "learning_rate": 0.00019925943155964856, | |
| "loss": 1.5238, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.026521578033893017, | |
| "grad_norm": 0.2336016446352005, | |
| "learning_rate": 0.00019924361097105623, | |
| "loss": 1.2218, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.026677587316445327, | |
| "grad_norm": 0.23773479461669922, | |
| "learning_rate": 0.00019922762381937878, | |
| "loss": 1.0842, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.02683359659899764, | |
| "grad_norm": 0.266222208738327, | |
| "learning_rate": 0.0001992114701314478, | |
| "loss": 1.2076, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.02698960588154995, | |
| "grad_norm": 0.29275181889533997, | |
| "learning_rate": 0.00019919514993437445, | |
| "loss": 1.3901, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.027145615164102264, | |
| "grad_norm": 0.2334383726119995, | |
| "learning_rate": 0.00019917866325554938, | |
| "loss": 1.2012, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.027301624446654574, | |
| "grad_norm": 0.293888121843338, | |
| "learning_rate": 0.00019916201012264254, | |
| "loss": 1.6131, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.027457633729206888, | |
| "grad_norm": 0.3042750954627991, | |
| "learning_rate": 0.0001991451905636033, | |
| "loss": 1.3144, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.027613643011759198, | |
| "grad_norm": 0.2652626633644104, | |
| "learning_rate": 0.00019912820460666044, | |
| "loss": 1.4368, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.02776965229431151, | |
| "grad_norm": 0.28741374611854553, | |
| "learning_rate": 0.00019911105228032186, | |
| "loss": 1.4643, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.027925661576863825, | |
| "grad_norm": 0.2808038890361786, | |
| "learning_rate": 0.00019909373361337476, | |
| "loss": 1.3013, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.028081670859416135, | |
| "grad_norm": 0.22930848598480225, | |
| "learning_rate": 0.0001990762486348855, | |
| "loss": 1.0587, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.02823768014196845, | |
| "grad_norm": 0.24289073050022125, | |
| "learning_rate": 0.00019905859737419956, | |
| "loss": 1.1174, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.02839368942452076, | |
| "grad_norm": 0.2626672685146332, | |
| "learning_rate": 0.00019904077986094152, | |
| "loss": 1.1746, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.028549698707073072, | |
| "grad_norm": 0.3174870014190674, | |
| "learning_rate": 0.00019902279612501493, | |
| "loss": 1.4464, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.028705707989625382, | |
| "grad_norm": 0.2851637303829193, | |
| "learning_rate": 0.0001990046461966024, | |
| "loss": 1.3527, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.028861717272177696, | |
| "grad_norm": 0.2576538622379303, | |
| "learning_rate": 0.00019898633010616542, | |
| "loss": 1.2546, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.029017726554730006, | |
| "grad_norm": 0.2922312319278717, | |
| "learning_rate": 0.0001989678478844443, | |
| "loss": 1.1445, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.02917373583728232, | |
| "grad_norm": 0.25312724709510803, | |
| "learning_rate": 0.00019894919956245824, | |
| "loss": 1.0533, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.02932974511983463, | |
| "grad_norm": 0.3193413019180298, | |
| "learning_rate": 0.00019893038517150525, | |
| "loss": 1.655, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.029485754402386943, | |
| "grad_norm": 0.26104092597961426, | |
| "learning_rate": 0.00019891140474316194, | |
| "loss": 1.5094, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.029641763684939253, | |
| "grad_norm": 0.2679871916770935, | |
| "learning_rate": 0.00019889225830928365, | |
| "loss": 1.3535, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.029797772967491566, | |
| "grad_norm": 0.2835332751274109, | |
| "learning_rate": 0.00019887294590200435, | |
| "loss": 1.647, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.029953782250043876, | |
| "grad_norm": 0.2309991866350174, | |
| "learning_rate": 0.00019885346755373656, | |
| "loss": 1.1869, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.03010979153259619, | |
| "grad_norm": 0.28801408410072327, | |
| "learning_rate": 0.00019883382329717128, | |
| "loss": 1.4037, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.0302658008151485, | |
| "grad_norm": 0.309851735830307, | |
| "learning_rate": 0.00019881401316527793, | |
| "loss": 1.2832, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.030421810097700813, | |
| "grad_norm": 0.27529048919677734, | |
| "learning_rate": 0.0001987940371913044, | |
| "loss": 1.5466, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.030577819380253127, | |
| "grad_norm": 0.25759854912757874, | |
| "learning_rate": 0.00019877389540877687, | |
| "loss": 1.2432, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.030733828662805437, | |
| "grad_norm": 0.27557173371315, | |
| "learning_rate": 0.0001987535878514998, | |
| "loss": 1.5681, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.03088983794535775, | |
| "grad_norm": 0.25760918855667114, | |
| "learning_rate": 0.0001987331145535559, | |
| "loss": 1.3067, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.03104584722791006, | |
| "grad_norm": 0.299180269241333, | |
| "learning_rate": 0.000198712475549306, | |
| "loss": 1.4642, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.031201856510462374, | |
| "grad_norm": 0.2398681640625, | |
| "learning_rate": 0.00019869167087338907, | |
| "loss": 1.0748, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.03135786579301469, | |
| "grad_norm": 0.2560211420059204, | |
| "learning_rate": 0.00019867070056072214, | |
| "loss": 1.2508, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.031513875075566994, | |
| "grad_norm": 0.25509408116340637, | |
| "learning_rate": 0.00019864956464650025, | |
| "loss": 1.4073, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.03166988435811931, | |
| "grad_norm": 0.27500587701797485, | |
| "learning_rate": 0.00019862826316619628, | |
| "loss": 1.3473, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.03182589364067162, | |
| "grad_norm": 0.2923906445503235, | |
| "learning_rate": 0.0001986067961555611, | |
| "loss": 1.4293, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.031981902923223934, | |
| "grad_norm": 0.24456267058849335, | |
| "learning_rate": 0.00019858516365062334, | |
| "loss": 1.2196, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.03213791220577624, | |
| "grad_norm": 0.3021962344646454, | |
| "learning_rate": 0.00019856336568768935, | |
| "loss": 1.5066, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.032293921488328554, | |
| "grad_norm": 0.2485729455947876, | |
| "learning_rate": 0.00019854140230334322, | |
| "loss": 1.2002, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.03244993077088087, | |
| "grad_norm": 0.26055216789245605, | |
| "learning_rate": 0.0001985192735344467, | |
| "loss": 1.3207, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.03260594005343318, | |
| "grad_norm": 0.2658592760562897, | |
| "learning_rate": 0.00019849697941813898, | |
| "loss": 0.9025, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.03276194933598549, | |
| "grad_norm": 0.30481112003326416, | |
| "learning_rate": 0.00019847451999183694, | |
| "loss": 1.5238, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.0329179586185378, | |
| "grad_norm": 0.28382736444473267, | |
| "learning_rate": 0.00019845189529323475, | |
| "loss": 1.3224, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.033073967901090115, | |
| "grad_norm": 0.2757686972618103, | |
| "learning_rate": 0.00019842910536030403, | |
| "loss": 1.3672, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.03322997718364243, | |
| "grad_norm": 0.2743508219718933, | |
| "learning_rate": 0.00019840615023129372, | |
| "loss": 1.3628, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.03338598646619474, | |
| "grad_norm": 0.26412197947502136, | |
| "learning_rate": 0.00019838302994472997, | |
| "loss": 1.141, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.03354199574874705, | |
| "grad_norm": 0.2859683632850647, | |
| "learning_rate": 0.0001983597445394162, | |
| "loss": 1.1566, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.03369800503129936, | |
| "grad_norm": 0.24881964921951294, | |
| "learning_rate": 0.00019833629405443284, | |
| "loss": 1.2038, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.033854014313851676, | |
| "grad_norm": 0.25597479939460754, | |
| "learning_rate": 0.0001983126785291375, | |
| "loss": 0.9913, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.03401002359640399, | |
| "grad_norm": 0.26771095395088196, | |
| "learning_rate": 0.00019828889800316466, | |
| "loss": 1.5417, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.034166032878956296, | |
| "grad_norm": 0.2678371071815491, | |
| "learning_rate": 0.00019826495251642578, | |
| "loss": 1.208, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.03432204216150861, | |
| "grad_norm": 0.2947763204574585, | |
| "learning_rate": 0.00019824084210910925, | |
| "loss": 1.3908, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.03447805144406092, | |
| "grad_norm": 0.2821643650531769, | |
| "learning_rate": 0.00019821656682168012, | |
| "loss": 1.6573, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.034634060726613236, | |
| "grad_norm": 0.24507346749305725, | |
| "learning_rate": 0.00019819212669488026, | |
| "loss": 1.0647, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.03479007000916554, | |
| "grad_norm": 0.2718466520309448, | |
| "learning_rate": 0.00019816752176972813, | |
| "loss": 1.3013, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.034946079291717856, | |
| "grad_norm": 0.2902746796607971, | |
| "learning_rate": 0.0001981427520875188, | |
| "loss": 1.2212, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.03510208857427017, | |
| "grad_norm": 0.25822389125823975, | |
| "learning_rate": 0.0001981178176898239, | |
| "loss": 1.4543, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.03525809785682248, | |
| "grad_norm": 0.3506292700767517, | |
| "learning_rate": 0.00019809271861849145, | |
| "loss": 1.8549, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.03541410713937479, | |
| "grad_norm": 0.2610777020454407, | |
| "learning_rate": 0.00019806745491564586, | |
| "loss": 1.3161, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.0355701164219271, | |
| "grad_norm": 0.29803603887557983, | |
| "learning_rate": 0.0001980420266236878, | |
| "loss": 1.2983, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.03572612570447942, | |
| "grad_norm": 0.24572676420211792, | |
| "learning_rate": 0.0001980164337852943, | |
| "loss": 1.291, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.03588213498703173, | |
| "grad_norm": 0.25573092699050903, | |
| "learning_rate": 0.00019799067644341844, | |
| "loss": 1.3207, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.036038144269584044, | |
| "grad_norm": 0.28766271471977234, | |
| "learning_rate": 0.00019796475464128942, | |
| "loss": 1.4527, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.03619415355213635, | |
| "grad_norm": 0.2636454701423645, | |
| "learning_rate": 0.00019793866842241243, | |
| "loss": 1.3899, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.036350162834688664, | |
| "grad_norm": 0.3094368577003479, | |
| "learning_rate": 0.00019791241783056874, | |
| "loss": 1.2935, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.03650617211724098, | |
| "grad_norm": 0.2588469088077545, | |
| "learning_rate": 0.00019788600290981525, | |
| "loss": 1.2457, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.03666218139979329, | |
| "grad_norm": 0.26457706093788147, | |
| "learning_rate": 0.0001978594237044849, | |
| "loss": 1.1753, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.0368181906823456, | |
| "grad_norm": 0.2559141516685486, | |
| "learning_rate": 0.0001978326802591862, | |
| "loss": 1.2004, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.03697419996489791, | |
| "grad_norm": 0.2815738320350647, | |
| "learning_rate": 0.00019780577261880336, | |
| "loss": 1.3706, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.037130209247450224, | |
| "grad_norm": 0.2584588825702667, | |
| "learning_rate": 0.0001977787008284962, | |
| "loss": 1.4192, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.03728621853000254, | |
| "grad_norm": 0.290865421295166, | |
| "learning_rate": 0.00019775146493369994, | |
| "loss": 1.2308, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.037442227812554844, | |
| "grad_norm": 0.2788088023662567, | |
| "learning_rate": 0.0001977240649801253, | |
| "loss": 1.2095, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.03759823709510716, | |
| "grad_norm": 0.28903988003730774, | |
| "learning_rate": 0.00019769650101375837, | |
| "loss": 1.5138, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.03775424637765947, | |
| "grad_norm": 0.29985305666923523, | |
| "learning_rate": 0.00019766877308086036, | |
| "loss": 1.4594, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.037910255660211785, | |
| "grad_norm": 0.3033303916454315, | |
| "learning_rate": 0.00019764088122796783, | |
| "loss": 1.6108, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.03806626494276409, | |
| "grad_norm": 0.2854767143726349, | |
| "learning_rate": 0.0001976128255018924, | |
| "loss": 1.377, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.038222274225316405, | |
| "grad_norm": 0.30725011229515076, | |
| "learning_rate": 0.00019758460594972068, | |
| "loss": 1.2651, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.03837828350786872, | |
| "grad_norm": 0.28218191862106323, | |
| "learning_rate": 0.00019755622261881427, | |
| "loss": 1.4354, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.03853429279042103, | |
| "grad_norm": 0.2794611155986786, | |
| "learning_rate": 0.00019752767555680968, | |
| "loss": 1.4666, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.03869030207297334, | |
| "grad_norm": 0.2824796736240387, | |
| "learning_rate": 0.00019749896481161808, | |
| "loss": 1.3645, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.03884631135552565, | |
| "grad_norm": 0.26165372133255005, | |
| "learning_rate": 0.00019747009043142555, | |
| "loss": 1.3445, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.039002320638077966, | |
| "grad_norm": 0.29985979199409485, | |
| "learning_rate": 0.00019744105246469263, | |
| "loss": 1.4558, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.03915832992063028, | |
| "grad_norm": 0.25439903140068054, | |
| "learning_rate": 0.00019741185096015448, | |
| "loss": 1.1075, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.03931433920318259, | |
| "grad_norm": 0.2533755898475647, | |
| "learning_rate": 0.00019738248596682078, | |
| "loss": 1.0891, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.0394703484857349, | |
| "grad_norm": 0.27487608790397644, | |
| "learning_rate": 0.0001973529575339755, | |
| "loss": 1.3128, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.03962635776828721, | |
| "grad_norm": 0.27824172377586365, | |
| "learning_rate": 0.00019732326571117703, | |
| "loss": 1.4045, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.039782367050839526, | |
| "grad_norm": 0.27959418296813965, | |
| "learning_rate": 0.00019729341054825782, | |
| "loss": 1.2169, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.03993837633339184, | |
| "grad_norm": 0.3103275001049042, | |
| "learning_rate": 0.00019726339209532462, | |
| "loss": 1.3043, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.040094385615944146, | |
| "grad_norm": 0.2712806463241577, | |
| "learning_rate": 0.00019723321040275815, | |
| "loss": 1.1747, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.04025039489849646, | |
| "grad_norm": 0.2961602210998535, | |
| "learning_rate": 0.0001972028655212131, | |
| "loss": 1.5744, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.04040640418104877, | |
| "grad_norm": 0.2686194181442261, | |
| "learning_rate": 0.00019717235750161806, | |
| "loss": 1.2442, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.04056241346360109, | |
| "grad_norm": 0.2742723822593689, | |
| "learning_rate": 0.00019714168639517544, | |
| "loss": 1.3225, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.04071842274615339, | |
| "grad_norm": 0.28742754459381104, | |
| "learning_rate": 0.00019711085225336132, | |
| "loss": 1.3711, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.04087443202870571, | |
| "grad_norm": 0.30374589562416077, | |
| "learning_rate": 0.00019707985512792543, | |
| "loss": 1.215, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.04103044131125802, | |
| "grad_norm": 0.2738686800003052, | |
| "learning_rate": 0.00019704869507089105, | |
| "loss": 1.4628, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.041186450593810334, | |
| "grad_norm": 0.2695278823375702, | |
| "learning_rate": 0.0001970173721345549, | |
| "loss": 1.4632, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.04134245987636264, | |
| "grad_norm": 0.2954547107219696, | |
| "learning_rate": 0.00019698588637148703, | |
| "loss": 1.2785, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.041498469158914954, | |
| "grad_norm": 0.2756305932998657, | |
| "learning_rate": 0.00019695423783453088, | |
| "loss": 1.4258, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.04165447844146727, | |
| "grad_norm": 0.2642769515514374, | |
| "learning_rate": 0.00019692242657680286, | |
| "loss": 1.3034, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.04181048772401958, | |
| "grad_norm": 0.2760365307331085, | |
| "learning_rate": 0.00019689045265169273, | |
| "loss": 1.5845, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.041966497006571894, | |
| "grad_norm": 0.23845522105693817, | |
| "learning_rate": 0.0001968583161128631, | |
| "loss": 1.113, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.0421225062891242, | |
| "grad_norm": 0.2855961322784424, | |
| "learning_rate": 0.0001968260170142496, | |
| "loss": 1.4019, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.042278515571676514, | |
| "grad_norm": 0.26462671160697937, | |
| "learning_rate": 0.00019679355541006054, | |
| "loss": 1.2425, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.04243452485422883, | |
| "grad_norm": 0.28468820452690125, | |
| "learning_rate": 0.00019676093135477713, | |
| "loss": 1.6525, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.04259053413678114, | |
| "grad_norm": 0.3233076333999634, | |
| "learning_rate": 0.0001967281449031531, | |
| "loss": 1.2168, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.04274654341933345, | |
| "grad_norm": 0.2688952684402466, | |
| "learning_rate": 0.00019669519611021486, | |
| "loss": 1.3948, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.04290255270188576, | |
| "grad_norm": 0.25911059975624084, | |
| "learning_rate": 0.00019666208503126112, | |
| "loss": 1.2875, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.043058561984438075, | |
| "grad_norm": 0.2789272964000702, | |
| "learning_rate": 0.00019662881172186313, | |
| "loss": 1.257, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.04321457126699039, | |
| "grad_norm": 0.26854726672172546, | |
| "learning_rate": 0.00019659537623786428, | |
| "loss": 1.4554, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.043370580549542695, | |
| "grad_norm": 0.31813284754753113, | |
| "learning_rate": 0.00019656177863538026, | |
| "loss": 1.667, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.04352658983209501, | |
| "grad_norm": 0.2801772356033325, | |
| "learning_rate": 0.00019652801897079869, | |
| "loss": 1.4555, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.04368259911464732, | |
| "grad_norm": 0.30256757140159607, | |
| "learning_rate": 0.00019649409730077935, | |
| "loss": 1.2628, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.043838608397199635, | |
| "grad_norm": 0.2807087302207947, | |
| "learning_rate": 0.00019646001368225382, | |
| "loss": 1.5143, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.04399461767975194, | |
| "grad_norm": 0.27217531204223633, | |
| "learning_rate": 0.0001964257681724255, | |
| "loss": 1.5372, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.044150626962304255, | |
| "grad_norm": 0.2996511459350586, | |
| "learning_rate": 0.00019639136082876953, | |
| "loss": 1.2692, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.04430663624485657, | |
| "grad_norm": 0.263231098651886, | |
| "learning_rate": 0.00019635679170903258, | |
| "loss": 1.2328, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.04446264552740888, | |
| "grad_norm": 0.3060413897037506, | |
| "learning_rate": 0.00019632206087123296, | |
| "loss": 1.5173, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.044618654809961196, | |
| "grad_norm": 0.25136467814445496, | |
| "learning_rate": 0.00019628716837366027, | |
| "loss": 1.1781, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.0447746640925135, | |
| "grad_norm": 0.27105534076690674, | |
| "learning_rate": 0.00019625211427487548, | |
| "loss": 1.4542, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.044930673375065816, | |
| "grad_norm": 0.27552956342697144, | |
| "learning_rate": 0.00019621689863371083, | |
| "loss": 1.3352, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.04508668265761813, | |
| "grad_norm": 0.26462072134017944, | |
| "learning_rate": 0.00019618152150926955, | |
| "loss": 1.2531, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.04524269194017044, | |
| "grad_norm": 0.2736480236053467, | |
| "learning_rate": 0.000196145982960926, | |
| "loss": 1.402, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.04539870122272275, | |
| "grad_norm": 0.2739974856376648, | |
| "learning_rate": 0.00019611028304832546, | |
| "loss": 1.4881, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.04555471050527506, | |
| "grad_norm": 0.25353673100471497, | |
| "learning_rate": 0.000196074421831384, | |
| "loss": 1.3935, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.04571071978782738, | |
| "grad_norm": 0.2595098614692688, | |
| "learning_rate": 0.00019603839937028838, | |
| "loss": 1.3306, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.04586672907037969, | |
| "grad_norm": 0.27779051661491394, | |
| "learning_rate": 0.00019600221572549606, | |
| "loss": 1.5111, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.046022738352932, | |
| "grad_norm": 0.26458942890167236, | |
| "learning_rate": 0.00019596587095773495, | |
| "loss": 1.1354, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.04617874763548431, | |
| "grad_norm": 0.3711000084877014, | |
| "learning_rate": 0.00019592936512800342, | |
| "loss": 1.387, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.046334756918036624, | |
| "grad_norm": 0.26172423362731934, | |
| "learning_rate": 0.00019589269829757008, | |
| "loss": 1.1995, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.04649076620058894, | |
| "grad_norm": 0.30684447288513184, | |
| "learning_rate": 0.00019585587052797389, | |
| "loss": 1.2853, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.046646775483141244, | |
| "grad_norm": 0.27383920550346375, | |
| "learning_rate": 0.00019581888188102375, | |
| "loss": 1.1397, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.04680278476569356, | |
| "grad_norm": 0.28926682472229004, | |
| "learning_rate": 0.00019578173241879872, | |
| "loss": 1.2977, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.04695879404824587, | |
| "grad_norm": 0.2573678195476532, | |
| "learning_rate": 0.00019574442220364767, | |
| "loss": 1.315, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.047114803330798184, | |
| "grad_norm": 0.286785751581192, | |
| "learning_rate": 0.00019570695129818926, | |
| "loss": 1.196, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.0472708126133505, | |
| "grad_norm": 0.26392433047294617, | |
| "learning_rate": 0.0001956693197653119, | |
| "loss": 1.067, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.047426821895902804, | |
| "grad_norm": 0.29351645708084106, | |
| "learning_rate": 0.00019563152766817354, | |
| "loss": 1.2977, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.04758283117845512, | |
| "grad_norm": 0.3556276857852936, | |
| "learning_rate": 0.00019559357507020162, | |
| "loss": 1.1268, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.04773884046100743, | |
| "grad_norm": 0.3044413924217224, | |
| "learning_rate": 0.00019555546203509297, | |
| "loss": 1.3528, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.047894849743559745, | |
| "grad_norm": 0.25455671548843384, | |
| "learning_rate": 0.00019551718862681364, | |
| "loss": 1.2099, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.04805085902611205, | |
| "grad_norm": 0.2863021492958069, | |
| "learning_rate": 0.00019547875490959885, | |
| "loss": 1.514, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.048206868308664365, | |
| "grad_norm": 0.2713131010532379, | |
| "learning_rate": 0.00019544016094795295, | |
| "loss": 1.2479, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.04836287759121668, | |
| "grad_norm": 0.28673309087753296, | |
| "learning_rate": 0.00019540140680664913, | |
| "loss": 1.4822, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.04851888687376899, | |
| "grad_norm": 0.28506314754486084, | |
| "learning_rate": 0.00019536249255072948, | |
| "loss": 1.1714, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.0486748961563213, | |
| "grad_norm": 0.2814370393753052, | |
| "learning_rate": 0.00019532341824550479, | |
| "loss": 1.3045, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.04883090543887361, | |
| "grad_norm": 0.2505611181259155, | |
| "learning_rate": 0.0001952841839565544, | |
| "loss": 1.1565, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.048986914721425925, | |
| "grad_norm": 0.27159830927848816, | |
| "learning_rate": 0.0001952447897497263, | |
| "loss": 1.0939, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.04914292400397824, | |
| "grad_norm": 0.27552008628845215, | |
| "learning_rate": 0.00019520523569113677, | |
| "loss": 1.4382, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.049298933286530545, | |
| "grad_norm": 0.2567708492279053, | |
| "learning_rate": 0.00019516552184717037, | |
| "loss": 1.2241, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.04945494256908286, | |
| "grad_norm": 0.27663713693618774, | |
| "learning_rate": 0.00019512564828447988, | |
| "loss": 1.2449, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.04961095185163517, | |
| "grad_norm": 0.2683660089969635, | |
| "learning_rate": 0.0001950856150699861, | |
| "loss": 1.1652, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.049766961134187486, | |
| "grad_norm": 0.25226572155952454, | |
| "learning_rate": 0.0001950454222708778, | |
| "loss": 1.1307, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.0499229704167398, | |
| "grad_norm": 0.23380513489246368, | |
| "learning_rate": 0.0001950050699546116, | |
| "loss": 1.1257, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.050078979699292106, | |
| "grad_norm": 0.2385280281305313, | |
| "learning_rate": 0.0001949645581889118, | |
| "loss": 0.9917, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.05023498898184442, | |
| "grad_norm": 0.23746567964553833, | |
| "learning_rate": 0.00019492388704177036, | |
| "loss": 1.1364, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.05039099826439673, | |
| "grad_norm": 0.27820831537246704, | |
| "learning_rate": 0.00019488305658144667, | |
| "loss": 1.3707, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.050547007546949047, | |
| "grad_norm": 0.2663419544696808, | |
| "learning_rate": 0.00019484206687646753, | |
| "loss": 1.3662, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.05070301682950135, | |
| "grad_norm": 0.27196773886680603, | |
| "learning_rate": 0.00019480091799562704, | |
| "loss": 1.2766, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.05085902611205367, | |
| "grad_norm": 0.296779602766037, | |
| "learning_rate": 0.00019475961000798645, | |
| "loss": 1.5789, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.05101503539460598, | |
| "grad_norm": 0.3267677128314972, | |
| "learning_rate": 0.0001947181429828739, | |
| "loss": 1.2782, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.051171044677158294, | |
| "grad_norm": 0.2852894067764282, | |
| "learning_rate": 0.00019467651698988462, | |
| "loss": 1.1466, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.0513270539597106, | |
| "grad_norm": 0.2959722876548767, | |
| "learning_rate": 0.0001946347320988806, | |
| "loss": 1.1929, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.051483063242262914, | |
| "grad_norm": 0.25998443365097046, | |
| "learning_rate": 0.00019459278837999046, | |
| "loss": 1.4104, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.05163907252481523, | |
| "grad_norm": 0.27319809794425964, | |
| "learning_rate": 0.00019455068590360942, | |
| "loss": 1.417, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.05179508180736754, | |
| "grad_norm": 0.22395959496498108, | |
| "learning_rate": 0.00019450842474039913, | |
| "loss": 1.2159, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.05195109108991985, | |
| "grad_norm": 0.24947980046272278, | |
| "learning_rate": 0.00019446600496128758, | |
| "loss": 1.1063, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.05210710037247216, | |
| "grad_norm": 0.235429584980011, | |
| "learning_rate": 0.00019442342663746902, | |
| "loss": 1.2234, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.052263109655024474, | |
| "grad_norm": 0.27443963289260864, | |
| "learning_rate": 0.00019438068984040365, | |
| "loss": 1.2038, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.05241911893757679, | |
| "grad_norm": 0.26688772439956665, | |
| "learning_rate": 0.00019433779464181778, | |
| "loss": 1.2956, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.052575128220129094, | |
| "grad_norm": 0.23804551362991333, | |
| "learning_rate": 0.00019429474111370352, | |
| "loss": 0.9525, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.05273113750268141, | |
| "grad_norm": 0.262890487909317, | |
| "learning_rate": 0.0001942515293283187, | |
| "loss": 1.2713, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.05288714678523372, | |
| "grad_norm": 0.29796820878982544, | |
| "learning_rate": 0.00019420815935818672, | |
| "loss": 1.5058, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.053043156067786035, | |
| "grad_norm": 0.275143563747406, | |
| "learning_rate": 0.00019416463127609656, | |
| "loss": 1.2604, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.05319916535033835, | |
| "grad_norm": 0.27801284193992615, | |
| "learning_rate": 0.00019412094515510248, | |
| "loss": 1.2588, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.053355174632890655, | |
| "grad_norm": 0.2604374885559082, | |
| "learning_rate": 0.00019407710106852404, | |
| "loss": 1.1432, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.05351118391544297, | |
| "grad_norm": 0.2863079011440277, | |
| "learning_rate": 0.00019403309908994586, | |
| "loss": 1.4854, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.05366719319799528, | |
| "grad_norm": 0.2515758275985718, | |
| "learning_rate": 0.00019398893929321761, | |
| "loss": 1.1682, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.053823202480547595, | |
| "grad_norm": 0.27037686109542847, | |
| "learning_rate": 0.00019394462175245381, | |
| "loss": 1.3679, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.0539792117630999, | |
| "grad_norm": 0.2368054836988449, | |
| "learning_rate": 0.00019390014654203369, | |
| "loss": 1.1406, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.054135221045652215, | |
| "grad_norm": 0.27759966254234314, | |
| "learning_rate": 0.0001938555137366011, | |
| "loss": 1.1669, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.05429123032820453, | |
| "grad_norm": 0.3004835546016693, | |
| "learning_rate": 0.00019381072341106452, | |
| "loss": 1.4811, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.05444723961075684, | |
| "grad_norm": 0.30656251311302185, | |
| "learning_rate": 0.0001937657756405966, | |
| "loss": 1.515, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.05460324889330915, | |
| "grad_norm": 0.31442925333976746, | |
| "learning_rate": 0.00019372067050063438, | |
| "loss": 1.4848, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.05475925817586146, | |
| "grad_norm": 0.2230207473039627, | |
| "learning_rate": 0.00019367540806687893, | |
| "loss": 0.9535, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.054915267458413776, | |
| "grad_norm": 0.2552795708179474, | |
| "learning_rate": 0.0001936299884152954, | |
| "loss": 1.2254, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.05507127674096609, | |
| "grad_norm": 0.29775241017341614, | |
| "learning_rate": 0.0001935844116221127, | |
| "loss": 1.3821, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.055227286023518396, | |
| "grad_norm": 0.24480530619621277, | |
| "learning_rate": 0.00019353867776382354, | |
| "loss": 1.1073, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.05538329530607071, | |
| "grad_norm": 0.2612270414829254, | |
| "learning_rate": 0.00019349278691718427, | |
| "loss": 1.3114, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.05553930458862302, | |
| "grad_norm": 0.307085245847702, | |
| "learning_rate": 0.0001934467391592146, | |
| "loss": 1.3602, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.055695313871175336, | |
| "grad_norm": 0.2688599228858948, | |
| "learning_rate": 0.00019340053456719768, | |
| "loss": 1.4347, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.05585132315372765, | |
| "grad_norm": 0.25372791290283203, | |
| "learning_rate": 0.00019335417321867987, | |
| "loss": 1.3468, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.05600733243627996, | |
| "grad_norm": 0.2706502377986908, | |
| "learning_rate": 0.0001933076551914706, | |
| "loss": 1.4489, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.05616334171883227, | |
| "grad_norm": 0.22997525334358215, | |
| "learning_rate": 0.00019326098056364222, | |
| "loss": 1.1305, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.056319351001384584, | |
| "grad_norm": 0.30573347210884094, | |
| "learning_rate": 0.00019321414941353003, | |
| "loss": 1.4231, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.0564753602839369, | |
| "grad_norm": 0.30873847007751465, | |
| "learning_rate": 0.00019316716181973188, | |
| "loss": 1.3478, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.056631369566489204, | |
| "grad_norm": 0.2514902651309967, | |
| "learning_rate": 0.00019312001786110828, | |
| "loss": 1.2094, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.05678737884904152, | |
| "grad_norm": 0.26067742705345154, | |
| "learning_rate": 0.00019307271761678213, | |
| "loss": 1.5841, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.05694338813159383, | |
| "grad_norm": 0.23508694767951965, | |
| "learning_rate": 0.00019302526116613864, | |
| "loss": 1.103, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.057099397414146144, | |
| "grad_norm": 0.24878567457199097, | |
| "learning_rate": 0.00019297764858882514, | |
| "loss": 1.0968, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.05725540669669845, | |
| "grad_norm": 0.23707476258277893, | |
| "learning_rate": 0.00019292987996475113, | |
| "loss": 1.0831, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.057411415979250764, | |
| "grad_norm": 0.2691617012023926, | |
| "learning_rate": 0.0001928819553740878, | |
| "loss": 1.2254, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.05756742526180308, | |
| "grad_norm": 0.26831138134002686, | |
| "learning_rate": 0.00019283387489726827, | |
| "loss": 1.3084, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.05772343454435539, | |
| "grad_norm": 0.281770259141922, | |
| "learning_rate": 0.00019278563861498723, | |
| "loss": 1.3377, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.0578794438269077, | |
| "grad_norm": 0.2634589970111847, | |
| "learning_rate": 0.00019273724660820088, | |
| "loss": 1.2648, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.05803545310946001, | |
| "grad_norm": 0.27592259645462036, | |
| "learning_rate": 0.00019268869895812672, | |
| "loss": 1.2751, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.058191462392012325, | |
| "grad_norm": 0.23107245564460754, | |
| "learning_rate": 0.00019263999574624355, | |
| "loss": 1.2651, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.05834747167456464, | |
| "grad_norm": 0.2582552134990692, | |
| "learning_rate": 0.0001925911370542912, | |
| "loss": 1.4914, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.05850348095711695, | |
| "grad_norm": 0.27152058482170105, | |
| "learning_rate": 0.00019254212296427044, | |
| "loss": 1.2227, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.05865949023966926, | |
| "grad_norm": 0.23554329574108124, | |
| "learning_rate": 0.00019249295355844285, | |
| "loss": 1.4113, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.05881549952222157, | |
| "grad_norm": 0.2793971300125122, | |
| "learning_rate": 0.00019244362891933077, | |
| "loss": 1.3325, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.058971508804773885, | |
| "grad_norm": 0.2800885736942291, | |
| "learning_rate": 0.00019239414912971696, | |
| "loss": 1.358, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.0591275180873262, | |
| "grad_norm": 0.27139201760292053, | |
| "learning_rate": 0.0001923445142726446, | |
| "loss": 1.2269, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.059283527369878505, | |
| "grad_norm": 0.276579886674881, | |
| "learning_rate": 0.0001922947244314172, | |
| "loss": 1.1521, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.05943953665243082, | |
| "grad_norm": 0.28917452692985535, | |
| "learning_rate": 0.0001922447796895982, | |
| "loss": 1.2803, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.05959554593498313, | |
| "grad_norm": 0.28668197989463806, | |
| "learning_rate": 0.00019219468013101124, | |
| "loss": 1.4025, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.059751555217535446, | |
| "grad_norm": 0.2973851263523102, | |
| "learning_rate": 0.00019214442583973966, | |
| "loss": 1.5472, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.05990756450008775, | |
| "grad_norm": 0.25934460759162903, | |
| "learning_rate": 0.00019209401690012653, | |
| "loss": 1.2496, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.060063573782640066, | |
| "grad_norm": 0.22885724902153015, | |
| "learning_rate": 0.00019204345339677442, | |
| "loss": 1.2088, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.06021958306519238, | |
| "grad_norm": 0.28346025943756104, | |
| "learning_rate": 0.00019199273541454538, | |
| "loss": 1.1561, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.06037559234774469, | |
| "grad_norm": 0.2574789822101593, | |
| "learning_rate": 0.00019194186303856067, | |
| "loss": 1.3209, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.060531601630297, | |
| "grad_norm": 0.26535728573799133, | |
| "learning_rate": 0.00019189083635420075, | |
| "loss": 1.3022, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.06068761091284931, | |
| "grad_norm": 0.2844642698764801, | |
| "learning_rate": 0.00019183965544710495, | |
| "loss": 1.3881, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.060843620195401626, | |
| "grad_norm": 0.24562187492847443, | |
| "learning_rate": 0.00019178832040317155, | |
| "loss": 1.159, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.06099962947795394, | |
| "grad_norm": 0.25778669118881226, | |
| "learning_rate": 0.0001917368313085574, | |
| "loss": 1.5154, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.061155638760506253, | |
| "grad_norm": 0.22877171635627747, | |
| "learning_rate": 0.00019168518824967795, | |
| "loss": 1.201, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.06131164804305856, | |
| "grad_norm": 0.2764502465724945, | |
| "learning_rate": 0.00019163339131320718, | |
| "loss": 1.4165, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.061467657325610874, | |
| "grad_norm": 0.23493847250938416, | |
| "learning_rate": 0.00019158144058607708, | |
| "loss": 1.1334, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.06162366660816319, | |
| "grad_norm": 0.2605098783969879, | |
| "learning_rate": 0.00019152933615547798, | |
| "loss": 1.1613, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.0617796758907155, | |
| "grad_norm": 0.23720701038837433, | |
| "learning_rate": 0.000191477078108858, | |
| "loss": 1.1966, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.06193568517326781, | |
| "grad_norm": 0.27043676376342773, | |
| "learning_rate": 0.00019142466653392318, | |
| "loss": 1.2793, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.06209169445582012, | |
| "grad_norm": 0.27630025148391724, | |
| "learning_rate": 0.0001913721015186372, | |
| "loss": 1.3858, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.062247703738372434, | |
| "grad_norm": 0.29454129934310913, | |
| "learning_rate": 0.0001913193831512213, | |
| "loss": 1.5234, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.06240371302092475, | |
| "grad_norm": 0.26943233609199524, | |
| "learning_rate": 0.00019126651152015403, | |
| "loss": 1.3181, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.06255972230347706, | |
| "grad_norm": 0.28831520676612854, | |
| "learning_rate": 0.0001912134867141712, | |
| "loss": 1.46, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.06271573158602937, | |
| "grad_norm": 0.26342567801475525, | |
| "learning_rate": 0.0001911603088222657, | |
| "loss": 1.4073, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.06287174086858167, | |
| "grad_norm": 0.2623300552368164, | |
| "learning_rate": 0.0001911069779336873, | |
| "loss": 1.3473, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.06302775015113399, | |
| "grad_norm": 0.25125861167907715, | |
| "learning_rate": 0.00019105349413794272, | |
| "loss": 1.0346, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.0631837594336863, | |
| "grad_norm": 0.30890092253685, | |
| "learning_rate": 0.00019099985752479506, | |
| "loss": 1.5751, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.06333976871623861, | |
| "grad_norm": 0.31404733657836914, | |
| "learning_rate": 0.00019094606818426403, | |
| "loss": 1.5458, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.06349577799879093, | |
| "grad_norm": 0.2684463858604431, | |
| "learning_rate": 0.00019089212620662568, | |
| "loss": 1.2342, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.06365178728134324, | |
| "grad_norm": 0.2748461365699768, | |
| "learning_rate": 0.00019083803168241223, | |
| "loss": 1.3353, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.06380779656389556, | |
| "grad_norm": 0.3061840832233429, | |
| "learning_rate": 0.00019078378470241183, | |
| "loss": 1.3197, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.06396380584644787, | |
| "grad_norm": 0.25601011514663696, | |
| "learning_rate": 0.00019072938535766865, | |
| "loss": 1.3904, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.06411981512900018, | |
| "grad_norm": 0.2844060957431793, | |
| "learning_rate": 0.00019067483373948243, | |
| "loss": 1.42, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.06427582441155248, | |
| "grad_norm": 0.2969295382499695, | |
| "learning_rate": 0.00019062012993940859, | |
| "loss": 1.4255, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.0644318336941048, | |
| "grad_norm": 0.2655050456523895, | |
| "learning_rate": 0.00019056527404925789, | |
| "loss": 1.1618, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.06458784297665711, | |
| "grad_norm": 0.2571544349193573, | |
| "learning_rate": 0.00019051026616109638, | |
| "loss": 1.2064, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.06474385225920942, | |
| "grad_norm": 0.29847028851509094, | |
| "learning_rate": 0.0001904551063672452, | |
| "loss": 1.2847, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.06489986154176174, | |
| "grad_norm": 0.24265627562999725, | |
| "learning_rate": 0.00019039979476028043, | |
| "loss": 1.2745, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.06505587082431405, | |
| "grad_norm": 0.24038730561733246, | |
| "learning_rate": 0.000190344331433033, | |
| "loss": 1.2761, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.06521188010686636, | |
| "grad_norm": 0.26194193959236145, | |
| "learning_rate": 0.00019028871647858834, | |
| "loss": 1.5021, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.06536788938941868, | |
| "grad_norm": 0.2636980712413788, | |
| "learning_rate": 0.00019023294999028653, | |
| "loss": 1.5029, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.06552389867197098, | |
| "grad_norm": 0.26995277404785156, | |
| "learning_rate": 0.00019017703206172185, | |
| "loss": 1.3068, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.06567990795452329, | |
| "grad_norm": 0.26835623383522034, | |
| "learning_rate": 0.0001901209627867428, | |
| "loss": 1.2868, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.0658359172370756, | |
| "grad_norm": 0.24785400927066803, | |
| "learning_rate": 0.0001900647422594519, | |
| "loss": 1.1875, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.06599192651962792, | |
| "grad_norm": 0.3184250593185425, | |
| "learning_rate": 0.0001900083705742054, | |
| "loss": 1.3802, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.06614793580218023, | |
| "grad_norm": 0.2850029766559601, | |
| "learning_rate": 0.00018995184782561345, | |
| "loss": 1.3043, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.06630394508473254, | |
| "grad_norm": 0.2940841317176819, | |
| "learning_rate": 0.00018989517410853955, | |
| "loss": 1.287, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.06645995436728486, | |
| "grad_norm": 0.2668844163417816, | |
| "learning_rate": 0.0001898383495181007, | |
| "loss": 1.3723, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.06661596364983717, | |
| "grad_norm": 0.2814147472381592, | |
| "learning_rate": 0.00018978137414966698, | |
| "loss": 1.2339, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.06677197293238948, | |
| "grad_norm": 0.3722403049468994, | |
| "learning_rate": 0.0001897242480988617, | |
| "loss": 1.2755, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.06692798221494178, | |
| "grad_norm": 0.2689428925514221, | |
| "learning_rate": 0.00018966697146156092, | |
| "loss": 1.4238, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.0670839914974941, | |
| "grad_norm": 0.29616808891296387, | |
| "learning_rate": 0.00018960954433389345, | |
| "loss": 1.3167, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.06724000078004641, | |
| "grad_norm": 0.2477925419807434, | |
| "learning_rate": 0.0001895519668122408, | |
| "loss": 1.1773, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.06739601006259872, | |
| "grad_norm": 0.23961544036865234, | |
| "learning_rate": 0.0001894942389932367, | |
| "loss": 1.1387, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.06755201934515104, | |
| "grad_norm": 0.26128751039505005, | |
| "learning_rate": 0.00018943636097376726, | |
| "loss": 1.0468, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.06770802862770335, | |
| "grad_norm": 0.33279022574424744, | |
| "learning_rate": 0.00018937833285097066, | |
| "loss": 1.8791, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.06786403791025566, | |
| "grad_norm": 0.2876769006252289, | |
| "learning_rate": 0.00018932015472223693, | |
| "loss": 1.3633, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.06802004719280798, | |
| "grad_norm": 0.24108922481536865, | |
| "learning_rate": 0.00018926182668520792, | |
| "loss": 1.2012, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.06817605647536028, | |
| "grad_norm": 0.29062169790267944, | |
| "learning_rate": 0.0001892033488377771, | |
| "loss": 1.3658, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.06833206575791259, | |
| "grad_norm": 0.26536259055137634, | |
| "learning_rate": 0.0001891447212780893, | |
| "loss": 1.2464, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.0684880750404649, | |
| "grad_norm": 0.2940811514854431, | |
| "learning_rate": 0.0001890859441045407, | |
| "loss": 1.4609, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.06864408432301722, | |
| "grad_norm": 0.27625903487205505, | |
| "learning_rate": 0.0001890270174157784, | |
| "loss": 1.4098, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.06880009360556953, | |
| "grad_norm": 0.2586573362350464, | |
| "learning_rate": 0.00018896794131070073, | |
| "loss": 1.3857, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.06895610288812185, | |
| "grad_norm": 0.28287774324417114, | |
| "learning_rate": 0.0001889087158884565, | |
| "loss": 1.2967, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.06911211217067416, | |
| "grad_norm": 0.2692122459411621, | |
| "learning_rate": 0.00018884934124844532, | |
| "loss": 1.5216, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.06926812145322647, | |
| "grad_norm": 0.3004090189933777, | |
| "learning_rate": 0.00018878981749031716, | |
| "loss": 1.1913, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.06942413073577879, | |
| "grad_norm": 0.253542423248291, | |
| "learning_rate": 0.00018873014471397224, | |
| "loss": 1.1299, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.06958014001833109, | |
| "grad_norm": 0.3034575283527374, | |
| "learning_rate": 0.00018867032301956088, | |
| "loss": 1.3577, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.0697361493008834, | |
| "grad_norm": 0.31302767992019653, | |
| "learning_rate": 0.00018861035250748343, | |
| "loss": 1.6029, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.06989215858343571, | |
| "grad_norm": 0.26993393898010254, | |
| "learning_rate": 0.00018855023327838983, | |
| "loss": 1.2035, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.07004816786598803, | |
| "grad_norm": 0.27148422598838806, | |
| "learning_rate": 0.00018848996543317982, | |
| "loss": 1.5843, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.07020417714854034, | |
| "grad_norm": 0.2631765305995941, | |
| "learning_rate": 0.00018842954907300236, | |
| "loss": 1.2641, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.07036018643109265, | |
| "grad_norm": 0.2621013820171356, | |
| "learning_rate": 0.00018836898429925585, | |
| "loss": 1.2167, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.07051619571364497, | |
| "grad_norm": 0.25064215064048767, | |
| "learning_rate": 0.0001883082712135877, | |
| "loss": 1.2631, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.07067220499619728, | |
| "grad_norm": 0.2558056712150574, | |
| "learning_rate": 0.00018824740991789415, | |
| "loss": 0.9964, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.07082821427874958, | |
| "grad_norm": 0.2675093412399292, | |
| "learning_rate": 0.00018818640051432035, | |
| "loss": 1.4953, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.07098422356130189, | |
| "grad_norm": 0.2550821006298065, | |
| "learning_rate": 0.0001881252431052599, | |
| "loss": 1.1283, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.0711402328438542, | |
| "grad_norm": 0.24893717467784882, | |
| "learning_rate": 0.00018806393779335483, | |
| "loss": 1.1725, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.07129624212640652, | |
| "grad_norm": 0.24471914768218994, | |
| "learning_rate": 0.00018800248468149543, | |
| "loss": 1.19, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.07145225140895883, | |
| "grad_norm": 0.27745166420936584, | |
| "learning_rate": 0.00018794088387282, | |
| "loss": 1.6347, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.07160826069151115, | |
| "grad_norm": 0.2930917739868164, | |
| "learning_rate": 0.00018787913547071484, | |
| "loss": 1.5139, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.07176426997406346, | |
| "grad_norm": 0.2656380534172058, | |
| "learning_rate": 0.00018781723957881372, | |
| "loss": 1.1726, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.07192027925661577, | |
| "grad_norm": 0.27983731031417847, | |
| "learning_rate": 0.0001877551963009982, | |
| "loss": 1.3818, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.07207628853916809, | |
| "grad_norm": 0.2744976580142975, | |
| "learning_rate": 0.0001876930057413971, | |
| "loss": 1.2756, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.07223229782172039, | |
| "grad_norm": 0.2684760093688965, | |
| "learning_rate": 0.00018763066800438636, | |
| "loss": 1.2302, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.0723883071042727, | |
| "grad_norm": 0.25079357624053955, | |
| "learning_rate": 0.00018756818319458907, | |
| "loss": 1.1575, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.07254431638682501, | |
| "grad_norm": 0.2802796959877014, | |
| "learning_rate": 0.000187505551416875, | |
| "loss": 1.3711, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.07270032566937733, | |
| "grad_norm": 0.7640414237976074, | |
| "learning_rate": 0.0001874427727763607, | |
| "loss": 1.3431, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.07285633495192964, | |
| "grad_norm": 0.265717089176178, | |
| "learning_rate": 0.0001873798473784092, | |
| "loss": 1.1778, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.07301234423448195, | |
| "grad_norm": 0.23273074626922607, | |
| "learning_rate": 0.00018731677532862976, | |
| "loss": 1.02, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.07316835351703427, | |
| "grad_norm": 0.248812735080719, | |
| "learning_rate": 0.00018725355673287778, | |
| "loss": 1.1423, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.07332436279958658, | |
| "grad_norm": 0.24919858574867249, | |
| "learning_rate": 0.00018719019169725472, | |
| "loss": 1.2377, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.07348037208213888, | |
| "grad_norm": 0.25503799319267273, | |
| "learning_rate": 0.00018712668032810768, | |
| "loss": 1.3236, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.0736363813646912, | |
| "grad_norm": 0.28893566131591797, | |
| "learning_rate": 0.00018706302273202943, | |
| "loss": 1.4662, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.07379239064724351, | |
| "grad_norm": 0.2384706735610962, | |
| "learning_rate": 0.00018699921901585813, | |
| "loss": 1.2817, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.07394839992979582, | |
| "grad_norm": 0.2527397572994232, | |
| "learning_rate": 0.0001869352692866772, | |
| "loss": 1.1766, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.07410440921234814, | |
| "grad_norm": 0.25340378284454346, | |
| "learning_rate": 0.00018687117365181512, | |
| "loss": 1.1876, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.07426041849490045, | |
| "grad_norm": 0.2570219039916992, | |
| "learning_rate": 0.00018680693221884517, | |
| "loss": 1.3472, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.07441642777745276, | |
| "grad_norm": 0.25267085433006287, | |
| "learning_rate": 0.00018674254509558544, | |
| "loss": 1.5048, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.07457243706000508, | |
| "grad_norm": 0.24603790044784546, | |
| "learning_rate": 0.00018667801239009846, | |
| "loss": 1.276, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.07472844634255738, | |
| "grad_norm": 0.2434520423412323, | |
| "learning_rate": 0.00018661333421069113, | |
| "loss": 1.3999, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.07488445562510969, | |
| "grad_norm": 0.27032792568206787, | |
| "learning_rate": 0.00018654851066591448, | |
| "loss": 1.3909, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.075040464907662, | |
| "grad_norm": 0.26559844613075256, | |
| "learning_rate": 0.00018648354186456348, | |
| "loss": 1.2931, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.07519647419021432, | |
| "grad_norm": 0.2563202679157257, | |
| "learning_rate": 0.000186418427915677, | |
| "loss": 1.2773, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.07535248347276663, | |
| "grad_norm": 0.2463751882314682, | |
| "learning_rate": 0.00018635316892853741, | |
| "loss": 1.4017, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.07550849275531894, | |
| "grad_norm": 0.26452189683914185, | |
| "learning_rate": 0.00018628776501267052, | |
| "loss": 1.2236, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.07566450203787126, | |
| "grad_norm": 0.48540955781936646, | |
| "learning_rate": 0.0001862222162778454, | |
| "loss": 1.1676, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.07582051132042357, | |
| "grad_norm": 0.2931404411792755, | |
| "learning_rate": 0.0001861565228340742, | |
| "loss": 1.3877, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.07597652060297588, | |
| "grad_norm": 0.2707270383834839, | |
| "learning_rate": 0.00018609068479161182, | |
| "loss": 1.2828, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.07613252988552818, | |
| "grad_norm": 0.25902295112609863, | |
| "learning_rate": 0.00018602470226095603, | |
| "loss": 1.2393, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.0762885391680805, | |
| "grad_norm": 0.27907291054725647, | |
| "learning_rate": 0.00018595857535284692, | |
| "loss": 1.1944, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.07644454845063281, | |
| "grad_norm": 0.3079850375652313, | |
| "learning_rate": 0.00018589230417826697, | |
| "loss": 1.3686, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.07660055773318512, | |
| "grad_norm": 0.250303715467453, | |
| "learning_rate": 0.00018582588884844084, | |
| "loss": 1.2497, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.07675656701573744, | |
| "grad_norm": 0.260257750749588, | |
| "learning_rate": 0.00018575932947483502, | |
| "loss": 1.4186, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.07691257629828975, | |
| "grad_norm": 0.2537723481655121, | |
| "learning_rate": 0.00018569262616915784, | |
| "loss": 1.28, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.07706858558084206, | |
| "grad_norm": 0.21861004829406738, | |
| "learning_rate": 0.00018562577904335912, | |
| "loss": 0.9705, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.07722459486339438, | |
| "grad_norm": 0.322566956281662, | |
| "learning_rate": 0.00018555878820963013, | |
| "loss": 1.4941, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.07738060414594668, | |
| "grad_norm": 0.24904873967170715, | |
| "learning_rate": 0.00018549165378040327, | |
| "loss": 1.2277, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.07753661342849899, | |
| "grad_norm": 0.2692057490348816, | |
| "learning_rate": 0.00018542437586835202, | |
| "loss": 1.3786, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.0776926227110513, | |
| "grad_norm": 0.27876508235931396, | |
| "learning_rate": 0.00018535695458639056, | |
| "loss": 1.3822, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.07784863199360362, | |
| "grad_norm": 0.2497495859861374, | |
| "learning_rate": 0.00018528939004767376, | |
| "loss": 1.1872, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.07800464127615593, | |
| "grad_norm": 0.28155678510665894, | |
| "learning_rate": 0.00018522168236559695, | |
| "loss": 1.2253, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.07800464127615593, | |
| "eval_loss": 1.3168833255767822, | |
| "eval_runtime": 110.9584, | |
| "eval_samples_per_second": 38.51, | |
| "eval_steps_per_second": 4.822, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.07816065055870824, | |
| "grad_norm": 0.25162461400032043, | |
| "learning_rate": 0.0001851538316537956, | |
| "loss": 1.2308, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.07831665984126056, | |
| "grad_norm": 0.33541133999824524, | |
| "learning_rate": 0.0001850858380261453, | |
| "loss": 1.2788, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.07847266912381287, | |
| "grad_norm": 0.29069721698760986, | |
| "learning_rate": 0.00018501770159676156, | |
| "loss": 1.4186, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.07862867840636519, | |
| "grad_norm": 0.24337412416934967, | |
| "learning_rate": 0.0001849494224799994, | |
| "loss": 1.2268, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.07878468768891748, | |
| "grad_norm": 0.2503622770309448, | |
| "learning_rate": 0.00018488100079045344, | |
| "loss": 1.1121, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.0789406969714698, | |
| "grad_norm": 0.3061240017414093, | |
| "learning_rate": 0.0001848124366429576, | |
| "loss": 1.4207, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.07909670625402211, | |
| "grad_norm": 0.3209320902824402, | |
| "learning_rate": 0.00018474373015258473, | |
| "loss": 1.3531, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.07925271553657443, | |
| "grad_norm": 0.26510298252105713, | |
| "learning_rate": 0.0001846748814346468, | |
| "loss": 1.1614, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.07940872481912674, | |
| "grad_norm": 0.24753335118293762, | |
| "learning_rate": 0.00018460589060469425, | |
| "loss": 1.2711, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.07956473410167905, | |
| "grad_norm": 0.2837298512458801, | |
| "learning_rate": 0.00018453675777851627, | |
| "loss": 1.2325, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.07972074338423137, | |
| "grad_norm": 0.30447372794151306, | |
| "learning_rate": 0.00018446748307214019, | |
| "loss": 1.2425, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.07987675266678368, | |
| "grad_norm": 0.27281391620635986, | |
| "learning_rate": 0.0001843980666018315, | |
| "loss": 1.3095, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.08003276194933598, | |
| "grad_norm": 0.27750325202941895, | |
| "learning_rate": 0.00018432850848409363, | |
| "loss": 1.5124, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.08018877123188829, | |
| "grad_norm": 0.32551145553588867, | |
| "learning_rate": 0.00018425880883566782, | |
| "loss": 1.5727, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.0803447805144406, | |
| "grad_norm": 0.29455453157424927, | |
| "learning_rate": 0.0001841889677735327, | |
| "loss": 1.1937, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.08050078979699292, | |
| "grad_norm": 0.271435022354126, | |
| "learning_rate": 0.00018411898541490434, | |
| "loss": 1.3523, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.08065679907954523, | |
| "grad_norm": 0.28192776441574097, | |
| "learning_rate": 0.0001840488618772359, | |
| "loss": 1.4196, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.08081280836209755, | |
| "grad_norm": 0.32622769474983215, | |
| "learning_rate": 0.00018397859727821748, | |
| "loss": 1.3939, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.08096881764464986, | |
| "grad_norm": 0.26916465163230896, | |
| "learning_rate": 0.00018390819173577598, | |
| "loss": 1.315, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.08112482692720217, | |
| "grad_norm": 0.2807716429233551, | |
| "learning_rate": 0.00018383764536807485, | |
| "loss": 1.4009, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.08128083620975449, | |
| "grad_norm": 0.2609405517578125, | |
| "learning_rate": 0.00018376695829351377, | |
| "loss": 0.9599, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.08143684549230679, | |
| "grad_norm": 0.27300071716308594, | |
| "learning_rate": 0.00018369613063072874, | |
| "loss": 1.2349, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.0815928547748591, | |
| "grad_norm": 0.26670917868614197, | |
| "learning_rate": 0.00018362516249859163, | |
| "loss": 1.2895, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.08174886405741141, | |
| "grad_norm": 0.2805304527282715, | |
| "learning_rate": 0.00018355405401621001, | |
| "loss": 1.3661, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.08190487333996373, | |
| "grad_norm": 0.25124502182006836, | |
| "learning_rate": 0.00018348280530292713, | |
| "loss": 1.2215, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.08206088262251604, | |
| "grad_norm": 0.2374117225408554, | |
| "learning_rate": 0.00018341141647832147, | |
| "loss": 1.1662, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.08221689190506835, | |
| "grad_norm": 0.2681942582130432, | |
| "learning_rate": 0.00018333988766220676, | |
| "loss": 1.3256, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.08237290118762067, | |
| "grad_norm": 0.26264506578445435, | |
| "learning_rate": 0.0001832682189746316, | |
| "loss": 1.1417, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.08252891047017298, | |
| "grad_norm": 0.2661115527153015, | |
| "learning_rate": 0.00018319641053587938, | |
| "loss": 1.2202, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.08268491975272528, | |
| "grad_norm": 0.23459146916866302, | |
| "learning_rate": 0.0001831244624664681, | |
| "loss": 1.0511, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.0828409290352776, | |
| "grad_norm": 0.31903690099716187, | |
| "learning_rate": 0.00018305237488714995, | |
| "loss": 1.565, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.08299693831782991, | |
| "grad_norm": 0.28528186678886414, | |
| "learning_rate": 0.00018298014791891137, | |
| "loss": 1.5023, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.08315294760038222, | |
| "grad_norm": 0.2572003901004791, | |
| "learning_rate": 0.00018290778168297277, | |
| "loss": 1.1518, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.08330895688293453, | |
| "grad_norm": 0.27797260880470276, | |
| "learning_rate": 0.00018283527630078825, | |
| "loss": 1.344, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.08346496616548685, | |
| "grad_norm": 0.3142591416835785, | |
| "learning_rate": 0.0001827626318940454, | |
| "loss": 1.4126, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.08362097544803916, | |
| "grad_norm": 0.2703491151332855, | |
| "learning_rate": 0.00018268984858466522, | |
| "loss": 1.2156, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.08377698473059147, | |
| "grad_norm": 0.29505112767219543, | |
| "learning_rate": 0.00018261692649480175, | |
| "loss": 1.421, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.08393299401314379, | |
| "grad_norm": 0.2756875157356262, | |
| "learning_rate": 0.00018254386574684204, | |
| "loss": 1.4858, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.08408900329569609, | |
| "grad_norm": 0.2744990885257721, | |
| "learning_rate": 0.0001824706664634058, | |
| "loss": 1.3441, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.0842450125782484, | |
| "grad_norm": 0.2834165096282959, | |
| "learning_rate": 0.00018239732876734527, | |
| "loss": 1.4142, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.08440102186080072, | |
| "grad_norm": 0.2717669904232025, | |
| "learning_rate": 0.0001823238527817449, | |
| "loss": 1.3199, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.08455703114335303, | |
| "grad_norm": 0.26433441042900085, | |
| "learning_rate": 0.00018225023862992142, | |
| "loss": 1.3197, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.08471304042590534, | |
| "grad_norm": 0.27460265159606934, | |
| "learning_rate": 0.00018217648643542323, | |
| "loss": 1.216, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.08486904970845766, | |
| "grad_norm": 0.26642194390296936, | |
| "learning_rate": 0.0001821025963220306, | |
| "loss": 1.1716, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.08502505899100997, | |
| "grad_norm": 0.2999640703201294, | |
| "learning_rate": 0.00018202856841375518, | |
| "loss": 1.394, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.08518106827356228, | |
| "grad_norm": 0.2676008641719818, | |
| "learning_rate": 0.00018195440283483988, | |
| "loss": 1.2725, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.08533707755611458, | |
| "grad_norm": 0.26116111874580383, | |
| "learning_rate": 0.0001818800997097587, | |
| "loss": 1.329, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.0854930868386669, | |
| "grad_norm": 0.26923874020576477, | |
| "learning_rate": 0.00018180565916321647, | |
| "loss": 1.2228, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.08564909612121921, | |
| "grad_norm": 0.2784603536128998, | |
| "learning_rate": 0.0001817310813201486, | |
| "loss": 1.1249, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.08580510540377152, | |
| "grad_norm": 0.27981552481651306, | |
| "learning_rate": 0.0001816563663057211, | |
| "loss": 1.2778, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.08596111468632384, | |
| "grad_norm": 0.2464422732591629, | |
| "learning_rate": 0.00018158151424533002, | |
| "loss": 1.0316, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.08611712396887615, | |
| "grad_norm": 0.23159442842006683, | |
| "learning_rate": 0.00018150652526460146, | |
| "loss": 0.9794, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.08627313325142846, | |
| "grad_norm": 0.28374752402305603, | |
| "learning_rate": 0.00018143139948939137, | |
| "loss": 1.0572, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.08642914253398078, | |
| "grad_norm": 0.28464943170547485, | |
| "learning_rate": 0.00018135613704578526, | |
| "loss": 1.024, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.08658515181653309, | |
| "grad_norm": 0.23248714208602905, | |
| "learning_rate": 0.000181280738060098, | |
| "loss": 0.9151, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.08674116109908539, | |
| "grad_norm": 0.2613517940044403, | |
| "learning_rate": 0.00018120520265887363, | |
| "loss": 1.2155, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.0868971703816377, | |
| "grad_norm": 0.2925867438316345, | |
| "learning_rate": 0.00018112953096888516, | |
| "loss": 1.2136, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.08705317966419002, | |
| "grad_norm": 0.3145943582057953, | |
| "learning_rate": 0.00018105372311713432, | |
| "loss": 1.4368, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.08720918894674233, | |
| "grad_norm": 0.29513052105903625, | |
| "learning_rate": 0.0001809777792308513, | |
| "loss": 1.4516, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.08736519822929464, | |
| "grad_norm": 0.22099293768405914, | |
| "learning_rate": 0.00018090169943749476, | |
| "loss": 1.0234, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.08752120751184696, | |
| "grad_norm": 0.24346297979354858, | |
| "learning_rate": 0.0001808254838647513, | |
| "loss": 1.3492, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.08767721679439927, | |
| "grad_norm": 0.2770818769931793, | |
| "learning_rate": 0.00018074913264053545, | |
| "loss": 1.4692, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.08783322607695158, | |
| "grad_norm": 0.2789641320705414, | |
| "learning_rate": 0.00018067264589298945, | |
| "loss": 1.3942, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.08798923535950388, | |
| "grad_norm": 0.2892186939716339, | |
| "learning_rate": 0.00018059602375048293, | |
| "loss": 1.3621, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.0881452446420562, | |
| "grad_norm": 0.28431588411331177, | |
| "learning_rate": 0.00018051926634161282, | |
| "loss": 1.3073, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.08830125392460851, | |
| "grad_norm": 0.3204723000526428, | |
| "learning_rate": 0.00018044237379520305, | |
| "loss": 1.8154, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.08845726320716082, | |
| "grad_norm": 0.2658674716949463, | |
| "learning_rate": 0.0001803653462403043, | |
| "loss": 1.1807, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.08861327248971314, | |
| "grad_norm": 0.2409079521894455, | |
| "learning_rate": 0.0001802881838061939, | |
| "loss": 1.2165, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.08876928177226545, | |
| "grad_norm": 0.25896573066711426, | |
| "learning_rate": 0.00018021088662237552, | |
| "loss": 1.1993, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.08892529105481776, | |
| "grad_norm": 0.27663204073905945, | |
| "learning_rate": 0.00018013345481857903, | |
| "loss": 1.1241, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.08908130033737008, | |
| "grad_norm": 0.2892790734767914, | |
| "learning_rate": 0.00018005588852476015, | |
| "loss": 1.6163, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.08923730961992239, | |
| "grad_norm": 0.30898550152778625, | |
| "learning_rate": 0.00017997818787110042, | |
| "loss": 1.2483, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.08939331890247469, | |
| "grad_norm": 0.23732271790504456, | |
| "learning_rate": 0.0001799003529880068, | |
| "loss": 1.1204, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.089549328185027, | |
| "grad_norm": 0.2597337067127228, | |
| "learning_rate": 0.0001798223840061116, | |
| "loss": 1.258, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.08970533746757932, | |
| "grad_norm": 0.31342512369155884, | |
| "learning_rate": 0.00017974428105627208, | |
| "loss": 1.4074, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.08986134675013163, | |
| "grad_norm": 0.30252331495285034, | |
| "learning_rate": 0.00017966604426957047, | |
| "loss": 1.2059, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.09001735603268395, | |
| "grad_norm": 0.29326415061950684, | |
| "learning_rate": 0.00017958767377731358, | |
| "loss": 1.4294, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.09017336531523626, | |
| "grad_norm": 0.2915484607219696, | |
| "learning_rate": 0.00017950916971103259, | |
| "loss": 1.3728, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.09032937459778857, | |
| "grad_norm": 0.2966526746749878, | |
| "learning_rate": 0.00017943053220248283, | |
| "loss": 1.5332, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.09048538388034089, | |
| "grad_norm": 0.24311012029647827, | |
| "learning_rate": 0.0001793517613836437, | |
| "loss": 1.1254, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.09064139316289319, | |
| "grad_norm": 0.2950594127178192, | |
| "learning_rate": 0.00017927285738671825, | |
| "loss": 1.7255, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.0907974024454455, | |
| "grad_norm": 0.24679097533226013, | |
| "learning_rate": 0.00017919382034413305, | |
| "loss": 1.2781, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.09095341172799781, | |
| "grad_norm": 0.2747292220592499, | |
| "learning_rate": 0.00017911465038853805, | |
| "loss": 1.3434, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.09110942101055013, | |
| "grad_norm": 0.30099523067474365, | |
| "learning_rate": 0.00017903534765280614, | |
| "loss": 1.4518, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.09126543029310244, | |
| "grad_norm": 0.2866073548793793, | |
| "learning_rate": 0.00017895591227003315, | |
| "loss": 1.1706, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.09142143957565475, | |
| "grad_norm": 0.28832805156707764, | |
| "learning_rate": 0.00017887634437353754, | |
| "loss": 1.2271, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.09157744885820707, | |
| "grad_norm": 0.3714962601661682, | |
| "learning_rate": 0.00017879664409686008, | |
| "loss": 1.4474, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.09173345814075938, | |
| "grad_norm": 0.30591243505477905, | |
| "learning_rate": 0.00017871681157376383, | |
| "loss": 1.0327, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.0918894674233117, | |
| "grad_norm": 0.3032775819301605, | |
| "learning_rate": 0.00017863684693823374, | |
| "loss": 1.6824, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.092045476705864, | |
| "grad_norm": 0.26961666345596313, | |
| "learning_rate": 0.00017855675032447648, | |
| "loss": 1.1249, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.0922014859884163, | |
| "grad_norm": 0.2679152488708496, | |
| "learning_rate": 0.00017847652186692026, | |
| "loss": 1.2182, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.09235749527096862, | |
| "grad_norm": 0.24089114367961884, | |
| "learning_rate": 0.00017839616170021452, | |
| "loss": 1.1095, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.09251350455352093, | |
| "grad_norm": 0.25100457668304443, | |
| "learning_rate": 0.00017831566995922985, | |
| "loss": 1.1441, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.09266951383607325, | |
| "grad_norm": 0.2766099274158478, | |
| "learning_rate": 0.0001782350467790575, | |
| "loss": 1.1893, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.09282552311862556, | |
| "grad_norm": 0.2666013240814209, | |
| "learning_rate": 0.00017815429229500946, | |
| "loss": 1.1802, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.09298153240117787, | |
| "grad_norm": 0.28148403763771057, | |
| "learning_rate": 0.00017807340664261802, | |
| "loss": 1.3232, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.09313754168373019, | |
| "grad_norm": 0.23684674501419067, | |
| "learning_rate": 0.00017799238995763568, | |
| "loss": 1.1869, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.09329355096628249, | |
| "grad_norm": 0.2614571154117584, | |
| "learning_rate": 0.00017791124237603477, | |
| "loss": 1.4023, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.0934495602488348, | |
| "grad_norm": 0.3051559329032898, | |
| "learning_rate": 0.00017782996403400736, | |
| "loss": 1.407, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.09360556953138711, | |
| "grad_norm": 0.2745681405067444, | |
| "learning_rate": 0.00017774855506796496, | |
| "loss": 1.3265, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.09376157881393943, | |
| "grad_norm": 0.2689257860183716, | |
| "learning_rate": 0.0001776670156145383, | |
| "loss": 1.3046, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.09391758809649174, | |
| "grad_norm": 0.29333195090293884, | |
| "learning_rate": 0.00017758534581057718, | |
| "loss": 1.2624, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.09407359737904405, | |
| "grad_norm": 0.30287420749664307, | |
| "learning_rate": 0.00017750354579315004, | |
| "loss": 1.28, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.09422960666159637, | |
| "grad_norm": 0.27796801924705505, | |
| "learning_rate": 0.00017742161569954398, | |
| "loss": 1.3305, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.09438561594414868, | |
| "grad_norm": 0.2703540325164795, | |
| "learning_rate": 0.0001773395556672644, | |
| "loss": 1.4356, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.094541625226701, | |
| "grad_norm": 0.26395589113235474, | |
| "learning_rate": 0.0001772573658340347, | |
| "loss": 1.1984, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.0946976345092533, | |
| "grad_norm": 0.2784560024738312, | |
| "learning_rate": 0.0001771750463377962, | |
| "loss": 1.3625, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.09485364379180561, | |
| "grad_norm": 0.31962451338768005, | |
| "learning_rate": 0.00017709259731670774, | |
| "loss": 1.3956, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.09500965307435792, | |
| "grad_norm": 0.274460107088089, | |
| "learning_rate": 0.00017701001890914572, | |
| "loss": 1.3071, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.09516566235691024, | |
| "grad_norm": 0.25924167037010193, | |
| "learning_rate": 0.00017692731125370354, | |
| "loss": 1.034, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.09532167163946255, | |
| "grad_norm": 0.3091680705547333, | |
| "learning_rate": 0.00017684447448919154, | |
| "loss": 1.4134, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.09547768092201486, | |
| "grad_norm": 0.25753480195999146, | |
| "learning_rate": 0.00017676150875463686, | |
| "loss": 1.2074, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.09563369020456718, | |
| "grad_norm": 0.27256032824516296, | |
| "learning_rate": 0.0001766784141892829, | |
| "loss": 1.3758, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.09578969948711949, | |
| "grad_norm": 0.24764277040958405, | |
| "learning_rate": 0.0001765951909325895, | |
| "loss": 1.0436, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.09594570876967179, | |
| "grad_norm": 0.2722652554512024, | |
| "learning_rate": 0.00017651183912423228, | |
| "loss": 1.3623, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.0961017180522241, | |
| "grad_norm": 0.27056217193603516, | |
| "learning_rate": 0.0001764283589041028, | |
| "loss": 1.2525, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.09625772733477642, | |
| "grad_norm": 0.27987945079803467, | |
| "learning_rate": 0.00017634475041230797, | |
| "loss": 1.5075, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.09641373661732873, | |
| "grad_norm": 0.29397958517074585, | |
| "learning_rate": 0.00017626101378917004, | |
| "loss": 1.3681, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.09656974589988104, | |
| "grad_norm": 0.2876337766647339, | |
| "learning_rate": 0.0001761771491752264, | |
| "loss": 1.5848, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.09672575518243336, | |
| "grad_norm": 0.237448051571846, | |
| "learning_rate": 0.0001760931567112291, | |
| "loss": 1.0918, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.09688176446498567, | |
| "grad_norm": 0.29513096809387207, | |
| "learning_rate": 0.0001760090365381449, | |
| "loss": 1.3236, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.09703777374753798, | |
| "grad_norm": 0.263920396566391, | |
| "learning_rate": 0.0001759247887971548, | |
| "loss": 1.4573, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.0971937830300903, | |
| "grad_norm": 0.31876271963119507, | |
| "learning_rate": 0.00017584041362965396, | |
| "loss": 1.3874, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.0973497923126426, | |
| "grad_norm": 0.30635690689086914, | |
| "learning_rate": 0.0001757559111772513, | |
| "loss": 1.2355, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.09750580159519491, | |
| "grad_norm": 0.25926241278648376, | |
| "learning_rate": 0.00017567128158176953, | |
| "loss": 1.2641, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.09766181087774722, | |
| "grad_norm": 0.2862091660499573, | |
| "learning_rate": 0.0001755865249852446, | |
| "loss": 1.3818, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.09781782016029954, | |
| "grad_norm": 0.2540535628795624, | |
| "learning_rate": 0.00017550164152992573, | |
| "loss": 1.3807, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.09797382944285185, | |
| "grad_norm": 0.30917900800704956, | |
| "learning_rate": 0.00017541663135827492, | |
| "loss": 1.1053, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.09812983872540416, | |
| "grad_norm": 0.30465036630630493, | |
| "learning_rate": 0.000175331494612967, | |
| "loss": 1.4489, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.09828584800795648, | |
| "grad_norm": 0.3043782711029053, | |
| "learning_rate": 0.00017524623143688902, | |
| "loss": 1.4544, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.09844185729050879, | |
| "grad_norm": 0.2681322991847992, | |
| "learning_rate": 0.00017516084197314046, | |
| "loss": 1.1926, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.09859786657306109, | |
| "grad_norm": 0.33450305461883545, | |
| "learning_rate": 0.00017507532636503256, | |
| "loss": 1.4383, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.0987538758556134, | |
| "grad_norm": 0.2626807987689972, | |
| "learning_rate": 0.00017498968475608838, | |
| "loss": 1.1565, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.09890988513816572, | |
| "grad_norm": 0.2553156912326813, | |
| "learning_rate": 0.00017490391729004244, | |
| "loss": 1.1327, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.09906589442071803, | |
| "grad_norm": 0.23390045762062073, | |
| "learning_rate": 0.00017481802411084042, | |
| "loss": 0.9856, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.09922190370327034, | |
| "grad_norm": 0.29881760478019714, | |
| "learning_rate": 0.00017473200536263905, | |
| "loss": 1.362, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.09937791298582266, | |
| "grad_norm": 0.2904150187969208, | |
| "learning_rate": 0.0001746458611898058, | |
| "loss": 1.242, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.09953392226837497, | |
| "grad_norm": 0.24842409789562225, | |
| "learning_rate": 0.00017455959173691863, | |
| "loss": 1.2694, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.09968993155092729, | |
| "grad_norm": 0.3337212800979614, | |
| "learning_rate": 0.00017447319714876579, | |
| "loss": 1.2554, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.0998459408334796, | |
| "grad_norm": 0.24105407297611237, | |
| "learning_rate": 0.00017438667757034546, | |
| "loss": 1.0582, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.1000019501160319, | |
| "grad_norm": 0.24266989529132843, | |
| "learning_rate": 0.00017430003314686569, | |
| "loss": 1.2125, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.10015795939858421, | |
| "grad_norm": 0.2654808461666107, | |
| "learning_rate": 0.00017421326402374405, | |
| "loss": 1.3229, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.10031396868113653, | |
| "grad_norm": 0.21931445598602295, | |
| "learning_rate": 0.00017412637034660734, | |
| "loss": 1.1168, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.10046997796368884, | |
| "grad_norm": 0.28860512375831604, | |
| "learning_rate": 0.0001740393522612915, | |
| "loss": 1.3681, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.10062598724624115, | |
| "grad_norm": 0.2736460566520691, | |
| "learning_rate": 0.0001739522099138411, | |
| "loss": 1.4054, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.10078199652879347, | |
| "grad_norm": 0.23222267627716064, | |
| "learning_rate": 0.00017386494345050942, | |
| "loss": 1.0973, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.10093800581134578, | |
| "grad_norm": 0.2684474587440491, | |
| "learning_rate": 0.000173777553017758, | |
| "loss": 1.0637, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.10109401509389809, | |
| "grad_norm": 0.2648880183696747, | |
| "learning_rate": 0.00017369003876225642, | |
| "loss": 1.5162, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.10125002437645039, | |
| "grad_norm": 0.26263687014579773, | |
| "learning_rate": 0.00017360240083088213, | |
| "loss": 1.3613, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.1014060336590027, | |
| "grad_norm": 0.2455459088087082, | |
| "learning_rate": 0.00017351463937072004, | |
| "loss": 1.3927, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.10156204294155502, | |
| "grad_norm": 0.273078590631485, | |
| "learning_rate": 0.00017342675452906248, | |
| "loss": 1.2485, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.10171805222410733, | |
| "grad_norm": 0.24480541050434113, | |
| "learning_rate": 0.00017333874645340884, | |
| "loss": 1.0656, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.10187406150665965, | |
| "grad_norm": 0.24994470179080963, | |
| "learning_rate": 0.0001732506152914653, | |
| "loss": 1.3653, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.10203007078921196, | |
| "grad_norm": 0.26110485196113586, | |
| "learning_rate": 0.00017316236119114463, | |
| "loss": 1.392, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.10218608007176427, | |
| "grad_norm": 0.30197709798812866, | |
| "learning_rate": 0.00017307398430056593, | |
| "loss": 1.5184, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.10234208935431659, | |
| "grad_norm": 0.26577743887901306, | |
| "learning_rate": 0.00017298548476805446, | |
| "loss": 1.4611, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.10249809863686889, | |
| "grad_norm": 0.2677333950996399, | |
| "learning_rate": 0.00017289686274214118, | |
| "loss": 1.3282, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.1026541079194212, | |
| "grad_norm": 0.2508523762226105, | |
| "learning_rate": 0.00017280811837156268, | |
| "loss": 1.1331, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.10281011720197351, | |
| "grad_norm": 0.24873429536819458, | |
| "learning_rate": 0.00017271925180526094, | |
| "loss": 1.1351, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.10296612648452583, | |
| "grad_norm": 0.2559413015842438, | |
| "learning_rate": 0.00017263026319238301, | |
| "loss": 1.245, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.10312213576707814, | |
| "grad_norm": 0.29988738894462585, | |
| "learning_rate": 0.0001725411526822807, | |
| "loss": 1.4004, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.10327814504963045, | |
| "grad_norm": 0.29719191789627075, | |
| "learning_rate": 0.0001724519204245105, | |
| "loss": 1.5687, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.10343415433218277, | |
| "grad_norm": 0.30810216069221497, | |
| "learning_rate": 0.0001723625665688331, | |
| "loss": 1.3712, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.10359016361473508, | |
| "grad_norm": 0.2754259407520294, | |
| "learning_rate": 0.00017227309126521348, | |
| "loss": 1.2083, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.1037461728972874, | |
| "grad_norm": 0.26548734307289124, | |
| "learning_rate": 0.00017218349466382023, | |
| "loss": 1.2657, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.1039021821798397, | |
| "grad_norm": 0.26369354128837585, | |
| "learning_rate": 0.00017209377691502565, | |
| "loss": 1.3359, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.10405819146239201, | |
| "grad_norm": 0.2526211440563202, | |
| "learning_rate": 0.0001720039381694053, | |
| "loss": 1.0633, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.10421420074494432, | |
| "grad_norm": 0.2874252498149872, | |
| "learning_rate": 0.00017191397857773788, | |
| "loss": 1.2833, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.10437021002749663, | |
| "grad_norm": 0.26982390880584717, | |
| "learning_rate": 0.00017182389829100485, | |
| "loss": 1.1843, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.10452621931004895, | |
| "grad_norm": 0.29615074396133423, | |
| "learning_rate": 0.00017173369746039025, | |
| "loss": 1.2992, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.10468222859260126, | |
| "grad_norm": 0.29073938727378845, | |
| "learning_rate": 0.00017164337623728045, | |
| "loss": 1.5432, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.10483823787515358, | |
| "grad_norm": 0.2858506143093109, | |
| "learning_rate": 0.00017155293477326384, | |
| "loss": 1.4446, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.10499424715770589, | |
| "grad_norm": 0.2399512678384781, | |
| "learning_rate": 0.00017146237322013068, | |
| "loss": 1.1643, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.10515025644025819, | |
| "grad_norm": 0.2796498239040375, | |
| "learning_rate": 0.00017137169172987268, | |
| "loss": 1.3158, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.1053062657228105, | |
| "grad_norm": 0.26859599351882935, | |
| "learning_rate": 0.00017128089045468294, | |
| "loss": 1.1761, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.10546227500536282, | |
| "grad_norm": 0.2749616503715515, | |
| "learning_rate": 0.00017118996954695553, | |
| "loss": 1.0586, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.10561828428791513, | |
| "grad_norm": 0.27312207221984863, | |
| "learning_rate": 0.00017109892915928535, | |
| "loss": 1.1367, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.10577429357046744, | |
| "grad_norm": 0.29626578092575073, | |
| "learning_rate": 0.00017100776944446781, | |
| "loss": 1.4223, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.10593030285301976, | |
| "grad_norm": 0.24335867166519165, | |
| "learning_rate": 0.00017091649055549855, | |
| "loss": 1.1041, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.10608631213557207, | |
| "grad_norm": 0.3017411530017853, | |
| "learning_rate": 0.0001708250926455733, | |
| "loss": 1.2854, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.10624232141812438, | |
| "grad_norm": 0.2864495515823364, | |
| "learning_rate": 0.00017073357586808752, | |
| "loss": 1.2539, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.1063983307006767, | |
| "grad_norm": 0.27407294511795044, | |
| "learning_rate": 0.0001706419403766361, | |
| "loss": 1.3136, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.106554339983229, | |
| "grad_norm": 0.3100734055042267, | |
| "learning_rate": 0.00017055018632501325, | |
| "loss": 1.3231, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.10671034926578131, | |
| "grad_norm": 0.3091520071029663, | |
| "learning_rate": 0.00017045831386721213, | |
| "loss": 1.3513, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.10686635854833362, | |
| "grad_norm": 0.2930145561695099, | |
| "learning_rate": 0.00017036632315742462, | |
| "loss": 1.3292, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.10702236783088594, | |
| "grad_norm": 0.30808883905410767, | |
| "learning_rate": 0.00017027421435004112, | |
| "loss": 1.6094, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.10717837711343825, | |
| "grad_norm": 0.2715398073196411, | |
| "learning_rate": 0.00017018198759965016, | |
| "loss": 1.3641, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.10733438639599056, | |
| "grad_norm": 0.2844456732273102, | |
| "learning_rate": 0.00017008964306103823, | |
| "loss": 1.3933, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.10749039567854288, | |
| "grad_norm": 0.258504718542099, | |
| "learning_rate": 0.00016999718088918955, | |
| "loss": 1.0621, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.10764640496109519, | |
| "grad_norm": 0.28674831986427307, | |
| "learning_rate": 0.00016990460123928575, | |
| "loss": 1.2759, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.10780241424364749, | |
| "grad_norm": 0.3062899708747864, | |
| "learning_rate": 0.0001698119042667056, | |
| "loss": 1.1537, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.1079584235261998, | |
| "grad_norm": 0.2539708614349365, | |
| "learning_rate": 0.00016971909012702483, | |
| "loss": 1.1463, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.10811443280875212, | |
| "grad_norm": 0.30207210779190063, | |
| "learning_rate": 0.00016962615897601573, | |
| "loss": 1.4219, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.10827044209130443, | |
| "grad_norm": 0.28675806522369385, | |
| "learning_rate": 0.00016953311096964705, | |
| "loss": 1.1476, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.10842645137385674, | |
| "grad_norm": 0.33274316787719727, | |
| "learning_rate": 0.00016943994626408363, | |
| "loss": 1.3351, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.10858246065640906, | |
| "grad_norm": 0.2725004553794861, | |
| "learning_rate": 0.00016934666501568617, | |
| "loss": 1.1795, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.10873846993896137, | |
| "grad_norm": 0.29064077138900757, | |
| "learning_rate": 0.00016925326738101098, | |
| "loss": 1.4255, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.10889447922151368, | |
| "grad_norm": 0.3007811903953552, | |
| "learning_rate": 0.00016915975351680968, | |
| "loss": 1.1951, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.109050488504066, | |
| "grad_norm": 0.26098549365997314, | |
| "learning_rate": 0.000169066123580029, | |
| "loss": 1.0585, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.1092064977866183, | |
| "grad_norm": 0.36355966329574585, | |
| "learning_rate": 0.00016897237772781044, | |
| "loss": 1.2911, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.10936250706917061, | |
| "grad_norm": 0.2830749750137329, | |
| "learning_rate": 0.00016887851611749005, | |
| "loss": 1.4469, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.10951851635172292, | |
| "grad_norm": 0.3175537884235382, | |
| "learning_rate": 0.00016878453890659814, | |
| "loss": 1.4589, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.10967452563427524, | |
| "grad_norm": 0.2898159623146057, | |
| "learning_rate": 0.0001686904462528591, | |
| "loss": 1.4318, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.10983053491682755, | |
| "grad_norm": 0.28991106152534485, | |
| "learning_rate": 0.000168596238314191, | |
| "loss": 1.3293, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.10998654419937987, | |
| "grad_norm": 0.27654772996902466, | |
| "learning_rate": 0.00016850191524870546, | |
| "loss": 1.4909, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.11014255348193218, | |
| "grad_norm": 0.29537513852119446, | |
| "learning_rate": 0.00016840747721470731, | |
| "loss": 1.4512, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.11029856276448449, | |
| "grad_norm": 0.2656291723251343, | |
| "learning_rate": 0.00016831292437069427, | |
| "loss": 1.0375, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.11045457204703679, | |
| "grad_norm": 0.3286688029766083, | |
| "learning_rate": 0.00016821825687535674, | |
| "loss": 1.3478, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.1106105813295891, | |
| "grad_norm": 0.2618601322174072, | |
| "learning_rate": 0.00016812347488757772, | |
| "loss": 1.3448, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.11076659061214142, | |
| "grad_norm": 0.29108762741088867, | |
| "learning_rate": 0.00016802857856643215, | |
| "loss": 1.3479, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.11092259989469373, | |
| "grad_norm": 0.3029685914516449, | |
| "learning_rate": 0.00016793356807118695, | |
| "loss": 1.2162, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.11107860917724605, | |
| "grad_norm": 0.2573980689048767, | |
| "learning_rate": 0.00016783844356130071, | |
| "loss": 1.0927, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.11123461845979836, | |
| "grad_norm": 0.2836451828479767, | |
| "learning_rate": 0.0001677432051964233, | |
| "loss": 1.2136, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.11139062774235067, | |
| "grad_norm": 0.2437037229537964, | |
| "learning_rate": 0.0001676478531363957, | |
| "loss": 1.0671, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.11154663702490299, | |
| "grad_norm": 0.2603608965873718, | |
| "learning_rate": 0.00016755238754124965, | |
| "loss": 1.2128, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.1117026463074553, | |
| "grad_norm": 0.2617943286895752, | |
| "learning_rate": 0.00016745680857120757, | |
| "loss": 1.3305, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.1118586555900076, | |
| "grad_norm": 0.27264609932899475, | |
| "learning_rate": 0.00016736111638668204, | |
| "loss": 1.3456, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.11201466487255991, | |
| "grad_norm": 0.33472567796707153, | |
| "learning_rate": 0.00016726531114827573, | |
| "loss": 1.2517, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.11217067415511223, | |
| "grad_norm": 0.2825791835784912, | |
| "learning_rate": 0.00016716939301678098, | |
| "loss": 1.3156, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.11232668343766454, | |
| "grad_norm": 0.2815983295440674, | |
| "learning_rate": 0.00016707336215317968, | |
| "loss": 1.2376, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.11248269272021685, | |
| "grad_norm": 0.3158409595489502, | |
| "learning_rate": 0.00016697721871864284, | |
| "loss": 1.5252, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.11263870200276917, | |
| "grad_norm": 0.27121129631996155, | |
| "learning_rate": 0.00016688096287453046, | |
| "loss": 1.3603, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.11279471128532148, | |
| "grad_norm": 0.2568758428096771, | |
| "learning_rate": 0.00016678459478239118, | |
| "loss": 1.1337, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.1129507205678738, | |
| "grad_norm": 0.26672929525375366, | |
| "learning_rate": 0.00016668811460396202, | |
| "loss": 1.1728, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.1131067298504261, | |
| "grad_norm": 0.2683919370174408, | |
| "learning_rate": 0.00016659152250116812, | |
| "loss": 1.2833, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.11326273913297841, | |
| "grad_norm": 0.2757527232170105, | |
| "learning_rate": 0.00016649481863612248, | |
| "loss": 1.0544, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.11341874841553072, | |
| "grad_norm": 0.2571371793746948, | |
| "learning_rate": 0.0001663980031711257, | |
| "loss": 1.1212, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.11357475769808303, | |
| "grad_norm": 0.2757047116756439, | |
| "learning_rate": 0.00016630107626866558, | |
| "loss": 1.1771, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.11373076698063535, | |
| "grad_norm": 0.262979120016098, | |
| "learning_rate": 0.00016620403809141705, | |
| "loss": 0.9962, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.11388677626318766, | |
| "grad_norm": 0.26567909121513367, | |
| "learning_rate": 0.00016610688880224178, | |
| "loss": 1.3037, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.11404278554573997, | |
| "grad_norm": 0.27931660413742065, | |
| "learning_rate": 0.00016600962856418782, | |
| "loss": 1.1863, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.11419879482829229, | |
| "grad_norm": 0.25071558356285095, | |
| "learning_rate": 0.00016591225754048963, | |
| "loss": 1.1437, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.1143548041108446, | |
| "grad_norm": 0.2775113880634308, | |
| "learning_rate": 0.00016581477589456734, | |
| "loss": 1.2152, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.1145108133933969, | |
| "grad_norm": 0.25055718421936035, | |
| "learning_rate": 0.00016571718379002705, | |
| "loss": 1.1479, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.11466682267594921, | |
| "grad_norm": 0.25468993186950684, | |
| "learning_rate": 0.00016561948139065996, | |
| "loss": 1.148, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.11482283195850153, | |
| "grad_norm": 0.26385918259620667, | |
| "learning_rate": 0.00016552166886044253, | |
| "loss": 1.3473, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.11497884124105384, | |
| "grad_norm": 0.27051180601119995, | |
| "learning_rate": 0.00016542374636353604, | |
| "loss": 1.196, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.11513485052360616, | |
| "grad_norm": 0.32731276750564575, | |
| "learning_rate": 0.0001653257140642863, | |
| "loss": 1.4514, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.11529085980615847, | |
| "grad_norm": 0.26046180725097656, | |
| "learning_rate": 0.00016522757212722344, | |
| "loss": 1.2186, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.11544686908871078, | |
| "grad_norm": 0.2661746144294739, | |
| "learning_rate": 0.00016512932071706152, | |
| "loss": 1.123, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.1156028783712631, | |
| "grad_norm": 0.25739923119544983, | |
| "learning_rate": 0.0001650309599986985, | |
| "loss": 1.1832, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.1157588876538154, | |
| "grad_norm": 0.30230990052223206, | |
| "learning_rate": 0.00016493249013721558, | |
| "loss": 1.5064, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.11591489693636771, | |
| "grad_norm": 0.25831449031829834, | |
| "learning_rate": 0.00016483391129787727, | |
| "loss": 1.1212, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.11607090621892002, | |
| "grad_norm": 0.24019654095172882, | |
| "learning_rate": 0.000164735223646131, | |
| "loss": 1.1555, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.11622691550147234, | |
| "grad_norm": 0.28396427631378174, | |
| "learning_rate": 0.0001646364273476067, | |
| "loss": 1.4754, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.11638292478402465, | |
| "grad_norm": 0.28211066126823425, | |
| "learning_rate": 0.00016453752256811674, | |
| "loss": 1.526, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.11653893406657696, | |
| "grad_norm": 0.2596474289894104, | |
| "learning_rate": 0.00016443850947365558, | |
| "loss": 1.2072, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.11669494334912928, | |
| "grad_norm": 0.25947293639183044, | |
| "learning_rate": 0.0001643393882303994, | |
| "loss": 1.3467, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.11685095263168159, | |
| "grad_norm": 0.30946600437164307, | |
| "learning_rate": 0.00016424015900470587, | |
| "loss": 1.3948, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.1170069619142339, | |
| "grad_norm": 0.3172161281108856, | |
| "learning_rate": 0.000164140821963114, | |
| "loss": 1.745, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.1171629711967862, | |
| "grad_norm": 0.26674196124076843, | |
| "learning_rate": 0.00016404137727234365, | |
| "loss": 1.5021, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.11731898047933852, | |
| "grad_norm": 0.26941999793052673, | |
| "learning_rate": 0.00016394182509929536, | |
| "loss": 1.2651, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.11747498976189083, | |
| "grad_norm": 0.29353249073028564, | |
| "learning_rate": 0.00016384216561105014, | |
| "loss": 1.2397, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.11763099904444314, | |
| "grad_norm": 0.2547638416290283, | |
| "learning_rate": 0.000163742398974869, | |
| "loss": 1.1032, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.11778700832699546, | |
| "grad_norm": 0.25621354579925537, | |
| "learning_rate": 0.00016364252535819282, | |
| "loss": 1.0842, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.11794301760954777, | |
| "grad_norm": 0.25465261936187744, | |
| "learning_rate": 0.00016354254492864211, | |
| "loss": 0.9941, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.11809902689210008, | |
| "grad_norm": 0.25726544857025146, | |
| "learning_rate": 0.00016344245785401653, | |
| "loss": 1.2613, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.1182550361746524, | |
| "grad_norm": 0.2696760594844818, | |
| "learning_rate": 0.00016334226430229475, | |
| "loss": 1.1349, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.1184110454572047, | |
| "grad_norm": 0.29465997219085693, | |
| "learning_rate": 0.00016324196444163423, | |
| "loss": 1.3099, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.11856705473975701, | |
| "grad_norm": 0.2854841351509094, | |
| "learning_rate": 0.00016314155844037074, | |
| "loss": 1.1648, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.11872306402230932, | |
| "grad_norm": 0.28557366132736206, | |
| "learning_rate": 0.0001630410464670182, | |
| "loss": 1.4045, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.11887907330486164, | |
| "grad_norm": 0.337882936000824, | |
| "learning_rate": 0.00016294042869026851, | |
| "loss": 1.4391, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.11903508258741395, | |
| "grad_norm": 0.25410857796669006, | |
| "learning_rate": 0.000162839705278991, | |
| "loss": 1.025, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.11919109186996626, | |
| "grad_norm": 0.2944369614124298, | |
| "learning_rate": 0.0001627388764022323, | |
| "loss": 1.3339, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.11934710115251858, | |
| "grad_norm": 0.30941835045814514, | |
| "learning_rate": 0.0001626379422292162, | |
| "loss": 1.5238, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.11950311043507089, | |
| "grad_norm": 0.2796765863895416, | |
| "learning_rate": 0.000162536902929343, | |
| "loss": 1.1711, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.1196591197176232, | |
| "grad_norm": 0.2882195711135864, | |
| "learning_rate": 0.00016243575867218958, | |
| "loss": 1.2852, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.1198151290001755, | |
| "grad_norm": 0.29050207138061523, | |
| "learning_rate": 0.00016233450962750893, | |
| "loss": 1.2789, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.11997113828272782, | |
| "grad_norm": 0.2745670974254608, | |
| "learning_rate": 0.00016223315596522987, | |
| "loss": 1.2741, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.12012714756528013, | |
| "grad_norm": 0.29764166474342346, | |
| "learning_rate": 0.0001621316978554569, | |
| "loss": 1.3636, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.12028315684783245, | |
| "grad_norm": 0.29131025075912476, | |
| "learning_rate": 0.00016203013546846966, | |
| "loss": 1.5137, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.12043916613038476, | |
| "grad_norm": 0.3370944857597351, | |
| "learning_rate": 0.00016192846897472297, | |
| "loss": 1.5541, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.12059517541293707, | |
| "grad_norm": 0.2678642272949219, | |
| "learning_rate": 0.0001618266985448463, | |
| "loss": 1.2024, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.12075118469548939, | |
| "grad_norm": 0.27655884623527527, | |
| "learning_rate": 0.00016172482434964353, | |
| "loss": 1.1084, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.1209071939780417, | |
| "grad_norm": 0.23235641419887543, | |
| "learning_rate": 0.00016162284656009274, | |
| "loss": 0.8548, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.121063203260594, | |
| "grad_norm": 0.2860414683818817, | |
| "learning_rate": 0.00016152076534734584, | |
| "loss": 1.5026, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.12121921254314631, | |
| "grad_norm": 0.2980406582355499, | |
| "learning_rate": 0.00016141858088272837, | |
| "loss": 1.3692, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.12137522182569863, | |
| "grad_norm": 0.29564347863197327, | |
| "learning_rate": 0.00016131629333773908, | |
| "loss": 1.6193, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.12153123110825094, | |
| "grad_norm": 0.250028520822525, | |
| "learning_rate": 0.0001612139028840498, | |
| "loss": 1.3295, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.12168724039080325, | |
| "grad_norm": 0.25812971591949463, | |
| "learning_rate": 0.00016111140969350503, | |
| "loss": 1.1061, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.12184324967335557, | |
| "grad_norm": 0.2702666223049164, | |
| "learning_rate": 0.0001610088139381217, | |
| "loss": 1.2846, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.12199925895590788, | |
| "grad_norm": 0.24256417155265808, | |
| "learning_rate": 0.00016090611579008888, | |
| "loss": 1.081, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.1221552682384602, | |
| "grad_norm": 0.3177904784679413, | |
| "learning_rate": 0.00016080331542176753, | |
| "loss": 1.5862, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.12231127752101251, | |
| "grad_norm": 0.25483664870262146, | |
| "learning_rate": 0.00016070041300569012, | |
| "loss": 1.1939, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.1224672868035648, | |
| "grad_norm": 0.23578673601150513, | |
| "learning_rate": 0.00016059740871456036, | |
| "loss": 1.0371, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.12262329608611712, | |
| "grad_norm": 0.28674736618995667, | |
| "learning_rate": 0.000160494302721253, | |
| "loss": 1.4739, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.12277930536866943, | |
| "grad_norm": 0.29090616106987, | |
| "learning_rate": 0.0001603910951988135, | |
| "loss": 1.3862, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.12293531465122175, | |
| "grad_norm": 0.2792899012565613, | |
| "learning_rate": 0.00016028778632045762, | |
| "loss": 1.3731, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.12309132393377406, | |
| "grad_norm": 0.2683924436569214, | |
| "learning_rate": 0.00016018437625957133, | |
| "loss": 1.4514, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.12324733321632637, | |
| "grad_norm": 0.331752747297287, | |
| "learning_rate": 0.00016008086518971037, | |
| "loss": 1.0936, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.12340334249887869, | |
| "grad_norm": 0.32185712456703186, | |
| "learning_rate": 0.0001599772532846, | |
| "loss": 1.7093, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.123559351781431, | |
| "grad_norm": 0.28801560401916504, | |
| "learning_rate": 0.0001598735407181347, | |
| "loss": 1.2923, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.1237153610639833, | |
| "grad_norm": 0.2626672387123108, | |
| "learning_rate": 0.00015976972766437795, | |
| "loss": 1.196, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.12387137034653561, | |
| "grad_norm": 0.30561795830726624, | |
| "learning_rate": 0.00015966581429756183, | |
| "loss": 1.5151, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.12402737962908793, | |
| "grad_norm": 0.2764839828014374, | |
| "learning_rate": 0.00015956180079208682, | |
| "loss": 1.231, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.12418338891164024, | |
| "grad_norm": 0.2506803870201111, | |
| "learning_rate": 0.00015945768732252144, | |
| "loss": 1.0394, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.12433939819419255, | |
| "grad_norm": 0.28655874729156494, | |
| "learning_rate": 0.00015935347406360192, | |
| "loss": 1.4689, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.12449540747674487, | |
| "grad_norm": 0.26048576831817627, | |
| "learning_rate": 0.00015924916119023212, | |
| "loss": 1.218, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.12465141675929718, | |
| "grad_norm": 0.26712656021118164, | |
| "learning_rate": 0.00015914474887748295, | |
| "loss": 1.232, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.1248074260418495, | |
| "grad_norm": 0.2652023434638977, | |
| "learning_rate": 0.00015904023730059228, | |
| "loss": 1.0205, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.12496343532440181, | |
| "grad_norm": 0.3364275097846985, | |
| "learning_rate": 0.0001589356266349645, | |
| "loss": 1.4919, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.12511944460695412, | |
| "grad_norm": 0.218467116355896, | |
| "learning_rate": 0.00015883091705617045, | |
| "loss": 0.8939, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.12527545388950642, | |
| "grad_norm": 0.2554807960987091, | |
| "learning_rate": 0.00015872610873994685, | |
| "loss": 1.2568, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.12543146317205875, | |
| "grad_norm": 0.2742806673049927, | |
| "learning_rate": 0.00015862120186219613, | |
| "loss": 1.0565, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.12558747245461105, | |
| "grad_norm": 0.23994481563568115, | |
| "learning_rate": 0.00015851619659898623, | |
| "loss": 0.9631, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.12574348173716335, | |
| "grad_norm": 0.29549404978752136, | |
| "learning_rate": 0.00015841109312655016, | |
| "loss": 1.2073, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.12589949101971568, | |
| "grad_norm": 0.27470991015434265, | |
| "learning_rate": 0.00015830589162128572, | |
| "loss": 1.2345, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.12605550030226798, | |
| "grad_norm": 0.27652519941329956, | |
| "learning_rate": 0.00015820059225975531, | |
| "loss": 1.2456, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.1262115095848203, | |
| "grad_norm": 0.2571077346801758, | |
| "learning_rate": 0.0001580951952186856, | |
| "loss": 1.0009, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.1263675188673726, | |
| "grad_norm": 0.27721402049064636, | |
| "learning_rate": 0.000157989700674967, | |
| "loss": 1.2101, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.12652352814992493, | |
| "grad_norm": 0.29823631048202515, | |
| "learning_rate": 0.00015788410880565379, | |
| "loss": 1.3992, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.12667953743247723, | |
| "grad_norm": 0.28366366028785706, | |
| "learning_rate": 0.00015777841978796347, | |
| "loss": 1.005, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.12683554671502956, | |
| "grad_norm": 0.3597376048564911, | |
| "learning_rate": 0.0001576726337992766, | |
| "loss": 1.6046, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.12699155599758186, | |
| "grad_norm": 0.27407100796699524, | |
| "learning_rate": 0.00015756675101713657, | |
| "loss": 1.0167, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.12714756528013416, | |
| "grad_norm": 0.3212680220603943, | |
| "learning_rate": 0.00015746077161924905, | |
| "loss": 1.4425, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.12730357456268648, | |
| "grad_norm": 0.25150859355926514, | |
| "learning_rate": 0.00015735469578348208, | |
| "loss": 1.2482, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.12745958384523878, | |
| "grad_norm": 0.2753000855445862, | |
| "learning_rate": 0.00015724852368786537, | |
| "loss": 1.3006, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.1276155931277911, | |
| "grad_norm": 0.27500027418136597, | |
| "learning_rate": 0.0001571422555105903, | |
| "loss": 1.2095, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.1277716024103434, | |
| "grad_norm": 0.2696485221385956, | |
| "learning_rate": 0.0001570358914300094, | |
| "loss": 1.1708, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.12792761169289574, | |
| "grad_norm": 0.2486962080001831, | |
| "learning_rate": 0.00015692943162463628, | |
| "loss": 1.0531, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.12808362097544804, | |
| "grad_norm": 0.265824556350708, | |
| "learning_rate": 0.00015682287627314515, | |
| "loss": 1.0712, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 0.12823963025800036, | |
| "grad_norm": 0.2963060140609741, | |
| "learning_rate": 0.00015671622555437053, | |
| "loss": 1.3806, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.12839563954055266, | |
| "grad_norm": 0.2849713861942291, | |
| "learning_rate": 0.00015660947964730708, | |
| "loss": 1.2242, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 0.12855164882310496, | |
| "grad_norm": 0.25108298659324646, | |
| "learning_rate": 0.0001565026387311092, | |
| "loss": 1.1128, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.1287076581056573, | |
| "grad_norm": 0.27622735500335693, | |
| "learning_rate": 0.00015639570298509064, | |
| "loss": 1.3599, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.1288636673882096, | |
| "grad_norm": 0.29195183515548706, | |
| "learning_rate": 0.0001562886725887245, | |
| "loss": 1.2931, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.12901967667076192, | |
| "grad_norm": 0.2943118214607239, | |
| "learning_rate": 0.00015618154772164256, | |
| "loss": 1.5802, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 0.12917568595331422, | |
| "grad_norm": 0.26325714588165283, | |
| "learning_rate": 0.00015607432856363525, | |
| "loss": 1.2455, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.12933169523586655, | |
| "grad_norm": 0.286743700504303, | |
| "learning_rate": 0.00015596701529465117, | |
| "loss": 1.3008, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 0.12948770451841884, | |
| "grad_norm": 0.2844702899456024, | |
| "learning_rate": 0.00015585960809479696, | |
| "loss": 1.3737, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.12964371380097114, | |
| "grad_norm": 0.25531789660453796, | |
| "learning_rate": 0.00015575210714433686, | |
| "loss": 1.1425, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 0.12979972308352347, | |
| "grad_norm": 0.26921185851097107, | |
| "learning_rate": 0.00015564451262369247, | |
| "loss": 1.106, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.12995573236607577, | |
| "grad_norm": 0.28271836042404175, | |
| "learning_rate": 0.00015553682471344238, | |
| "loss": 1.3681, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 0.1301117416486281, | |
| "grad_norm": 0.26876282691955566, | |
| "learning_rate": 0.00015542904359432198, | |
| "loss": 1.112, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.1302677509311804, | |
| "grad_norm": 0.2895980179309845, | |
| "learning_rate": 0.00015532116944722308, | |
| "loss": 1.1285, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.13042376021373273, | |
| "grad_norm": 0.2612462639808655, | |
| "learning_rate": 0.00015521320245319363, | |
| "loss": 1.2669, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.13057976949628503, | |
| "grad_norm": 0.30689284205436707, | |
| "learning_rate": 0.00015510514279343734, | |
| "loss": 1.3512, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.13073577877883735, | |
| "grad_norm": 0.2981073558330536, | |
| "learning_rate": 0.00015499699064931355, | |
| "loss": 1.1284, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.13089178806138965, | |
| "grad_norm": 0.2637684643268585, | |
| "learning_rate": 0.00015488874620233674, | |
| "loss": 1.0698, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 0.13104779734394195, | |
| "grad_norm": 0.3048469126224518, | |
| "learning_rate": 0.0001547804096341763, | |
| "loss": 1.5209, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.13120380662649428, | |
| "grad_norm": 0.2396387904882431, | |
| "learning_rate": 0.00015467198112665632, | |
| "loss": 0.9584, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 0.13135981590904658, | |
| "grad_norm": 0.27103736996650696, | |
| "learning_rate": 0.0001545634608617551, | |
| "loss": 1.2846, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.1315158251915989, | |
| "grad_norm": 0.2971721589565277, | |
| "learning_rate": 0.00015445484902160491, | |
| "loss": 1.6074, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 0.1316718344741512, | |
| "grad_norm": 0.2440243512392044, | |
| "learning_rate": 0.00015434614578849188, | |
| "loss": 1.045, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.13182784375670353, | |
| "grad_norm": 0.30210787057876587, | |
| "learning_rate": 0.00015423735134485536, | |
| "loss": 1.2948, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.13198385303925583, | |
| "grad_norm": 0.25344711542129517, | |
| "learning_rate": 0.00015412846587328782, | |
| "loss": 1.2089, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.13213986232180816, | |
| "grad_norm": 0.2884974479675293, | |
| "learning_rate": 0.0001540194895565346, | |
| "loss": 1.1123, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 0.13229587160436046, | |
| "grad_norm": 0.28012582659721375, | |
| "learning_rate": 0.00015391042257749336, | |
| "loss": 1.2269, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.13245188088691276, | |
| "grad_norm": 0.26394879817962646, | |
| "learning_rate": 0.00015380126511921403, | |
| "loss": 1.4469, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 0.1326078901694651, | |
| "grad_norm": 0.2717582583427429, | |
| "learning_rate": 0.0001536920173648984, | |
| "loss": 1.1494, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.1327638994520174, | |
| "grad_norm": 0.2968549132347107, | |
| "learning_rate": 0.00015358267949789966, | |
| "loss": 1.1903, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.13291990873456971, | |
| "grad_norm": 0.2570381164550781, | |
| "learning_rate": 0.00015347325170172245, | |
| "loss": 1.1035, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.133075918017122, | |
| "grad_norm": 0.3070929944515228, | |
| "learning_rate": 0.0001533637341600221, | |
| "loss": 1.4062, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 0.13323192729967434, | |
| "grad_norm": 0.2886407971382141, | |
| "learning_rate": 0.0001532541270566049, | |
| "loss": 1.3491, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.13338793658222664, | |
| "grad_norm": 0.2572009861469269, | |
| "learning_rate": 0.00015314443057542703, | |
| "loss": 1.2643, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.13354394586477897, | |
| "grad_norm": 0.2768828272819519, | |
| "learning_rate": 0.00015303464490059506, | |
| "loss": 1.1444, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.13369995514733127, | |
| "grad_norm": 0.3006720542907715, | |
| "learning_rate": 0.00015292477021636497, | |
| "loss": 1.2172, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 0.13385596442988357, | |
| "grad_norm": 0.24407751858234406, | |
| "learning_rate": 0.0001528148067071423, | |
| "loss": 0.9457, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.1340119737124359, | |
| "grad_norm": 0.25638723373413086, | |
| "learning_rate": 0.00015270475455748166, | |
| "loss": 1.1478, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 0.1341679829949882, | |
| "grad_norm": 0.24834637343883514, | |
| "learning_rate": 0.00015259461395208628, | |
| "loss": 0.9835, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.13432399227754052, | |
| "grad_norm": 0.2611735463142395, | |
| "learning_rate": 0.00015248438507580806, | |
| "loss": 1.125, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 0.13448000156009282, | |
| "grad_norm": 0.3239066004753113, | |
| "learning_rate": 0.00015237406811364682, | |
| "loss": 1.1973, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.13463601084264515, | |
| "grad_norm": 0.2662723958492279, | |
| "learning_rate": 0.0001522636632507504, | |
| "loss": 1.1115, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 0.13479202012519745, | |
| "grad_norm": 0.26053330302238464, | |
| "learning_rate": 0.00015215317067241414, | |
| "loss": 1.0885, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.13494802940774975, | |
| "grad_norm": 0.337984561920166, | |
| "learning_rate": 0.00015204259056408046, | |
| "loss": 0.8782, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.13510403869030208, | |
| "grad_norm": 0.2965889871120453, | |
| "learning_rate": 0.00015193192311133884, | |
| "loss": 1.3198, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.13526004797285437, | |
| "grad_norm": 0.3056474030017853, | |
| "learning_rate": 0.00015182116849992526, | |
| "loss": 1.5133, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 0.1354160572554067, | |
| "grad_norm": 0.29193446040153503, | |
| "learning_rate": 0.00015171032691572206, | |
| "loss": 1.2365, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.135572066537959, | |
| "grad_norm": 0.28123265504837036, | |
| "learning_rate": 0.00015159939854475743, | |
| "loss": 1.1654, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 0.13572807582051133, | |
| "grad_norm": 0.3033466041088104, | |
| "learning_rate": 0.00015148838357320537, | |
| "loss": 1.5473, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.13588408510306363, | |
| "grad_norm": 0.26069045066833496, | |
| "learning_rate": 0.00015137728218738502, | |
| "loss": 1.2213, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 0.13604009438561596, | |
| "grad_norm": 0.3010377883911133, | |
| "learning_rate": 0.0001512660945737608, | |
| "loss": 1.1906, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.13619610366816826, | |
| "grad_norm": 0.2615121304988861, | |
| "learning_rate": 0.00015115482091894165, | |
| "loss": 1.0807, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 0.13635211295072056, | |
| "grad_norm": 0.27064162492752075, | |
| "learning_rate": 0.00015104346140968095, | |
| "loss": 1.3376, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.13650812223327288, | |
| "grad_norm": 0.26106327772140503, | |
| "learning_rate": 0.00015093201623287631, | |
| "loss": 1.2357, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.13666413151582518, | |
| "grad_norm": 0.26505109667778015, | |
| "learning_rate": 0.00015082048557556893, | |
| "loss": 1.4311, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.1368201407983775, | |
| "grad_norm": 0.2965877950191498, | |
| "learning_rate": 0.00015070886962494358, | |
| "loss": 1.3246, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 0.1369761500809298, | |
| "grad_norm": 0.3173799216747284, | |
| "learning_rate": 0.0001505971685683282, | |
| "loss": 1.4795, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.13713215936348214, | |
| "grad_norm": 0.2562354505062103, | |
| "learning_rate": 0.00015048538259319346, | |
| "loss": 1.0112, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 0.13728816864603444, | |
| "grad_norm": 0.2736887037754059, | |
| "learning_rate": 0.00015037351188715265, | |
| "loss": 1.3539, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.13744417792858676, | |
| "grad_norm": 0.30376073718070984, | |
| "learning_rate": 0.00015026155663796123, | |
| "loss": 1.2837, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 0.13760018721113906, | |
| "grad_norm": 0.3052879869937897, | |
| "learning_rate": 0.00015014951703351653, | |
| "loss": 1.3994, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.13775619649369136, | |
| "grad_norm": 0.25414812564849854, | |
| "learning_rate": 0.00015003739326185751, | |
| "loss": 0.9258, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 0.1379122057762437, | |
| "grad_norm": 0.33165043592453003, | |
| "learning_rate": 0.00014992518551116434, | |
| "loss": 1.4427, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.138068215058796, | |
| "grad_norm": 0.2764113247394562, | |
| "learning_rate": 0.00014981289396975817, | |
| "loss": 1.3084, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.13822422434134832, | |
| "grad_norm": 0.3221314251422882, | |
| "learning_rate": 0.0001497005188261007, | |
| "loss": 1.0262, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.13838023362390062, | |
| "grad_norm": 0.24285611510276794, | |
| "learning_rate": 0.0001495880602687941, | |
| "loss": 1.1275, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 0.13853624290645294, | |
| "grad_norm": 0.27305787801742554, | |
| "learning_rate": 0.00014947551848658034, | |
| "loss": 1.3409, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.13869225218900524, | |
| "grad_norm": 0.29822468757629395, | |
| "learning_rate": 0.00014936289366834123, | |
| "loss": 1.3696, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 0.13884826147155757, | |
| "grad_norm": 0.259112685918808, | |
| "learning_rate": 0.00014925018600309785, | |
| "loss": 1.2456, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.13900427075410987, | |
| "grad_norm": 0.28749990463256836, | |
| "learning_rate": 0.00014913739568001033, | |
| "loss": 1.2809, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 0.13916028003666217, | |
| "grad_norm": 0.24120725691318512, | |
| "learning_rate": 0.0001490245228883776, | |
| "loss": 1.1092, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.1393162893192145, | |
| "grad_norm": 0.2791595160961151, | |
| "learning_rate": 0.0001489115678176369, | |
| "loss": 1.024, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 0.1394722986017668, | |
| "grad_norm": 0.260062038898468, | |
| "learning_rate": 0.00014879853065736365, | |
| "loss": 1.1766, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.13962830788431912, | |
| "grad_norm": 0.2642684280872345, | |
| "learning_rate": 0.00014868541159727096, | |
| "loss": 1.3869, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.13978431716687142, | |
| "grad_norm": 0.2463667243719101, | |
| "learning_rate": 0.00014857221082720948, | |
| "loss": 1.0662, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.13994032644942375, | |
| "grad_norm": 0.2916738986968994, | |
| "learning_rate": 0.0001484589285371669, | |
| "loss": 1.3209, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 0.14009633573197605, | |
| "grad_norm": 0.27236512303352356, | |
| "learning_rate": 0.0001483455649172678, | |
| "loss": 1.1833, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.14025234501452835, | |
| "grad_norm": 0.2619946002960205, | |
| "learning_rate": 0.0001482321201577733, | |
| "loss": 1.3137, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 0.14040835429708068, | |
| "grad_norm": 0.31396883726119995, | |
| "learning_rate": 0.00014811859444908052, | |
| "loss": 1.3727, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.14056436357963298, | |
| "grad_norm": 0.25572189688682556, | |
| "learning_rate": 0.0001480049879817226, | |
| "loss": 1.1046, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 0.1407203728621853, | |
| "grad_norm": 0.2937905490398407, | |
| "learning_rate": 0.0001478913009463682, | |
| "loss": 1.3542, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.1408763821447376, | |
| "grad_norm": 0.253520131111145, | |
| "learning_rate": 0.00014777753353382119, | |
| "loss": 1.2329, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 0.14103239142728993, | |
| "grad_norm": 0.32491999864578247, | |
| "learning_rate": 0.00014766368593502026, | |
| "loss": 1.3285, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.14118840070984223, | |
| "grad_norm": 0.2527139484882355, | |
| "learning_rate": 0.00014754975834103877, | |
| "loss": 1.1277, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.14134440999239456, | |
| "grad_norm": 0.275272399187088, | |
| "learning_rate": 0.00014743575094308431, | |
| "loss": 1.4177, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.14150041927494686, | |
| "grad_norm": 0.26013612747192383, | |
| "learning_rate": 0.0001473216639324984, | |
| "loss": 1.2476, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 0.14165642855749916, | |
| "grad_norm": 0.28431418538093567, | |
| "learning_rate": 0.0001472074975007562, | |
| "loss": 1.3947, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.1418124378400515, | |
| "grad_norm": 0.2629927396774292, | |
| "learning_rate": 0.0001470932518394661, | |
| "loss": 1.1587, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 0.14196844712260379, | |
| "grad_norm": 0.2944284975528717, | |
| "learning_rate": 0.00014697892714036958, | |
| "loss": 1.342, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.1421244564051561, | |
| "grad_norm": 0.31365662813186646, | |
| "learning_rate": 0.00014686452359534066, | |
| "loss": 1.4326, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 0.1422804656877084, | |
| "grad_norm": 0.255875825881958, | |
| "learning_rate": 0.0001467500413963857, | |
| "loss": 1.2305, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.14243647497026074, | |
| "grad_norm": 0.2717350423336029, | |
| "learning_rate": 0.00014663548073564316, | |
| "loss": 1.1965, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 0.14259248425281304, | |
| "grad_norm": 0.28059136867523193, | |
| "learning_rate": 0.00014652084180538302, | |
| "loss": 1.3361, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.14274849353536537, | |
| "grad_norm": 0.2790951430797577, | |
| "learning_rate": 0.00014640612479800686, | |
| "loss": 1.2785, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.14290450281791767, | |
| "grad_norm": 0.24599488079547882, | |
| "learning_rate": 0.00014629132990604706, | |
| "loss": 1.2433, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.14306051210046997, | |
| "grad_norm": 0.288792222738266, | |
| "learning_rate": 0.00014617645732216685, | |
| "loss": 1.1779, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 0.1432165213830223, | |
| "grad_norm": 0.3035881221294403, | |
| "learning_rate": 0.00014606150723915984, | |
| "loss": 1.3885, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.1433725306655746, | |
| "grad_norm": 0.28884077072143555, | |
| "learning_rate": 0.00014594647984994964, | |
| "loss": 1.3079, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 0.14352853994812692, | |
| "grad_norm": 0.26054033637046814, | |
| "learning_rate": 0.00014583137534758967, | |
| "loss": 1.1897, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.14368454923067922, | |
| "grad_norm": 0.31249237060546875, | |
| "learning_rate": 0.00014571619392526278, | |
| "loss": 1.4518, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 0.14384055851323155, | |
| "grad_norm": 0.27947118878364563, | |
| "learning_rate": 0.0001456009357762809, | |
| "loss": 1.2305, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.14399656779578385, | |
| "grad_norm": 0.2928619980812073, | |
| "learning_rate": 0.00014548560109408466, | |
| "loss": 1.3645, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 0.14415257707833617, | |
| "grad_norm": 0.2735868990421295, | |
| "learning_rate": 0.00014537019007224324, | |
| "loss": 1.4351, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.14430858636088847, | |
| "grad_norm": 0.30757883191108704, | |
| "learning_rate": 0.00014525470290445392, | |
| "loss": 1.4073, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.14446459564344077, | |
| "grad_norm": 0.28719013929367065, | |
| "learning_rate": 0.00014513913978454168, | |
| "loss": 1.2918, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.1446206049259931, | |
| "grad_norm": 0.2720332145690918, | |
| "learning_rate": 0.00014502350090645917, | |
| "loss": 1.2763, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 0.1447766142085454, | |
| "grad_norm": 0.24720966815948486, | |
| "learning_rate": 0.000144907786464286, | |
| "loss": 1.0549, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.14493262349109773, | |
| "grad_norm": 0.3164946138858795, | |
| "learning_rate": 0.0001447919966522287, | |
| "loss": 1.1007, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 0.14508863277365003, | |
| "grad_norm": 0.2940044105052948, | |
| "learning_rate": 0.00014467613166462023, | |
| "loss": 1.2818, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.14524464205620236, | |
| "grad_norm": 0.34050655364990234, | |
| "learning_rate": 0.00014456019169591978, | |
| "loss": 1.2618, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 0.14540065133875466, | |
| "grad_norm": 0.24612417817115784, | |
| "learning_rate": 0.0001444441769407124, | |
| "loss": 0.991, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.14555666062130695, | |
| "grad_norm": 0.2636529505252838, | |
| "learning_rate": 0.00014432808759370854, | |
| "loss": 1.4259, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 0.14571266990385928, | |
| "grad_norm": 0.2628234624862671, | |
| "learning_rate": 0.00014421192384974396, | |
| "loss": 1.2545, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.14586867918641158, | |
| "grad_norm": 0.2733708918094635, | |
| "learning_rate": 0.00014409568590377918, | |
| "loss": 1.1442, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.1460246884689639, | |
| "grad_norm": 0.24912774562835693, | |
| "learning_rate": 0.0001439793739508994, | |
| "loss": 1.039, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.1461806977515162, | |
| "grad_norm": 0.2927952706813812, | |
| "learning_rate": 0.00014386298818631386, | |
| "loss": 1.179, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 0.14633670703406854, | |
| "grad_norm": 0.29066377878189087, | |
| "learning_rate": 0.0001437465288053558, | |
| "loss": 1.2024, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.14649271631662084, | |
| "grad_norm": 0.2862846553325653, | |
| "learning_rate": 0.00014362999600348196, | |
| "loss": 1.1401, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 0.14664872559917316, | |
| "grad_norm": 0.3009769022464752, | |
| "learning_rate": 0.00014351338997627234, | |
| "loss": 1.3966, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.14680473488172546, | |
| "grad_norm": 0.31753668189048767, | |
| "learning_rate": 0.00014339671091942978, | |
| "loss": 1.4626, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 0.14696074416427776, | |
| "grad_norm": 0.28623080253601074, | |
| "learning_rate": 0.0001432799590287797, | |
| "loss": 1.2841, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.1471167534468301, | |
| "grad_norm": 0.3344881534576416, | |
| "learning_rate": 0.00014316313450026986, | |
| "loss": 1.5589, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 0.1472727627293824, | |
| "grad_norm": 0.3132301867008209, | |
| "learning_rate": 0.00014304623752996973, | |
| "loss": 1.4286, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.14742877201193472, | |
| "grad_norm": 0.299078106880188, | |
| "learning_rate": 0.00014292926831407061, | |
| "loss": 1.2099, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.14758478129448702, | |
| "grad_norm": 0.27058905363082886, | |
| "learning_rate": 0.0001428122270488848, | |
| "loss": 1.2331, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.14774079057703934, | |
| "grad_norm": 0.3202461004257202, | |
| "learning_rate": 0.00014269511393084572, | |
| "loss": 1.0677, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 0.14789679985959164, | |
| "grad_norm": 0.3005964756011963, | |
| "learning_rate": 0.00014257792915650728, | |
| "loss": 1.3382, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.14805280914214397, | |
| "grad_norm": 0.28587067127227783, | |
| "learning_rate": 0.00014246067292254366, | |
| "loss": 1.2216, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 0.14820881842469627, | |
| "grad_norm": 0.27515730261802673, | |
| "learning_rate": 0.00014234334542574906, | |
| "loss": 1.1608, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.14836482770724857, | |
| "grad_norm": 0.26588740944862366, | |
| "learning_rate": 0.00014222594686303706, | |
| "loss": 1.1547, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 0.1485208369898009, | |
| "grad_norm": 0.3122014105319977, | |
| "learning_rate": 0.00014210847743144087, | |
| "loss": 1.3642, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.1486768462723532, | |
| "grad_norm": 0.34852224588394165, | |
| "learning_rate": 0.00014199093732811225, | |
| "loss": 1.4751, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 0.14883285555490552, | |
| "grad_norm": 0.2674144208431244, | |
| "learning_rate": 0.00014187332675032188, | |
| "loss": 1.2941, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.14898886483745782, | |
| "grad_norm": 0.30863744020462036, | |
| "learning_rate": 0.00014175564589545854, | |
| "loss": 1.298, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.14914487412001015, | |
| "grad_norm": 0.26412221789360046, | |
| "learning_rate": 0.00014163789496102902, | |
| "loss": 1.218, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.14930088340256245, | |
| "grad_norm": 0.2920873761177063, | |
| "learning_rate": 0.0001415200741446577, | |
| "loss": 1.5198, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 0.14945689268511475, | |
| "grad_norm": 0.29869547486305237, | |
| "learning_rate": 0.00014140218364408632, | |
| "loss": 1.3896, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.14961290196766708, | |
| "grad_norm": 0.2696417570114136, | |
| "learning_rate": 0.00014128422365717347, | |
| "loss": 1.2046, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 0.14976891125021938, | |
| "grad_norm": 0.27298402786254883, | |
| "learning_rate": 0.0001411661943818944, | |
| "loss": 1.3599, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.1499249205327717, | |
| "grad_norm": 0.27962544560432434, | |
| "learning_rate": 0.0001410480960163407, | |
| "loss": 1.25, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 0.150080929815324, | |
| "grad_norm": 0.2612510323524475, | |
| "learning_rate": 0.00014092992875871979, | |
| "loss": 1.1053, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.15023693909787633, | |
| "grad_norm": 0.27618667483329773, | |
| "learning_rate": 0.00014081169280735488, | |
| "loss": 1.3871, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 0.15039294838042863, | |
| "grad_norm": 0.24976608157157898, | |
| "learning_rate": 0.00014069338836068433, | |
| "loss": 1.2613, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.15054895766298096, | |
| "grad_norm": 0.267610102891922, | |
| "learning_rate": 0.00014057501561726157, | |
| "loss": 1.0631, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.15070496694553326, | |
| "grad_norm": 0.29677531123161316, | |
| "learning_rate": 0.00014045657477575448, | |
| "loss": 1.3567, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.15086097622808556, | |
| "grad_norm": 0.29539185762405396, | |
| "learning_rate": 0.0001403380660349455, | |
| "loss": 1.1386, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 0.15101698551063789, | |
| "grad_norm": 0.2691122889518738, | |
| "learning_rate": 0.00014021948959373076, | |
| "loss": 1.1089, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.15117299479319019, | |
| "grad_norm": 0.24394790828227997, | |
| "learning_rate": 0.0001401008456511202, | |
| "loss": 1.1893, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 0.1513290040757425, | |
| "grad_norm": 0.2849481403827667, | |
| "learning_rate": 0.0001399821344062369, | |
| "loss": 1.4775, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.1514850133582948, | |
| "grad_norm": 0.2634568512439728, | |
| "learning_rate": 0.00013986335605831705, | |
| "loss": 1.1655, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 0.15164102264084714, | |
| "grad_norm": 0.269879013299942, | |
| "learning_rate": 0.00013974451080670934, | |
| "loss": 1.2047, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.15179703192339944, | |
| "grad_norm": 0.27636033296585083, | |
| "learning_rate": 0.0001396255988508748, | |
| "loss": 1.2987, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 0.15195304120595177, | |
| "grad_norm": 0.2572225332260132, | |
| "learning_rate": 0.00013950662039038643, | |
| "loss": 1.3322, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.15210905048850407, | |
| "grad_norm": 0.2573801279067993, | |
| "learning_rate": 0.00013938757562492873, | |
| "loss": 1.2547, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.15226505977105637, | |
| "grad_norm": 0.3160158395767212, | |
| "learning_rate": 0.00013926846475429766, | |
| "loss": 1.5537, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.1524210690536087, | |
| "grad_norm": 0.30125337839126587, | |
| "learning_rate": 0.00013914928797839995, | |
| "loss": 1.0853, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 0.152577078336161, | |
| "grad_norm": 0.25772640109062195, | |
| "learning_rate": 0.0001390300454972531, | |
| "loss": 1.198, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.15273308761871332, | |
| "grad_norm": 0.257586270570755, | |
| "learning_rate": 0.0001389107375109848, | |
| "loss": 1.086, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 0.15288909690126562, | |
| "grad_norm": 0.2763863205909729, | |
| "learning_rate": 0.00013879136421983266, | |
| "loss": 1.2639, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.15304510618381795, | |
| "grad_norm": 0.2751125991344452, | |
| "learning_rate": 0.00013867192582414393, | |
| "loss": 1.2473, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 0.15320111546637025, | |
| "grad_norm": 0.3138543367385864, | |
| "learning_rate": 0.0001385524225243751, | |
| "loss": 1.3107, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.15335712474892257, | |
| "grad_norm": 0.27820733189582825, | |
| "learning_rate": 0.00013843285452109166, | |
| "loss": 1.048, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 0.15351313403147487, | |
| "grad_norm": 0.25756746530532837, | |
| "learning_rate": 0.00013831322201496757, | |
| "loss": 1.0374, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.15366914331402717, | |
| "grad_norm": 0.332603394985199, | |
| "learning_rate": 0.0001381935252067852, | |
| "loss": 1.3359, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.1538251525965795, | |
| "grad_norm": 0.33936744928359985, | |
| "learning_rate": 0.00013807376429743467, | |
| "loss": 1.5814, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.1539811618791318, | |
| "grad_norm": 0.2748062014579773, | |
| "learning_rate": 0.00013795393948791383, | |
| "loss": 1.201, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 0.15413717116168413, | |
| "grad_norm": 0.26038771867752075, | |
| "learning_rate": 0.0001378340509793277, | |
| "loss": 1.2087, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.15429318044423643, | |
| "grad_norm": 0.24746748805046082, | |
| "learning_rate": 0.00013771409897288822, | |
| "loss": 1.0487, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 0.15444918972678875, | |
| "grad_norm": 0.270280122756958, | |
| "learning_rate": 0.0001375940836699139, | |
| "loss": 1.1529, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.15460519900934105, | |
| "grad_norm": 0.28278234601020813, | |
| "learning_rate": 0.00013747400527182953, | |
| "loss": 1.4292, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 0.15476120829189335, | |
| "grad_norm": 0.3091171681880951, | |
| "learning_rate": 0.0001373538639801657, | |
| "loss": 1.2118, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.15491721757444568, | |
| "grad_norm": 0.264275461435318, | |
| "learning_rate": 0.0001372336599965586, | |
| "loss": 1.2727, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 0.15507322685699798, | |
| "grad_norm": 0.3125738799571991, | |
| "learning_rate": 0.00013711339352274966, | |
| "loss": 1.3389, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.1552292361395503, | |
| "grad_norm": 0.2750801146030426, | |
| "learning_rate": 0.0001369930647605852, | |
| "loss": 1.1031, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.1553852454221026, | |
| "grad_norm": 0.274777889251709, | |
| "learning_rate": 0.00013687267391201605, | |
| "loss": 1.4329, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.15554125470465494, | |
| "grad_norm": 0.28475117683410645, | |
| "learning_rate": 0.00013675222117909717, | |
| "loss": 1.1914, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 0.15569726398720724, | |
| "grad_norm": 0.27364879846572876, | |
| "learning_rate": 0.00013663170676398752, | |
| "loss": 1.1511, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.15585327326975956, | |
| "grad_norm": 0.310995489358902, | |
| "learning_rate": 0.00013651113086894952, | |
| "loss": 1.0349, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 0.15600928255231186, | |
| "grad_norm": 0.2910314202308655, | |
| "learning_rate": 0.00013639049369634876, | |
| "loss": 1.3302, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.15600928255231186, | |
| "eval_loss": 1.2771576642990112, | |
| "eval_runtime": 110.8263, | |
| "eval_samples_per_second": 38.556, | |
| "eval_steps_per_second": 4.827, | |
| "step": 1000 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 2500, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 4.144559113202074e+16, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |