diff --git "a/trainer_state.json" "b/trainer_state.json" --- "a/trainer_state.json" +++ "b/trainer_state.json" @@ -1,8635 +1,13615 @@ { + "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, - "epoch": 4.982278481012658, + "epoch": 5.0, "eval_steps": 500, - "global_step": 1230, + "global_step": 1235, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.004050632911392405, - "grad_norm": 6.616540841176689, - "learning_rate": 3.2520325203252037e-07, - "loss": 0.799, - "step": 1 + "grad_norm": 6.420329598097033, + "learning_rate": 0.0, + "loss": 0.7925, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.08521983027458191, + "step": 1, + "valid_targets_mean": 4176.2, + "valid_targets_min": 1259 }, { "epoch": 0.00810126582278481, - "grad_norm": 6.474323454882261, - "learning_rate": 6.504065040650407e-07, - "loss": 0.7966, - "step": 2 + "grad_norm": 6.227117765992412, + "learning_rate": 3.2258064516129035e-07, + "loss": 0.7813, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0910545140504837, + "step": 2, + "valid_targets_mean": 4499.6, + "valid_targets_min": 1332 }, { "epoch": 0.012151898734177215, - "grad_norm": 6.590202665967519, - "learning_rate": 9.75609756097561e-07, - "loss": 0.8157, - "step": 3 + "grad_norm": 6.456985603229851, + "learning_rate": 6.451612903225807e-07, + "loss": 0.8104, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.08171398937702179, + "step": 3, + "valid_targets_mean": 3918.3, + "valid_targets_min": 1195 }, { "epoch": 0.01620253164556962, - "grad_norm": 6.4407197325404395, - "learning_rate": 1.3008130081300815e-06, - "loss": 0.8138, - "step": 4 + "grad_norm": 6.249791581499031, + "learning_rate": 9.67741935483871e-07, + "loss": 0.7958, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0754520520567894, + "step": 4, + "valid_targets_mean": 3975.1, + "valid_targets_min": 1365 }, { "epoch": 0.020253164556962026, - "grad_norm": 6.158479270927061, - "learning_rate": 1.6260162601626018e-06, - "loss": 0.7721, - "step": 5 + "grad_norm": 5.810678301715652, + "learning_rate": 1.2903225806451614e-06, + "loss": 0.7392, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.09185856580734253, + "step": 5, + "valid_targets_mean": 4353.0, + "valid_targets_min": 1210 }, { "epoch": 0.02430379746835443, - "grad_norm": 5.677102963188688, - "learning_rate": 1.951219512195122e-06, - "loss": 0.7531, - "step": 6 + "grad_norm": 5.41391649813407, + "learning_rate": 1.6129032258064516e-06, + "loss": 0.7529, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.08282750844955444, + "step": 6, + "valid_targets_mean": 3698.6, + "valid_targets_min": 1283 }, { "epoch": 0.028354430379746835, - "grad_norm": 4.676756291527093, - "learning_rate": 2.2764227642276426e-06, - "loss": 0.7373, - "step": 7 + "grad_norm": 4.424776890998065, + "learning_rate": 1.935483870967742e-06, + "loss": 0.7235, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.11384673416614532, + "step": 7, + "valid_targets_mean": 5868.5, + "valid_targets_min": 939 }, { "epoch": 0.03240506329113924, - "grad_norm": 4.343262501135752, - "learning_rate": 2.601626016260163e-06, - "loss": 0.7121, - "step": 8 + "grad_norm": 4.077353043076277, + "learning_rate": 2.2580645161290324e-06, + "loss": 0.7017, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.10583869367837906, + "step": 8, + "valid_targets_mean": 5800.6, + "valid_targets_min": 1707 }, { "epoch": 0.03645569620253165, - "grad_norm": 2.6756698165724946, - "learning_rate": 2.926829268292683e-06, - "loss": 0.6824, - "step": 9 + "grad_norm": 2.2756792698357886, + "learning_rate": 2.580645161290323e-06, + "loss": 0.6685, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06618285179138184, + "step": 9, + "valid_targets_mean": 3563.2, + "valid_targets_min": 875 }, { "epoch": 0.04050632911392405, - "grad_norm": 2.3675911494991886, - "learning_rate": 3.2520325203252037e-06, - "loss": 0.6685, - "step": 10 + "grad_norm": 2.1200029277675614, + "learning_rate": 2.903225806451613e-06, + "loss": 0.6746, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.08664921671152115, + "step": 10, + "valid_targets_mean": 5192.1, + "valid_targets_min": 1433 }, { "epoch": 0.044556962025316456, - "grad_norm": 2.3311962373874815, - "learning_rate": 3.577235772357724e-06, - "loss": 0.645, - "step": 11 + "grad_norm": 1.9809407766421865, + "learning_rate": 3.225806451612903e-06, + "loss": 0.6465, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.07697989046573639, + "step": 11, + "valid_targets_mean": 4453.6, + "valid_targets_min": 1029 }, { "epoch": 0.04860759493670886, - "grad_norm": 3.205846081925055, - "learning_rate": 3.902439024390244e-06, - "loss": 0.6237, - "step": 12 + "grad_norm": 3.3136354683998746, + "learning_rate": 3.548387096774194e-06, + "loss": 0.6338, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.08591389656066895, + "step": 12, + "valid_targets_mean": 4982.2, + "valid_targets_min": 1056 }, { "epoch": 0.052658227848101265, - "grad_norm": 3.3083916500841353, - "learning_rate": 4.227642276422765e-06, - "loss": 0.6387, - "step": 13 + "grad_norm": 3.6619988249855, + "learning_rate": 3.870967741935484e-06, + "loss": 0.6646, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.09026002883911133, + "step": 13, + "valid_targets_mean": 4544.6, + "valid_targets_min": 722 }, { "epoch": 0.05670886075949367, - "grad_norm": 3.0986106516707808, - "learning_rate": 4.552845528455285e-06, - "loss": 0.5988, - "step": 14 + "grad_norm": 3.2077224891053713, + "learning_rate": 4.193548387096774e-06, + "loss": 0.6125, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.07291078567504883, + "step": 14, + "valid_targets_mean": 4802.9, + "valid_targets_min": 1215 }, { "epoch": 0.060759493670886074, - "grad_norm": 3.0968269938757107, - "learning_rate": 4.8780487804878055e-06, - "loss": 0.6203, - "step": 15 + "grad_norm": 3.4045427950280165, + "learning_rate": 4.516129032258065e-06, + "loss": 0.646, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.08019945025444031, + "step": 15, + "valid_targets_mean": 4426.5, + "valid_targets_min": 1411 }, { "epoch": 0.06481012658227848, - "grad_norm": 2.460478449014551, - "learning_rate": 5.203252032520326e-06, - "loss": 0.5746, - "step": 16 + "grad_norm": 2.636319161292139, + "learning_rate": 4.838709677419355e-06, + "loss": 0.5962, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.07337251305580139, + "step": 16, + "valid_targets_mean": 4542.8, + "valid_targets_min": 1060 }, { "epoch": 0.06886075949367089, - "grad_norm": 2.3493430269108955, - "learning_rate": 5.528455284552846e-06, - "loss": 0.5795, - "step": 17 + "grad_norm": 2.542957173923556, + "learning_rate": 5.161290322580646e-06, + "loss": 0.601, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.10810995101928711, + "step": 17, + "valid_targets_mean": 6028.9, + "valid_targets_min": 1560 }, { "epoch": 0.0729113924050633, - "grad_norm": 1.9918845141289823, - "learning_rate": 5.853658536585366e-06, - "loss": 0.5698, - "step": 18 + "grad_norm": 2.0489652247347863, + "learning_rate": 5.483870967741935e-06, + "loss": 0.5887, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.08178861439228058, + "step": 18, + "valid_targets_mean": 5927.2, + "valid_targets_min": 1886 }, { "epoch": 0.0769620253164557, - "grad_norm": 1.6489527830554676, - "learning_rate": 6.178861788617887e-06, - "loss": 0.5487, - "step": 19 + "grad_norm": 1.5141315525278707, + "learning_rate": 5.806451612903226e-06, + "loss": 0.5803, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.07792671024799347, + "step": 19, + "valid_targets_mean": 5318.4, + "valid_targets_min": 1014 }, { "epoch": 0.0810126582278481, - "grad_norm": 1.572885949701176, - "learning_rate": 6.504065040650407e-06, - "loss": 0.5224, - "step": 20 + "grad_norm": 1.4189201796367261, + "learning_rate": 6.129032258064517e-06, + "loss": 0.5588, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05552545189857483, + "step": 20, + "valid_targets_mean": 3590.8, + "valid_targets_min": 996 }, { "epoch": 0.08506329113924051, - "grad_norm": 1.451712249307161, - "learning_rate": 6.829268292682928e-06, - "loss": 0.5247, - "step": 21 + "grad_norm": 1.4030946456260018, + "learning_rate": 6.451612903225806e-06, + "loss": 0.5513, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06783956289291382, + "step": 21, + "valid_targets_mean": 4338.8, + "valid_targets_min": 1382 }, { "epoch": 0.08911392405063291, - "grad_norm": 1.4054656377299188, - "learning_rate": 7.154471544715448e-06, - "loss": 0.5041, - "step": 22 + "grad_norm": 1.3837911133448608, + "learning_rate": 6.774193548387097e-06, + "loss": 0.5383, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0490473248064518, + "step": 22, + "valid_targets_mean": 3343.5, + "valid_targets_min": 1526 }, { "epoch": 0.09316455696202532, - "grad_norm": 1.2530723859668358, - "learning_rate": 7.4796747967479676e-06, - "loss": 0.5149, - "step": 23 + "grad_norm": 1.1409375146335552, + "learning_rate": 7.096774193548388e-06, + "loss": 0.5404, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.07465570420026779, + "step": 23, + "valid_targets_mean": 5280.9, + "valid_targets_min": 891 }, { "epoch": 0.09721518987341772, - "grad_norm": 1.0837888171188292, - "learning_rate": 7.804878048780489e-06, - "loss": 0.5091, - "step": 24 + "grad_norm": 0.921464895025812, + "learning_rate": 7.4193548387096784e-06, + "loss": 0.5473, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.07144070416688919, + "step": 24, + "valid_targets_mean": 4877.6, + "valid_targets_min": 1659 }, { "epoch": 0.10126582278481013, - "grad_norm": 1.0771838844672532, - "learning_rate": 8.130081300813009e-06, - "loss": 0.4944, - "step": 25 + "grad_norm": 1.1027739754347268, + "learning_rate": 7.741935483870968e-06, + "loss": 0.5181, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05358847230672836, + "step": 25, + "valid_targets_mean": 4054.9, + "valid_targets_min": 1409 }, { "epoch": 0.10531645569620253, - "grad_norm": 0.9066662533989501, - "learning_rate": 8.45528455284553e-06, - "loss": 0.479, - "step": 26 + "grad_norm": 0.978422105693308, + "learning_rate": 8.064516129032258e-06, + "loss": 0.5043, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05265155807137489, + "step": 26, + "valid_targets_mean": 4199.1, + "valid_targets_min": 1048 }, { "epoch": 0.10936708860759493, - "grad_norm": 0.8248319063901174, - "learning_rate": 8.78048780487805e-06, - "loss": 0.4866, - "step": 27 + "grad_norm": 0.8264186306920859, + "learning_rate": 8.387096774193549e-06, + "loss": 0.5111, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.047893133014440536, + "step": 27, + "valid_targets_mean": 4003.8, + "valid_targets_min": 1485 }, { "epoch": 0.11341772151898734, - "grad_norm": 0.8417065051344589, - "learning_rate": 9.10569105691057e-06, - "loss": 0.488, - "step": 28 + "grad_norm": 0.8702558388946918, + "learning_rate": 8.70967741935484e-06, + "loss": 0.5237, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.07174102216959, + "step": 28, + "valid_targets_mean": 4451.9, + "valid_targets_min": 1003 }, { "epoch": 0.11746835443037974, - "grad_norm": 0.7213171607490497, - "learning_rate": 9.43089430894309e-06, - "loss": 0.4747, - "step": 29 + "grad_norm": 0.7688421729857465, + "learning_rate": 9.03225806451613e-06, + "loss": 0.5, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.07423195987939835, + "step": 29, + "valid_targets_mean": 5388.8, + "valid_targets_min": 1313 }, { "epoch": 0.12151898734177215, - "grad_norm": 0.750611509191151, - "learning_rate": 9.756097560975611e-06, - "loss": 0.4627, - "step": 30 + "grad_norm": 0.8462036679650163, + "learning_rate": 9.35483870967742e-06, + "loss": 0.4917, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06195066124200821, + "step": 30, + "valid_targets_mean": 4936.8, + "valid_targets_min": 1281 }, { "epoch": 0.12556962025316457, - "grad_norm": 0.6754163606343294, - "learning_rate": 1.008130081300813e-05, - "loss": 0.4588, - "step": 31 + "grad_norm": 0.6471729036969938, + "learning_rate": 9.67741935483871e-06, + "loss": 0.4856, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.052095141261816025, + "step": 31, + "valid_targets_mean": 3925.6, + "valid_targets_min": 982 }, { "epoch": 0.12962025316455697, - "grad_norm": 0.6472243449527941, - "learning_rate": 1.0406504065040652e-05, - "loss": 0.4663, - "step": 32 + "grad_norm": 0.6236907607706728, + "learning_rate": 1e-05, + "loss": 0.4888, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.07329850643873215, + "step": 32, + "valid_targets_mean": 4460.2, + "valid_targets_min": 834 }, { "epoch": 0.13367088607594937, - "grad_norm": 0.6183284435767364, - "learning_rate": 1.0731707317073172e-05, - "loss": 0.4635, - "step": 33 + "grad_norm": 0.7166373529372694, + "learning_rate": 1.0322580645161291e-05, + "loss": 0.4859, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05221894755959511, + "step": 33, + "valid_targets_mean": 5000.0, + "valid_targets_min": 1107 }, { "epoch": 0.13772151898734178, - "grad_norm": 0.6515160702962011, - "learning_rate": 1.1056910569105692e-05, - "loss": 0.4483, - "step": 34 + "grad_norm": 0.6552483827885665, + "learning_rate": 1.0645161290322582e-05, + "loss": 0.4849, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04771258682012558, + "step": 34, + "valid_targets_mean": 3255.8, + "valid_targets_min": 1309 }, { "epoch": 0.14177215189873418, - "grad_norm": 0.5986807187059076, - "learning_rate": 1.1382113821138213e-05, - "loss": 0.4393, - "step": 35 + "grad_norm": 0.5471801158923242, + "learning_rate": 1.096774193548387e-05, + "loss": 0.4738, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05435044690966606, + "step": 35, + "valid_targets_mean": 3794.1, + "valid_targets_min": 1102 }, { "epoch": 0.1458227848101266, - "grad_norm": 0.5715501569153342, - "learning_rate": 1.1707317073170731e-05, - "loss": 0.4402, - "step": 36 + "grad_norm": 0.5734883495938603, + "learning_rate": 1.1290322580645164e-05, + "loss": 0.4693, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0718671903014183, + "step": 36, + "valid_targets_mean": 4801.4, + "valid_targets_min": 970 }, { "epoch": 0.149873417721519, - "grad_norm": 0.5723777681221434, - "learning_rate": 1.2032520325203254e-05, - "loss": 0.4322, - "step": 37 + "grad_norm": 0.6395737529832362, + "learning_rate": 1.1612903225806453e-05, + "loss": 0.4678, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04163165017962456, + "step": 37, + "valid_targets_mean": 3100.6, + "valid_targets_min": 1101 }, { "epoch": 0.1539240506329114, - "grad_norm": 0.5936945690667882, - "learning_rate": 1.2357723577235774e-05, - "loss": 0.4405, - "step": 38 + "grad_norm": 0.5491272351602188, + "learning_rate": 1.1935483870967743e-05, + "loss": 0.4638, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0640963688492775, + "step": 38, + "valid_targets_mean": 5124.9, + "valid_targets_min": 1311 }, { "epoch": 0.1579746835443038, - "grad_norm": 1.569697710037054, - "learning_rate": 1.2682926829268294e-05, - "loss": 0.4337, - "step": 39 + "grad_norm": 0.5858951225783975, + "learning_rate": 1.2258064516129034e-05, + "loss": 0.4633, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0465809591114521, + "step": 39, + "valid_targets_mean": 3319.8, + "valid_targets_min": 1454 }, { "epoch": 0.1620253164556962, - "grad_norm": 0.5830272144805612, - "learning_rate": 1.3008130081300815e-05, - "loss": 0.4425, - "step": 40 + "grad_norm": 0.5078493446863747, + "learning_rate": 1.2580645161290324e-05, + "loss": 0.4794, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.07410909980535507, + "step": 40, + "valid_targets_mean": 5386.2, + "valid_targets_min": 1146 }, { "epoch": 0.1660759493670886, - "grad_norm": 0.5025898006391891, - "learning_rate": 1.3333333333333333e-05, - "loss": 0.4403, - "step": 41 + "grad_norm": 0.6556197909112473, + "learning_rate": 1.2903225806451613e-05, + "loss": 0.4669, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04371774196624756, + "step": 41, + "valid_targets_mean": 3323.4, + "valid_targets_min": 1429 }, { "epoch": 0.17012658227848101, - "grad_norm": 0.5187222659956625, - "learning_rate": 1.3658536585365855e-05, - "loss": 0.4255, - "step": 42 + "grad_norm": 0.4954070482937292, + "learning_rate": 1.3225806451612903e-05, + "loss": 0.4518, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06891210377216339, + "step": 42, + "valid_targets_mean": 4534.8, + "valid_targets_min": 1056 }, { "epoch": 0.17417721518987342, - "grad_norm": 0.5094946140456181, - "learning_rate": 1.3983739837398376e-05, - "loss": 0.4284, - "step": 43 + "grad_norm": 0.6069525492861939, + "learning_rate": 1.3548387096774194e-05, + "loss": 0.4593, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06779725104570389, + "step": 43, + "valid_targets_mean": 4790.6, + "valid_targets_min": 1597 }, { "epoch": 0.17822784810126582, - "grad_norm": 0.48752983370039016, - "learning_rate": 1.4308943089430896e-05, - "loss": 0.4248, - "step": 44 + "grad_norm": 0.5243882432164961, + "learning_rate": 1.3870967741935486e-05, + "loss": 0.4504, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.044036492705345154, + "step": 44, + "valid_targets_mean": 3547.5, + "valid_targets_min": 1343 }, { "epoch": 0.18227848101265823, - "grad_norm": 0.5625781112475009, - "learning_rate": 1.4634146341463415e-05, - "loss": 0.4347, - "step": 45 + "grad_norm": 0.5006000360790126, + "learning_rate": 1.4193548387096776e-05, + "loss": 0.4666, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04936465620994568, + "step": 45, + "valid_targets_mean": 3365.9, + "valid_targets_min": 901 }, { "epoch": 0.18632911392405063, - "grad_norm": 0.5130642860249767, - "learning_rate": 1.4959349593495935e-05, - "loss": 0.432, - "step": 46 + "grad_norm": 0.6455809632367707, + "learning_rate": 1.4516129032258066e-05, + "loss": 0.4608, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05207091197371483, + "step": 46, + "valid_targets_mean": 4055.6, + "valid_targets_min": 925 }, { "epoch": 0.19037974683544304, - "grad_norm": 0.559949497276752, - "learning_rate": 1.528455284552846e-05, - "loss": 0.4338, - "step": 47 + "grad_norm": 0.5346587224021531, + "learning_rate": 1.4838709677419357e-05, + "loss": 0.4522, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05411246418952942, + "step": 47, + "valid_targets_mean": 5068.6, + "valid_targets_min": 1370 }, { "epoch": 0.19443037974683544, - "grad_norm": 0.47266595492741803, - "learning_rate": 1.5609756097560978e-05, - "loss": 0.4171, - "step": 48 + "grad_norm": 0.4627343813521031, + "learning_rate": 1.5161290322580646e-05, + "loss": 0.44, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.07487696409225464, + "step": 48, + "valid_targets_mean": 4657.8, + "valid_targets_min": 775 }, { "epoch": 0.19848101265822785, - "grad_norm": 0.4913030476775072, - "learning_rate": 1.5934959349593496e-05, - "loss": 0.4309, - "step": 49 + "grad_norm": 0.5934753153922505, + "learning_rate": 1.5483870967741936e-05, + "loss": 0.4625, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05100386589765549, + "step": 49, + "valid_targets_mean": 4084.1, + "valid_targets_min": 854 }, { "epoch": 0.20253164556962025, - "grad_norm": 0.47745991231812274, - "learning_rate": 1.6260162601626018e-05, - "loss": 0.4267, - "step": 50 + "grad_norm": 0.5498959721358364, + "learning_rate": 1.5806451612903226e-05, + "loss": 0.4518, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.048061393201351166, + "step": 50, + "valid_targets_mean": 4037.7, + "valid_targets_min": 1066 }, { "epoch": 0.20658227848101265, - "grad_norm": 0.5380315600361899, - "learning_rate": 1.6585365853658537e-05, - "loss": 0.4111, - "step": 51 + "grad_norm": 0.5153204230281868, + "learning_rate": 1.6129032258064517e-05, + "loss": 0.4322, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0640299916267395, + "step": 51, + "valid_targets_mean": 5563.1, + "valid_targets_min": 1095 }, { "epoch": 0.21063291139240506, - "grad_norm": 0.5080920562171014, - "learning_rate": 1.691056910569106e-05, - "loss": 0.4201, - "step": 52 + "grad_norm": 0.4838663333830291, + "learning_rate": 1.6451612903225807e-05, + "loss": 0.4507, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04705619812011719, + "step": 52, + "valid_targets_mean": 4156.1, + "valid_targets_min": 1509 }, { "epoch": 0.21468354430379746, - "grad_norm": 0.5128546367455661, - "learning_rate": 1.7235772357723578e-05, - "loss": 0.4278, - "step": 53 + "grad_norm": 0.5662923684942928, + "learning_rate": 1.6774193548387098e-05, + "loss": 0.454, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06045686453580856, + "step": 53, + "valid_targets_mean": 4749.8, + "valid_targets_min": 1347 }, { "epoch": 0.21873417721518987, - "grad_norm": 0.5460161968046872, - "learning_rate": 1.75609756097561e-05, - "loss": 0.4199, - "step": 54 + "grad_norm": 0.49698527661841424, + "learning_rate": 1.7096774193548388e-05, + "loss": 0.4434, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04809245467185974, + "step": 54, + "valid_targets_mean": 4022.0, + "valid_targets_min": 1166 }, { "epoch": 0.22278481012658227, - "grad_norm": 0.4745231340547568, - "learning_rate": 1.788617886178862e-05, - "loss": 0.4005, - "step": 55 + "grad_norm": 0.5745339911953643, + "learning_rate": 1.741935483870968e-05, + "loss": 0.4205, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05455455929040909, + "step": 55, + "valid_targets_mean": 5457.8, + "valid_targets_min": 1092 }, { "epoch": 0.22683544303797468, - "grad_norm": 0.5331283405788664, - "learning_rate": 1.821138211382114e-05, - "loss": 0.4025, - "step": 56 + "grad_norm": 0.45417778196816644, + "learning_rate": 1.774193548387097e-05, + "loss": 0.4309, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0891185849905014, + "step": 56, + "valid_targets_mean": 6185.5, + "valid_targets_min": 846 }, { "epoch": 0.23088607594936708, - "grad_norm": 0.5344188550194445, - "learning_rate": 1.8536585365853663e-05, - "loss": 0.4074, - "step": 57 + "grad_norm": 0.5571432526973803, + "learning_rate": 1.806451612903226e-05, + "loss": 0.4281, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04635221138596535, + "step": 57, + "valid_targets_mean": 4196.2, + "valid_targets_min": 1384 }, { "epoch": 0.23493670886075949, - "grad_norm": 0.5651363932284691, - "learning_rate": 1.886178861788618e-05, - "loss": 0.4065, - "step": 58 + "grad_norm": 0.5064870878860718, + "learning_rate": 1.838709677419355e-05, + "loss": 0.4379, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04302515089511871, + "step": 58, + "valid_targets_mean": 3178.4, + "valid_targets_min": 995 }, { "epoch": 0.2389873417721519, - "grad_norm": 0.5020133690315554, - "learning_rate": 1.91869918699187e-05, - "loss": 0.4058, - "step": 59 + "grad_norm": 0.5103891846350429, + "learning_rate": 1.870967741935484e-05, + "loss": 0.4347, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06216754764318466, + "step": 59, + "valid_targets_mean": 4990.6, + "valid_targets_min": 1463 }, { "epoch": 0.2430379746835443, - "grad_norm": 0.5536966308046732, - "learning_rate": 1.9512195121951222e-05, - "loss": 0.3977, - "step": 60 + "grad_norm": 0.4708904538255792, + "learning_rate": 1.903225806451613e-05, + "loss": 0.4123, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0597774013876915, + "step": 60, + "valid_targets_mean": 4607.7, + "valid_targets_min": 1447 }, { "epoch": 0.2470886075949367, - "grad_norm": 0.6457517735601038, - "learning_rate": 1.983739837398374e-05, - "loss": 0.4089, - "step": 61 + "grad_norm": 0.42765870481541046, + "learning_rate": 1.935483870967742e-05, + "loss": 0.4434, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0484953448176384, + "step": 61, + "valid_targets_mean": 3862.9, + "valid_targets_min": 894 }, { "epoch": 0.25113924050632913, - "grad_norm": 0.5675906894408289, - "learning_rate": 2.016260162601626e-05, - "loss": 0.4093, - "step": 62 + "grad_norm": 0.48927429380395465, + "learning_rate": 1.967741935483871e-05, + "loss": 0.4343, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.035868145525455475, + "step": 62, + "valid_targets_mean": 3523.4, + "valid_targets_min": 1009 }, { "epoch": 0.25518987341772154, - "grad_norm": 0.6612100414239813, - "learning_rate": 2.048780487804878e-05, - "loss": 0.4114, - "step": 63 + "grad_norm": 0.48899816472824276, + "learning_rate": 2e-05, + "loss": 0.4381, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05651945248246193, + "step": 63, + "valid_targets_mean": 4210.9, + "valid_targets_min": 1536 }, { "epoch": 0.25924050632911394, - "grad_norm": 0.5300873182013642, - "learning_rate": 2.0813008130081303e-05, - "loss": 0.4149, - "step": 64 + "grad_norm": 0.49834976591205526, + "learning_rate": 2.0322580645161292e-05, + "loss": 0.4396, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05972347408533096, + "step": 64, + "valid_targets_mean": 5685.6, + "valid_targets_min": 1421 }, { "epoch": 0.26329113924050634, - "grad_norm": 0.7085856534538018, - "learning_rate": 2.1138211382113822e-05, - "loss": 0.4163, - "step": 65 + "grad_norm": 0.5160400190152858, + "learning_rate": 2.0645161290322582e-05, + "loss": 0.4395, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06172401085495949, + "step": 65, + "valid_targets_mean": 5001.4, + "valid_targets_min": 861 }, { "epoch": 0.26734177215189875, - "grad_norm": 0.5413408512162725, - "learning_rate": 2.1463414634146344e-05, - "loss": 0.3785, - "step": 66 + "grad_norm": 0.5490855404405546, + "learning_rate": 2.0967741935483873e-05, + "loss": 0.4079, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05667848885059357, + "step": 66, + "valid_targets_mean": 3961.4, + "valid_targets_min": 1510 }, { "epoch": 0.27139240506329115, - "grad_norm": 0.6114204625352642, - "learning_rate": 2.1788617886178863e-05, - "loss": 0.3987, - "step": 67 + "grad_norm": 0.5187677100204087, + "learning_rate": 2.1290322580645163e-05, + "loss": 0.4285, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04428575187921524, + "step": 67, + "valid_targets_mean": 3353.8, + "valid_targets_min": 992 }, { "epoch": 0.27544303797468356, - "grad_norm": 0.7978747283598397, - "learning_rate": 2.2113821138211385e-05, - "loss": 0.3976, - "step": 68 + "grad_norm": 0.5524242109730821, + "learning_rate": 2.161290322580645e-05, + "loss": 0.423, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.056597426533699036, + "step": 68, + "valid_targets_mean": 4378.8, + "valid_targets_min": 852 }, { "epoch": 0.27949367088607596, - "grad_norm": 0.6219038558687726, - "learning_rate": 2.2439024390243907e-05, - "loss": 0.3942, - "step": 69 + "grad_norm": 0.6044648675848316, + "learning_rate": 2.193548387096774e-05, + "loss": 0.4135, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.037096381187438965, + "step": 69, + "valid_targets_mean": 3259.7, + "valid_targets_min": 1333 }, { "epoch": 0.28354430379746837, - "grad_norm": 0.8831256567553301, - "learning_rate": 2.2764227642276426e-05, - "loss": 0.4062, - "step": 70 + "grad_norm": 0.6375788523437971, + "learning_rate": 2.225806451612903e-05, + "loss": 0.4445, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.044327497482299805, + "step": 70, + "valid_targets_mean": 3402.1, + "valid_targets_min": 1017 }, { "epoch": 0.28759493670886077, - "grad_norm": 0.8701458499016784, - "learning_rate": 2.3089430894308948e-05, - "loss": 0.4017, - "step": 71 + "grad_norm": 0.7420187981275729, + "learning_rate": 2.2580645161290328e-05, + "loss": 0.4288, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0775013118982315, + "step": 71, + "valid_targets_mean": 5517.9, + "valid_targets_min": 1512 }, { "epoch": 0.2916455696202532, - "grad_norm": 0.6423135673887835, - "learning_rate": 2.3414634146341463e-05, - "loss": 0.3836, - "step": 72 + "grad_norm": 0.594231468226171, + "learning_rate": 2.2903225806451615e-05, + "loss": 0.3961, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05749458447098732, + "step": 72, + "valid_targets_mean": 4559.2, + "valid_targets_min": 1204 }, { "epoch": 0.2956962025316456, - "grad_norm": 0.7863473594259498, - "learning_rate": 2.3739837398373985e-05, - "loss": 0.3998, - "step": 73 + "grad_norm": 0.6876765721414142, + "learning_rate": 2.3225806451612906e-05, + "loss": 0.4297, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.059393130242824554, + "step": 73, + "valid_targets_mean": 4895.0, + "valid_targets_min": 1425 }, { "epoch": 0.299746835443038, - "grad_norm": 0.6856234082896128, - "learning_rate": 2.4065040650406507e-05, - "loss": 0.3802, - "step": 74 + "grad_norm": 0.7313997502730675, + "learning_rate": 2.3548387096774196e-05, + "loss": 0.4106, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.055824242532253265, + "step": 74, + "valid_targets_mean": 5374.9, + "valid_targets_min": 958 }, { "epoch": 0.3037974683544304, - "grad_norm": 0.6427282075061846, - "learning_rate": 2.4390243902439026e-05, - "loss": 0.4136, - "step": 75 + "grad_norm": 0.5852457029426907, + "learning_rate": 2.3870967741935486e-05, + "loss": 0.4299, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06372793763875961, + "step": 75, + "valid_targets_mean": 5972.1, + "valid_targets_min": 842 }, { "epoch": 0.3078481012658228, - "grad_norm": 0.763928377690798, - "learning_rate": 2.4715447154471548e-05, - "loss": 0.3932, - "step": 76 + "grad_norm": 0.6251085254030976, + "learning_rate": 2.4193548387096777e-05, + "loss": 0.4299, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.044161923229694366, + "step": 76, + "valid_targets_mean": 4310.9, + "valid_targets_min": 770 }, { "epoch": 0.3118987341772152, - "grad_norm": 0.6659477877128168, - "learning_rate": 2.5040650406504066e-05, - "loss": 0.3767, - "step": 77 + "grad_norm": 0.5948103719382858, + "learning_rate": 2.4516129032258067e-05, + "loss": 0.4053, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05264626815915108, + "step": 77, + "valid_targets_mean": 4628.0, + "valid_targets_min": 1385 }, { "epoch": 0.3159493670886076, - "grad_norm": 0.652228378642887, - "learning_rate": 2.536585365853659e-05, - "loss": 0.4151, - "step": 78 + "grad_norm": 0.604033977230391, + "learning_rate": 2.4838709677419358e-05, + "loss": 0.4179, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05454590916633606, + "step": 78, + "valid_targets_mean": 5104.9, + "valid_targets_min": 1642 }, { "epoch": 0.32, - "grad_norm": 0.7116692577132699, - "learning_rate": 2.569105691056911e-05, - "loss": 0.3862, - "step": 79 + "grad_norm": 0.6370412410362083, + "learning_rate": 2.5161290322580648e-05, + "loss": 0.4143, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05205424875020981, + "step": 79, + "valid_targets_mean": 3922.1, + "valid_targets_min": 896 }, { "epoch": 0.3240506329113924, - "grad_norm": 0.6139313448609756, - "learning_rate": 2.601626016260163e-05, - "loss": 0.4019, - "step": 80 + "grad_norm": 0.47243769878404057, + "learning_rate": 2.548387096774194e-05, + "loss": 0.4225, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05829702690243721, + "step": 80, + "valid_targets_mean": 5457.8, + "valid_targets_min": 1868 }, { "epoch": 0.3281012658227848, - "grad_norm": 0.7106418148735457, - "learning_rate": 2.634146341463415e-05, - "loss": 0.3816, - "step": 81 + "grad_norm": 0.5815081489368484, + "learning_rate": 2.5806451612903226e-05, + "loss": 0.4065, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.042076289653778076, + "step": 81, + "valid_targets_mean": 4102.0, + "valid_targets_min": 1411 }, { "epoch": 0.3321518987341772, - "grad_norm": 0.5455098402049849, - "learning_rate": 2.6666666666666667e-05, - "loss": 0.388, - "step": 82 + "grad_norm": 0.5225876956461049, + "learning_rate": 2.6129032258064516e-05, + "loss": 0.4081, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.046641163527965546, + "step": 82, + "valid_targets_mean": 4011.5, + "valid_targets_min": 1200 }, { "epoch": 0.3362025316455696, - "grad_norm": 0.5833625742760852, - "learning_rate": 2.699186991869919e-05, - "loss": 0.3926, - "step": 83 + "grad_norm": 0.4677338333165761, + "learning_rate": 2.6451612903225806e-05, + "loss": 0.4089, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06551804393529892, + "step": 83, + "valid_targets_mean": 5499.8, + "valid_targets_min": 1338 }, { "epoch": 0.34025316455696203, - "grad_norm": 0.6976430625248523, - "learning_rate": 2.731707317073171e-05, - "loss": 0.4148, - "step": 84 + "grad_norm": 0.614764413868001, + "learning_rate": 2.6774193548387097e-05, + "loss": 0.441, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05876109376549721, + "step": 84, + "valid_targets_mean": 4796.1, + "valid_targets_min": 1326 }, { "epoch": 0.34430379746835443, - "grad_norm": 0.5631256669723129, - "learning_rate": 2.764227642276423e-05, - "loss": 0.4038, - "step": 85 + "grad_norm": 0.6457657603329191, + "learning_rate": 2.7096774193548387e-05, + "loss": 0.4321, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06244055926799774, + "step": 85, + "valid_targets_mean": 4897.1, + "valid_targets_min": 1076 }, { "epoch": 0.34835443037974684, - "grad_norm": 0.6297177773650035, - "learning_rate": 2.796747967479675e-05, - "loss": 0.3814, - "step": 86 + "grad_norm": 0.8450721977557598, + "learning_rate": 2.7419354838709678e-05, + "loss": 0.4159, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02325458824634552, + "step": 86, + "valid_targets_mean": 2360.2, + "valid_targets_min": 1064 }, { "epoch": 0.35240506329113924, - "grad_norm": 0.6530498529887747, - "learning_rate": 2.829268292682927e-05, - "loss": 0.3897, - "step": 87 + "grad_norm": 0.9376224307797884, + "learning_rate": 2.774193548387097e-05, + "loss": 0.4275, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.07406799495220184, + "step": 87, + "valid_targets_mean": 4577.8, + "valid_targets_min": 1642 }, { "epoch": 0.35645569620253165, - "grad_norm": 0.660758390996932, - "learning_rate": 2.8617886178861792e-05, - "loss": 0.3971, - "step": 88 + "grad_norm": 0.8814792501538088, + "learning_rate": 2.8064516129032262e-05, + "loss": 0.428, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.07024728506803513, + "step": 88, + "valid_targets_mean": 6062.3, + "valid_targets_min": 1235 }, { "epoch": 0.36050632911392405, - "grad_norm": 0.6432981711779486, - "learning_rate": 2.8943089430894314e-05, - "loss": 0.4082, - "step": 89 + "grad_norm": 0.9349922865807039, + "learning_rate": 2.8387096774193552e-05, + "loss": 0.4334, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0638594776391983, + "step": 89, + "valid_targets_mean": 5731.2, + "valid_targets_min": 1500 }, { "epoch": 0.36455696202531646, - "grad_norm": 0.6769527644089639, - "learning_rate": 2.926829268292683e-05, - "loss": 0.3858, - "step": 90 + "grad_norm": 0.7683719797658665, + "learning_rate": 2.8709677419354843e-05, + "loss": 0.4095, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0370856411755085, + "step": 90, + "valid_targets_mean": 2962.9, + "valid_targets_min": 1186 }, { "epoch": 0.36860759493670886, - "grad_norm": 0.7317177578010697, - "learning_rate": 2.959349593495935e-05, - "loss": 0.3942, - "step": 91 + "grad_norm": 0.829622404761209, + "learning_rate": 2.9032258064516133e-05, + "loss": 0.4239, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06027737632393837, + "step": 91, + "valid_targets_mean": 5650.5, + "valid_targets_min": 1201 }, { "epoch": 0.37265822784810126, - "grad_norm": 0.6561649562895105, - "learning_rate": 2.991869918699187e-05, - "loss": 0.4103, - "step": 92 + "grad_norm": 0.8014182001734745, + "learning_rate": 2.9354838709677423e-05, + "loss": 0.4352, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05477523431181908, + "step": 92, + "valid_targets_mean": 4282.9, + "valid_targets_min": 1167 }, { "epoch": 0.37670886075949367, - "grad_norm": 0.6370332145085758, - "learning_rate": 3.0243902439024392e-05, - "loss": 0.3763, - "step": 93 + "grad_norm": 0.6733027367606301, + "learning_rate": 2.9677419354838714e-05, + "loss": 0.4088, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.057788364589214325, + "step": 93, + "valid_targets_mean": 4786.1, + "valid_targets_min": 1266 }, { "epoch": 0.3807594936708861, - "grad_norm": 0.6994856265474081, - "learning_rate": 3.056910569105692e-05, - "loss": 0.379, - "step": 94 + "grad_norm": 0.727168788839206, + "learning_rate": 3.0000000000000004e-05, + "loss": 0.4044, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04125818610191345, + "step": 94, + "valid_targets_mean": 3322.1, + "valid_targets_min": 1565 }, { "epoch": 0.3848101265822785, - "grad_norm": 0.7534532810506789, - "learning_rate": 3.089430894308943e-05, - "loss": 0.3967, - "step": 95 + "grad_norm": 0.6935011193456103, + "learning_rate": 3.032258064516129e-05, + "loss": 0.4165, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04848355054855347, + "step": 95, + "valid_targets_mean": 4021.0, + "valid_targets_min": 1580 }, { "epoch": 0.3888607594936709, - "grad_norm": 0.8013673825606735, - "learning_rate": 3.1219512195121955e-05, - "loss": 0.3974, - "step": 96 + "grad_norm": 0.6682798377596265, + "learning_rate": 3.0645161290322585e-05, + "loss": 0.415, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04788028076291084, + "step": 96, + "valid_targets_mean": 4199.7, + "valid_targets_min": 1270 }, { "epoch": 0.3929113924050633, - "grad_norm": 0.6601186985117031, - "learning_rate": 3.154471544715447e-05, - "loss": 0.3932, - "step": 97 + "grad_norm": 0.6246593721293577, + "learning_rate": 3.096774193548387e-05, + "loss": 0.417, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.038797035813331604, + "step": 97, + "valid_targets_mean": 3773.2, + "valid_targets_min": 925 }, { "epoch": 0.3969620253164557, - "grad_norm": 0.8816268253435977, - "learning_rate": 3.186991869918699e-05, - "loss": 0.389, - "step": 98 + "grad_norm": 0.5861274809105809, + "learning_rate": 3.1290322580645166e-05, + "loss": 0.4017, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04115966707468033, + "step": 98, + "valid_targets_mean": 3950.2, + "valid_targets_min": 1396 }, { "epoch": 0.4010126582278481, - "grad_norm": 0.6502491759480651, - "learning_rate": 3.2195121951219514e-05, - "loss": 0.3931, - "step": 99 + "grad_norm": 0.7103528215779562, + "learning_rate": 3.161290322580645e-05, + "loss": 0.431, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05843929946422577, + "step": 99, + "valid_targets_mean": 4570.2, + "valid_targets_min": 1129 }, { "epoch": 0.4050632911392405, - "grad_norm": 0.6405538480362688, - "learning_rate": 3.2520325203252037e-05, - "loss": 0.3857, - "step": 100 + "grad_norm": 0.5907411603216484, + "learning_rate": 3.193548387096774e-05, + "loss": 0.4079, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.041239671409130096, + "step": 100, + "valid_targets_mean": 3490.6, + "valid_targets_min": 1166 }, { "epoch": 0.4091139240506329, - "grad_norm": 0.7194466820059928, - "learning_rate": 3.284552845528456e-05, - "loss": 0.4032, - "step": 101 + "grad_norm": 0.6217864930174553, + "learning_rate": 3.2258064516129034e-05, + "loss": 0.4313, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0475708544254303, + "step": 101, + "valid_targets_mean": 4086.8, + "valid_targets_min": 1129 }, { "epoch": 0.4131645569620253, - "grad_norm": 0.5720845890175896, - "learning_rate": 3.3170731707317074e-05, - "loss": 0.3885, - "step": 102 + "grad_norm": 0.5399807835544768, + "learning_rate": 3.258064516129033e-05, + "loss": 0.4161, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06466293334960938, + "step": 102, + "valid_targets_mean": 5693.0, + "valid_targets_min": 808 }, { "epoch": 0.4172151898734177, - "grad_norm": 0.6136400802482502, - "learning_rate": 3.3495934959349596e-05, - "loss": 0.3889, - "step": 103 + "grad_norm": 0.610285856962528, + "learning_rate": 3.2903225806451614e-05, + "loss": 0.4095, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04695362597703934, + "step": 103, + "valid_targets_mean": 3831.8, + "valid_targets_min": 1211 }, { "epoch": 0.4212658227848101, - "grad_norm": 0.6483481006307124, - "learning_rate": 3.382113821138212e-05, - "loss": 0.4024, - "step": 104 + "grad_norm": 0.7251098492949469, + "learning_rate": 3.322580645161291e-05, + "loss": 0.4314, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05233541131019592, + "step": 104, + "valid_targets_mean": 4645.6, + "valid_targets_min": 1205 }, { "epoch": 0.4253164556962025, - "grad_norm": 0.7609866812145318, - "learning_rate": 3.414634146341463e-05, - "loss": 0.4102, - "step": 105 + "grad_norm": 0.5634018976126205, + "learning_rate": 3.3548387096774195e-05, + "loss": 0.42, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06966076046228409, + "step": 105, + "valid_targets_mean": 6022.4, + "valid_targets_min": 1222 }, { "epoch": 0.4293670886075949, - "grad_norm": 0.5910885647277584, - "learning_rate": 3.4471544715447155e-05, - "loss": 0.3815, - "step": 106 + "grad_norm": 0.5798410623158287, + "learning_rate": 3.387096774193549e-05, + "loss": 0.4009, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05769410356879234, + "step": 106, + "valid_targets_mean": 6083.4, + "valid_targets_min": 1292 }, { "epoch": 0.43341772151898733, - "grad_norm": 0.7632752159812108, - "learning_rate": 3.479674796747968e-05, - "loss": 0.391, - "step": 107 + "grad_norm": 0.5843707202381763, + "learning_rate": 3.4193548387096776e-05, + "loss": 0.4214, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04835837706923485, + "step": 107, + "valid_targets_mean": 3786.8, + "valid_targets_min": 1461 }, { "epoch": 0.43746835443037974, - "grad_norm": 0.6537989518615785, - "learning_rate": 3.51219512195122e-05, - "loss": 0.3857, - "step": 108 + "grad_norm": 0.5780833413976226, + "learning_rate": 3.451612903225807e-05, + "loss": 0.4085, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05581241473555565, + "step": 108, + "valid_targets_mean": 5162.2, + "valid_targets_min": 1565 }, { "epoch": 0.44151898734177214, - "grad_norm": 0.45132763454677677, - "learning_rate": 3.544715447154472e-05, - "loss": 0.3728, - "step": 109 + "grad_norm": 0.5940606633232612, + "learning_rate": 3.483870967741936e-05, + "loss": 0.4078, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06986302137374878, + "step": 109, + "valid_targets_mean": 5780.5, + "valid_targets_min": 1673 }, { "epoch": 0.44556962025316454, - "grad_norm": 0.5502432025399826, - "learning_rate": 3.577235772357724e-05, - "loss": 0.3875, - "step": 110 + "grad_norm": 0.5425947044504619, + "learning_rate": 3.516129032258065e-05, + "loss": 0.4127, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04262430965900421, + "step": 110, + "valid_targets_mean": 4362.6, + "valid_targets_min": 1120 }, { "epoch": 0.44962025316455695, - "grad_norm": 0.5482141122008402, - "learning_rate": 3.609756097560976e-05, - "loss": 0.3931, - "step": 111 + "grad_norm": 0.5681614069237215, + "learning_rate": 3.548387096774194e-05, + "loss": 0.4118, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.07122272253036499, + "step": 111, + "valid_targets_mean": 5264.9, + "valid_targets_min": 1212 }, { "epoch": 0.45367088607594935, - "grad_norm": 0.5042579700603617, - "learning_rate": 3.642276422764228e-05, - "loss": 0.3923, - "step": 112 + "grad_norm": 0.5261871695912, + "learning_rate": 3.5806451612903225e-05, + "loss": 0.415, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.049640655517578125, + "step": 112, + "valid_targets_mean": 4602.9, + "valid_targets_min": 1322 }, { "epoch": 0.45772151898734176, - "grad_norm": 0.48080252398491274, - "learning_rate": 3.67479674796748e-05, - "loss": 0.3939, - "step": 113 + "grad_norm": 0.5753944494929996, + "learning_rate": 3.612903225806452e-05, + "loss": 0.4088, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05468757823109627, + "step": 113, + "valid_targets_mean": 3892.3, + "valid_targets_min": 1006 }, { "epoch": 0.46177215189873416, - "grad_norm": 0.5819181191826066, - "learning_rate": 3.7073170731707325e-05, - "loss": 0.3922, - "step": 114 + "grad_norm": 0.510769138277729, + "learning_rate": 3.6451612903225805e-05, + "loss": 0.4168, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05141526088118553, + "step": 114, + "valid_targets_mean": 4239.3, + "valid_targets_min": 1187 }, { "epoch": 0.46582278481012657, - "grad_norm": 0.5161246323531038, - "learning_rate": 3.739837398373984e-05, - "loss": 0.3754, - "step": 115 + "grad_norm": 0.6220564717227475, + "learning_rate": 3.67741935483871e-05, + "loss": 0.3929, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05481129512190819, + "step": 115, + "valid_targets_mean": 5462.4, + "valid_targets_min": 1259 }, { "epoch": 0.46987341772151897, - "grad_norm": 0.5401719710859738, - "learning_rate": 3.772357723577236e-05, - "loss": 0.3877, - "step": 116 + "grad_norm": 0.4626762378034261, + "learning_rate": 3.7096774193548386e-05, + "loss": 0.4059, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.07054520398378372, + "step": 116, + "valid_targets_mean": 5666.0, + "valid_targets_min": 994 }, { "epoch": 0.4739240506329114, - "grad_norm": 0.5850809539882289, - "learning_rate": 3.804878048780488e-05, - "loss": 0.3942, - "step": 117 + "grad_norm": 0.7539378367732781, + "learning_rate": 3.741935483870968e-05, + "loss": 0.4216, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06530013680458069, + "step": 117, + "valid_targets_mean": 5578.9, + "valid_targets_min": 1256 }, { "epoch": 0.4779746835443038, - "grad_norm": 0.5375210050579635, - "learning_rate": 3.83739837398374e-05, - "loss": 0.3701, - "step": 118 + "grad_norm": 0.592412643415671, + "learning_rate": 3.7741935483870974e-05, + "loss": 0.3899, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02674018032848835, + "step": 118, + "valid_targets_mean": 3557.8, + "valid_targets_min": 970 }, { "epoch": 0.4820253164556962, - "grad_norm": 0.5704604164558655, - "learning_rate": 3.869918699186992e-05, - "loss": 0.3886, - "step": 119 + "grad_norm": 0.7086278514415122, + "learning_rate": 3.806451612903226e-05, + "loss": 0.4068, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03874162957072258, + "step": 119, + "valid_targets_mean": 3451.5, + "valid_targets_min": 1390 }, { "epoch": 0.4860759493670886, - "grad_norm": 0.6367835721745447, - "learning_rate": 3.9024390243902444e-05, - "loss": 0.3838, - "step": 120 + "grad_norm": 0.8676643374155362, + "learning_rate": 3.8387096774193555e-05, + "loss": 0.4195, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05797410011291504, + "step": 120, + "valid_targets_mean": 5209.5, + "valid_targets_min": 1236 }, { "epoch": 0.490126582278481, - "grad_norm": 0.6276071899311296, - "learning_rate": 3.9349593495934966e-05, - "loss": 0.3719, - "step": 121 + "grad_norm": 0.9409109428169281, + "learning_rate": 3.870967741935484e-05, + "loss": 0.3931, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04078049957752228, + "step": 121, + "valid_targets_mean": 3740.5, + "valid_targets_min": 1056 }, { "epoch": 0.4941772151898734, - "grad_norm": 0.5100499219774005, - "learning_rate": 3.967479674796748e-05, - "loss": 0.3917, - "step": 122 + "grad_norm": 0.7804190583394841, + "learning_rate": 3.9032258064516135e-05, + "loss": 0.4203, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04113798215985298, + "step": 122, + "valid_targets_mean": 3469.4, + "valid_targets_min": 1310 }, { "epoch": 0.4982278481012658, - "grad_norm": 0.621723133019605, - "learning_rate": 4e-05, - "loss": 0.3955, - "step": 123 + "grad_norm": 0.6976532711479004, + "learning_rate": 3.935483870967742e-05, + "loss": 0.413, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04702591896057129, + "step": 123, + "valid_targets_mean": 4705.9, + "valid_targets_min": 1178 }, { "epoch": 0.5022784810126583, - "grad_norm": 0.5629741730086855, - "learning_rate": 3.999991946137476e-05, - "loss": 0.3922, - "step": 124 + "grad_norm": 0.6970124689802406, + "learning_rate": 3.9677419354838716e-05, + "loss": 0.4229, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0377708300948143, + "step": 124, + "valid_targets_mean": 3650.1, + "valid_targets_min": 1443 }, { "epoch": 0.5063291139240507, - "grad_norm": 0.5406316246600198, - "learning_rate": 3.999967784614766e-05, - "loss": 0.3606, - "step": 125 + "grad_norm": 0.6067040651055834, + "learning_rate": 4e-05, + "loss": 0.3816, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.044166043400764465, + "step": 125, + "valid_targets_mean": 3421.4, + "valid_targets_min": 999 }, { "epoch": 0.5103797468354431, - "grad_norm": 0.5600740792620695, - "learning_rate": 3.9999275156264656e-05, - "loss": 0.386, - "step": 126 + "grad_norm": 0.7312519375483681, + "learning_rate": 3.9999920040266474e-05, + "loss": 0.4079, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05348426103591919, + "step": 126, + "valid_targets_mean": 3932.2, + "valid_targets_min": 944 }, { "epoch": 0.5144303797468355, - "grad_norm": 0.6258097987727942, - "learning_rate": 3.999871139496895e-05, - "loss": 0.3862, - "step": 127 + "grad_norm": 0.6629876905385923, + "learning_rate": 3.999968016170525e-05, + "loss": 0.408, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04648736119270325, + "step": 127, + "valid_targets_mean": 3955.1, + "valid_targets_min": 972 }, { "epoch": 0.5184810126582279, - "grad_norm": 0.511933762069151, - "learning_rate": 3.9997986566800995e-05, - "loss": 0.3948, - "step": 128 + "grad_norm": 0.8204905739876648, + "learning_rate": 3.9999280366234394e-05, + "loss": 0.4183, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06762972474098206, + "step": 128, + "valid_targets_mean": 6666.4, + "valid_targets_min": 1389 }, { "epoch": 0.5225316455696203, - "grad_norm": 0.504332229912149, - "learning_rate": 3.999710067759846e-05, - "loss": 0.3932, - "step": 129 + "grad_norm": 1.7319807050242546, + "learning_rate": 3.999872065705065e-05, + "loss": 0.4227, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04184423387050629, + "step": 129, + "valid_targets_mean": 3844.4, + "valid_targets_min": 1319 }, { "epoch": 0.5265822784810127, - "grad_norm": 0.5105578811506033, - "learning_rate": 3.999605373449617e-05, - "loss": 0.4006, - "step": 130 + "grad_norm": 0.9634703858631996, + "learning_rate": 3.999800103862945e-05, + "loss": 0.4335, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04888691380620003, + "step": 130, + "valid_targets_mean": 4168.3, + "valid_targets_min": 1386 }, { "epoch": 0.5306329113924051, - "grad_norm": 0.5205593904868228, - "learning_rate": 3.9994845745926075e-05, - "loss": 0.3891, - "step": 131 + "grad_norm": 0.784517576483665, + "learning_rate": 3.9997121516724834e-05, + "loss": 0.4119, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.044563088566064835, + "step": 131, + "valid_targets_mean": 4179.5, + "valid_targets_min": 1230 }, { "epoch": 0.5346835443037975, - "grad_norm": 0.6152102208120602, - "learning_rate": 3.999347672161713e-05, - "loss": 0.3821, - "step": 132 + "grad_norm": 0.7533377576610076, + "learning_rate": 3.999608209836944e-05, + "loss": 0.4039, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04797458276152611, + "step": 132, + "valid_targets_mean": 4516.6, + "valid_targets_min": 1349 }, { "epoch": 0.5387341772151899, - "grad_norm": 0.5205473151605777, - "learning_rate": 3.999194667259528e-05, - "loss": 0.3855, - "step": 133 + "grad_norm": 0.678560635739266, + "learning_rate": 3.999488279187443e-05, + "loss": 0.4042, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05512206628918648, + "step": 133, + "valid_targets_mean": 4947.1, + "valid_targets_min": 964 }, { "epoch": 0.5427848101265823, - "grad_norm": 0.5838400985938534, - "learning_rate": 3.999025561118334e-05, - "loss": 0.393, - "step": 134 + "grad_norm": 0.8073678337167025, + "learning_rate": 3.9993523606829425e-05, + "loss": 0.4247, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026869136840105057, + "step": 134, + "valid_targets_mean": 2812.9, + "valid_targets_min": 1261 }, { "epoch": 0.5468354430379747, - "grad_norm": 0.5605239709503395, - "learning_rate": 3.998840355100086e-05, - "loss": 0.3789, - "step": 135 + "grad_norm": 0.7196854610319208, + "learning_rate": 3.9992004554102435e-05, + "loss": 0.4098, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04661894962191582, + "step": 135, + "valid_targets_mean": 3934.5, + "valid_targets_min": 1198 }, { "epoch": 0.5508860759493671, - "grad_norm": 0.49019950538882867, - "learning_rate": 3.998639050696409e-05, - "loss": 0.3847, - "step": 136 + "grad_norm": 0.7897514744361768, + "learning_rate": 3.999032564583976e-05, + "loss": 0.4093, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.052173689007759094, + "step": 136, + "valid_targets_mean": 4249.8, + "valid_targets_min": 1415 }, { "epoch": 0.5549367088607595, - "grad_norm": 0.4548532481780974, - "learning_rate": 3.998421649528582e-05, - "loss": 0.3844, - "step": 137 + "grad_norm": 0.71909049194181, + "learning_rate": 3.998848689546591e-05, + "loss": 0.4006, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06791016459465027, + "step": 137, + "valid_targets_mean": 5992.1, + "valid_targets_min": 899 }, { "epoch": 0.5589873417721519, - "grad_norm": 0.5261329592085209, - "learning_rate": 3.9981881533475234e-05, - "loss": 0.3825, - "step": 138 + "grad_norm": 0.5433203728561666, + "learning_rate": 3.9986488317683496e-05, + "loss": 0.405, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04295790195465088, + "step": 138, + "valid_targets_mean": 3433.3, + "valid_targets_min": 1075 }, { "epoch": 0.5630379746835443, - "grad_norm": 0.43722984117577496, - "learning_rate": 3.997938564033779e-05, - "loss": 0.3819, - "step": 139 + "grad_norm": 0.5480032491903007, + "learning_rate": 3.9984329928473067e-05, + "loss": 0.3995, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0497143492102623, + "step": 139, + "valid_targets_mean": 4588.2, + "valid_targets_min": 1152 }, { "epoch": 0.5670886075949367, - "grad_norm": 0.5330799402756228, - "learning_rate": 3.9976728835975064e-05, - "loss": 0.3916, - "step": 140 + "grad_norm": 0.6103982990072461, + "learning_rate": 3.998201174509306e-05, + "loss": 0.4037, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.049686968326568604, + "step": 140, + "valid_targets_mean": 4468.4, + "valid_targets_min": 1476 }, { "epoch": 0.5711392405063291, - "grad_norm": 0.5565583472937724, - "learning_rate": 3.9973911141784605e-05, - "loss": 0.3739, - "step": 141 + "grad_norm": 0.5629988818331454, + "learning_rate": 3.997953378607961e-05, + "loss": 0.397, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04076606035232544, + "step": 141, + "valid_targets_mean": 3553.6, + "valid_targets_min": 1049 }, { "epoch": 0.5751898734177215, - "grad_norm": 0.6079760230334288, - "learning_rate": 3.997093258045973e-05, - "loss": 0.3803, - "step": 142 + "grad_norm": 0.5359259818177005, + "learning_rate": 3.9976896071246403e-05, + "loss": 0.3994, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04303029179573059, + "step": 142, + "valid_targets_mean": 3950.3, + "valid_targets_min": 1532 }, { "epoch": 0.579240506329114, - "grad_norm": 0.494759642886606, - "learning_rate": 3.996779317598936e-05, - "loss": 0.3907, - "step": 143 + "grad_norm": 0.6529004994577806, + "learning_rate": 3.997409862168454e-05, + "loss": 0.4124, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06024843454360962, + "step": 143, + "valid_targets_mean": 5404.2, + "valid_targets_min": 1553 }, { "epoch": 0.5832911392405064, - "grad_norm": 0.5946891150411486, - "learning_rate": 3.996449295365782e-05, - "loss": 0.3905, - "step": 144 + "grad_norm": 0.45493544439121325, + "learning_rate": 3.997114145976237e-05, + "loss": 0.4152, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05036636069417, + "step": 144, + "valid_targets_mean": 4548.0, + "valid_targets_min": 1389 }, { "epoch": 0.5873417721518988, - "grad_norm": 0.4412945363216492, - "learning_rate": 3.996103194004467e-05, - "loss": 0.3808, - "step": 145 + "grad_norm": 0.549691947022771, + "learning_rate": 3.9968024609125245e-05, + "loss": 0.3962, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.043455820530653, + "step": 145, + "valid_targets_mean": 4322.4, + "valid_targets_min": 1112 }, { "epoch": 0.5913924050632912, - "grad_norm": 0.5901528370321416, - "learning_rate": 3.995741016302441e-05, - "loss": 0.3897, - "step": 146 + "grad_norm": 0.5403024847204195, + "learning_rate": 3.996474809469545e-05, + "loss": 0.4095, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03951553627848625, + "step": 146, + "valid_targets_mean": 3759.8, + "valid_targets_min": 1422 }, { "epoch": 0.5954430379746836, - "grad_norm": 0.4899374115359715, - "learning_rate": 3.9953627651766364e-05, - "loss": 0.3727, - "step": 147 + "grad_norm": 0.51919871990684, + "learning_rate": 3.996131194267189e-05, + "loss": 0.392, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03196530044078827, + "step": 147, + "valid_targets_mean": 2806.6, + "valid_targets_min": 1104 }, { "epoch": 0.599493670886076, - "grad_norm": 0.6123773529216578, - "learning_rate": 3.9949684436734325e-05, - "loss": 0.3619, - "step": 148 + "grad_norm": 0.5576879996511324, + "learning_rate": 3.995771618052995e-05, + "loss": 0.387, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.043483369052410126, + "step": 148, + "valid_targets_mean": 3857.6, + "valid_targets_min": 1281 }, { "epoch": 0.6035443037974684, - "grad_norm": 0.4853453351537688, - "learning_rate": 3.994558054968643e-05, - "loss": 0.3732, - "step": 149 + "grad_norm": 0.6205996474132962, + "learning_rate": 3.9953960837021256e-05, + "loss": 0.4022, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04536616429686546, + "step": 149, + "valid_targets_mean": 3949.1, + "valid_targets_min": 1340 }, { "epoch": 0.6075949367088608, - "grad_norm": 0.4977697463339369, - "learning_rate": 3.994131602367481e-05, - "loss": 0.3752, - "step": 150 + "grad_norm": 0.5287455565975784, + "learning_rate": 3.995004594217343e-05, + "loss": 0.4003, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0511779822409153, + "step": 150, + "valid_targets_mean": 4899.2, + "valid_targets_min": 1345 }, { "epoch": 0.6116455696202532, - "grad_norm": 0.5414166984338413, - "learning_rate": 3.9936890893045376e-05, - "loss": 0.3744, - "step": 151 + "grad_norm": 0.5286353613893214, + "learning_rate": 3.994597152728987e-05, + "loss": 0.3984, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.048138149082660675, + "step": 151, + "valid_targets_mean": 4724.3, + "valid_targets_min": 1146 }, { "epoch": 0.6156962025316456, - "grad_norm": 0.48749139915438855, - "learning_rate": 3.993230519343752e-05, - "loss": 0.3971, - "step": 152 + "grad_norm": 0.45837696459121535, + "learning_rate": 3.994173762494947e-05, + "loss": 0.4264, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05765845626592636, + "step": 152, + "valid_targets_mean": 4681.2, + "valid_targets_min": 1914 }, { "epoch": 0.619746835443038, - "grad_norm": 0.5976258378379328, - "learning_rate": 3.992755896178383e-05, - "loss": 0.3826, - "step": 153 + "grad_norm": 0.49883723421511145, + "learning_rate": 3.9937344269006424e-05, + "loss": 0.4049, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.031516727060079575, + "step": 153, + "valid_targets_mean": 3224.6, + "valid_targets_min": 1286 }, { "epoch": 0.6237974683544304, - "grad_norm": 0.42523904094458825, - "learning_rate": 3.992265223630981e-05, - "loss": 0.3687, - "step": 154 + "grad_norm": 0.39660923865273007, + "learning_rate": 3.993279149458988e-05, + "loss": 0.3913, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.036629438400268555, + "step": 154, + "valid_targets_mean": 3398.7, + "valid_targets_min": 1294 }, { "epoch": 0.6278481012658228, - "grad_norm": 0.528902294404794, - "learning_rate": 3.991758505653355e-05, - "loss": 0.3967, - "step": 155 + "grad_norm": 0.5360988652258376, + "learning_rate": 3.9928079338103695e-05, + "loss": 0.4152, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06448749452829361, + "step": 155, + "valid_targets_mean": 5875.6, + "valid_targets_min": 1524 }, { "epoch": 0.6318987341772152, - "grad_norm": 0.48770937311698753, - "learning_rate": 3.991235746326543e-05, - "loss": 0.3815, - "step": 156 + "grad_norm": 0.5226602694069594, + "learning_rate": 3.992320783722616e-05, + "loss": 0.4017, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05161647871136665, + "step": 156, + "valid_targets_mean": 4747.9, + "valid_targets_min": 1278 }, { "epoch": 0.6359493670886076, - "grad_norm": 0.445358901306249, - "learning_rate": 3.9906969498607745e-05, - "loss": 0.3717, - "step": 157 + "grad_norm": 0.40436815356541295, + "learning_rate": 3.991817703090965e-05, + "loss": 0.3868, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04100307077169418, + "step": 157, + "valid_targets_mean": 3910.7, + "valid_targets_min": 1557 }, { "epoch": 0.64, - "grad_norm": 0.5388773039788232, - "learning_rate": 3.990142120595444e-05, - "loss": 0.382, - "step": 158 + "grad_norm": 0.3926704120594189, + "learning_rate": 3.991298695938038e-05, + "loss": 0.4043, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.042444128543138504, + "step": 158, + "valid_targets_mean": 3522.6, + "valid_targets_min": 1096 }, { "epoch": 0.6440506329113924, - "grad_norm": 0.4974551837610294, - "learning_rate": 3.98957126299907e-05, - "loss": 0.3763, - "step": 159 + "grad_norm": 0.5076623774445436, + "learning_rate": 3.9907637664138e-05, + "loss": 0.3993, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04470229893922806, + "step": 159, + "valid_targets_mean": 3696.1, + "valid_targets_min": 1064 }, { "epoch": 0.6481012658227848, - "grad_norm": 0.48145696573304353, - "learning_rate": 3.9889843816692596e-05, - "loss": 0.3885, - "step": 160 + "grad_norm": 0.5337854961710667, + "learning_rate": 3.990212918795534e-05, + "loss": 0.4051, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05759483948349953, + "step": 160, + "valid_targets_mean": 5352.3, + "valid_targets_min": 1560 }, { "epoch": 0.6521518987341772, - "grad_norm": 0.4852697094770394, - "learning_rate": 3.9883814813326766e-05, - "loss": 0.3857, - "step": 161 + "grad_norm": 0.49232979360802614, + "learning_rate": 3.9896461574878036e-05, + "loss": 0.4012, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04649016261100769, + "step": 161, + "valid_targets_mean": 4706.9, + "valid_targets_min": 1022 }, { "epoch": 0.6562025316455696, - "grad_norm": 0.46635533311708754, - "learning_rate": 3.9877625668449956e-05, - "loss": 0.3776, - "step": 162 + "grad_norm": 0.4492209651055195, + "learning_rate": 3.989063487022416e-05, + "loss": 0.4099, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.045562755316495895, + "step": 162, + "valid_targets_mean": 4150.2, + "valid_targets_min": 1119 }, { "epoch": 0.660253164556962, - "grad_norm": 0.5151311130783601, - "learning_rate": 3.98712764319087e-05, - "loss": 0.3804, - "step": 163 + "grad_norm": 0.5807853378786956, + "learning_rate": 3.9884649120583903e-05, + "loss": 0.4069, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04843149334192276, + "step": 163, + "valid_targets_mean": 3811.6, + "valid_targets_min": 807 }, { "epoch": 0.6643037974683544, - "grad_norm": 0.4264588202621475, - "learning_rate": 3.9864767154838864e-05, - "loss": 0.384, - "step": 164 + "grad_norm": 0.4421586230983281, + "learning_rate": 3.987850437381915e-05, + "loss": 0.392, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03646192327141762, + "step": 164, + "valid_targets_mean": 3694.9, + "valid_targets_min": 884 }, { "epoch": 0.6683544303797468, - "grad_norm": 0.46551804296424865, - "learning_rate": 3.9858097889665277e-05, - "loss": 0.3713, - "step": 165 + "grad_norm": 0.4801528491517805, + "learning_rate": 3.987220067906313e-05, + "loss": 0.3917, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.051381587982177734, + "step": 165, + "valid_targets_mean": 5180.6, + "valid_targets_min": 1305 }, { "epoch": 0.6724050632911392, - "grad_norm": 0.4744091958623959, - "learning_rate": 3.985126869010129e-05, - "loss": 0.3803, - "step": 166 + "grad_norm": 0.5294172160595377, + "learning_rate": 3.986573808672002e-05, + "loss": 0.3953, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03182698041200638, + "step": 166, + "valid_targets_mean": 3104.5, + "valid_targets_min": 1123 }, { "epoch": 0.6764556962025317, - "grad_norm": 0.4732573885033216, - "learning_rate": 3.984427961114833e-05, - "loss": 0.3665, - "step": 167 + "grad_norm": 0.5134011377182386, + "learning_rate": 3.9859116648464534e-05, + "loss": 0.3955, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06264036148786545, + "step": 167, + "valid_targets_mean": 5248.3, + "valid_targets_min": 1209 }, { "epoch": 0.6805063291139241, - "grad_norm": 0.4655474646655494, - "learning_rate": 3.9837130709095475e-05, - "loss": 0.3677, - "step": 168 + "grad_norm": 0.4858291424933584, + "learning_rate": 3.985233641724152e-05, + "loss": 0.3959, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04595412313938141, + "step": 168, + "valid_targets_mean": 3716.1, + "valid_targets_min": 1065 }, { "epoch": 0.6845569620253165, - "grad_norm": 0.498662203183722, - "learning_rate": 3.982982204151901e-05, - "loss": 0.3801, - "step": 169 + "grad_norm": 0.46922241556163324, + "learning_rate": 3.984539744726554e-05, + "loss": 0.412, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04231109470129013, + "step": 169, + "valid_targets_mean": 4030.9, + "valid_targets_min": 1320 }, { "epoch": 0.6886075949367089, - "grad_norm": 0.4548444564791788, - "learning_rate": 3.982235366728193e-05, - "loss": 0.3714, - "step": 170 + "grad_norm": 0.4787874437935905, + "learning_rate": 3.983829979402038e-05, + "loss": 0.3949, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.053731370717287064, + "step": 170, + "valid_targets_mean": 4688.7, + "valid_targets_min": 1498 }, { "epoch": 0.6926582278481013, - "grad_norm": 0.5674550155900002, - "learning_rate": 3.9814725646533505e-05, - "loss": 0.373, - "step": 171 + "grad_norm": 0.46699510261304283, + "learning_rate": 3.983104351425871e-05, + "loss": 0.4031, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03717079013586044, + "step": 171, + "valid_targets_mean": 2916.2, + "valid_targets_min": 1284 }, { "epoch": 0.6967088607594937, - "grad_norm": 0.4568990388963496, - "learning_rate": 3.9806938040708746e-05, - "loss": 0.3545, - "step": 172 + "grad_norm": 0.5286391464842389, + "learning_rate": 3.982362866600154e-05, + "loss": 0.3805, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05524525046348572, + "step": 172, + "valid_targets_mean": 4783.1, + "valid_targets_min": 1617 }, { "epoch": 0.7007594936708861, - "grad_norm": 0.4854290944250977, - "learning_rate": 3.9798990912527976e-05, - "loss": 0.3664, - "step": 173 + "grad_norm": 0.44158987757459944, + "learning_rate": 3.9816055308537806e-05, + "loss": 0.3981, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.059007737785577774, + "step": 173, + "valid_targets_mean": 4662.8, + "valid_targets_min": 902 }, { "epoch": 0.7048101265822785, - "grad_norm": 0.5068358545778741, - "learning_rate": 3.979088432599627e-05, - "loss": 0.3791, - "step": 174 + "grad_norm": 0.465458920311661, + "learning_rate": 3.980832350242388e-05, + "loss": 0.4016, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.053797923028469086, + "step": 174, + "valid_targets_mean": 5012.7, + "valid_targets_min": 1519 }, { "epoch": 0.7088607594936709, - "grad_norm": 0.44246109952716284, - "learning_rate": 3.9782618346402964e-05, - "loss": 0.3971, - "step": 175 + "grad_norm": 0.4868379216742597, + "learning_rate": 3.980043330948306e-05, + "loss": 0.4135, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06050734966993332, + "step": 175, + "valid_targets_mean": 5410.9, + "valid_targets_min": 1412 }, { "epoch": 0.7129113924050633, - "grad_norm": 0.5105198118923095, - "learning_rate": 3.977419304032111e-05, - "loss": 0.3769, - "step": 176 + "grad_norm": 0.4412776613977876, + "learning_rate": 3.9792384792805124e-05, + "loss": 0.3905, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.049749113619327545, + "step": 176, + "valid_targets_mean": 4090.3, + "valid_targets_min": 1200 }, { "epoch": 0.7169620253164557, - "grad_norm": 0.4504255417270314, - "learning_rate": 3.976560847560697e-05, - "loss": 0.3839, - "step": 177 + "grad_norm": 0.42469125011897935, + "learning_rate": 3.978417801674579e-05, + "loss": 0.4029, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04124998301267624, + "step": 177, + "valid_targets_mean": 3514.7, + "valid_targets_min": 1021 }, { "epoch": 0.7210126582278481, - "grad_norm": 0.47083549172542233, - "learning_rate": 3.9756864721399456e-05, - "loss": 0.3825, - "step": 178 + "grad_norm": 0.4989418958189537, + "learning_rate": 3.9775813046926244e-05, + "loss": 0.4007, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.045302219688892365, + "step": 178, + "valid_targets_mean": 5219.8, + "valid_targets_min": 1081 }, { "epoch": 0.7250632911392405, - "grad_norm": 0.4938812526861007, - "learning_rate": 3.974796184811956e-05, - "loss": 0.3655, - "step": 179 + "grad_norm": 0.38298541539289754, + "learning_rate": 3.976728995023254e-05, + "loss": 0.3936, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03838229179382324, + "step": 179, + "valid_targets_mean": 3586.2, + "valid_targets_min": 1157 }, { "epoch": 0.7291139240506329, - "grad_norm": 0.6121123595102624, - "learning_rate": 3.973889992746979e-05, - "loss": 0.3867, - "step": 180 + "grad_norm": 0.4784940401420178, + "learning_rate": 3.975860879481514e-05, + "loss": 0.4053, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04661254584789276, + "step": 180, + "valid_targets_mean": 6014.8, + "valid_targets_min": 1968 }, { "epoch": 0.7331645569620253, - "grad_norm": 0.3935469525704463, - "learning_rate": 3.972967903243361e-05, - "loss": 0.3812, - "step": 181 + "grad_norm": 0.4762470844905524, + "learning_rate": 3.974976965008833e-05, + "loss": 0.3996, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06391244381666183, + "step": 181, + "valid_targets_mean": 6059.4, + "valid_targets_min": 1415 }, { "epoch": 0.7372151898734177, - "grad_norm": 0.5665964439606438, - "learning_rate": 3.972029923727486e-05, - "loss": 0.3835, - "step": 182 + "grad_norm": 0.4155266684565921, + "learning_rate": 3.9740772586729674e-05, + "loss": 0.3994, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04258221387863159, + "step": 182, + "valid_targets_mean": 3626.9, + "valid_targets_min": 1583 }, { "epoch": 0.7412658227848101, - "grad_norm": 0.40364607683557086, - "learning_rate": 3.971076061753709e-05, - "loss": 0.3605, - "step": 183 + "grad_norm": 0.4197532740492893, + "learning_rate": 3.973161767667946e-05, + "loss": 0.3947, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0651676282286644, + "step": 183, + "valid_targets_mean": 5145.3, + "valid_targets_min": 1610 }, { "epoch": 0.7453164556962025, - "grad_norm": 0.5981224928226465, - "learning_rate": 3.9701063250043066e-05, - "loss": 0.3588, - "step": 184 + "grad_norm": 0.44567348029714937, + "learning_rate": 3.972230499314009e-05, + "loss": 0.3831, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03477749228477478, + "step": 184, + "valid_targets_mean": 3506.1, + "valid_targets_min": 1272 }, { "epoch": 0.7493670886075949, - "grad_norm": 0.3856239194304908, - "learning_rate": 3.969120721289402e-05, - "loss": 0.3734, - "step": 185 + "grad_norm": 0.4660019920816794, + "learning_rate": 3.9712834610575545e-05, + "loss": 0.3914, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06879770755767822, + "step": 185, + "valid_targets_mean": 6329.2, + "valid_targets_min": 1450 }, { "epoch": 0.7534177215189873, - "grad_norm": 0.5908369710632956, - "learning_rate": 3.9681192585469146e-05, - "loss": 0.3685, - "step": 186 + "grad_norm": 0.35466972491540216, + "learning_rate": 3.9703206604710746e-05, + "loss": 0.3884, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04594149440526962, + "step": 186, + "valid_targets_mean": 4308.9, + "valid_targets_min": 1318 }, { "epoch": 0.7574683544303797, - "grad_norm": 0.3894179164448138, - "learning_rate": 3.9671019448424865e-05, - "loss": 0.3593, - "step": 187 + "grad_norm": 0.4043292780387038, + "learning_rate": 3.969342105253098e-05, + "loss": 0.378, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04979319870471954, + "step": 187, + "valid_targets_mean": 4318.1, + "valid_targets_min": 1072 }, { "epoch": 0.7615189873417721, - "grad_norm": 0.5215780390811954, - "learning_rate": 3.966068788369422e-05, - "loss": 0.3824, - "step": 188 + "grad_norm": 0.43762304614629416, + "learning_rate": 3.9683478032281246e-05, + "loss": 0.4087, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05901230499148369, + "step": 188, + "valid_targets_mean": 5515.2, + "valid_targets_min": 1401 }, { "epoch": 0.7655696202531646, - "grad_norm": 0.4154544733655475, - "learning_rate": 3.965019797448622e-05, - "loss": 0.3722, - "step": 189 + "grad_norm": 0.35257897927594767, + "learning_rate": 3.967337762346568e-05, + "loss": 0.3971, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.036184344440698624, + "step": 189, + "valid_targets_mean": 3648.5, + "valid_targets_min": 1316 }, { "epoch": 0.769620253164557, - "grad_norm": 0.48384942940680387, - "learning_rate": 3.963954980528515e-05, - "loss": 0.371, - "step": 190 + "grad_norm": 0.4586516555619478, + "learning_rate": 3.9663119906846885e-05, + "loss": 0.3994, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0621817409992218, + "step": 190, + "valid_targets_mean": 5998.1, + "valid_targets_min": 1407 }, { "epoch": 0.7736708860759494, - "grad_norm": 0.4151020053728562, - "learning_rate": 3.9628743461849905e-05, - "loss": 0.3628, - "step": 191 + "grad_norm": 0.4133487608785662, + "learning_rate": 3.965270496444528e-05, + "loss": 0.3918, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04161564260721207, + "step": 191, + "valid_targets_mean": 4036.8, + "valid_targets_min": 1020 }, { "epoch": 0.7777215189873418, - "grad_norm": 0.5605697944605167, - "learning_rate": 3.961777903121329e-05, - "loss": 0.3725, - "step": 192 + "grad_norm": 0.4450796537280172, + "learning_rate": 3.9642132879538465e-05, + "loss": 0.3957, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03854193910956383, + "step": 192, + "valid_targets_mean": 3875.6, + "valid_targets_min": 1243 }, { "epoch": 0.7817721518987342, - "grad_norm": 0.43406064369137337, - "learning_rate": 3.960665660168131e-05, - "loss": 0.3762, - "step": 193 + "grad_norm": 0.3901632045904612, + "learning_rate": 3.9631403736660566e-05, + "loss": 0.4021, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.037209074944257736, + "step": 193, + "valid_targets_mean": 3762.1, + "valid_targets_min": 1137 }, { "epoch": 0.7858227848101266, - "grad_norm": 0.4543925214390367, - "learning_rate": 3.9595376262832485e-05, - "loss": 0.3716, - "step": 194 + "grad_norm": 0.3849767613588487, + "learning_rate": 3.962051762160151e-05, + "loss": 0.3937, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.044705405831336975, + "step": 194, + "valid_targets_mean": 4061.6, + "valid_targets_min": 1186 }, { "epoch": 0.789873417721519, - "grad_norm": 0.42761526807896044, - "learning_rate": 3.9583938105517127e-05, - "loss": 0.3656, - "step": 195 + "grad_norm": 0.38977830003176955, + "learning_rate": 3.960947462140638e-05, + "loss": 0.39, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.043201059103012085, + "step": 195, + "valid_targets_mean": 4334.1, + "valid_targets_min": 865 }, { "epoch": 0.7939240506329114, - "grad_norm": 0.425195946163417, - "learning_rate": 3.957234222185657e-05, - "loss": 0.3685, - "step": 196 + "grad_norm": 0.4206313908143476, + "learning_rate": 3.959827482437472e-05, + "loss": 0.4019, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05456563085317612, + "step": 196, + "valid_targets_mean": 4199.2, + "valid_targets_min": 1019 }, { "epoch": 0.7979746835443038, - "grad_norm": 0.4620078808266855, - "learning_rate": 3.9560588705242474e-05, - "loss": 0.3658, - "step": 197 + "grad_norm": 0.4099403626766528, + "learning_rate": 3.958691832005981e-05, + "loss": 0.3886, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.032608289271593094, + "step": 197, + "valid_targets_mean": 3207.4, + "valid_targets_min": 943 }, { "epoch": 0.8020253164556962, - "grad_norm": 0.5274449762131294, - "learning_rate": 3.954867765033605e-05, - "loss": 0.3596, - "step": 198 + "grad_norm": 0.41275822414873825, + "learning_rate": 3.9575405199267944e-05, + "loss": 0.3859, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04315009340643883, + "step": 198, + "valid_targets_mean": 4100.4, + "valid_targets_min": 1125 }, { "epoch": 0.8060759493670886, - "grad_norm": 0.37957278413317347, - "learning_rate": 3.953660915306728e-05, - "loss": 0.3479, - "step": 199 + "grad_norm": 0.42788263964263656, + "learning_rate": 3.956373555405774e-05, + "loss": 0.3734, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.040834665298461914, + "step": 199, + "valid_targets_mean": 3672.3, + "valid_targets_min": 1106 }, { "epoch": 0.810126582278481, - "grad_norm": 0.4799711325269261, - "learning_rate": 3.952438331063419e-05, - "loss": 0.3669, - "step": 200 + "grad_norm": 0.3843454886819529, + "learning_rate": 3.955190947773937e-05, + "loss": 0.3747, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03427031636238098, + "step": 200, + "valid_targets_mean": 3138.8, + "valid_targets_min": 971 }, { "epoch": 0.8141772151898734, - "grad_norm": 0.4514457644026668, - "learning_rate": 3.951200022150205e-05, - "loss": 0.3726, - "step": 201 + "grad_norm": 0.45090978500288503, + "learning_rate": 3.9539927064873824e-05, + "loss": 0.3986, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0527331605553627, + "step": 201, + "valid_targets_mean": 4807.4, + "valid_targets_min": 1351 }, { "epoch": 0.8182278481012658, - "grad_norm": 0.39922943671244554, - "learning_rate": 3.949945998540253e-05, - "loss": 0.36, - "step": 202 + "grad_norm": 0.3969147927096043, + "learning_rate": 3.952778841127214e-05, + "loss": 0.3778, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030247759073972702, + "step": 202, + "valid_targets_mean": 3209.6, + "valid_targets_min": 1731 }, { "epoch": 0.8222784810126582, - "grad_norm": 0.5341218610008884, - "learning_rate": 3.9486762703332993e-05, - "loss": 0.3626, - "step": 203 + "grad_norm": 0.39455231931822077, + "learning_rate": 3.95154936139947e-05, + "loss": 0.3692, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.034564059227705, + "step": 203, + "valid_targets_mean": 3379.9, + "valid_targets_min": 1167 }, { "epoch": 0.8263291139240506, - "grad_norm": 0.39994061049090884, - "learning_rate": 3.947390847755559e-05, - "loss": 0.3616, - "step": 204 + "grad_norm": 0.4082760818606499, + "learning_rate": 3.950304277135034e-05, + "loss": 0.3811, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.031288061290979385, + "step": 204, + "valid_targets_mean": 3503.6, + "valid_targets_min": 869 }, { "epoch": 0.830379746835443, - "grad_norm": 0.5087763598480363, - "learning_rate": 3.946089741159648e-05, - "loss": 0.3723, - "step": 205 + "grad_norm": 0.401292855029016, + "learning_rate": 3.94904359828957e-05, + "loss": 0.3875, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03167634457349777, + "step": 205, + "valid_targets_mean": 2991.9, + "valid_targets_min": 1527 }, { "epoch": 0.8344303797468354, - "grad_norm": 0.571082065233194, - "learning_rate": 3.944772961024501e-05, - "loss": 0.3647, - "step": 206 + "grad_norm": 0.4712089367640346, + "learning_rate": 3.94776733494343e-05, + "loss": 0.3928, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030103646218776703, + "step": 206, + "valid_targets_mean": 2564.2, + "valid_targets_min": 1003 }, { "epoch": 0.8384810126582278, - "grad_norm": 0.40769972678351213, - "learning_rate": 3.943440517955285e-05, - "loss": 0.3635, - "step": 207 + "grad_norm": 0.40947408231322907, + "learning_rate": 3.9464754973015826e-05, + "loss": 0.3837, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04522818326950073, + "step": 207, + "valid_targets_mean": 4203.4, + "valid_targets_min": 1026 }, { "epoch": 0.8425316455696202, - "grad_norm": 0.49410582685700866, - "learning_rate": 3.9420924226833126e-05, - "loss": 0.3645, - "step": 208 + "grad_norm": 0.4151402680031294, + "learning_rate": 3.9451680956935275e-05, + "loss": 0.3866, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03589247912168503, + "step": 208, + "valid_targets_mean": 3352.5, + "valid_targets_min": 857 }, { "epoch": 0.8465822784810126, - "grad_norm": 0.4387525444885117, - "learning_rate": 3.9407286860659566e-05, - "loss": 0.3598, - "step": 209 + "grad_norm": 0.4503235990787147, + "learning_rate": 3.943845140573213e-05, + "loss": 0.3957, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.029857059940695763, + "step": 209, + "valid_targets_mean": 2977.6, + "valid_targets_min": 1329 }, { "epoch": 0.850632911392405, - "grad_norm": 0.44273304130244956, - "learning_rate": 3.9393493190865657e-05, - "loss": 0.3611, - "step": 210 + "grad_norm": 0.36151361416077155, + "learning_rate": 3.942506642518952e-05, + "loss": 0.3801, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.040279075503349304, + "step": 210, + "valid_targets_mean": 3750.4, + "valid_targets_min": 1110 }, { "epoch": 0.8546835443037974, - "grad_norm": 0.43175110862933125, - "learning_rate": 3.937954332854371e-05, - "loss": 0.3584, - "step": 211 + "grad_norm": 0.46511902089306856, + "learning_rate": 3.9411526122333396e-05, + "loss": 0.3801, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05088287591934204, + "step": 211, + "valid_targets_mean": 5016.0, + "valid_targets_min": 1202 }, { "epoch": 0.8587341772151899, - "grad_norm": 0.4564656033730629, - "learning_rate": 3.9365437386044016e-05, - "loss": 0.371, - "step": 212 + "grad_norm": 0.44263610411173665, + "learning_rate": 3.939783060543167e-05, + "loss": 0.403, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03430716320872307, + "step": 212, + "valid_targets_mean": 3398.6, + "valid_targets_min": 875 }, { "epoch": 0.8627848101265823, - "grad_norm": 0.40961026413253804, - "learning_rate": 3.935117547697387e-05, - "loss": 0.3767, - "step": 213 + "grad_norm": 0.4269425843758099, + "learning_rate": 3.938397998399333e-05, + "loss": 0.3875, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04929053410887718, + "step": 213, + "valid_targets_mean": 4928.9, + "valid_targets_min": 992 }, { "epoch": 0.8668354430379747, - "grad_norm": 0.4406667742487598, - "learning_rate": 3.933675771619675e-05, - "loss": 0.3724, - "step": 214 + "grad_norm": 0.4507167614566644, + "learning_rate": 3.936997436876756e-05, + "loss": 0.3931, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05837477371096611, + "step": 214, + "valid_targets_mean": 5017.2, + "valid_targets_min": 1566 }, { "epoch": 0.8708860759493671, - "grad_norm": 0.4322305722373633, - "learning_rate": 3.932218421983131e-05, - "loss": 0.3745, - "step": 215 + "grad_norm": 0.4178256937125121, + "learning_rate": 3.935581387174289e-05, + "loss": 0.4008, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.054189737886190414, + "step": 215, + "valid_targets_mean": 4754.5, + "valid_targets_min": 1273 }, { "epoch": 0.8749367088607595, - "grad_norm": 0.39692139508630897, - "learning_rate": 3.9307455105250484e-05, - "loss": 0.3792, - "step": 216 + "grad_norm": 0.4424970209938244, + "learning_rate": 3.9341498606146295e-05, + "loss": 0.4035, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.07565446197986603, + "step": 216, + "valid_targets_mean": 5814.6, + "valid_targets_min": 1326 }, { "epoch": 0.8789873417721519, - "grad_norm": 0.39211956930153335, - "learning_rate": 3.929257049108054e-05, - "loss": 0.3592, - "step": 217 + "grad_norm": 0.4163177338806734, + "learning_rate": 3.932702868644224e-05, + "loss": 0.3875, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04473324120044708, + "step": 217, + "valid_targets_mean": 3958.6, + "valid_targets_min": 1551 }, { "epoch": 0.8830379746835443, - "grad_norm": 0.4168765456638921, - "learning_rate": 3.927753049720011e-05, - "loss": 0.3661, - "step": 218 + "grad_norm": 0.46312477737105173, + "learning_rate": 3.931240422833181e-05, + "loss": 0.3834, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04033679515123367, + "step": 218, + "valid_targets_mean": 3933.8, + "valid_targets_min": 1506 }, { "epoch": 0.8870886075949367, - "grad_norm": 0.4261728514980525, - "learning_rate": 3.9262335244739234e-05, - "loss": 0.3611, - "step": 219 + "grad_norm": 0.4231326125132253, + "learning_rate": 3.9297625348751806e-05, + "loss": 0.3806, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04238021373748779, + "step": 219, + "valid_targets_mean": 4357.4, + "valid_targets_min": 1142 }, { "epoch": 0.8911392405063291, - "grad_norm": 0.42900491455321205, - "learning_rate": 3.92469848560784e-05, - "loss": 0.3561, - "step": 220 + "grad_norm": 0.3722935718035281, + "learning_rate": 3.9282692165873744e-05, + "loss": 0.3786, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06605974584817886, + "step": 220, + "valid_targets_mean": 5542.3, + "valid_targets_min": 1528 }, { "epoch": 0.8951898734177215, - "grad_norm": 0.46058162414644876, - "learning_rate": 3.923147945484751e-05, - "loss": 0.3694, - "step": 221 + "grad_norm": 0.4198846670074615, + "learning_rate": 3.9267604799102945e-05, + "loss": 0.3972, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04203680902719498, + "step": 221, + "valid_targets_mean": 3785.0, + "valid_targets_min": 1059 }, { "epoch": 0.8992405063291139, - "grad_norm": 0.44165633680367516, - "learning_rate": 3.9215819165924956e-05, - "loss": 0.3549, - "step": 222 + "grad_norm": 0.3935157764748133, + "learning_rate": 3.9252363369077615e-05, + "loss": 0.3751, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03383598476648331, + "step": 222, + "valid_targets_mean": 3525.6, + "valid_targets_min": 971 }, { "epoch": 0.9032911392405063, - "grad_norm": 0.5460218821199697, - "learning_rate": 3.920000411543654e-05, - "loss": 0.3731, - "step": 223 + "grad_norm": 0.5234426202913071, + "learning_rate": 3.9236967997667796e-05, + "loss": 0.3866, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06510619819164276, + "step": 223, + "valid_targets_mean": 5984.1, + "valid_targets_min": 822 }, { "epoch": 0.9073417721518987, - "grad_norm": 0.4938065359019653, - "learning_rate": 3.9184034430754495e-05, - "loss": 0.3505, - "step": 224 + "grad_norm": 0.4503790413013757, + "learning_rate": 3.9221418807974496e-05, + "loss": 0.3746, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.045053329318761826, + "step": 224, + "valid_targets_mean": 4381.6, + "valid_targets_min": 892 }, { "epoch": 0.9113924050632911, - "grad_norm": 0.4508355784135828, - "learning_rate": 3.916791024049648e-05, - "loss": 0.3553, - "step": 225 + "grad_norm": 0.4530961552486848, + "learning_rate": 3.92057159243286e-05, + "loss": 0.3766, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05290493741631508, + "step": 225, + "valid_targets_mean": 5423.4, + "valid_targets_min": 1471 }, { "epoch": 0.9154430379746835, - "grad_norm": 0.48359403997241646, - "learning_rate": 3.91516316745245e-05, - "loss": 0.3644, - "step": 226 + "grad_norm": 0.4629682871370494, + "learning_rate": 3.9189859472289956e-05, + "loss": 0.3785, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04433508589863777, + "step": 226, + "valid_targets_mean": 4834.6, + "valid_targets_min": 1760 }, { "epoch": 0.9194936708860759, - "grad_norm": 0.544720421516036, - "learning_rate": 3.913519886394389e-05, - "loss": 0.3876, - "step": 227 + "grad_norm": 0.4791394492679248, + "learning_rate": 3.9173849578646324e-05, + "loss": 0.4097, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05587176978588104, + "step": 227, + "valid_targets_mean": 5096.2, + "valid_targets_min": 1041 }, { "epoch": 0.9235443037974683, - "grad_norm": 0.4118678349885707, - "learning_rate": 3.911861194110225e-05, - "loss": 0.3713, - "step": 228 + "grad_norm": 0.5245116702704936, + "learning_rate": 3.9157686371412395e-05, + "loss": 0.3854, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0511966198682785, + "step": 228, + "valid_targets_mean": 4770.9, + "valid_targets_min": 1217 }, { "epoch": 0.9275949367088607, - "grad_norm": 0.5112105119981865, - "learning_rate": 3.910187103958837e-05, - "loss": 0.3681, - "step": 229 + "grad_norm": 0.4262501396573589, + "learning_rate": 3.914136997982874e-05, + "loss": 0.3961, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04806748032569885, + "step": 229, + "valid_targets_mean": 3857.7, + "valid_targets_min": 1186 }, { "epoch": 0.9316455696202531, - "grad_norm": 0.4105441927327187, - "learning_rate": 3.908497629423117e-05, - "loss": 0.3715, - "step": 230 + "grad_norm": 0.4264988999563365, + "learning_rate": 3.912490053436079e-05, + "loss": 0.3945, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04742566496133804, + "step": 230, + "valid_targets_mean": 4552.2, + "valid_targets_min": 1180 }, { "epoch": 0.9356962025316455, - "grad_norm": 0.4644045076064998, - "learning_rate": 3.9067927841098614e-05, - "loss": 0.3597, - "step": 231 + "grad_norm": 0.5040072717810649, + "learning_rate": 3.9108278166697805e-05, + "loss": 0.3938, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04683136194944382, + "step": 231, + "valid_targets_mean": 4043.7, + "valid_targets_min": 1310 }, { "epoch": 0.9397468354430379, - "grad_norm": 0.3989283424653455, - "learning_rate": 3.9050725817496594e-05, - "loss": 0.355, - "step": 232 + "grad_norm": 0.35864337523854756, + "learning_rate": 3.909150300975178e-05, + "loss": 0.3785, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.044740431010723114, + "step": 232, + "valid_targets_mean": 4040.9, + "valid_targets_min": 1382 }, { "epoch": 0.9437974683544303, - "grad_norm": 0.45332897262360705, - "learning_rate": 3.9033370361967844e-05, - "loss": 0.3668, - "step": 233 + "grad_norm": 0.4341125223055753, + "learning_rate": 3.907457519765642e-05, + "loss": 0.3944, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04963330924510956, + "step": 233, + "valid_targets_mean": 4824.9, + "valid_targets_min": 932 }, { "epoch": 0.9478481012658228, - "grad_norm": 0.41886635544985723, - "learning_rate": 3.901586161429081e-05, - "loss": 0.3713, - "step": 234 + "grad_norm": 0.4214466788514605, + "learning_rate": 3.905749486576607e-05, + "loss": 0.39, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05568487197160721, + "step": 234, + "valid_targets_mean": 4694.1, + "valid_targets_min": 1123 }, { "epoch": 0.9518987341772152, - "grad_norm": 0.40715278318792036, - "learning_rate": 3.8998199715478545e-05, - "loss": 0.3625, - "step": 235 + "grad_norm": 0.4490916948287539, + "learning_rate": 3.90402621506546e-05, + "loss": 0.382, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.045049626380205154, + "step": 235, + "valid_targets_mean": 4411.8, + "valid_targets_min": 698 }, { "epoch": 0.9559493670886076, - "grad_norm": 0.4729078180980691, - "learning_rate": 3.8980384807777564e-05, - "loss": 0.36, - "step": 236 + "grad_norm": 0.4469433625250188, + "learning_rate": 3.902287719011435e-05, + "loss": 0.3805, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.046605631709098816, + "step": 236, + "valid_targets_mean": 4418.7, + "valid_targets_min": 1229 }, { "epoch": 0.96, - "grad_norm": 0.4102015941268841, - "learning_rate": 3.896241703466667e-05, - "loss": 0.3814, - "step": 237 + "grad_norm": 0.4116996636586392, + "learning_rate": 3.900534012315499e-05, + "loss": 0.3984, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05655079334974289, + "step": 237, + "valid_targets_mean": 5258.9, + "valid_targets_min": 1299 }, { "epoch": 0.9640506329113924, - "grad_norm": 0.4305575571401939, - "learning_rate": 3.894429654085585e-05, - "loss": 0.3653, - "step": 238 + "grad_norm": 0.43553038453231674, + "learning_rate": 3.8987651090002454e-05, + "loss": 0.3955, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.041694432497024536, + "step": 238, + "valid_targets_mean": 4233.6, + "valid_targets_min": 1580 }, { "epoch": 0.9681012658227848, - "grad_norm": 0.433837803321305, - "learning_rate": 3.892602347228505e-05, - "loss": 0.3794, - "step": 239 + "grad_norm": 0.3897273392949861, + "learning_rate": 3.896981023209777e-05, + "loss": 0.3958, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06257250905036926, + "step": 239, + "valid_targets_mean": 6301.8, + "valid_targets_min": 1614 }, { "epoch": 0.9721518987341772, - "grad_norm": 0.43435638402327614, - "learning_rate": 3.890759797612307e-05, - "loss": 0.3502, - "step": 240 + "grad_norm": 0.40644566913890406, + "learning_rate": 3.8951817692095966e-05, + "loss": 0.3799, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05280187726020813, + "step": 240, + "valid_targets_mean": 4621.9, + "valid_targets_min": 1158 }, { "epoch": 0.9762025316455696, - "grad_norm": 0.38073355766199574, - "learning_rate": 3.888902020076632e-05, - "loss": 0.3728, - "step": 241 + "grad_norm": 0.4031570803146659, + "learning_rate": 3.893367361386491e-05, + "loss": 0.3937, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06152264028787613, + "step": 241, + "valid_targets_mean": 5334.7, + "valid_targets_min": 1224 }, { "epoch": 0.980253164556962, - "grad_norm": 0.4242913532591757, - "learning_rate": 3.887029029583764e-05, - "loss": 0.36, - "step": 242 + "grad_norm": 0.40558630009805174, + "learning_rate": 3.891537814248417e-05, + "loss": 0.3883, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04186493530869484, + "step": 242, + "valid_targets_mean": 3455.3, + "valid_targets_min": 1365 }, { "epoch": 0.9843037974683544, - "grad_norm": 0.36490913133345676, - "learning_rate": 3.8851408412185125e-05, - "loss": 0.3746, - "step": 243 + "grad_norm": 0.3926160221949178, + "learning_rate": 3.889693142424384e-05, + "loss": 0.3994, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04740064591169357, + "step": 243, + "valid_targets_mean": 4265.6, + "valid_targets_min": 1488 }, { "epoch": 0.9883544303797468, - "grad_norm": 0.4115648846497593, - "learning_rate": 3.8832374701880855e-05, - "loss": 0.3732, - "step": 244 + "grad_norm": 0.40598571265510414, + "learning_rate": 3.8878333606643405e-05, + "loss": 0.3852, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03236037492752075, + "step": 244, + "valid_targets_mean": 3068.9, + "valid_targets_min": 1181 }, { "epoch": 0.9924050632911392, - "grad_norm": 0.382223367664726, - "learning_rate": 3.881318931821972e-05, - "loss": 0.3631, - "step": 245 + "grad_norm": 0.36199236830013165, + "learning_rate": 3.885958483839049e-05, + "loss": 0.3896, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06099804490804672, + "step": 245, + "valid_targets_mean": 4501.0, + "valid_targets_min": 1186 }, { "epoch": 0.9964556962025316, - "grad_norm": 0.4269495405065219, - "learning_rate": 3.879385241571817e-05, - "loss": 0.3651, - "step": 246 + "grad_norm": 0.4415810554714419, + "learning_rate": 3.8840685269399786e-05, + "loss": 0.3885, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05218656361103058, + "step": 246, + "valid_targets_mean": 4923.1, + "valid_targets_min": 1119 }, { - "epoch": 1.000506329113924, - "grad_norm": 0.3857743605396771, - "learning_rate": 3.8774364150112955e-05, - "loss": 0.3439, - "step": 247 + "epoch": 1.0, + "grad_norm": 0.4935612926224235, + "learning_rate": 3.882163505079171e-05, + "loss": 0.3867, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.040381744503974915, + "step": 247, + "valid_targets_mean": 3306.5, + "valid_targets_min": 811 + }, + { + "epoch": 1.0040506329113925, + "grad_norm": 0.43881757415111144, + "learning_rate": 3.880243433489132e-05, + "loss": 0.3333, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.043302834033966064, + "step": 248, + "valid_targets_mean": 5179.9, + "valid_targets_min": 1324 + }, + { + "epoch": 1.0081012658227848, + "grad_norm": 0.4768130337286285, + "learning_rate": 3.878308327522702e-05, + "loss": 0.3403, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03876558691263199, + "step": 249, + "valid_targets_mean": 3617.6, + "valid_targets_min": 1289 + }, + { + "epoch": 1.0121518987341773, + "grad_norm": 0.46379437071742236, + "learning_rate": 3.876358202652936e-05, + "loss": 0.3445, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.042863473296165466, + "step": 250, + "valid_targets_mean": 4636.4, + "valid_targets_min": 1324 + }, + { + "epoch": 1.0162025316455696, + "grad_norm": 0.479982525586596, + "learning_rate": 3.8743930744729835e-05, + "loss": 0.3278, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04512117803096771, + "step": 251, + "valid_targets_mean": 5158.1, + "valid_targets_min": 1022 + }, + { + "epoch": 1.0202531645569621, + "grad_norm": 0.4073467211154948, + "learning_rate": 3.8724129586959535e-05, + "loss": 0.3314, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.050884395837783813, + "step": 252, + "valid_targets_mean": 5163.6, + "valid_targets_min": 1070 + }, + { + "epoch": 1.0243037974683544, + "grad_norm": 0.49364056497777786, + "learning_rate": 3.870417871154801e-05, + "loss": 0.3368, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05375582352280617, + "step": 253, + "valid_targets_mean": 5155.7, + "valid_targets_min": 1448 + }, + { + "epoch": 1.028354430379747, + "grad_norm": 0.42136875982681915, + "learning_rate": 3.868407827802192e-05, + "loss": 0.3279, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.043836839497089386, + "step": 254, + "valid_targets_mean": 4467.7, + "valid_targets_min": 1257 + }, + { + "epoch": 1.0324050632911392, + "grad_norm": 0.49067221875631695, + "learning_rate": 3.8663828447103804e-05, + "loss": 0.3467, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0388471893966198, + "step": 255, + "valid_targets_mean": 4225.1, + "valid_targets_min": 860 + }, + { + "epoch": 1.0364556962025318, + "grad_norm": 0.400665256592623, + "learning_rate": 3.8643429380710765e-05, + "loss": 0.3411, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.036631450057029724, + "step": 256, + "valid_targets_mean": 4231.9, + "valid_targets_min": 1610 + }, + { + "epoch": 1.040506329113924, + "grad_norm": 0.49979735329761005, + "learning_rate": 3.862288124195319e-05, + "loss": 0.3425, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04496927186846733, + "step": 257, + "valid_targets_mean": 5401.9, + "valid_targets_min": 1217 + }, + { + "epoch": 1.0445569620253166, + "grad_norm": 0.40250127249183065, + "learning_rate": 3.8602184195133464e-05, + "loss": 0.3291, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03760378807783127, + "step": 258, + "valid_targets_mean": 3832.8, + "valid_targets_min": 1619 + }, + { + "epoch": 1.0486075949367089, + "grad_norm": 0.4993152725336139, + "learning_rate": 3.8581338405744604e-05, + "loss": 0.3409, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.047988615930080414, + "step": 259, + "valid_targets_mean": 5089.2, + "valid_targets_min": 1207 + }, + { + "epoch": 1.0526582278481014, + "grad_norm": 0.4833916443613953, + "learning_rate": 3.8560344040468996e-05, + "loss": 0.3466, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04081379622220993, + "step": 260, + "valid_targets_mean": 4535.2, + "valid_targets_min": 1403 + }, + { + "epoch": 1.0567088607594937, + "grad_norm": 0.49185705402569097, + "learning_rate": 3.853920126717703e-05, + "loss": 0.3506, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04410317540168762, + "step": 261, + "valid_targets_mean": 4016.5, + "valid_targets_min": 1143 + }, + { + "epoch": 1.0607594936708862, + "grad_norm": 0.4732534934525123, + "learning_rate": 3.851791025492574e-05, + "loss": 0.3408, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.048802439123392105, + "step": 262, + "valid_targets_mean": 5977.5, + "valid_targets_min": 1318 + }, + { + "epoch": 1.0648101265822785, + "grad_norm": 0.4651821986477631, + "learning_rate": 3.8496471173957515e-05, + "loss": 0.3429, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04655998945236206, + "step": 263, + "valid_targets_mean": 5134.2, + "valid_targets_min": 906 + }, + { + "epoch": 1.068860759493671, + "grad_norm": 0.44444401244665843, + "learning_rate": 3.847488419569866e-05, + "loss": 0.3326, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05208176374435425, + "step": 264, + "valid_targets_mean": 5735.6, + "valid_targets_min": 1299 + }, + { + "epoch": 1.0729113924050633, + "grad_norm": 0.43965189163949886, + "learning_rate": 3.8453149492758077e-05, + "loss": 0.3336, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04766218364238739, + "step": 265, + "valid_targets_mean": 5380.4, + "valid_targets_min": 1442 + }, + { + "epoch": 1.0769620253164558, + "grad_norm": 0.44147871823418006, + "learning_rate": 3.8431267238925886e-05, + "loss": 0.3274, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06004750728607178, + "step": 266, + "valid_targets_mean": 5674.8, + "valid_targets_min": 1688 + }, + { + "epoch": 1.081012658227848, + "grad_norm": 0.40035732042631894, + "learning_rate": 3.840923760917199e-05, + "loss": 0.334, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028999630361795425, + "step": 267, + "valid_targets_mean": 4041.1, + "valid_targets_min": 1463 + }, + { + "epoch": 1.0850632911392406, + "grad_norm": 0.4399057296115138, + "learning_rate": 3.838706077964473e-05, + "loss": 0.3262, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.041777171194553375, + "step": 268, + "valid_targets_mean": 5098.9, + "valid_targets_min": 1078 + }, + { + "epoch": 1.089113924050633, + "grad_norm": 0.4193239485777806, + "learning_rate": 3.836473692766944e-05, + "loss": 0.3288, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04640112817287445, + "step": 269, + "valid_targets_mean": 4746.3, + "valid_targets_min": 1270 + }, + { + "epoch": 1.0931645569620254, + "grad_norm": 0.4178067626741526, + "learning_rate": 3.834226623174705e-05, + "loss": 0.3229, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03233584016561508, + "step": 270, + "valid_targets_mean": 3726.8, + "valid_targets_min": 688 }, { - "epoch": 1.0045569620253165, - "grad_norm": 0.5182363861417547, - "learning_rate": 3.8754724678359884e-05, - "loss": 0.3144, - "step": 248 + "epoch": 1.0972151898734177, + "grad_norm": 0.4277304389800791, + "learning_rate": 3.831964887155264e-05, + "loss": 0.3194, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03970283269882202, + "step": 271, + "valid_targets_mean": 4546.9, + "valid_targets_min": 1055 + }, + { + "epoch": 1.1012658227848102, + "grad_norm": 0.39751963960825176, + "learning_rate": 3.829688502793403e-05, + "loss": 0.332, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.050614580512046814, + "step": 272, + "valid_targets_mean": 6069.5, + "valid_targets_min": 1198 + }, + { + "epoch": 1.1053164556962025, + "grad_norm": 0.39667785576607584, + "learning_rate": 3.827397488291029e-05, + "loss": 0.3365, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.052200429141521454, + "step": 273, + "valid_targets_mean": 4859.7, + "valid_targets_min": 1591 + }, + { + "epoch": 1.109367088607595, + "grad_norm": 0.4631771987391291, + "learning_rate": 3.8250918619670343e-05, + "loss": 0.3294, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05346325412392616, + "step": 274, + "valid_targets_mean": 5889.3, + "valid_targets_min": 1285 + }, + { + "epoch": 1.1134177215189873, + "grad_norm": 0.4095769895472331, + "learning_rate": 3.822771642257145e-05, + "loss": 0.3167, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.042441610246896744, + "step": 275, + "valid_targets_mean": 4516.6, + "valid_targets_min": 1607 + }, + { + "epoch": 1.1174683544303798, + "grad_norm": 0.42832336914382285, + "learning_rate": 3.820436847713776e-05, + "loss": 0.319, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.033627428114414215, + "step": 276, + "valid_targets_mean": 3252.9, + "valid_targets_min": 1235 + }, + { + "epoch": 1.1215189873417721, + "grad_norm": 0.41935447022143024, + "learning_rate": 3.8180874970058826e-05, + "loss": 0.3395, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.046168241649866104, + "step": 277, + "valid_targets_mean": 4573.6, + "valid_targets_min": 1322 + }, + { + "epoch": 1.1255696202531646, + "grad_norm": 0.42766634759524186, + "learning_rate": 3.8157236089188103e-05, + "loss": 0.3336, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03643731027841568, + "step": 278, + "valid_targets_mean": 4285.5, + "valid_targets_min": 1489 + }, + { + "epoch": 1.129620253164557, + "grad_norm": 0.4137474754288218, + "learning_rate": 3.813345202354145e-05, + "loss": 0.3324, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03984470292925835, + "step": 279, + "valid_targets_mean": 4170.9, + "valid_targets_min": 920 + }, + { + "epoch": 1.1336708860759495, + "grad_norm": 0.5587127024191587, + "learning_rate": 3.810952296329563e-05, + "loss": 0.3199, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03912029415369034, + "step": 280, + "valid_targets_mean": 4313.5, + "valid_targets_min": 1575 + }, + { + "epoch": 1.1377215189873418, + "grad_norm": 0.38188114391239336, + "learning_rate": 3.808544909978676e-05, + "loss": 0.3357, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05175253003835678, + "step": 281, + "valid_targets_mean": 5816.8, + "valid_targets_min": 1139 + }, + { + "epoch": 1.1417721518987343, + "grad_norm": 0.48292556377655704, + "learning_rate": 3.806123062550882e-05, + "loss": 0.3164, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05137088894844055, + "step": 282, + "valid_targets_mean": 5902.8, + "valid_targets_min": 1175 + }, + { + "epoch": 1.1458227848101266, + "grad_norm": 0.4688024194245287, + "learning_rate": 3.803686773411208e-05, + "loss": 0.3259, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.046909987926483154, + "step": 283, + "valid_targets_mean": 5458.0, + "valid_targets_min": 961 + }, + { + "epoch": 1.149873417721519, + "grad_norm": 0.4010452184684908, + "learning_rate": 3.801236062040158e-05, + "loss": 0.3475, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.033218979835510254, + "step": 284, + "valid_targets_mean": 3854.1, + "valid_targets_min": 1119 + }, + { + "epoch": 1.1539240506329114, + "grad_norm": 0.48370238933037013, + "learning_rate": 3.798770948033553e-05, + "loss": 0.3474, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.025871790945529938, + "step": 285, + "valid_targets_mean": 3305.8, + "valid_targets_min": 956 + }, + { + "epoch": 1.1579746835443039, + "grad_norm": 0.4471042978546612, + "learning_rate": 3.796291451102381e-05, + "loss": 0.3272, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04039275646209717, + "step": 286, + "valid_targets_mean": 4557.0, + "valid_targets_min": 1476 + }, + { + "epoch": 1.1620253164556962, + "grad_norm": 0.38851396235800023, + "learning_rate": 3.793797591072631e-05, + "loss": 0.3399, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03040398843586445, + "step": 287, + "valid_targets_mean": 3757.9, + "valid_targets_min": 1198 + }, + { + "epoch": 1.1660759493670887, + "grad_norm": 0.4377443454251537, + "learning_rate": 3.791289387885144e-05, + "loss": 0.3307, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04426418989896774, + "step": 288, + "valid_targets_mean": 4752.1, + "valid_targets_min": 1596 + }, + { + "epoch": 1.170126582278481, + "grad_norm": 0.40605885786934004, + "learning_rate": 3.788766861595443e-05, + "loss": 0.3366, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03723035007715225, + "step": 289, + "valid_targets_mean": 4943.8, + "valid_targets_min": 1541 + }, + { + "epoch": 1.1741772151898735, + "grad_norm": 0.4138525326746093, + "learning_rate": 3.7862300323735835e-05, + "loss": 0.3372, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.039631813764572144, + "step": 290, + "valid_targets_mean": 4613.4, + "valid_targets_min": 1194 + }, + { + "epoch": 1.1782278481012658, + "grad_norm": 0.36085473699571674, + "learning_rate": 3.783678920503983e-05, + "loss": 0.3236, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03367921710014343, + "step": 291, + "valid_targets_mean": 3855.4, + "valid_targets_min": 1662 + }, + { + "epoch": 1.1822784810126583, + "grad_norm": 0.40026413984436504, + "learning_rate": 3.781113546385265e-05, + "loss": 0.3384, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04290143400430679, + "step": 292, + "valid_targets_mean": 4215.2, + "valid_targets_min": 1378 + }, + { + "epoch": 1.1863291139240506, + "grad_norm": 0.38610007418503284, + "learning_rate": 3.7785339305300916e-05, + "loss": 0.3284, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03657708317041397, + "step": 293, + "valid_targets_mean": 4263.8, + "valid_targets_min": 1159 + }, + { + "epoch": 1.1903797468354431, + "grad_norm": 0.40277194203840083, + "learning_rate": 3.7759400935650025e-05, + "loss": 0.3315, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06683148443698883, + "step": 294, + "valid_targets_mean": 6487.7, + "valid_targets_min": 1705 + }, + { + "epoch": 1.1944303797468354, + "grad_norm": 0.380575022478197, + "learning_rate": 3.77333205623025e-05, + "loss": 0.3234, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.035810552537441254, + "step": 295, + "valid_targets_mean": 4024.1, + "valid_targets_min": 951 + }, + { + "epoch": 1.198481012658228, + "grad_norm": 0.4241047155197534, + "learning_rate": 3.7707098393796295e-05, + "loss": 0.3259, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03711569681763649, + "step": 296, + "valid_targets_mean": 3789.1, + "valid_targets_min": 1619 + }, + { + "epoch": 1.2025316455696202, + "grad_norm": 0.42925599564732586, + "learning_rate": 3.7680734639803185e-05, + "loss": 0.3362, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06396347284317017, + "step": 297, + "valid_targets_mean": 6671.0, + "valid_targets_min": 1322 + }, + { + "epoch": 1.2065822784810127, + "grad_norm": 0.42717609021903125, + "learning_rate": 3.765422951112704e-05, + "loss": 0.3147, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05779135972261429, + "step": 298, + "valid_targets_mean": 6074.8, + "valid_targets_min": 1801 + }, + { + "epoch": 1.210632911392405, + "grad_norm": 0.5503385227348256, + "learning_rate": 3.762758321970216e-05, + "loss": 0.3374, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0427449569106102, + "step": 299, + "valid_targets_mean": 5239.9, + "valid_targets_min": 1335 + }, + { + "epoch": 1.2146835443037975, + "grad_norm": 0.390158538869697, + "learning_rate": 3.7600795978591584e-05, + "loss": 0.3337, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0317438542842865, + "step": 300, + "valid_targets_mean": 3869.2, + "valid_targets_min": 1430 + }, + { + "epoch": 1.2187341772151898, + "grad_norm": 0.46847184119408986, + "learning_rate": 3.757386800198538e-05, + "loss": 0.3343, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.045495979487895966, + "step": 301, + "valid_targets_mean": 5395.8, + "valid_targets_min": 1545 + }, + { + "epoch": 1.2227848101265824, + "grad_norm": 0.4428823267611126, + "learning_rate": 3.7546799505198925e-05, + "loss": 0.3519, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.055398546159267426, + "step": 302, + "valid_targets_mean": 6161.7, + "valid_targets_min": 1237 + }, + { + "epoch": 1.2268354430379746, + "grad_norm": 0.4064436411344038, + "learning_rate": 3.7519590704671197e-05, + "loss": 0.3441, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04253986105322838, + "step": 303, + "valid_targets_mean": 4458.0, + "valid_targets_min": 1259 + }, + { + "epoch": 1.2308860759493672, + "grad_norm": 0.39611039274123905, + "learning_rate": 3.749224181796305e-05, + "loss": 0.3447, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.036902233958244324, + "step": 304, + "valid_targets_mean": 4446.3, + "valid_targets_min": 1412 + }, + { + "epoch": 1.2349367088607595, + "grad_norm": 0.539713223604662, + "learning_rate": 3.746475306375545e-05, + "loss": 0.3396, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.043336015194654465, + "step": 305, + "valid_targets_mean": 4629.6, + "valid_targets_min": 1188 + }, + { + "epoch": 1.238987341772152, + "grad_norm": 0.31666258051552043, + "learning_rate": 3.743712466184774e-05, + "loss": 0.3117, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04101824015378952, + "step": 306, + "valid_targets_mean": 4542.2, + "valid_targets_min": 1250 + }, + { + "epoch": 1.2430379746835443, + "grad_norm": 0.4603641684309312, + "learning_rate": 3.7409356833155885e-05, + "loss": 0.3392, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03385325148701668, + "step": 307, + "valid_targets_mean": 3908.0, + "valid_targets_min": 1289 + }, + { + "epoch": 1.2470886075949368, + "grad_norm": 0.38131629891769525, + "learning_rate": 3.73814497997107e-05, + "loss": 0.3533, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04682439565658569, + "step": 308, + "valid_targets_mean": 5508.6, + "valid_targets_min": 1640 + }, + { + "epoch": 1.251139240506329, + "grad_norm": 0.3977678912424595, + "learning_rate": 3.7353403784656095e-05, + "loss": 0.3348, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.047584839165210724, + "step": 309, + "valid_targets_mean": 6236.3, + "valid_targets_min": 1146 + }, + { + "epoch": 1.2551898734177216, + "grad_norm": 0.3775688600802151, + "learning_rate": 3.732521901224724e-05, + "loss": 0.3181, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04272007197141647, + "step": 310, + "valid_targets_mean": 4600.5, + "valid_targets_min": 1083 + }, + { + "epoch": 1.2592405063291139, + "grad_norm": 0.35415078868301875, + "learning_rate": 3.7296895707848844e-05, + "loss": 0.3266, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05276188254356384, + "step": 311, + "valid_targets_mean": 4910.4, + "valid_targets_min": 1292 + }, + { + "epoch": 1.2632911392405064, + "grad_norm": 0.37644340464983783, + "learning_rate": 3.7268434097933275e-05, + "loss": 0.3174, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05105176940560341, + "step": 312, + "valid_targets_mean": 5747.3, + "valid_targets_min": 1282 + }, + { + "epoch": 1.2673417721518987, + "grad_norm": 0.4146221684659416, + "learning_rate": 3.723983441007882e-05, + "loss": 0.3451, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.045349009335041046, + "step": 313, + "valid_targets_mean": 4974.2, + "valid_targets_min": 1128 + }, + { + "epoch": 1.2713924050632912, + "grad_norm": 0.3823656488056918, + "learning_rate": 3.721109687296781e-05, + "loss": 0.32, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05362933874130249, + "step": 314, + "valid_targets_mean": 5927.6, + "valid_targets_min": 1286 + }, + { + "epoch": 1.2754430379746835, + "grad_norm": 0.3980190324687026, + "learning_rate": 3.718222171638484e-05, + "loss": 0.3467, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03732815384864807, + "step": 315, + "valid_targets_mean": 4203.2, + "valid_targets_min": 1017 + }, + { + "epoch": 1.279493670886076, + "grad_norm": 0.35730350158693447, + "learning_rate": 3.715320917121488e-05, + "loss": 0.3508, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0718134194612503, + "step": 316, + "valid_targets_mean": 5351.9, + "valid_targets_min": 932 + }, + { + "epoch": 1.2835443037974683, + "grad_norm": 0.38034364154189443, + "learning_rate": 3.712405946944149e-05, + "loss": 0.3243, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.043744880706071854, + "step": 317, + "valid_targets_mean": 4461.6, + "valid_targets_min": 1355 + }, + { + "epoch": 1.2875949367088608, + "grad_norm": 0.37238286698954254, + "learning_rate": 3.709477284414488e-05, + "loss": 0.3317, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04839417338371277, + "step": 318, + "valid_targets_mean": 5919.4, + "valid_targets_min": 1416 + }, + { + "epoch": 1.2916455696202531, + "grad_norm": 0.3694152773764771, + "learning_rate": 3.7065349529500136e-05, + "loss": 0.3329, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05601927638053894, + "step": 319, + "valid_targets_mean": 6470.4, + "valid_targets_min": 1380 + }, + { + "epoch": 1.2956962025316456, + "grad_norm": 0.40603377803421475, + "learning_rate": 3.7035789760775305e-05, + "loss": 0.3364, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03035106509923935, + "step": 320, + "valid_targets_mean": 3595.9, + "valid_targets_min": 1353 + }, + { + "epoch": 1.299746835443038, + "grad_norm": 0.42842444607862246, + "learning_rate": 3.70060937743295e-05, + "loss": 0.3397, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04682677239179611, + "step": 321, + "valid_targets_mean": 5207.8, + "valid_targets_min": 811 + }, + { + "epoch": 1.3037974683544304, + "grad_norm": 0.3504215938773099, + "learning_rate": 3.6976261807611045e-05, + "loss": 0.3443, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05075697600841522, + "step": 322, + "valid_targets_mean": 5177.2, + "valid_targets_min": 1577 + }, + { + "epoch": 1.3078481012658227, + "grad_norm": 0.4032879194995496, + "learning_rate": 3.694629409915555e-05, + "loss": 0.3293, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.045434221625328064, + "step": 323, + "valid_targets_mean": 4798.3, + "valid_targets_min": 851 + }, + { + "epoch": 1.3118987341772153, + "grad_norm": 0.3539200936827596, + "learning_rate": 3.691619088858401e-05, + "loss": 0.3338, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05697557330131531, + "step": 324, + "valid_targets_mean": 6687.8, + "valid_targets_min": 1313 + }, + { + "epoch": 1.3159493670886075, + "grad_norm": 0.4349578347430558, + "learning_rate": 3.68859524166009e-05, + "loss": 0.3375, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.029414789751172066, + "step": 325, + "valid_targets_mean": 3496.2, + "valid_targets_min": 1144 + }, + { + "epoch": 1.32, + "grad_norm": 0.3891623595744905, + "learning_rate": 3.6855578924992226e-05, + "loss": 0.329, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03618660196661949, + "step": 326, + "valid_targets_mean": 4056.2, + "valid_targets_min": 1196 + }, + { + "epoch": 1.3240506329113924, + "grad_norm": 0.3901311810771893, + "learning_rate": 3.6825070656623626e-05, + "loss": 0.3331, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05077538639307022, + "step": 327, + "valid_targets_mean": 5577.2, + "valid_targets_min": 1710 + }, + { + "epoch": 1.3281012658227849, + "grad_norm": 0.3795594032117602, + "learning_rate": 3.67944278554384e-05, + "loss": 0.3324, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.041123196482658386, + "step": 328, + "valid_targets_mean": 4052.9, + "valid_targets_min": 1029 + }, + { + "epoch": 1.3321518987341772, + "grad_norm": 0.3837261513398238, + "learning_rate": 3.676365076645557e-05, + "loss": 0.3375, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0361938551068306, + "step": 329, + "valid_targets_mean": 3657.7, + "valid_targets_min": 1281 + }, + { + "epoch": 1.3362025316455697, + "grad_norm": 0.3872825195453024, + "learning_rate": 3.673273963576791e-05, + "loss": 0.3322, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05670418217778206, + "step": 330, + "valid_targets_mean": 6128.2, + "valid_targets_min": 1762 + }, + { + "epoch": 1.340253164556962, + "grad_norm": 0.40953104436741, + "learning_rate": 3.670169471054001e-05, + "loss": 0.3351, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.048498839139938354, + "step": 331, + "valid_targets_mean": 4052.1, + "valid_targets_min": 1276 + }, + { + "epoch": 1.3443037974683545, + "grad_norm": 0.3895846012883171, + "learning_rate": 3.6670516239006254e-05, + "loss": 0.3282, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06334970146417618, + "step": 332, + "valid_targets_mean": 5396.1, + "valid_targets_min": 1551 + }, + { + "epoch": 1.3483544303797468, + "grad_norm": 0.4223997770807982, + "learning_rate": 3.663920447046888e-05, + "loss": 0.3318, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03903364762663841, + "step": 333, + "valid_targets_mean": 4347.9, + "valid_targets_min": 1491 + }, + { + "epoch": 1.3524050632911393, + "grad_norm": 0.3390409761293719, + "learning_rate": 3.6607759655295954e-05, + "loss": 0.3199, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04412960261106491, + "step": 334, + "valid_targets_mean": 4938.9, + "valid_targets_min": 733 + }, + { + "epoch": 1.3564556962025316, + "grad_norm": 0.36219279290904527, + "learning_rate": 3.657618204491937e-05, + "loss": 0.3282, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.043282829225063324, + "step": 335, + "valid_targets_mean": 4905.5, + "valid_targets_min": 1403 + }, + { + "epoch": 1.360506329113924, + "grad_norm": 0.3303574390032699, + "learning_rate": 3.6544471891832875e-05, + "loss": 0.3432, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03686103969812393, + "step": 336, + "valid_targets_mean": 4191.9, + "valid_targets_min": 1037 + }, + { + "epoch": 1.3645569620253164, + "grad_norm": 0.37333045985626745, + "learning_rate": 3.6512629449589983e-05, + "loss": 0.3412, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.044646792113780975, + "step": 337, + "valid_targets_mean": 4578.7, + "valid_targets_min": 974 + }, + { + "epoch": 1.368607594936709, + "grad_norm": 0.34535259322026063, + "learning_rate": 3.6480654972802045e-05, + "loss": 0.3283, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03553847223520279, + "step": 338, + "valid_targets_mean": 3828.4, + "valid_targets_min": 1676 + }, + { + "epoch": 1.3726582278481012, + "grad_norm": 0.35059742097235924, + "learning_rate": 3.644854871713611e-05, + "loss": 0.328, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04590984433889389, + "step": 339, + "valid_targets_mean": 4716.9, + "valid_targets_min": 847 + }, + { + "epoch": 1.3767088607594937, + "grad_norm": 0.3850539252862224, + "learning_rate": 3.6416310939312924e-05, + "loss": 0.3319, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05009511113166809, + "step": 340, + "valid_targets_mean": 4437.9, + "valid_targets_min": 1058 + }, + { + "epoch": 1.380759493670886, + "grad_norm": 0.35551096939232074, + "learning_rate": 3.638394189710493e-05, + "loss": 0.33, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.034703344106674194, + "step": 341, + "valid_targets_mean": 3578.6, + "valid_targets_min": 1017 + }, + { + "epoch": 1.3848101265822785, + "grad_norm": 0.46909030035234556, + "learning_rate": 3.635144184933412e-05, + "loss": 0.3488, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.032210104167461395, + "step": 342, + "valid_targets_mean": 3431.7, + "valid_targets_min": 1125 + }, + { + "epoch": 1.3888607594936708, + "grad_norm": 0.36339571305087914, + "learning_rate": 3.6318811055869994e-05, + "loss": 0.3322, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04059069603681564, + "step": 343, + "valid_targets_mean": 5441.1, + "valid_targets_min": 1225 + }, + { + "epoch": 1.3929113924050633, + "grad_norm": 0.40886661117023054, + "learning_rate": 3.628604977762752e-05, + "loss": 0.3403, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04117385670542717, + "step": 344, + "valid_targets_mean": 4661.5, + "valid_targets_min": 1282 + }, + { + "epoch": 1.3969620253164556, + "grad_norm": 0.3479116261748228, + "learning_rate": 3.6253158276565004e-05, + "loss": 0.3484, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04562729597091675, + "step": 345, + "valid_targets_mean": 4236.6, + "valid_targets_min": 1080 + }, + { + "epoch": 1.4010126582278482, + "grad_norm": 0.3708622197529606, + "learning_rate": 3.6220136815682016e-05, + "loss": 0.3311, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.053136058151721954, + "step": 346, + "valid_targets_mean": 4996.6, + "valid_targets_min": 1328 + }, + { + "epoch": 1.4050632911392404, + "grad_norm": 0.3675849834932444, + "learning_rate": 3.618698565901727e-05, + "loss": 0.3378, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06611038744449615, + "step": 347, + "valid_targets_mean": 7163.4, + "valid_targets_min": 1242 + }, + { + "epoch": 1.409113924050633, + "grad_norm": 0.3218894297367269, + "learning_rate": 3.615370507164654e-05, + "loss": 0.3277, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02884174883365631, + "step": 348, + "valid_targets_mean": 3643.3, + "valid_targets_min": 1126 + }, + { + "epoch": 1.4131645569620253, + "grad_norm": 0.36402217004287496, + "learning_rate": 3.61202953196805e-05, + "loss": 0.3332, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030001427978277206, + "step": 349, + "valid_targets_mean": 3776.5, + "valid_targets_min": 1279 + }, + { + "epoch": 1.4172151898734178, + "grad_norm": 0.35653994311682313, + "learning_rate": 3.6086756670262654e-05, + "loss": 0.3458, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06715628504753113, + "step": 350, + "valid_targets_mean": 6776.6, + "valid_targets_min": 1473 + }, + { + "epoch": 1.42126582278481, + "grad_norm": 0.35604711232033076, + "learning_rate": 3.6053089391567135e-05, + "loss": 0.3249, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03905327618122101, + "step": 351, + "valid_targets_mean": 4154.4, + "valid_targets_min": 1459 + }, + { + "epoch": 1.4253164556962026, + "grad_norm": 0.35567231926138243, + "learning_rate": 3.601929375279662e-05, + "loss": 0.3385, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04926755651831627, + "step": 352, + "valid_targets_mean": 4944.0, + "valid_targets_min": 909 + }, + { + "epoch": 1.4293670886075949, + "grad_norm": 0.32837476644100877, + "learning_rate": 3.598537002418012e-05, + "loss": 0.3266, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04684937745332718, + "step": 353, + "valid_targets_mean": 4363.1, + "valid_targets_min": 1286 + }, + { + "epoch": 1.4334177215189874, + "grad_norm": 0.2981576463639634, + "learning_rate": 3.5951318476970885e-05, + "loss": 0.3256, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04351592808961868, + "step": 354, + "valid_targets_mean": 5082.2, + "valid_targets_min": 1073 + }, + { + "epoch": 1.4374683544303797, + "grad_norm": 0.3298612706995704, + "learning_rate": 3.591713938344416e-05, + "loss": 0.3224, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04313363507390022, + "step": 355, + "valid_targets_mean": 4193.2, + "valid_targets_min": 1051 + }, + { + "epoch": 1.4415189873417722, + "grad_norm": 0.3254732866947206, + "learning_rate": 3.588283301689507e-05, + "loss": 0.3278, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.045036278665065765, + "step": 356, + "valid_targets_mean": 5073.4, + "valid_targets_min": 1495 + }, + { + "epoch": 1.4455696202531645, + "grad_norm": 0.3316068107096375, + "learning_rate": 3.5848399651636424e-05, + "loss": 0.3263, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03079894930124283, + "step": 357, + "valid_targets_mean": 3537.6, + "valid_targets_min": 1461 + }, + { + "epoch": 1.449620253164557, + "grad_norm": 0.3604454920738615, + "learning_rate": 3.5813839562996473e-05, + "loss": 0.3317, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.054758235812187195, + "step": 358, + "valid_targets_mean": 5031.2, + "valid_targets_min": 1342 + }, + { + "epoch": 1.4536708860759493, + "grad_norm": 0.33420365879889447, + "learning_rate": 3.5779153027316775e-05, + "loss": 0.3294, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.037421490997076035, + "step": 359, + "valid_targets_mean": 4054.1, + "valid_targets_min": 1186 + }, + { + "epoch": 1.4577215189873418, + "grad_norm": 0.4027058616416431, + "learning_rate": 3.574434032194994e-05, + "loss": 0.3316, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.027023347094655037, + "step": 360, + "valid_targets_mean": 2857.1, + "valid_targets_min": 1384 + }, + { + "epoch": 1.461772151898734, + "grad_norm": 0.46987798450445356, + "learning_rate": 3.5709401725257436e-05, + "loss": 0.3449, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03215904161334038, + "step": 361, + "valid_targets_mean": 3671.6, + "valid_targets_min": 1490 + }, + { + "epoch": 1.4658227848101266, + "grad_norm": 0.3272406858775468, + "learning_rate": 3.5674337516607346e-05, + "loss": 0.3448, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028709491714835167, + "step": 362, + "valid_targets_mean": 3303.6, + "valid_targets_min": 1155 + }, + { + "epoch": 1.469873417721519, + "grad_norm": 0.4371735016087626, + "learning_rate": 3.563914797637216e-05, + "loss": 0.3279, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03896871954202652, + "step": 363, + "valid_targets_mean": 3658.6, + "valid_targets_min": 1059 + }, + { + "epoch": 1.4739240506329114, + "grad_norm": 0.37907120545170914, + "learning_rate": 3.560383338592649e-05, + "loss": 0.3408, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.041140444576740265, + "step": 364, + "valid_targets_mean": 4692.1, + "valid_targets_min": 1321 + }, + { + "epoch": 1.4779746835443037, + "grad_norm": 0.4335216000616064, + "learning_rate": 3.556839402764487e-05, + "loss": 0.3365, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.034600693732500076, + "step": 365, + "valid_targets_mean": 3900.6, + "valid_targets_min": 751 + }, + { + "epoch": 1.4820253164556962, + "grad_norm": 0.3948803916787027, + "learning_rate": 3.553283018489946e-05, + "loss": 0.3293, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.033790454268455505, + "step": 366, + "valid_targets_mean": 3697.8, + "valid_targets_min": 1236 + }, + { + "epoch": 1.4860759493670885, + "grad_norm": 0.34938116772620914, + "learning_rate": 3.54971421420578e-05, + "loss": 0.3092, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030311282724142075, + "step": 367, + "valid_targets_mean": 3477.8, + "valid_targets_min": 1283 + }, + { + "epoch": 1.490126582278481, + "grad_norm": 0.35498446996451205, + "learning_rate": 3.5461330184480524e-05, + "loss": 0.3252, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030405160039663315, + "step": 368, + "valid_targets_mean": 3928.5, + "valid_targets_min": 1395 + }, + { + "epoch": 1.4941772151898733, + "grad_norm": 0.32826187877829, + "learning_rate": 3.54253945985191e-05, + "loss": 0.3283, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.047535236924886703, + "step": 369, + "valid_targets_mean": 5368.7, + "valid_targets_min": 1139 + }, + { + "epoch": 1.4982278481012659, + "grad_norm": 0.3797114148319085, + "learning_rate": 3.5389335671513524e-05, + "loss": 0.3283, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04450792446732521, + "step": 370, + "valid_targets_mean": 4625.2, + "valid_targets_min": 1062 + }, + { + "epoch": 1.5022784810126582, + "grad_norm": 0.303916602209006, + "learning_rate": 3.5353153691789986e-05, + "loss": 0.3293, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.041045308113098145, + "step": 371, + "valid_targets_mean": 4232.3, + "valid_targets_min": 952 + }, + { + "epoch": 1.5063291139240507, + "grad_norm": 0.3496379847439436, + "learning_rate": 3.531684894865866e-05, + "loss": 0.3472, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0392393060028553, + "step": 372, + "valid_targets_mean": 4212.8, + "valid_targets_min": 1134 + }, + { + "epoch": 1.5103797468354432, + "grad_norm": 0.30989553544793547, + "learning_rate": 3.528042173241128e-05, + "loss": 0.3321, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.041263651102781296, + "step": 373, + "valid_targets_mean": 4612.2, + "valid_targets_min": 1511 + }, + { + "epoch": 1.5144303797468355, + "grad_norm": 0.3379482122421573, + "learning_rate": 3.524387233431892e-05, + "loss": 0.3163, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04224957525730133, + "step": 374, + "valid_targets_mean": 4965.3, + "valid_targets_min": 1138 + }, + { + "epoch": 1.5184810126582278, + "grad_norm": 0.3074788070468125, + "learning_rate": 3.520720104662958e-05, + "loss": 0.3398, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030848810449242592, + "step": 375, + "valid_targets_mean": 4269.1, + "valid_targets_min": 1068 + }, + { + "epoch": 1.5225316455696203, + "grad_norm": 0.36045776072321795, + "learning_rate": 3.5170408162565904e-05, + "loss": 0.3423, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03196258470416069, + "step": 376, + "valid_targets_mean": 3321.8, + "valid_targets_min": 875 + }, + { + "epoch": 1.5265822784810128, + "grad_norm": 0.32727093276382224, + "learning_rate": 3.513349397632281e-05, + "loss": 0.3351, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04221861809492111, + "step": 377, + "valid_targets_mean": 4121.6, + "valid_targets_min": 1013 + }, + { + "epoch": 1.530632911392405, + "grad_norm": 0.37888143338164704, + "learning_rate": 3.5096458783065145e-05, + "loss": 0.3304, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.053419679403305054, + "step": 378, + "valid_targets_mean": 5939.3, + "valid_targets_min": 1360 + }, + { + "epoch": 1.5346835443037974, + "grad_norm": 0.34124282562131614, + "learning_rate": 3.505930287892533e-05, + "loss": 0.3308, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.031853239983320236, + "step": 379, + "valid_targets_mean": 3459.0, + "valid_targets_min": 1516 + }, + { + "epoch": 1.53873417721519, + "grad_norm": 0.3267953485708573, + "learning_rate": 3.502202656100099e-05, + "loss": 0.3269, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.035576820373535156, + "step": 380, + "valid_targets_mean": 3330.2, + "valid_targets_min": 1476 + }, + { + "epoch": 1.5427848101265824, + "grad_norm": 0.31760612631955587, + "learning_rate": 3.498463012735256e-05, + "loss": 0.3142, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03268013894557953, + "step": 381, + "valid_targets_mean": 3766.8, + "valid_targets_min": 1185 + }, + { + "epoch": 1.5468354430379747, + "grad_norm": 0.3173788992702674, + "learning_rate": 3.494711387700094e-05, + "loss": 0.3322, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05025570094585419, + "step": 382, + "valid_targets_mean": 5228.7, + "valid_targets_min": 1490 + }, + { + "epoch": 1.550886075949367, + "grad_norm": 0.3039081459849771, + "learning_rate": 3.490947810992505e-05, + "loss": 0.3334, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.042354974895715714, + "step": 383, + "valid_targets_mean": 5028.8, + "valid_targets_min": 1195 + }, + { + "epoch": 1.5549367088607595, + "grad_norm": 0.3297043324355368, + "learning_rate": 3.48717231270595e-05, + "loss": 0.3348, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.032343845814466476, + "step": 384, + "valid_targets_mean": 3612.1, + "valid_targets_min": 1624 + }, + { + "epoch": 1.558987341772152, + "grad_norm": 0.3118757725648861, + "learning_rate": 3.483384923029211e-05, + "loss": 0.3315, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.024401690810918808, + "step": 385, + "valid_targets_mean": 2570.1, + "valid_targets_min": 1189 + }, + { + "epoch": 1.5630379746835443, + "grad_norm": 0.3342509367926826, + "learning_rate": 3.479585672246156e-05, + "loss": 0.3392, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01928572542965412, + "step": 386, + "valid_targets_mean": 2404.0, + "valid_targets_min": 951 + }, + { + "epoch": 1.5670886075949366, + "grad_norm": 0.33034978184042413, + "learning_rate": 3.4757745907354924e-05, + "loss": 0.3401, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04664783552289009, + "step": 387, + "valid_targets_mean": 4955.7, + "valid_targets_min": 1696 + }, + { + "epoch": 1.5711392405063291, + "grad_norm": 0.3128272729667946, + "learning_rate": 3.471951708970527e-05, + "loss": 0.314, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.032305166125297546, + "step": 388, + "valid_targets_mean": 4388.4, + "valid_targets_min": 1445 + }, + { + "epoch": 1.5751898734177217, + "grad_norm": 0.3278781730514049, + "learning_rate": 3.468117057518921e-05, + "loss": 0.3394, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023931726813316345, + "step": 389, + "valid_targets_mean": 3438.9, + "valid_targets_min": 801 + }, + { + "epoch": 1.579240506329114, + "grad_norm": 0.3035670296435243, + "learning_rate": 3.4642706670424426e-05, + "loss": 0.309, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028833506628870964, + "step": 390, + "valid_targets_mean": 3001.4, + "valid_targets_min": 1320 + }, + { + "epoch": 1.5832911392405062, + "grad_norm": 0.30047887062670176, + "learning_rate": 3.460412568296731e-05, + "loss": 0.3329, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03830994665622711, + "step": 391, + "valid_targets_mean": 4628.4, + "valid_targets_min": 828 + }, + { + "epoch": 1.5873417721518988, + "grad_norm": 0.31495946252083773, + "learning_rate": 3.456542792131039e-05, + "loss": 0.3336, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.040600620210170746, + "step": 392, + "valid_targets_mean": 4804.1, + "valid_targets_min": 1048 + }, + { + "epoch": 1.5913924050632913, + "grad_norm": 0.30159967585448116, + "learning_rate": 3.4526613694879936e-05, + "loss": 0.3296, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.048246171325445175, + "step": 393, + "valid_targets_mean": 5419.3, + "valid_targets_min": 2118 + }, + { + "epoch": 1.5954430379746836, + "grad_norm": 0.30585703916726353, + "learning_rate": 3.448768331403348e-05, + "loss": 0.335, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0464690625667572, + "step": 394, + "valid_targets_mean": 4379.9, + "valid_targets_min": 1679 + }, + { + "epoch": 1.5994936708860759, + "grad_norm": 0.3096463737816418, + "learning_rate": 3.44486370900573e-05, + "loss": 0.3306, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028461167588829994, + "step": 395, + "valid_targets_mean": 3153.1, + "valid_targets_min": 968 + }, + { + "epoch": 1.6035443037974684, + "grad_norm": 0.2923490186836879, + "learning_rate": 3.440947533516396e-05, + "loss": 0.3284, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0476490780711174, + "step": 396, + "valid_targets_mean": 4962.8, + "valid_targets_min": 1331 + }, + { + "epoch": 1.6075949367088609, + "grad_norm": 0.3032365316110196, + "learning_rate": 3.4370198362489816e-05, + "loss": 0.339, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03698112815618515, + "step": 397, + "valid_targets_mean": 3731.7, + "valid_targets_min": 1119 + }, + { + "epoch": 1.6116455696202532, + "grad_norm": 0.3014456358442913, + "learning_rate": 3.43308064860925e-05, + "loss": 0.3237, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03145356476306915, + "step": 398, + "valid_targets_mean": 3412.6, + "valid_targets_min": 889 + }, + { + "epoch": 1.6156962025316455, + "grad_norm": 0.29376436754109375, + "learning_rate": 3.4291300020948393e-05, + "loss": 0.3269, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05502167344093323, + "step": 399, + "valid_targets_mean": 5827.9, + "valid_targets_min": 1170 + }, + { + "epoch": 1.619746835443038, + "grad_norm": 0.32590073619863685, + "learning_rate": 3.425167928295015e-05, + "loss": 0.3342, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04351133480668068, + "step": 400, + "valid_targets_mean": 4124.1, + "valid_targets_min": 967 + }, + { + "epoch": 1.6237974683544305, + "grad_norm": 0.314874821138269, + "learning_rate": 3.421194458890411e-05, + "loss": 0.3357, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03858516365289688, + "step": 401, + "valid_targets_mean": 4577.7, + "valid_targets_min": 923 + }, + { + "epoch": 1.6278481012658228, + "grad_norm": 0.3270329953263554, + "learning_rate": 3.4172096256527865e-05, + "loss": 0.3241, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028715766966342926, + "step": 402, + "valid_targets_mean": 3371.6, + "valid_targets_min": 1516 + }, + { + "epoch": 1.631898734177215, + "grad_norm": 0.37463633917697076, + "learning_rate": 3.4132134604447595e-05, + "loss": 0.3397, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.027709491550922394, + "step": 403, + "valid_targets_mean": 3559.0, + "valid_targets_min": 1318 + }, + { + "epoch": 1.6359493670886076, + "grad_norm": 0.3520789571435681, + "learning_rate": 3.409205995219561e-05, + "loss": 0.3208, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03921286016702652, + "step": 404, + "valid_targets_mean": 3807.2, + "valid_targets_min": 1176 + }, + { + "epoch": 1.6400000000000001, + "grad_norm": 0.3436752269222773, + "learning_rate": 3.4051872620207765e-05, + "loss": 0.3433, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.040631406009197235, + "step": 405, + "valid_targets_mean": 4202.3, + "valid_targets_min": 1173 + }, + { + "epoch": 1.6440506329113924, + "grad_norm": 0.34364647000492415, + "learning_rate": 3.40115729298209e-05, + "loss": 0.3225, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.058425869792699814, + "step": 406, + "valid_targets_mean": 5856.3, + "valid_targets_min": 1262 + }, + { + "epoch": 1.6481012658227847, + "grad_norm": 0.3331724959905717, + "learning_rate": 3.397116120327025e-05, + "loss": 0.326, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04639708250761032, + "step": 407, + "valid_targets_mean": 4204.1, + "valid_targets_min": 1464 + }, + { + "epoch": 1.6521518987341772, + "grad_norm": 0.3137296205528992, + "learning_rate": 3.3930637763686916e-05, + "loss": 0.3421, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04129298776388168, + "step": 408, + "valid_targets_mean": 4056.5, + "valid_targets_min": 1449 + }, + { + "epoch": 1.6562025316455697, + "grad_norm": 0.3289054917261365, + "learning_rate": 3.389000293509524e-05, + "loss": 0.3485, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0556643083691597, + "step": 409, + "valid_targets_mean": 5869.9, + "valid_targets_min": 1631 + }, + { + "epoch": 1.660253164556962, + "grad_norm": 0.33517747820106003, + "learning_rate": 3.3849257042410225e-05, + "loss": 0.3417, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06025295704603195, + "step": 410, + "valid_targets_mean": 5569.8, + "valid_targets_min": 1372 + }, + { + "epoch": 1.6643037974683543, + "grad_norm": 0.3662995924142495, + "learning_rate": 3.380840041143494e-05, + "loss": 0.3475, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05463819205760956, + "step": 411, + "valid_targets_mean": 5496.8, + "valid_targets_min": 940 + }, + { + "epoch": 1.6683544303797468, + "grad_norm": 0.34477861177946845, + "learning_rate": 3.376743336885792e-05, + "loss": 0.3232, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05034920573234558, + "step": 412, + "valid_targets_mean": 5222.1, + "valid_targets_min": 1362 + }, + { + "epoch": 1.6724050632911394, + "grad_norm": 0.36968040457564083, + "learning_rate": 3.372635624225055e-05, + "loss": 0.3387, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04692748934030533, + "step": 413, + "valid_targets_mean": 4201.8, + "valid_targets_min": 1148 + }, + { + "epoch": 1.6764556962025317, + "grad_norm": 0.35143575068405164, + "learning_rate": 3.3685169360064445e-05, + "loss": 0.3264, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04613827168941498, + "step": 414, + "valid_targets_mean": 5270.6, + "valid_targets_min": 1208 + }, + { + "epoch": 1.680506329113924, + "grad_norm": 0.3590426224257718, + "learning_rate": 3.3643873051628804e-05, + "loss": 0.3456, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.06139792501926422, + "step": 415, + "valid_targets_mean": 6543.7, + "valid_targets_min": 1696 + }, + { + "epoch": 1.6845569620253165, + "grad_norm": 0.36142747750980425, + "learning_rate": 3.3602467647147814e-05, + "loss": 0.3385, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04478755593299866, + "step": 416, + "valid_targets_mean": 4169.5, + "valid_targets_min": 1095 + }, + { + "epoch": 1.688607594936709, + "grad_norm": 0.32833771591232075, + "learning_rate": 3.356095347769799e-05, + "loss": 0.3295, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03753453865647316, + "step": 417, + "valid_targets_mean": 4762.8, + "valid_targets_min": 1921 + }, + { + "epoch": 1.6926582278481013, + "grad_norm": 0.32356028745922505, + "learning_rate": 3.351933087522552e-05, + "loss": 0.3348, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.043473415076732635, + "step": 418, + "valid_targets_mean": 4554.4, + "valid_targets_min": 1133 + }, + { + "epoch": 1.6967088607594936, + "grad_norm": 0.33419384875899516, + "learning_rate": 3.3477600172543626e-05, + "loss": 0.3399, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04143213480710983, + "step": 419, + "valid_targets_mean": 4895.8, + "valid_targets_min": 1700 + }, + { + "epoch": 1.700759493670886, + "grad_norm": 0.3107542790151482, + "learning_rate": 3.3435761703329894e-05, + "loss": 0.3377, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.038589876145124435, + "step": 420, + "valid_targets_mean": 3990.7, + "valid_targets_min": 1266 + }, + { + "epoch": 1.7048101265822786, + "grad_norm": 0.3316401272152589, + "learning_rate": 3.3393815802123606e-05, + "loss": 0.3395, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04931402951478958, + "step": 421, + "valid_targets_mean": 5523.5, + "valid_targets_min": 1650 + }, + { + "epoch": 1.7088607594936709, + "grad_norm": 0.31553689061241225, + "learning_rate": 3.335176280432308e-05, + "loss": 0.3211, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028230369091033936, + "step": 422, + "valid_targets_mean": 3250.0, + "valid_targets_min": 1178 + }, + { + "epoch": 1.7129113924050632, + "grad_norm": 0.3758417635262789, + "learning_rate": 3.330960304618295e-05, + "loss": 0.3437, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.049489326775074005, + "step": 423, + "valid_targets_mean": 4991.7, + "valid_targets_min": 1106 + }, + { + "epoch": 1.7169620253164557, + "grad_norm": 0.3936550836798475, + "learning_rate": 3.326733686481153e-05, + "loss": 0.3315, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0469365231692791, + "step": 424, + "valid_targets_mean": 5517.9, + "valid_targets_min": 849 + }, + { + "epoch": 1.7210126582278482, + "grad_norm": 0.34400213540431585, + "learning_rate": 3.3224964598168084e-05, + "loss": 0.3324, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04376298189163208, + "step": 425, + "valid_targets_mean": 5182.9, + "valid_targets_min": 1349 + }, + { + "epoch": 1.7250632911392405, + "grad_norm": 0.33882223743314704, + "learning_rate": 3.318248658506012e-05, + "loss": 0.3319, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05328531190752983, + "step": 426, + "valid_targets_mean": 5232.5, + "valid_targets_min": 1236 + }, + { + "epoch": 1.7291139240506328, + "grad_norm": 0.3243709127485186, + "learning_rate": 3.313990316514069e-05, + "loss": 0.3299, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.035082388669252396, + "step": 427, + "valid_targets_mean": 3614.2, + "valid_targets_min": 1523 + }, + { + "epoch": 1.7331645569620253, + "grad_norm": 0.3660746372986188, + "learning_rate": 3.309721467890571e-05, + "loss": 0.3253, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04420848935842514, + "step": 428, + "valid_targets_mean": 4702.2, + "valid_targets_min": 830 + }, + { + "epoch": 1.7372151898734178, + "grad_norm": 0.29622838869567936, + "learning_rate": 3.305442146769115e-05, + "loss": 0.336, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03971245139837265, + "step": 429, + "valid_targets_mean": 4418.9, + "valid_targets_min": 1136 + }, + { + "epoch": 1.7412658227848101, + "grad_norm": 0.3637366045478612, + "learning_rate": 3.30115238736704e-05, + "loss": 0.3277, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04427309334278107, + "step": 430, + "valid_targets_mean": 4616.5, + "valid_targets_min": 1148 + }, + { + "epoch": 1.7453164556962024, + "grad_norm": 0.3736217004489942, + "learning_rate": 3.296852223985148e-05, + "loss": 0.3139, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03867566958069801, + "step": 431, + "valid_targets_mean": 3565.4, + "valid_targets_min": 1374 + }, + { + "epoch": 1.749367088607595, + "grad_norm": 0.34223310955896535, + "learning_rate": 3.2925416910074315e-05, + "loss": 0.3373, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04823065549135208, + "step": 432, + "valid_targets_mean": 5013.8, + "valid_targets_min": 1178 + }, + { + "epoch": 1.7534177215189874, + "grad_norm": 0.3525464599213832, + "learning_rate": 3.288220822900796e-05, + "loss": 0.3018, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.035914428532123566, + "step": 433, + "valid_targets_mean": 4390.2, + "valid_targets_min": 1513 + }, + { + "epoch": 1.7574683544303797, + "grad_norm": 0.33891873554132135, + "learning_rate": 3.283889654214788e-05, + "loss": 0.3258, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.025227509438991547, + "step": 434, + "valid_targets_mean": 3230.1, + "valid_targets_min": 1312 + }, + { + "epoch": 1.761518987341772, + "grad_norm": 0.3795304300630568, + "learning_rate": 3.279548219581317e-05, + "loss": 0.3332, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02923169732093811, + "step": 435, + "valid_targets_mean": 3928.2, + "valid_targets_min": 1827 + }, + { + "epoch": 1.7655696202531646, + "grad_norm": 0.35314291509357043, + "learning_rate": 3.275196553714379e-05, + "loss": 0.322, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.027506470680236816, + "step": 436, + "valid_targets_mean": 3589.4, + "valid_targets_min": 1689 + }, + { + "epoch": 1.769620253164557, + "grad_norm": 0.3403736589530196, + "learning_rate": 3.270834691409778e-05, + "loss": 0.3271, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.043082475662231445, + "step": 437, + "valid_targets_mean": 4697.5, + "valid_targets_min": 1003 + }, + { + "epoch": 1.7736708860759494, + "grad_norm": 0.3470890475541193, + "learning_rate": 3.266462667544849e-05, + "loss": 0.326, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04425528645515442, + "step": 438, + "valid_targets_mean": 4046.9, + "valid_targets_min": 1103 + }, + { + "epoch": 1.7777215189873417, + "grad_norm": 0.33460584037667124, + "learning_rate": 3.262080517078178e-05, + "loss": 0.3336, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05333739146590233, + "step": 439, + "valid_targets_mean": 6205.1, + "valid_targets_min": 2031 + }, + { + "epoch": 1.7817721518987342, + "grad_norm": 0.332223846117327, + "learning_rate": 3.257688275049323e-05, + "loss": 0.3291, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05119897425174713, + "step": 440, + "valid_targets_mean": 5728.1, + "valid_targets_min": 1444 + }, + { + "epoch": 1.7858227848101267, + "grad_norm": 0.3438853865299442, + "learning_rate": 3.253285976578535e-05, + "loss": 0.3443, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.040141962468624115, + "step": 441, + "valid_targets_mean": 4616.1, + "valid_targets_min": 1306 + }, + { + "epoch": 1.789873417721519, + "grad_norm": 0.3634757057983563, + "learning_rate": 3.2488736568664756e-05, + "loss": 0.3396, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05128256976604462, + "step": 442, + "valid_targets_mean": 4661.9, + "valid_targets_min": 1422 + }, + { + "epoch": 1.7939240506329113, + "grad_norm": 0.35226806103777647, + "learning_rate": 3.244451351193935e-05, + "loss": 0.3228, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030896876007318497, + "step": 443, + "valid_targets_mean": 3503.1, + "valid_targets_min": 786 + }, + { + "epoch": 1.7979746835443038, + "grad_norm": 0.33185212160462785, + "learning_rate": 3.240019094921551e-05, + "loss": 0.3247, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.037942226976156235, + "step": 444, + "valid_targets_mean": 3962.8, + "valid_targets_min": 1230 + }, + { + "epoch": 1.8020253164556963, + "grad_norm": 0.36700704498321507, + "learning_rate": 3.235576923489527e-05, + "loss": 0.3393, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04156826436519623, + "step": 445, + "valid_targets_mean": 4232.2, + "valid_targets_min": 1134 + }, + { + "epoch": 1.8060759493670886, + "grad_norm": 0.32868655839505095, + "learning_rate": 3.2311248724173476e-05, + "loss": 0.3314, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0416606068611145, + "step": 446, + "valid_targets_mean": 4821.2, + "valid_targets_min": 1406 + }, + { + "epoch": 1.810126582278481, + "grad_norm": 0.3718438171345449, + "learning_rate": 3.226662977303494e-05, + "loss": 0.332, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030391667038202286, + "step": 447, + "valid_targets_mean": 3128.0, + "valid_targets_min": 972 + }, + { + "epoch": 1.8141772151898734, + "grad_norm": 0.373982059055943, + "learning_rate": 3.2221912738251626e-05, + "loss": 0.338, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05585087835788727, + "step": 448, + "valid_targets_mean": 5579.8, + "valid_targets_min": 1371 + }, + { + "epoch": 1.818227848101266, + "grad_norm": 0.32577144664806534, + "learning_rate": 3.2177097977379736e-05, + "loss": 0.3202, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.043629519641399384, + "step": 449, + "valid_targets_mean": 5300.1, + "valid_targets_min": 868 + }, + { + "epoch": 1.8222784810126582, + "grad_norm": 0.3881428987379395, + "learning_rate": 3.21321858487569e-05, + "loss": 0.3256, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.045537859201431274, + "step": 450, + "valid_targets_mean": 5358.4, + "valid_targets_min": 1494 + }, + { + "epoch": 1.8263291139240505, + "grad_norm": 0.33679408936863786, + "learning_rate": 3.208717671149931e-05, + "loss": 0.3325, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05366458743810654, + "step": 451, + "valid_targets_mean": 4847.1, + "valid_targets_min": 1025 + }, + { + "epoch": 1.830379746835443, + "grad_norm": 0.3164653551106017, + "learning_rate": 3.204207092549882e-05, + "loss": 0.3497, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04051554203033447, + "step": 452, + "valid_targets_mean": 4086.8, + "valid_targets_min": 1593 + }, + { + "epoch": 1.8344303797468355, + "grad_norm": 0.33504380345770385, + "learning_rate": 3.19968688514201e-05, + "loss": 0.338, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05457250401377678, + "step": 453, + "valid_targets_mean": 6232.4, + "valid_targets_min": 1295 + }, + { + "epoch": 1.8384810126582278, + "grad_norm": 0.2839096408910801, + "learning_rate": 3.195157085069774e-05, + "loss": 0.3432, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.029980182647705078, + "step": 454, + "valid_targets_mean": 3421.4, + "valid_targets_min": 1506 + }, + { + "epoch": 1.8425316455696201, + "grad_norm": 0.3726394710042998, + "learning_rate": 3.190617728553332e-05, + "loss": 0.3129, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0307072214782238, + "step": 455, + "valid_targets_mean": 3492.6, + "valid_targets_min": 1467 + }, + { + "epoch": 1.8465822784810126, + "grad_norm": 0.31640186998479036, + "learning_rate": 3.1860688518892606e-05, + "loss": 0.3455, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04523278772830963, + "step": 456, + "valid_targets_mean": 5002.6, + "valid_targets_min": 1141 + }, + { + "epoch": 1.8506329113924052, + "grad_norm": 0.34971430216593696, + "learning_rate": 3.181510491450254e-05, + "loss": 0.3302, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03884712979197502, + "step": 457, + "valid_targets_mean": 4064.7, + "valid_targets_min": 1235 + }, + { + "epoch": 1.8546835443037974, + "grad_norm": 0.32742203770785705, + "learning_rate": 3.176942683684842e-05, + "loss": 0.354, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.047533608973026276, + "step": 458, + "valid_targets_mean": 5261.7, + "valid_targets_min": 1702 + }, + { + "epoch": 1.8587341772151897, + "grad_norm": 0.3131402584468333, + "learning_rate": 3.1723654651170934e-05, + "loss": 0.3155, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.039669573307037354, + "step": 459, + "valid_targets_mean": 4605.7, + "valid_targets_min": 1480 + }, + { + "epoch": 1.8627848101265823, + "grad_norm": 0.32060824625839895, + "learning_rate": 3.167778872346327e-05, + "loss": 0.3217, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0460028350353241, + "step": 460, + "valid_targets_mean": 4227.1, + "valid_targets_min": 1103 + }, + { + "epoch": 1.8668354430379748, + "grad_norm": 0.2997542609453774, + "learning_rate": 3.163182942046815e-05, + "loss": 0.324, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026472963392734528, + "step": 461, + "valid_targets_mean": 3085.8, + "valid_targets_min": 1151 + }, + { + "epoch": 1.870886075949367, + "grad_norm": 0.3679982728380354, + "learning_rate": 3.158577710967494e-05, + "loss": 0.326, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03438718616962433, + "step": 462, + "valid_targets_mean": 3432.3, + "valid_targets_min": 1345 + }, + { + "epoch": 1.8749367088607594, + "grad_norm": 0.2938267435171152, + "learning_rate": 3.153963215931669e-05, + "loss": 0.3215, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03502851724624634, + "step": 463, + "valid_targets_mean": 4178.2, + "valid_targets_min": 963 + }, + { + "epoch": 1.8789873417721519, + "grad_norm": 0.3585044764978308, + "learning_rate": 3.14933949383672e-05, + "loss": 0.3363, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0659424364566803, + "step": 464, + "valid_targets_mean": 8122.4, + "valid_targets_min": 1108 + }, + { + "epoch": 1.8830379746835444, + "grad_norm": 0.3118791436215761, + "learning_rate": 3.144706581653805e-05, + "loss": 0.3333, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05027894675731659, + "step": 465, + "valid_targets_mean": 5247.6, + "valid_targets_min": 1063 + }, + { + "epoch": 1.8870886075949367, + "grad_norm": 0.3761180103785145, + "learning_rate": 3.140064516427566e-05, + "loss": 0.3463, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03511781617999077, + "step": 466, + "valid_targets_mean": 3987.2, + "valid_targets_min": 893 + }, + { + "epoch": 1.891139240506329, + "grad_norm": 0.42711588287783636, + "learning_rate": 3.135413335275833e-05, + "loss": 0.3368, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04904717206954956, + "step": 467, + "valid_targets_mean": 4935.4, + "valid_targets_min": 1291 + }, + { + "epoch": 1.8951898734177215, + "grad_norm": 0.36317986690264864, + "learning_rate": 3.130753075389327e-05, + "loss": 0.3435, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.042481452226638794, + "step": 468, + "valid_targets_mean": 4847.0, + "valid_targets_min": 1741 + }, + { + "epoch": 1.899240506329114, + "grad_norm": 0.29204436191953415, + "learning_rate": 3.1260837740313625e-05, + "loss": 0.3218, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03251627832651138, + "step": 469, + "valid_targets_mean": 4043.2, + "valid_targets_min": 1435 + }, + { + "epoch": 1.9032911392405063, + "grad_norm": 0.361688803214787, + "learning_rate": 3.121405468537547e-05, + "loss": 0.3417, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04765615239739418, + "step": 470, + "valid_targets_mean": 5419.4, + "valid_targets_min": 1151 + }, + { + "epoch": 1.9073417721518986, + "grad_norm": 0.3191407264558652, + "learning_rate": 3.116718196315488e-05, + "loss": 0.3314, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.050547726452350616, + "step": 471, + "valid_targets_mean": 5128.0, + "valid_targets_min": 1240 + }, + { + "epoch": 1.9113924050632911, + "grad_norm": 0.34633636808004736, + "learning_rate": 3.1120219948444894e-05, + "loss": 0.3439, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03523585945367813, + "step": 472, + "valid_targets_mean": 4280.5, + "valid_targets_min": 1036 + }, + { + "epoch": 1.9154430379746836, + "grad_norm": 0.2968650475339727, + "learning_rate": 3.107316901675252e-05, + "loss": 0.3249, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05320196598768234, + "step": 473, + "valid_targets_mean": 5309.4, + "valid_targets_min": 1250 + }, + { + "epoch": 1.919493670886076, + "grad_norm": 0.3030978413371183, + "learning_rate": 3.1026029544295745e-05, + "loss": 0.3264, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03609578311443329, + "step": 474, + "valid_targets_mean": 4145.1, + "valid_targets_min": 915 + }, + { + "epoch": 1.9235443037974682, + "grad_norm": 0.33869189366733815, + "learning_rate": 3.097880190800057e-05, + "loss": 0.3365, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.051121585071086884, + "step": 475, + "valid_targets_mean": 5403.0, + "valid_targets_min": 1633 + }, + { + "epoch": 1.9275949367088607, + "grad_norm": 0.334386515109002, + "learning_rate": 3.093148648549788e-05, + "loss": 0.3424, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05128683149814606, + "step": 476, + "valid_targets_mean": 4943.8, + "valid_targets_min": 1134 + }, + { + "epoch": 1.9316455696202532, + "grad_norm": 0.3430991389462553, + "learning_rate": 3.088408365512055e-05, + "loss": 0.3201, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04292987287044525, + "step": 477, + "valid_targets_mean": 5557.6, + "valid_targets_min": 1453 + }, + { + "epoch": 1.9356962025316455, + "grad_norm": 0.3639481087556812, + "learning_rate": 3.083659379590034e-05, + "loss": 0.3384, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04263174161314964, + "step": 478, + "valid_targets_mean": 5070.9, + "valid_targets_min": 1419 + }, + { + "epoch": 1.9397468354430378, + "grad_norm": 0.34507267926661267, + "learning_rate": 3.0789017287564913e-05, + "loss": 0.3294, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.045663461089134216, + "step": 479, + "valid_targets_mean": 4628.5, + "valid_targets_min": 869 + }, + { + "epoch": 1.9437974683544303, + "grad_norm": 0.3515284965881423, + "learning_rate": 3.074135451053475e-05, + "loss": 0.3351, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04386589676141739, + "step": 480, + "valid_targets_mean": 4744.3, + "valid_targets_min": 1600 + }, + { + "epoch": 1.9478481012658229, + "grad_norm": 0.37366993391351117, + "learning_rate": 3.069360584592015e-05, + "loss": 0.3183, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04510515183210373, + "step": 481, + "valid_targets_mean": 4698.1, + "valid_targets_min": 1135 + }, + { + "epoch": 1.9518987341772152, + "grad_norm": 0.3289061989714357, + "learning_rate": 3.064577167551816e-05, + "loss": 0.3538, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03470800444483757, + "step": 482, + "valid_targets_mean": 4177.1, + "valid_targets_min": 1143 + }, + { + "epoch": 1.9559493670886074, + "grad_norm": 0.2961429018385786, + "learning_rate": 3.059785238180953e-05, + "loss": 0.3134, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05565746873617172, + "step": 483, + "valid_targets_mean": 6605.4, + "valid_targets_min": 1223 + }, + { + "epoch": 1.96, + "grad_norm": 0.3193374118216837, + "learning_rate": 3.054984834795566e-05, + "loss": 0.3284, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05232729762792587, + "step": 484, + "valid_targets_mean": 6810.3, + "valid_targets_min": 1217 + }, + { + "epoch": 1.9640506329113925, + "grad_norm": 0.2970041554551343, + "learning_rate": 3.0501759957795537e-05, + "loss": 0.3274, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04064800962805748, + "step": 485, + "valid_targets_mean": 4291.6, + "valid_targets_min": 1209 + }, + { + "epoch": 1.9681012658227848, + "grad_norm": 0.3222203224742733, + "learning_rate": 3.0453587595842627e-05, + "loss": 0.3253, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03722073882818222, + "step": 486, + "valid_targets_mean": 3872.8, + "valid_targets_min": 1251 + }, + { + "epoch": 1.972151898734177, + "grad_norm": 0.32905816289022255, + "learning_rate": 3.0405331647281856e-05, + "loss": 0.328, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04163438081741333, + "step": 487, + "valid_targets_mean": 4996.4, + "valid_targets_min": 1017 + }, + { + "epoch": 1.9762025316455696, + "grad_norm": 0.3033504131427964, + "learning_rate": 3.0356992497966508e-05, + "loss": 0.3192, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02825050801038742, + "step": 488, + "valid_targets_mean": 3304.9, + "valid_targets_min": 1152 + }, + { + "epoch": 1.980253164556962, + "grad_norm": 0.3467979919349524, + "learning_rate": 3.0308570534415137e-05, + "loss": 0.3329, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0449945442378521, + "step": 489, + "valid_targets_mean": 5895.3, + "valid_targets_min": 1437 + }, + { + "epoch": 1.9843037974683544, + "grad_norm": 0.32266276242539704, + "learning_rate": 3.0260066143808467e-05, + "loss": 0.3316, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03618864715099335, + "step": 490, + "valid_targets_mean": 3618.2, + "valid_targets_min": 1060 + }, + { + "epoch": 1.9883544303797467, + "grad_norm": 0.30250022892082834, + "learning_rate": 3.021147971398632e-05, + "loss": 0.324, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04590712860226631, + "step": 491, + "valid_targets_mean": 5090.6, + "valid_targets_min": 1261 + }, + { + "epoch": 1.9924050632911392, + "grad_norm": 0.29843459484025175, + "learning_rate": 3.0162811633444478e-05, + "loss": 0.325, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03130924701690674, + "step": 492, + "valid_targets_mean": 3311.7, + "valid_targets_min": 1693 + }, + { + "epoch": 1.9964556962025317, + "grad_norm": 0.3185193871458022, + "learning_rate": 3.0114062291331635e-05, + "loss": 0.3261, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03464746102690697, + "step": 493, + "valid_targets_mean": 3960.8, + "valid_targets_min": 1462 + }, + { + "epoch": 2.0, + "grad_norm": 0.30644069464206103, + "learning_rate": 3.0065232077446217e-05, + "loss": 0.3114, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030792618170380592, + "step": 494, + "valid_targets_mean": 3500.9, + "valid_targets_min": 901 + }, + { + "epoch": 2.0040506329113925, + "grad_norm": 0.4743719160458842, + "learning_rate": 3.001632138223332e-05, + "loss": 0.2585, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030326060950756073, + "step": 495, + "valid_targets_mean": 4105.0, + "valid_targets_min": 1372 + }, + { + "epoch": 2.008101265822785, + "grad_norm": 0.3171536640082535, + "learning_rate": 2.9967330596781565e-05, + "loss": 0.2619, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.040422745048999786, + "step": 496, + "valid_targets_mean": 5578.3, + "valid_targets_min": 1982 + }, + { + "epoch": 2.012151898734177, + "grad_norm": 0.5811372726698385, + "learning_rate": 2.9918260112819952e-05, + "loss": 0.26, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03700966387987137, + "step": 497, + "valid_targets_mean": 5488.2, + "valid_targets_min": 1924 + }, + { + "epoch": 2.0162025316455696, + "grad_norm": 0.35672827420284275, + "learning_rate": 2.9869110322714778e-05, + "loss": 0.2416, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.025780418887734413, + "step": 498, + "valid_targets_mean": 4780.2, + "valid_targets_min": 1272 + }, + { + "epoch": 2.020253164556962, + "grad_norm": 0.39321861478004955, + "learning_rate": 2.9819881619466447e-05, + "loss": 0.2402, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023468071594834328, + "step": 499, + "valid_targets_mean": 4412.1, + "valid_targets_min": 1296 + }, + { + "epoch": 2.0243037974683546, + "grad_norm": 0.3669361535720418, + "learning_rate": 2.9770574396706364e-05, + "loss": 0.2558, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026013996452093124, + "step": 500, + "valid_targets_mean": 3958.3, + "valid_targets_min": 1253 + }, + { + "epoch": 2.0283544303797467, + "grad_norm": 0.42664137168830724, + "learning_rate": 2.972118904869376e-05, + "loss": 0.2475, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028630584478378296, + "step": 501, + "valid_targets_mean": 3853.2, + "valid_targets_min": 1320 + }, + { + "epoch": 2.0324050632911392, + "grad_norm": 0.3340542010102885, + "learning_rate": 2.9671725970312563e-05, + "loss": 0.2527, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028039783239364624, + "step": 502, + "valid_targets_mean": 3687.6, + "valid_targets_min": 1307 + }, + { + "epoch": 2.0364556962025318, + "grad_norm": 0.38777171194781507, + "learning_rate": 2.9622185557068238e-05, + "loss": 0.25, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02727580815553665, + "step": 503, + "valid_targets_mean": 3820.6, + "valid_targets_min": 1367 + }, + { + "epoch": 2.0405063291139243, + "grad_norm": 0.32924842344596694, + "learning_rate": 2.95725682050846e-05, + "loss": 0.2439, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02716241590678692, + "step": 504, + "valid_targets_mean": 3934.8, + "valid_targets_min": 1618 + }, + { + "epoch": 2.0445569620253163, + "grad_norm": 0.3039367974293122, + "learning_rate": 2.9522874311100684e-05, + "loss": 0.2498, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03982719033956528, + "step": 505, + "valid_targets_mean": 6318.9, + "valid_targets_min": 1567 + }, + { + "epoch": 2.048607594936709, + "grad_norm": 0.30628238080051057, + "learning_rate": 2.9473104272467526e-05, + "loss": 0.2443, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.031220629811286926, + "step": 506, + "valid_targets_mean": 4749.3, + "valid_targets_min": 1159 + }, + { + "epoch": 2.0526582278481014, + "grad_norm": 0.3423732745997802, + "learning_rate": 2.942325848714505e-05, + "loss": 0.2571, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.025803158059716225, + "step": 507, + "valid_targets_mean": 3670.8, + "valid_targets_min": 1155 + }, + { + "epoch": 2.056708860759494, + "grad_norm": 0.3181443981098617, + "learning_rate": 2.9373337353698804e-05, + "loss": 0.2512, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02923399582505226, + "step": 508, + "valid_targets_mean": 3626.4, + "valid_targets_min": 1134 + }, + { + "epoch": 2.060759493670886, + "grad_norm": 0.3188060851658609, + "learning_rate": 2.932334127129686e-05, + "loss": 0.2508, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03547723591327667, + "step": 509, + "valid_targets_mean": 4740.8, + "valid_targets_min": 1428 + }, + { + "epoch": 2.0648101265822785, + "grad_norm": 0.35017866716587204, + "learning_rate": 2.9273270639706547e-05, + "loss": 0.2522, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.042312368750572205, + "step": 510, + "valid_targets_mean": 6287.3, + "valid_targets_min": 728 + }, + { + "epoch": 2.068860759493671, + "grad_norm": 0.3314228937290726, + "learning_rate": 2.922312585929131e-05, + "loss": 0.2594, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.029436534270644188, + "step": 511, + "valid_targets_mean": 4389.8, + "valid_targets_min": 2048 + }, + { + "epoch": 2.0729113924050635, + "grad_norm": 0.36123869120637675, + "learning_rate": 2.9172907331007478e-05, + "loss": 0.2598, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03139062598347664, + "step": 512, + "valid_targets_mean": 4490.1, + "valid_targets_min": 1400 + }, + { + "epoch": 2.0769620253164556, + "grad_norm": 0.3152518397145963, + "learning_rate": 2.9122615456401053e-05, + "loss": 0.2554, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.033695369958877563, + "step": 513, + "valid_targets_mean": 4604.3, + "valid_targets_min": 939 + }, + { + "epoch": 2.081012658227848, + "grad_norm": 0.353497447966068, + "learning_rate": 2.9072250637604535e-05, + "loss": 0.2596, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.043300822377204895, + "step": 514, + "valid_targets_mean": 6654.4, + "valid_targets_min": 1476 + }, + { + "epoch": 2.0850632911392406, + "grad_norm": 0.30130567529114766, + "learning_rate": 2.9021813277333667e-05, + "loss": 0.2581, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.027196582406759262, + "step": 515, + "valid_targets_mean": 4558.4, + "valid_targets_min": 1535 + }, + { + "epoch": 2.089113924050633, + "grad_norm": 0.32168385547845024, + "learning_rate": 2.8971303778884244e-05, + "loss": 0.2486, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023042596876621246, + "step": 516, + "valid_targets_mean": 2868.8, + "valid_targets_min": 1010 + }, + { + "epoch": 2.093164556962025, + "grad_norm": 0.3578906280639518, + "learning_rate": 2.8920722546128863e-05, + "loss": 0.2616, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.046286728233098984, + "step": 517, + "valid_targets_mean": 5850.9, + "valid_targets_min": 1198 + }, + { + "epoch": 2.0972151898734177, + "grad_norm": 0.290499018540267, + "learning_rate": 2.887006998351372e-05, + "loss": 0.247, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021764980629086494, + "step": 518, + "valid_targets_mean": 3495.7, + "valid_targets_min": 1256 + }, + { + "epoch": 2.1012658227848102, + "grad_norm": 0.35911454856145447, + "learning_rate": 2.881934649605535e-05, + "loss": 0.2515, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02156168967485428, + "step": 519, + "valid_targets_mean": 4110.6, + "valid_targets_min": 1325 + }, + { + "epoch": 2.1053164556962027, + "grad_norm": 0.33183904920020607, + "learning_rate": 2.8768552489337408e-05, + "loss": 0.2643, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03599020093679428, + "step": 520, + "valid_targets_mean": 4835.8, + "valid_targets_min": 1515 + }, + { + "epoch": 2.109367088607595, + "grad_norm": 0.32230216170063813, + "learning_rate": 2.8717688369507425e-05, + "loss": 0.2555, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.029470087960362434, + "step": 521, + "valid_targets_mean": 4240.4, + "valid_targets_min": 1175 + }, + { + "epoch": 2.1134177215189873, + "grad_norm": 0.34079290172519544, + "learning_rate": 2.8666754543273535e-05, + "loss": 0.2596, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03355570137500763, + "step": 522, + "valid_targets_mean": 4991.8, + "valid_targets_min": 977 + }, + { + "epoch": 2.11746835443038, + "grad_norm": 0.3287057012336886, + "learning_rate": 2.8615751417901273e-05, + "loss": 0.2614, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04670153558254242, + "step": 523, + "valid_targets_mean": 6104.7, + "valid_targets_min": 1441 + }, + { + "epoch": 2.1215189873417724, + "grad_norm": 0.39226120428182537, + "learning_rate": 2.856467940121025e-05, + "loss": 0.2537, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.031637925654649734, + "step": 524, + "valid_targets_mean": 4412.3, + "valid_targets_min": 1529 + }, + { + "epoch": 2.1255696202531644, + "grad_norm": 0.365395614656608, + "learning_rate": 2.8513538901570964e-05, + "loss": 0.254, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03924763202667236, + "step": 525, + "valid_targets_mean": 5742.6, + "valid_targets_min": 1811 + }, + { + "epoch": 2.129620253164557, + "grad_norm": 0.3435845959967996, + "learning_rate": 2.846233032790149e-05, + "loss": 0.25, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028139624744653702, + "step": 526, + "valid_targets_mean": 3931.2, + "valid_targets_min": 1297 + }, + { + "epoch": 2.1336708860759495, + "grad_norm": 0.3397642853173752, + "learning_rate": 2.84110540896642e-05, + "loss": 0.2534, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02817264385521412, + "step": 527, + "valid_targets_mean": 4044.0, + "valid_targets_min": 1334 + }, + { + "epoch": 2.137721518987342, + "grad_norm": 0.3470751268385581, + "learning_rate": 2.8359710596862555e-05, + "loss": 0.2408, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.037513479590415955, + "step": 528, + "valid_targets_mean": 4566.8, + "valid_targets_min": 1166 + }, + { + "epoch": 2.141772151898734, + "grad_norm": 0.3078086766463351, + "learning_rate": 2.8308300260037734e-05, + "loss": 0.2419, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03615918010473251, + "step": 529, + "valid_targets_mean": 5390.9, + "valid_targets_min": 1652 + }, + { + "epoch": 2.1458227848101266, + "grad_norm": 0.32227431988680616, + "learning_rate": 2.8256823490265424e-05, + "loss": 0.245, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030684471130371094, + "step": 530, + "valid_targets_mean": 4033.6, + "valid_targets_min": 1149 + }, + { + "epoch": 2.149873417721519, + "grad_norm": 0.3200750213307613, + "learning_rate": 2.8205280699152515e-05, + "loss": 0.2508, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.019646242260932922, + "step": 531, + "valid_targets_mean": 3203.2, + "valid_targets_min": 1081 + }, + { + "epoch": 2.1539240506329116, + "grad_norm": 0.30413373085344425, + "learning_rate": 2.8153672298833777e-05, + "loss": 0.2605, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0323205292224884, + "step": 532, + "valid_targets_mean": 4365.4, + "valid_targets_min": 1320 + }, + { + "epoch": 2.1579746835443037, + "grad_norm": 0.3227746772569516, + "learning_rate": 2.8101998701968615e-05, + "loss": 0.254, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02868857979774475, + "step": 533, + "valid_targets_mean": 4597.7, + "valid_targets_min": 974 + }, + { + "epoch": 2.162025316455696, + "grad_norm": 0.33095219812272253, + "learning_rate": 2.805026032173773e-05, + "loss": 0.2559, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03351202979683876, + "step": 534, + "valid_targets_mean": 4203.1, + "valid_targets_min": 1186 + }, + { + "epoch": 2.1660759493670887, + "grad_norm": 0.31803856832646626, + "learning_rate": 2.7998457571839832e-05, + "loss": 0.2571, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028351303189992905, + "step": 535, + "valid_targets_mean": 4174.6, + "valid_targets_min": 961 + }, + { + "epoch": 2.170126582278481, + "grad_norm": 0.3098804337345109, + "learning_rate": 2.7946590866488322e-05, + "loss": 0.2502, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.025098508223891258, + "step": 536, + "valid_targets_mean": 3296.6, + "valid_targets_min": 1194 + }, + { + "epoch": 2.1741772151898733, + "grad_norm": 0.30957451812750464, + "learning_rate": 2.7894660620408e-05, + "loss": 0.2583, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023833472281694412, + "step": 537, + "valid_targets_mean": 3825.5, + "valid_targets_min": 1270 + }, + { + "epoch": 2.178227848101266, + "grad_norm": 0.32285871900224744, + "learning_rate": 2.784266724883173e-05, + "loss": 0.2546, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03730406612157822, + "step": 538, + "valid_targets_mean": 5946.6, + "valid_targets_min": 917 + }, + { + "epoch": 2.1822784810126583, + "grad_norm": 0.30315455729748625, + "learning_rate": 2.7790611167497128e-05, + "loss": 0.2551, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.027421358972787857, + "step": 539, + "valid_targets_mean": 4410.9, + "valid_targets_min": 1703 + }, + { + "epoch": 2.186329113924051, + "grad_norm": 0.3101908957957333, + "learning_rate": 2.7738492792643224e-05, + "loss": 0.2522, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03136130049824715, + "step": 540, + "valid_targets_mean": 3916.8, + "valid_targets_min": 1137 + }, + { + "epoch": 2.190379746835443, + "grad_norm": 0.33793475672526513, + "learning_rate": 2.768631254100716e-05, + "loss": 0.2554, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03737739101052284, + "step": 541, + "valid_targets_mean": 4827.4, + "valid_targets_min": 1447 + }, + { + "epoch": 2.1944303797468354, + "grad_norm": 0.3135772042063124, + "learning_rate": 2.7634070829820836e-05, + "loss": 0.2486, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.025419535115361214, + "step": 542, + "valid_targets_mean": 3488.4, + "valid_targets_min": 1217 + }, + { + "epoch": 2.198481012658228, + "grad_norm": 0.3204768128524193, + "learning_rate": 2.7581768076807592e-05, + "loss": 0.2373, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014784103259444237, + "step": 543, + "valid_targets_mean": 2494.2, + "valid_targets_min": 991 + }, + { + "epoch": 2.2025316455696204, + "grad_norm": 0.33193823700912034, + "learning_rate": 2.7529404700178832e-05, + "loss": 0.2448, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03723284602165222, + "step": 544, + "valid_targets_mean": 4625.8, + "valid_targets_min": 1372 + }, + { + "epoch": 2.2065822784810125, + "grad_norm": 0.32807630280677497, + "learning_rate": 2.7476981118630735e-05, + "loss": 0.2586, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02511894702911377, + "step": 545, + "valid_targets_mean": 3786.2, + "valid_targets_min": 1642 + }, + { + "epoch": 2.210632911392405, + "grad_norm": 0.31380484555706084, + "learning_rate": 2.742449775134085e-05, + "loss": 0.2536, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01563825085759163, + "step": 546, + "valid_targets_mean": 2965.8, + "valid_targets_min": 1310 + }, + { + "epoch": 2.2146835443037975, + "grad_norm": 0.32866936967105825, + "learning_rate": 2.7371955017964788e-05, + "loss": 0.2601, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.024734562262892723, + "step": 547, + "valid_targets_mean": 3240.2, + "valid_targets_min": 1380 + }, + { + "epoch": 2.21873417721519, + "grad_norm": 0.3521184862713378, + "learning_rate": 2.7319353338632845e-05, + "loss": 0.2459, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02412096969783306, + "step": 548, + "valid_targets_mean": 3382.1, + "valid_targets_min": 1026 + }, + { + "epoch": 2.222784810126582, + "grad_norm": 0.3221306823431023, + "learning_rate": 2.7266693133946645e-05, + "loss": 0.2419, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020261388272047043, + "step": 549, + "valid_targets_mean": 3593.0, + "valid_targets_min": 1223 + }, + { + "epoch": 2.2268354430379746, + "grad_norm": 0.3368087467122894, + "learning_rate": 2.7213974824975785e-05, + "loss": 0.2625, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04048336297273636, + "step": 550, + "valid_targets_mean": 6000.6, + "valid_targets_min": 1353 + }, + { + "epoch": 2.230886075949367, + "grad_norm": 0.2954368338935399, + "learning_rate": 2.7161198833254464e-05, + "loss": 0.255, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.024658389389514923, + "step": 551, + "valid_targets_mean": 3828.2, + "valid_targets_min": 1487 + }, + { + "epoch": 2.2349367088607597, + "grad_norm": 0.3349581200749901, + "learning_rate": 2.7108365580778095e-05, + "loss": 0.2616, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.044930506497621536, + "step": 552, + "valid_targets_mean": 5943.4, + "valid_targets_min": 1021 + }, + { + "epoch": 2.2389873417721518, + "grad_norm": 0.32044780080787033, + "learning_rate": 2.705547548999996e-05, + "loss": 0.2682, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0389753133058548, + "step": 553, + "valid_targets_mean": 4595.6, + "valid_targets_min": 1392 + }, + { + "epoch": 2.2430379746835443, + "grad_norm": 0.31328225966606027, + "learning_rate": 2.7002528983827817e-05, + "loss": 0.2525, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0390719473361969, + "step": 554, + "valid_targets_mean": 5455.2, + "valid_targets_min": 1115 + }, + { + "epoch": 2.247088607594937, + "grad_norm": 0.3350003475019475, + "learning_rate": 2.6949526485620517e-05, + "loss": 0.2556, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0258515402674675, + "step": 555, + "valid_targets_mean": 3440.7, + "valid_targets_min": 1175 + }, + { + "epoch": 2.2511392405063293, + "grad_norm": 0.29418896974560066, + "learning_rate": 2.6896468419184623e-05, + "loss": 0.2494, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04345124214887619, + "step": 556, + "valid_targets_mean": 6255.0, + "valid_targets_min": 1503 + }, + { + "epoch": 2.2551898734177214, + "grad_norm": 0.31284133074228326, + "learning_rate": 2.6843355208771024e-05, + "loss": 0.2651, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0307228434830904, + "step": 557, + "valid_targets_mean": 4361.1, + "valid_targets_min": 1051 + }, + { + "epoch": 2.259240506329114, + "grad_norm": 0.29849006850574383, + "learning_rate": 2.6790187279071536e-05, + "loss": 0.2478, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03746002912521362, + "step": 558, + "valid_targets_mean": 5214.0, + "valid_targets_min": 1284 + }, + { + "epoch": 2.2632911392405064, + "grad_norm": 0.3481923927123726, + "learning_rate": 2.67369650552155e-05, + "loss": 0.2453, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026629457250237465, + "step": 559, + "valid_targets_mean": 3566.5, + "valid_targets_min": 1076 + }, + { + "epoch": 2.267341772151899, + "grad_norm": 0.3109132395106645, + "learning_rate": 2.6683688962766407e-05, + "loss": 0.2471, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03283403813838959, + "step": 560, + "valid_targets_mean": 4456.2, + "valid_targets_min": 1371 + }, + { + "epoch": 2.271392405063291, + "grad_norm": 0.3371083146820319, + "learning_rate": 2.6630359427718468e-05, + "loss": 0.2573, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.033645160496234894, + "step": 561, + "valid_targets_mean": 4629.2, + "valid_targets_min": 1251 + }, + { + "epoch": 2.2754430379746835, + "grad_norm": 0.3004647288482063, + "learning_rate": 2.6576976876493225e-05, + "loss": 0.2519, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03325767070055008, + "step": 562, + "valid_targets_mean": 4771.6, + "valid_targets_min": 1227 + }, + { + "epoch": 2.279493670886076, + "grad_norm": 0.3084492314730116, + "learning_rate": 2.652354173593614e-05, + "loss": 0.2545, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03239452838897705, + "step": 563, + "valid_targets_mean": 4820.6, + "valid_targets_min": 1212 + }, + { + "epoch": 2.2835443037974685, + "grad_norm": 0.29720381711643024, + "learning_rate": 2.647005443331317e-05, + "loss": 0.2509, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04225707799196243, + "step": 564, + "valid_targets_mean": 6034.4, + "valid_targets_min": 1778 + }, + { + "epoch": 2.2875949367088606, + "grad_norm": 0.32464412279319016, + "learning_rate": 2.6416515396307354e-05, + "loss": 0.2561, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03437012434005737, + "step": 565, + "valid_targets_mean": 5410.6, + "valid_targets_min": 1184 + }, + { + "epoch": 2.291645569620253, + "grad_norm": 0.28582335898985517, + "learning_rate": 2.636292505301542e-05, + "loss": 0.2428, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04065122827887535, + "step": 566, + "valid_targets_mean": 5343.1, + "valid_targets_min": 967 + }, + { + "epoch": 2.2956962025316456, + "grad_norm": 0.3090492224591957, + "learning_rate": 2.630928383194431e-05, + "loss": 0.2616, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0343312993645668, + "step": 567, + "valid_targets_mean": 4817.2, + "valid_targets_min": 1044 + }, + { + "epoch": 2.299746835443038, + "grad_norm": 0.30265240010177236, + "learning_rate": 2.625559216200781e-05, + "loss": 0.2431, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04373444989323616, + "step": 568, + "valid_targets_mean": 4855.6, + "valid_targets_min": 1664 + }, + { + "epoch": 2.3037974683544302, + "grad_norm": 0.33054956711731676, + "learning_rate": 2.620185047252308e-05, + "loss": 0.2492, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028334802016615868, + "step": 569, + "valid_targets_mean": 3796.7, + "valid_targets_min": 1634 + }, + { + "epoch": 2.3078481012658227, + "grad_norm": 0.2893427649819799, + "learning_rate": 2.614805919320723e-05, + "loss": 0.2513, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03859302029013634, + "step": 570, + "valid_targets_mean": 5215.1, + "valid_targets_min": 1381 + }, + { + "epoch": 2.3118987341772153, + "grad_norm": 0.3194258038200415, + "learning_rate": 2.6094218754173895e-05, + "loss": 0.2371, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030152734369039536, + "step": 571, + "valid_targets_mean": 4297.3, + "valid_targets_min": 1471 + }, + { + "epoch": 2.3159493670886078, + "grad_norm": 0.280776302266346, + "learning_rate": 2.60403295859298e-05, + "loss": 0.2615, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026337942108511925, + "step": 572, + "valid_targets_mean": 3656.6, + "valid_targets_min": 1274 + }, + { + "epoch": 2.32, + "grad_norm": 0.33440416375508986, + "learning_rate": 2.5986392119371297e-05, + "loss": 0.2513, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02118188515305519, + "step": 573, + "valid_targets_mean": 3200.6, + "valid_targets_min": 941 + }, + { + "epoch": 2.3240506329113924, + "grad_norm": 0.31817968317557005, + "learning_rate": 2.5932406785780928e-05, + "loss": 0.262, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03111189603805542, + "step": 574, + "valid_targets_mean": 4663.5, + "valid_targets_min": 1166 + }, + { + "epoch": 2.328101265822785, + "grad_norm": 0.3240466981779466, + "learning_rate": 2.5878374016823987e-05, + "loss": 0.2563, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03127945959568024, + "step": 575, + "valid_targets_mean": 3908.9, + "valid_targets_min": 1131 + }, + { + "epoch": 2.3321518987341774, + "grad_norm": 0.2814225317214882, + "learning_rate": 2.5824294244545045e-05, + "loss": 0.2551, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03601471334695816, + "step": 576, + "valid_targets_mean": 5604.6, + "valid_targets_min": 1460 + }, + { + "epoch": 2.3362025316455695, + "grad_norm": 0.2925319005446987, + "learning_rate": 2.577016790136453e-05, + "loss": 0.2413, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02899722382426262, + "step": 577, + "valid_targets_mean": 4115.4, + "valid_targets_min": 1293 + }, + { + "epoch": 2.340253164556962, + "grad_norm": 0.299581537856646, + "learning_rate": 2.5715995420075225e-05, + "loss": 0.2561, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03020840510725975, + "step": 578, + "valid_targets_mean": 3873.1, + "valid_targets_min": 981 + }, + { + "epoch": 2.3443037974683545, + "grad_norm": 0.29720958694672484, + "learning_rate": 2.5661777233838864e-05, + "loss": 0.2575, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03779207915067673, + "step": 579, + "valid_targets_mean": 6122.3, + "valid_targets_min": 1270 + }, + { + "epoch": 2.348354430379747, + "grad_norm": 0.3339333851319194, + "learning_rate": 2.5607513776182616e-05, + "loss": 0.263, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.033767811954021454, + "step": 580, + "valid_targets_mean": 3967.4, + "valid_targets_min": 1203 + }, + { + "epoch": 2.352405063291139, + "grad_norm": 0.29003044027962066, + "learning_rate": 2.5553205480995633e-05, + "loss": 0.2498, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0322946161031723, + "step": 581, + "valid_targets_mean": 4815.2, + "valid_targets_min": 1555 + }, + { + "epoch": 2.3564556962025316, + "grad_norm": 0.3183024209535512, + "learning_rate": 2.5498852782525604e-05, + "loss": 0.2516, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030187206342816353, + "step": 582, + "valid_targets_mean": 4114.4, + "valid_targets_min": 1224 + }, + { + "epoch": 2.360506329113924, + "grad_norm": 0.314168755966548, + "learning_rate": 2.5444456115375252e-05, + "loss": 0.2633, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028827618807554245, + "step": 583, + "valid_targets_mean": 4781.1, + "valid_targets_min": 1182 + }, + { + "epoch": 2.3645569620253166, + "grad_norm": 0.3343463864058334, + "learning_rate": 2.539001591449889e-05, + "loss": 0.2627, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04334329441189766, + "step": 584, + "valid_targets_mean": 5263.4, + "valid_targets_min": 1096 + }, + { + "epoch": 2.3686075949367087, + "grad_norm": 0.3110837175415601, + "learning_rate": 2.53355326151989e-05, + "loss": 0.2553, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021545108407735825, + "step": 585, + "valid_targets_mean": 3364.5, + "valid_targets_min": 1045 + }, + { + "epoch": 2.372658227848101, + "grad_norm": 0.32372691675059556, + "learning_rate": 2.5281006653122288e-05, + "loss": 0.2764, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.05526130646467209, + "step": 586, + "valid_targets_mean": 6967.8, + "valid_targets_min": 1101 + }, + { + "epoch": 2.3767088607594937, + "grad_norm": 0.3012800068851953, + "learning_rate": 2.5226438464257217e-05, + "loss": 0.2638, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030668795108795166, + "step": 587, + "valid_targets_mean": 4461.7, + "valid_targets_min": 946 + }, + { + "epoch": 2.3807594936708862, + "grad_norm": 0.3124888285643379, + "learning_rate": 2.5171828484929447e-05, + "loss": 0.2603, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03976716846227646, + "step": 588, + "valid_targets_mean": 4932.5, + "valid_targets_min": 1154 + }, + { + "epoch": 2.3848101265822783, + "grad_norm": 0.2691816871691649, + "learning_rate": 2.5117177151798934e-05, + "loss": 0.2558, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03441830724477768, + "step": 589, + "valid_targets_mean": 5091.1, + "valid_targets_min": 1419 + }, + { + "epoch": 2.388860759493671, + "grad_norm": 0.3215830136868613, + "learning_rate": 2.506248490185627e-05, + "loss": 0.2631, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030570250004529953, + "step": 590, + "valid_targets_mean": 3887.0, + "valid_targets_min": 1148 + }, + { + "epoch": 2.3929113924050633, + "grad_norm": 0.31485583061901584, + "learning_rate": 2.500775217241923e-05, + "loss": 0.261, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.038104347884655, + "step": 591, + "valid_targets_mean": 6018.2, + "valid_targets_min": 1546 + }, + { + "epoch": 2.396962025316456, + "grad_norm": 0.3080284475460948, + "learning_rate": 2.4952979401129266e-05, + "loss": 0.2537, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02784135192632675, + "step": 592, + "valid_targets_mean": 3782.8, + "valid_targets_min": 861 + }, + { + "epoch": 2.401012658227848, + "grad_norm": 0.304473757906764, + "learning_rate": 2.4898167025948e-05, + "loss": 0.257, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.025928225368261337, + "step": 593, + "valid_targets_mean": 3888.6, + "valid_targets_min": 1261 + }, + { + "epoch": 2.4050632911392404, + "grad_norm": 0.3231693169855003, + "learning_rate": 2.4843315485153703e-05, + "loss": 0.2577, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.029293615370988846, + "step": 594, + "valid_targets_mean": 4102.2, + "valid_targets_min": 1643 + }, + { + "epoch": 2.409113924050633, + "grad_norm": 0.30696065623367774, + "learning_rate": 2.4788425217337856e-05, + "loss": 0.2618, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03263610601425171, + "step": 595, + "valid_targets_mean": 3837.7, + "valid_targets_min": 1478 + }, + { + "epoch": 2.4131645569620255, + "grad_norm": 0.2760337527277468, + "learning_rate": 2.4733496661401572e-05, + "loss": 0.2501, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04267909377813339, + "step": 596, + "valid_targets_mean": 6636.6, + "valid_targets_min": 1240 + }, + { + "epoch": 2.4172151898734175, + "grad_norm": 0.30165615231190473, + "learning_rate": 2.4678530256552114e-05, + "loss": 0.2604, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021362274885177612, + "step": 597, + "valid_targets_mean": 3170.9, + "valid_targets_min": 737 + }, + { + "epoch": 2.42126582278481, + "grad_norm": 0.3280741446956329, + "learning_rate": 2.46235264422994e-05, + "loss": 0.265, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03641419857740402, + "step": 598, + "valid_targets_mean": 5095.3, + "valid_targets_min": 1164 + }, + { + "epoch": 2.4253164556962026, + "grad_norm": 0.28204803788757965, + "learning_rate": 2.4568485658452452e-05, + "loss": 0.2584, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03064592182636261, + "step": 599, + "valid_targets_mean": 3911.2, + "valid_targets_min": 945 + }, + { + "epoch": 2.429367088607595, + "grad_norm": 0.29615169855204987, + "learning_rate": 2.451340834511592e-05, + "loss": 0.25, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.031110865995287895, + "step": 600, + "valid_targets_mean": 4780.8, + "valid_targets_min": 1196 + }, + { + "epoch": 2.433417721518987, + "grad_norm": 0.3188121688952584, + "learning_rate": 2.4458294942686528e-05, + "loss": 0.2637, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04329856485128403, + "step": 601, + "valid_targets_mean": 5275.9, + "valid_targets_min": 985 + }, + { + "epoch": 2.4374683544303797, + "grad_norm": 0.27951189958912137, + "learning_rate": 2.440314589184958e-05, + "loss": 0.2478, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.024103689938783646, + "step": 602, + "valid_targets_mean": 4180.4, + "valid_targets_min": 1375 + }, + { + "epoch": 2.441518987341772, + "grad_norm": 0.3029822254490272, + "learning_rate": 2.434796163357541e-05, + "loss": 0.2592, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03358926996588707, + "step": 603, + "valid_targets_mean": 4572.4, + "valid_targets_min": 1281 + }, + { + "epoch": 2.4455696202531647, + "grad_norm": 0.3031290979928384, + "learning_rate": 2.429274260911588e-05, + "loss": 0.2454, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026303090155124664, + "step": 604, + "valid_targets_mean": 3897.2, + "valid_targets_min": 1052 + }, + { + "epoch": 2.449620253164557, + "grad_norm": 0.28066556297235906, + "learning_rate": 2.423748926000084e-05, + "loss": 0.2531, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02926298975944519, + "step": 605, + "valid_targets_mean": 3421.1, + "valid_targets_min": 885 + }, + { + "epoch": 2.4536708860759493, + "grad_norm": 0.3193581360060043, + "learning_rate": 2.418220202803459e-05, + "loss": 0.2507, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02278808504343033, + "step": 606, + "valid_targets_mean": 3550.6, + "valid_targets_min": 851 + }, + { + "epoch": 2.457721518987342, + "grad_norm": 0.3047526662043712, + "learning_rate": 2.412688135529237e-05, + "loss": 0.2587, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.031472377479076385, + "step": 607, + "valid_targets_mean": 4140.6, + "valid_targets_min": 1186 + }, + { + "epoch": 2.4617721518987343, + "grad_norm": 0.3030005726177253, + "learning_rate": 2.4071527684116807e-05, + "loss": 0.2555, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03665117174386978, + "step": 608, + "valid_targets_mean": 4628.6, + "valid_targets_min": 1018 + }, + { + "epoch": 2.4658227848101264, + "grad_norm": 0.2852849326847587, + "learning_rate": 2.4016141457114372e-05, + "loss": 0.263, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030085954815149307, + "step": 609, + "valid_targets_mean": 4607.9, + "valid_targets_min": 838 + }, + { + "epoch": 2.469873417721519, + "grad_norm": 0.29629453300425185, + "learning_rate": 2.396072311715187e-05, + "loss": 0.2536, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03866127133369446, + "step": 610, + "valid_targets_mean": 4362.0, + "valid_targets_min": 1135 + }, + { + "epoch": 2.4739240506329114, + "grad_norm": 0.3143382515410931, + "learning_rate": 2.3905273107352863e-05, + "loss": 0.2616, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03561249375343323, + "step": 611, + "valid_targets_mean": 4431.0, + "valid_targets_min": 851 + }, + { + "epoch": 2.477974683544304, + "grad_norm": 0.31019023346681507, + "learning_rate": 2.3849791871094157e-05, + "loss": 0.2507, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03153450787067413, + "step": 612, + "valid_targets_mean": 4573.3, + "valid_targets_min": 1572 + }, + { + "epoch": 2.482025316455696, + "grad_norm": 0.32844579891221176, + "learning_rate": 2.379427985200223e-05, + "loss": 0.2675, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03163554146885872, + "step": 613, + "valid_targets_mean": 4459.7, + "valid_targets_min": 1269 + }, + { + "epoch": 2.4860759493670885, + "grad_norm": 0.3329061332621241, + "learning_rate": 2.373873749394972e-05, + "loss": 0.257, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02794114127755165, + "step": 614, + "valid_targets_mean": 4042.6, + "valid_targets_min": 920 + }, + { + "epoch": 2.490126582278481, + "grad_norm": 0.2943763735771222, + "learning_rate": 2.3683165241051836e-05, + "loss": 0.2532, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.031979337334632874, + "step": 615, + "valid_targets_mean": 4640.8, + "valid_targets_min": 1653 + }, + { + "epoch": 2.4941772151898736, + "grad_norm": 0.2833662624998012, + "learning_rate": 2.3627563537662823e-05, + "loss": 0.2539, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02792017161846161, + "step": 616, + "valid_targets_mean": 4450.8, + "valid_targets_min": 1192 + }, + { + "epoch": 2.4982278481012656, + "grad_norm": 0.29721603364616905, + "learning_rate": 2.3571932828372433e-05, + "loss": 0.2653, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028793741017580032, + "step": 617, + "valid_targets_mean": 4491.8, + "valid_targets_min": 1621 + }, + { + "epoch": 2.502278481012658, + "grad_norm": 0.33444596381029734, + "learning_rate": 2.351627355800233e-05, + "loss": 0.2655, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01843031868338585, + "step": 618, + "valid_targets_mean": 2783.9, + "valid_targets_min": 1379 + }, + { + "epoch": 2.5063291139240507, + "grad_norm": 0.27517479471822953, + "learning_rate": 2.346058617160256e-05, + "loss": 0.2591, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026295827701687813, + "step": 619, + "valid_targets_mean": 3806.1, + "valid_targets_min": 1204 + }, + { + "epoch": 2.510379746835443, + "grad_norm": 0.2916338910826666, + "learning_rate": 2.3404871114447976e-05, + "loss": 0.2548, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020635364577174187, + "step": 620, + "valid_targets_mean": 3079.8, + "valid_targets_min": 1064 + }, + { + "epoch": 2.5144303797468357, + "grad_norm": 0.30894555244757066, + "learning_rate": 2.334912883203469e-05, + "loss": 0.2701, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04036060720682144, + "step": 621, + "valid_targets_mean": 5933.9, + "valid_targets_min": 1190 + }, + { + "epoch": 2.5184810126582278, + "grad_norm": 0.30113451870141916, + "learning_rate": 2.3293359770076505e-05, + "loss": 0.2661, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023307915776968002, + "step": 622, + "valid_targets_mean": 3366.1, + "valid_targets_min": 1684 + }, + { + "epoch": 2.5225316455696203, + "grad_norm": 0.32708924994767863, + "learning_rate": 2.323756437450136e-05, + "loss": 0.2605, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03131123632192612, + "step": 623, + "valid_targets_mean": 4423.8, + "valid_targets_min": 1800 + }, + { + "epoch": 2.526582278481013, + "grad_norm": 0.3213327010704033, + "learning_rate": 2.3181743091447754e-05, + "loss": 0.2557, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028300907462835312, + "step": 624, + "valid_targets_mean": 4199.7, + "valid_targets_min": 1425 + }, + { + "epoch": 2.530632911392405, + "grad_norm": 0.29768752466827214, + "learning_rate": 2.3125896367261164e-05, + "loss": 0.2612, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028280166909098625, + "step": 625, + "valid_targets_mean": 4495.2, + "valid_targets_min": 1336 + }, + { + "epoch": 2.5346835443037974, + "grad_norm": 0.29324733188037844, + "learning_rate": 2.307002464849052e-05, + "loss": 0.2582, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03605882078409195, + "step": 626, + "valid_targets_mean": 5320.2, + "valid_targets_min": 1240 + }, + { + "epoch": 2.53873417721519, + "grad_norm": 0.3005243259009893, + "learning_rate": 2.30141283818846e-05, + "loss": 0.2539, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026947829872369766, + "step": 627, + "valid_targets_mean": 3318.6, + "valid_targets_min": 1248 + }, + { + "epoch": 2.5427848101265824, + "grad_norm": 0.27350582442900084, + "learning_rate": 2.2958208014388462e-05, + "loss": 0.2514, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03252295404672623, + "step": 628, + "valid_targets_mean": 5166.3, + "valid_targets_min": 1345 + }, + { + "epoch": 2.546835443037975, + "grad_norm": 0.28854390431284904, + "learning_rate": 2.2902263993139857e-05, + "loss": 0.2462, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03174077346920967, + "step": 629, + "valid_targets_mean": 4279.8, + "valid_targets_min": 983 + }, + { + "epoch": 2.550886075949367, + "grad_norm": 0.28372611723663116, + "learning_rate": 2.2846296765465708e-05, + "loss": 0.2467, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0290265791118145, + "step": 630, + "valid_targets_mean": 4876.6, + "valid_targets_min": 1103 + }, + { + "epoch": 2.5549367088607595, + "grad_norm": 0.27799669957215767, + "learning_rate": 2.2790306778878465e-05, + "loss": 0.2577, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03007456660270691, + "step": 631, + "valid_targets_mean": 4479.5, + "valid_targets_min": 1483 + }, + { + "epoch": 2.558987341772152, + "grad_norm": 0.282275352024946, + "learning_rate": 2.273429448107257e-05, + "loss": 0.2535, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0425049290060997, + "step": 632, + "valid_targets_mean": 5926.8, + "valid_targets_min": 1197 + }, + { + "epoch": 2.563037974683544, + "grad_norm": 0.25916687600374244, + "learning_rate": 2.2678260319920868e-05, + "loss": 0.2573, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.033781878650188446, + "step": 633, + "valid_targets_mean": 4686.0, + "valid_targets_min": 1132 + }, + { + "epoch": 2.5670886075949366, + "grad_norm": 0.2705411998780125, + "learning_rate": 2.262220474347102e-05, + "loss": 0.2544, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02024194225668907, + "step": 634, + "valid_targets_mean": 3486.6, + "valid_targets_min": 1290 + }, + { + "epoch": 2.571139240506329, + "grad_norm": 0.2697006089406536, + "learning_rate": 2.2566128199941918e-05, + "loss": 0.2553, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03225478157401085, + "step": 635, + "valid_targets_mean": 5035.3, + "valid_targets_min": 1411 + }, + { + "epoch": 2.5751898734177217, + "grad_norm": 0.2686734454523371, + "learning_rate": 2.2510031137720104e-05, + "loss": 0.2625, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026321150362491608, + "step": 636, + "valid_targets_mean": 3903.1, + "valid_targets_min": 1029 + }, + { + "epoch": 2.579240506329114, + "grad_norm": 0.31013089351460327, + "learning_rate": 2.2453914005356206e-05, + "loss": 0.2546, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.037434741854667664, + "step": 637, + "valid_targets_mean": 5300.7, + "valid_targets_min": 1202 + }, + { + "epoch": 2.5832911392405062, + "grad_norm": 0.2786446450188604, + "learning_rate": 2.2397777251561302e-05, + "loss": 0.2544, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.031491074711084366, + "step": 638, + "valid_targets_mean": 4333.8, + "valid_targets_min": 1201 + }, + { + "epoch": 2.5873417721518988, + "grad_norm": 0.29864961232770876, + "learning_rate": 2.2341621325203393e-05, + "loss": 0.2466, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.027186639606952667, + "step": 639, + "valid_targets_mean": 3333.9, + "valid_targets_min": 1120 + }, + { + "epoch": 2.5913924050632913, + "grad_norm": 0.3215698124780734, + "learning_rate": 2.2285446675303764e-05, + "loss": 0.2637, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03543717414140701, + "step": 640, + "valid_targets_mean": 4715.6, + "valid_targets_min": 1411 + }, + { + "epoch": 2.5954430379746833, + "grad_norm": 0.3050945512291773, + "learning_rate": 2.2229253751033423e-05, + "loss": 0.2601, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.024836286902427673, + "step": 641, + "valid_targets_mean": 4104.9, + "valid_targets_min": 1240 + }, + { + "epoch": 2.599493670886076, + "grad_norm": 0.3050775033486984, + "learning_rate": 2.2173043001709492e-05, + "loss": 0.262, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030903708189725876, + "step": 642, + "valid_targets_mean": 4335.6, + "valid_targets_min": 1410 + }, + { + "epoch": 2.6035443037974684, + "grad_norm": 0.2997633406509843, + "learning_rate": 2.2116814876791627e-05, + "loss": 0.264, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.032921142876148224, + "step": 643, + "valid_targets_mean": 4964.4, + "valid_targets_min": 1126 + }, + { + "epoch": 2.607594936708861, + "grad_norm": 0.3087476373401668, + "learning_rate": 2.2060569825878413e-05, + "loss": 0.262, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03322776407003403, + "step": 644, + "valid_targets_mean": 4452.0, + "valid_targets_min": 920 + }, + { + "epoch": 2.6116455696202534, + "grad_norm": 0.2828106261268277, + "learning_rate": 2.2004308298703777e-05, + "loss": 0.2516, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.031175900250673294, + "step": 645, + "valid_targets_mean": 4302.5, + "valid_targets_min": 780 + }, + { + "epoch": 2.6156962025316455, + "grad_norm": 0.3008393389781136, + "learning_rate": 2.1948030745133396e-05, + "loss": 0.2586, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.037226781249046326, + "step": 646, + "valid_targets_mean": 4957.2, + "valid_targets_min": 1283 + }, + { + "epoch": 2.619746835443038, + "grad_norm": 0.29503272638846983, + "learning_rate": 2.1891737615161083e-05, + "loss": 0.2514, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02325281873345375, + "step": 647, + "valid_targets_mean": 3224.1, + "valid_targets_min": 1214 + }, + { + "epoch": 2.6237974683544305, + "grad_norm": 0.29677005409578583, + "learning_rate": 2.1835429358905218e-05, + "loss": 0.2541, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030813105404376984, + "step": 648, + "valid_targets_mean": 4649.4, + "valid_targets_min": 1340 + }, + { + "epoch": 2.6278481012658226, + "grad_norm": 0.2693461924966625, + "learning_rate": 2.17791064266051e-05, + "loss": 0.2629, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03964713588356972, + "step": 649, + "valid_targets_mean": 5541.3, + "valid_targets_min": 1081 + }, + { + "epoch": 2.631898734177215, + "grad_norm": 0.302821999547357, + "learning_rate": 2.172276926861741e-05, + "loss": 0.2504, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028620591387152672, + "step": 650, + "valid_targets_mean": 4284.2, + "valid_targets_min": 814 + }, + { + "epoch": 2.6359493670886076, + "grad_norm": 0.29782965749403933, + "learning_rate": 2.166641833541255e-05, + "loss": 0.2579, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03146466612815857, + "step": 651, + "valid_targets_mean": 4254.1, + "valid_targets_min": 1112 + }, + { + "epoch": 2.64, + "grad_norm": 0.2842374042867517, + "learning_rate": 2.161005407757109e-05, + "loss": 0.2448, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02665664814412594, + "step": 652, + "valid_targets_mean": 3843.0, + "valid_targets_min": 1300 + }, + { + "epoch": 2.6440506329113926, + "grad_norm": 0.318201022567993, + "learning_rate": 2.155367694578013e-05, + "loss": 0.2676, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030992835760116577, + "step": 653, + "valid_targets_mean": 4395.9, + "valid_targets_min": 1175 + }, + { + "epoch": 2.6481012658227847, + "grad_norm": 0.3003631747307125, + "learning_rate": 2.1497287390829712e-05, + "loss": 0.2576, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03913872689008713, + "step": 654, + "valid_targets_mean": 5126.5, + "valid_targets_min": 1052 + }, + { + "epoch": 2.6521518987341772, + "grad_norm": 0.2867285062665385, + "learning_rate": 2.1440885863609208e-05, + "loss": 0.2583, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01915227808058262, + "step": 655, + "valid_targets_mean": 3469.7, + "valid_targets_min": 783 + }, + { + "epoch": 2.6562025316455697, + "grad_norm": 0.30572330910607615, + "learning_rate": 2.1384472815103737e-05, + "loss": 0.2556, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03280739486217499, + "step": 656, + "valid_targets_mean": 4244.4, + "valid_targets_min": 1440 + }, + { + "epoch": 2.660253164556962, + "grad_norm": 0.29116636191372325, + "learning_rate": 2.1328048696390534e-05, + "loss": 0.2554, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02809366211295128, + "step": 657, + "valid_targets_mean": 4139.7, + "valid_targets_min": 1281 + }, + { + "epoch": 2.6643037974683543, + "grad_norm": 0.29492182773411035, + "learning_rate": 2.1271613958635342e-05, + "loss": 0.2415, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0448947437107563, + "step": 658, + "valid_targets_mean": 6116.7, + "valid_targets_min": 2123 + }, + { + "epoch": 2.668354430379747, + "grad_norm": 0.2897066616049584, + "learning_rate": 2.1215169053088825e-05, + "loss": 0.2455, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03614652901887894, + "step": 659, + "valid_targets_mean": 5847.3, + "valid_targets_min": 1304 + }, + { + "epoch": 2.6724050632911394, + "grad_norm": 0.328253562044885, + "learning_rate": 2.1158714431082937e-05, + "loss": 0.2698, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026467427611351013, + "step": 660, + "valid_targets_mean": 4154.1, + "valid_targets_min": 1534 + }, + { + "epoch": 2.676455696202532, + "grad_norm": 0.29549581880007425, + "learning_rate": 2.110225054402733e-05, + "loss": 0.2611, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04412643983960152, + "step": 661, + "valid_targets_mean": 5657.8, + "valid_targets_min": 1309 + }, + { + "epoch": 2.680506329113924, + "grad_norm": 0.31282829359365777, + "learning_rate": 2.1045777843405747e-05, + "loss": 0.2531, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02726035937666893, + "step": 662, + "valid_targets_mean": 3428.0, + "valid_targets_min": 900 + }, + { + "epoch": 2.6845569620253165, + "grad_norm": 0.3009950377222184, + "learning_rate": 2.09892967807724e-05, + "loss": 0.2579, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03681063652038574, + "step": 663, + "valid_targets_mean": 4952.4, + "valid_targets_min": 778 + }, + { + "epoch": 2.688607594936709, + "grad_norm": 0.30137309251881117, + "learning_rate": 2.093280780774835e-05, + "loss": 0.2615, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03740883618593216, + "step": 664, + "valid_targets_mean": 5394.9, + "valid_targets_min": 1522 + }, + { + "epoch": 2.692658227848101, + "grad_norm": 0.28601586593852013, + "learning_rate": 2.087631137601793e-05, + "loss": 0.2502, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03194505721330643, + "step": 665, + "valid_targets_mean": 4750.3, + "valid_targets_min": 1325 + }, + { + "epoch": 2.6967088607594936, + "grad_norm": 0.32148793052456576, + "learning_rate": 2.0819807937325104e-05, + "loss": 0.2639, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04088940471410751, + "step": 666, + "valid_targets_mean": 5782.1, + "valid_targets_min": 1447 + }, + { + "epoch": 2.700759493670886, + "grad_norm": 0.29417424569091954, + "learning_rate": 2.0763297943469847e-05, + "loss": 0.2619, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017188072204589844, + "step": 667, + "valid_targets_mean": 3231.1, + "valid_targets_min": 1094 + }, + { + "epoch": 2.7048101265822786, + "grad_norm": 0.31965059393490125, + "learning_rate": 2.070678184630458e-05, + "loss": 0.2403, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.033367276191711426, + "step": 668, + "valid_targets_mean": 4940.9, + "valid_targets_min": 1223 }, { - "epoch": 1.0086075949367088, - "grad_norm": 0.40471323867861103, - "learning_rate": 3.873493415863256e-05, - "loss": 0.3177, - "step": 249 + "epoch": 2.708860759493671, + "grad_norm": 0.3027836295255975, + "learning_rate": 2.06502600977305e-05, + "loss": 0.246, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03082066960632801, + "step": 669, + "valid_targets_mean": 4925.0, + "valid_targets_min": 1102 + }, + { + "epoch": 2.712911392405063, + "grad_norm": 0.32643311209170656, + "learning_rate": 2.0593733149694008e-05, + "loss": 0.2603, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.027547260746359825, + "step": 670, + "valid_targets_mean": 4204.4, + "valid_targets_min": 1152 + }, + { + "epoch": 2.7169620253164557, + "grad_norm": 0.3072630507340393, + "learning_rate": 2.0537201454183074e-05, + "loss": 0.2572, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02958519198000431, + "step": 671, + "valid_targets_mean": 4499.6, + "valid_targets_min": 975 + }, + { + "epoch": 2.721012658227848, + "grad_norm": 0.2872594254641355, + "learning_rate": 2.0480665463223626e-05, + "loss": 0.2532, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03325200825929642, + "step": 672, + "valid_targets_mean": 5680.6, + "valid_targets_min": 1443 + }, + { + "epoch": 2.7250632911392403, + "grad_norm": 0.34577995147391477, + "learning_rate": 2.042412562887594e-05, + "loss": 0.2631, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02334584668278694, + "step": 673, + "valid_targets_mean": 3676.2, + "valid_targets_min": 1157 + }, + { + "epoch": 2.729113924050633, + "grad_norm": 0.28381380010863383, + "learning_rate": 2.0367582403231033e-05, + "loss": 0.2656, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03436291217803955, + "step": 674, + "valid_targets_mean": 5062.5, + "valid_targets_min": 1572 + }, + { + "epoch": 2.7331645569620253, + "grad_norm": 0.3107063992450581, + "learning_rate": 2.0311036238407023e-05, + "loss": 0.2571, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030485117807984352, + "step": 675, + "valid_targets_mean": 4593.1, + "valid_targets_min": 1677 + }, + { + "epoch": 2.737215189873418, + "grad_norm": 0.30380115780354106, + "learning_rate": 2.0254487586545534e-05, + "loss": 0.2629, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03379993885755539, + "step": 676, + "valid_targets_mean": 4763.4, + "valid_targets_min": 1064 + }, + { + "epoch": 2.7412658227848103, + "grad_norm": 0.3116411566769037, + "learning_rate": 2.0197936899808088e-05, + "loss": 0.2451, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0327700711786747, + "step": 677, + "valid_targets_mean": 5130.9, + "valid_targets_min": 1643 + }, + { + "epoch": 2.7453164556962024, + "grad_norm": 0.2834334524400567, + "learning_rate": 2.0141384630372466e-05, + "loss": 0.2582, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.022598637267947197, + "step": 678, + "valid_targets_mean": 3332.0, + "valid_targets_min": 1457 + }, + { + "epoch": 2.749367088607595, + "grad_norm": 0.2954144930897787, + "learning_rate": 2.0084831230429106e-05, + "loss": 0.2484, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02883639745414257, + "step": 679, + "valid_targets_mean": 4062.5, + "valid_targets_min": 842 + }, + { + "epoch": 2.7534177215189874, + "grad_norm": 0.28743408729324493, + "learning_rate": 2.0028277152177492e-05, + "loss": 0.2482, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.032943643629550934, + "step": 680, + "valid_targets_mean": 5628.5, + "valid_targets_min": 897 + }, + { + "epoch": 2.7574683544303795, + "grad_norm": 0.2661613276585499, + "learning_rate": 1.9971722847822518e-05, + "loss": 0.243, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.042223528027534485, + "step": 681, + "valid_targets_mean": 5950.6, + "valid_targets_min": 1513 + }, + { + "epoch": 2.761518987341772, + "grad_norm": 0.32422655919166515, + "learning_rate": 1.99151687695709e-05, + "loss": 0.2567, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.038189806044101715, + "step": 682, + "valid_targets_mean": 4266.8, + "valid_targets_min": 957 + }, + { + "epoch": 2.7655696202531646, + "grad_norm": 0.28388437468562894, + "learning_rate": 1.9858615369627537e-05, + "loss": 0.2598, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.029368355870246887, + "step": 683, + "valid_targets_mean": 4718.3, + "valid_targets_min": 915 + }, + { + "epoch": 2.769620253164557, + "grad_norm": 0.27844480657474535, + "learning_rate": 1.9802063100191916e-05, + "loss": 0.2679, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030729448422789574, + "step": 684, + "valid_targets_mean": 4384.0, + "valid_targets_min": 1178 + }, + { + "epoch": 2.7736708860759496, + "grad_norm": 0.30984977356577154, + "learning_rate": 1.974551241345447e-05, + "loss": 0.2539, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02934480458498001, + "step": 685, + "valid_targets_mean": 3798.8, + "valid_targets_min": 1493 + }, + { + "epoch": 2.7777215189873417, + "grad_norm": 0.27448884582560484, + "learning_rate": 1.9688963761592984e-05, + "loss": 0.2547, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02799350395798683, + "step": 686, + "valid_targets_mean": 3781.4, + "valid_targets_min": 1855 + }, + { + "epoch": 2.781772151898734, + "grad_norm": 0.3077186286572531, + "learning_rate": 1.963241759676897e-05, + "loss": 0.2604, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03678155690431595, + "step": 687, + "valid_targets_mean": 5446.1, + "valid_targets_min": 1235 + }, + { + "epoch": 2.7858227848101267, + "grad_norm": 0.273877614979815, + "learning_rate": 1.9575874371124066e-05, + "loss": 0.2405, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.029743792489171028, + "step": 688, + "valid_targets_mean": 4165.7, + "valid_targets_min": 915 + }, + { + "epoch": 2.7898734177215188, + "grad_norm": 0.2939071744294542, + "learning_rate": 1.951933453677638e-05, + "loss": 0.2543, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.040034182369709015, + "step": 689, + "valid_targets_mean": 5826.3, + "valid_targets_min": 1081 + }, + { + "epoch": 2.7939240506329113, + "grad_norm": 0.338200305666527, + "learning_rate": 1.9462798545816932e-05, + "loss": 0.2635, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020607519894838333, + "step": 690, + "valid_targets_mean": 3353.9, + "valid_targets_min": 1111 + }, + { + "epoch": 2.797974683544304, + "grad_norm": 0.2693609927482392, + "learning_rate": 1.9406266850305996e-05, + "loss": 0.2556, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030511412769556046, + "step": 691, + "valid_targets_mean": 5535.2, + "valid_targets_min": 1391 + }, + { + "epoch": 2.8020253164556963, + "grad_norm": 0.32107879586411486, + "learning_rate": 1.9349739902269503e-05, + "loss": 0.2447, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.025336556136608124, + "step": 692, + "valid_targets_mean": 4009.9, + "valid_targets_min": 1289 + }, + { + "epoch": 2.806075949367089, + "grad_norm": 0.2821704052610864, + "learning_rate": 1.9293218153695425e-05, + "loss": 0.2623, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03870566934347153, + "step": 693, + "valid_targets_mean": 6418.9, + "valid_targets_min": 913 + }, + { + "epoch": 2.810126582278481, + "grad_norm": 0.2947465266050635, + "learning_rate": 1.9236702056530153e-05, + "loss": 0.2515, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03403696417808533, + "step": 694, + "valid_targets_mean": 4783.3, + "valid_targets_min": 1247 + }, + { + "epoch": 2.8141772151898734, + "grad_norm": 0.3224972914915537, + "learning_rate": 1.9180192062674907e-05, + "loss": 0.2639, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04721870645880699, + "step": 695, + "valid_targets_mean": 6279.4, + "valid_targets_min": 1563 + }, + { + "epoch": 2.818227848101266, + "grad_norm": 0.2885491555150038, + "learning_rate": 1.9123688623982076e-05, + "loss": 0.2522, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018630696460604668, + "step": 696, + "valid_targets_mean": 3067.3, + "valid_targets_min": 957 + }, + { + "epoch": 2.822278481012658, + "grad_norm": 0.2705441196421664, + "learning_rate": 1.906719219225165e-05, + "loss": 0.2406, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01688503846526146, + "step": 697, + "valid_targets_mean": 2701.2, + "valid_targets_min": 973 + }, + { + "epoch": 2.8263291139240505, + "grad_norm": 0.3168281330219779, + "learning_rate": 1.9010703219227603e-05, + "loss": 0.2527, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03765212744474411, + "step": 698, + "valid_targets_mean": 4885.8, + "valid_targets_min": 1524 + }, + { + "epoch": 2.830379746835443, + "grad_norm": 0.2807198194935173, + "learning_rate": 1.8954222156594256e-05, + "loss": 0.265, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03330278396606445, + "step": 699, + "valid_targets_mean": 4268.5, + "valid_targets_min": 1146 + }, + { + "epoch": 2.8344303797468355, + "grad_norm": 0.3256073602246422, + "learning_rate": 1.8897749455972673e-05, + "loss": 0.2527, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028569743037223816, + "step": 700, + "valid_targets_mean": 3776.6, + "valid_targets_min": 733 + }, + { + "epoch": 2.838481012658228, + "grad_norm": 0.27646963358428234, + "learning_rate": 1.884128556891707e-05, + "loss": 0.2478, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04450641945004463, + "step": 701, + "valid_targets_mean": 4887.8, + "valid_targets_min": 1283 + }, + { + "epoch": 2.84253164556962, + "grad_norm": 0.2741209798208164, + "learning_rate": 1.878483094691119e-05, + "loss": 0.2482, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0251961387693882, + "step": 702, + "valid_targets_mean": 3683.0, + "valid_targets_min": 1118 + }, + { + "epoch": 2.8465822784810126, + "grad_norm": 0.28110770884148384, + "learning_rate": 1.8728386041364664e-05, + "loss": 0.2544, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.038397304713726044, + "step": 703, + "valid_targets_mean": 5510.7, + "valid_targets_min": 1309 + }, + { + "epoch": 2.850632911392405, + "grad_norm": 0.25860615600142395, + "learning_rate": 1.867195130360947e-05, + "loss": 0.2436, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.045691635459661484, + "step": 704, + "valid_targets_mean": 6492.9, + "valid_targets_min": 1402 + }, + { + "epoch": 2.8546835443037972, + "grad_norm": 0.2910171774254887, + "learning_rate": 1.8615527184896263e-05, + "loss": 0.2527, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03464601933956146, + "step": 705, + "valid_targets_mean": 4857.5, + "valid_targets_min": 961 + }, + { + "epoch": 2.8587341772151897, + "grad_norm": 0.2658774941052945, + "learning_rate": 1.8559114136390795e-05, + "loss": 0.248, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.039772506803274155, + "step": 706, + "valid_targets_mean": 5122.4, + "valid_targets_min": 1059 + }, + { + "epoch": 2.8627848101265823, + "grad_norm": 0.25838343326049434, + "learning_rate": 1.8502712609170298e-05, + "loss": 0.2539, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03241926431655884, + "step": 707, + "valid_targets_mean": 4657.2, + "valid_targets_min": 992 + }, + { + "epoch": 2.8668354430379748, + "grad_norm": 0.29961837015268405, + "learning_rate": 1.8446323054219876e-05, + "loss": 0.2603, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03099638596177101, + "step": 708, + "valid_targets_mean": 4535.9, + "valid_targets_min": 1544 + }, + { + "epoch": 2.8708860759493673, + "grad_norm": 0.2826577983451744, + "learning_rate": 1.838994592242891e-05, + "loss": 0.2494, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.022791540250182152, + "step": 709, + "valid_targets_mean": 3683.8, + "valid_targets_min": 1101 + }, + { + "epoch": 2.8749367088607594, + "grad_norm": 0.2789431626612818, + "learning_rate": 1.8333581664587453e-05, + "loss": 0.2399, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03397709131240845, + "step": 710, + "valid_targets_mean": 4410.2, + "valid_targets_min": 1095 + }, + { + "epoch": 2.878987341772152, + "grad_norm": 0.30434622336390565, + "learning_rate": 1.8277230731382593e-05, + "loss": 0.2535, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028042972087860107, + "step": 711, + "valid_targets_mean": 4426.9, + "valid_targets_min": 1666 + }, + { + "epoch": 2.8830379746835444, + "grad_norm": 0.27371088369925034, + "learning_rate": 1.82208935733949e-05, + "loss": 0.2419, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.031526222825050354, + "step": 712, + "valid_targets_mean": 4659.2, + "valid_targets_min": 784 + }, + { + "epoch": 2.8870886075949365, + "grad_norm": 0.2685540373810642, + "learning_rate": 1.8164570641094793e-05, + "loss": 0.2588, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.029739882797002792, + "step": 713, + "valid_targets_mean": 4144.9, + "valid_targets_min": 915 + }, + { + "epoch": 2.891139240506329, + "grad_norm": 0.2759997112656724, + "learning_rate": 1.810826238483892e-05, + "loss": 0.2436, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02822181209921837, + "step": 714, + "valid_targets_mean": 4396.9, + "valid_targets_min": 1350 + }, + { + "epoch": 2.8951898734177215, + "grad_norm": 0.2643249992462491, + "learning_rate": 1.8051969254866608e-05, + "loss": 0.2537, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.033480800688266754, + "step": 715, + "valid_targets_mean": 4895.6, + "valid_targets_min": 1502 + }, + { + "epoch": 2.899240506329114, + "grad_norm": 0.2646348469873884, + "learning_rate": 1.7995691701296226e-05, + "loss": 0.2657, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030401896685361862, + "step": 716, + "valid_targets_mean": 4391.8, + "valid_targets_min": 1157 + }, + { + "epoch": 2.9032911392405065, + "grad_norm": 0.2597089268403161, + "learning_rate": 1.7939430174121593e-05, + "loss": 0.2472, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02844076231122017, + "step": 717, + "valid_targets_mean": 4140.0, + "valid_targets_min": 1646 + }, + { + "epoch": 2.9073417721518986, + "grad_norm": 0.26283592691048274, + "learning_rate": 1.788318512320838e-05, + "loss": 0.262, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03905324637889862, + "step": 718, + "valid_targets_mean": 5344.9, + "valid_targets_min": 1717 + }, + { + "epoch": 2.911392405063291, + "grad_norm": 0.2716776578167852, + "learning_rate": 1.782695699829051e-05, + "loss": 0.239, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021692726761102676, + "step": 719, + "valid_targets_mean": 2800.7, + "valid_targets_min": 1130 + }, + { + "epoch": 2.9154430379746836, + "grad_norm": 0.26509502419429243, + "learning_rate": 1.7770746248966587e-05, + "loss": 0.2532, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030680546537041664, + "step": 720, + "valid_targets_mean": 4364.1, + "valid_targets_min": 899 + }, + { + "epoch": 2.9194936708860757, + "grad_norm": 0.28772245040534483, + "learning_rate": 1.7714553324696243e-05, + "loss": 0.2698, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03688710182905197, + "step": 721, + "valid_targets_mean": 4688.4, + "valid_targets_min": 1281 + }, + { + "epoch": 2.923544303797468, + "grad_norm": 0.2810229272917195, + "learning_rate": 1.7658378674796614e-05, + "loss": 0.2622, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02813960425555706, + "step": 722, + "valid_targets_mean": 4541.4, + "valid_targets_min": 1352 + }, + { + "epoch": 2.9275949367088607, + "grad_norm": 0.30002042412455127, + "learning_rate": 1.76022227484387e-05, + "loss": 0.2644, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03764427825808525, + "step": 723, + "valid_targets_mean": 5619.6, + "valid_targets_min": 1286 + }, + { + "epoch": 2.9316455696202532, + "grad_norm": 0.25999917059624383, + "learning_rate": 1.7546085994643807e-05, + "loss": 0.2476, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020120546221733093, + "step": 724, + "valid_targets_mean": 3131.7, + "valid_targets_min": 1088 + }, + { + "epoch": 2.9356962025316458, + "grad_norm": 0.24301031977739251, + "learning_rate": 1.7489968862279902e-05, + "loss": 0.2401, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030156666412949562, + "step": 725, + "valid_targets_mean": 4536.0, + "valid_targets_min": 889 + }, + { + "epoch": 2.939746835443038, + "grad_norm": 0.2834351965141156, + "learning_rate": 1.743387180005809e-05, + "loss": 0.2653, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.052154481410980225, + "step": 726, + "valid_targets_mean": 7368.1, + "valid_targets_min": 1318 + }, + { + "epoch": 2.9437974683544303, + "grad_norm": 0.27572670971938895, + "learning_rate": 1.737779525652899e-05, + "loss": 0.2646, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.024445312097668648, + "step": 727, + "valid_targets_mean": 3959.4, + "valid_targets_min": 1066 + }, + { + "epoch": 2.947848101265823, + "grad_norm": 0.2704592646522639, + "learning_rate": 1.732173968007914e-05, + "loss": 0.2597, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02838369458913803, + "step": 728, + "valid_targets_mean": 3935.4, + "valid_targets_min": 1243 + }, + { + "epoch": 2.951898734177215, + "grad_norm": 0.2685193260978328, + "learning_rate": 1.7265705518927437e-05, + "loss": 0.2485, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02375240996479988, + "step": 729, + "valid_targets_mean": 3453.5, + "valid_targets_min": 1300 + }, + { + "epoch": 2.9559493670886074, + "grad_norm": 0.2578427303145989, + "learning_rate": 1.7209693221121542e-05, + "loss": 0.254, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023774366825819016, + "step": 730, + "valid_targets_mean": 3456.4, + "valid_targets_min": 1458 + }, + { + "epoch": 2.96, + "grad_norm": 0.2546217929799125, + "learning_rate": 1.7153703234534302e-05, + "loss": 0.2468, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02758769877254963, + "step": 731, + "valid_targets_mean": 4289.4, + "valid_targets_min": 976 + }, + { + "epoch": 2.9640506329113925, + "grad_norm": 0.2654879948700078, + "learning_rate": 1.709773600686015e-05, + "loss": 0.2595, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.04343331605195999, + "step": 732, + "valid_targets_mean": 5717.3, + "valid_targets_min": 1722 + }, + { + "epoch": 2.968101265822785, + "grad_norm": 0.25748033286051897, + "learning_rate": 1.7041791985611548e-05, + "loss": 0.2487, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.050244733691215515, + "step": 733, + "valid_targets_mean": 6772.3, + "valid_targets_min": 1419 + }, + { + "epoch": 2.972151898734177, + "grad_norm": 0.2542798539074249, + "learning_rate": 1.69858716181154e-05, + "loss": 0.2452, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026083718985319138, + "step": 734, + "valid_targets_mean": 3278.6, + "valid_targets_min": 770 + }, + { + "epoch": 2.9762025316455696, + "grad_norm": 0.2572335702204434, + "learning_rate": 1.692997535150948e-05, + "loss": 0.2444, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03589995950460434, + "step": 735, + "valid_targets_mean": 5458.4, + "valid_targets_min": 1202 + }, + { + "epoch": 2.980253164556962, + "grad_norm": 0.254374544775352, + "learning_rate": 1.687410363273884e-05, + "loss": 0.2538, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.029865849763154984, + "step": 736, + "valid_targets_mean": 4263.7, + "valid_targets_min": 1376 + }, + { + "epoch": 2.984303797468354, + "grad_norm": 0.26497805309307315, + "learning_rate": 1.6818256908552257e-05, + "loss": 0.2593, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.038646139204502106, + "step": 737, + "valid_targets_mean": 5068.5, + "valid_targets_min": 1461 + }, + { + "epoch": 2.9883544303797467, + "grad_norm": 0.2569838835553243, + "learning_rate": 1.6762435625498646e-05, + "loss": 0.2453, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02871629409492016, + "step": 738, + "valid_targets_mean": 4765.9, + "valid_targets_min": 1531 + }, + { + "epoch": 2.992405063291139, + "grad_norm": 0.2530215798640709, + "learning_rate": 1.67066402299235e-05, + "loss": 0.2662, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.036961380392313004, + "step": 739, + "valid_targets_mean": 5032.2, + "valid_targets_min": 1194 + }, + { + "epoch": 2.9964556962025317, + "grad_norm": 0.26245999391543, + "learning_rate": 1.6650871167965313e-05, + "loss": 0.2544, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02777264080941677, + "step": 740, + "valid_targets_mean": 3679.3, + "valid_targets_min": 1109 + }, + { + "epoch": 3.0, + "grad_norm": 0.2793670359330504, + "learning_rate": 1.6595128885552028e-05, + "loss": 0.2518, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03139267489314079, + "step": 741, + "valid_targets_mean": 3354.4, + "valid_targets_min": 977 + }, + { + "epoch": 3.0040506329113925, + "grad_norm": 0.4518355029730338, + "learning_rate": 1.653941382839745e-05, + "loss": 0.1875, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.029589787125587463, + "step": 742, + "valid_targets_mean": 5513.1, + "valid_targets_min": 1195 + }, + { + "epoch": 3.008101265822785, + "grad_norm": 0.3131036035935897, + "learning_rate": 1.6483726441997673e-05, + "loss": 0.1707, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.022383853793144226, + "step": 743, + "valid_targets_mean": 5414.9, + "valid_targets_min": 1062 + }, + { + "epoch": 3.012151898734177, + "grad_norm": 0.43578496818594536, + "learning_rate": 1.642806717162757e-05, + "loss": 0.1798, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018505379557609558, + "step": 744, + "valid_targets_mean": 3892.1, + "valid_targets_min": 1224 + }, + { + "epoch": 3.0162025316455696, + "grad_norm": 0.5303532489283547, + "learning_rate": 1.637243646233718e-05, + "loss": 0.1797, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02654384821653366, + "step": 745, + "valid_targets_mean": 4359.8, + "valid_targets_min": 1391 + }, + { + "epoch": 3.020253164556962, + "grad_norm": 0.36041463705304083, + "learning_rate": 1.6316834758948174e-05, + "loss": 0.1723, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01781952753663063, + "step": 746, + "valid_targets_mean": 4052.8, + "valid_targets_min": 1279 + }, + { + "epoch": 3.0243037974683546, + "grad_norm": 0.3558315031597273, + "learning_rate": 1.6261262506050282e-05, + "loss": 0.1742, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01695730909705162, + "step": 747, + "valid_targets_mean": 3121.8, + "valid_targets_min": 788 + }, + { + "epoch": 3.0283544303797467, + "grad_norm": 0.38372608971688293, + "learning_rate": 1.620572014799777e-05, + "loss": 0.1711, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026282865554094315, + "step": 748, + "valid_targets_mean": 5316.9, + "valid_targets_min": 1490 + }, + { + "epoch": 3.0324050632911392, + "grad_norm": 0.3293468845617277, + "learning_rate": 1.6150208128905857e-05, + "loss": 0.1635, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02892507240176201, + "step": 749, + "valid_targets_mean": 5845.9, + "valid_targets_min": 1405 + }, + { + "epoch": 3.0364556962025318, + "grad_norm": 0.31716354074304054, + "learning_rate": 1.6094726892647147e-05, + "loss": 0.1717, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.024220801889896393, + "step": 750, + "valid_targets_mean": 5615.8, + "valid_targets_min": 1425 + }, + { + "epoch": 3.0405063291139243, + "grad_norm": 0.3664828568121349, + "learning_rate": 1.6039276882848135e-05, + "loss": 0.1762, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013677317649126053, + "step": 751, + "valid_targets_mean": 3208.0, + "valid_targets_min": 970 }, { - "epoch": 1.0126582278481013, - "grad_norm": 0.5153235955639972, - "learning_rate": 3.871499275032111e-05, - "loss": 0.3223, - "step": 250 + "epoch": 3.0445569620253163, + "grad_norm": 0.33903502187956835, + "learning_rate": 1.5983858542885635e-05, + "loss": 0.1735, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018305784091353416, + "step": 752, + "valid_targets_mean": 3494.2, + "valid_targets_min": 1546 }, { - "epoch": 1.0167088607594936, - "grad_norm": 0.38156584074959976, - "learning_rate": 3.869490061403091e-05, - "loss": 0.2978, - "step": 251 + "epoch": 3.048607594936709, + "grad_norm": 0.3498757139361586, + "learning_rate": 1.5928472315883203e-05, + "loss": 0.1754, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014736996963620186, + "step": 753, + "valid_targets_mean": 3807.8, + "valid_targets_min": 1288 + }, + { + "epoch": 3.0526582278481014, + "grad_norm": 0.34193040921745776, + "learning_rate": 1.5873118644707633e-05, + "loss": 0.1822, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02990785986185074, + "step": 754, + "valid_targets_mean": 5325.8, + "valid_targets_min": 1245 + }, + { + "epoch": 3.056708860759494, + "grad_norm": 0.33397160073921883, + "learning_rate": 1.5817797971965413e-05, + "loss": 0.1821, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014044428244233131, + "step": 755, + "valid_targets_mean": 3148.8, + "valid_targets_min": 1382 + }, + { + "epoch": 3.060759493670886, + "grad_norm": 0.3323577994417775, + "learning_rate": 1.576251073999917e-05, + "loss": 0.1682, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01589152216911316, + "step": 756, + "valid_targets_mean": 3347.2, + "valid_targets_min": 1044 + }, + { + "epoch": 3.0648101265822785, + "grad_norm": 0.33502996544962327, + "learning_rate": 1.5707257390884126e-05, + "loss": 0.1638, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017870556563138962, + "step": 757, + "valid_targets_mean": 3450.2, + "valid_targets_min": 745 }, { - "epoch": 1.0207594936708861, - "grad_norm": 0.4694601899432551, - "learning_rate": 3.867465791158124e-05, - "loss": 0.3117, - "step": 252 + "epoch": 3.068860759493671, + "grad_norm": 0.30464635685328645, + "learning_rate": 1.5652038366424595e-05, + "loss": 0.1663, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.024972693994641304, + "step": 758, + "valid_targets_mean": 5675.8, + "valid_targets_min": 1103 }, { - "epoch": 1.0248101265822784, - "grad_norm": 0.4472606457664077, - "learning_rate": 3.865426480600407e-05, - "loss": 0.3052, - "step": 253 + "epoch": 3.0729113924050635, + "grad_norm": 0.317761137059346, + "learning_rate": 1.5596854108150424e-05, + "loss": 0.1715, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01955476403236389, + "step": 759, + "valid_targets_mean": 4004.8, + "valid_targets_min": 1108 + }, + { + "epoch": 3.0769620253164556, + "grad_norm": 0.29358636528638604, + "learning_rate": 1.5541705057313476e-05, + "loss": 0.1712, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.019088895991444588, + "step": 760, + "valid_targets_mean": 4483.4, + "valid_targets_min": 1176 + }, + { + "epoch": 3.081012658227848, + "grad_norm": 0.3073330115701503, + "learning_rate": 1.5486591654884086e-05, + "loss": 0.1718, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015020224265754223, + "step": 761, + "valid_targets_mean": 3669.2, + "valid_targets_min": 1121 + }, + { + "epoch": 3.0850632911392406, + "grad_norm": 0.299608173186423, + "learning_rate": 1.543151434154755e-05, + "loss": 0.1778, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02530589886009693, + "step": 762, + "valid_targets_mean": 5061.1, + "valid_targets_min": 1440 + }, + { + "epoch": 3.089113924050633, + "grad_norm": 0.27825960482122103, + "learning_rate": 1.5376473557700615e-05, + "loss": 0.1802, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015576534904539585, + "step": 763, + "valid_targets_mean": 3578.2, + "valid_targets_min": 1242 + }, + { + "epoch": 3.093164556962025, + "grad_norm": 0.3316565116707584, + "learning_rate": 1.5321469743447893e-05, + "loss": 0.1749, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.024011485278606415, + "step": 764, + "valid_targets_mean": 4485.5, + "valid_targets_min": 1289 + }, + { + "epoch": 3.0972151898734177, + "grad_norm": 0.29137811908746475, + "learning_rate": 1.5266503338598434e-05, + "loss": 0.1722, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.025936177000403404, + "step": 765, + "valid_targets_mean": 4847.2, + "valid_targets_min": 1336 + }, + { + "epoch": 3.1012658227848102, + "grad_norm": 0.29389891839150223, + "learning_rate": 1.5211574782662149e-05, + "loss": 0.1728, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021893944591283798, + "step": 766, + "valid_targets_mean": 5340.1, + "valid_targets_min": 1699 + }, + { + "epoch": 3.1053164556962027, + "grad_norm": 0.29807222422096724, + "learning_rate": 1.5156684514846307e-05, + "loss": 0.1695, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02523096464574337, + "step": 767, + "valid_targets_mean": 4752.1, + "valid_targets_min": 875 + }, + { + "epoch": 3.109367088607595, + "grad_norm": 0.30795542223986466, + "learning_rate": 1.5101832974052012e-05, + "loss": 0.1809, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026390738785266876, + "step": 768, + "valid_targets_mean": 4690.3, + "valid_targets_min": 885 + }, + { + "epoch": 3.1134177215189873, + "grad_norm": 0.31729949382152767, + "learning_rate": 1.5047020598870737e-05, + "loss": 0.18, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02073775604367256, + "step": 769, + "valid_targets_mean": 4297.0, + "valid_targets_min": 1332 + }, + { + "epoch": 3.11746835443038, + "grad_norm": 0.3112664533988072, + "learning_rate": 1.4992247827580778e-05, + "loss": 0.1779, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02091725915670395, + "step": 770, + "valid_targets_mean": 3788.1, + "valid_targets_min": 778 + }, + { + "epoch": 3.1215189873417724, + "grad_norm": 0.30766965175005956, + "learning_rate": 1.493751509814374e-05, + "loss": 0.1772, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016842763870954514, + "step": 771, + "valid_targets_mean": 3957.2, + "valid_targets_min": 1020 }, { - "epoch": 1.028860759493671, - "grad_norm": 0.4213830764206453, - "learning_rate": 3.863372146154264e-05, - "loss": 0.3055, - "step": 254 + "epoch": 3.1255696202531644, + "grad_norm": 0.3033325549159383, + "learning_rate": 1.4882822848201073e-05, + "loss": 0.1714, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03626658022403717, + "step": 772, + "valid_targets_mean": 7330.2, + "valid_targets_min": 1320 + }, + { + "epoch": 3.129620253164557, + "grad_norm": 0.3087107442257974, + "learning_rate": 1.4828171515070553e-05, + "loss": 0.1776, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02217162773013115, + "step": 773, + "valid_targets_mean": 4305.2, + "valid_targets_min": 1379 + }, + { + "epoch": 3.1336708860759495, + "grad_norm": 0.2828472592017574, + "learning_rate": 1.4773561535742793e-05, + "loss": 0.1567, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.025130566209554672, + "step": 774, + "valid_targets_mean": 5140.2, + "valid_targets_min": 1524 }, { - "epoch": 1.0329113924050632, - "grad_norm": 0.4520073084621344, - "learning_rate": 3.861302804365024e-05, - "loss": 0.3175, - "step": 255 + "epoch": 3.137721518987342, + "grad_norm": 0.2879324146158943, + "learning_rate": 1.4718993346877715e-05, + "loss": 0.1734, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01935396157205105, + "step": 775, + "valid_targets_mean": 4015.8, + "valid_targets_min": 1750 + }, + { + "epoch": 3.141772151898734, + "grad_norm": 0.2861796742039433, + "learning_rate": 1.466446738480111e-05, + "loss": 0.1625, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023186706006526947, + "step": 776, + "valid_targets_mean": 5214.1, + "valid_targets_min": 1322 + }, + { + "epoch": 3.1458227848101266, + "grad_norm": 0.2797425568919168, + "learning_rate": 1.460998408550112e-05, + "loss": 0.1775, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02018989436328411, + "step": 777, + "valid_targets_mean": 4186.8, + "valid_targets_min": 1263 + }, + { + "epoch": 3.149873417721519, + "grad_norm": 0.28560455048945466, + "learning_rate": 1.4555543884624751e-05, + "loss": 0.1679, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014386219903826714, + "step": 778, + "valid_targets_mean": 3198.4, + "valid_targets_min": 979 + }, + { + "epoch": 3.1539240506329116, + "grad_norm": 0.29510716090586336, + "learning_rate": 1.4501147217474402e-05, + "loss": 0.1673, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018298868089914322, + "step": 779, + "valid_targets_mean": 3162.6, + "valid_targets_min": 1179 + }, + { + "epoch": 3.1579746835443037, + "grad_norm": 0.29166498949590175, + "learning_rate": 1.444679451900437e-05, + "loss": 0.1723, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03829861432313919, + "step": 780, + "valid_targets_mean": 8741.9, + "valid_targets_min": 1108 + }, + { + "epoch": 3.162025316455696, + "grad_norm": 0.27182783348812456, + "learning_rate": 1.4392486223817397e-05, + "loss": 0.173, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.024232111871242523, + "step": 781, + "valid_targets_mean": 5093.5, + "valid_targets_min": 1537 + }, + { + "epoch": 3.1660759493670887, + "grad_norm": 0.29577602878611126, + "learning_rate": 1.4338222766161143e-05, + "loss": 0.1805, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01894107460975647, + "step": 782, + "valid_targets_mean": 3929.9, + "valid_targets_min": 1014 + }, + { + "epoch": 3.170126582278481, + "grad_norm": 0.2899090365502566, + "learning_rate": 1.4284004579924777e-05, + "loss": 0.1637, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021671906113624573, + "step": 783, + "valid_targets_mean": 4770.2, + "valid_targets_min": 1293 + }, + { + "epoch": 3.1741772151898733, + "grad_norm": 0.27733370216002134, + "learning_rate": 1.4229832098635479e-05, + "loss": 0.1721, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.024629831314086914, + "step": 784, + "valid_targets_mean": 6038.6, + "valid_targets_min": 1443 + }, + { + "epoch": 3.178227848101266, + "grad_norm": 0.3000817308258268, + "learning_rate": 1.4175705755454963e-05, + "loss": 0.1679, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0261150561273098, + "step": 785, + "valid_targets_mean": 4464.9, + "valid_targets_min": 1302 + }, + { + "epoch": 3.1822784810126583, + "grad_norm": 0.2785628374210658, + "learning_rate": 1.412162598317602e-05, + "loss": 0.1725, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020167309790849686, + "step": 786, + "valid_targets_mean": 4205.3, + "valid_targets_min": 1347 }, { - "epoch": 1.0369620253164558, - "grad_norm": 0.39518451005870114, - "learning_rate": 3.85921847189888e-05, - "loss": 0.3141, - "step": 256 + "epoch": 3.186329113924051, + "grad_norm": 0.2766193388956131, + "learning_rate": 1.406759321421907e-05, + "loss": 0.1722, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012921443209052086, + "step": 787, + "valid_targets_mean": 3072.4, + "valid_targets_min": 1502 + }, + { + "epoch": 3.190379746835443, + "grad_norm": 0.2835705291271444, + "learning_rate": 1.401360788062871e-05, + "loss": 0.1718, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012393220327794552, + "step": 788, + "valid_targets_mean": 3018.7, + "valid_targets_min": 879 + }, + { + "epoch": 3.1944303797468354, + "grad_norm": 0.29540883185162925, + "learning_rate": 1.3959670414070204e-05, + "loss": 0.1649, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026593632996082306, + "step": 789, + "valid_targets_mean": 4636.0, + "valid_targets_min": 1379 + }, + { + "epoch": 3.198481012658228, + "grad_norm": 0.2911916198878864, + "learning_rate": 1.3905781245826108e-05, + "loss": 0.1701, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.027367331087589264, + "step": 790, + "valid_targets_mean": 4947.6, + "valid_targets_min": 1267 + }, + { + "epoch": 3.2025316455696204, + "grad_norm": 0.29329356188410366, + "learning_rate": 1.3851940806792778e-05, + "loss": 0.1624, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017680518329143524, + "step": 791, + "valid_targets_mean": 3998.1, + "valid_targets_min": 870 + }, + { + "epoch": 3.2065822784810125, + "grad_norm": 0.3083830348494984, + "learning_rate": 1.379814952747693e-05, + "loss": 0.1696, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.025335486978292465, + "step": 792, + "valid_targets_mean": 5041.7, + "valid_targets_min": 1133 + }, + { + "epoch": 3.210632911392405, + "grad_norm": 0.2798046695321331, + "learning_rate": 1.3744407837992193e-05, + "loss": 0.1762, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020378734916448593, + "step": 793, + "valid_targets_mean": 4000.8, + "valid_targets_min": 1305 + }, + { + "epoch": 3.2146835443037975, + "grad_norm": 0.2781061493945947, + "learning_rate": 1.3690716168055692e-05, + "loss": 0.1723, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.029875535517930984, + "step": 794, + "valid_targets_mean": 6269.0, + "valid_targets_min": 1199 + }, + { + "epoch": 3.21873417721519, + "grad_norm": 0.28425247889799193, + "learning_rate": 1.363707494698459e-05, + "loss": 0.1747, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01849631406366825, + "step": 795, + "valid_targets_mean": 4173.1, + "valid_targets_min": 1121 + }, + { + "epoch": 3.222784810126582, + "grad_norm": 0.29167790078536165, + "learning_rate": 1.358348460369265e-05, + "loss": 0.1648, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016716711223125458, + "step": 796, + "valid_targets_mean": 3498.2, + "valid_targets_min": 1028 + }, + { + "epoch": 3.2268354430379746, + "grad_norm": 0.2706793655937343, + "learning_rate": 1.352994556668684e-05, + "loss": 0.176, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02109239064157009, + "step": 797, + "valid_targets_mean": 4928.1, + "valid_targets_min": 1692 + }, + { + "epoch": 3.230886075949367, + "grad_norm": 0.2858053934845192, + "learning_rate": 1.3476458264063865e-05, + "loss": 0.1646, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012462573125958443, + "step": 798, + "valid_targets_mean": 3088.8, + "valid_targets_min": 875 + }, + { + "epoch": 3.2349367088607597, + "grad_norm": 0.30039221272535604, + "learning_rate": 1.342302312350678e-05, + "loss": 0.17, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016949564218521118, + "step": 799, + "valid_targets_mean": 3712.4, + "valid_targets_min": 1407 + }, + { + "epoch": 3.2389873417721518, + "grad_norm": 0.2787693166001655, + "learning_rate": 1.3369640572281537e-05, + "loss": 0.1693, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020679127424955368, + "step": 800, + "valid_targets_mean": 4078.4, + "valid_targets_min": 1072 + }, + { + "epoch": 3.2430379746835443, + "grad_norm": 0.29520428907315616, + "learning_rate": 1.3316311037233596e-05, + "loss": 0.1719, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018096426501870155, + "step": 801, + "valid_targets_mean": 3506.6, + "valid_targets_min": 1152 }, { - "epoch": 1.041012658227848, - "grad_norm": 0.4593623563226846, - "learning_rate": 3.85711916554276e-05, - "loss": 0.3093, - "step": 257 + "epoch": 3.247088607594937, + "grad_norm": 0.2667786897026877, + "learning_rate": 1.326303494478451e-05, + "loss": 0.1734, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023975923657417297, + "step": 802, + "valid_targets_mean": 5160.4, + "valid_targets_min": 1551 + }, + { + "epoch": 3.2511392405063293, + "grad_norm": 0.2770929210536329, + "learning_rate": 1.3209812720928472e-05, + "loss": 0.1756, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.027329906821250916, + "step": 803, + "valid_targets_mean": 5768.4, + "valid_targets_min": 1146 + }, + { + "epoch": 3.2551898734177214, + "grad_norm": 0.29416739016414467, + "learning_rate": 1.315664479122898e-05, + "loss": 0.1738, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023470068350434303, + "step": 804, + "valid_targets_mean": 3871.2, + "valid_targets_min": 1226 + }, + { + "epoch": 3.259240506329114, + "grad_norm": 0.2620201865177205, + "learning_rate": 1.3103531580815378e-05, + "loss": 0.1752, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.022001316770911217, + "step": 805, + "valid_targets_mean": 5060.7, + "valid_targets_min": 1445 + }, + { + "epoch": 3.2632911392405064, + "grad_norm": 0.28078996386465055, + "learning_rate": 1.3050473514379488e-05, + "loss": 0.1706, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02083248645067215, + "step": 806, + "valid_targets_mean": 4420.8, + "valid_targets_min": 1243 }, { - "epoch": 1.0450632911392406, - "grad_norm": 0.4548290497711387, - "learning_rate": 3.85500490220419e-05, - "loss": 0.3079, - "step": 258 + "epoch": 3.267341772151899, + "grad_norm": 0.27940811510151586, + "learning_rate": 1.2997471016172188e-05, + "loss": 0.1715, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023604262620210648, + "step": 807, + "valid_targets_mean": 5010.2, + "valid_targets_min": 1028 }, { - "epoch": 1.0491139240506329, - "grad_norm": 0.4029758807372885, - "learning_rate": 3.852875698911154e-05, - "loss": 0.294, - "step": 259 + "epoch": 3.271392405063291, + "grad_norm": 0.2792666068611182, + "learning_rate": 1.2944524510000042e-05, + "loss": 0.1678, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02203407883644104, + "step": 808, + "valid_targets_mean": 4796.6, + "valid_targets_min": 1529 + }, + { + "epoch": 3.2754430379746835, + "grad_norm": 0.2770883931596026, + "learning_rate": 1.2891634419221908e-05, + "loss": 0.1798, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.024457737803459167, + "step": 809, + "valid_targets_mean": 5265.9, + "valid_targets_min": 1340 + }, + { + "epoch": 3.279493670886076, + "grad_norm": 0.28475756468903735, + "learning_rate": 1.2838801166745545e-05, + "loss": 0.1717, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0277986042201519, + "step": 810, + "valid_targets_mean": 4750.3, + "valid_targets_min": 1003 + }, + { + "epoch": 3.2835443037974685, + "grad_norm": 0.28703035190856646, + "learning_rate": 1.2786025175024216e-05, + "loss": 0.172, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028684845194220543, + "step": 811, + "valid_targets_mean": 5778.2, + "valid_targets_min": 1094 + }, + { + "epoch": 3.2875949367088606, + "grad_norm": 0.2575044676747565, + "learning_rate": 1.2733306866053357e-05, + "loss": 0.1786, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016220424324274063, + "step": 812, + "valid_targets_mean": 3783.6, + "valid_targets_min": 939 + }, + { + "epoch": 3.291645569620253, + "grad_norm": 0.2668553861153468, + "learning_rate": 1.2680646661367163e-05, + "loss": 0.1779, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0175221785902977, + "step": 813, + "valid_targets_mean": 4105.9, + "valid_targets_min": 1251 + }, + { + "epoch": 3.2956962025316456, + "grad_norm": 0.2898200450052223, + "learning_rate": 1.2628044982035219e-05, + "loss": 0.1715, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021266207098960876, + "step": 814, + "valid_targets_mean": 4000.8, + "valid_targets_min": 1101 }, { - "epoch": 1.0531645569620254, - "grad_norm": 0.44801762962474895, - "learning_rate": 3.850731572811963e-05, - "loss": 0.311, - "step": 260 + "epoch": 3.299746835443038, + "grad_norm": 0.2777976709411515, + "learning_rate": 1.2575502248659155e-05, + "loss": 0.1737, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01444375328719616, + "step": 815, + "valid_targets_mean": 3121.9, + "valid_targets_min": 1417 + }, + { + "epoch": 3.3037974683544302, + "grad_norm": 0.2556688244111565, + "learning_rate": 1.2523018881369269e-05, + "loss": 0.1747, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02820611745119095, + "step": 816, + "valid_targets_mean": 6372.2, + "valid_targets_min": 1123 + }, + { + "epoch": 3.3078481012658227, + "grad_norm": 0.2828983328250475, + "learning_rate": 1.2470595299821176e-05, + "loss": 0.1729, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02478950098156929, + "step": 817, + "valid_targets_mean": 5102.2, + "valid_targets_min": 515 + }, + { + "epoch": 3.3118987341772153, + "grad_norm": 0.2663885169819304, + "learning_rate": 1.2418231923192416e-05, + "loss": 0.1725, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03402366489171982, + "step": 818, + "valid_targets_mean": 7563.4, + "valid_targets_min": 1038 }, { - "epoch": 1.0572151898734177, - "grad_norm": 0.42512004600221576, - "learning_rate": 3.848572541175116e-05, - "loss": 0.3213, - "step": 261 + "epoch": 3.3159493670886078, + "grad_norm": 0.28119186515935485, + "learning_rate": 1.2365929170179165e-05, + "loss": 0.1658, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018654121086001396, + "step": 819, + "valid_targets_mean": 4249.9, + "valid_targets_min": 1297 + }, + { + "epoch": 3.32, + "grad_norm": 0.2650192587281139, + "learning_rate": 1.2313687458992848e-05, + "loss": 0.1682, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014561069197952747, + "step": 820, + "valid_targets_mean": 3717.8, + "valid_targets_min": 1111 + }, + { + "epoch": 3.3240506329113924, + "grad_norm": 0.2677650578515307, + "learning_rate": 1.2261507207356785e-05, + "loss": 0.1667, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018807832151651382, + "step": 821, + "valid_targets_mean": 4245.8, + "valid_targets_min": 1257 + }, + { + "epoch": 3.328101265822785, + "grad_norm": 0.28086766974582916, + "learning_rate": 1.220938883250288e-05, + "loss": 0.1779, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02263430505990982, + "step": 822, + "valid_targets_mean": 4172.3, + "valid_targets_min": 1631 + }, + { + "epoch": 3.3321518987341774, + "grad_norm": 0.26787693516554345, + "learning_rate": 1.215733275116827e-05, + "loss": 0.1644, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020411396399140358, + "step": 823, + "valid_targets_mean": 4443.4, + "valid_targets_min": 1111 + }, + { + "epoch": 3.3362025316455695, + "grad_norm": 0.291485421895171, + "learning_rate": 1.2105339379592005e-05, + "loss": 0.1754, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023871038109064102, + "step": 824, + "valid_targets_mean": 4766.5, + "valid_targets_min": 1127 + }, + { + "epoch": 3.340253164556962, + "grad_norm": 0.2620819618225538, + "learning_rate": 1.2053409133511681e-05, + "loss": 0.1691, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.019396033138036728, + "step": 825, + "valid_targets_mean": 4995.4, + "valid_targets_min": 1031 + }, + { + "epoch": 3.3443037974683545, + "grad_norm": 0.26617063956170284, + "learning_rate": 1.2001542428160173e-05, + "loss": 0.1668, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01592746004462242, + "step": 826, + "valid_targets_mean": 3031.6, + "valid_targets_min": 1360 + }, + { + "epoch": 3.348354430379747, + "grad_norm": 0.2791441439661935, + "learning_rate": 1.1949739678262278e-05, + "loss": 0.159, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01481454074382782, + "step": 827, + "valid_targets_mean": 3479.9, + "valid_targets_min": 1489 + }, + { + "epoch": 3.352405063291139, + "grad_norm": 0.2788019727068902, + "learning_rate": 1.189800129803139e-05, + "loss": 0.1652, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014004124328494072, + "step": 828, + "valid_targets_mean": 3138.6, + "valid_targets_min": 1117 + }, + { + "epoch": 3.3564556962025316, + "grad_norm": 0.27843872235351913, + "learning_rate": 1.1846327701166228e-05, + "loss": 0.1662, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018882954493165016, + "step": 829, + "valid_targets_mean": 3859.1, + "valid_targets_min": 1149 + }, + { + "epoch": 3.360506329113924, + "grad_norm": 0.27454904391431767, + "learning_rate": 1.1794719300847493e-05, + "loss": 0.1706, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014589566737413406, + "step": 830, + "valid_targets_mean": 3264.4, + "valid_targets_min": 1556 + }, + { + "epoch": 3.3645569620253166, + "grad_norm": 0.28428328898750455, + "learning_rate": 1.1743176509734582e-05, + "loss": 0.1635, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021249011158943176, + "step": 831, + "valid_targets_mean": 4815.2, + "valid_targets_min": 1592 + }, + { + "epoch": 3.3686075949367087, + "grad_norm": 0.2925951397601006, + "learning_rate": 1.1691699739962275e-05, + "loss": 0.1711, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016693107783794403, + "step": 832, + "valid_targets_mean": 3980.7, + "valid_targets_min": 1435 }, { - "epoch": 1.0612658227848102, - "grad_norm": 0.42043840299184576, - "learning_rate": 3.846398621389154e-05, - "loss": 0.3183, - "step": 262 + "epoch": 3.372658227848101, + "grad_norm": 0.2907892696205683, + "learning_rate": 1.1640289403137456e-05, + "loss": 0.1617, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02736719697713852, + "step": 833, + "valid_targets_mean": 5829.9, + "valid_targets_min": 913 }, { - "epoch": 1.0653164556962025, - "grad_norm": 0.3892962415472906, - "learning_rate": 3.84420983096253e-05, - "loss": 0.3051, - "step": 263 + "epoch": 3.3767088607594937, + "grad_norm": 0.2887674410665316, + "learning_rate": 1.1588945910335802e-05, + "loss": 0.1661, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.029096435755491257, + "step": 834, + "valid_targets_mean": 5260.5, + "valid_targets_min": 1214 + }, + { + "epoch": 3.3807594936708862, + "grad_norm": 0.2724871828550331, + "learning_rate": 1.1537669672098517e-05, + "loss": 0.168, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02586859092116356, + "step": 835, + "valid_targets_mean": 5010.1, + "valid_targets_min": 1034 + }, + { + "epoch": 3.3848101265822783, + "grad_norm": 0.28626628867865767, + "learning_rate": 1.1486461098429036e-05, + "loss": 0.1749, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015445518307387829, + "step": 836, + "valid_targets_mean": 4448.1, + "valid_targets_min": 819 + }, + { + "epoch": 3.388860759493671, + "grad_norm": 0.26727985163226237, + "learning_rate": 1.1435320598789747e-05, + "loss": 0.181, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020043276250362396, + "step": 837, + "valid_targets_mean": 4645.7, + "valid_targets_min": 1255 + }, + { + "epoch": 3.3929113924050633, + "grad_norm": 0.2726651228835583, + "learning_rate": 1.1384248582098736e-05, + "loss": 0.1696, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016197284683585167, + "step": 838, + "valid_targets_mean": 3677.2, + "valid_targets_min": 1308 + }, + { + "epoch": 3.396962025316456, + "grad_norm": 0.27571953261817456, + "learning_rate": 1.1333245456726467e-05, + "loss": 0.1722, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.029869263991713524, + "step": 839, + "valid_targets_mean": 5654.1, + "valid_targets_min": 1572 + }, + { + "epoch": 3.401012658227848, + "grad_norm": 0.27600908329749335, + "learning_rate": 1.128231163049258e-05, + "loss": 0.1731, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016275284811854362, + "step": 840, + "valid_targets_mean": 4025.7, + "valid_targets_min": 1054 + }, + { + "epoch": 3.4050632911392404, + "grad_norm": 0.25713871920251963, + "learning_rate": 1.123144751066259e-05, + "loss": 0.1746, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.027817074209451675, + "step": 841, + "valid_targets_mean": 5710.4, + "valid_targets_min": 1580 + }, + { + "epoch": 3.409113924050633, + "grad_norm": 0.2825201455489712, + "learning_rate": 1.1180653503944657e-05, + "loss": 0.1723, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.027706153690814972, + "step": 842, + "valid_targets_mean": 5161.7, + "valid_targets_min": 1480 + }, + { + "epoch": 3.4131645569620255, + "grad_norm": 0.2858818538810382, + "learning_rate": 1.1129930016486287e-05, + "loss": 0.1712, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023515108972787857, + "step": 843, + "valid_targets_mean": 4839.2, + "valid_targets_min": 1786 + }, + { + "epoch": 3.4172151898734175, + "grad_norm": 0.2638128074444224, + "learning_rate": 1.107927745387114e-05, + "loss": 0.1707, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01564435288310051, + "step": 844, + "valid_targets_mean": 3269.4, + "valid_targets_min": 920 + }, + { + "epoch": 3.42126582278481, + "grad_norm": 0.2695433501417177, + "learning_rate": 1.1028696221115766e-05, + "loss": 0.1731, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03138864040374756, + "step": 845, + "valid_targets_mean": 6159.3, + "valid_targets_min": 1067 + }, + { + "epoch": 3.4253164556962026, + "grad_norm": 0.266064573836415, + "learning_rate": 1.0978186722666341e-05, + "loss": 0.1696, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026126345619559288, + "step": 846, + "valid_targets_mean": 5643.2, + "valid_targets_min": 1514 + }, + { + "epoch": 3.429367088607595, + "grad_norm": 0.2758817540755521, + "learning_rate": 1.0927749362395475e-05, + "loss": 0.1726, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015297709964215755, + "step": 847, + "valid_targets_mean": 3850.0, + "valid_targets_min": 1189 + }, + { + "epoch": 3.433417721518987, + "grad_norm": 0.26901779395475367, + "learning_rate": 1.0877384543598954e-05, + "loss": 0.1774, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021081728860735893, + "step": 848, + "valid_targets_mean": 4777.1, + "valid_targets_min": 1144 }, { - "epoch": 1.069367088607595, - "grad_norm": 0.42278867827373684, - "learning_rate": 3.8420061875234606e-05, - "loss": 0.3004, - "step": 264 + "epoch": 3.4374683544303797, + "grad_norm": 0.27724722669699176, + "learning_rate": 1.0827092668992529e-05, + "loss": 0.178, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02179073728621006, + "step": 849, + "valid_targets_mean": 5052.2, + "valid_targets_min": 1071 }, { - "epoch": 1.0734177215189873, - "grad_norm": 0.3988295654730013, - "learning_rate": 3.839787708819787e-05, - "loss": 0.3154, - "step": 265 + "epoch": 3.441518987341772, + "grad_norm": 0.2641162290763693, + "learning_rate": 1.077687414070869e-05, + "loss": 0.1687, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02662193402647972, + "step": 850, + "valid_targets_mean": 5878.2, + "valid_targets_min": 1193 }, { - "epoch": 1.0774683544303798, - "grad_norm": 0.3992434575808868, - "learning_rate": 3.8375544127188325e-05, - "loss": 0.3136, - "step": 266 + "epoch": 3.4455696202531647, + "grad_norm": 0.25669905999278697, + "learning_rate": 1.0726729360293451e-05, + "loss": 0.1698, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021715868264436722, + "step": 851, + "valid_targets_mean": 5160.4, + "valid_targets_min": 1104 + }, + { + "epoch": 3.449620253164557, + "grad_norm": 0.25970259826495146, + "learning_rate": 1.0676658728703147e-05, + "loss": 0.1628, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018533840775489807, + "step": 852, + "valid_targets_mean": 3411.4, + "valid_targets_min": 1243 + }, + { + "epoch": 3.4536708860759493, + "grad_norm": 0.25457353336282595, + "learning_rate": 1.0626662646301198e-05, + "loss": 0.1716, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014588287100195885, + "step": 853, + "valid_targets_mean": 4137.9, + "valid_targets_min": 1310 + }, + { + "epoch": 3.457721518987342, + "grad_norm": 0.24824852868665856, + "learning_rate": 1.0576741512854958e-05, + "loss": 0.1697, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017550557851791382, + "step": 854, + "valid_targets_mean": 4064.2, + "valid_targets_min": 1137 + }, + { + "epoch": 3.4617721518987343, + "grad_norm": 0.26572522596932885, + "learning_rate": 1.052689572753247e-05, + "loss": 0.1725, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.028516778722405434, + "step": 855, + "valid_targets_mean": 5678.3, + "valid_targets_min": 1583 }, { - "epoch": 1.081518987341772, - "grad_norm": 0.44220015426617215, - "learning_rate": 3.8353063172072564e-05, - "loss": 0.316, - "step": 267 + "epoch": 3.4658227848101264, + "grad_norm": 0.26457860513561665, + "learning_rate": 1.0477125688899328e-05, + "loss": 0.1692, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020979564636945724, + "step": 856, + "valid_targets_mean": 4732.6, + "valid_targets_min": 886 + }, + { + "epoch": 3.469873417721519, + "grad_norm": 0.25261090215661935, + "learning_rate": 1.0427431794915406e-05, + "loss": 0.1619, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01421372964978218, + "step": 857, + "valid_targets_mean": 2682.5, + "valid_targets_min": 1229 + }, + { + "epoch": 3.4739240506329114, + "grad_norm": 0.25156130882787403, + "learning_rate": 1.0377814442931769e-05, + "loss": 0.1707, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0210129264742136, + "step": 858, + "valid_targets_mean": 4471.9, + "valid_targets_min": 1063 + }, + { + "epoch": 3.477974683544304, + "grad_norm": 0.274113112378801, + "learning_rate": 1.0328274029687439e-05, + "loss": 0.1753, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021651387214660645, + "step": 859, + "valid_targets_mean": 4218.9, + "valid_targets_min": 772 }, { - "epoch": 1.0855696202531646, - "grad_norm": 0.4187428826568106, - "learning_rate": 3.8330434403909105e-05, - "loss": 0.3168, - "step": 268 + "epoch": 3.482025316455696, + "grad_norm": 0.26377135539628055, + "learning_rate": 1.0278810951306253e-05, + "loss": 0.1658, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02407141774892807, + "step": 860, + "valid_targets_mean": 5740.3, + "valid_targets_min": 1286 }, { - "epoch": 1.089620253164557, - "grad_norm": 0.37810889035188694, - "learning_rate": 3.8307658004946934e-05, - "loss": 0.289, - "step": 269 + "epoch": 3.4860759493670885, + "grad_norm": 0.25760012748068395, + "learning_rate": 1.0229425603293647e-05, + "loss": 0.1673, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016751490533351898, + "step": 861, + "valid_targets_mean": 3865.2, + "valid_targets_min": 1616 + }, + { + "epoch": 3.490126582278481, + "grad_norm": 0.2536224368663142, + "learning_rate": 1.0180118380533558e-05, + "loss": 0.1765, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.019702903926372528, + "step": 862, + "valid_targets_mean": 4204.1, + "valid_targets_min": 1170 + }, + { + "epoch": 3.4941772151898736, + "grad_norm": 0.2858251933318063, + "learning_rate": 1.0130889677285225e-05, + "loss": 0.1702, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030851341784000397, + "step": 863, + "valid_targets_mean": 4783.8, + "valid_targets_min": 1175 + }, + { + "epoch": 3.4982278481012656, + "grad_norm": 0.2588318742109123, + "learning_rate": 1.008173988718005e-05, + "loss": 0.1736, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01750990003347397, + "step": 864, + "valid_targets_mean": 3935.4, + "valid_targets_min": 921 + }, + { + "epoch": 3.502278481012658, + "grad_norm": 0.2761022183429568, + "learning_rate": 1.0032669403218439e-05, + "loss": 0.1706, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023192841559648514, + "step": 865, + "valid_targets_mean": 4564.4, + "valid_targets_min": 1572 }, { - "epoch": 1.0936708860759494, - "grad_norm": 0.3623707862537442, - "learning_rate": 3.8284734158624046e-05, - "loss": 0.3028, - "step": 270 + "epoch": 3.5063291139240507, + "grad_norm": 0.2769011221878136, + "learning_rate": 9.983678617766677e-06, + "loss": 0.1741, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014613568782806396, + "step": 866, + "valid_targets_mean": 3809.8, + "valid_targets_min": 1867 }, { - "epoch": 1.0977215189873417, - "grad_norm": 0.373036044593057, - "learning_rate": 3.826166304956594e-05, - "loss": 0.3024, - "step": 271 + "epoch": 3.510379746835443, + "grad_norm": 0.27145474600848807, + "learning_rate": 9.934767922553788e-06, + "loss": 0.1683, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016787836328148842, + "step": 867, + "valid_targets_mean": 3861.8, + "valid_targets_min": 980 + }, + { + "epoch": 3.5144303797468357, + "grad_norm": 0.28588768149583893, + "learning_rate": 9.885937708668371e-06, + "loss": 0.1724, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.030413717031478882, + "step": 868, + "valid_targets_mean": 5795.8, + "valid_targets_min": 1129 + }, + { + "epoch": 3.5184810126582278, + "grad_norm": 0.27730137605077443, + "learning_rate": 9.837188366555524e-06, + "loss": 0.1703, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023630110546946526, + "step": 869, + "valid_targets_mean": 4854.9, + "valid_targets_min": 1130 + }, + { + "epoch": 3.5225316455696203, + "grad_norm": 0.26416487489880813, + "learning_rate": 9.788520286013691e-06, + "loss": 0.1707, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018917936831712723, + "step": 870, + "valid_targets_mean": 3930.0, + "valid_targets_min": 1135 + }, + { + "epoch": 3.526582278481013, + "grad_norm": 0.2705474279313182, + "learning_rate": 9.739933856191537e-06, + "loss": 0.1753, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01810012012720108, + "step": 871, + "valid_targets_mean": 3565.3, + "valid_targets_min": 1006 + }, + { + "epoch": 3.530632911392405, + "grad_norm": 0.2634513861085076, + "learning_rate": 9.691429465584866e-06, + "loss": 0.1716, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.024609217420220375, + "step": 872, + "valid_targets_mean": 4994.6, + "valid_targets_min": 1070 + }, + { + "epoch": 3.5346835443037974, + "grad_norm": 0.2701987987515959, + "learning_rate": 9.643007502033492e-06, + "loss": 0.1644, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.019655343145132065, + "step": 873, + "valid_targets_mean": 4291.4, + "valid_targets_min": 1074 + }, + { + "epoch": 3.53873417721519, + "grad_norm": 0.2845364092602511, + "learning_rate": 9.594668352718152e-06, + "loss": 0.1774, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014933746308088303, + "step": 874, + "valid_targets_mean": 3257.0, + "valid_targets_min": 820 + }, + { + "epoch": 3.5427848101265824, + "grad_norm": 0.27663014137321357, + "learning_rate": 9.546412404157383e-06, + "loss": 0.1707, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01676180027425289, + "step": 875, + "valid_targets_mean": 3740.1, + "valid_targets_min": 910 + }, + { + "epoch": 3.546835443037975, + "grad_norm": 0.2869278908470468, + "learning_rate": 9.498240042204471e-06, + "loss": 0.1826, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017435681074857712, + "step": 876, + "valid_targets_mean": 3564.0, + "valid_targets_min": 942 + }, + { + "epoch": 3.550886075949367, + "grad_norm": 0.2525536913472693, + "learning_rate": 9.450151652044339e-06, + "loss": 0.1682, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02018481120467186, + "step": 877, + "valid_targets_mean": 5247.9, + "valid_targets_min": 1322 + }, + { + "epoch": 3.5549367088607595, + "grad_norm": 0.26135470090742097, + "learning_rate": 9.402147618190472e-06, + "loss": 0.1752, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.024542193859815598, + "step": 878, + "valid_targets_mean": 4611.9, + "valid_targets_min": 1269 + }, + { + "epoch": 3.558987341772152, + "grad_norm": 0.26586642166295227, + "learning_rate": 9.354228324481841e-06, + "loss": 0.1678, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026726705953478813, + "step": 879, + "valid_targets_mean": 5407.9, + "valid_targets_min": 1536 + }, + { + "epoch": 3.563037974683544, + "grad_norm": 0.27983180794676255, + "learning_rate": 9.306394154079852e-06, + "loss": 0.1763, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015666600316762924, + "step": 880, + "valid_targets_mean": 3015.9, + "valid_targets_min": 1245 + }, + { + "epoch": 3.5670886075949366, + "grad_norm": 0.2575519741288266, + "learning_rate": 9.258645489465254e-06, + "loss": 0.17, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020282434299588203, + "step": 881, + "valid_targets_mean": 3948.4, + "valid_targets_min": 1163 + }, + { + "epoch": 3.571139240506329, + "grad_norm": 0.2653082828382865, + "learning_rate": 9.21098271243509e-06, + "loss": 0.178, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01743317022919655, + "step": 882, + "valid_targets_mean": 4008.4, + "valid_targets_min": 1868 }, { - "epoch": 1.1017721518987342, - "grad_norm": 0.3579610036865689, - "learning_rate": 3.8238444863584164e-05, - "loss": 0.3063, - "step": 272 + "epoch": 3.5751898734177217, + "grad_norm": 0.27186636332643244, + "learning_rate": 9.163406204099659e-06, + "loss": 0.1708, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.019136447459459305, + "step": 883, + "valid_targets_mean": 4040.6, + "valid_targets_min": 897 }, { - "epoch": 1.1058227848101265, - "grad_norm": 0.373698605628248, - "learning_rate": 3.821507978767479e-05, - "loss": 0.3107, - "step": 273 + "epoch": 3.579240506329114, + "grad_norm": 0.257092667145981, + "learning_rate": 9.115916344879452e-06, + "loss": 0.158, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014980191364884377, + "step": 884, + "valid_targets_mean": 3076.8, + "valid_targets_min": 1085 + }, + { + "epoch": 3.5832911392405062, + "grad_norm": 0.27330926886995605, + "learning_rate": 9.068513514502126e-06, + "loss": 0.1807, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03542338311672211, + "step": 885, + "valid_targets_mean": 4332.9, + "valid_targets_min": 1017 + }, + { + "epoch": 3.5873417721518988, + "grad_norm": 0.2668971182685187, + "learning_rate": 9.021198091999439e-06, + "loss": 0.1725, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015454728156328201, + "step": 886, + "valid_targets_mean": 3425.2, + "valid_targets_min": 1040 + }, + { + "epoch": 3.5913924050632913, + "grad_norm": 0.27206850542620964, + "learning_rate": 8.973970455704251e-06, + "loss": 0.1787, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014722486957907677, + "step": 887, + "valid_targets_mean": 3363.7, + "valid_targets_min": 893 + }, + { + "epoch": 3.5954430379746833, + "grad_norm": 0.2575898216184259, + "learning_rate": 8.926830983247495e-06, + "loss": 0.1633, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009300259873270988, + "step": 888, + "valid_targets_mean": 2336.5, + "valid_targets_min": 1117 + }, + { + "epoch": 3.599493670886076, + "grad_norm": 0.28148287174224407, + "learning_rate": 8.879780051555116e-06, + "loss": 0.1656, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01779111847281456, + "step": 889, + "valid_targets_mean": 3630.5, + "valid_targets_min": 1085 + }, + { + "epoch": 3.6035443037974684, + "grad_norm": 0.26848570602365607, + "learning_rate": 8.83281803684512e-06, + "loss": 0.1776, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012078113853931427, + "step": 890, + "valid_targets_mean": 2879.1, + "valid_targets_min": 1218 + }, + { + "epoch": 3.607594936708861, + "grad_norm": 0.26625379992305637, + "learning_rate": 8.785945314624532e-06, + "loss": 0.1697, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01545766368508339, + "step": 891, + "valid_targets_mean": 3677.1, + "valid_targets_min": 1415 + }, + { + "epoch": 3.6116455696202534, + "grad_norm": 0.2796316691103621, + "learning_rate": 8.739162259686385e-06, + "loss": 0.1756, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023361319676041603, + "step": 892, + "valid_targets_mean": 4513.4, + "valid_targets_min": 1128 + }, + { + "epoch": 3.6156962025316455, + "grad_norm": 0.2732692790433263, + "learning_rate": 8.692469246106734e-06, + "loss": 0.1774, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021602444350719452, + "step": 893, + "valid_targets_mean": 4360.2, + "valid_targets_min": 996 + }, + { + "epoch": 3.619746835443038, + "grad_norm": 0.254548372810956, + "learning_rate": 8.645866647241677e-06, + "loss": 0.1716, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014873235486447811, + "step": 894, + "valid_targets_mean": 4151.7, + "valid_targets_min": 1002 + }, + { + "epoch": 3.6237974683544305, + "grad_norm": 0.25854002807195725, + "learning_rate": 8.599354835724348e-06, + "loss": 0.1719, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01778283342719078, + "step": 895, + "valid_targets_mean": 3965.2, + "valid_targets_min": 1304 }, { - "epoch": 1.109873417721519, - "grad_norm": 0.36314227762444573, - "learning_rate": 3.819156801001693e-05, - "loss": 0.2985, - "step": 274 + "epoch": 3.6278481012658226, + "grad_norm": 0.25770645960418515, + "learning_rate": 8.552934183461958e-06, + "loss": 0.1647, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016672223806381226, + "step": 896, + "valid_targets_mean": 4062.9, + "valid_targets_min": 949 + }, + { + "epoch": 3.631898734177215, + "grad_norm": 0.2760455013295443, + "learning_rate": 8.506605061632805e-06, + "loss": 0.1717, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03020300716161728, + "step": 897, + "valid_targets_mean": 5710.8, + "valid_targets_min": 1398 + }, + { + "epoch": 3.6359493670886076, + "grad_norm": 0.2601624470475137, + "learning_rate": 8.460367840683308e-06, + "loss": 0.1638, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.027632739394903183, + "step": 898, + "valid_targets_mean": 6040.9, + "valid_targets_min": 1168 + }, + { + "epoch": 3.64, + "grad_norm": 0.25673009327136503, + "learning_rate": 8.414222890325065e-06, + "loss": 0.1712, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02003858983516693, + "step": 899, + "valid_targets_mean": 4638.9, + "valid_targets_min": 1044 + }, + { + "epoch": 3.6440506329113926, + "grad_norm": 0.267767166225715, + "learning_rate": 8.368170579531855e-06, + "loss": 0.1731, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02521287277340889, + "step": 900, + "valid_targets_mean": 4739.3, + "valid_targets_min": 982 + }, + { + "epoch": 3.6481012658227847, + "grad_norm": 0.2682121961553667, + "learning_rate": 8.322211276536733e-06, + "loss": 0.1734, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023730196058750153, + "step": 901, + "valid_targets_mean": 4744.9, + "valid_targets_min": 1014 + }, + { + "epoch": 3.6521518987341772, + "grad_norm": 0.253369373219694, + "learning_rate": 8.27634534882907e-06, + "loss": 0.1686, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021989701315760612, + "step": 902, + "valid_targets_mean": 4789.8, + "valid_targets_min": 940 + }, + { + "epoch": 3.6562025316455697, + "grad_norm": 0.27301076748001013, + "learning_rate": 8.230573163151585e-06, + "loss": 0.1746, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02975517511367798, + "step": 903, + "valid_targets_mean": 4628.7, + "valid_targets_min": 929 + }, + { + "epoch": 3.660253164556962, + "grad_norm": 0.2497290882389554, + "learning_rate": 8.184895085497466e-06, + "loss": 0.1635, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01518967468291521, + "step": 904, + "valid_targets_mean": 3753.2, + "valid_targets_min": 992 + }, + { + "epoch": 3.6643037974683543, + "grad_norm": 0.2598974280536631, + "learning_rate": 8.139311481107401e-06, + "loss": 0.1602, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02183011919260025, + "step": 905, + "valid_targets_mean": 4992.8, + "valid_targets_min": 1164 + }, + { + "epoch": 3.668354430379747, + "grad_norm": 0.2643137313815488, + "learning_rate": 8.093822714466686e-06, + "loss": 0.1768, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.025400429964065552, + "step": 906, + "valid_targets_mean": 4411.9, + "valid_targets_min": 992 + }, + { + "epoch": 3.6724050632911394, + "grad_norm": 0.2653001465753879, + "learning_rate": 8.048429149302271e-06, + "loss": 0.1683, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02079555206000805, + "step": 907, + "valid_targets_mean": 4470.9, + "valid_targets_min": 1324 + }, + { + "epoch": 3.676455696202532, + "grad_norm": 0.26895487702039644, + "learning_rate": 8.003131148579902e-06, + "loss": 0.1873, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.024797216057777405, + "step": 908, + "valid_targets_mean": 5130.2, + "valid_targets_min": 946 + }, + { + "epoch": 3.680506329113924, + "grad_norm": 0.2666756575494181, + "learning_rate": 7.957929074501182e-06, + "loss": 0.167, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02008213847875595, + "step": 909, + "valid_targets_mean": 4111.8, + "valid_targets_min": 932 }, { - "epoch": 1.1139240506329113, - "grad_norm": 0.388679686159093, - "learning_rate": 3.816790971997121e-05, - "loss": 0.2904, - "step": 275 + "epoch": 3.6845569620253165, + "grad_norm": 0.2540007439992368, + "learning_rate": 7.912823288500695e-06, + "loss": 0.1675, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.025224491953849792, + "step": 910, + "valid_targets_mean": 5069.3, + "valid_targets_min": 1430 }, { - "epoch": 1.1179746835443038, - "grad_norm": 0.40794205230160135, - "learning_rate": 3.8144105108078246e-05, - "loss": 0.2998, - "step": 276 + "epoch": 3.688607594936709, + "grad_norm": 0.26073487502324005, + "learning_rate": 7.867814151243102e-06, + "loss": 0.1705, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018188446760177612, + "step": 911, + "valid_targets_mean": 4627.6, + "valid_targets_min": 1025 }, { - "epoch": 1.1220253164556961, - "grad_norm": 0.39930400463981613, - "learning_rate": 3.81201543660571e-05, - "loss": 0.3118, - "step": 277 + "epoch": 3.692658227848101, + "grad_norm": 0.2728741992837641, + "learning_rate": 7.822902022620267e-06, + "loss": 0.1703, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023914583027362823, + "step": 912, + "valid_targets_mean": 4483.4, + "valid_targets_min": 1166 }, { - "epoch": 1.1260759493670887, - "grad_norm": 0.4255774371377535, - "learning_rate": 3.809605768680377e-05, - "loss": 0.3202, - "step": 278 + "epoch": 3.6967088607594936, + "grad_norm": 0.26399975045933377, + "learning_rate": 7.778087261748377e-06, + "loss": 0.1774, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01853068359196186, + "step": 913, + "valid_targets_mean": 4006.7, + "valid_targets_min": 1153 + }, + { + "epoch": 3.700759493670886, + "grad_norm": 0.2533150709773523, + "learning_rate": 7.733370226965056e-06, + "loss": 0.1746, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01628236100077629, + "step": 914, + "valid_targets_mean": 4182.9, + "valid_targets_min": 1054 + }, + { + "epoch": 3.7048101265822786, + "grad_norm": 0.26525652186299153, + "learning_rate": 7.688751275826527e-06, + "loss": 0.1716, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020979510620236397, + "step": 915, + "valid_targets_mean": 4063.8, + "valid_targets_min": 1211 + }, + { + "epoch": 3.708860759493671, + "grad_norm": 0.25620575937396894, + "learning_rate": 7.644230765104735e-06, + "loss": 0.1615, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02342154271900654, + "step": 916, + "valid_targets_mean": 5562.9, + "valid_targets_min": 1385 + }, + { + "epoch": 3.712911392405063, + "grad_norm": 0.2725936348721527, + "learning_rate": 7.599809050784502e-06, + "loss": 0.1635, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01897498033940792, + "step": 917, + "valid_targets_mean": 4265.4, + "valid_targets_min": 1208 + }, + { + "epoch": 3.7169620253164557, + "grad_norm": 0.271718242795319, + "learning_rate": 7.555486488060661e-06, + "loss": 0.1695, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02727007493376732, + "step": 918, + "valid_targets_mean": 4823.9, + "valid_targets_min": 1198 + }, + { + "epoch": 3.721012658227848, + "grad_norm": 0.25621091668575013, + "learning_rate": 7.511263431335249e-06, + "loss": 0.1715, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017735548317432404, + "step": 919, + "valid_targets_mean": 4271.2, + "valid_targets_min": 1546 + }, + { + "epoch": 3.7250632911392403, + "grad_norm": 0.2530190312257486, + "learning_rate": 7.467140234214656e-06, + "loss": 0.1695, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01537634152919054, + "step": 920, + "valid_targets_mean": 4448.5, + "valid_targets_min": 1213 + }, + { + "epoch": 3.729113924050633, + "grad_norm": 0.2721300806889746, + "learning_rate": 7.423117249506777e-06, + "loss": 0.1735, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02418111450970173, + "step": 921, + "valid_targets_mean": 5360.5, + "valid_targets_min": 1055 }, { - "epoch": 1.130126582278481, - "grad_norm": 0.38735615562552056, - "learning_rate": 3.807181526438958e-05, - "loss": 0.309, - "step": 279 + "epoch": 3.7331645569620253, + "grad_norm": 0.2780701753257114, + "learning_rate": 7.379194829218231e-06, + "loss": 0.1636, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01901012286543846, + "step": 922, + "valid_targets_mean": 3735.7, + "valid_targets_min": 897 + }, + { + "epoch": 3.737215189873418, + "grad_norm": 0.25856564395672654, + "learning_rate": 7.335373324551518e-06, + "loss": 0.1756, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01937105692923069, + "step": 923, + "valid_targets_mean": 4076.9, + "valid_targets_min": 945 + }, + { + "epoch": 3.7412658227848103, + "grad_norm": 0.2629253903683393, + "learning_rate": 7.291653085902224e-06, + "loss": 0.1733, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012313663959503174, + "step": 924, + "valid_targets_mean": 3137.3, + "valid_targets_min": 1172 + }, + { + "epoch": 3.7453164556962024, + "grad_norm": 0.2787227707451909, + "learning_rate": 7.248034462856213e-06, + "loss": 0.1679, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.022976160049438477, + "step": 925, + "valid_targets_mean": 4435.9, + "valid_targets_min": 1428 + }, + { + "epoch": 3.749367088607595, + "grad_norm": 0.2688756768110731, + "learning_rate": 7.20451780418683e-06, + "loss": 0.1656, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02019648440182209, + "step": 926, + "valid_targets_mean": 3506.9, + "valid_targets_min": 861 }, { - "epoch": 1.1341772151898735, - "grad_norm": 0.4466492250913194, - "learning_rate": 3.8047427294059697e-05, - "loss": 0.3043, - "step": 280 + "epoch": 3.7534177215189874, + "grad_norm": 0.24455271551147317, + "learning_rate": 7.161103457852128e-06, + "loss": 0.1653, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021572839468717575, + "step": 927, + "valid_targets_mean": 5081.4, + "valid_targets_min": 1330 }, { - "epoch": 1.1382278481012658, - "grad_norm": 0.42640365815915254, - "learning_rate": 3.802289397223145e-05, - "loss": 0.3031, - "step": 281 + "epoch": 3.7574683544303795, + "grad_norm": 0.25971636467078635, + "learning_rate": 7.117791770992046e-06, + "loss": 0.169, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02314162254333496, + "step": 928, + "valid_targets_mean": 4839.4, + "valid_targets_min": 1052 + }, + { + "epoch": 3.761518987341772, + "grad_norm": 0.280975447445582, + "learning_rate": 7.074583089925688e-06, + "loss": 0.182, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01538354717195034, + "step": 929, + "valid_targets_mean": 3658.2, + "valid_targets_min": 1409 + }, + { + "epoch": 3.7655696202531646, + "grad_norm": 0.26892822036099945, + "learning_rate": 7.0314777601485155e-06, + "loss": 0.1707, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016586199402809143, + "step": 930, + "valid_targets_mean": 3859.4, + "valid_targets_min": 1343 + }, + { + "epoch": 3.769620253164557, + "grad_norm": 0.25282373989841805, + "learning_rate": 6.988476126329604e-06, + "loss": 0.1729, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.027788568288087845, + "step": 931, + "valid_targets_mean": 6460.5, + "valid_targets_min": 1253 + }, + { + "epoch": 3.7736708860759496, + "grad_norm": 0.2741741674704985, + "learning_rate": 6.945578532308857e-06, + "loss": 0.1757, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03157555311918259, + "step": 932, + "valid_targets_mean": 5928.9, + "valid_targets_min": 1063 + }, + { + "epoch": 3.7777215189873417, + "grad_norm": 0.25125778944512545, + "learning_rate": 6.902785321094301e-06, + "loss": 0.1628, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016016250476241112, + "step": 933, + "valid_targets_mean": 3399.4, + "valid_targets_min": 1410 + }, + { + "epoch": 3.781772151898734, + "grad_norm": 0.28015343257100844, + "learning_rate": 6.8600968348593115e-06, + "loss": 0.1786, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012400506064295769, + "step": 934, + "valid_targets_mean": 2638.9, + "valid_targets_min": 899 + }, + { + "epoch": 3.7858227848101267, + "grad_norm": 0.24892331479213567, + "learning_rate": 6.817513414939892e-06, + "loss": 0.1511, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013685108162462711, + "step": 935, + "valid_targets_mean": 3465.3, + "valid_targets_min": 1601 }, { - "epoch": 1.1422784810126583, - "grad_norm": 0.4409857000647316, - "learning_rate": 3.7998215496492854e-05, - "loss": 0.3055, - "step": 282 + "epoch": 3.7898734177215188, + "grad_norm": 0.2646257029306078, + "learning_rate": 6.775035401831924e-06, + "loss": 0.1681, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012793563306331635, + "step": 936, + "valid_targets_mean": 3330.2, + "valid_targets_min": 1394 }, { - "epoch": 1.1463291139240506, - "grad_norm": 0.39427645212429807, - "learning_rate": 3.797339206560096e-05, - "loss": 0.3043, - "step": 283 + "epoch": 3.7939240506329113, + "grad_norm": 0.266412647881079, + "learning_rate": 6.732663135188473e-06, + "loss": 0.166, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020142555236816406, + "step": 937, + "valid_targets_mean": 4423.8, + "valid_targets_min": 1298 }, { - "epoch": 1.150379746835443, - "grad_norm": 0.48099247250782884, - "learning_rate": 3.794842387948027e-05, - "loss": 0.3201, - "step": 284 + "epoch": 3.797974683544304, + "grad_norm": 0.26291465669937136, + "learning_rate": 6.690396953817055e-06, + "loss": 0.1711, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013688749633729458, + "step": 938, + "valid_targets_mean": 3101.6, + "valid_targets_min": 1313 + }, + { + "epoch": 3.8020253164556963, + "grad_norm": 0.26546325021532996, + "learning_rate": 6.648237195676927e-06, + "loss": 0.171, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016537221148610115, + "step": 939, + "valid_targets_mean": 3412.6, + "valid_targets_min": 1393 + }, + { + "epoch": 3.806075949367089, + "grad_norm": 0.2567377674015171, + "learning_rate": 6.606184197876395e-06, + "loss": 0.1632, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02425518073141575, + "step": 940, + "valid_targets_mean": 4925.3, + "valid_targets_min": 1084 + }, + { + "epoch": 3.810126582278481, + "grad_norm": 0.2616122882966271, + "learning_rate": 6.564238296670109e-06, + "loss": 0.1613, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0212604571133852, + "step": 941, + "valid_targets_mean": 4302.8, + "valid_targets_min": 1324 + }, + { + "epoch": 3.8141772151898734, + "grad_norm": 0.259336042662043, + "learning_rate": 6.52239982745638e-06, + "loss": 0.1636, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.025171024724841118, + "step": 942, + "valid_targets_mean": 5372.2, + "valid_targets_min": 1588 + }, + { + "epoch": 3.818227848101266, + "grad_norm": 0.26152603888356707, + "learning_rate": 6.480669124774484e-06, + "loss": 0.1749, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02495972439646721, + "step": 943, + "valid_targets_mean": 4591.2, + "valid_targets_min": 958 + }, + { + "epoch": 3.822278481012658, + "grad_norm": 0.25718843187745966, + "learning_rate": 6.439046522302013e-06, + "loss": 0.1679, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017802279442548752, + "step": 944, + "valid_targets_mean": 4441.4, + "valid_targets_min": 1063 + }, + { + "epoch": 3.8263291139240505, + "grad_norm": 0.24864279505089587, + "learning_rate": 6.397532352852191e-06, + "loss": 0.1694, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017683859914541245, + "step": 945, + "valid_targets_mean": 4570.5, + "valid_targets_min": 973 + }, + { + "epoch": 3.830379746835443, + "grad_norm": 0.24182498223350302, + "learning_rate": 6.356126948371202e-06, + "loss": 0.1662, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0314711257815361, + "step": 946, + "valid_targets_mean": 7331.1, + "valid_targets_min": 834 + }, + { + "epoch": 3.8344303797468355, + "grad_norm": 0.26373174301768926, + "learning_rate": 6.314830639935559e-06, + "loss": 0.163, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.019661404192447662, + "step": 947, + "valid_targets_mean": 3925.5, + "valid_targets_min": 929 }, { - "epoch": 1.1544303797468354, - "grad_norm": 0.4683806591681051, - "learning_rate": 3.7923311139221114e-05, - "loss": 0.3194, - "step": 285 + "epoch": 3.838481012658228, + "grad_norm": 0.25151186493467503, + "learning_rate": 6.273643757749448e-06, + "loss": 0.1596, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.019310811534523964, + "step": 948, + "valid_targets_mean": 4121.4, + "valid_targets_min": 1296 + }, + { + "epoch": 3.84253164556962, + "grad_norm": 0.2525157270897604, + "learning_rate": 6.2325666311420856e-06, + "loss": 0.1795, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02261164039373398, + "step": 949, + "valid_targets_mean": 5320.1, + "valid_targets_min": 1075 + }, + { + "epoch": 3.8465822784810126, + "grad_norm": 0.2514022583146415, + "learning_rate": 6.1915995885650675e-06, + "loss": 0.1758, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02208195999264717, + "step": 950, + "valid_targets_mean": 4716.1, + "valid_targets_min": 972 + }, + { + "epoch": 3.850632911392405, + "grad_norm": 0.25474420781507917, + "learning_rate": 6.150742957589784e-06, + "loss": 0.1708, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.026663821190595627, + "step": 951, + "valid_targets_mean": 5850.9, + "valid_targets_min": 1609 + }, + { + "epoch": 3.8546835443037972, + "grad_norm": 0.25778279169226703, + "learning_rate": 6.109997064904774e-06, + "loss": 0.1702, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02077229507267475, + "step": 952, + "valid_targets_mean": 4348.4, + "valid_targets_min": 1185 + }, + { + "epoch": 3.8587341772151897, + "grad_norm": 0.26348476292614187, + "learning_rate": 6.069362236313092e-06, + "loss": 0.1659, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01722509041428566, + "step": 953, + "valid_targets_mean": 3411.2, + "valid_targets_min": 998 + }, + { + "epoch": 3.8627848101265823, + "grad_norm": 0.24558839076433475, + "learning_rate": 6.0288387967297565e-06, + "loss": 0.1672, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021953541785478592, + "step": 954, + "valid_targets_mean": 4563.6, + "valid_targets_min": 1152 + }, + { + "epoch": 3.8668354430379748, + "grad_norm": 0.2562829506926769, + "learning_rate": 5.988427070179108e-06, + "loss": 0.1705, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016076896339654922, + "step": 955, + "valid_targets_mean": 3906.6, + "valid_targets_min": 1348 + }, + { + "epoch": 3.8708860759493673, + "grad_norm": 0.26383340936593336, + "learning_rate": 5.948127379792234e-06, + "loss": 0.1723, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020764939486980438, + "step": 956, + "valid_targets_mean": 4350.1, + "valid_targets_min": 1420 + }, + { + "epoch": 3.8749367088607594, + "grad_norm": 0.2712332484036904, + "learning_rate": 5.90794004780439e-06, + "loss": 0.1686, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.022838331758975983, + "step": 957, + "valid_targets_mean": 4404.2, + "valid_targets_min": 1378 + }, + { + "epoch": 3.878987341772152, + "grad_norm": 0.2622111952679663, + "learning_rate": 5.867865395552405e-06, + "loss": 0.163, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02169284224510193, + "step": 958, + "valid_targets_mean": 3874.9, + "valid_targets_min": 1088 + }, + { + "epoch": 3.8830379746835444, + "grad_norm": 0.25352172374047266, + "learning_rate": 5.8279037434721364e-06, + "loss": 0.1682, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.019588781520724297, + "step": 959, + "valid_targets_mean": 4924.4, + "valid_targets_min": 1371 + }, + { + "epoch": 3.8870886075949365, + "grad_norm": 0.2544134305559441, + "learning_rate": 5.788055411095892e-06, + "loss": 0.16, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.022651158273220062, + "step": 960, + "valid_targets_mean": 5013.6, + "valid_targets_min": 1538 + }, + { + "epoch": 3.891139240506329, + "grad_norm": 0.2598511789733668, + "learning_rate": 5.7483207170498625e-06, + "loss": 0.1767, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018454328179359436, + "step": 961, + "valid_targets_mean": 3996.5, + "valid_targets_min": 1382 + }, + { + "epoch": 3.8951898734177215, + "grad_norm": 0.2777136565674028, + "learning_rate": 5.708699979051613e-06, + "loss": 0.1695, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01723894290626049, + "step": 962, + "valid_targets_mean": 3204.9, + "valid_targets_min": 979 + }, + { + "epoch": 3.899240506329114, + "grad_norm": 0.24515877257866994, + "learning_rate": 5.66919351390751e-06, + "loss": 0.1708, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018098827451467514, + "step": 963, + "valid_targets_mean": 5105.1, + "valid_targets_min": 1065 }, { - "epoch": 1.158481012658228, - "grad_norm": 0.3718883562363268, - "learning_rate": 3.7898054047078054e-05, - "loss": 0.2945, - "step": 286 + "epoch": 3.9032911392405065, + "grad_norm": 0.2609791537507237, + "learning_rate": 5.629801637510188e-06, + "loss": 0.1666, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009696192108094692, + "step": 964, + "valid_targets_mean": 2726.9, + "valid_targets_min": 1328 + }, + { + "epoch": 3.9073417721518986, + "grad_norm": 0.2801700145562046, + "learning_rate": 5.590524664836043e-06, + "loss": 0.1739, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03406285494565964, + "step": 965, + "valid_targets_mean": 6347.1, + "valid_targets_min": 1495 + }, + { + "epoch": 3.911392405063291, + "grad_norm": 0.2543961381253017, + "learning_rate": 5.551362909942708e-06, + "loss": 0.1662, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014446750283241272, + "step": 966, + "valid_targets_mean": 3029.8, + "valid_targets_min": 868 + }, + { + "epoch": 3.9154430379746836, + "grad_norm": 0.2604047583655427, + "learning_rate": 5.51231668596653e-06, + "loss": 0.1817, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020005282014608383, + "step": 967, + "valid_targets_mean": 4006.3, + "valid_targets_min": 1141 + }, + { + "epoch": 3.9194936708860757, + "grad_norm": 0.2569034631060701, + "learning_rate": 5.47338630512007e-06, + "loss": 0.1712, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013550623320043087, + "step": 968, + "valid_targets_mean": 3199.0, + "valid_targets_min": 956 + }, + { + "epoch": 3.923544303797468, + "grad_norm": 0.26580168189369774, + "learning_rate": 5.434572078689617e-06, + "loss": 0.1656, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015013760887086391, + "step": 969, + "valid_targets_mean": 3526.7, + "valid_targets_min": 1301 }, { - "epoch": 1.1625316455696202, - "grad_norm": 0.48855101618719643, - "learning_rate": 3.787265280646825e-05, - "loss": 0.3074, - "step": 287 + "epoch": 3.9275949367088607, + "grad_norm": 0.2344212801278269, + "learning_rate": 5.395874317032694e-06, + "loss": 0.1664, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017263196408748627, + "step": 970, + "valid_targets_mean": 4849.8, + "valid_targets_min": 929 }, { - "epoch": 1.1665822784810127, - "grad_norm": 0.488001904655089, - "learning_rate": 3.7847107621969786e-05, - "loss": 0.3098, - "step": 288 + "epoch": 3.9316455696202532, + "grad_norm": 0.2575740382944395, + "learning_rate": 5.357293329575573e-06, + "loss": 0.1534, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.024109669029712677, + "step": 971, + "valid_targets_mean": 5135.2, + "valid_targets_min": 1642 + }, + { + "epoch": 3.9356962025316458, + "grad_norm": 0.27920946943071534, + "learning_rate": 5.318829424810797e-06, + "loss": 0.1716, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017383450642228127, + "step": 972, + "valid_targets_mean": 4036.1, + "valid_targets_min": 1577 + }, + { + "epoch": 3.939746835443038, + "grad_norm": 0.2674779028295388, + "learning_rate": 5.280482910294728e-06, + "loss": 0.1731, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014004931785166264, + "step": 973, + "valid_targets_mean": 3164.8, + "valid_targets_min": 865 + }, + { + "epoch": 3.9437974683544303, + "grad_norm": 0.2558046574136687, + "learning_rate": 5.242254092645078e-06, + "loss": 0.1657, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021920736879110336, + "step": 974, + "valid_targets_mean": 4416.2, + "valid_targets_min": 1186 + }, + { + "epoch": 3.947848101265823, + "grad_norm": 0.27037859086507426, + "learning_rate": 5.2041432775384445e-06, + "loss": 0.1728, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.03166478872299194, + "step": 975, + "valid_targets_mean": 5821.2, + "valid_targets_min": 915 + }, + { + "epoch": 3.951898734177215, + "grad_norm": 0.2568070232064629, + "learning_rate": 5.166150769707892e-06, + "loss": 0.1618, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01807749643921852, + "step": 976, + "valid_targets_mean": 4275.6, + "valid_targets_min": 1151 + }, + { + "epoch": 3.9559493670886074, + "grad_norm": 0.2610467303453652, + "learning_rate": 5.128276872940503e-06, + "loss": 0.1585, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02191128209233284, + "step": 977, + "valid_targets_mean": 4305.9, + "valid_targets_min": 1243 }, { - "epoch": 1.170632911392405, - "grad_norm": 0.41149185666387744, - "learning_rate": 3.7821418699320064e-05, - "loss": 0.3018, - "step": 289 + "epoch": 3.96, + "grad_norm": 0.24641444939717588, + "learning_rate": 5.090521890074953e-06, + "loss": 0.1715, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02309935726225376, + "step": 978, + "valid_targets_mean": 4733.2, + "valid_targets_min": 1227 + }, + { + "epoch": 3.9640506329113925, + "grad_norm": 0.25286992119518464, + "learning_rate": 5.0528861229990675e-06, + "loss": 0.1695, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01703287474811077, + "step": 979, + "valid_targets_mean": 4090.1, + "valid_targets_min": 943 + }, + { + "epoch": 3.968101265822785, + "grad_norm": 0.2515046419810229, + "learning_rate": 5.015369872647442e-06, + "loss": 0.1697, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.023712750524282455, + "step": 980, + "valid_targets_mean": 4041.8, + "valid_targets_min": 1425 + }, + { + "epoch": 3.972151898734177, + "grad_norm": 0.24882079439249843, + "learning_rate": 4.9779734389990195e-06, + "loss": 0.1572, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02589654177427292, + "step": 981, + "valid_targets_mean": 5828.3, + "valid_targets_min": 1577 }, { - "epoch": 1.1746835443037975, - "grad_norm": 0.5049317655930331, - "learning_rate": 3.7795586245414145e-05, - "loss": 0.3198, - "step": 290 + "epoch": 3.9762025316455696, + "grad_norm": 0.25927214688896627, + "learning_rate": 4.940697121074675e-06, + "loss": 0.1655, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02023175358772278, + "step": 982, + "valid_targets_mean": 4153.8, + "valid_targets_min": 1031 }, { - "epoch": 1.1787341772151898, - "grad_norm": 0.3638509943809286, - "learning_rate": 3.776961046830306e-05, - "loss": 0.3128, - "step": 291 + "epoch": 3.980253164556962, + "grad_norm": 0.2578774726920398, + "learning_rate": 4.903541216934864e-06, + "loss": 0.1658, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.019639194011688232, + "step": 983, + "valid_targets_mean": 4131.6, + "valid_targets_min": 869 + }, + { + "epoch": 3.984303797468354, + "grad_norm": 0.25118875230741106, + "learning_rate": 4.8665060236772e-06, + "loss": 0.162, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014539085328578949, + "step": 984, + "valid_targets_mean": 4011.5, + "valid_targets_min": 1210 + }, + { + "epoch": 3.9883544303797467, + "grad_norm": 0.2498271509027046, + "learning_rate": 4.829591837434102e-06, + "loss": 0.1723, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018505923449993134, + "step": 985, + "valid_targets_mean": 4007.4, + "valid_targets_min": 1684 + }, + { + "epoch": 3.992405063291139, + "grad_norm": 0.2569733312313395, + "learning_rate": 4.792798953370424e-06, + "loss": 0.1796, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.027533505111932755, + "step": 986, + "valid_targets_mean": 5823.2, + "valid_targets_min": 1688 + }, + { + "epoch": 3.9964556962025317, + "grad_norm": 0.25646292717570135, + "learning_rate": 4.756127665681083e-06, + "loss": 0.1673, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021060016006231308, + "step": 987, + "valid_targets_mean": 4374.2, + "valid_targets_min": 971 }, { - "epoch": 1.1827848101265823, - "grad_norm": 0.512119104956546, - "learning_rate": 3.774349157719215e-05, - "loss": 0.3047, - "step": 292 + "epoch": 4.0, + "grad_norm": 0.2771544532900968, + "learning_rate": 4.719578267588725e-06, + "loss": 0.165, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021999431774020195, + "step": 988, + "valid_targets_mean": 4322.3, + "valid_targets_min": 1295 }, { - "epoch": 1.1868354430379746, - "grad_norm": 0.41507224029396195, - "learning_rate": 3.7717229782439365e-05, - "loss": 0.2938, - "step": 293 + "epoch": 4.004050632911392, + "grad_norm": 0.4003732826881182, + "learning_rate": 4.6831510513413525e-06, + "loss": 0.1201, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014306396245956421, + "step": 989, + "valid_targets_mean": 3664.5, + "valid_targets_min": 1216 }, { - "epoch": 1.1908860759493671, - "grad_norm": 0.4458602180586056, - "learning_rate": 3.769082529555359e-05, - "loss": 0.3143, - "step": 294 + "epoch": 4.008101265822785, + "grad_norm": 0.3655726493503762, + "learning_rate": 4.6468463082100176e-06, + "loss": 0.1199, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012987561523914337, + "step": 990, + "valid_targets_mean": 4540.7, + "valid_targets_min": 765 + }, + { + "epoch": 4.012151898734177, + "grad_norm": 0.30964833471583064, + "learning_rate": 4.610664328486485e-06, + "loss": 0.1104, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0126019436866045, + "step": 991, + "valid_targets_mean": 4332.8, + "valid_targets_min": 838 + }, + { + "epoch": 4.01620253164557, + "grad_norm": 0.250996175520195, + "learning_rate": 4.5746054014809e-06, + "loss": 0.1135, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014090234413743019, + "step": 992, + "valid_targets_mean": 4519.7, + "valid_targets_min": 1248 + }, + { + "epoch": 4.020253164556962, + "grad_norm": 0.2892432889212748, + "learning_rate": 4.5386698155194766e-06, + "loss": 0.1146, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013059249147772789, + "step": 993, + "valid_targets_mean": 4145.2, + "valid_targets_min": 1426 + }, + { + "epoch": 4.024303797468354, + "grad_norm": 0.34040346346515926, + "learning_rate": 4.502857857942204e-06, + "loss": 0.1099, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013071644119918346, + "step": 994, + "valid_targets_mean": 4365.7, + "valid_targets_min": 1116 }, { - "epoch": 1.1949367088607594, - "grad_norm": 0.45200586849839086, - "learning_rate": 3.766427832919294e-05, - "loss": 0.2907, - "step": 295 + "epoch": 4.028354430379747, + "grad_norm": 0.43935845584745803, + "learning_rate": 4.467169815100549e-06, + "loss": 0.1175, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01606411673128605, + "step": 995, + "valid_targets_mean": 5574.1, + "valid_targets_min": 1533 }, { - "epoch": 1.198987341772152, - "grad_norm": 0.4185441323047734, - "learning_rate": 3.7637589097163024e-05, - "loss": 0.2996, - "step": 296 + "epoch": 4.032405063291139, + "grad_norm": 0.38111330979551494, + "learning_rate": 4.431605972355137e-06, + "loss": 0.1166, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014245737344026566, + "step": 996, + "valid_targets_mean": 4346.1, + "valid_targets_min": 1263 }, { - "epoch": 1.2030379746835442, - "grad_norm": 0.47620238078848565, - "learning_rate": 3.761075781441526e-05, - "loss": 0.3204, - "step": 297 + "epoch": 4.036455696202531, + "grad_norm": 0.30287401074217873, + "learning_rate": 4.396166614073516e-06, + "loss": 0.1124, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013161606155335903, + "step": 997, + "valid_targets_mean": 4571.4, + "valid_targets_min": 1194 + }, + { + "epoch": 4.040506329113924, + "grad_norm": 0.2798009529690627, + "learning_rate": 4.360852023627846e-06, + "loss": 0.1081, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016295261681079865, + "step": 998, + "valid_targets_mean": 5411.4, + "valid_targets_min": 1057 + }, + { + "epoch": 4.044556962025316, + "grad_norm": 0.2922199601998903, + "learning_rate": 4.325662483392656e-06, + "loss": 0.1153, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.010619858279824257, + "step": 999, + "valid_targets_mean": 3356.9, + "valid_targets_min": 1193 + }, + { + "epoch": 4.048607594936709, + "grad_norm": 0.2787040016448312, + "learning_rate": 4.29059827474257e-06, + "loss": 0.116, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021953728049993515, + "step": 1000, + "valid_targets_mean": 6415.1, + "valid_targets_min": 1294 + }, + { + "epoch": 4.052658227848101, + "grad_norm": 0.28719331022465855, + "learning_rate": 4.255659678050063e-06, + "loss": 0.1144, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015153765678405762, + "step": 1001, + "valid_targets_mean": 4424.2, + "valid_targets_min": 1264 + }, + { + "epoch": 4.056708860759493, + "grad_norm": 0.2885840564817545, + "learning_rate": 4.22084697268323e-06, + "loss": 0.1147, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016420746222138405, + "step": 1002, + "valid_targets_mean": 4203.6, + "valid_targets_min": 1221 + }, + { + "epoch": 4.060759493670886, + "grad_norm": 0.2889212669563016, + "learning_rate": 4.186160437003535e-06, + "loss": 0.1136, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.007882167585194111, + "step": 1003, + "valid_targets_mean": 2971.5, + "valid_targets_min": 1138 + }, + { + "epoch": 4.0648101265822785, + "grad_norm": 0.2708558928023269, + "learning_rate": 4.151600348363582e-06, + "loss": 0.1154, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011313255876302719, + "step": 1004, + "valid_targets_mean": 4321.1, + "valid_targets_min": 1264 + }, + { + "epoch": 4.0688607594936705, + "grad_norm": 0.2589101403089754, + "learning_rate": 4.11716698310493e-06, + "loss": 0.1067, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012739719823002815, + "step": 1005, + "valid_targets_mean": 4941.8, + "valid_targets_min": 1269 + }, + { + "epoch": 4.0729113924050635, + "grad_norm": 0.2795822652210067, + "learning_rate": 4.082860616555848e-06, + "loss": 0.1143, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011751806363463402, + "step": 1006, + "valid_targets_mean": 3975.0, + "valid_targets_min": 1238 + }, + { + "epoch": 4.076962025316456, + "grad_norm": 0.27901499487691106, + "learning_rate": 4.048681523029125e-06, + "loss": 0.1046, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013513721525669098, + "step": 1007, + "valid_targets_mean": 4067.1, + "valid_targets_min": 1093 + }, + { + "epoch": 4.0810126582278485, + "grad_norm": 0.28199971187759815, + "learning_rate": 4.014629975819882e-06, + "loss": 0.1167, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013382020406425, + "step": 1008, + "valid_targets_mean": 4960.6, + "valid_targets_min": 809 + }, + { + "epoch": 4.085063291139241, + "grad_norm": 0.2739291802210153, + "learning_rate": 3.9807062472033855e-06, + "loss": 0.1153, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011161049827933311, + "step": 1009, + "valid_targets_mean": 3603.4, + "valid_targets_min": 1360 + }, + { + "epoch": 4.089113924050633, + "grad_norm": 0.2608212672568236, + "learning_rate": 3.94691060843287e-06, + "loss": 0.1155, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02048046886920929, + "step": 1010, + "valid_targets_mean": 6277.9, + "valid_targets_min": 1750 }, { - "epoch": 1.2070886075949367, - "grad_norm": 0.37681597816293283, - "learning_rate": 3.75837846970451e-05, - "loss": 0.295, - "step": 298 + "epoch": 4.093164556962026, + "grad_norm": 0.2529424848219501, + "learning_rate": 3.913243329737353e-06, + "loss": 0.1131, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.00706963986158371, + "step": 1011, + "valid_targets_mean": 3304.1, + "valid_targets_min": 1428 }, { - "epoch": 1.211139240506329, - "grad_norm": 0.44045998419606053, - "learning_rate": 3.755666996229032e-05, - "loss": 0.3156, - "step": 299 + "epoch": 4.097215189873418, + "grad_norm": 0.26109890093820065, + "learning_rate": 3.879704680319503e-06, + "loss": 0.1142, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.007938157767057419, + "step": 1012, + "valid_targets_mean": 3203.9, + "valid_targets_min": 1186 }, { - "epoch": 1.2151898734177216, - "grad_norm": 0.42622523087891967, - "learning_rate": 3.752941382852927e-05, - "loss": 0.3056, - "step": 300 + "epoch": 4.10126582278481, + "grad_norm": 0.25760137757483825, + "learning_rate": 3.846294928353471e-06, + "loss": 0.1086, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01943141408264637, + "step": 1013, + "valid_targets_mean": 5626.9, + "valid_targets_min": 1068 + }, + { + "epoch": 4.105316455696203, + "grad_norm": 0.2479626910788689, + "learning_rate": 3.8130143409827345e-06, + "loss": 0.1112, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013259506784379482, + "step": 1014, + "valid_targets_mean": 4291.4, + "valid_targets_min": 1322 + }, + { + "epoch": 4.109367088607595, + "grad_norm": 0.2569955392936143, + "learning_rate": 3.7798631843179887e-06, + "loss": 0.1071, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.019261609762907028, + "step": 1015, + "valid_targets_mean": 4930.9, + "valid_targets_min": 828 + }, + { + "epoch": 4.113417721518988, + "grad_norm": 0.25253897584899043, + "learning_rate": 3.7468417234349973e-06, + "loss": 0.1171, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018999673426151276, + "step": 1016, + "valid_targets_mean": 5932.4, + "valid_targets_min": 1791 + }, + { + "epoch": 4.11746835443038, + "grad_norm": 0.2611560587823719, + "learning_rate": 3.7139502223724823e-06, + "loss": 0.1104, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01204339787364006, + "step": 1017, + "valid_targets_mean": 3853.3, + "valid_targets_min": 1330 + }, + { + "epoch": 4.121518987341772, + "grad_norm": 0.2586414238324103, + "learning_rate": 3.6811889441300095e-06, + "loss": 0.1065, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015708964318037033, + "step": 1018, + "valid_targets_mean": 4885.8, + "valid_targets_min": 994 + }, + { + "epoch": 4.125569620253165, + "grad_norm": 0.25967544297221273, + "learning_rate": 3.648558150665886e-06, + "loss": 0.1186, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01149682980030775, + "step": 1019, + "valid_targets_mean": 3851.9, + "valid_targets_min": 1121 + }, + { + "epoch": 4.129620253164557, + "grad_norm": 0.25572555469489316, + "learning_rate": 3.616058102895075e-06, + "loss": 0.1126, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016392933204770088, + "step": 1020, + "valid_targets_mean": 4871.8, + "valid_targets_min": 1100 + }, + { + "epoch": 4.133670886075949, + "grad_norm": 0.2441110532126236, + "learning_rate": 3.583689060687079e-06, + "loss": 0.1157, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012376191094517708, + "step": 1021, + "valid_targets_mean": 4524.1, + "valid_targets_min": 1028 + }, + { + "epoch": 4.137721518987342, + "grad_norm": 0.24398646370891114, + "learning_rate": 3.5514512828639023e-06, + "loss": 0.1091, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018081197515130043, + "step": 1022, + "valid_targets_mean": 6087.4, + "valid_targets_min": 1117 + }, + { + "epoch": 4.141772151898734, + "grad_norm": 0.2546021016376095, + "learning_rate": 3.51934502719796e-06, + "loss": 0.1039, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01941712014377117, + "step": 1023, + "valid_targets_mean": 5920.0, + "valid_targets_min": 1266 + }, + { + "epoch": 4.145822784810127, + "grad_norm": 0.25101564084424466, + "learning_rate": 3.4873705504100185e-06, + "loss": 0.1031, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011633863672614098, + "step": 1024, + "valid_targets_mean": 3833.2, + "valid_targets_min": 1357 + }, + { + "epoch": 4.149873417721519, + "grad_norm": 0.24006292387114583, + "learning_rate": 3.4555281081671345e-06, + "loss": 0.1106, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01572991907596588, + "step": 1025, + "valid_targets_mean": 5546.1, + "valid_targets_min": 1024 }, { - "epoch": 1.2192405063291138, - "grad_norm": 0.39145017259592085, - "learning_rate": 3.7502016515279115e-05, - "loss": 0.32, - "step": 301 + "epoch": 4.153924050632911, + "grad_norm": 0.24629627209906968, + "learning_rate": 3.4238179550806326e-06, + "loss": 0.108, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012446759268641472, + "step": 1026, + "valid_targets_mean": 4453.8, + "valid_targets_min": 975 }, { - "epoch": 1.2232911392405064, - "grad_norm": 0.38906613244202576, - "learning_rate": 3.7474478243194043e-05, - "loss": 0.3301, - "step": 302 + "epoch": 4.157974683544304, + "grad_norm": 0.25615612600541154, + "learning_rate": 3.3922403447040497e-06, + "loss": 0.108, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014301635324954987, + "step": 1027, + "valid_targets_mean": 4307.7, + "valid_targets_min": 1391 + }, + { + "epoch": 4.162025316455696, + "grad_norm": 0.23812708306917799, + "learning_rate": 3.3607955295311247e-06, + "loss": 0.101, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011930042877793312, + "step": 1028, + "valid_targets_mean": 4049.1, + "valid_targets_min": 1088 + }, + { + "epoch": 4.166075949367088, + "grad_norm": 0.24105681140422938, + "learning_rate": 3.32948376099375e-06, + "loss": 0.1081, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012245169840753078, + "step": 1029, + "valid_targets_mean": 4256.4, + "valid_targets_min": 902 + }, + { + "epoch": 4.170126582278481, + "grad_norm": 0.2603579396276263, + "learning_rate": 3.2983052894599976e-06, + "loss": 0.1119, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009018047712743282, + "step": 1030, + "valid_targets_mean": 2847.9, + "valid_targets_min": 1455 }, { - "epoch": 1.2273417721518987, - "grad_norm": 0.3784912113621531, - "learning_rate": 3.744679923406351e-05, - "loss": 0.3111, - "step": 303 + "epoch": 4.174177215189873, + "grad_norm": 0.24094462126263408, + "learning_rate": 3.2672603642320945e-06, + "loss": 0.1131, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009460234083235264, + "step": 1031, + "valid_targets_mean": 3995.7, + "valid_targets_min": 1381 }, { - "epoch": 1.2313924050632912, - "grad_norm": 0.4216892242894796, - "learning_rate": 3.741897971081043e-05, - "loss": 0.3203, - "step": 304 + "epoch": 4.178227848101266, + "grad_norm": 0.2539272303216648, + "learning_rate": 3.236349233544438e-06, + "loss": 0.1101, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.00956416130065918, + "step": 1032, + "valid_targets_mean": 3387.9, + "valid_targets_min": 1578 + }, + { + "epoch": 4.182278481012658, + "grad_norm": 0.24607206464657877, + "learning_rate": 3.205572144561604e-06, + "loss": 0.1109, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009433125145733356, + "step": 1033, + "valid_targets_mean": 3179.5, + "valid_targets_min": 1212 + }, + { + "epoch": 4.18632911392405, + "grad_norm": 0.2369043979584509, + "learning_rate": 3.174929343376374e-06, + "loss": 0.1104, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01051153615117073, + "step": 1034, + "valid_targets_mean": 3981.8, + "valid_targets_min": 1298 + }, + { + "epoch": 4.190379746835443, + "grad_norm": 0.24736908966042873, + "learning_rate": 3.1444210750077777e-06, + "loss": 0.1088, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012088462710380554, + "step": 1035, + "valid_targets_mean": 4654.1, + "valid_targets_min": 1300 + }, + { + "epoch": 4.194430379746835, + "grad_norm": 0.24641644567177923, + "learning_rate": 3.1140475833991046e-06, + "loss": 0.1111, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018200930207967758, + "step": 1036, + "valid_targets_mean": 5480.0, + "valid_targets_min": 1518 + }, + { + "epoch": 4.1984810126582275, + "grad_norm": 0.2473167962312305, + "learning_rate": 3.0838091114159895e-06, + "loss": 0.1113, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014702276326715946, + "step": 1037, + "valid_targets_mean": 4763.0, + "valid_targets_min": 914 + }, + { + "epoch": 4.2025316455696204, + "grad_norm": 0.2380808023314683, + "learning_rate": 3.0537059008444524e-06, + "loss": 0.1131, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012161606922745705, + "step": 1038, + "valid_targets_mean": 4509.5, + "valid_targets_min": 1325 }, { - "epoch": 1.2354430379746835, - "grad_norm": 0.3768836360285952, - "learning_rate": 3.739101989748946e-05, - "loss": 0.3102, - "step": 305 + "epoch": 4.2065822784810125, + "grad_norm": 0.2371915388592596, + "learning_rate": 3.0237381923889563e-06, + "loss": 0.1019, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0148830721154809, + "step": 1039, + "valid_targets_mean": 5838.4, + "valid_targets_min": 1214 + }, + { + "epoch": 4.2106329113924055, + "grad_norm": 0.2459439242197879, + "learning_rate": 2.9939062256704997e-06, + "loss": 0.1104, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014685314148664474, + "step": 1040, + "valid_targets_mean": 4390.6, + "valid_targets_min": 1164 + }, + { + "epoch": 4.2146835443037975, + "grad_norm": 0.23226846355518324, + "learning_rate": 2.964210239224696e-06, + "loss": 0.1131, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01576303504407406, + "step": 1041, + "valid_targets_mean": 5250.2, + "valid_targets_min": 1423 + }, + { + "epoch": 4.21873417721519, + "grad_norm": 0.2426028729765493, + "learning_rate": 2.934650470499869e-06, + "loss": 0.1036, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.02107350155711174, + "step": 1042, + "valid_targets_mean": 6033.2, + "valid_targets_min": 1302 + }, + { + "epoch": 4.222784810126583, + "grad_norm": 0.2387287478909716, + "learning_rate": 2.905227155855128e-06, + "loss": 0.1163, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011402295902371407, + "step": 1043, + "valid_targets_mean": 3917.2, + "valid_targets_min": 1283 + }, + { + "epoch": 4.226835443037975, + "grad_norm": 0.24908362936328057, + "learning_rate": 2.875940530558519e-06, + "loss": 0.112, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016240276396274567, + "step": 1044, + "valid_targets_mean": 5248.2, + "valid_targets_min": 1146 + }, + { + "epoch": 4.230886075949367, + "grad_norm": 0.23423752746429335, + "learning_rate": 2.8467908287851154e-06, + "loss": 0.1082, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012546324171125889, + "step": 1045, + "valid_targets_mean": 5506.6, + "valid_targets_min": 1132 + }, + { + "epoch": 4.23493670886076, + "grad_norm": 0.2520296188756494, + "learning_rate": 2.8177782836151603e-06, + "loss": 0.1073, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014122463762760162, + "step": 1046, + "valid_targets_mean": 4423.8, + "valid_targets_min": 1436 + }, + { + "epoch": 4.238987341772152, + "grad_norm": 0.23557866922129664, + "learning_rate": 2.7889031270321874e-06, + "loss": 0.1092, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017405740916728973, + "step": 1047, + "valid_targets_mean": 5047.8, + "valid_targets_min": 1672 + }, + { + "epoch": 4.243037974683545, + "grad_norm": 0.24293451283336756, + "learning_rate": 2.760165589921184e-06, + "loss": 0.1042, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01629508100450039, + "step": 1048, + "valid_targets_mean": 5036.8, + "valid_targets_min": 1141 + }, + { + "epoch": 4.247088607594937, + "grad_norm": 0.23988649700489975, + "learning_rate": 2.7315659020667305e-06, + "loss": 0.1097, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.010395625606179237, + "step": 1049, + "valid_targets_mean": 4051.7, + "valid_targets_min": 1243 + }, + { + "epoch": 4.251139240506329, + "grad_norm": 0.24862472182244583, + "learning_rate": 2.703104292151164e-06, + "loss": 0.1106, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01229317206889391, + "step": 1050, + "valid_targets_mean": 3529.8, + "valid_targets_min": 1096 + }, + { + "epoch": 4.255189873417722, + "grad_norm": 0.2349536501847763, + "learning_rate": 2.6747809877527607e-06, + "loss": 0.1164, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01483272947371006, + "step": 1051, + "valid_targets_mean": 4799.9, + "valid_targets_min": 1106 + }, + { + "epoch": 4.259240506329114, + "grad_norm": 0.23067383757803828, + "learning_rate": 2.6465962153439082e-06, + "loss": 0.106, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012178324162960052, + "step": 1052, + "valid_targets_mean": 4971.4, + "valid_targets_min": 1405 }, { - "epoch": 1.239493670886076, - "grad_norm": 0.4008715996205904, - "learning_rate": 3.7362920019285066e-05, - "loss": 0.2882, - "step": 306 + "epoch": 4.263291139240506, + "grad_norm": 0.2658122118801973, + "learning_rate": 2.618550200289303e-06, + "loss": 0.1074, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012370210140943527, + "step": 1053, + "valid_targets_mean": 4161.8, + "valid_targets_min": 1260 }, { - "epoch": 1.2435443037974683, - "grad_norm": 0.4169500727114222, - "learning_rate": 3.73346803025098e-05, - "loss": 0.3196, - "step": 307 + "epoch": 4.267341772151899, + "grad_norm": 0.2385990623874725, + "learning_rate": 2.59064316684412e-06, + "loss": 0.1074, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.010678352788090706, + "step": 1054, + "valid_targets_mean": 3915.4, + "valid_targets_min": 1049 }, { - "epoch": 1.2475949367088608, - "grad_norm": 0.3596736817631642, - "learning_rate": 3.730630097460247e-05, - "loss": 0.3169, - "step": 308 + "epoch": 4.271392405063291, + "grad_norm": 0.24493657061517968, + "learning_rate": 2.562875338152264e-06, + "loss": 0.1121, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011886542662978172, + "step": 1055, + "valid_targets_mean": 4704.1, + "valid_targets_min": 1066 }, { - "epoch": 1.251645569620253, - "grad_norm": 0.45383372875838074, - "learning_rate": 3.727778226412628e-05, - "loss": 0.3188, - "step": 309 + "epoch": 4.275443037974684, + "grad_norm": 0.24932341757544677, + "learning_rate": 2.5352469362445552e-06, + "loss": 0.1162, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014495152980089188, + "step": 1056, + "valid_targets_mean": 4415.4, + "valid_targets_min": 1253 + }, + { + "epoch": 4.279493670886076, + "grad_norm": 0.23423551396029124, + "learning_rate": 2.5077581820369523e-06, + "loss": 0.1083, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01119888387620449, + "step": 1057, + "valid_targets_mean": 3970.0, + "valid_targets_min": 1026 + }, + { + "epoch": 4.283544303797468, + "grad_norm": 0.23860357313003533, + "learning_rate": 2.480409295328803e-06, + "loss": 0.1034, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.010069448500871658, + "step": 1058, + "valid_targets_mean": 3870.6, + "valid_targets_min": 1099 + }, + { + "epoch": 4.287594936708861, + "grad_norm": 0.24458807065991509, + "learning_rate": 2.45320049480108e-06, + "loss": 0.1099, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015278888866305351, + "step": 1059, + "valid_targets_mean": 4747.9, + "valid_targets_min": 1143 + }, + { + "epoch": 4.291645569620253, + "grad_norm": 0.23154765850425468, + "learning_rate": 2.4261319980146293e-06, + "loss": 0.1095, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01711427979171276, + "step": 1060, + "valid_targets_mean": 5362.8, + "valid_targets_min": 1402 + }, + { + "epoch": 4.295696202531645, + "grad_norm": 0.2348947040709094, + "learning_rate": 2.399204021408421e-06, + "loss": 0.1048, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017931368201971054, + "step": 1061, + "valid_targets_mean": 5377.5, + "valid_targets_min": 1164 + }, + { + "epoch": 4.299746835443038, + "grad_norm": 0.24187219348589478, + "learning_rate": 2.3724167802978395e-06, + "loss": 0.1144, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009700765833258629, + "step": 1062, + "valid_targets_mean": 3637.3, + "valid_targets_min": 1719 + }, + { + "epoch": 4.30379746835443, + "grad_norm": 0.24712355847455172, + "learning_rate": 2.3457704888729625e-06, + "loss": 0.1034, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011837814934551716, + "step": 1063, + "valid_targets_mean": 4285.8, + "valid_targets_min": 1364 + }, + { + "epoch": 4.307848101265823, + "grad_norm": 0.25482267306274325, + "learning_rate": 2.3192653601968162e-06, + "loss": 0.1087, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017915062606334686, + "step": 1064, + "valid_targets_mean": 4704.2, + "valid_targets_min": 1391 }, { - "epoch": 1.2556962025316456, - "grad_norm": 0.3708696452154138, - "learning_rate": 3.7249124400767006e-05, - "loss": 0.2932, - "step": 310 - }, - { - "epoch": 1.259746835443038, - "grad_norm": 0.38017714206275194, - "learning_rate": 3.722032761533114e-05, - "loss": 0.3044, - "step": 311 - }, - { - "epoch": 1.2637974683544304, - "grad_norm": 0.38903080040893345, - "learning_rate": 3.719139213974403e-05, - "loss": 0.2859, - "step": 312 - }, - { - "epoch": 1.2678481012658227, - "grad_norm": 0.4237171809133419, - "learning_rate": 3.7162318207048006e-05, - "loss": 0.3218, - "step": 313 - }, - { - "epoch": 1.2718987341772152, - "grad_norm": 0.3911093364068046, - "learning_rate": 3.713310605140055e-05, - "loss": 0.3003, - "step": 314 - }, - { - "epoch": 1.2759493670886077, - "grad_norm": 0.40094745664913267, - "learning_rate": 3.710375590807233e-05, - "loss": 0.3167, - "step": 315 - }, - { - "epoch": 1.28, - "grad_norm": 0.388554840588802, - "learning_rate": 3.7074268013445365e-05, - "loss": 0.3072, - "step": 316 - }, - { - "epoch": 1.2840506329113923, - "grad_norm": 0.3528133424702583, - "learning_rate": 3.7044642605011114e-05, - "loss": 0.2975, - "step": 317 - }, - { - "epoch": 1.2881012658227848, - "grad_norm": 0.37292530385294026, - "learning_rate": 3.701487992136854e-05, - "loss": 0.3025, - "step": 318 - }, - { - "epoch": 1.2921518987341774, - "grad_norm": 0.3517577174738942, - "learning_rate": 3.69849802022222e-05, - "loss": 0.3191, - "step": 319 - }, - { - "epoch": 1.2962025316455696, - "grad_norm": 0.3683039864288579, - "learning_rate": 3.6954943688380334e-05, - "loss": 0.306, - "step": 320 - }, - { - "epoch": 1.300253164556962, - "grad_norm": 0.3934537659862702, - "learning_rate": 3.692477062175289e-05, - "loss": 0.3101, - "step": 321 - }, - { - "epoch": 1.3043037974683545, - "grad_norm": 0.3709759260295349, - "learning_rate": 3.689446124534958e-05, - "loss": 0.3115, - "step": 322 - }, - { - "epoch": 1.308354430379747, - "grad_norm": 0.4078998532827138, - "learning_rate": 3.686401580327799e-05, - "loss": 0.3138, - "step": 323 - }, - { - "epoch": 1.3124050632911393, - "grad_norm": 0.38497107622568194, - "learning_rate": 3.683343454074149e-05, - "loss": 0.3081, - "step": 324 - }, - { - "epoch": 1.3164556962025316, - "grad_norm": 0.39117870137121996, - "learning_rate": 3.6802717704037386e-05, - "loss": 0.3115, - "step": 325 - }, - { - "epoch": 1.320506329113924, - "grad_norm": 0.3573059313663357, - "learning_rate": 3.6771865540554855e-05, - "loss": 0.2992, - "step": 326 - }, - { - "epoch": 1.3245569620253166, - "grad_norm": 0.40203363003379106, - "learning_rate": 3.674087829877297e-05, - "loss": 0.3107, - "step": 327 - }, - { - "epoch": 1.3286075949367089, - "grad_norm": 0.38962655382917494, - "learning_rate": 3.6709756228258735e-05, - "loss": 0.3114, - "step": 328 - }, - { - "epoch": 1.3326582278481012, - "grad_norm": 0.37511970994837557, - "learning_rate": 3.667849957966501e-05, - "loss": 0.3047, - "step": 329 - }, - { - "epoch": 1.3367088607594937, - "grad_norm": 0.36457850547592774, - "learning_rate": 3.6647108604728546e-05, - "loss": 0.2971, - "step": 330 - }, - { - "epoch": 1.3407594936708862, - "grad_norm": 0.33774521726771906, - "learning_rate": 3.661558355626795e-05, - "loss": 0.2932, - "step": 331 - }, - { - "epoch": 1.3448101265822785, - "grad_norm": 0.3636656105219239, - "learning_rate": 3.658392468818163e-05, - "loss": 0.3002, - "step": 332 - }, - { - "epoch": 1.3488607594936708, - "grad_norm": 0.31321411197159726, - "learning_rate": 3.655213225544574e-05, - "loss": 0.3083, - "step": 333 - }, - { - "epoch": 1.3529113924050633, - "grad_norm": 0.32657355911635233, - "learning_rate": 3.652020651411218e-05, - "loss": 0.298, - "step": 334 - }, - { - "epoch": 1.3569620253164558, - "grad_norm": 0.3380579487426661, - "learning_rate": 3.6488147721306474e-05, - "loss": 0.3124, - "step": 335 - }, - { - "epoch": 1.3610126582278481, - "grad_norm": 0.35873855755908063, - "learning_rate": 3.645595613522574e-05, - "loss": 0.3173, - "step": 336 - }, - { - "epoch": 1.3650632911392404, - "grad_norm": 0.3582562735927334, - "learning_rate": 3.642363201513657e-05, - "loss": 0.3115, - "step": 337 - }, - { - "epoch": 1.369113924050633, - "grad_norm": 0.3687602502711562, - "learning_rate": 3.6391175621373006e-05, - "loss": 0.3045, - "step": 338 - }, - { - "epoch": 1.3731645569620254, - "grad_norm": 0.41965199699997474, - "learning_rate": 3.6358587215334355e-05, - "loss": 0.308, - "step": 339 - }, - { - "epoch": 1.3772151898734177, - "grad_norm": 0.3368401903867559, - "learning_rate": 3.632586705948318e-05, - "loss": 0.3025, - "step": 340 - }, - { - "epoch": 1.38126582278481, - "grad_norm": 0.3496207224304805, - "learning_rate": 3.629301541734311e-05, - "loss": 0.3242, - "step": 341 - }, - { - "epoch": 1.3853164556962025, - "grad_norm": 0.35937586417012435, - "learning_rate": 3.626003255349676e-05, - "loss": 0.3218, - "step": 342 - }, - { - "epoch": 1.389367088607595, - "grad_norm": 0.30954197599652233, - "learning_rate": 3.622691873358357e-05, - "loss": 0.3151, - "step": 343 - }, - { - "epoch": 1.3934177215189874, - "grad_norm": 0.3796695377233597, - "learning_rate": 3.61936742242977e-05, - "loss": 0.3171, - "step": 344 - }, - { - "epoch": 1.3974683544303796, - "grad_norm": 0.33671418084622534, - "learning_rate": 3.6160299293385864e-05, - "loss": 0.3087, - "step": 345 - }, - { - "epoch": 1.4015189873417722, - "grad_norm": 0.34008801775337266, - "learning_rate": 3.612679420964516e-05, - "loss": 0.3141, - "step": 346 - }, - { - "epoch": 1.4055696202531647, - "grad_norm": 0.35438680409491535, - "learning_rate": 3.609315924292092e-05, - "loss": 0.3047, - "step": 347 - }, - { - "epoch": 1.409620253164557, - "grad_norm": 0.3309026533894053, - "learning_rate": 3.6059394664104554e-05, - "loss": 0.3046, - "step": 348 - }, - { - "epoch": 1.4136708860759493, - "grad_norm": 0.37041288531262945, - "learning_rate": 3.602550074513133e-05, - "loss": 0.3022, - "step": 349 - }, - { - "epoch": 1.4177215189873418, - "grad_norm": 0.36671584499785154, - "learning_rate": 3.599147775897822e-05, - "loss": 0.3076, - "step": 350 - }, - { - "epoch": 1.4217721518987343, - "grad_norm": 0.3683879520241095, - "learning_rate": 3.595732597966167e-05, - "loss": 0.3083, - "step": 351 - }, - { - "epoch": 1.4258227848101266, - "grad_norm": 0.33413301296680736, - "learning_rate": 3.592304568223542e-05, - "loss": 0.3062, - "step": 352 - }, - { - "epoch": 1.4298734177215189, - "grad_norm": 0.37380631139938114, - "learning_rate": 3.588863714278826e-05, - "loss": 0.2958, - "step": 353 - }, - { - "epoch": 1.4339240506329114, - "grad_norm": 0.36062326488543883, - "learning_rate": 3.585410063844186e-05, - "loss": 0.2931, - "step": 354 - }, - { - "epoch": 1.437974683544304, - "grad_norm": 0.3622229125164926, - "learning_rate": 3.581943644734846e-05, - "loss": 0.2996, - "step": 355 - }, - { - "epoch": 1.4420253164556962, - "grad_norm": 0.3819719196404846, - "learning_rate": 3.578464484868869e-05, - "loss": 0.309, - "step": 356 - }, - { - "epoch": 1.4460759493670885, - "grad_norm": 0.352125928219145, - "learning_rate": 3.5749726122669316e-05, - "loss": 0.3036, - "step": 357 - }, - { - "epoch": 1.450126582278481, - "grad_norm": 0.3726055203774872, - "learning_rate": 3.5714680550520943e-05, - "loss": 0.31, - "step": 358 - }, - { - "epoch": 1.4541772151898735, - "grad_norm": 0.3776051700137096, - "learning_rate": 3.5679508414495794e-05, - "loss": 0.314, - "step": 359 - }, - { - "epoch": 1.4582278481012658, - "grad_norm": 0.3767577519707251, - "learning_rate": 3.564420999786543e-05, - "loss": 0.3094, - "step": 360 - }, - { - "epoch": 1.4622784810126581, - "grad_norm": 0.4074339338683601, - "learning_rate": 3.560878558491842e-05, - "loss": 0.3162, - "step": 361 - }, - { - "epoch": 1.4663291139240506, - "grad_norm": 0.33414754474626074, - "learning_rate": 3.5573235460958145e-05, - "loss": 0.3087, - "step": 362 - }, - { - "epoch": 1.4703797468354431, - "grad_norm": 0.40276270930067204, - "learning_rate": 3.553755991230039e-05, - "loss": 0.3019, - "step": 363 - }, - { - "epoch": 1.4744303797468354, - "grad_norm": 0.3569737364499927, - "learning_rate": 3.5501759226271144e-05, - "loss": 0.3095, - "step": 364 - }, - { - "epoch": 1.4784810126582277, - "grad_norm": 0.3951990904713934, - "learning_rate": 3.546583369120419e-05, - "loss": 0.3149, - "step": 365 - }, - { - "epoch": 1.4825316455696202, - "grad_norm": 0.42006591448621794, - "learning_rate": 3.5429783596438864e-05, - "loss": 0.3036, - "step": 366 - }, - { - "epoch": 1.4865822784810128, - "grad_norm": 0.33040874615111787, - "learning_rate": 3.539360923231766e-05, - "loss": 0.292, - "step": 367 - }, - { - "epoch": 1.490632911392405, - "grad_norm": 0.3778169988859503, - "learning_rate": 3.535731089018394e-05, - "loss": 0.3022, - "step": 368 - }, - { - "epoch": 1.4946835443037974, - "grad_norm": 0.37768212944799073, - "learning_rate": 3.532088886237956e-05, - "loss": 0.2984, - "step": 369 - }, - { - "epoch": 1.4987341772151899, - "grad_norm": 0.35620827314210374, - "learning_rate": 3.528434344224253e-05, - "loss": 0.3002, - "step": 370 - }, - { - "epoch": 1.5027848101265824, - "grad_norm": 0.43037923315563587, - "learning_rate": 3.524767492410464e-05, - "loss": 0.3211, - "step": 371 - }, - { - "epoch": 1.5068354430379747, - "grad_norm": 0.35195909697612277, - "learning_rate": 3.521088360328908e-05, - "loss": 0.311, - "step": 372 - }, - { - "epoch": 1.510886075949367, - "grad_norm": 0.3711450229276612, - "learning_rate": 3.517396977610811e-05, - "loss": 0.3001, - "step": 373 - }, - { - "epoch": 1.5149367088607595, - "grad_norm": 0.36729472410650504, - "learning_rate": 3.5136933739860595e-05, - "loss": 0.2988, - "step": 374 - }, - { - "epoch": 1.518987341772152, - "grad_norm": 0.40554013217426027, - "learning_rate": 3.509977579282971e-05, - "loss": 0.3219, - "step": 375 - }, - { - "epoch": 1.5230379746835443, - "grad_norm": 0.4495547020923126, - "learning_rate": 3.5062496234280424e-05, - "loss": 0.3081, - "step": 376 - }, - { - "epoch": 1.5270886075949366, - "grad_norm": 0.4153316219102672, - "learning_rate": 3.502509536445719e-05, - "loss": 0.3124, - "step": 377 - }, - { - "epoch": 1.531139240506329, - "grad_norm": 0.3846414856431148, - "learning_rate": 3.498757348458147e-05, - "loss": 0.3036, - "step": 378 - }, - { - "epoch": 1.5351898734177216, - "grad_norm": 0.3967641070642658, - "learning_rate": 3.4949930896849324e-05, - "loss": 0.3075, - "step": 379 - }, - { - "epoch": 1.539240506329114, - "grad_norm": 0.3554207096813955, - "learning_rate": 3.491216790442899e-05, - "loss": 0.2922, - "step": 380 - }, - { - "epoch": 1.5432911392405062, - "grad_norm": 0.37608270198564364, - "learning_rate": 3.487428481145839e-05, - "loss": 0.299, - "step": 381 - }, - { - "epoch": 1.5473417721518987, - "grad_norm": 0.4242786975157559, - "learning_rate": 3.483628192304278e-05, - "loss": 0.3075, - "step": 382 - }, - { - "epoch": 1.5513924050632912, - "grad_norm": 0.35626465601849155, - "learning_rate": 3.479815954525219e-05, - "loss": 0.3248, - "step": 383 - }, - { - "epoch": 1.5554430379746835, - "grad_norm": 0.37477137297898305, - "learning_rate": 3.475991798511899e-05, - "loss": 0.2997, - "step": 384 - }, - { - "epoch": 1.5594936708860758, - "grad_norm": 0.3640621366264976, - "learning_rate": 3.4721557550635464e-05, - "loss": 0.316, - "step": 385 - }, - { - "epoch": 1.5635443037974683, - "grad_norm": 0.3484506899595033, - "learning_rate": 3.468307855075128e-05, - "loss": 0.3102, - "step": 386 - }, - { - "epoch": 1.5675949367088609, - "grad_norm": 0.3708071599479605, - "learning_rate": 3.4644481295371005e-05, - "loss": 0.3229, - "step": 387 - }, - { - "epoch": 1.5716455696202531, - "grad_norm": 0.3288023071698783, - "learning_rate": 3.460576609535163e-05, - "loss": 0.3, - "step": 388 - }, - { - "epoch": 1.5756962025316454, - "grad_norm": 0.3518968564337706, - "learning_rate": 3.456693326250006e-05, - "loss": 0.3111, - "step": 389 - }, - { - "epoch": 1.579746835443038, - "grad_norm": 0.3245852088532655, - "learning_rate": 3.452798310957058e-05, - "loss": 0.2918, - "step": 390 - }, - { - "epoch": 1.5837974683544305, - "grad_norm": 0.3358093803637686, - "learning_rate": 3.4488915950262386e-05, - "loss": 0.2993, - "step": 391 - }, - { - "epoch": 1.5878481012658228, - "grad_norm": 0.35882399915393237, - "learning_rate": 3.4449732099216985e-05, - "loss": 0.3145, - "step": 392 - }, - { - "epoch": 1.591898734177215, - "grad_norm": 0.33067413016730346, - "learning_rate": 3.441043187201574e-05, - "loss": 0.3115, - "step": 393 - }, - { - "epoch": 1.5959493670886076, - "grad_norm": 0.32778330362095626, - "learning_rate": 3.437101558517728e-05, - "loss": 0.3071, - "step": 394 - }, - { - "epoch": 1.6, - "grad_norm": 0.31413419589106195, - "learning_rate": 3.433148355615496e-05, - "loss": 0.3073, - "step": 395 - }, - { - "epoch": 1.6040506329113924, - "grad_norm": 0.31471642804486183, - "learning_rate": 3.4291836103334294e-05, - "loss": 0.3011, - "step": 396 - }, - { - "epoch": 1.6081012658227847, - "grad_norm": 0.34047956205692265, - "learning_rate": 3.425207354603043e-05, - "loss": 0.3069, - "step": 397 - }, - { - "epoch": 1.6121518987341772, - "grad_norm": 0.3626339218829372, - "learning_rate": 3.421219620448553e-05, - "loss": 0.3017, - "step": 398 - }, - { - "epoch": 1.6162025316455697, - "grad_norm": 0.34868182854954943, - "learning_rate": 3.417220439986623e-05, - "loss": 0.3012, - "step": 399 - }, - { - "epoch": 1.620253164556962, - "grad_norm": 0.3480734460766342, - "learning_rate": 3.4132098454261024e-05, - "loss": 0.3037, - "step": 400 - }, - { - "epoch": 1.6243037974683543, - "grad_norm": 0.33681019616820673, - "learning_rate": 3.4091878690677676e-05, - "loss": 0.312, - "step": 401 - }, - { - "epoch": 1.6283544303797468, - "grad_norm": 0.32409164592458173, - "learning_rate": 3.405154543304065e-05, - "loss": 0.2994, - "step": 402 - }, - { - "epoch": 1.6324050632911393, - "grad_norm": 0.3351476462826444, - "learning_rate": 3.401109900618843e-05, - "loss": 0.3088, - "step": 403 - }, - { - "epoch": 1.6364556962025316, - "grad_norm": 0.3722931928591337, - "learning_rate": 3.3970539735870996e-05, - "loss": 0.2945, - "step": 404 - }, - { - "epoch": 1.640506329113924, - "grad_norm": 0.32738075064384836, - "learning_rate": 3.392986794874714e-05, - "loss": 0.3104, - "step": 405 - }, - { - "epoch": 1.6445569620253164, - "grad_norm": 0.3807225132103434, - "learning_rate": 3.388908397238184e-05, - "loss": 0.3016, - "step": 406 - }, - { - "epoch": 1.648607594936709, - "grad_norm": 0.3882093765887405, - "learning_rate": 3.384818813524362e-05, - "loss": 0.3099, - "step": 407 - }, - { - "epoch": 1.6526582278481012, - "grad_norm": 0.32245071553202787, - "learning_rate": 3.380718076670195e-05, - "loss": 0.3099, - "step": 408 - }, - { - "epoch": 1.6567088607594935, - "grad_norm": 0.38340179945418557, - "learning_rate": 3.376606219702454e-05, - "loss": 0.3125, - "step": 409 - }, - { - "epoch": 1.660759493670886, - "grad_norm": 0.3249376884421805, - "learning_rate": 3.372483275737468e-05, - "loss": 0.3064, - "step": 410 - }, - { - "epoch": 1.6648101265822786, - "grad_norm": 0.35158598633902166, - "learning_rate": 3.368349277980861e-05, - "loss": 0.3177, - "step": 411 - }, - { - "epoch": 1.6688607594936709, - "grad_norm": 0.3538849132777769, - "learning_rate": 3.3642042597272844e-05, - "loss": 0.3101, - "step": 412 - }, - { - "epoch": 1.6729113924050631, - "grad_norm": 0.33348812206396083, - "learning_rate": 3.360048254360144e-05, - "loss": 0.3107, - "step": 413 - }, - { - "epoch": 1.6769620253164557, - "grad_norm": 0.38120476907145756, - "learning_rate": 3.355881295351336e-05, - "loss": 0.3045, - "step": 414 - }, - { - "epoch": 1.6810126582278482, - "grad_norm": 0.3186755410913132, - "learning_rate": 3.351703416260975e-05, - "loss": 0.3184, - "step": 415 - }, - { - "epoch": 1.6850632911392405, - "grad_norm": 0.4028436539339753, - "learning_rate": 3.347514650737126e-05, - "loss": 0.3163, - "step": 416 - }, - { - "epoch": 1.6891139240506328, - "grad_norm": 0.35998718747286695, - "learning_rate": 3.3433150325155295e-05, - "loss": 0.3154, - "step": 417 - }, - { - "epoch": 1.6931645569620253, - "grad_norm": 0.3419052451854561, - "learning_rate": 3.339104595419334e-05, - "loss": 0.3072, - "step": 418 - }, - { - "epoch": 1.6972151898734178, - "grad_norm": 0.32571003218071815, - "learning_rate": 3.3348833733588204e-05, - "loss": 0.3221, - "step": 419 - }, - { - "epoch": 1.70126582278481, - "grad_norm": 0.37665502467793605, - "learning_rate": 3.3306514003311305e-05, - "loss": 0.3134, - "step": 420 - }, - { - "epoch": 1.7053164556962024, - "grad_norm": 0.32960721576607344, - "learning_rate": 3.326408710419996e-05, - "loss": 0.3118, - "step": 421 - }, - { - "epoch": 1.709367088607595, - "grad_norm": 0.31986035746301433, - "learning_rate": 3.322155337795454e-05, - "loss": 0.3035, - "step": 422 - }, - { - "epoch": 1.7134177215189874, - "grad_norm": 0.29131743585380454, - "learning_rate": 3.317891316713587e-05, - "loss": 0.309, - "step": 423 - }, - { - "epoch": 1.7174683544303797, - "grad_norm": 0.34580285455860643, - "learning_rate": 3.313616681516231e-05, - "loss": 0.3045, - "step": 424 - }, - { - "epoch": 1.721518987341772, - "grad_norm": 0.3065576659134362, - "learning_rate": 3.309331466630713e-05, - "loss": 0.3182, - "step": 425 - }, - { - "epoch": 1.7255696202531645, - "grad_norm": 0.34846853798629507, - "learning_rate": 3.305035706569563e-05, - "loss": 0.3047, - "step": 426 - }, - { - "epoch": 1.729620253164557, - "grad_norm": 0.30321531766671955, - "learning_rate": 3.3007294359302433e-05, - "loss": 0.3044, - "step": 427 - }, - { - "epoch": 1.7336708860759493, - "grad_norm": 0.41512873547163487, - "learning_rate": 3.296412689394864e-05, - "loss": 0.3043, - "step": 428 - }, - { - "epoch": 1.7377215189873416, - "grad_norm": 0.3425970937025628, - "learning_rate": 3.292085501729909e-05, - "loss": 0.3083, - "step": 429 - }, - { - "epoch": 1.7417721518987341, - "grad_norm": 0.3600679162185109, - "learning_rate": 3.2877479077859534e-05, - "loss": 0.3036, - "step": 430 - }, - { - "epoch": 1.7458227848101266, - "grad_norm": 0.3455753775334914, - "learning_rate": 3.283399942497381e-05, - "loss": 0.288, - "step": 431 - }, - { - "epoch": 1.749873417721519, - "grad_norm": 0.3560226655218737, - "learning_rate": 3.279041640882108e-05, - "loss": 0.3027, - "step": 432 - }, - { - "epoch": 1.7539240506329112, - "grad_norm": 0.3508361229146194, - "learning_rate": 3.2746730380412964e-05, - "loss": 0.2919, - "step": 433 - }, - { - "epoch": 1.7579746835443038, - "grad_norm": 0.3320865940991662, - "learning_rate": 3.2702941691590726e-05, - "loss": 0.2953, - "step": 434 - }, - { - "epoch": 1.7620253164556963, - "grad_norm": 0.36987211858137564, - "learning_rate": 3.265905069502244e-05, - "loss": 0.3203, - "step": 435 - }, - { - "epoch": 1.7660759493670886, - "grad_norm": 0.33831485507546677, - "learning_rate": 3.261505774420016e-05, - "loss": 0.2977, - "step": 436 - }, - { - "epoch": 1.7701265822784809, - "grad_norm": 0.4040583165106916, - "learning_rate": 3.257096319343707e-05, - "loss": 0.301, - "step": 437 - }, - { - "epoch": 1.7741772151898734, - "grad_norm": 0.35598305455242285, - "learning_rate": 3.2526767397864614e-05, - "loss": 0.2935, - "step": 438 - }, - { - "epoch": 1.7782278481012659, - "grad_norm": 0.4059701179135238, - "learning_rate": 3.248247071342966e-05, - "loss": 0.3219, - "step": 439 - }, - { - "epoch": 1.7822784810126582, - "grad_norm": 0.3510217153208696, - "learning_rate": 3.243807349689161e-05, - "loss": 0.3043, - "step": 440 - }, - { - "epoch": 1.7863291139240505, - "grad_norm": 0.3512658956120526, - "learning_rate": 3.2393576105819544e-05, - "loss": 0.3192, - "step": 441 - }, - { - "epoch": 1.790379746835443, - "grad_norm": 0.35895811010040746, - "learning_rate": 3.2348978898589333e-05, - "loss": 0.307, - "step": 442 - }, - { - "epoch": 1.7944303797468355, - "grad_norm": 0.4249513099091344, - "learning_rate": 3.230428223438075e-05, - "loss": 0.3013, - "step": 443 - }, - { - "epoch": 1.7984810126582278, - "grad_norm": 0.3352076104746956, - "learning_rate": 3.225948647317459e-05, - "loss": 0.2987, - "step": 444 - }, - { - "epoch": 1.80253164556962, - "grad_norm": 0.4289409058939938, - "learning_rate": 3.2214591975749745e-05, - "loss": 0.3181, - "step": 445 - }, - { - "epoch": 1.8065822784810126, - "grad_norm": 0.39341694347057665, - "learning_rate": 3.216959910368034e-05, - "loss": 0.311, - "step": 446 - }, - { - "epoch": 1.8106329113924051, - "grad_norm": 0.417025530824109, - "learning_rate": 3.212450821933277e-05, - "loss": 0.3049, - "step": 447 - }, - { - "epoch": 1.8146835443037974, - "grad_norm": 0.3887723170385768, - "learning_rate": 3.207931968586281e-05, - "loss": 0.3179, - "step": 448 - }, - { - "epoch": 1.81873417721519, - "grad_norm": 0.38974756267927496, - "learning_rate": 3.203403386721272e-05, - "loss": 0.298, - "step": 449 - }, - { - "epoch": 1.8227848101265822, - "grad_norm": 0.4040405016056989, - "learning_rate": 3.1988651128108245e-05, - "loss": 0.2983, - "step": 450 - }, - { - "epoch": 1.8268354430379747, - "grad_norm": 0.4082509500711886, - "learning_rate": 3.194317183405573e-05, - "loss": 0.3091, - "step": 451 - }, - { - "epoch": 1.8308860759493673, - "grad_norm": 0.36798266778857186, - "learning_rate": 3.189759635133914e-05, - "loss": 0.3316, - "step": 452 - }, - { - "epoch": 1.8349367088607595, - "grad_norm": 0.3532281353860484, - "learning_rate": 3.185192504701718e-05, - "loss": 0.3108, - "step": 453 - }, - { - "epoch": 1.8389873417721518, - "grad_norm": 0.347162741022233, - "learning_rate": 3.1806158288920234e-05, - "loss": 0.3136, - "step": 454 - }, - { - "epoch": 1.8430379746835444, - "grad_norm": 0.346705743547188, - "learning_rate": 3.1760296445647477e-05, - "loss": 0.3, - "step": 455 - }, - { - "epoch": 1.8470886075949369, - "grad_norm": 0.33622617828683116, - "learning_rate": 3.1714339886563896e-05, - "loss": 0.3128, - "step": 456 - }, - { - "epoch": 1.8511392405063292, - "grad_norm": 0.3441087707959669, - "learning_rate": 3.166828898179731e-05, - "loss": 0.3036, - "step": 457 - }, - { - "epoch": 1.8551898734177215, - "grad_norm": 0.3280173255061695, - "learning_rate": 3.162214410223536e-05, - "loss": 0.3289, - "step": 458 - }, - { - "epoch": 1.859240506329114, - "grad_norm": 0.3478782448564192, - "learning_rate": 3.157590561952257e-05, - "loss": 0.2899, - "step": 459 - }, - { - "epoch": 1.8632911392405065, - "grad_norm": 0.31582728388215037, - "learning_rate": 3.152957390605732e-05, - "loss": 0.2951, - "step": 460 - }, - { - "epoch": 1.8673417721518988, - "grad_norm": 0.31975948851908287, - "learning_rate": 3.148314933498886e-05, - "loss": 0.3005, - "step": 461 - }, - { - "epoch": 1.871392405063291, - "grad_norm": 0.33305479142753014, - "learning_rate": 3.143663228021431e-05, - "loss": 0.3019, - "step": 462 - }, - { - "epoch": 1.8754430379746836, - "grad_norm": 0.29771055186034223, - "learning_rate": 3.1390023116375624e-05, - "loss": 0.3114, - "step": 463 - }, - { - "epoch": 1.879493670886076, - "grad_norm": 0.34254152813331346, - "learning_rate": 3.134332221885661e-05, - "loss": 0.3072, - "step": 464 - }, - { - "epoch": 1.8835443037974684, - "grad_norm": 0.3030677988687901, - "learning_rate": 3.129652996377987e-05, - "loss": 0.3063, - "step": 465 - }, - { - "epoch": 1.8875949367088607, - "grad_norm": 0.33846526093021156, - "learning_rate": 3.12496467280038e-05, - "loss": 0.314, - "step": 466 - }, - { - "epoch": 1.8916455696202532, - "grad_norm": 0.3558736795532902, - "learning_rate": 3.120267288911952e-05, - "loss": 0.3082, - "step": 467 - }, - { - "epoch": 1.8956962025316457, - "grad_norm": 0.3182751704662135, - "learning_rate": 3.11556088254479e-05, - "loss": 0.3147, - "step": 468 - }, - { - "epoch": 1.899746835443038, - "grad_norm": 0.3285375429365072, - "learning_rate": 3.11084549160364e-05, - "loss": 0.2885, - "step": 469 - }, - { - "epoch": 1.9037974683544303, - "grad_norm": 0.361452802984596, - "learning_rate": 3.106121154065615e-05, - "loss": 0.3234, - "step": 470 - }, - { - "epoch": 1.9078481012658228, - "grad_norm": 0.33561875479342007, - "learning_rate": 3.1013879079798805e-05, - "loss": 0.3093, - "step": 471 - }, - { - "epoch": 1.9118987341772153, - "grad_norm": 0.37255876812371796, - "learning_rate": 3.096645791467348e-05, - "loss": 0.3123, - "step": 472 - }, - { - "epoch": 1.9159493670886076, - "grad_norm": 0.3499580467284806, - "learning_rate": 3.091894842720373e-05, - "loss": 0.3043, - "step": 473 - }, - { - "epoch": 1.92, - "grad_norm": 0.33157066558634335, - "learning_rate": 3.0871351000024425e-05, - "loss": 0.2966, - "step": 474 - }, - { - "epoch": 1.9240506329113924, - "grad_norm": 0.40534424699598226, - "learning_rate": 3.0823666016478716e-05, - "loss": 0.3073, - "step": 475 - }, - { - "epoch": 1.928101265822785, - "grad_norm": 0.30211307801700543, - "learning_rate": 3.0775893860614896e-05, - "loss": 0.3041, - "step": 476 - }, - { - "epoch": 1.9321518987341773, - "grad_norm": 0.37154144186317534, - "learning_rate": 3.0728034917183336e-05, - "loss": 0.3075, - "step": 477 - }, - { - "epoch": 1.9362025316455695, - "grad_norm": 0.3975354357050741, - "learning_rate": 3.06800895716334e-05, - "loss": 0.3076, - "step": 478 - }, - { - "epoch": 1.940253164556962, - "grad_norm": 0.375381675859087, - "learning_rate": 3.063205821011029e-05, - "loss": 0.3109, - "step": 479 - }, - { - "epoch": 1.9443037974683546, - "grad_norm": 0.35385791064965955, - "learning_rate": 3.0583941219452016e-05, - "loss": 0.314, - "step": 480 - }, - { - "epoch": 1.9483544303797469, - "grad_norm": 0.35131466246152804, - "learning_rate": 3.053573898718618e-05, - "loss": 0.2942, - "step": 481 - }, - { - "epoch": 1.9524050632911392, - "grad_norm": 0.3882910092558706, - "learning_rate": 3.0487451901526956e-05, - "loss": 0.3218, - "step": 482 - }, - { - "epoch": 1.9564556962025317, - "grad_norm": 0.3427505221427456, - "learning_rate": 3.0439080351371875e-05, - "loss": 0.2968, - "step": 483 - }, - { - "epoch": 1.9605063291139242, - "grad_norm": 0.3819014089455782, - "learning_rate": 3.0390624726298764e-05, - "loss": 0.3068, - "step": 484 - }, - { - "epoch": 1.9645569620253165, - "grad_norm": 0.33721238680975085, - "learning_rate": 3.034208541656255e-05, - "loss": 0.2962, - "step": 485 - }, - { - "epoch": 1.9686075949367088, - "grad_norm": 0.3391141070069093, - "learning_rate": 3.029346281309218e-05, - "loss": 0.3039, - "step": 486 - }, - { - "epoch": 1.9726582278481013, - "grad_norm": 0.3272795261940452, - "learning_rate": 3.0244757307487415e-05, - "loss": 0.3013, - "step": 487 - }, - { - "epoch": 1.9767088607594938, - "grad_norm": 0.3062386641155375, - "learning_rate": 3.019596929201569e-05, - "loss": 0.2863, - "step": 488 - }, - { - "epoch": 1.980759493670886, - "grad_norm": 0.3475874933998762, - "learning_rate": 3.0147099159608985e-05, - "loss": 0.3092, - "step": 489 - }, - { - "epoch": 1.9848101265822784, - "grad_norm": 0.32914917810414906, - "learning_rate": 3.0098147303860616e-05, - "loss": 0.301, - "step": 490 - }, - { - "epoch": 1.988860759493671, - "grad_norm": 0.32867483006849474, - "learning_rate": 3.0049114119022117e-05, - "loss": 0.3022, - "step": 491 - }, - { - "epoch": 1.9929113924050634, - "grad_norm": 0.32880097025368576, - "learning_rate": 3.0000000000000004e-05, - "loss": 0.2939, - "step": 492 - }, - { - "epoch": 1.9969620253164557, - "grad_norm": 0.3381942447692117, - "learning_rate": 2.995080534235264e-05, - "loss": 0.2969, - "step": 493 - }, - { - "epoch": 2.001012658227848, - "grad_norm": 0.3186874182696045, - "learning_rate": 2.9901530542287044e-05, - "loss": 0.2839, - "step": 494 - }, - { - "epoch": 2.0050632911392405, - "grad_norm": 0.4202143661111709, - "learning_rate": 2.9852175996655676e-05, - "loss": 0.2299, - "step": 495 - }, - { - "epoch": 2.009113924050633, - "grad_norm": 0.421940030363095, - "learning_rate": 2.980274210295326e-05, - "loss": 0.2529, - "step": 496 - }, - { - "epoch": 2.013164556962025, - "grad_norm": 0.5029481786190206, - "learning_rate": 2.9753229259313578e-05, - "loss": 0.2316, - "step": 497 - }, - { - "epoch": 2.0172151898734176, - "grad_norm": 0.38965584880344234, - "learning_rate": 2.9703637864506274e-05, - "loss": 0.225, - "step": 498 - }, - { - "epoch": 2.02126582278481, - "grad_norm": 0.36436934740828836, - "learning_rate": 2.965396831793362e-05, - "loss": 0.2322, - "step": 499 - }, - { - "epoch": 2.0253164556962027, - "grad_norm": 0.36999122350211794, - "learning_rate": 2.9604221019627316e-05, - "loss": 0.2298, - "step": 500 - }, - { - "epoch": 2.0293670886075947, - "grad_norm": 0.3716459623187574, - "learning_rate": 2.955439637024526e-05, - "loss": 0.2237, - "step": 501 - }, - { - "epoch": 2.0334177215189873, - "grad_norm": 0.40076888138912475, - "learning_rate": 2.9504494771068334e-05, - "loss": 0.2233, - "step": 502 - }, - { - "epoch": 2.0374683544303798, - "grad_norm": 0.3540927619182084, - "learning_rate": 2.9454516623997156e-05, - "loss": 0.2154, - "step": 503 - }, - { - "epoch": 2.0415189873417723, - "grad_norm": 0.442313865919713, - "learning_rate": 2.9404462331548847e-05, - "loss": 0.2224, - "step": 504 - }, - { - "epoch": 2.0455696202531644, - "grad_norm": 0.3261808547505268, - "learning_rate": 2.93543322968538e-05, - "loss": 0.2343, - "step": 505 - }, - { - "epoch": 2.049620253164557, - "grad_norm": 0.422051468390874, - "learning_rate": 2.9304126923652428e-05, - "loss": 0.2241, - "step": 506 - }, - { - "epoch": 2.0536708860759494, - "grad_norm": 0.35311825171712996, - "learning_rate": 2.9253846616291896e-05, - "loss": 0.2354, - "step": 507 - }, - { - "epoch": 2.057721518987342, - "grad_norm": 0.3912566361021335, - "learning_rate": 2.9203491779722896e-05, - "loss": 0.2166, - "step": 508 - }, - { - "epoch": 2.061772151898734, - "grad_norm": 0.35866533690807656, - "learning_rate": 2.9153062819496357e-05, - "loss": 0.2198, - "step": 509 - }, - { - "epoch": 2.0658227848101265, - "grad_norm": 0.41608012138400047, - "learning_rate": 2.9102560141760178e-05, - "loss": 0.2375, - "step": 510 - }, - { - "epoch": 2.069873417721519, - "grad_norm": 0.3077940784724969, - "learning_rate": 2.9051984153256004e-05, - "loss": 0.2362, - "step": 511 - }, - { - "epoch": 2.0739240506329115, - "grad_norm": 0.42065743908520087, - "learning_rate": 2.900133526131588e-05, - "loss": 0.2403, - "step": 512 - }, - { - "epoch": 2.0779746835443036, - "grad_norm": 0.33933903281834593, - "learning_rate": 2.8950613873859025e-05, - "loss": 0.2303, - "step": 513 - }, - { - "epoch": 2.082025316455696, - "grad_norm": 0.342172208506659, - "learning_rate": 2.8899820399388515e-05, - "loss": 0.2423, - "step": 514 - }, - { - "epoch": 2.0860759493670886, - "grad_norm": 0.382036033143454, - "learning_rate": 2.8848955246988012e-05, - "loss": 0.2305, - "step": 515 - }, - { - "epoch": 2.090126582278481, - "grad_norm": 0.32557128773846483, - "learning_rate": 2.879801882631847e-05, - "loss": 0.2218, - "step": 516 - }, - { - "epoch": 2.094177215189873, - "grad_norm": 0.3767529172465701, - "learning_rate": 2.8747011547614808e-05, - "loss": 0.2394, - "step": 517 - }, - { - "epoch": 2.0982278481012657, - "grad_norm": 0.3867110994320598, - "learning_rate": 2.8695933821682635e-05, - "loss": 0.2211, - "step": 518 - }, - { - "epoch": 2.1022784810126582, - "grad_norm": 0.3108020881301015, - "learning_rate": 2.864478605989494e-05, - "loss": 0.2334, - "step": 519 - }, - { - "epoch": 2.1063291139240508, - "grad_norm": 0.343336721151281, - "learning_rate": 2.8593568674188765e-05, - "loss": 0.236, - "step": 520 - }, - { - "epoch": 2.110379746835443, - "grad_norm": 0.3667039556169178, - "learning_rate": 2.8542282077061892e-05, - "loss": 0.2318, - "step": 521 - }, - { - "epoch": 2.1144303797468353, - "grad_norm": 0.34807687715788505, - "learning_rate": 2.8490926681569523e-05, - "loss": 0.2289, - "step": 522 - }, - { - "epoch": 2.118481012658228, - "grad_norm": 0.35076105289609055, - "learning_rate": 2.8439502901320956e-05, - "loss": 0.2443, - "step": 523 - }, - { - "epoch": 2.1225316455696204, - "grad_norm": 0.34075041057557043, - "learning_rate": 2.8388011150476237e-05, - "loss": 0.2205, - "step": 524 - }, - { - "epoch": 2.1265822784810124, - "grad_norm": 0.3818341449276432, - "learning_rate": 2.8336451843742866e-05, - "loss": 0.2326, - "step": 525 - }, - { - "epoch": 2.130632911392405, - "grad_norm": 0.3003569304567432, - "learning_rate": 2.8284825396372387e-05, - "loss": 0.2191, - "step": 526 - }, - { - "epoch": 2.1346835443037975, - "grad_norm": 0.370402792510317, - "learning_rate": 2.8233132224157132e-05, - "loss": 0.2317, - "step": 527 - }, - { - "epoch": 2.13873417721519, - "grad_norm": 0.3279957066011223, - "learning_rate": 2.8181372743426805e-05, - "loss": 0.2131, - "step": 528 - }, - { - "epoch": 2.1427848101265825, - "grad_norm": 0.33356932643747456, - "learning_rate": 2.8129547371045128e-05, - "loss": 0.2322, - "step": 529 - }, - { - "epoch": 2.1468354430379746, - "grad_norm": 0.3583470956277111, - "learning_rate": 2.8077656524406534e-05, - "loss": 0.2182, - "step": 530 - }, - { - "epoch": 2.150886075949367, - "grad_norm": 0.33005145912038103, - "learning_rate": 2.802570062143278e-05, - "loss": 0.2147, - "step": 531 - }, - { - "epoch": 2.1549367088607596, - "grad_norm": 0.33952235565565364, - "learning_rate": 2.7973680080569555e-05, - "loss": 0.2445, - "step": 532 - }, - { - "epoch": 2.1589873417721517, - "grad_norm": 0.3982486914977662, - "learning_rate": 2.792159532078314e-05, - "loss": 0.235, - "step": 533 - }, - { - "epoch": 2.163037974683544, - "grad_norm": 0.31376903305711296, - "learning_rate": 2.7869446761557033e-05, - "loss": 0.2305, - "step": 534 - }, - { - "epoch": 2.1670886075949367, - "grad_norm": 0.40324962441140294, - "learning_rate": 2.781723482288857e-05, - "loss": 0.2231, - "step": 535 - }, - { - "epoch": 2.1711392405063292, - "grad_norm": 0.30543611061645193, - "learning_rate": 2.7764959925285517e-05, - "loss": 0.2304, - "step": 536 - }, - { - "epoch": 2.1751898734177217, - "grad_norm": 0.36582683865338333, - "learning_rate": 2.771262248976272e-05, - "loss": 0.2247, - "step": 537 - }, - { - "epoch": 2.179240506329114, - "grad_norm": 0.30045849876187175, - "learning_rate": 2.7660222937838677e-05, - "loss": 0.2365, - "step": 538 - }, - { - "epoch": 2.1832911392405063, - "grad_norm": 0.3310106943877157, - "learning_rate": 2.7607761691532186e-05, - "loss": 0.2296, - "step": 539 - }, - { - "epoch": 2.187341772151899, - "grad_norm": 0.3222424600684214, - "learning_rate": 2.7555239173358916e-05, - "loss": 0.2253, - "step": 540 - }, - { - "epoch": 2.191392405063291, - "grad_norm": 0.3331263485269118, - "learning_rate": 2.7502655806328e-05, - "loss": 0.2177, - "step": 541 - }, - { - "epoch": 2.1954430379746834, - "grad_norm": 0.3226078619705582, - "learning_rate": 2.7450012013938648e-05, - "loss": 0.2181, - "step": 542 - }, - { - "epoch": 2.199493670886076, - "grad_norm": 0.34647964811167686, - "learning_rate": 2.739730822017673e-05, - "loss": 0.2174, - "step": 543 - }, - { - "epoch": 2.2035443037974685, - "grad_norm": 0.3100039945841827, - "learning_rate": 2.7344544849511355e-05, - "loss": 0.227, - "step": 544 - }, - { - "epoch": 2.207594936708861, - "grad_norm": 0.35351912932360496, - "learning_rate": 2.7291722326891456e-05, - "loss": 0.2406, - "step": 545 - }, - { - "epoch": 2.211645569620253, - "grad_norm": 0.34145910345911307, - "learning_rate": 2.723884107774236e-05, - "loss": 0.2246, - "step": 546 - }, - { - "epoch": 2.2156962025316456, - "grad_norm": 0.34103007269174834, - "learning_rate": 2.718590152796239e-05, - "loss": 0.238, - "step": 547 - }, - { - "epoch": 2.219746835443038, - "grad_norm": 0.37752222785639417, - "learning_rate": 2.71329041039194e-05, - "loss": 0.2146, - "step": 548 - }, - { - "epoch": 2.22379746835443, - "grad_norm": 0.3106685632570866, - "learning_rate": 2.7079849232447357e-05, - "loss": 0.2255, - "step": 549 - }, - { - "epoch": 2.2278481012658227, - "grad_norm": 0.3565144379699296, - "learning_rate": 2.7026737340842895e-05, - "loss": 0.2468, - "step": 550 - }, - { - "epoch": 2.231898734177215, - "grad_norm": 0.32066486779989345, - "learning_rate": 2.697356885686189e-05, - "loss": 0.2213, - "step": 551 - }, - { - "epoch": 2.2359493670886077, - "grad_norm": 0.3571471990643203, - "learning_rate": 2.6920344208716014e-05, - "loss": 0.234, - "step": 552 - }, - { - "epoch": 2.24, - "grad_norm": 0.34387486472214907, - "learning_rate": 2.6867063825069252e-05, - "loss": 0.244, - "step": 553 - }, - { - "epoch": 2.2440506329113923, - "grad_norm": 0.3183468678976038, - "learning_rate": 2.6813728135034494e-05, - "loss": 0.2292, - "step": 554 - }, - { - "epoch": 2.248101265822785, - "grad_norm": 0.3517469907430876, - "learning_rate": 2.6760337568170056e-05, - "loss": 0.2334, - "step": 555 - }, - { - "epoch": 2.2521518987341773, - "grad_norm": 0.3139780522459836, - "learning_rate": 2.6706892554476226e-05, - "loss": 0.2319, - "step": 556 - }, - { - "epoch": 2.2562025316455694, - "grad_norm": 0.3319448150834173, - "learning_rate": 2.6653393524391795e-05, - "loss": 0.2319, - "step": 557 - }, - { - "epoch": 2.260253164556962, - "grad_norm": 0.3474796980649211, - "learning_rate": 2.6599840908790592e-05, - "loss": 0.2391, - "step": 558 - }, - { - "epoch": 2.2643037974683544, - "grad_norm": 0.31059297441143163, - "learning_rate": 2.6546235138978028e-05, - "loss": 0.2015, - "step": 559 - }, - { - "epoch": 2.268354430379747, - "grad_norm": 0.3358312632552793, - "learning_rate": 2.6492576646687597e-05, - "loss": 0.2276, - "step": 560 - }, - { - "epoch": 2.2724050632911394, - "grad_norm": 0.37064924468074395, - "learning_rate": 2.6438865864077425e-05, - "loss": 0.2224, - "step": 561 - }, - { - "epoch": 2.2764556962025315, - "grad_norm": 0.3003172244276441, - "learning_rate": 2.6385103223726766e-05, - "loss": 0.2394, - "step": 562 - }, - { - "epoch": 2.280506329113924, - "grad_norm": 0.3540759021212988, - "learning_rate": 2.6331289158632537e-05, - "loss": 0.232, - "step": 563 - }, - { - "epoch": 2.2845569620253166, - "grad_norm": 0.3343498062647225, - "learning_rate": 2.6277424102205817e-05, - "loss": 0.2349, - "step": 564 - }, - { - "epoch": 2.2886075949367086, - "grad_norm": 0.3205503466207593, - "learning_rate": 2.6223508488268374e-05, - "loss": 0.2348, - "step": 565 - }, - { - "epoch": 2.292658227848101, - "grad_norm": 0.34886547389591654, - "learning_rate": 2.6169542751049148e-05, - "loss": 0.2217, - "step": 566 - }, - { - "epoch": 2.2967088607594937, - "grad_norm": 0.30657664739681834, - "learning_rate": 2.6115527325180754e-05, - "loss": 0.2411, - "step": 567 - }, - { - "epoch": 2.300759493670886, - "grad_norm": 0.36195707990008286, - "learning_rate": 2.606146264569603e-05, - "loss": 0.2075, - "step": 568 - }, - { - "epoch": 2.3048101265822787, - "grad_norm": 0.3312578144682574, - "learning_rate": 2.6007349148024447e-05, - "loss": 0.2279, - "step": 569 - }, - { - "epoch": 2.3088607594936708, - "grad_norm": 0.32303200348689703, - "learning_rate": 2.5953187267988694e-05, - "loss": 0.2263, - "step": 570 - }, - { - "epoch": 2.3129113924050633, - "grad_norm": 0.3539564156273888, - "learning_rate": 2.5898977441801097e-05, - "loss": 0.2238, - "step": 571 - }, - { - "epoch": 2.316962025316456, - "grad_norm": 0.3175429448707806, - "learning_rate": 2.584472010606015e-05, - "loss": 0.2196, - "step": 572 - }, - { - "epoch": 2.321012658227848, - "grad_norm": 0.3427743621699918, - "learning_rate": 2.5790415697746976e-05, - "loss": 0.2231, - "step": 573 - }, - { - "epoch": 2.3250632911392404, - "grad_norm": 0.33047355749585616, - "learning_rate": 2.5736064654221808e-05, - "loss": 0.2343, - "step": 574 - }, - { - "epoch": 2.329113924050633, - "grad_norm": 0.3084207181707558, - "learning_rate": 2.568166741322048e-05, - "loss": 0.2313, - "step": 575 - }, - { - "epoch": 2.3331645569620254, - "grad_norm": 0.29696309945301413, - "learning_rate": 2.56272244128509e-05, - "loss": 0.2247, - "step": 576 - }, - { - "epoch": 2.337215189873418, - "grad_norm": 0.33442276560415224, - "learning_rate": 2.55727360915895e-05, - "loss": 0.2307, - "step": 577 - }, - { - "epoch": 2.34126582278481, - "grad_norm": 0.33916975183181786, - "learning_rate": 2.5518202888277734e-05, - "loss": 0.2263, - "step": 578 - }, - { - "epoch": 2.3453164556962025, - "grad_norm": 0.32041529521824186, - "learning_rate": 2.5463625242118523e-05, - "loss": 0.2378, - "step": 579 - }, - { - "epoch": 2.349367088607595, - "grad_norm": 0.3363192682354492, - "learning_rate": 2.5409003592672723e-05, - "loss": 0.232, - "step": 580 - }, - { - "epoch": 2.353417721518987, - "grad_norm": 0.3125051262879329, - "learning_rate": 2.535433837985559e-05, - "loss": 0.229, - "step": 581 - }, - { - "epoch": 2.3574683544303796, - "grad_norm": 0.35146861684417946, - "learning_rate": 2.529963004393324e-05, - "loss": 0.2295, - "step": 582 - }, - { - "epoch": 2.361518987341772, - "grad_norm": 0.3343526977980359, - "learning_rate": 2.524487902551908e-05, - "loss": 0.2456, - "step": 583 - }, - { - "epoch": 2.3655696202531646, - "grad_norm": 0.3344390038543155, - "learning_rate": 2.519008576557029e-05, - "loss": 0.2331, - "step": 584 - }, - { - "epoch": 2.369620253164557, - "grad_norm": 0.37276725143202716, - "learning_rate": 2.5135250705384254e-05, - "loss": 0.2388, - "step": 585 - }, - { - "epoch": 2.3736708860759492, - "grad_norm": 0.3038000050999679, - "learning_rate": 2.5080374286595007e-05, - "loss": 0.2429, - "step": 586 - }, - { - "epoch": 2.3777215189873417, - "grad_norm": 0.3837563815303415, - "learning_rate": 2.5025456951169677e-05, - "loss": 0.246, - "step": 587 - }, - { - "epoch": 2.3817721518987343, - "grad_norm": 0.3052384603062137, - "learning_rate": 2.4970499141404942e-05, - "loss": 0.2313, - "step": 588 - }, - { - "epoch": 2.3858227848101268, - "grad_norm": 0.3427864640613424, - "learning_rate": 2.491550129992345e-05, - "loss": 0.2378, - "step": 589 - }, - { - "epoch": 2.389873417721519, - "grad_norm": 0.3377584439736947, - "learning_rate": 2.486046386967024e-05, - "loss": 0.23, - "step": 590 - }, - { - "epoch": 2.3939240506329114, - "grad_norm": 0.3209817438957892, - "learning_rate": 2.4805387293909214e-05, - "loss": 0.2407, - "step": 591 - }, - { - "epoch": 2.397974683544304, - "grad_norm": 0.3396306654692784, - "learning_rate": 2.4750272016219552e-05, - "loss": 0.2245, - "step": 592 - }, - { - "epoch": 2.4020253164556964, - "grad_norm": 0.3044799039387763, - "learning_rate": 2.4695118480492114e-05, - "loss": 0.2395, - "step": 593 - }, - { - "epoch": 2.4060759493670885, - "grad_norm": 0.3244853017704809, - "learning_rate": 2.4639927130925898e-05, - "loss": 0.2267, - "step": 594 - }, - { - "epoch": 2.410126582278481, - "grad_norm": 0.3242814335344009, - "learning_rate": 2.458469841202444e-05, - "loss": 0.2186, - "step": 595 - }, - { - "epoch": 2.4141772151898735, - "grad_norm": 0.29210813570125793, - "learning_rate": 2.452943276859226e-05, - "loss": 0.2329, - "step": 596 - }, - { - "epoch": 2.418227848101266, - "grad_norm": 0.3667952132266445, - "learning_rate": 2.447413064573125e-05, - "loss": 0.2389, - "step": 597 - }, - { - "epoch": 2.422278481012658, - "grad_norm": 0.3236306693329527, - "learning_rate": 2.4418792488837095e-05, - "loss": 0.2373, - "step": 598 - }, - { - "epoch": 2.4263291139240506, - "grad_norm": 0.34364296794999205, - "learning_rate": 2.4363418743595713e-05, - "loss": 0.2265, - "step": 599 - }, - { - "epoch": 2.430379746835443, - "grad_norm": 0.3178166876200261, - "learning_rate": 2.430800985597963e-05, - "loss": 0.2324, - "step": 600 - }, - { - "epoch": 2.4344303797468356, - "grad_norm": 0.33390152246493937, - "learning_rate": 2.4252566272244415e-05, - "loss": 0.2327, - "step": 601 - }, - { - "epoch": 2.4384810126582277, - "grad_norm": 0.3132143786510946, - "learning_rate": 2.4197088438925063e-05, - "loss": 0.2249, - "step": 602 - }, - { - "epoch": 2.44253164556962, - "grad_norm": 0.3144053814144751, - "learning_rate": 2.4141576802832417e-05, - "loss": 0.2316, - "step": 603 - }, - { - "epoch": 2.4465822784810127, - "grad_norm": 0.3066758354289081, - "learning_rate": 2.408603181104957e-05, - "loss": 0.2127, - "step": 604 - }, - { - "epoch": 2.4506329113924052, - "grad_norm": 0.33590284205673726, - "learning_rate": 2.4030453910928245e-05, - "loss": 0.2283, - "step": 605 - }, - { - "epoch": 2.4546835443037973, - "grad_norm": 0.2929689449460151, - "learning_rate": 2.397484355008521e-05, - "loss": 0.2275, - "step": 606 - }, - { - "epoch": 2.45873417721519, - "grad_norm": 0.3317807036650442, - "learning_rate": 2.3919201176398662e-05, - "loss": 0.2226, - "step": 607 - }, - { - "epoch": 2.4627848101265823, - "grad_norm": 0.32008657883466285, - "learning_rate": 2.3863527238004633e-05, - "loss": 0.2316, - "step": 608 - }, - { - "epoch": 2.466835443037975, - "grad_norm": 0.3114178326090441, - "learning_rate": 2.380782218329337e-05, - "loss": 0.234, - "step": 609 - }, - { - "epoch": 2.470886075949367, - "grad_norm": 0.3233869789345488, - "learning_rate": 2.3752086460905725e-05, - "loss": 0.234, - "step": 610 - }, - { - "epoch": 2.4749367088607594, - "grad_norm": 0.3309863242340352, - "learning_rate": 2.3696320519729544e-05, - "loss": 0.2417, - "step": 611 - }, - { - "epoch": 2.478987341772152, - "grad_norm": 0.32844562271316874, - "learning_rate": 2.3640524808896045e-05, - "loss": 0.2348, - "step": 612 - }, - { - "epoch": 2.4830379746835445, - "grad_norm": 0.3369904678729884, - "learning_rate": 2.3584699777776222e-05, - "loss": 0.2398, - "step": 613 - }, - { - "epoch": 2.4870886075949366, - "grad_norm": 0.2860776951333733, - "learning_rate": 2.3528845875977195e-05, - "loss": 0.2324, - "step": 614 - }, - { - "epoch": 2.491139240506329, - "grad_norm": 0.32477149257158855, - "learning_rate": 2.3472963553338614e-05, - "loss": 0.2261, - "step": 615 - }, - { - "epoch": 2.4951898734177216, - "grad_norm": 0.29667596682905584, - "learning_rate": 2.341705325992901e-05, - "loss": 0.2368, - "step": 616 - }, - { - "epoch": 2.499240506329114, - "grad_norm": 0.3024219913294697, - "learning_rate": 2.336111544604222e-05, - "loss": 0.2323, - "step": 617 - }, - { - "epoch": 2.503291139240506, - "grad_norm": 0.3136508071085885, - "learning_rate": 2.33051505621937e-05, - "loss": 0.2395, - "step": 618 - }, - { - "epoch": 2.5073417721518987, - "grad_norm": 0.29577653614805455, - "learning_rate": 2.324915905911693e-05, - "loss": 0.2338, - "step": 619 - }, - { - "epoch": 2.511392405063291, - "grad_norm": 0.30366453062617665, - "learning_rate": 2.319314138775977e-05, - "loss": 0.2435, - "step": 620 - }, - { - "epoch": 2.5154430379746833, - "grad_norm": 0.31821408416911096, - "learning_rate": 2.3137097999280856e-05, - "loss": 0.2581, - "step": 621 - }, - { - "epoch": 2.519493670886076, - "grad_norm": 0.28811472658974696, - "learning_rate": 2.308102934504593e-05, - "loss": 0.2273, - "step": 622 - }, - { - "epoch": 2.5235443037974683, - "grad_norm": 0.32020178339840244, - "learning_rate": 2.3024935876624222e-05, - "loss": 0.2391, - "step": 623 - }, - { - "epoch": 2.527594936708861, - "grad_norm": 0.307084993447976, - "learning_rate": 2.2968818045784813e-05, - "loss": 0.233, - "step": 624 - }, - { - "epoch": 2.5316455696202533, - "grad_norm": 0.29015370749820446, - "learning_rate": 2.2912676304493006e-05, - "loss": 0.2489, - "step": 625 - }, - { - "epoch": 2.5356962025316454, - "grad_norm": 0.2999431888526454, - "learning_rate": 2.2856511104906668e-05, - "loss": 0.2259, - "step": 626 - }, - { - "epoch": 2.539746835443038, - "grad_norm": 0.3317474664144101, - "learning_rate": 2.2800322899372586e-05, - "loss": 0.2194, - "step": 627 - }, - { - "epoch": 2.5437974683544304, - "grad_norm": 0.2834874033188405, - "learning_rate": 2.2744112140422844e-05, - "loss": 0.2338, - "step": 628 - }, - { - "epoch": 2.547848101265823, - "grad_norm": 0.29730279435675405, - "learning_rate": 2.2687879280771177e-05, - "loss": 0.2324, - "step": 629 - }, - { - "epoch": 2.5518987341772155, - "grad_norm": 0.3114087763338251, - "learning_rate": 2.26316247733093e-05, - "loss": 0.2335, - "step": 630 - }, - { - "epoch": 2.5559493670886075, - "grad_norm": 0.28827782752869485, - "learning_rate": 2.257534907110328e-05, - "loss": 0.2294, - "step": 631 - }, - { - "epoch": 2.56, - "grad_norm": 0.28708836411935695, - "learning_rate": 2.2519052627389882e-05, - "loss": 0.2395, - "step": 632 - }, - { - "epoch": 2.5640506329113926, - "grad_norm": 0.2887297457314167, - "learning_rate": 2.246273589557294e-05, - "loss": 0.2416, - "step": 633 - }, - { - "epoch": 2.5681012658227846, - "grad_norm": 0.286262059121319, - "learning_rate": 2.240639932921966e-05, - "loss": 0.238, - "step": 634 - }, - { - "epoch": 2.572151898734177, - "grad_norm": 0.27024567578020364, - "learning_rate": 2.2350043382056995e-05, - "loss": 0.2466, - "step": 635 - }, - { - "epoch": 2.5762025316455697, - "grad_norm": 0.3025562806557872, - "learning_rate": 2.2293668507968015e-05, - "loss": 0.2422, - "step": 636 - }, - { - "epoch": 2.580253164556962, - "grad_norm": 0.28661160948762593, - "learning_rate": 2.2237275160988186e-05, - "loss": 0.2358, - "step": 637 - }, - { - "epoch": 2.5843037974683547, - "grad_norm": 0.2929293274627698, - "learning_rate": 2.2180863795301787e-05, - "loss": 0.2236, - "step": 638 - }, - { - "epoch": 2.5883544303797468, - "grad_norm": 0.28877826219946695, - "learning_rate": 2.212443486523819e-05, - "loss": 0.2248, - "step": 639 - }, - { - "epoch": 2.5924050632911393, - "grad_norm": 0.2923062100120105, - "learning_rate": 2.2067988825268243e-05, - "loss": 0.2433, - "step": 640 - }, - { - "epoch": 2.596455696202532, - "grad_norm": 0.2927163765388562, - "learning_rate": 2.2011526130000596e-05, - "loss": 0.2345, - "step": 641 - }, - { - "epoch": 2.600506329113924, - "grad_norm": 0.3057658365762034, - "learning_rate": 2.1955047234178038e-05, - "loss": 0.2389, - "step": 642 - }, - { - "epoch": 2.6045569620253164, - "grad_norm": 0.2972381289521406, - "learning_rate": 2.1898552592673825e-05, - "loss": 0.2393, - "step": 643 - }, - { - "epoch": 2.608607594936709, - "grad_norm": 0.3264507687670856, - "learning_rate": 2.184204266048803e-05, - "loss": 0.238, - "step": 644 - }, - { - "epoch": 2.6126582278481014, - "grad_norm": 0.3081470129365603, - "learning_rate": 2.1785517892743887e-05, - "loss": 0.2139, - "step": 645 - }, - { - "epoch": 2.616708860759494, - "grad_norm": 0.3464173287109854, - "learning_rate": 2.17289787446841e-05, - "loss": 0.2341, - "step": 646 - }, - { - "epoch": 2.620759493670886, - "grad_norm": 0.3060514959264648, - "learning_rate": 2.1672425671667198e-05, - "loss": 0.2225, - "step": 647 - }, - { - "epoch": 2.6248101265822785, - "grad_norm": 0.30792753435196085, - "learning_rate": 2.161585912916385e-05, - "loss": 0.2463, - "step": 648 - }, - { - "epoch": 2.628860759493671, - "grad_norm": 0.3396525801751568, - "learning_rate": 2.1559279572753214e-05, - "loss": 0.2351, - "step": 649 - }, - { - "epoch": 2.632911392405063, - "grad_norm": 0.2911893715014043, - "learning_rate": 2.1502687458119268e-05, - "loss": 0.2185, - "step": 650 - }, - { - "epoch": 2.6369620253164556, - "grad_norm": 0.3514074632282774, - "learning_rate": 2.1446083241047116e-05, - "loss": 0.2239, - "step": 651 - }, - { - "epoch": 2.641012658227848, - "grad_norm": 0.3077437004270069, - "learning_rate": 2.1389467377419333e-05, - "loss": 0.2274, - "step": 652 - }, - { - "epoch": 2.6450632911392407, - "grad_norm": 0.29631837947427786, - "learning_rate": 2.133284032321232e-05, - "loss": 0.2372, - "step": 653 - }, - { - "epoch": 2.649113924050633, - "grad_norm": 0.3019119801912528, - "learning_rate": 2.1276202534492566e-05, - "loss": 0.2479, - "step": 654 - }, - { - "epoch": 2.6531645569620252, - "grad_norm": 0.31960189583156856, - "learning_rate": 2.121955446741306e-05, - "loss": 0.2338, - "step": 655 - }, - { - "epoch": 2.6572151898734178, - "grad_norm": 0.3187735220545649, - "learning_rate": 2.1162896578209517e-05, - "loss": 0.2185, - "step": 656 - }, - { - "epoch": 2.6612658227848103, - "grad_norm": 0.3305830311295424, - "learning_rate": 2.1106229323196813e-05, - "loss": 0.2286, - "step": 657 - }, - { - "epoch": 2.6653164556962023, - "grad_norm": 0.33401267312879923, - "learning_rate": 2.1049553158765214e-05, - "loss": 0.2182, - "step": 658 - }, - { - "epoch": 2.669367088607595, - "grad_norm": 0.3173485119264029, - "learning_rate": 2.0992868541376764e-05, - "loss": 0.2325, - "step": 659 - }, - { - "epoch": 2.6734177215189874, - "grad_norm": 0.3221766612292656, - "learning_rate": 2.093617592756158e-05, - "loss": 0.2382, - "step": 660 - }, - { - "epoch": 2.67746835443038, - "grad_norm": 0.3317191399796401, - "learning_rate": 2.0879475773914167e-05, - "loss": 0.2284, - "step": 661 - }, - { - "epoch": 2.6815189873417724, - "grad_norm": 0.3070225612509448, - "learning_rate": 2.082276853708978e-05, - "loss": 0.2298, - "step": 662 - }, - { - "epoch": 2.6855696202531645, - "grad_norm": 0.31677634285509093, - "learning_rate": 2.076605467380071e-05, - "loss": 0.2277, - "step": 663 - }, - { - "epoch": 2.689620253164557, - "grad_norm": 0.31824172182379085, - "learning_rate": 2.0709334640812613e-05, - "loss": 0.2404, - "step": 664 - }, - { - "epoch": 2.6936708860759495, - "grad_norm": 0.2942177303440855, - "learning_rate": 2.0652608894940824e-05, - "loss": 0.2314, - "step": 665 - }, - { - "epoch": 2.6977215189873416, - "grad_norm": 0.29885634544731843, - "learning_rate": 2.0595877893046722e-05, - "loss": 0.2455, - "step": 666 - }, - { - "epoch": 2.701772151898734, - "grad_norm": 0.304670070991131, - "learning_rate": 2.0539142092033985e-05, - "loss": 0.2272, - "step": 667 - }, - { - "epoch": 2.7058227848101266, - "grad_norm": 0.2959052841778297, - "learning_rate": 2.048240194884496e-05, - "loss": 0.2243, - "step": 668 - }, - { - "epoch": 2.709873417721519, - "grad_norm": 0.3341247923500945, - "learning_rate": 2.042565792045695e-05, - "loss": 0.2292, - "step": 669 - }, - { - "epoch": 2.7139240506329116, - "grad_norm": 0.3141099074496931, - "learning_rate": 2.036891046387857e-05, - "loss": 0.2354, - "step": 670 - }, - { - "epoch": 2.7179746835443037, - "grad_norm": 0.3242743822642043, - "learning_rate": 2.0312160036146036e-05, - "loss": 0.2301, - "step": 671 - }, - { - "epoch": 2.7220253164556962, - "grad_norm": 0.3332099266280377, - "learning_rate": 2.025540709431948e-05, - "loss": 0.2438, - "step": 672 - }, - { - "epoch": 2.7260759493670887, - "grad_norm": 0.3295532005466524, - "learning_rate": 2.0198652095479298e-05, - "loss": 0.2536, - "step": 673 - }, - { - "epoch": 2.730126582278481, - "grad_norm": 0.33331789931730993, - "learning_rate": 2.014189549672245e-05, - "loss": 0.2385, - "step": 674 - }, - { - "epoch": 2.7341772151898733, - "grad_norm": 0.3324722495958164, - "learning_rate": 2.0085137755158776e-05, - "loss": 0.2314, - "step": 675 - }, - { - "epoch": 2.738227848101266, - "grad_norm": 0.3236170103075277, - "learning_rate": 2.0028379327907327e-05, - "loss": 0.2412, - "step": 676 - }, - { - "epoch": 2.7422784810126584, - "grad_norm": 0.31790139845186743, - "learning_rate": 1.9971620672092676e-05, - "loss": 0.2241, - "step": 677 - }, - { - "epoch": 2.746329113924051, - "grad_norm": 0.30257758385673583, - "learning_rate": 1.991486224484123e-05, - "loss": 0.2187, - "step": 678 - }, - { - "epoch": 2.750379746835443, - "grad_norm": 0.30373666544695865, - "learning_rate": 1.985810450327756e-05, - "loss": 0.2287, - "step": 679 - }, - { - "epoch": 2.7544303797468355, - "grad_norm": 0.3085964878334211, - "learning_rate": 1.9801347904520706e-05, - "loss": 0.2401, - "step": 680 - }, - { - "epoch": 2.758481012658228, - "grad_norm": 0.27497435294521877, - "learning_rate": 1.974459290568053e-05, - "loss": 0.2092, - "step": 681 - }, - { - "epoch": 2.76253164556962, - "grad_norm": 0.3215781053713376, - "learning_rate": 1.968783996385397e-05, - "loss": 0.23, - "step": 682 - }, - { - "epoch": 2.7665822784810126, - "grad_norm": 0.3074045343659693, - "learning_rate": 1.963108953612143e-05, - "loss": 0.2359, - "step": 683 - }, - { - "epoch": 2.770632911392405, - "grad_norm": 0.2919988041419382, - "learning_rate": 1.9574342079543056e-05, - "loss": 0.2412, - "step": 684 - }, - { - "epoch": 2.7746835443037976, - "grad_norm": 0.32384774822913615, - "learning_rate": 1.9517598051155046e-05, - "loss": 0.2269, - "step": 685 - }, - { - "epoch": 2.77873417721519, - "grad_norm": 0.31168931913535425, - "learning_rate": 1.9460857907966025e-05, - "loss": 0.2288, - "step": 686 - }, - { - "epoch": 2.782784810126582, - "grad_norm": 0.27446813238880874, - "learning_rate": 1.9404122106953285e-05, - "loss": 0.2366, - "step": 687 - }, - { - "epoch": 2.7868354430379747, - "grad_norm": 0.338853314742463, - "learning_rate": 1.9347391105059176e-05, - "loss": 0.2222, - "step": 688 - }, - { - "epoch": 2.790886075949367, - "grad_norm": 0.31577529708831825, - "learning_rate": 1.92906653591874e-05, - "loss": 0.2372, - "step": 689 - }, - { - "epoch": 2.7949367088607593, - "grad_norm": 0.2858206562054224, - "learning_rate": 1.9233945326199295e-05, - "loss": 0.2357, - "step": 690 - }, - { - "epoch": 2.798987341772152, - "grad_norm": 0.31468871310816077, - "learning_rate": 1.917723146291022e-05, - "loss": 0.2356, - "step": 691 - }, - { - "epoch": 2.8030379746835443, - "grad_norm": 0.3220685990666963, - "learning_rate": 1.912052422608584e-05, - "loss": 0.2367, - "step": 692 - }, - { - "epoch": 2.807088607594937, - "grad_norm": 0.3049555792340489, - "learning_rate": 1.9063824072438428e-05, - "loss": 0.258, - "step": 693 - }, - { - "epoch": 2.8111392405063294, - "grad_norm": 0.32314734654617416, - "learning_rate": 1.9007131458623246e-05, - "loss": 0.2266, - "step": 694 - }, - { - "epoch": 2.8151898734177214, - "grad_norm": 0.33811331139430184, - "learning_rate": 1.895044684123479e-05, - "loss": 0.2377, - "step": 695 - }, - { - "epoch": 2.819240506329114, - "grad_norm": 0.2901367320286777, - "learning_rate": 1.8893770676803194e-05, - "loss": 0.221, - "step": 696 - }, - { - "epoch": 2.8232911392405065, - "grad_norm": 0.2923434569382237, - "learning_rate": 1.8837103421790486e-05, - "loss": 0.229, - "step": 697 - }, - { - "epoch": 2.8273417721518985, - "grad_norm": 0.34319938178192727, - "learning_rate": 1.8780445532586952e-05, - "loss": 0.235, - "step": 698 - }, - { - "epoch": 2.831392405063291, - "grad_norm": 0.29899766300741853, - "learning_rate": 1.872379746550743e-05, - "loss": 0.2236, - "step": 699 - }, - { - "epoch": 2.8354430379746836, - "grad_norm": 0.29920529165177634, - "learning_rate": 1.866715967678769e-05, - "loss": 0.2284, - "step": 700 - }, - { - "epoch": 2.839493670886076, - "grad_norm": 0.3203336061991541, - "learning_rate": 1.861053262258067e-05, - "loss": 0.2265, - "step": 701 - }, - { - "epoch": 2.8435443037974686, - "grad_norm": 0.2927777622315442, - "learning_rate": 1.8553916758952897e-05, - "loss": 0.2205, - "step": 702 - }, - { - "epoch": 2.8475949367088607, - "grad_norm": 0.29665631189719466, - "learning_rate": 1.8497312541880735e-05, - "loss": 0.2244, - "step": 703 - }, - { - "epoch": 2.851645569620253, - "grad_norm": 0.278546753617254, - "learning_rate": 1.8440720427246786e-05, - "loss": 0.2215, - "step": 704 - }, - { - "epoch": 2.8556962025316457, - "grad_norm": 0.27576053637433146, - "learning_rate": 1.8384140870836157e-05, - "loss": 0.2311, - "step": 705 - }, - { - "epoch": 2.8597468354430378, - "grad_norm": 0.28842207805624825, - "learning_rate": 1.8327574328332806e-05, - "loss": 0.2279, - "step": 706 - }, - { - "epoch": 2.8637974683544303, - "grad_norm": 0.2853800299665492, - "learning_rate": 1.8271021255315906e-05, - "loss": 0.2269, - "step": 707 - }, - { - "epoch": 2.867848101265823, - "grad_norm": 0.29992224522148414, - "learning_rate": 1.8214482107256117e-05, - "loss": 0.2327, - "step": 708 - }, - { - "epoch": 2.8718987341772153, - "grad_norm": 0.28498835940422096, - "learning_rate": 1.8157957339511968e-05, - "loss": 0.2166, - "step": 709 - }, - { - "epoch": 2.875949367088608, - "grad_norm": 0.28990475967898827, - "learning_rate": 1.8101447407326182e-05, - "loss": 0.2175, - "step": 710 - }, - { - "epoch": 2.88, - "grad_norm": 0.296267780247893, - "learning_rate": 1.8044952765821966e-05, - "loss": 0.2306, - "step": 711 - }, - { - "epoch": 2.8840506329113924, - "grad_norm": 0.27978590112411217, - "learning_rate": 1.7988473869999407e-05, - "loss": 0.2345, - "step": 712 - }, - { - "epoch": 2.888101265822785, - "grad_norm": 0.29646088061093234, - "learning_rate": 1.7932011174731764e-05, - "loss": 0.2295, - "step": 713 - }, - { - "epoch": 2.892151898734177, - "grad_norm": 0.28546996263123725, - "learning_rate": 1.7875565134761817e-05, - "loss": 0.2212, - "step": 714 - }, - { - "epoch": 2.8962025316455695, - "grad_norm": 0.2931315302261362, - "learning_rate": 1.7819136204698226e-05, - "loss": 0.2266, - "step": 715 - }, - { - "epoch": 2.900253164556962, - "grad_norm": 0.30361105735540095, - "learning_rate": 1.776272483901182e-05, - "loss": 0.2431, - "step": 716 - }, - { - "epoch": 2.9043037974683545, - "grad_norm": 0.287661548824662, - "learning_rate": 1.7706331492031995e-05, - "loss": 0.2248, - "step": 717 - }, - { - "epoch": 2.908354430379747, - "grad_norm": 0.2835093836474247, - "learning_rate": 1.764995661794301e-05, - "loss": 0.2291, - "step": 718 - }, - { - "epoch": 2.912405063291139, - "grad_norm": 0.29842256431288633, - "learning_rate": 1.759360067078035e-05, - "loss": 0.2149, - "step": 719 - }, - { - "epoch": 2.9164556962025316, - "grad_norm": 0.3030425200800445, - "learning_rate": 1.7537264104427064e-05, - "loss": 0.229, - "step": 720 - }, - { - "epoch": 2.920506329113924, - "grad_norm": 0.2854031957945021, - "learning_rate": 1.748094737261012e-05, - "loss": 0.2364, - "step": 721 - }, - { - "epoch": 2.9245569620253162, - "grad_norm": 0.2894403790882635, - "learning_rate": 1.7424650928896726e-05, - "loss": 0.2246, - "step": 722 - }, - { - "epoch": 2.9286075949367087, - "grad_norm": 0.3235362848666229, - "learning_rate": 1.7368375226690712e-05, - "loss": 0.2442, - "step": 723 - }, - { - "epoch": 2.9326582278481013, - "grad_norm": 0.2869789964507192, - "learning_rate": 1.731212071922883e-05, - "loss": 0.219, - "step": 724 - }, - { - "epoch": 2.9367088607594938, - "grad_norm": 0.2924959355001303, - "learning_rate": 1.7255887859577156e-05, - "loss": 0.2337, - "step": 725 - }, - { - "epoch": 2.9407594936708863, - "grad_norm": 0.3237147366649734, - "learning_rate": 1.7199677100627427e-05, - "loss": 0.2404, - "step": 726 - }, - { - "epoch": 2.9448101265822784, - "grad_norm": 0.2942224292902647, - "learning_rate": 1.7143488895093343e-05, - "loss": 0.2458, - "step": 727 - }, - { - "epoch": 2.948860759493671, - "grad_norm": 0.31212122955483235, - "learning_rate": 1.7087323695506994e-05, - "loss": 0.2306, - "step": 728 - }, - { - "epoch": 2.9529113924050634, - "grad_norm": 0.2792245722798537, - "learning_rate": 1.7031181954215194e-05, - "loss": 0.2237, - "step": 729 - }, - { - "epoch": 2.9569620253164555, - "grad_norm": 0.29860037140954665, - "learning_rate": 1.6975064123375788e-05, - "loss": 0.2352, - "step": 730 - }, - { - "epoch": 2.961012658227848, - "grad_norm": 0.2782302264236621, - "learning_rate": 1.6918970654954084e-05, - "loss": 0.2201, - "step": 731 - }, - { - "epoch": 2.9650632911392405, - "grad_norm": 0.28914524601595876, - "learning_rate": 1.686290200071915e-05, - "loss": 0.2422, - "step": 732 - }, - { - "epoch": 2.969113924050633, - "grad_norm": 0.28470459164380324, - "learning_rate": 1.6806858612240234e-05, - "loss": 0.222, - "step": 733 - }, - { - "epoch": 2.9731645569620255, - "grad_norm": 0.2679451075393547, - "learning_rate": 1.6750840940883078e-05, - "loss": 0.2119, - "step": 734 - }, - { - "epoch": 2.9772151898734176, - "grad_norm": 0.28370523303601225, - "learning_rate": 1.6694849437806305e-05, - "loss": 0.227, - "step": 735 - }, - { - "epoch": 2.98126582278481, - "grad_norm": 0.2690847647682115, - "learning_rate": 1.663888455395778e-05, - "loss": 0.2242, - "step": 736 - }, - { - "epoch": 2.9853164556962026, - "grad_norm": 0.2840641582274788, - "learning_rate": 1.6582946740070995e-05, - "loss": 0.2372, - "step": 737 - }, - { - "epoch": 2.9893670886075947, - "grad_norm": 0.28313134940882745, - "learning_rate": 1.6527036446661396e-05, - "loss": 0.2073, - "step": 738 - }, - { - "epoch": 2.993417721518987, - "grad_norm": 0.28042052444691495, - "learning_rate": 1.6471154124022818e-05, - "loss": 0.2292, - "step": 739 - }, - { - "epoch": 2.9974683544303797, - "grad_norm": 0.2780981778825185, - "learning_rate": 1.6415300222223788e-05, - "loss": 0.2199, - "step": 740 - }, - { - "epoch": 3.0015189873417722, - "grad_norm": 0.3060274509454366, - "learning_rate": 1.6359475191103958e-05, - "loss": 0.2063, - "step": 741 - }, - { - "epoch": 3.0055696202531648, - "grad_norm": 0.36200049117108607, - "learning_rate": 1.6303679480270466e-05, - "loss": 0.1735, - "step": 742 - }, - { - "epoch": 3.009620253164557, - "grad_norm": 0.29139065567740846, - "learning_rate": 1.624791353909428e-05, - "loss": 0.1651, - "step": 743 - }, - { - "epoch": 3.0136708860759494, - "grad_norm": 0.552809595589079, - "learning_rate": 1.619217781670663e-05, - "loss": 0.1622, - "step": 744 - }, - { - "epoch": 3.017721518987342, - "grad_norm": 0.3077459681145641, - "learning_rate": 1.6136472761995373e-05, - "loss": 0.1536, - "step": 745 - }, - { - "epoch": 3.0217721518987344, - "grad_norm": 0.39515053362386904, - "learning_rate": 1.608079882360134e-05, - "loss": 0.1654, - "step": 746 - }, - { - "epoch": 3.0258227848101265, - "grad_norm": 0.3667221489412283, - "learning_rate": 1.60251564499148e-05, - "loss": 0.145, - "step": 747 - }, - { - "epoch": 3.029873417721519, - "grad_norm": 0.30166240937442923, - "learning_rate": 1.596954608907176e-05, - "loss": 0.1504, - "step": 748 - }, - { - "epoch": 3.0339240506329115, - "grad_norm": 0.34557963011605447, - "learning_rate": 1.591396818895043e-05, - "loss": 0.178, - "step": 749 - }, - { - "epoch": 3.037974683544304, - "grad_norm": 0.32585853695194056, - "learning_rate": 1.585842319716759e-05, - "loss": 0.1623, - "step": 750 - }, - { - "epoch": 3.042025316455696, - "grad_norm": 0.2905305194110086, - "learning_rate": 1.5802911561074944e-05, - "loss": 0.1722, - "step": 751 - }, - { - "epoch": 3.0460759493670886, - "grad_norm": 0.30064416636759667, - "learning_rate": 1.5747433727755595e-05, - "loss": 0.1743, - "step": 752 - }, - { - "epoch": 3.050126582278481, - "grad_norm": 0.2866589520138042, - "learning_rate": 1.5691990144020376e-05, - "loss": 0.1601, - "step": 753 - }, - { - "epoch": 3.0541772151898736, - "grad_norm": 0.28816265537993224, - "learning_rate": 1.5636581256404297e-05, - "loss": 0.1737, - "step": 754 - }, - { - "epoch": 3.0582278481012657, - "grad_norm": 0.29241602740225925, - "learning_rate": 1.558120751116291e-05, - "loss": 0.1774, - "step": 755 - }, - { - "epoch": 3.062278481012658, - "grad_norm": 0.32084543061779264, - "learning_rate": 1.552586935426876e-05, - "loss": 0.1664, - "step": 756 - }, - { - "epoch": 3.0663291139240507, - "grad_norm": 0.3005010703992188, - "learning_rate": 1.547056723140774e-05, - "loss": 0.1351, - "step": 757 - }, - { - "epoch": 3.0703797468354432, - "grad_norm": 0.29300814112506124, - "learning_rate": 1.5415301587975565e-05, - "loss": 0.1658, - "step": 758 - }, - { - "epoch": 3.0744303797468353, - "grad_norm": 0.27870219956972886, - "learning_rate": 1.536007286907411e-05, - "loss": 0.1711, - "step": 759 - }, - { - "epoch": 3.078481012658228, - "grad_norm": 0.2720793934517818, - "learning_rate": 1.5304881519507896e-05, - "loss": 0.1596, - "step": 760 - }, - { - "epoch": 3.0825316455696203, - "grad_norm": 0.2899767816627878, - "learning_rate": 1.5249727983780453e-05, - "loss": 0.1657, - "step": 761 - }, - { - "epoch": 3.086582278481013, - "grad_norm": 0.2637903993928107, - "learning_rate": 1.5194612706090786e-05, - "loss": 0.1833, - "step": 762 - }, - { - "epoch": 3.090632911392405, - "grad_norm": 0.27524005418064934, - "learning_rate": 1.5139536130329771e-05, - "loss": 0.1585, - "step": 763 - }, - { - "epoch": 3.0946835443037974, - "grad_norm": 0.2749368536956698, - "learning_rate": 1.508449870007656e-05, - "loss": 0.159, - "step": 764 - }, - { - "epoch": 3.09873417721519, - "grad_norm": 0.29371154109453923, - "learning_rate": 1.5029500858595056e-05, - "loss": 0.1688, - "step": 765 - }, - { - "epoch": 3.1027848101265825, - "grad_norm": 0.2715792930644393, - "learning_rate": 1.4974543048830328e-05, - "loss": 0.1601, - "step": 766 - }, - { - "epoch": 3.1068354430379745, - "grad_norm": 0.2760931252285285, - "learning_rate": 1.4919625713405e-05, - "loss": 0.1683, - "step": 767 - }, - { - "epoch": 3.110886075949367, - "grad_norm": 0.2862316730292663, - "learning_rate": 1.4864749294615756e-05, - "loss": 0.1589, - "step": 768 - }, - { - "epoch": 3.1149367088607596, - "grad_norm": 0.2813630153689391, - "learning_rate": 1.4809914234429716e-05, - "loss": 0.167, - "step": 769 - }, - { - "epoch": 3.118987341772152, - "grad_norm": 0.2936114861512913, - "learning_rate": 1.4755120974480923e-05, - "loss": 0.1938, - "step": 770 - }, - { - "epoch": 3.123037974683544, - "grad_norm": 0.26514449715469973, - "learning_rate": 1.4700369956066771e-05, - "loss": 0.1719, - "step": 771 - }, - { - "epoch": 3.1270886075949367, - "grad_norm": 0.2970807274290257, - "learning_rate": 1.4645661620144413e-05, - "loss": 0.1677, - "step": 772 - }, - { - "epoch": 3.131139240506329, - "grad_norm": 0.28503235828734624, - "learning_rate": 1.4590996407327284e-05, - "loss": 0.1378, - "step": 773 - }, - { - "epoch": 3.1351898734177217, - "grad_norm": 0.2797121676564152, - "learning_rate": 1.4536374757881487e-05, - "loss": 0.1637, - "step": 774 - }, - { - "epoch": 3.1392405063291138, - "grad_norm": 0.2771464596220541, - "learning_rate": 1.4481797111722271e-05, - "loss": 0.1568, - "step": 775 - }, - { - "epoch": 3.1432911392405063, - "grad_norm": 0.2623454910146439, - "learning_rate": 1.4427263908410507e-05, - "loss": 0.1664, - "step": 776 - }, - { - "epoch": 3.147341772151899, - "grad_norm": 0.2688475800669211, - "learning_rate": 1.4372775587149108e-05, - "loss": 0.1648, - "step": 777 - }, - { - "epoch": 3.1513924050632913, - "grad_norm": 0.26394204865784615, - "learning_rate": 1.4318332586779522e-05, - "loss": 0.1523, - "step": 778 - }, - { - "epoch": 3.1554430379746834, - "grad_norm": 0.2843726582000157, - "learning_rate": 1.4263935345778202e-05, - "loss": 0.1476, - "step": 779 - }, - { - "epoch": 3.159493670886076, - "grad_norm": 0.2813034614077088, - "learning_rate": 1.420958430225303e-05, - "loss": 0.1754, - "step": 780 - }, - { - "epoch": 3.1635443037974684, - "grad_norm": 0.28225866238992053, - "learning_rate": 1.415527989393985e-05, - "loss": 0.1762, - "step": 781 - }, - { - "epoch": 3.167594936708861, - "grad_norm": 0.29136345317324325, - "learning_rate": 1.410102255819891e-05, - "loss": 0.1508, - "step": 782 - }, - { - "epoch": 3.171645569620253, - "grad_norm": 0.28828969148301964, - "learning_rate": 1.404681273201131e-05, - "loss": 0.1615, - "step": 783 - }, - { - "epoch": 3.1756962025316455, - "grad_norm": 0.2824882045613792, - "learning_rate": 1.399265085197556e-05, - "loss": 0.1549, - "step": 784 - }, - { - "epoch": 3.179746835443038, - "grad_norm": 0.28164846694763584, - "learning_rate": 1.393853735430398e-05, - "loss": 0.158, - "step": 785 - }, - { - "epoch": 3.1837974683544306, - "grad_norm": 0.2715068507902323, - "learning_rate": 1.3884472674819246e-05, - "loss": 0.1784, - "step": 786 - }, - { - "epoch": 3.1878481012658226, - "grad_norm": 0.2826962543702162, - "learning_rate": 1.3830457248950864e-05, - "loss": 0.1574, - "step": 787 - }, - { - "epoch": 3.191898734177215, - "grad_norm": 0.2771333748317222, - "learning_rate": 1.377649151173163e-05, - "loss": 0.1586, - "step": 788 - }, - { - "epoch": 3.1959493670886077, - "grad_norm": 0.2923828090801464, - "learning_rate": 1.3722575897794181e-05, - "loss": 0.1542, - "step": 789 - }, - { - "epoch": 3.2, - "grad_norm": 0.2721248174056244, - "learning_rate": 1.3668710841367472e-05, - "loss": 0.1532, - "step": 790 - }, - { - "epoch": 3.2040506329113922, - "grad_norm": 0.28059024731526755, - "learning_rate": 1.361489677627324e-05, - "loss": 0.1435, - "step": 791 - }, - { - "epoch": 3.2081012658227848, - "grad_norm": 0.29150260293484526, - "learning_rate": 1.3561134135922585e-05, - "loss": 0.1692, - "step": 792 - }, - { - "epoch": 3.2121518987341773, - "grad_norm": 0.274049392809611, - "learning_rate": 1.350742335331241e-05, - "loss": 0.1606, - "step": 793 - }, - { - "epoch": 3.21620253164557, - "grad_norm": 0.288976421924057, - "learning_rate": 1.345376486102198e-05, - "loss": 0.1674, - "step": 794 - }, - { - "epoch": 3.220253164556962, - "grad_norm": 0.2667104566333963, - "learning_rate": 1.3400159091209414e-05, - "loss": 0.1561, - "step": 795 - }, - { - "epoch": 3.2243037974683544, - "grad_norm": 0.26712839050702614, - "learning_rate": 1.3346606475608216e-05, - "loss": 0.1639, - "step": 796 - }, - { - "epoch": 3.228354430379747, - "grad_norm": 0.27315109482276695, - "learning_rate": 1.3293107445523781e-05, - "loss": 0.1688, - "step": 797 - }, - { - "epoch": 3.2324050632911394, - "grad_norm": 0.26405267634975127, - "learning_rate": 1.3239662431829949e-05, - "loss": 0.1566, - "step": 798 - }, - { - "epoch": 3.2364556962025315, - "grad_norm": 0.26998958631664255, - "learning_rate": 1.3186271864965509e-05, - "loss": 0.1483, - "step": 799 - }, - { - "epoch": 3.240506329113924, - "grad_norm": 0.26731679471473246, - "learning_rate": 1.3132936174930756e-05, - "loss": 0.1643, - "step": 800 - }, - { - "epoch": 3.2445569620253165, - "grad_norm": 0.2710451129219873, - "learning_rate": 1.3079655791283995e-05, - "loss": 0.1542, - "step": 801 - }, - { - "epoch": 3.248607594936709, - "grad_norm": 0.26708794900151694, - "learning_rate": 1.3026431143138108e-05, - "loss": 0.169, - "step": 802 - }, - { - "epoch": 3.252658227848101, - "grad_norm": 0.2824829343783079, - "learning_rate": 1.2973262659157114e-05, - "loss": 0.1709, - "step": 803 - }, - { - "epoch": 3.2567088607594936, - "grad_norm": 0.26994341552354434, - "learning_rate": 1.2920150767552651e-05, - "loss": 0.1576, - "step": 804 - }, - { - "epoch": 3.260759493670886, - "grad_norm": 0.28109100373791535, - "learning_rate": 1.2867095896080607e-05, - "loss": 0.18, - "step": 805 - }, - { - "epoch": 3.2648101265822786, - "grad_norm": 0.2702272439526788, - "learning_rate": 1.2814098472037612e-05, - "loss": 0.1572, - "step": 806 - }, - { - "epoch": 3.2688607594936707, - "grad_norm": 0.26364607289168573, - "learning_rate": 1.276115892225764e-05, - "loss": 0.1753, - "step": 807 - }, - { - "epoch": 3.2729113924050632, - "grad_norm": 0.25186717760782223, - "learning_rate": 1.2708277673108555e-05, - "loss": 0.1636, - "step": 808 - }, - { - "epoch": 3.2769620253164558, - "grad_norm": 0.28016137440435945, - "learning_rate": 1.2655455150488649e-05, - "loss": 0.1607, - "step": 809 - }, - { - "epoch": 3.2810126582278483, - "grad_norm": 0.25492211703388984, - "learning_rate": 1.2602691779823272e-05, - "loss": 0.1637, - "step": 810 - }, - { - "epoch": 3.2850632911392403, - "grad_norm": 0.26982822566380793, - "learning_rate": 1.2549987986061355e-05, - "loss": 0.1705, - "step": 811 - }, - { - "epoch": 3.289113924050633, - "grad_norm": 0.2765062465349065, - "learning_rate": 1.2497344193672005e-05, - "loss": 0.1864, - "step": 812 - }, - { - "epoch": 3.2931645569620254, - "grad_norm": 0.2832962810055447, - "learning_rate": 1.2444760826641092e-05, - "loss": 0.1603, - "step": 813 - }, - { - "epoch": 3.297215189873418, - "grad_norm": 0.27482883677138903, - "learning_rate": 1.2392238308467817e-05, - "loss": 0.1673, - "step": 814 - }, - { - "epoch": 3.30126582278481, - "grad_norm": 0.2754201565091245, - "learning_rate": 1.2339777062161326e-05, - "loss": 0.1734, - "step": 815 - }, - { - "epoch": 3.3053164556962025, - "grad_norm": 0.2727530905418694, - "learning_rate": 1.2287377510237293e-05, - "loss": 0.1676, - "step": 816 - }, - { - "epoch": 3.309367088607595, - "grad_norm": 0.2632611201992197, - "learning_rate": 1.2235040074714488e-05, - "loss": 0.1608, - "step": 817 - }, - { - "epoch": 3.3134177215189875, - "grad_norm": 0.2746162876177735, - "learning_rate": 1.2182765177111434e-05, - "loss": 0.1715, - "step": 818 - }, - { - "epoch": 3.3174683544303796, - "grad_norm": 0.2664981827021515, - "learning_rate": 1.213055323844297e-05, - "loss": 0.1604, - "step": 819 - }, - { - "epoch": 3.321518987341772, - "grad_norm": 0.26182640199339563, - "learning_rate": 1.2078404679216864e-05, - "loss": 0.1509, - "step": 820 - }, - { - "epoch": 3.3255696202531646, - "grad_norm": 0.27011924857934344, - "learning_rate": 1.2026319919430458e-05, - "loss": 0.1559, - "step": 821 - }, - { - "epoch": 3.329620253164557, - "grad_norm": 0.2665968876031193, - "learning_rate": 1.1974299378567227e-05, - "loss": 0.1617, - "step": 822 - }, - { - "epoch": 3.333670886075949, - "grad_norm": 0.2531737345534062, - "learning_rate": 1.1922343475593462e-05, - "loss": 0.1547, - "step": 823 - }, - { - "epoch": 3.3377215189873417, - "grad_norm": 0.26813349453528385, - "learning_rate": 1.187045262895488e-05, - "loss": 0.1755, - "step": 824 - }, - { - "epoch": 3.3417721518987342, - "grad_norm": 0.2612534758453089, - "learning_rate": 1.1818627256573203e-05, - "loss": 0.1591, - "step": 825 - }, - { - "epoch": 3.3458227848101267, - "grad_norm": 0.2532972604472537, - "learning_rate": 1.1766867775842864e-05, - "loss": 0.1467, - "step": 826 - }, - { - "epoch": 3.349873417721519, - "grad_norm": 0.2617337455859381, - "learning_rate": 1.1715174603627615e-05, - "loss": 0.1505, - "step": 827 - }, - { - "epoch": 3.3539240506329113, - "grad_norm": 0.2734240076400232, - "learning_rate": 1.1663548156257147e-05, - "loss": 0.151, - "step": 828 - }, - { - "epoch": 3.357974683544304, - "grad_norm": 0.2605088936753551, - "learning_rate": 1.161198884952377e-05, - "loss": 0.1549, - "step": 829 - }, - { - "epoch": 3.3620253164556964, - "grad_norm": 0.25811178491616976, - "learning_rate": 1.1560497098679056e-05, - "loss": 0.153, - "step": 830 - }, - { - "epoch": 3.3660759493670884, - "grad_norm": 0.264060797707371, - "learning_rate": 1.1509073318430479e-05, - "loss": 0.1572, - "step": 831 - }, - { - "epoch": 3.370126582278481, - "grad_norm": 0.2637885910448637, - "learning_rate": 1.1457717922938116e-05, - "loss": 0.1582, - "step": 832 - }, - { - "epoch": 3.3741772151898735, - "grad_norm": 0.26511174310555086, - "learning_rate": 1.1406431325811233e-05, - "loss": 0.1489, - "step": 833 - }, - { - "epoch": 3.378227848101266, - "grad_norm": 0.2569367447847199, - "learning_rate": 1.135521394010506e-05, - "loss": 0.1553, - "step": 834 - }, - { - "epoch": 3.382278481012658, - "grad_norm": 0.25883756922850437, - "learning_rate": 1.1304066178317367e-05, - "loss": 0.1566, - "step": 835 - }, - { - "epoch": 3.3863291139240506, - "grad_norm": 0.26305626535383103, - "learning_rate": 1.1252988452385199e-05, - "loss": 0.1741, - "step": 836 - }, - { - "epoch": 3.390379746835443, - "grad_norm": 0.2476710870101896, - "learning_rate": 1.1201981173681536e-05, - "loss": 0.1863, - "step": 837 - }, - { - "epoch": 3.3944303797468356, - "grad_norm": 0.2795290027009559, - "learning_rate": 1.1151044753011991e-05, - "loss": 0.1443, - "step": 838 - }, - { - "epoch": 3.3984810126582277, - "grad_norm": 0.2697790172731661, - "learning_rate": 1.1100179600611491e-05, - "loss": 0.1656, - "step": 839 - }, - { - "epoch": 3.40253164556962, - "grad_norm": 0.26452246824122355, - "learning_rate": 1.1049386126140985e-05, - "loss": 0.1714, - "step": 840 - }, - { - "epoch": 3.4065822784810127, - "grad_norm": 0.2666929959990581, - "learning_rate": 1.0998664738684128e-05, - "loss": 0.1706, - "step": 841 - }, - { - "epoch": 3.410632911392405, - "grad_norm": 0.25790067564700403, - "learning_rate": 1.0948015846744e-05, - "loss": 0.1547, - "step": 842 - }, - { - "epoch": 3.4146835443037973, - "grad_norm": 0.27596472166554403, - "learning_rate": 1.0897439858239832e-05, - "loss": 0.1525, - "step": 843 - }, - { - "epoch": 3.41873417721519, - "grad_norm": 0.2632876978898246, - "learning_rate": 1.0846937180503652e-05, - "loss": 0.1603, - "step": 844 - }, - { - "epoch": 3.4227848101265823, - "grad_norm": 0.2616944596796812, - "learning_rate": 1.0796508220277117e-05, - "loss": 0.1604, - "step": 845 - }, - { - "epoch": 3.426835443037975, - "grad_norm": 0.26895280100145114, - "learning_rate": 1.0746153383708107e-05, - "loss": 0.15, - "step": 846 - }, - { - "epoch": 3.430886075949367, - "grad_norm": 0.2603983195953071, - "learning_rate": 1.0695873076347579e-05, - "loss": 0.1683, - "step": 847 - }, - { - "epoch": 3.4349367088607594, - "grad_norm": 0.25783974892111267, - "learning_rate": 1.0645667703146205e-05, - "loss": 0.1621, - "step": 848 - }, - { - "epoch": 3.438987341772152, - "grad_norm": 0.2664377200624605, - "learning_rate": 1.0595537668451161e-05, - "loss": 0.1646, - "step": 849 - }, - { - "epoch": 3.4430379746835444, - "grad_norm": 0.2507921000791203, - "learning_rate": 1.0545483376002854e-05, - "loss": 0.1761, - "step": 850 - }, - { - "epoch": 3.4470886075949365, - "grad_norm": 0.24727966359970321, - "learning_rate": 1.0495505228931676e-05, - "loss": 0.1575, - "step": 851 - }, - { - "epoch": 3.451139240506329, - "grad_norm": 0.2503863859844233, - "learning_rate": 1.044560362975474e-05, - "loss": 0.1638, - "step": 852 - }, - { - "epoch": 3.4551898734177215, - "grad_norm": 0.25425801714150653, - "learning_rate": 1.0395778980372695e-05, - "loss": 0.1704, - "step": 853 - }, - { - "epoch": 3.459240506329114, - "grad_norm": 0.25099140954760923, - "learning_rate": 1.0346031682066381e-05, - "loss": 0.179, - "step": 854 - }, - { - "epoch": 3.463291139240506, - "grad_norm": 0.2562649829244321, - "learning_rate": 1.0296362135493724e-05, - "loss": 0.1534, - "step": 855 - }, - { - "epoch": 3.4673417721518986, - "grad_norm": 0.2504488814660078, - "learning_rate": 1.0246770740686422e-05, - "loss": 0.1508, - "step": 856 - }, - { - "epoch": 3.471392405063291, - "grad_norm": 0.24980955902959262, - "learning_rate": 1.0197257897046743e-05, - "loss": 0.1645, - "step": 857 - }, - { - "epoch": 3.4754430379746837, - "grad_norm": 0.25644277016623085, - "learning_rate": 1.014782400334433e-05, - "loss": 0.1602, - "step": 858 - }, - { - "epoch": 3.479493670886076, - "grad_norm": 0.2736284721150555, - "learning_rate": 1.009846945771296e-05, - "loss": 0.1468, - "step": 859 - }, - { - "epoch": 3.4835443037974683, - "grad_norm": 0.26431620043307835, - "learning_rate": 1.0049194657647363e-05, - "loss": 0.161, - "step": 860 - }, - { - "epoch": 3.487594936708861, - "grad_norm": 0.25157764784583264, - "learning_rate": 1.0000000000000006e-05, - "loss": 0.1769, - "step": 861 - }, - { - "epoch": 3.4916455696202533, - "grad_norm": 0.2513244077822873, - "learning_rate": 9.950885880977891e-06, - "loss": 0.1495, - "step": 862 - }, - { - "epoch": 3.4956962025316454, - "grad_norm": 0.2600404661140625, - "learning_rate": 9.901852696139382e-06, - "loss": 0.1499, - "step": 863 - }, - { - "epoch": 3.499746835443038, - "grad_norm": 0.24766211299666183, - "learning_rate": 9.852900840391027e-06, - "loss": 0.1601, - "step": 864 - }, - { - "epoch": 3.5037974683544304, - "grad_norm": 0.25459478267005264, - "learning_rate": 9.804030707984313e-06, - "loss": 0.1666, - "step": 865 - }, - { - "epoch": 3.507848101265823, - "grad_norm": 0.25551337650680533, - "learning_rate": 9.755242692512599e-06, - "loss": 0.1522, - "step": 866 - }, - { - "epoch": 3.5118987341772154, - "grad_norm": 0.2686448292917107, - "learning_rate": 9.70653718690782e-06, - "loss": 0.1537, - "step": 867 - }, - { - "epoch": 3.5159493670886075, - "grad_norm": 0.2637220992910812, - "learning_rate": 9.657914583437454e-06, - "loss": 0.1583, - "step": 868 - }, - { - "epoch": 3.52, - "grad_norm": 0.27194163884586703, - "learning_rate": 9.609375273701246e-06, - "loss": 0.1616, - "step": 869 - }, - { - "epoch": 3.5240506329113925, - "grad_norm": 0.26587082551037833, - "learning_rate": 9.560919648628133e-06, - "loss": 0.1609, - "step": 870 - }, - { - "epoch": 3.5281012658227846, - "grad_norm": 0.2598823215000778, - "learning_rate": 9.512548098473047e-06, - "loss": 0.1714, - "step": 871 - }, - { - "epoch": 3.532151898734177, - "grad_norm": 0.26568758375422563, - "learning_rate": 9.464261012813825e-06, - "loss": 0.1572, - "step": 872 - }, - { - "epoch": 3.5362025316455696, - "grad_norm": 0.2580110536237401, - "learning_rate": 9.416058780547987e-06, - "loss": 0.1604, - "step": 873 - }, - { - "epoch": 3.540253164556962, - "grad_norm": 0.26724647140895236, - "learning_rate": 9.367941789889714e-06, - "loss": 0.1565, - "step": 874 - }, - { - "epoch": 3.5443037974683547, - "grad_norm": 0.2554052505751437, - "learning_rate": 9.319910428366607e-06, - "loss": 0.1583, - "step": 875 - }, - { - "epoch": 3.5483544303797467, - "grad_norm": 0.27077633235334864, - "learning_rate": 9.271965082816667e-06, - "loss": 0.1667, - "step": 876 - }, - { - "epoch": 3.5524050632911393, - "grad_norm": 0.24902228668988838, - "learning_rate": 9.224106139385111e-06, - "loss": 0.1617, - "step": 877 - }, - { - "epoch": 3.5564556962025318, - "grad_norm": 0.25829664660334706, - "learning_rate": 9.176333983521291e-06, - "loss": 0.163, - "step": 878 - }, - { - "epoch": 3.560506329113924, - "grad_norm": 0.25982362872171905, - "learning_rate": 9.12864899997558e-06, - "loss": 0.1645, - "step": 879 - }, - { - "epoch": 3.5645569620253164, - "grad_norm": 0.27312349793829543, - "learning_rate": 9.08105157279628e-06, - "loss": 0.1492, - "step": 880 - }, - { - "epoch": 3.568607594936709, - "grad_norm": 0.2629289373979584, - "learning_rate": 9.03354208532653e-06, - "loss": 0.1611, - "step": 881 - }, - { - "epoch": 3.5726582278481014, - "grad_norm": 0.26342271900968806, - "learning_rate": 8.986120920201205e-06, - "loss": 0.1725, - "step": 882 - }, - { - "epoch": 3.576708860759494, - "grad_norm": 0.2573326384384912, - "learning_rate": 8.938788459343852e-06, - "loss": 0.1503, - "step": 883 - }, - { - "epoch": 3.580759493670886, - "grad_norm": 0.254933662034643, - "learning_rate": 8.8915450839636e-06, - "loss": 0.1447, - "step": 884 - }, - { - "epoch": 3.5848101265822785, - "grad_norm": 0.26528432513607414, - "learning_rate": 8.844391174552116e-06, - "loss": 0.1471, - "step": 885 - }, - { - "epoch": 3.588860759493671, - "grad_norm": 0.24412207701039962, - "learning_rate": 8.797327110880479e-06, - "loss": 0.1616, - "step": 886 - }, - { - "epoch": 3.592911392405063, - "grad_norm": 0.25765974263196645, - "learning_rate": 8.750353271996206e-06, - "loss": 0.157, - "step": 887 - }, - { - "epoch": 3.5969620253164556, - "grad_norm": 0.26300833085224196, - "learning_rate": 8.703470036220132e-06, - "loss": 0.1571, - "step": 888 - }, - { - "epoch": 3.601012658227848, - "grad_norm": 0.2543082166393287, - "learning_rate": 8.656677781143394e-06, - "loss": 0.154, - "step": 889 - }, - { - "epoch": 3.6050632911392406, - "grad_norm": 0.2646333150767995, - "learning_rate": 8.609976883624377e-06, - "loss": 0.1627, - "step": 890 - }, - { - "epoch": 3.609113924050633, - "grad_norm": 0.26294403036811204, - "learning_rate": 8.563367719785698e-06, - "loss": 0.1562, - "step": 891 - }, - { - "epoch": 3.613164556962025, - "grad_norm": 0.2537339884786861, - "learning_rate": 8.516850665011138e-06, - "loss": 0.1703, - "step": 892 - }, - { - "epoch": 3.6172151898734177, - "grad_norm": 0.2503776872498334, - "learning_rate": 8.47042609394269e-06, - "loss": 0.1782, - "step": 893 - }, - { - "epoch": 3.6212658227848102, - "grad_norm": 0.24695938707334175, - "learning_rate": 8.424094380477432e-06, - "loss": 0.1745, - "step": 894 - }, - { - "epoch": 3.6253164556962023, - "grad_norm": 0.2585438113825923, - "learning_rate": 8.37785589776465e-06, - "loss": 0.1553, - "step": 895 - }, - { - "epoch": 3.629367088607595, - "grad_norm": 0.25144708706099395, - "learning_rate": 8.331711018202694e-06, - "loss": 0.1536, - "step": 896 - }, - { - "epoch": 3.6334177215189873, - "grad_norm": 0.25498137052430847, - "learning_rate": 8.285660113436104e-06, - "loss": 0.164, - "step": 897 - }, - { - "epoch": 3.63746835443038, - "grad_norm": 0.24662897995461333, - "learning_rate": 8.239703554352527e-06, - "loss": 0.1558, - "step": 898 - }, - { - "epoch": 3.6415189873417724, - "grad_norm": 0.25294462267597984, - "learning_rate": 8.193841711079775e-06, - "loss": 0.1656, - "step": 899 - }, - { - "epoch": 3.6455696202531644, - "grad_norm": 0.26255345053131307, - "learning_rate": 8.148074952982828e-06, - "loss": 0.1566, - "step": 900 - }, - { - "epoch": 3.649620253164557, - "grad_norm": 0.25137851755900076, - "learning_rate": 8.102403648660859e-06, - "loss": 0.1537, - "step": 901 - }, - { - "epoch": 3.6536708860759495, - "grad_norm": 0.25173240400442926, - "learning_rate": 8.056828165944282e-06, - "loss": 0.1683, - "step": 902 - }, - { - "epoch": 3.6577215189873415, - "grad_norm": 0.26893119901402424, - "learning_rate": 8.011348871891762e-06, - "loss": 0.1476, - "step": 903 - }, - { - "epoch": 3.661772151898734, - "grad_norm": 0.25370461352083684, - "learning_rate": 7.965966132787287e-06, - "loss": 0.1517, - "step": 904 - }, - { - "epoch": 3.6658227848101266, - "grad_norm": 0.2592657674137614, - "learning_rate": 7.920680314137189e-06, - "loss": 0.1543, - "step": 905 - }, - { - "epoch": 3.669873417721519, - "grad_norm": 0.26904350172396935, - "learning_rate": 7.875491780667246e-06, - "loss": 0.1579, - "step": 906 - }, - { - "epoch": 3.6739240506329116, - "grad_norm": 0.2563642222287851, - "learning_rate": 7.830400896319667e-06, - "loss": 0.1634, - "step": 907 - }, - { - "epoch": 3.6779746835443037, - "grad_norm": 0.252001696487572, - "learning_rate": 7.785408024250259e-06, - "loss": 0.172, - "step": 908 - }, - { - "epoch": 3.682025316455696, - "grad_norm": 0.25501007372280865, - "learning_rate": 7.74051352682542e-06, - "loss": 0.145, - "step": 909 - }, - { - "epoch": 3.6860759493670887, - "grad_norm": 0.25574545342675387, - "learning_rate": 7.695717765619257e-06, - "loss": 0.1576, - "step": 910 - }, - { - "epoch": 3.690126582278481, - "grad_norm": 0.2689826734109508, - "learning_rate": 7.651021101410673e-06, - "loss": 0.1681, - "step": 911 - }, - { - "epoch": 3.6941772151898733, - "grad_norm": 0.26280527011809796, - "learning_rate": 7.606423894180464e-06, - "loss": 0.152, - "step": 912 - }, - { - "epoch": 3.698227848101266, - "grad_norm": 0.24575540960862094, - "learning_rate": 7.56192650310839e-06, - "loss": 0.1738, - "step": 913 - }, - { - "epoch": 3.7022784810126583, - "grad_norm": 0.26201779693822813, - "learning_rate": 7.517529286570349e-06, - "loss": 0.1829, - "step": 914 - }, - { - "epoch": 3.706329113924051, - "grad_norm": 0.25715506327145493, - "learning_rate": 7.473232602135387e-06, - "loss": 0.1571, - "step": 915 - }, - { - "epoch": 3.710379746835443, - "grad_norm": 0.25651673921895973, - "learning_rate": 7.429036806562935e-06, - "loss": 0.1542, - "step": 916 - }, - { - "epoch": 3.7144303797468354, - "grad_norm": 0.26890002819348385, - "learning_rate": 7.3849422557998455e-06, - "loss": 0.151, - "step": 917 - }, - { - "epoch": 3.718481012658228, - "grad_norm": 0.23887823313256512, - "learning_rate": 7.340949304977567e-06, - "loss": 0.1638, - "step": 918 - }, - { - "epoch": 3.72253164556962, - "grad_norm": 0.2609412277247185, - "learning_rate": 7.297058308409282e-06, - "loss": 0.1687, - "step": 919 - }, - { - "epoch": 3.7265822784810125, - "grad_norm": 0.24915967002903086, - "learning_rate": 7.25326961958704e-06, - "loss": 0.1519, - "step": 920 - }, - { - "epoch": 3.730632911392405, - "grad_norm": 0.24613298106351236, - "learning_rate": 7.209583591178921e-06, - "loss": 0.1785, - "step": 921 - }, - { - "epoch": 3.7346835443037976, - "grad_norm": 0.2549832671818719, - "learning_rate": 7.1660005750261925e-06, - "loss": 0.1534, - "step": 922 - }, - { - "epoch": 3.73873417721519, - "grad_norm": 0.2588558511375635, - "learning_rate": 7.1225209221404765e-06, - "loss": 0.1678, - "step": 923 - }, - { - "epoch": 3.742784810126582, - "grad_norm": 0.2595025463453933, - "learning_rate": 7.079144982700909e-06, - "loss": 0.164, - "step": 924 - }, - { - "epoch": 3.7468354430379747, - "grad_norm": 0.25301945521507274, - "learning_rate": 7.0358731060513695e-06, - "loss": 0.1468, - "step": 925 - }, - { - "epoch": 3.750886075949367, - "grad_norm": 0.2682988450021254, - "learning_rate": 6.99270564069757e-06, - "loss": 0.165, - "step": 926 - }, - { - "epoch": 3.7549367088607593, - "grad_norm": 0.2651277878082875, - "learning_rate": 6.949642934304375e-06, - "loss": 0.1611, - "step": 927 - }, - { - "epoch": 3.7589873417721518, - "grad_norm": 0.27212629687356005, - "learning_rate": 6.906685333692871e-06, - "loss": 0.1559, - "step": 928 - }, - { - "epoch": 3.7630379746835443, - "grad_norm": 0.25918895431616457, - "learning_rate": 6.86383318483769e-06, - "loss": 0.175, - "step": 929 - }, - { - "epoch": 3.767088607594937, - "grad_norm": 0.2513515923164042, - "learning_rate": 6.821086832864139e-06, - "loss": 0.1643, - "step": 930 - }, - { - "epoch": 3.7711392405063293, - "grad_norm": 0.2636534796012303, - "learning_rate": 6.77844662204546e-06, - "loss": 0.1539, - "step": 931 - }, - { - "epoch": 3.7751898734177214, - "grad_norm": 0.2590482063373185, - "learning_rate": 6.7359128958000455e-06, - "loss": 0.1675, - "step": 932 - }, - { - "epoch": 3.779240506329114, - "grad_norm": 0.24214244261885493, - "learning_rate": 6.693485996688695e-06, - "loss": 0.1708, - "step": 933 - }, - { - "epoch": 3.7832911392405064, - "grad_norm": 0.24885286145042437, - "learning_rate": 6.651166266411801e-06, - "loss": 0.1454, - "step": 934 - }, - { - "epoch": 3.7873417721518985, - "grad_norm": 0.25639296535596184, - "learning_rate": 6.6089540458066725e-06, - "loss": 0.1534, - "step": 935 - }, - { - "epoch": 3.791392405063291, - "grad_norm": 0.2454823023479455, - "learning_rate": 6.566849674844711e-06, - "loss": 0.1487, - "step": 936 - }, - { - "epoch": 3.7954430379746835, - "grad_norm": 0.24995289547525043, - "learning_rate": 6.524853492628747e-06, - "loss": 0.1661, - "step": 937 - }, - { - "epoch": 3.799493670886076, - "grad_norm": 0.24871013416158672, - "learning_rate": 6.4829658373902536e-06, - "loss": 0.1483, - "step": 938 - }, - { - "epoch": 3.8035443037974686, - "grad_norm": 0.2564676971680992, - "learning_rate": 6.441187046486648e-06, - "loss": 0.1449, - "step": 939 - }, - { - "epoch": 3.8075949367088606, - "grad_norm": 0.25408841558086237, - "learning_rate": 6.399517456398567e-06, - "loss": 0.151, - "step": 940 - }, - { - "epoch": 3.811645569620253, - "grad_norm": 0.24553866719178202, - "learning_rate": 6.357957402727164e-06, - "loss": 0.1386, - "step": 941 - }, - { - "epoch": 3.8156962025316457, - "grad_norm": 0.25238766164802734, - "learning_rate": 6.316507220191395e-06, - "loss": 0.1491, - "step": 942 - }, - { - "epoch": 3.8197468354430377, - "grad_norm": 0.24249730208629064, - "learning_rate": 6.275167242625331e-06, - "loss": 0.1653, - "step": 943 - }, - { - "epoch": 3.8237974683544302, - "grad_norm": 0.25595514239465644, - "learning_rate": 6.233937802975471e-06, - "loss": 0.1615, - "step": 944 - }, - { - "epoch": 3.8278481012658228, - "grad_norm": 0.25168076676429946, - "learning_rate": 6.192819233298046e-06, - "loss": 0.1636, - "step": 945 - }, - { - "epoch": 3.8318987341772153, - "grad_norm": 0.24912997906881862, - "learning_rate": 6.151811864756383e-06, - "loss": 0.1681, - "step": 946 - }, - { - "epoch": 3.835949367088608, - "grad_norm": 0.2586493335469311, - "learning_rate": 6.1109160276181655e-06, - "loss": 0.1292, - "step": 947 - }, - { - "epoch": 3.84, - "grad_norm": 0.2510417615265298, - "learning_rate": 6.070132051252868e-06, - "loss": 0.1637, - "step": 948 - }, - { - "epoch": 3.8440506329113924, - "grad_norm": 0.24791258276792308, - "learning_rate": 6.0294602641290034e-06, - "loss": 0.1771, - "step": 949 - }, - { - "epoch": 3.848101265822785, - "grad_norm": 0.25194700112199614, - "learning_rate": 5.988900993811575e-06, - "loss": 0.1741, - "step": 950 - }, - { - "epoch": 3.852151898734177, - "grad_norm": 0.23820939426186255, - "learning_rate": 5.948454566959363e-06, - "loss": 0.164, - "step": 951 - }, - { - "epoch": 3.8562025316455695, - "grad_norm": 0.2408436276487035, - "learning_rate": 5.908121309322328e-06, - "loss": 0.1511, - "step": 952 - }, - { - "epoch": 3.860253164556962, - "grad_norm": 0.24745522516161628, - "learning_rate": 5.867901545738976e-06, - "loss": 0.1545, - "step": 953 - }, - { - "epoch": 3.8643037974683545, - "grad_norm": 0.2488410445991274, - "learning_rate": 5.827795600133774e-06, - "loss": 0.1735, - "step": 954 - }, - { - "epoch": 3.868354430379747, - "grad_norm": 0.24700430953241254, - "learning_rate": 5.787803795514466e-06, - "loss": 0.156, - "step": 955 - }, - { - "epoch": 3.872405063291139, - "grad_norm": 0.24172489200686792, - "learning_rate": 5.747926453969576e-06, - "loss": 0.156, - "step": 956 - }, - { - "epoch": 3.8764556962025316, - "grad_norm": 0.25016605363013683, - "learning_rate": 5.708163896665708e-06, - "loss": 0.1503, - "step": 957 - }, - { - "epoch": 3.880506329113924, - "grad_norm": 0.2505268401010105, - "learning_rate": 5.668516443845047e-06, - "loss": 0.1514, - "step": 958 - }, - { - "epoch": 3.884556962025316, - "grad_norm": 0.23397148882392907, - "learning_rate": 5.6289844148227225e-06, - "loss": 0.1547, - "step": 959 - }, - { - "epoch": 3.8886075949367087, - "grad_norm": 0.24839991533585884, - "learning_rate": 5.5895681279842615e-06, - "loss": 0.166, - "step": 960 - }, - { - "epoch": 3.8926582278481012, - "grad_norm": 0.2515908714602789, - "learning_rate": 5.550267900783019e-06, - "loss": 0.1564, - "step": 961 - }, - { - "epoch": 3.8967088607594937, - "grad_norm": 0.24054860826873142, - "learning_rate": 5.511084049737623e-06, - "loss": 0.1506, - "step": 962 - }, - { - "epoch": 3.9007594936708863, - "grad_norm": 0.24243524633870642, - "learning_rate": 5.4720168904294215e-06, - "loss": 0.1647, - "step": 963 - }, - { - "epoch": 3.9048101265822783, - "grad_norm": 0.23721814393073262, - "learning_rate": 5.433066737499948e-06, - "loss": 0.164, - "step": 964 - }, - { - "epoch": 3.908860759493671, - "grad_norm": 0.25360314027534153, - "learning_rate": 5.394233904648376e-06, - "loss": 0.1466, - "step": 965 - }, - { - "epoch": 3.9129113924050634, - "grad_norm": 0.24223205317647217, - "learning_rate": 5.355518704628997e-06, - "loss": 0.1626, - "step": 966 - }, - { - "epoch": 3.9169620253164554, - "grad_norm": 0.24803890780467708, - "learning_rate": 5.316921449248731e-06, - "loss": 0.1772, - "step": 967 - }, - { - "epoch": 3.921012658227848, - "grad_norm": 0.24828158524718116, - "learning_rate": 5.278442449364538e-06, - "loss": 0.1464, - "step": 968 - }, - { - "epoch": 3.9250632911392405, - "grad_norm": 0.25132187353390767, - "learning_rate": 5.240082014881016e-06, - "loss": 0.1663, - "step": 969 - }, - { - "epoch": 3.929113924050633, - "grad_norm": 0.23833499831406454, - "learning_rate": 5.201840454747822e-06, - "loss": 0.1534, - "step": 970 - }, - { - "epoch": 3.9331645569620255, - "grad_norm": 0.24977835699120865, - "learning_rate": 5.163718076957223e-06, - "loss": 0.1483, - "step": 971 - }, - { - "epoch": 3.9372151898734176, - "grad_norm": 0.24809800617318303, - "learning_rate": 5.125715188541609e-06, - "loss": 0.1631, - "step": 972 - }, - { - "epoch": 3.94126582278481, - "grad_norm": 0.25573283701244554, - "learning_rate": 5.087832095571021e-06, - "loss": 0.1417, - "step": 973 - }, - { - "epoch": 3.9453164556962026, - "grad_norm": 0.25998208286045466, - "learning_rate": 5.0500691031506766e-06, - "loss": 0.1655, - "step": 974 - }, - { - "epoch": 3.9493670886075947, - "grad_norm": 0.25074200303320515, - "learning_rate": 5.01242651541854e-06, - "loss": 0.1661, - "step": 975 - }, - { - "epoch": 3.953417721518987, - "grad_norm": 0.2595153696780746, - "learning_rate": 4.974904635542815e-06, - "loss": 0.1462, - "step": 976 - }, - { - "epoch": 3.9574683544303797, - "grad_norm": 0.2463018573381016, - "learning_rate": 4.937503765719582e-06, - "loss": 0.1589, - "step": 977 - }, - { - "epoch": 3.961518987341772, - "grad_norm": 0.2521897661099928, - "learning_rate": 4.900224207170299e-06, - "loss": 0.1698, - "step": 978 - }, - { - "epoch": 3.9655696202531647, - "grad_norm": 0.27432207410824794, - "learning_rate": 4.8630662601394065e-06, - "loss": 0.1445, - "step": 979 - }, - { - "epoch": 3.969620253164557, - "grad_norm": 0.24196881622190164, - "learning_rate": 4.8260302238918995e-06, - "loss": 0.1576, - "step": 980 - }, - { - "epoch": 3.9736708860759493, - "grad_norm": 0.24618525514853487, - "learning_rate": 4.789116396710924e-06, - "loss": 0.1472, - "step": 981 - }, - { - "epoch": 3.977721518987342, - "grad_norm": 0.25540689964575236, - "learning_rate": 4.752325075895368e-06, - "loss": 0.1523, - "step": 982 - }, - { - "epoch": 3.981772151898734, - "grad_norm": 0.259029620573096, - "learning_rate": 4.715656557757473e-06, - "loss": 0.164, - "step": 983 - }, - { - "epoch": 3.9858227848101264, - "grad_norm": 0.24416716324394266, - "learning_rate": 4.679111137620442e-06, - "loss": 0.1624, - "step": 984 - }, - { - "epoch": 3.989873417721519, - "grad_norm": 0.24455449312611538, - "learning_rate": 4.6426891098160585e-06, - "loss": 0.1567, - "step": 985 - }, - { - "epoch": 3.9939240506329114, - "grad_norm": 0.2672936255859228, - "learning_rate": 4.6063907676823474e-06, - "loss": 0.1658, - "step": 986 - }, - { - "epoch": 3.997974683544304, - "grad_norm": 0.25292377519422765, - "learning_rate": 4.570216403561141e-06, - "loss": 0.1537, - "step": 987 - }, - { - "epoch": 4.002025316455696, - "grad_norm": 0.2630643821793862, - "learning_rate": 4.534166308795815e-06, - "loss": 0.1314, - "step": 988 - }, - { - "epoch": 4.006075949367089, - "grad_norm": 0.33880314687057583, - "learning_rate": 4.498240773728859e-06, - "loss": 0.1235, - "step": 989 - }, - { - "epoch": 4.010126582278481, - "grad_norm": 0.27208439979546833, - "learning_rate": 4.462440087699609e-06, - "loss": 0.1166, - "step": 990 - }, - { - "epoch": 4.014177215189873, - "grad_norm": 0.20534963200806275, - "learning_rate": 4.426764539041861e-06, - "loss": 0.1135, - "step": 991 - }, - { - "epoch": 4.018227848101266, - "grad_norm": 0.2704665900696553, - "learning_rate": 4.391214415081582e-06, - "loss": 0.1124, - "step": 992 - }, - { - "epoch": 4.022278481012658, - "grad_norm": 0.3754796314299987, - "learning_rate": 4.355790002134579e-06, - "loss": 0.118, - "step": 993 - }, - { - "epoch": 4.02632911392405, - "grad_norm": 0.36722700935941166, - "learning_rate": 4.320491585504207e-06, - "loss": 0.1195, - "step": 994 - }, - { - "epoch": 4.030379746835443, - "grad_norm": 0.2691072819760026, - "learning_rate": 4.2853194494790615e-06, - "loss": 0.1232, - "step": 995 - }, - { - "epoch": 4.034430379746835, - "grad_norm": 0.24776192744305042, - "learning_rate": 4.250273877330691e-06, - "loss": 0.1093, - "step": 996 - }, - { - "epoch": 4.038481012658228, - "grad_norm": 0.3282723338526353, - "learning_rate": 4.215355151311313e-06, - "loss": 0.1228, - "step": 997 - }, - { - "epoch": 4.04253164556962, - "grad_norm": 0.3248037904612766, - "learning_rate": 4.180563552651542e-06, - "loss": 0.1283, - "step": 998 - }, - { - "epoch": 4.046582278481012, - "grad_norm": 0.2676506944728411, - "learning_rate": 4.145899361558147e-06, - "loss": 0.1128, - "step": 999 - }, - { - "epoch": 4.050632911392405, - "grad_norm": 0.21153746241567573, - "learning_rate": 4.111362857211738e-06, - "loss": 0.1231, - "step": 1000 - }, - { - "epoch": 4.054683544303797, - "grad_norm": 0.2216142981544128, - "learning_rate": 4.076954317764592e-06, - "loss": 0.1181, - "step": 1001 - }, - { - "epoch": 4.0587341772151895, - "grad_norm": 0.25186408890903295, - "learning_rate": 4.042674020338335e-06, - "loss": 0.1272, - "step": 1002 - }, - { - "epoch": 4.062784810126582, - "grad_norm": 0.2862467211159429, - "learning_rate": 4.0085222410217835e-06, - "loss": 0.1178, - "step": 1003 - }, - { - "epoch": 4.0668354430379745, - "grad_norm": 0.2584655362217925, - "learning_rate": 3.974499254868674e-06, - "loss": 0.1202, - "step": 1004 - }, - { - "epoch": 4.0708860759493675, - "grad_norm": 0.21287397999311464, - "learning_rate": 3.940605335895451e-06, - "loss": 0.1158, - "step": 1005 - }, - { - "epoch": 4.0749367088607595, - "grad_norm": 0.2272606799300811, - "learning_rate": 3.90684075707908e-06, - "loss": 0.1167, - "step": 1006 - }, - { - "epoch": 4.078987341772152, - "grad_norm": 0.23460983905126037, - "learning_rate": 3.8732057903548505e-06, - "loss": 0.1183, - "step": 1007 - }, - { - "epoch": 4.083037974683545, - "grad_norm": 0.24355711245616457, - "learning_rate": 3.8397007066141375e-06, - "loss": 0.1174, - "step": 1008 - }, - { - "epoch": 4.087088607594937, - "grad_norm": 0.22953984106875466, - "learning_rate": 3.806325775702304e-06, - "loss": 0.1235, - "step": 1009 - }, - { - "epoch": 4.091139240506329, - "grad_norm": 0.21988789633880235, - "learning_rate": 3.773081266416434e-06, - "loss": 0.1255, - "step": 1010 - }, - { - "epoch": 4.095189873417722, - "grad_norm": 0.22299107697306728, - "learning_rate": 3.739967446503245e-06, - "loss": 0.126, - "step": 1011 - }, - { - "epoch": 4.099240506329114, - "grad_norm": 0.22849862539071805, - "learning_rate": 3.706984582656894e-06, - "loss": 0.1111, - "step": 1012 - }, - { - "epoch": 4.103291139240507, - "grad_norm": 0.23103160714784965, - "learning_rate": 3.6741329405168237e-06, - "loss": 0.1152, - "step": 1013 - }, - { - "epoch": 4.107341772151899, - "grad_norm": 0.225094717670364, - "learning_rate": 3.641412784665648e-06, - "loss": 0.1038, - "step": 1014 - }, - { - "epoch": 4.111392405063291, - "grad_norm": 0.22321826082979954, - "learning_rate": 3.608824378627005e-06, - "loss": 0.1292, - "step": 1015 - }, - { - "epoch": 4.115443037974684, - "grad_norm": 0.2230738656578773, - "learning_rate": 3.5763679848634337e-06, - "loss": 0.1172, - "step": 1016 - }, - { - "epoch": 4.119493670886076, - "grad_norm": 0.21839372704933116, - "learning_rate": 3.544043864774269e-06, - "loss": 0.1059, - "step": 1017 - }, - { - "epoch": 4.123544303797468, - "grad_norm": 0.2286119411270651, - "learning_rate": 3.5118522786935282e-06, - "loss": 0.1237, - "step": 1018 - }, - { - "epoch": 4.127594936708861, - "grad_norm": 0.21380841486349964, - "learning_rate": 3.479793485887819e-06, - "loss": 0.1183, - "step": 1019 - }, - { - "epoch": 4.131645569620253, - "grad_norm": 0.2081825307671998, - "learning_rate": 3.4478677445542653e-06, - "loss": 0.1283, - "step": 1020 - }, - { - "epoch": 4.135696202531646, - "grad_norm": 0.20569889756285226, - "learning_rate": 3.4160753118183767e-06, - "loss": 0.1188, - "step": 1021 - }, - { - "epoch": 4.139746835443038, - "grad_norm": 0.2170529063060265, - "learning_rate": 3.3844164437320527e-06, - "loss": 0.1165, - "step": 1022 - }, - { - "epoch": 4.14379746835443, - "grad_norm": 0.2117667452477069, - "learning_rate": 3.3528913952714558e-06, - "loss": 0.1015, - "step": 1023 - }, - { - "epoch": 4.147848101265823, - "grad_norm": 0.21430017179735353, - "learning_rate": 3.321500420335e-06, - "loss": 0.1059, - "step": 1024 - }, - { - "epoch": 4.151898734177215, - "grad_norm": 0.21470745995743878, - "learning_rate": 3.290243771741275e-06, - "loss": 0.1283, - "step": 1025 - }, - { - "epoch": 4.155949367088607, - "grad_norm": 0.2030013839294294, - "learning_rate": 3.2591217012270325e-06, - "loss": 0.1183, - "step": 1026 - }, - { - "epoch": 4.16, - "grad_norm": 0.22519001699224542, - "learning_rate": 3.228134459445149e-06, - "loss": 0.1001, - "step": 1027 - }, - { - "epoch": 4.164050632911392, - "grad_norm": 0.20670809055762013, - "learning_rate": 3.1972822959626205e-06, - "loss": 0.1151, - "step": 1028 - }, - { - "epoch": 4.168101265822785, - "grad_norm": 0.21987863043333009, - "learning_rate": 3.166565459258513e-06, - "loss": 0.1167, - "step": 1029 - }, - { - "epoch": 4.172151898734177, - "grad_norm": 0.21189346202964568, - "learning_rate": 3.1359841967220193e-06, - "loss": 0.1158, - "step": 1030 - }, - { - "epoch": 4.176202531645569, - "grad_norm": 0.21345713130052868, - "learning_rate": 3.105538754650419e-06, - "loss": 0.1082, - "step": 1031 - }, - { - "epoch": 4.180253164556962, - "grad_norm": 0.22010486394357115, - "learning_rate": 3.07522937824712e-06, - "loss": 0.1058, - "step": 1032 - }, - { - "epoch": 4.184303797468354, - "grad_norm": 0.21336919310878416, - "learning_rate": 3.0450563116196697e-06, - "loss": 0.1348, - "step": 1033 - }, - { - "epoch": 4.188354430379746, - "grad_norm": 0.21401173426009146, - "learning_rate": 3.0150197977778008e-06, - "loss": 0.107, - "step": 1034 - }, - { - "epoch": 4.192405063291139, - "grad_norm": 0.2105273688417726, - "learning_rate": 2.985120078631465e-06, - "loss": 0.1274, - "step": 1035 - }, - { - "epoch": 4.1964556962025314, - "grad_norm": 0.21352221483537187, - "learning_rate": 2.9553573949888893e-06, - "loss": 0.1172, - "step": 1036 - }, - { - "epoch": 4.200506329113924, - "grad_norm": 0.2130753836403766, - "learning_rate": 2.9257319865546384e-06, - "loss": 0.121, - "step": 1037 - }, - { - "epoch": 4.2045569620253165, - "grad_norm": 0.1985612497120337, - "learning_rate": 2.896244091927678e-06, - "loss": 0.1271, - "step": 1038 - }, - { - "epoch": 4.2086075949367086, - "grad_norm": 0.2089819131698675, - "learning_rate": 2.8668939485994584e-06, - "loss": 0.1125, - "step": 1039 - }, - { - "epoch": 4.2126582278481015, - "grad_norm": 0.20674932442009689, - "learning_rate": 2.837681792951994e-06, - "loss": 0.1339, - "step": 1040 - }, - { - "epoch": 4.216708860759494, - "grad_norm": 0.20694356951579712, - "learning_rate": 2.808607860255981e-06, - "loss": 0.1165, - "step": 1041 - }, - { - "epoch": 4.220759493670886, - "grad_norm": 0.20817915565327672, - "learning_rate": 2.7796723846688634e-06, - "loss": 0.1122, - "step": 1042 - }, - { - "epoch": 4.224810126582279, - "grad_norm": 0.21245203026818976, - "learning_rate": 2.7508755992329937e-06, - "loss": 0.1237, - "step": 1043 - }, - { - "epoch": 4.228860759493671, - "grad_norm": 0.20551464168371134, - "learning_rate": 2.722217735873718e-06, - "loss": 0.132, - "step": 1044 - }, - { - "epoch": 4.232911392405064, - "grad_norm": 0.21495236065423326, - "learning_rate": 2.6936990253975315e-06, - "loss": 0.1183, - "step": 1045 - }, - { - "epoch": 4.236962025316456, - "grad_norm": 0.20062676615951858, - "learning_rate": 2.665319697490205e-06, - "loss": 0.1064, - "step": 1046 - }, - { - "epoch": 4.241012658227848, - "grad_norm": 0.2123075948988442, - "learning_rate": 2.637079980714945e-06, - "loss": 0.1329, - "step": 1047 - }, - { - "epoch": 4.245063291139241, - "grad_norm": 0.20341629001007777, - "learning_rate": 2.6089801025105453e-06, - "loss": 0.1119, - "step": 1048 - }, - { - "epoch": 4.249113924050633, - "grad_norm": 0.20713083837574972, - "learning_rate": 2.581020289189571e-06, - "loss": 0.1121, - "step": 1049 - }, - { - "epoch": 4.253164556962025, - "grad_norm": 0.2157678977321124, - "learning_rate": 2.553200765936501e-06, - "loss": 0.1237, - "step": 1050 - }, - { - "epoch": 4.257215189873418, - "grad_norm": 0.2217107081643587, - "learning_rate": 2.525521756805962e-06, - "loss": 0.133, - "step": 1051 - }, - { - "epoch": 4.26126582278481, - "grad_norm": 0.20744164058612227, - "learning_rate": 2.497983484720885e-06, - "loss": 0.1297, - "step": 1052 - }, - { - "epoch": 4.265316455696203, - "grad_norm": 0.21715298158620164, - "learning_rate": 2.470586171470728e-06, - "loss": 0.1105, - "step": 1053 - }, - { - "epoch": 4.269367088607595, - "grad_norm": 0.21088198002146902, - "learning_rate": 2.4433300377096836e-06, - "loss": 0.1217, - "step": 1054 - }, - { - "epoch": 4.273417721518987, - "grad_norm": 0.213936525667129, - "learning_rate": 2.4162153029549073e-06, - "loss": 0.1316, - "step": 1055 - }, - { - "epoch": 4.27746835443038, - "grad_norm": 0.21696396428842615, - "learning_rate": 2.3892421855847458e-06, - "loss": 0.1123, - "step": 1056 - }, - { - "epoch": 4.281518987341772, - "grad_norm": 0.20443173310555152, - "learning_rate": 2.362410902836978e-06, - "loss": 0.118, - "step": 1057 - }, - { - "epoch": 4.285569620253165, - "grad_norm": 0.21461082139433513, - "learning_rate": 2.3357216708070653e-06, - "loss": 0.1078, - "step": 1058 - }, - { - "epoch": 4.289620253164557, - "grad_norm": 0.2087392070133489, - "learning_rate": 2.309174704446411e-06, - "loss": 0.1219, - "step": 1059 - }, - { - "epoch": 4.293670886075949, - "grad_norm": 0.20754734076427592, - "learning_rate": 2.2827702175606437e-06, - "loss": 0.1206, - "step": 1060 - }, - { - "epoch": 4.297721518987342, - "grad_norm": 0.20075152583335018, - "learning_rate": 2.256508422807855e-06, - "loss": 0.1223, - "step": 1061 - }, - { - "epoch": 4.301772151898734, - "grad_norm": 0.21696826977527567, - "learning_rate": 2.230389531696946e-06, - "loss": 0.1047, - "step": 1062 - }, - { - "epoch": 4.305822784810126, - "grad_norm": 0.21041400641960695, - "learning_rate": 2.204413754585857e-06, - "loss": 0.0975, - "step": 1063 - }, - { - "epoch": 4.309873417721519, - "grad_norm": 0.21178496521510046, - "learning_rate": 2.1785813006799406e-06, - "loss": 0.1167, - "step": 1064 - }, - { - "epoch": 4.313924050632911, - "grad_norm": 0.2330828758422294, - "learning_rate": 2.1528923780302224e-06, - "loss": 0.1119, - "step": 1065 - }, - { - "epoch": 4.317974683544303, - "grad_norm": 0.22699012410415995, - "learning_rate": 2.127347193531757e-06, - "loss": 0.116, - "step": 1066 - }, - { - "epoch": 4.322025316455696, - "grad_norm": 0.21138016027001472, - "learning_rate": 2.101945952921942e-06, - "loss": 0.1158, - "step": 1067 - }, - { - "epoch": 4.326075949367088, - "grad_norm": 0.20151910317051508, - "learning_rate": 2.0766888607788906e-06, - "loss": 0.1168, - "step": 1068 - }, - { - "epoch": 4.330126582278481, - "grad_norm": 0.2125072515748105, - "learning_rate": 2.0515761205197337e-06, - "loss": 0.0979, - "step": 1069 - }, - { - "epoch": 4.334177215189873, - "grad_norm": 0.2107698407842126, - "learning_rate": 2.0266079343990453e-06, - "loss": 0.1244, - "step": 1070 - }, - { - "epoch": 4.3382278481012655, - "grad_norm": 0.21027002659203328, - "learning_rate": 2.0017845035071494e-06, - "loss": 0.1116, - "step": 1071 - }, - { - "epoch": 4.3422784810126585, - "grad_norm": 0.21996085057991177, - "learning_rate": 1.9771060277685537e-06, - "loss": 0.1186, - "step": 1072 - }, - { - "epoch": 4.3463291139240505, - "grad_norm": 0.2103809729797862, - "learning_rate": 1.95257270594031e-06, - "loss": 0.1019, - "step": 1073 - }, - { - "epoch": 4.3503797468354435, - "grad_norm": 0.20627715439090055, - "learning_rate": 1.9281847356104188e-06, - "loss": 0.1183, - "step": 1074 - }, - { - "epoch": 4.3544303797468356, - "grad_norm": 0.20589562887736468, - "learning_rate": 1.9039423131962365e-06, - "loss": 0.1227, - "step": 1075 - }, - { - "epoch": 4.358481012658228, - "grad_norm": 0.21336148947063796, - "learning_rate": 1.8798456339429027e-06, - "loss": 0.0859, - "step": 1076 - }, - { - "epoch": 4.362531645569621, - "grad_norm": 0.20831422816400133, - "learning_rate": 1.8558948919217612e-06, - "loss": 0.1162, - "step": 1077 - }, - { - "epoch": 4.366582278481013, - "grad_norm": 0.21145219119477301, - "learning_rate": 1.8320902800287954e-06, - "loss": 0.1267, - "step": 1078 - }, - { - "epoch": 4.370632911392405, - "grad_norm": 0.2136018809141807, - "learning_rate": 1.8084319899830726e-06, - "loss": 0.1143, - "step": 1079 - }, - { - "epoch": 4.374683544303798, - "grad_norm": 0.21949543447597544, - "learning_rate": 1.7849202123252097e-06, - "loss": 0.1257, - "step": 1080 - }, - { - "epoch": 4.37873417721519, - "grad_norm": 0.2140850337924572, - "learning_rate": 1.7615551364158401e-06, - "loss": 0.1228, - "step": 1081 - }, - { - "epoch": 4.382784810126582, - "grad_norm": 0.21026595117652644, - "learning_rate": 1.738336950434061e-06, - "loss": 0.11, - "step": 1082 - }, - { - "epoch": 4.386835443037975, - "grad_norm": 0.20738756716663165, - "learning_rate": 1.715265841375957e-06, - "loss": 0.1358, - "step": 1083 - }, - { - "epoch": 4.390886075949367, - "grad_norm": 0.21865162793635812, - "learning_rate": 1.6923419950530684e-06, - "loss": 0.1051, - "step": 1084 - }, - { - "epoch": 4.39493670886076, - "grad_norm": 0.2054525486858317, - "learning_rate": 1.6695655960909008e-06, - "loss": 0.1184, - "step": 1085 - }, - { - "epoch": 4.398987341772152, - "grad_norm": 0.21244269696954227, - "learning_rate": 1.646936827927441e-06, - "loss": 0.1316, - "step": 1086 - }, - { - "epoch": 4.403037974683544, - "grad_norm": 0.20580273370728408, - "learning_rate": 1.6244558728116766e-06, - "loss": 0.1222, - "step": 1087 - }, - { - "epoch": 4.407088607594937, - "grad_norm": 0.21336703945479918, - "learning_rate": 1.6021229118021265e-06, - "loss": 0.1209, - "step": 1088 - }, - { - "epoch": 4.411139240506329, - "grad_norm": 0.2012888307312296, - "learning_rate": 1.5799381247653967e-06, - "loss": 0.11, - "step": 1089 - }, - { - "epoch": 4.415189873417722, - "grad_norm": 0.20989183896767596, - "learning_rate": 1.5579016903747013e-06, - "loss": 0.1123, - "step": 1090 - }, - { - "epoch": 4.419240506329114, - "grad_norm": 0.20811171237500875, - "learning_rate": 1.5360137861084656e-06, - "loss": 0.1066, - "step": 1091 - }, - { - "epoch": 4.423291139240506, - "grad_norm": 0.20996562964240167, - "learning_rate": 1.5142745882488475e-06, - "loss": 0.1172, - "step": 1092 - }, - { - "epoch": 4.427341772151899, - "grad_norm": 0.21055298826319746, - "learning_rate": 1.4926842718803691e-06, - "loss": 0.1227, - "step": 1093 - }, - { - "epoch": 4.431392405063291, - "grad_norm": 0.21826211484484484, - "learning_rate": 1.4712430108884657e-06, - "loss": 0.1129, - "step": 1094 - }, - { - "epoch": 4.435443037974683, - "grad_norm": 0.20575334368513748, - "learning_rate": 1.4499509779581078e-06, - "loss": 0.1147, - "step": 1095 - }, - { - "epoch": 4.439493670886076, - "grad_norm": 0.20683016805102505, - "learning_rate": 1.4288083445723988e-06, - "loss": 0.1291, - "step": 1096 - }, - { - "epoch": 4.443544303797468, - "grad_norm": 0.21522887137057378, - "learning_rate": 1.4078152810112045e-06, - "loss": 0.1203, - "step": 1097 - }, - { - "epoch": 4.44759493670886, - "grad_norm": 0.21198371267282254, - "learning_rate": 1.3869719563497697e-06, - "loss": 0.111, - "step": 1098 - }, - { - "epoch": 4.451645569620253, - "grad_norm": 0.20300825250127186, - "learning_rate": 1.3662785384573663e-06, - "loss": 0.1175, - "step": 1099 - }, - { - "epoch": 4.455696202531645, - "grad_norm": 0.20681093665476818, - "learning_rate": 1.3457351939959383e-06, - "loss": 0.1143, - "step": 1100 - }, - { - "epoch": 4.459746835443038, - "grad_norm": 0.20407933733347589, - "learning_rate": 1.3253420884187551e-06, - "loss": 0.1347, - "step": 1101 - }, - { - "epoch": 4.46379746835443, - "grad_norm": 0.20555664919896124, - "learning_rate": 1.3050993859690953e-06, - "loss": 0.1131, - "step": 1102 - }, - { - "epoch": 4.467848101265822, - "grad_norm": 0.21494197075788835, - "learning_rate": 1.2850072496788869e-06, - "loss": 0.0946, - "step": 1103 - }, - { - "epoch": 4.471898734177215, - "grad_norm": 0.20360098681010508, - "learning_rate": 1.2650658413674434e-06, - "loss": 0.1201, - "step": 1104 - }, - { - "epoch": 4.4759493670886075, - "grad_norm": 0.2040934099996908, - "learning_rate": 1.2452753216401226e-06, - "loss": 0.1163, - "step": 1105 - }, - { - "epoch": 4.48, - "grad_norm": 0.211373206426806, - "learning_rate": 1.2256358498870503e-06, - "loss": 0.1176, - "step": 1106 - }, - { - "epoch": 4.4840506329113925, - "grad_norm": 0.20926912822238875, - "learning_rate": 1.2061475842818337e-06, - "loss": 0.1142, - "step": 1107 - }, - { - "epoch": 4.488101265822785, - "grad_norm": 0.2159979372098844, - "learning_rate": 1.1868106817802816e-06, - "loss": 0.1086, - "step": 1108 - }, - { - "epoch": 4.4921518987341775, - "grad_norm": 0.2105402047851378, - "learning_rate": 1.1676252981191482e-06, - "loss": 0.1128, - "step": 1109 - }, - { - "epoch": 4.49620253164557, - "grad_norm": 0.20316782582784526, - "learning_rate": 1.1485915878148823e-06, - "loss": 0.1157, - "step": 1110 - }, - { - "epoch": 4.500253164556962, - "grad_norm": 0.22405189785088125, - "learning_rate": 1.1297097041623584e-06, - "loss": 0.1101, - "step": 1111 - }, - { - "epoch": 4.504303797468355, - "grad_norm": 0.21294812237958868, - "learning_rate": 1.1109797992336847e-06, - "loss": 0.1154, - "step": 1112 - }, - { - "epoch": 4.508354430379747, - "grad_norm": 0.21814642926193412, - "learning_rate": 1.092402023876933e-06, - "loss": 0.104, - "step": 1113 - }, - { - "epoch": 4.512405063291139, - "grad_norm": 0.20926351251232184, - "learning_rate": 1.0739765277149527e-06, - "loss": 0.0972, - "step": 1114 - }, - { - "epoch": 4.516455696202532, - "grad_norm": 0.21393826956472128, - "learning_rate": 1.0557034591441596e-06, - "loss": 0.1242, - "step": 1115 - }, - { - "epoch": 4.520506329113924, - "grad_norm": 0.19864420070866448, - "learning_rate": 1.0375829653333324e-06, - "loss": 0.1149, - "step": 1116 - }, - { - "epoch": 4.524556962025317, - "grad_norm": 0.20980522179760241, - "learning_rate": 1.0196151922224385e-06, - "loss": 0.1207, - "step": 1117 - }, - { - "epoch": 4.528607594936709, - "grad_norm": 0.20963704558047147, - "learning_rate": 1.0018002845214526e-06, - "loss": 0.1158, - "step": 1118 - }, - { - "epoch": 4.532658227848101, - "grad_norm": 0.21167444116505832, - "learning_rate": 9.841383857091947e-07, - "loss": 0.1185, - "step": 1119 - }, - { - "epoch": 4.536708860759494, - "grad_norm": 0.2025059229430548, - "learning_rate": 9.666296380321616e-07, - "loss": 0.1165, - "step": 1120 + "epoch": 4.311898734177215, + "grad_norm": 0.2491622467734568, + "learning_rate": 2.2929016062037057e-06, + "loss": 0.1082, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009641005657613277, + "step": 1065, + "valid_targets_mean": 2723.6, + "valid_targets_min": 915 }, { - "epoch": 4.540759493670886, - "grad_norm": 0.19894280074565418, - "learning_rate": 9.492741825034124e-07, + "epoch": 4.315949367088607, + "grad_norm": 0.23291800635203916, + "learning_rate": 2.2666794376975055e-06, "loss": 0.1099, - "step": 1121 - }, - { - "epoch": 4.544810126582279, - "grad_norm": 0.21171253336046708, - "learning_rate": 9.320721589013892e-07, - "loss": 0.1177, - "step": 1122 - }, - { - "epoch": 4.548860759493671, - "grad_norm": 0.20105006257948016, - "learning_rate": 9.150237057688339e-07, - "loss": 0.1047, - "step": 1123 - }, - { - "epoch": 4.552911392405063, - "grad_norm": 0.2032328366210336, - "learning_rate": 8.981289604116328e-07, - "loss": 0.1213, - "step": 1124 - }, - { - "epoch": 4.556962025316456, - "grad_norm": 0.19853846094696845, - "learning_rate": 8.813880588977542e-07, - "loss": 0.1195, - "step": 1125 - }, - { - "epoch": 4.561012658227848, - "grad_norm": 0.2074935286115484, - "learning_rate": 8.648011360561126e-07, - "loss": 0.114, - "step": 1126 - }, - { - "epoch": 4.56506329113924, - "grad_norm": 0.22013640458218453, - "learning_rate": 8.483683254755037e-07, - "loss": 0.1281, - "step": 1127 - }, - { - "epoch": 4.569113924050633, - "grad_norm": 0.21339634951818434, - "learning_rate": 8.320897595035227e-07, - "loss": 0.1123, - "step": 1128 + "loss_nan_ranks": 0, + "loss_rank_avg": 0.010794535279273987, + "step": 1066, + "valid_targets_mean": 4469.6, + "valid_targets_min": 1237 + }, + { + "epoch": 4.32, + "grad_norm": 0.24684772398855878, + "learning_rate": 2.24059906434998e-06, + "loss": 0.1075, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015624083578586578, + "step": 1067, + "valid_targets_mean": 5676.4, + "valid_targets_min": 1377 + }, + { + "epoch": 4.324050632911392, + "grad_norm": 0.24639223893807155, + "learning_rate": 2.214660694699091e-06, + "loss": 0.1178, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.021386027336120605, + "step": 1068, + "valid_targets_mean": 6623.7, + "valid_targets_min": 1466 }, { - "epoch": 4.573164556962025, - "grad_norm": 0.19926989858502814, - "learning_rate": 8.159655692455093e-07, - "loss": 0.1224, - "step": 1129 + "epoch": 4.328101265822784, + "grad_norm": 0.25355308895164, + "learning_rate": 2.188864536147355e-06, + "loss": 0.1101, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012787914834916592, + "step": 1069, + "valid_targets_mean": 3562.9, + "valid_targets_min": 1581 + }, + { + "epoch": 4.332151898734177, + "grad_norm": 0.2438523661180247, + "learning_rate": 2.1632107949601757e-06, + "loss": 0.1073, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013902321457862854, + "step": 1070, + "valid_targets_mean": 4379.1, + "valid_targets_min": 1419 + }, + { + "epoch": 4.3362025316455695, + "grad_norm": 0.23567280343654579, + "learning_rate": 2.137699676264171e-06, + "loss": 0.1144, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012037249282002449, + "step": 1071, + "valid_targets_mean": 4784.9, + "valid_targets_min": 1578 + }, + { + "epoch": 4.340253164556962, + "grad_norm": 0.2510640878527952, + "learning_rate": 2.1123313840455738e-06, + "loss": 0.1082, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.008996009826660156, + "step": 1072, + "valid_targets_mean": 3449.2, + "valid_targets_min": 1279 + }, + { + "epoch": 4.3443037974683545, + "grad_norm": 0.25292754399485057, + "learning_rate": 2.08710612114857e-06, + "loss": 0.1068, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.010600311681628227, + "step": 1073, + "valid_targets_mean": 3625.1, + "valid_targets_min": 1295 + }, + { + "epoch": 4.348354430379747, + "grad_norm": 0.23233271567825875, + "learning_rate": 2.0620240892736953e-06, + "loss": 0.1151, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01687842421233654, + "step": 1074, + "valid_targets_mean": 5275.1, + "valid_targets_min": 1080 }, { - "epoch": 4.577215189873417, - "grad_norm": 0.21057777132930375, - "learning_rate": 7.999958845634648e-07, - "loss": 0.1199, - "step": 1130 + "epoch": 4.3524050632911395, + "grad_norm": 0.23516768581742103, + "learning_rate": 2.0370854889762e-06, + "loss": 0.1074, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014397460035979748, + "step": 1075, + "valid_targets_mean": 4645.4, + "valid_targets_min": 1009 + }, + { + "epoch": 4.356455696202532, + "grad_norm": 0.2407611209148767, + "learning_rate": 2.0122905196644726e-06, + "loss": 0.0982, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009776914492249489, + "step": 1076, + "valid_targets_mean": 3878.9, + "valid_targets_min": 1328 + }, + { + "epoch": 4.360506329113924, + "grad_norm": 0.24693246517008668, + "learning_rate": 1.987639379598427e-06, + "loss": 0.1074, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.019608918577432632, + "step": 1077, + "valid_targets_mean": 5656.2, + "valid_targets_min": 1097 }, { - "epoch": 4.58126582278481, - "grad_norm": 0.21230670196649526, - "learning_rate": 7.841808340750478e-07, + "epoch": 4.364556962025317, + "grad_norm": 0.23917342823384077, + "learning_rate": 1.963132265887919e-06, + "loss": 0.1186, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015440532006323338, + "step": 1078, + "valid_targets_mean": 4235.8, + "valid_targets_min": 1298 + }, + { + "epoch": 4.368607594936709, + "grad_norm": 0.2476402893022485, + "learning_rate": 1.9387693744911784e-06, + "loss": 0.1148, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0155026875436306, + "step": 1079, + "valid_targets_mean": 4751.9, + "valid_targets_min": 1181 + }, + { + "epoch": 4.372658227848102, + "grad_norm": 0.24046448814498173, + "learning_rate": 1.9145509002132368e-06, + "loss": 0.1149, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0111662857234478, + "step": 1080, + "valid_targets_mean": 3625.0, + "valid_targets_min": 1172 + }, + { + "epoch": 4.376708860759494, + "grad_norm": 0.2508759488943957, + "learning_rate": 1.8904770367043723e-06, + "loss": 0.1161, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015941239893436432, + "step": 1081, + "valid_targets_mean": 4548.8, + "valid_targets_min": 897 + }, + { + "epoch": 4.380759493670886, + "grad_norm": 0.2417593143201865, + "learning_rate": 1.8665479764585504e-06, + "loss": 0.1192, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011951446533203125, + "step": 1082, + "valid_targets_mean": 4621.5, + "valid_targets_min": 1320 + }, + { + "epoch": 4.384810126582279, + "grad_norm": 0.24355761102946463, + "learning_rate": 1.8427639108118977e-06, + "loss": 0.1137, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014804190024733543, + "step": 1083, + "valid_targets_mean": 4951.3, + "valid_targets_min": 1288 + }, + { + "epoch": 4.388860759493671, + "grad_norm": 0.24633384198377473, + "learning_rate": 1.8191250299411755e-06, + "loss": 0.1064, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015205018222332, + "step": 1084, + "valid_targets_mean": 4959.1, + "valid_targets_min": 1053 + }, + { + "epoch": 4.392911392405063, + "grad_norm": 0.24678792381766704, + "learning_rate": 1.7956315228622446e-06, + "loss": 0.1035, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.010629641823470592, + "step": 1085, + "valid_targets_mean": 3940.3, + "valid_targets_min": 1206 + }, + { + "epoch": 4.396962025316456, + "grad_norm": 0.23023111044490813, + "learning_rate": 1.7722835774285553e-06, "loss": 0.1138, - "step": 1131 - }, - { - "epoch": 4.585316455696202, - "grad_norm": 0.21718564894895476, - "learning_rate": 7.685205451524869e-07, - "loss": 0.125, - "step": 1132 + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01395136397331953, + "step": 1086, + "valid_targets_mean": 5731.8, + "valid_targets_min": 1487 }, { - "epoch": 4.589367088607595, - "grad_norm": 0.20942737386401283, - "learning_rate": 7.530151439216027e-07, - "loss": 0.1106, - "step": 1133 - }, - { - "epoch": 4.593417721518987, - "grad_norm": 0.20594666622404534, - "learning_rate": 7.376647552607675e-07, - "loss": 0.1174, - "step": 1134 - }, - { - "epoch": 4.597468354430379, - "grad_norm": 0.20526587440873395, - "learning_rate": 7.224695027998963e-07, - "loss": 0.1212, - "step": 1135 - }, - { - "epoch": 4.601518987341772, - "grad_norm": 0.21472834819077183, - "learning_rate": 7.07429508919466e-07, - "loss": 0.1019, - "step": 1136 - }, - { - "epoch": 4.605569620253164, - "grad_norm": 0.20581667163388773, - "learning_rate": 6.925448947495206e-07, - "loss": 0.1404, - "step": 1137 - }, - { - "epoch": 4.609620253164557, - "grad_norm": 0.21338707700272144, - "learning_rate": 6.778157801686936e-07, - "loss": 0.1152, - "step": 1138 - }, - { - "epoch": 4.613670886075949, - "grad_norm": 0.2053462512645686, - "learning_rate": 6.632422838032515e-07, - "loss": 0.1051, - "step": 1139 - }, - { - "epoch": 4.6177215189873415, - "grad_norm": 0.2041771303195351, - "learning_rate": 6.488245230261281e-07, - "loss": 0.1259, - "step": 1140 - }, - { - "epoch": 4.6217721518987345, - "grad_norm": 0.20335070817851897, - "learning_rate": 6.345626139559868e-07, - "loss": 0.115, - "step": 1141 + "epoch": 4.401012658227848, + "grad_norm": 0.2254694696708648, + "learning_rate": 1.7490813803296624e-06, + "loss": 0.1074, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018306240439414978, + "step": 1087, + "valid_targets_mean": 6387.6, + "valid_targets_min": 1149 + }, + { + "epoch": 4.405063291139241, + "grad_norm": 0.23602745246431867, + "learning_rate": 1.726025117089718e-06, + "loss": 0.1092, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01539086364209652, + "step": 1088, + "valid_targets_mean": 5928.7, + "valid_targets_min": 981 + }, + { + "epoch": 4.409113924050633, + "grad_norm": 0.24677515872589578, + "learning_rate": 1.70311497206598e-06, + "loss": 0.1107, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01074039563536644, + "step": 1089, + "valid_targets_mean": 3698.3, + "valid_targets_min": 1212 + }, + { + "epoch": 4.413164556962025, + "grad_norm": 0.24193020498990295, + "learning_rate": 1.6803511284473638e-06, + "loss": 0.1034, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012303371913731098, + "step": 1090, + "valid_targets_mean": 3728.2, + "valid_targets_min": 1434 + }, + { + "epoch": 4.417215189873418, + "grad_norm": 0.25697776287635726, + "learning_rate": 1.6577337682529538e-06, + "loss": 0.1151, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015065351501107216, + "step": 1091, + "valid_targets_mean": 4298.5, + "valid_targets_min": 1137 + }, + { + "epoch": 4.42126582278481, + "grad_norm": 0.24763499135416595, + "learning_rate": 1.6352630723305617e-06, + "loss": 0.1077, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017283417284488678, + "step": 1092, + "valid_targets_mean": 4962.7, + "valid_targets_min": 1041 + }, + { + "epoch": 4.425316455696202, + "grad_norm": 0.23550719317841343, + "learning_rate": 1.612939220355272e-06, + "loss": 0.1128, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014940792694687843, + "step": 1093, + "valid_targets_mean": 4944.0, + "valid_targets_min": 1344 + }, + { + "epoch": 4.429367088607595, + "grad_norm": 0.23589340772446932, + "learning_rate": 1.5907623908280112e-06, + "loss": 0.1061, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015425451099872589, + "step": 1094, + "valid_targets_mean": 5052.8, + "valid_targets_min": 1142 + }, + { + "epoch": 4.433417721518987, + "grad_norm": 0.2354380195701554, + "learning_rate": 1.568732761074121e-06, + "loss": 0.1081, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.010928714647889137, + "step": 1095, + "valid_targets_mean": 4296.2, + "valid_targets_min": 1321 + }, + { + "epoch": 4.43746835443038, + "grad_norm": 0.23904734455838542, + "learning_rate": 1.5468505072419237e-06, + "loss": 0.1122, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018300985917448997, + "step": 1096, + "valid_targets_mean": 5111.7, + "valid_targets_min": 1443 }, { - "epoch": 4.6258227848101265, - "grad_norm": 0.20335588641121002, - "learning_rate": 6.204566714562866e-07, + "epoch": 4.441518987341772, + "grad_norm": 0.24199872653993765, + "learning_rate": 1.5251158043013448e-06, "loss": 0.1105, - "step": 1142 - }, - { - "epoch": 4.629873417721519, - "grad_norm": 0.1978284104826405, - "learning_rate": 6.06506809134344e-07, - "loss": 0.1256, - "step": 1143 - }, - { - "epoch": 4.633924050632912, - "grad_norm": 0.19588668465530068, - "learning_rate": 5.927131393404373e-07, - "loss": 0.1229, - "step": 1144 - }, - { - "epoch": 4.637974683544304, - "grad_norm": 0.2112026075669013, - "learning_rate": 5.790757731668817e-07, - "loss": 0.1048, - "step": 1145 - }, - { - "epoch": 4.642025316455696, - "grad_norm": 0.1988884909158249, - "learning_rate": 5.655948204471507e-07, - "loss": 0.1252, - "step": 1146 - }, - { - "epoch": 4.646075949367089, - "grad_norm": 0.2052698975996843, - "learning_rate": 5.522703897549875e-07, - "loss": 0.13, - "step": 1147 - }, - { - "epoch": 4.650126582278481, - "grad_norm": 0.2113471441983246, - "learning_rate": 5.391025884035239e-07, - "loss": 0.118, - "step": 1148 + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01229251828044653, + "step": 1097, + "valid_targets_mean": 5098.9, + "valid_targets_min": 515 + }, + { + "epoch": 4.445569620253164, + "grad_norm": 0.23200296960339453, + "learning_rate": 1.5035288260424885e-06, + "loss": 0.1035, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.010072460398077965, + "step": 1098, + "valid_targets_mean": 3966.8, + "valid_targets_min": 868 + }, + { + "epoch": 4.449620253164557, + "grad_norm": 0.2462493968454483, + "learning_rate": 1.4820897450742623e-06, + "loss": 0.108, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.00930652767419815, + "step": 1099, + "valid_targets_mean": 2841.6, + "valid_targets_min": 1123 + }, + { + "epoch": 4.453670886075949, + "grad_norm": 0.23785633041846221, + "learning_rate": 1.4607987328229767e-06, + "loss": 0.1132, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01593669131398201, + "step": 1100, + "valid_targets_mean": 5572.1, + "valid_targets_min": 1144 + }, + { + "epoch": 4.457721518987341, + "grad_norm": 0.25294580176951953, + "learning_rate": 1.4396559595310056e-06, + "loss": 0.1105, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01751577854156494, + "step": 1101, + "valid_targets_mean": 5425.2, + "valid_targets_min": 1279 + }, + { + "epoch": 4.461772151898734, + "grad_norm": 0.2258972905675386, + "learning_rate": 1.4186615942553973e-06, + "loss": 0.1104, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01287138182669878, + "step": 1102, + "valid_targets_mean": 4567.1, + "valid_targets_min": 1567 + }, + { + "epoch": 4.465822784810126, + "grad_norm": 0.26232752630138473, + "learning_rate": 1.3978158048665424e-06, + "loss": 0.1128, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015547772869467735, + "step": 1103, + "valid_targets_mean": 4495.8, + "valid_targets_min": 1144 + }, + { + "epoch": 4.469873417721519, + "grad_norm": 0.23988110774615404, + "learning_rate": 1.3771187580468115e-06, + "loss": 0.1108, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012048361822962761, + "step": 1104, + "valid_targets_mean": 4554.2, + "valid_targets_min": 1062 + }, + { + "epoch": 4.473924050632911, + "grad_norm": 0.23875743605498156, + "learning_rate": 1.3565706192892392e-06, + "loss": 0.1098, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012100529856979847, + "step": 1105, + "valid_targets_mean": 4961.4, + "valid_targets_min": 1118 + }, + { + "epoch": 4.4779746835443035, + "grad_norm": 0.22977235459445125, + "learning_rate": 1.3361715528961993e-06, + "loss": 0.1057, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01663079485297203, + "step": 1106, + "valid_targets_mean": 5048.4, + "valid_targets_min": 1415 + }, + { + "epoch": 4.4820253164556965, + "grad_norm": 0.23543453318246152, + "learning_rate": 1.3159217219780817e-06, + "loss": 0.1098, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013302445411682129, + "step": 1107, + "valid_targets_mean": 4229.1, + "valid_targets_min": 1511 + }, + { + "epoch": 4.4860759493670885, + "grad_norm": 0.25576008740685763, + "learning_rate": 1.2958212884519949e-06, + "loss": 0.1151, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015795301645994186, + "step": 1108, + "valid_targets_mean": 5122.8, + "valid_targets_min": 1139 + }, + { + "epoch": 4.490126582278481, + "grad_norm": 0.23384858874585748, + "learning_rate": 1.2758704130404675e-06, + "loss": 0.1037, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014064125716686249, + "step": 1109, + "valid_targets_mean": 4712.9, + "valid_targets_min": 741 + }, + { + "epoch": 4.494177215189874, + "grad_norm": 0.23135518159291688, + "learning_rate": 1.2560692552701738e-06, + "loss": 0.1071, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009383934549987316, + "step": 1110, + "valid_targets_mean": 3435.5, + "valid_targets_min": 1257 + }, + { + "epoch": 4.498227848101266, + "grad_norm": 0.23853269674283112, + "learning_rate": 1.2364179734706382e-06, + "loss": 0.1076, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013448908925056458, + "step": 1111, + "valid_targets_mean": 4762.3, + "valid_targets_min": 1467 + }, + { + "epoch": 4.502278481012659, + "grad_norm": 0.2469096524547987, + "learning_rate": 1.2169167247729852e-06, + "loss": 0.1109, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011553686112165451, + "step": 1112, + "valid_targets_mean": 3743.1, + "valid_targets_min": 1108 + }, + { + "epoch": 4.506329113924051, + "grad_norm": 0.2424742627143951, + "learning_rate": 1.1975656651086864e-06, + "loss": 0.1123, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.019136445596814156, + "step": 1113, + "valid_targets_mean": 5195.6, + "valid_targets_min": 1037 }, { - "epoch": 4.654177215189874, - "grad_norm": 0.20811912693391285, - "learning_rate": 5.260915224444207e-07, - "loss": 0.0994, - "step": 1149 + "epoch": 4.510379746835443, + "grad_norm": 0.25201664133307256, + "learning_rate": 1.1783649492082926e-06, + "loss": 0.1101, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020832877606153488, + "step": 1114, + "valid_targets_mean": 5313.9, + "valid_targets_min": 1113 + }, + { + "epoch": 4.514430379746836, + "grad_norm": 0.23965971627973182, + "learning_rate": 1.1593147306002183e-06, + "loss": 0.1081, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015099755488336086, + "step": 1115, + "valid_targets_mean": 4789.8, + "valid_targets_min": 1380 + }, + { + "epoch": 4.518481012658228, + "grad_norm": 0.24374472934816327, + "learning_rate": 1.140415161609505e-06, + "loss": 0.1093, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01824544370174408, + "step": 1116, + "valid_targets_mean": 5898.2, + "valid_targets_min": 1408 + }, + { + "epoch": 4.52253164556962, + "grad_norm": 0.22621971438759939, + "learning_rate": 1.121666393356604e-06, + "loss": 0.1067, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.006982675287872553, + "step": 1117, + "valid_targets_mean": 3357.9, + "valid_targets_min": 1230 + }, + { + "epoch": 4.526582278481013, + "grad_norm": 0.24277785111288838, + "learning_rate": 1.1030685757561632e-06, + "loss": 0.1105, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014613626524806023, + "step": 1118, + "valid_targets_mean": 4052.4, + "valid_targets_min": 1621 + }, + { + "epoch": 4.530632911392405, + "grad_norm": 0.24715457555021894, + "learning_rate": 1.0846218575158373e-06, + "loss": 0.1173, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014865981414914131, + "step": 1119, + "valid_targets_mean": 4405.4, + "valid_targets_min": 1484 + }, + { + "epoch": 4.534683544303798, + "grad_norm": 0.2350923562542679, + "learning_rate": 1.066326386135097e-06, + "loss": 0.1034, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013338493183255196, + "step": 1120, + "valid_targets_mean": 5251.7, + "valid_targets_min": 1338 + }, + { + "epoch": 4.53873417721519, + "grad_norm": 0.23261782507815473, + "learning_rate": 1.0481823079040398e-06, + "loss": 0.107, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018727660179138184, + "step": 1121, + "valid_targets_mean": 6229.1, + "valid_targets_min": 1371 + }, + { + "epoch": 4.542784810126582, + "grad_norm": 0.2372217839868558, + "learning_rate": 1.0301897679022344e-06, + "loss": 0.1065, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015288677997887135, + "step": 1122, + "valid_targets_mean": 4868.1, + "valid_targets_min": 1328 + }, + { + "epoch": 4.546835443037975, + "grad_norm": 0.22851145118582236, + "learning_rate": 1.0123489099975491e-06, + "loss": 0.106, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009553518146276474, + "step": 1123, + "valid_targets_mean": 3740.2, + "valid_targets_min": 1374 + }, + { + "epoch": 4.550886075949367, + "grad_norm": 0.24789698005851352, + "learning_rate": 9.9465987684501e-07, + "loss": 0.1114, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014984434470534325, + "step": 1124, + "valid_targets_mean": 4818.4, + "valid_targets_min": 1139 + }, + { + "epoch": 4.55493670886076, + "grad_norm": 0.2352459318761055, + "learning_rate": 9.771228098856534e-07, + "loss": 0.1102, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012403848581016064, + "step": 1125, + "valid_targets_mean": 3469.1, + "valid_targets_min": 828 + }, + { + "epoch": 4.558987341772152, + "grad_norm": 0.23193523725933549, + "learning_rate": 9.597378493454013e-07, + "loss": 0.1083, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01025470346212387, + "step": 1126, + "valid_targets_mean": 3658.0, + "valid_targets_min": 985 + }, + { + "epoch": 4.563037974683544, + "grad_norm": 0.24340338611260912, + "learning_rate": 9.425051342339353e-07, + "loss": 0.1157, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011842905543744564, + "step": 1127, + "valid_targets_mean": 4074.4, + "valid_targets_min": 1395 }, { - "epoch": 4.658227848101266, - "grad_norm": 0.21073259866741254, - "learning_rate": 5.132372966670129e-07, - "loss": 0.1184, - "step": 1150 + "epoch": 4.567088607594937, + "grad_norm": 0.23813693989298826, + "learning_rate": 9.254248023435864e-07, + "loss": 0.1064, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014363069087266922, + "step": 1128, + "valid_targets_mean": 5131.9, + "valid_targets_min": 1460 + }, + { + "epoch": 4.571139240506329, + "grad_norm": 0.2314982637199714, + "learning_rate": 9.084969902482288e-07, + "loss": 0.113, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017669495195150375, + "step": 1129, + "valid_targets_mean": 5292.9, + "valid_targets_min": 1350 + }, + { + "epoch": 4.575189873417721, + "grad_norm": 0.24205208281186263, + "learning_rate": 8.917218333021993e-07, + "loss": 0.1043, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015519630163908005, + "step": 1130, + "valid_targets_mean": 4128.9, + "valid_targets_min": 868 + }, + { + "epoch": 4.579240506329114, + "grad_norm": 0.2261304617829508, + "learning_rate": 8.750994656392109e-07, + "loss": 0.111, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016634684056043625, + "step": 1131, + "valid_targets_mean": 5589.6, + "valid_targets_min": 1178 }, { - "epoch": 4.662278481012658, - "grad_norm": 0.2122659373264562, - "learning_rate": 5.005400145974704e-07, - "loss": 0.1233, - "step": 1151 + "epoch": 4.583291139240506, + "grad_norm": 0.23952329967135425, + "learning_rate": 8.58630020171265e-07, + "loss": 0.1105, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014186002314090729, + "step": 1132, + "valid_targets_mean": 4555.3, + "valid_targets_min": 1872 }, { - "epoch": 4.666329113924051, - "grad_norm": 0.21095871472177763, - "learning_rate": 4.879997784979562e-07, - "loss": 0.1309, - "step": 1152 + "epoch": 4.587341772151898, + "grad_norm": 0.2442885441237708, + "learning_rate": 8.423136285876099e-07, + "loss": 0.1121, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012748418375849724, + "step": 1133, + "valid_targets_mean": 4353.7, + "valid_targets_min": 1734 }, { - "epoch": 4.670379746835443, - "grad_norm": 0.20313814160332547, - "learning_rate": 4.7561668936580984e-07, + "epoch": 4.591392405063291, + "grad_norm": 0.2551715498481697, + "learning_rate": 8.26150421353682e-07, "loss": 0.1074, - "step": 1153 - }, - { - "epoch": 4.674430379746836, - "grad_norm": 0.2085712200415366, - "learning_rate": 4.6339084693272306e-07, - "loss": 0.117, - "step": 1154 - }, - { - "epoch": 4.678481012658228, - "grad_norm": 0.19007347926883442, - "learning_rate": 4.5132234966395847e-07, - "loss": 0.1174, - "step": 1155 - }, - { - "epoch": 4.68253164556962, - "grad_norm": 0.21950506053306648, - "learning_rate": 4.3941129475752795e-07, - "loss": 0.1201, - "step": 1156 - }, - { - "epoch": 4.686582278481013, - "grad_norm": 0.20730175876913146, - "learning_rate": 4.27657778143431e-07, - "loss": 0.1279, - "step": 1157 - }, - { - "epoch": 4.690632911392405, - "grad_norm": 0.20838929170563114, - "learning_rate": 4.1606189448287757e-07, - "loss": 0.1283, - "step": 1158 - }, - { - "epoch": 4.694683544303797, - "grad_norm": 0.20839511874075492, - "learning_rate": 4.046237371675177e-07, - "loss": 0.1059, - "step": 1159 - }, - { - "epoch": 4.69873417721519, - "grad_norm": 0.20355928247494684, - "learning_rate": 3.9334339831869963e-07, - "loss": 0.1246, - "step": 1160 - }, - { - "epoch": 4.702784810126582, - "grad_norm": 0.20633094453618736, - "learning_rate": 3.8222096878671955e-07, - "loss": 0.1166, - "step": 1161 - }, - { - "epoch": 4.706835443037974, - "grad_norm": 0.20431024970942696, - "learning_rate": 3.7125653815009545e-07, - "loss": 0.1282, - "step": 1162 - }, - { - "epoch": 4.710886075949367, - "grad_norm": 0.20623206191042928, - "learning_rate": 3.6045019471484974e-07, - "loss": 0.1177, - "step": 1163 - }, - { - "epoch": 4.714936708860759, - "grad_norm": 0.2144716107826845, - "learning_rate": 3.498020255137813e-07, - "loss": 0.1192, - "step": 1164 - }, - { - "epoch": 4.718987341772152, - "grad_norm": 0.21757555870967651, - "learning_rate": 3.393121163057811e-07, - "loss": 0.1155, - "step": 1165 - }, - { - "epoch": 4.723037974683544, - "grad_norm": 0.2129426113170404, - "learning_rate": 3.289805515751399e-07, - "loss": 0.13, - "step": 1166 - }, - { - "epoch": 4.727088607594936, - "grad_norm": 0.20367334408369145, - "learning_rate": 3.188074145308573e-07, - "loss": 0.1072, - "step": 1167 - }, - { - "epoch": 4.731139240506329, - "grad_norm": 0.20341722591197034, - "learning_rate": 3.087927871059804e-07, - "loss": 0.1174, - "step": 1168 - }, - { - "epoch": 4.735189873417721, - "grad_norm": 0.20920904027456105, - "learning_rate": 2.989367499569418e-07, - "loss": 0.12, - "step": 1169 - }, - { - "epoch": 4.739240506329114, - "grad_norm": 0.20504474871296832, - "learning_rate": 2.8923938246290917e-07, - "loss": 0.1405, - "step": 1170 - }, - { - "epoch": 4.743291139240506, - "grad_norm": 0.2042104154739712, - "learning_rate": 2.7970076272514804e-07, - "loss": 0.1175, - "step": 1171 - }, - { - "epoch": 4.7473417721518985, - "grad_norm": 0.2017294394627823, - "learning_rate": 2.703209675663887e-07, - "loss": 0.1224, - "step": 1172 - }, - { - "epoch": 4.751392405063291, - "grad_norm": 0.1980570388527049, - "learning_rate": 2.6110007253021374e-07, - "loss": 0.1159, - "step": 1173 - }, - { - "epoch": 4.7554430379746835, - "grad_norm": 0.20035308721355388, - "learning_rate": 2.520381518804471e-07, - "loss": 0.0996, - "step": 1174 - }, - { - "epoch": 4.759493670886076, - "grad_norm": 0.20241333870484432, - "learning_rate": 2.4313527860054585e-07, - "loss": 0.1152, - "step": 1175 + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01069591473788023, + "step": 1134, + "valid_targets_mean": 3888.2, + "valid_targets_min": 1501 }, { - "epoch": 4.7635443037974685, - "grad_norm": 0.20014095559607262, - "learning_rate": 2.343915243930317e-07, - "loss": 0.108, - "step": 1176 - }, - { - "epoch": 4.767594936708861, - "grad_norm": 0.2128426050120164, - "learning_rate": 2.2580695967889367e-07, - "loss": 0.1127, - "step": 1177 - }, - { - "epoch": 4.7716455696202535, - "grad_norm": 0.19967549439905458, - "learning_rate": 2.1738165359704189e-07, - "loss": 0.1178, - "step": 1178 + "epoch": 4.595443037974683, + "grad_norm": 0.24467054996033144, + "learning_rate": 8.101405277100549e-07, + "loss": 0.1145, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012983351945877075, + "step": 1135, + "valid_targets_mean": 4395.8, + "valid_targets_min": 983 + }, + { + "epoch": 4.599493670886076, + "grad_norm": 0.23952615604287592, + "learning_rate": 7.942840756714077e-07, + "loss": 0.1089, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011914948001503944, + "step": 1136, + "valid_targets_mean": 3921.2, + "valid_targets_min": 1482 + }, + { + "epoch": 4.603544303797468, + "grad_norm": 0.24743934229288847, + "learning_rate": 7.785811920255093e-07, + "loss": 0.1117, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01661062426865101, + "step": 1137, + "valid_targets_mean": 4857.1, + "valid_targets_min": 915 }, { - "epoch": 4.775696202531646, - "grad_norm": 0.21115778749209888, - "learning_rate": 2.0911567400373257e-07, + "epoch": 4.6075949367088604, + "grad_norm": 0.243337849066111, + "learning_rate": 7.630320023322024e-07, "loss": 0.1168, - "step": 1179 + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009833608753979206, + "step": 1138, + "valid_targets_mean": 3292.8, + "valid_targets_min": 639 }, { - "epoch": 4.779746835443038, - "grad_norm": 0.21039004378596235, - "learning_rate": 2.0100908747202607e-07, - "loss": 0.0961, - "step": 1180 - }, - { - "epoch": 4.783797468354431, - "grad_norm": 0.19953766867394895, - "learning_rate": 1.9306195929125638e-07, - "loss": 0.11, - "step": 1181 - }, - { - "epoch": 4.787848101265823, - "grad_norm": 0.20760586525440083, - "learning_rate": 1.8527435346650247e-07, - "loss": 0.1224, - "step": 1182 - }, - { - "epoch": 4.791898734177215, - "grad_norm": 0.2060238614534191, - "learning_rate": 1.7764633271807108e-07, - "loss": 0.12, - "step": 1183 - }, - { - "epoch": 4.795949367088608, - "grad_norm": 0.21099171015572574, - "learning_rate": 1.7017795848099262e-07, - "loss": 0.1163, - "step": 1184 - }, - { - "epoch": 4.8, - "grad_norm": 0.20425209892239116, - "learning_rate": 1.6286929090452596e-07, - "loss": 0.107, - "step": 1185 - }, - { - "epoch": 4.804050632911393, - "grad_norm": 0.20896229654474527, - "learning_rate": 1.557203888516745e-07, + "epoch": 4.611645569620253, + "grad_norm": 0.24721697537601303, + "learning_rate": 7.476366309223903e-07, "loss": 0.1118, - "step": 1186 - }, - { - "epoch": 4.808101265822785, - "grad_norm": 0.20505541101005698, - "learning_rate": 1.487313098987131e-07, - "loss": 0.1188, - "step": 1187 - }, - { - "epoch": 4.812151898734177, - "grad_norm": 0.20922577673677953, - "learning_rate": 1.4190211033472402e-07, - "loss": 0.1274, - "step": 1188 - }, - { - "epoch": 4.81620253164557, - "grad_norm": 0.21116939541376195, - "learning_rate": 1.3523284516113955e-07, - "loss": 0.106, - "step": 1189 - }, - { - "epoch": 4.820253164556962, - "grad_norm": 0.20768343006102155, - "learning_rate": 1.2872356809130682e-07, - "loss": 0.1076, - "step": 1190 - }, - { - "epoch": 4.824303797468355, - "grad_norm": 0.20468417782779227, - "learning_rate": 1.2237433155004807e-07, - "loss": 0.1223, - "step": 1191 - }, - { - "epoch": 4.828354430379747, - "grad_norm": 0.2098082124146172, - "learning_rate": 1.1618518667323886e-07, - "loss": 0.1039, - "step": 1192 - }, - { - "epoch": 4.832405063291139, - "grad_norm": 0.21051427840192627, - "learning_rate": 1.1015618330740385e-07, - "loss": 0.1292, - "step": 1193 + "loss_nan_ranks": 0, + "loss_rank_avg": 0.008912472054362297, + "step": 1139, + "valid_targets_mean": 2869.4, + "valid_targets_min": 1178 }, { - "epoch": 4.836455696202532, - "grad_norm": 0.19749426940217815, - "learning_rate": 1.042873700093061e-07, - "loss": 0.1318, - "step": 1194 - }, - { - "epoch": 4.840506329113924, - "grad_norm": 0.19820803691814487, - "learning_rate": 9.857879404556291e-08, - "loss": 0.1231, - "step": 1195 + "epoch": 4.6156962025316455, + "grad_norm": 0.24031897463539212, + "learning_rate": 7.323952008970537e-07, + "loss": 0.1111, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013830740936100483, + "step": 1140, + "valid_targets_mean": 4143.6, + "valid_targets_min": 1243 + }, + { + "epoch": 4.619746835443038, + "grad_norm": 0.2470119895734213, + "learning_rate": 7.173078341262618e-07, + "loss": 0.1109, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.010928130708634853, + "step": 1141, + "valid_targets_mean": 3865.4, + "valid_targets_min": 1214 + }, + { + "epoch": 4.6237974683544305, + "grad_norm": 0.2409629382375663, + "learning_rate": 7.023746512481989e-07, + "loss": 0.1121, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011756358668208122, + "step": 1142, + "valid_targets_mean": 3866.8, + "valid_targets_min": 745 }, { - "epoch": 4.844556962025316, - "grad_norm": 0.19930057255770592, - "learning_rate": 9.303050139225722e-08, + "epoch": 4.627848101265823, + "grad_norm": 0.23363037887291022, + "learning_rate": 6.875957716681902e-07, + "loss": 0.1119, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01830505207180977, + "step": 1143, + "valid_targets_mean": 5421.4, + "valid_targets_min": 995 + }, + { + "epoch": 4.6318987341772155, + "grad_norm": 0.24180508844289644, + "learning_rate": 6.729713135577687e-07, + "loss": 0.1075, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009786557406187057, + "step": 1144, + "valid_targets_mean": 3802.4, + "valid_targets_min": 904 + }, + { + "epoch": 4.635949367088608, + "grad_norm": 0.23052849419457697, + "learning_rate": 6.585013938537143e-07, + "loss": 0.105, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01294481847435236, + "step": 1145, + "valid_targets_mean": 4412.9, + "valid_targets_min": 1270 + }, + { + "epoch": 4.64, + "grad_norm": 0.2309435146050416, + "learning_rate": 6.441861282571115e-07, + "loss": 0.1088, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014942819252610207, + "step": 1146, + "valid_targets_mean": 5691.0, + "valid_targets_min": 1458 + }, + { + "epoch": 4.644050632911393, + "grad_norm": 0.23113695571145013, + "learning_rate": 6.300256312324471e-07, "loss": 0.1049, - "step": 1196 + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015250153839588165, + "step": 1147, + "valid_targets_mean": 5221.1, + "valid_targets_min": 1262 }, { - "epoch": 4.848607594936709, - "grad_norm": 0.20497167366532656, - "learning_rate": 8.76425367345779e-08, - "loss": 0.1171, - "step": 1197 + "epoch": 4.648101265822785, + "grad_norm": 0.23932963878845193, + "learning_rate": 6.160200160066776e-07, + "loss": 0.1175, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013173839077353477, + "step": 1148, + "valid_targets_mean": 4475.6, + "valid_targets_min": 1448 }, { - "epoch": 4.852658227848101, - "grad_norm": 0.2087849578411844, - "learning_rate": 8.241494346644897e-08, - "loss": 0.1256, - "step": 1198 + "epoch": 4.652151898734177, + "grad_norm": 0.2604647410402613, + "learning_rate": 6.021693945683327e-07, + "loss": 0.1078, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012512031942605972, + "step": 1149, + "valid_targets_mean": 3842.2, + "valid_targets_min": 1361 }, { - "epoch": 4.856708860759493, - "grad_norm": 0.20960047207966875, - "learning_rate": 7.734776369019204e-08, - "loss": 0.1145, - "step": 1199 + "epoch": 4.65620253164557, + "grad_norm": 0.2523420469817438, + "learning_rate": 5.884738776666044e-07, + "loss": 0.1102, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013408460654318333, + "step": 1150, + "valid_targets_mean": 4382.5, + "valid_targets_min": 1186 }, { - "epoch": 4.860759493670886, - "grad_norm": 0.20709045703359283, - "learning_rate": 7.244103821617332e-08, - "loss": 0.1276, - "step": 1200 + "epoch": 4.660253164556962, + "grad_norm": 0.2435123644982836, + "learning_rate": 5.749335748104878e-07, + "loss": 0.117, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013424522243440151, + "step": 1151, + "valid_targets_mean": 4520.6, + "valid_targets_min": 1361 }, { - "epoch": 4.864810126582278, - "grad_norm": 0.20764505959342205, - "learning_rate": 6.769480656248606e-08, - "loss": 0.1127, - "step": 1201 + "epoch": 4.664303797468355, + "grad_norm": 0.2381173828939779, + "learning_rate": 5.615485942678778e-07, + "loss": 0.1154, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01639263704419136, + "step": 1152, + "valid_targets_mean": 5444.4, + "valid_targets_min": 1579 }, { - "epoch": 4.868860759493671, - "grad_norm": 0.20732384965327175, - "learning_rate": 6.310910695462635e-08, - "loss": 0.1133, - "step": 1202 + "epoch": 4.668354430379747, + "grad_norm": 0.23270087042794557, + "learning_rate": 5.483190430647289e-07, + "loss": 0.1072, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01745782047510147, + "step": 1153, + "valid_targets_mean": 6098.2, + "valid_targets_min": 1126 }, { - "epoch": 4.872911392405063, - "grad_norm": 0.20349684428929318, - "learning_rate": 5.8683976325191185e-08, - "loss": 0.1142, - "step": 1203 + "epoch": 4.672405063291139, + "grad_norm": 0.24265278992980324, + "learning_rate": 5.352450269841747e-07, + "loss": 0.108, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014374315738677979, + "step": 1154, + "valid_targets_mean": 4428.9, + "valid_targets_min": 1319 + }, + { + "epoch": 4.676455696202532, + "grad_norm": 0.22989885206535504, + "learning_rate": 5.22326650565701e-07, + "loss": 0.1107, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01609790325164795, + "step": 1155, + "valid_targets_mean": 4971.1, + "valid_targets_min": 1240 + }, + { + "epoch": 4.680506329113924, + "grad_norm": 0.22750941745027642, + "learning_rate": 5.095640171043048e-07, + "loss": 0.1066, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009911238215863705, + "step": 1156, + "valid_targets_mean": 3801.1, + "valid_targets_min": 1198 }, { - "epoch": 4.876962025316455, - "grad_norm": 0.20097560676224976, - "learning_rate": 5.4419450313571984e-08, - "loss": 0.1149, - "step": 1204 + "epoch": 4.684556962025317, + "grad_norm": 0.24098958521920144, + "learning_rate": 4.969572286496594e-07, + "loss": 0.1119, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.022853152826428413, + "step": 1157, + "valid_targets_mean": 5341.6, + "valid_targets_min": 1436 }, { - "epoch": 4.881012658227848, - "grad_norm": 0.2023305140232061, - "learning_rate": 5.031556326567488e-08, - "loss": 0.1242, - "step": 1205 + "epoch": 4.688607594936709, + "grad_norm": 0.2392105757928696, + "learning_rate": 4.845063860053079e-07, + "loss": 0.1129, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016803249716758728, + "step": 1158, + "valid_targets_mean": 5111.2, + "valid_targets_min": 1302 }, { - "epoch": 4.88506329113924, - "grad_norm": 0.20571007500161037, - "learning_rate": 4.637234823364312e-08, - "loss": 0.1094, - "step": 1206 + "epoch": 4.692658227848101, + "grad_norm": 0.24206555145149455, + "learning_rate": 4.7221158872785997e-07, + "loss": 0.1129, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011856446042656898, + "step": 1159, + "valid_targets_mean": 4355.4, + "valid_targets_min": 1073 + }, + { + "epoch": 4.696708860759494, + "grad_norm": 0.24045095578781692, + "learning_rate": 4.600729351261857e-07, + "loss": 0.1108, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017727192491292953, + "step": 1160, + "valid_targets_mean": 5086.9, + "valid_targets_min": 1256 + }, + { + "epoch": 4.700759493670886, + "grad_norm": 0.23286053027211254, + "learning_rate": 4.4809052226063355e-07, + "loss": 0.1101, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009751537814736366, + "step": 1161, + "valid_targets_mean": 3784.0, + "valid_targets_min": 1066 }, { - "epoch": 4.889113924050633, - "grad_norm": 0.2025404660661232, - "learning_rate": 4.258983697558838e-08, - "loss": 0.1087, - "step": 1207 + "epoch": 4.704810126582278, + "grad_norm": 0.25441081024171247, + "learning_rate": 4.3626444594226046e-07, + "loss": 0.1066, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015385766513645649, + "step": 1162, + "valid_targets_mean": 4484.6, + "valid_targets_min": 1134 }, { - "epoch": 4.8931645569620255, - "grad_norm": 0.20330971685864385, - "learning_rate": 3.896805995533548e-08, - "loss": 0.1147, - "step": 1208 + "epoch": 4.708860759493671, + "grad_norm": 0.24446375599635572, + "learning_rate": 4.245948007320588e-07, + "loss": 0.1175, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012111935764551163, + "step": 1163, + "valid_targets_mean": 3709.1, + "valid_targets_min": 1099 }, { - "epoch": 4.8972151898734175, - "grad_norm": 0.21059128788978063, - "learning_rate": 3.550704634218028e-08, - "loss": 0.1117, - "step": 1209 + "epoch": 4.712911392405063, + "grad_norm": 0.23130476730298755, + "learning_rate": 4.130816799401971e-07, + "loss": 0.1058, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.008287204429507256, + "step": 1164, + "valid_targets_mean": 3054.6, + "valid_targets_min": 1392 + }, + { + "epoch": 4.716962025316455, + "grad_norm": 0.24435314164443975, + "learning_rate": 4.017251756252827e-07, + "loss": 0.1164, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013173110783100128, + "step": 1165, + "valid_targets_mean": 3855.7, + "valid_targets_min": 1333 + }, + { + "epoch": 4.721012658227848, + "grad_norm": 0.223425818597529, + "learning_rate": 3.9052537859362473e-07, + "loss": 0.1154, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012818815186619759, + "step": 1166, + "valid_targets_mean": 4592.1, + "valid_targets_min": 1370 + }, + { + "epoch": 4.72506329113924, + "grad_norm": 0.2345797216984834, + "learning_rate": 3.7948237839849687e-07, + "loss": 0.1077, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.020206928253173828, + "step": 1167, + "valid_targets_mean": 5562.1, + "valid_targets_min": 1174 + }, + { + "epoch": 4.729113924050633, + "grad_norm": 0.24694843961663482, + "learning_rate": 3.6859626333943797e-07, + "loss": 0.1094, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011120930314064026, + "step": 1168, + "valid_targets_mean": 4122.5, + "valid_targets_min": 1409 }, { - "epoch": 4.9012658227848105, - "grad_norm": 0.20481831153392321, - "learning_rate": 3.2206824010647676e-08, - "loss": 0.1172, - "step": 1210 + "epoch": 4.733164556962025, + "grad_norm": 0.23916710669800414, + "learning_rate": 3.578671204615347e-07, + "loss": 0.1155, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.010612422600388527, + "step": 1169, + "valid_targets_mean": 3807.1, + "valid_targets_min": 1177 }, { - "epoch": 4.905316455696203, - "grad_norm": 0.20538447024850046, - "learning_rate": 2.9067419540278476e-08, - "loss": 0.122, - "step": 1211 + "epoch": 4.737215189873417, + "grad_norm": 0.23534285201891775, + "learning_rate": 3.472950355547267e-07, + "loss": 0.117, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.017033474519848824, + "step": 1170, + "valid_targets_mean": 5603.9, + "valid_targets_min": 1779 + }, + { + "epoch": 4.74126582278481, + "grad_norm": 0.21909731195941826, + "learning_rate": 3.3688009315312244e-07, + "loss": 0.1056, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01086464338004589, + "step": 1171, + "valid_targets_mean": 4267.8, + "valid_targets_min": 1056 + }, + { + "epoch": 4.745316455696202, + "grad_norm": 0.24388805731338395, + "learning_rate": 3.266223765343224e-07, + "loss": 0.11, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015588557347655296, + "step": 1172, + "valid_targets_mean": 4333.6, + "valid_targets_min": 1172 + }, + { + "epoch": 4.749367088607595, + "grad_norm": 0.22955891283346133, + "learning_rate": 3.16521967718757e-07, + "loss": 0.1115, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016997043043375015, + "step": 1173, + "valid_targets_mean": 5823.1, + "valid_targets_min": 1068 + }, + { + "epoch": 4.7534177215189874, + "grad_norm": 0.23474462024092194, + "learning_rate": 3.0657894746902726e-07, + "loss": 0.098, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0072157541289925575, + "step": 1174, + "valid_targets_mean": 2820.7, + "valid_targets_min": 986 + }, + { + "epoch": 4.7574683544303795, + "grad_norm": 0.23196623024831636, + "learning_rate": 2.9679339528925654e-07, + "loss": 0.1042, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01756441406905651, + "step": 1175, + "valid_targets_mean": 5168.0, + "valid_targets_min": 956 + }, + { + "epoch": 4.7615189873417725, + "grad_norm": 0.2459616528038731, + "learning_rate": 2.871653894244597e-07, + "loss": 0.108, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.010087177157402039, + "step": 1176, + "valid_targets_mean": 3164.2, + "valid_targets_min": 1003 + }, + { + "epoch": 4.7655696202531646, + "grad_norm": 0.23554609620846492, + "learning_rate": 2.7769500685991266e-07, + "loss": 0.1085, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013887660577893257, + "step": 1177, + "valid_targets_mean": 4648.5, + "valid_targets_min": 876 + }, + { + "epoch": 4.769620253164557, + "grad_norm": 0.23193207006204014, + "learning_rate": 2.683823233205485e-07, + "loss": 0.1046, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012032478116452694, + "step": 1178, + "valid_targets_mean": 3907.8, + "valid_targets_min": 1086 + }, + { + "epoch": 4.77367088607595, + "grad_norm": 0.2510058757439523, + "learning_rate": 2.592274132703287e-07, + "loss": 0.1052, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014164448715746403, + "step": 1179, + "valid_targets_mean": 4645.2, + "valid_targets_min": 998 + }, + { + "epoch": 4.777721518987342, + "grad_norm": 0.23175267801304952, + "learning_rate": 2.502303499116754e-07, + "loss": 0.1069, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009768212214112282, + "step": 1180, + "valid_targets_mean": 3876.7, + "valid_targets_min": 967 + }, + { + "epoch": 4.781772151898734, + "grad_norm": 0.24370100149211857, + "learning_rate": 2.413912051848666e-07, + "loss": 0.0988, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009045436978340149, + "step": 1181, + "valid_targets_mean": 3097.1, + "valid_targets_min": 1008 + }, + { + "epoch": 4.785822784810127, + "grad_norm": 0.2370454588564179, + "learning_rate": 2.3271004976746613e-07, + "loss": 0.1159, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011587612330913544, + "step": 1182, + "valid_targets_mean": 3934.2, + "valid_targets_min": 1371 }, { - "epoch": 4.909367088607595, - "grad_norm": 0.20292283079286413, - "learning_rate": 2.6088858215400638e-08, - "loss": 0.1245, - "step": 1212 + "epoch": 4.789873417721519, + "grad_norm": 0.23815121483817714, + "learning_rate": 2.2418695307376393e-07, + "loss": 0.1128, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014157026074826717, + "step": 1183, + "valid_targets_mean": 4409.0, + "valid_targets_min": 1453 }, { - "epoch": 4.913417721518988, - "grad_norm": 0.21145456513897448, - "learning_rate": 2.3271164024940564e-08, - "loss": 0.1318, - "step": 1213 + "epoch": 4.793924050632912, + "grad_norm": 0.23783821363379468, + "learning_rate": 2.1582198325420968e-07, + "loss": 0.1133, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.010271416045725346, + "step": 1184, + "valid_targets_mean": 3274.3, + "valid_targets_min": 1086 + }, + { + "epoch": 4.797974683544304, + "grad_norm": 0.23229256280874674, + "learning_rate": 2.0761520719488447e-07, + "loss": 0.0997, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.006490450352430344, + "step": 1185, + "valid_targets_mean": 2632.8, + "valid_targets_min": 1157 + }, + { + "epoch": 4.802025316455696, + "grad_norm": 0.24419609664439176, + "learning_rate": 1.9956669051694798e-07, + "loss": 0.1104, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012339601293206215, + "step": 1186, + "valid_targets_mean": 3850.8, + "valid_targets_min": 1617 + }, + { + "epoch": 4.806075949367089, + "grad_norm": 0.24897996532180455, + "learning_rate": 1.916764975761254e-07, + "loss": 0.1164, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012995798140764236, + "step": 1187, + "valid_targets_mean": 3230.2, + "valid_targets_min": 688 + }, + { + "epoch": 4.810126582278481, + "grad_norm": 0.24236818776265567, + "learning_rate": 1.839446914621923e-07, + "loss": 0.1144, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01723683439195156, + "step": 1188, + "valid_targets_mean": 5158.0, + "valid_targets_min": 1305 + }, + { + "epoch": 4.814177215189874, + "grad_norm": 0.23727219276751094, + "learning_rate": 1.7637133399845962e-07, + "loss": 0.1071, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012703627347946167, + "step": 1189, + "valid_targets_mean": 4795.9, + "valid_targets_min": 1235 + }, + { + "epoch": 4.818227848101266, + "grad_norm": 0.24391400967421373, + "learning_rate": 1.6895648574129397e-07, + "loss": 0.1121, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01454693078994751, + "step": 1190, + "valid_targets_mean": 4082.4, + "valid_targets_min": 1370 }, { - "epoch": 4.91746835443038, - "grad_norm": 0.210870267562045, - "learning_rate": 2.061435966221881e-08, - "loss": 0.1167, - "step": 1214 + "epoch": 4.822278481012658, + "grad_norm": 0.24677221947667136, + "learning_rate": 1.6170020597962465e-07, + "loss": 0.1078, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01703988015651703, + "step": 1191, + "valid_targets_mean": 4989.6, + "valid_targets_min": 1047 }, { - "epoch": 4.921518987341772, - "grad_norm": 0.21228156298120943, - "learning_rate": 1.811846652477245e-08, - "loss": 0.1123, - "step": 1215 + "epoch": 4.826329113924051, + "grad_norm": 0.23896025862435885, + "learning_rate": 1.5460255273447078e-07, + "loss": 0.1125, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015895096585154533, + "step": 1192, + "valid_targets_mean": 5156.8, + "valid_targets_min": 957 }, { - "epoch": 4.925569620253165, - "grad_norm": 0.20151352196850916, - "learning_rate": 1.5783504714184106e-08, - "loss": 0.1129, - "step": 1216 + "epoch": 4.830379746835443, + "grad_norm": 0.23823056699462464, + "learning_rate": 1.4766358275847715e-07, + "loss": 0.1047, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.007428309414535761, + "step": 1193, + "valid_targets_mean": 2911.4, + "valid_targets_min": 1249 }, { - "epoch": 4.929620253164557, - "grad_norm": 0.2099959540784521, - "learning_rate": 1.360949303591097e-08, + "epoch": 4.834430379746835, + "grad_norm": 0.2247744160174395, + "learning_rate": 1.4088335153546573e-07, "loss": 0.1202, - "step": 1217 - }, - { - "epoch": 4.93367088607595, - "grad_norm": 0.20824123991142346, - "learning_rate": 1.1596448999144916e-08, - "loss": 0.1212, - "step": 1218 - }, - { - "epoch": 4.937721518987342, - "grad_norm": 0.20295686813461908, - "learning_rate": 9.744388816668172e-09, - "loss": 0.1275, - "step": 1219 + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01684490591287613, + "step": 1194, + "valid_targets_mean": 5795.6, + "valid_targets_min": 1354 }, { - "epoch": 4.941772151898734, - "grad_norm": 0.2005956711554205, - "learning_rate": 8.05332740472009e-09, + "epoch": 4.838481012658228, + "grad_norm": 0.2272848805922521, + "learning_rate": 1.3426191327998496e-07, "loss": 0.1122, - "step": 1220 - }, - { - "epoch": 4.945822784810127, - "grad_norm": 0.20202159349472282, - "learning_rate": 6.523278382872811e-09, - "loss": 0.1275, - "step": 1221 - }, - { - "epoch": 4.949873417721519, - "grad_norm": 0.21763566583306004, - "learning_rate": 5.15425407393133e-09, - "loss": 0.1111, - "step": 1222 + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015583060681819916, + "step": 1195, + "valid_targets_mean": 5315.5, + "valid_targets_min": 1446 }, { - "epoch": 4.953924050632912, - "grad_norm": 0.20255252844405391, - "learning_rate": 3.94626550383137e-09, - "loss": 0.1253, - "step": 1223 + "epoch": 4.84253164556962, + "grad_norm": 0.24493188674129462, + "learning_rate": 1.277993209368744e-07, + "loss": 0.106, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011442995630204678, + "step": 1196, + "valid_targets_mean": 3736.6, + "valid_targets_min": 1506 + }, + { + "epoch": 4.846582278481012, + "grad_norm": 0.2382142438109627, + "learning_rate": 1.214956261808542e-07, + "loss": 0.1069, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013297231867909431, + "step": 1197, + "valid_targets_mean": 4526.4, + "valid_targets_min": 1470 + }, + { + "epoch": 4.850632911392405, + "grad_norm": 0.2264295628229462, + "learning_rate": 1.1535087941609846e-07, + "loss": 0.1076, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012269329279661179, + "step": 1198, + "valid_targets_mean": 4186.8, + "valid_targets_min": 1153 }, { - "epoch": 4.957974683544304, - "grad_norm": 0.21164634900002333, - "learning_rate": 2.899322401546112e-09, - "loss": 0.1102, - "step": 1224 + "epoch": 4.854683544303797, + "grad_norm": 0.2307526058124495, + "learning_rate": 1.0936512977583802e-07, + "loss": 0.1121, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.010853808373212814, + "step": 1199, + "valid_targets_mean": 3549.2, + "valid_targets_min": 1597 + }, + { + "epoch": 4.85873417721519, + "grad_norm": 0.23868588844023084, + "learning_rate": 1.0353842512196732e-07, + "loss": 0.113, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009980385191738605, + "step": 1200, + "valid_targets_mean": 3069.8, + "valid_targets_min": 1044 + }, + { + "epoch": 4.862784810126582, + "grad_norm": 0.23795196443442018, + "learning_rate": 9.787081204466253e-08, + "loss": 0.1183, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015351405367255211, + "step": 1201, + "valid_targets_mean": 4512.4, + "valid_targets_min": 1229 }, { - "epoch": 4.962025316455696, - "grad_norm": 0.2106747639379142, - "learning_rate": 2.013433199010706e-09, - "loss": 0.1129, - "step": 1225 + "epoch": 4.866835443037974, + "grad_norm": 0.25403745028440267, + "learning_rate": 9.236233586200627e-08, + "loss": 0.1181, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01124050747603178, + "step": 1202, + "valid_targets_mean": 3405.4, + "valid_targets_min": 812 + }, + { + "epoch": 4.870886075949367, + "grad_norm": 0.23091009126628026, + "learning_rate": 8.70130406196279e-08, + "loss": 0.1053, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015791581943631172, + "step": 1203, + "valid_targets_mean": 5422.2, + "valid_targets_min": 1240 + }, + { + "epoch": 4.874936708860759, + "grad_norm": 0.24714675801931965, + "learning_rate": 8.182296909035047e-08, + "loss": 0.1097, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013445362448692322, + "step": 1204, + "valid_targets_mean": 3916.1, + "valid_targets_min": 1263 + }, + { + "epoch": 4.878987341772152, + "grad_norm": 0.23238400031620596, + "learning_rate": 7.679216277384439e-08, + "loss": 0.1089, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013845035806298256, + "step": 1205, + "valid_targets_mean": 4798.1, + "valid_targets_min": 1292 + }, + { + "epoch": 4.883037974683544, + "grad_norm": 0.2433693133823994, + "learning_rate": 7.19206618963053e-08, + "loss": 0.1141, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.008713947609066963, + "step": 1206, + "valid_targets_mean": 3317.1, + "valid_targets_min": 1387 + }, + { + "epoch": 4.8870886075949365, + "grad_norm": 0.24241122111292934, + "learning_rate": 6.720850541012347e-08, + "loss": 0.1089, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013348542153835297, + "step": 1207, + "valid_targets_mean": 4618.2, + "valid_targets_min": 1733 + }, + { + "epoch": 4.891139240506329, + "grad_norm": 0.2304352158805518, + "learning_rate": 6.265573099357714e-08, + "loss": 0.1072, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011430079117417336, + "step": 1208, + "valid_targets_mean": 4107.3, + "valid_targets_min": 1089 }, { - "epoch": 4.966075949367089, - "grad_norm": 0.20158062763575213, - "learning_rate": 1.2886050310556563e-09, - "loss": 0.1321, - "step": 1226 + "epoch": 4.8951898734177215, + "grad_norm": 0.24145238610331443, + "learning_rate": 5.826237505053067e-08, + "loss": 0.108, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011784245260059834, + "step": 1209, + "valid_targets_mean": 4086.8, + "valid_targets_min": 1417 + }, + { + "epoch": 4.899240506329114, + "grad_norm": 0.2450341203448637, + "learning_rate": 5.4028472710137e-08, + "loss": 0.1136, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014564326032996178, + "step": 1210, + "valid_targets_mean": 4164.2, + "valid_targets_min": 1538 + }, + { + "epoch": 4.9032911392405065, + "grad_norm": 0.24354061349344405, + "learning_rate": 4.9954057826571146e-08, + "loss": 0.1142, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011334458366036415, + "step": 1211, + "valid_targets_mean": 4018.5, + "valid_targets_min": 1525 + }, + { + "epoch": 4.907341772151899, + "grad_norm": 0.2315542455299904, + "learning_rate": 4.6039162978743776e-08, + "loss": 0.1104, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.0195978581905365, + "step": 1212, + "valid_targets_mean": 6231.7, + "valid_targets_min": 1774 + }, + { + "epoch": 4.911392405063291, + "grad_norm": 0.23313479210470323, + "learning_rate": 4.228381947004812e-08, + "loss": 0.1116, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014019984751939774, + "step": 1213, + "valid_targets_mean": 4620.9, + "valid_targets_min": 1882 + }, + { + "epoch": 4.915443037974684, + "grad_norm": 0.22855727314511912, + "learning_rate": 3.868805732811343e-08, + "loss": 0.1112, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.007369753904640675, + "step": 1214, + "valid_targets_mean": 3527.6, + "valid_targets_min": 1183 + }, + { + "epoch": 4.919493670886076, + "grad_norm": 0.25236996871942824, + "learning_rate": 3.525190530455636e-08, + "loss": 0.1108, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01704399660229683, + "step": 1215, + "valid_targets_mean": 5038.8, + "valid_targets_min": 1361 + }, + { + "epoch": 4.923544303797469, + "grad_norm": 0.2339802304650147, + "learning_rate": 3.197539087475887e-08, + "loss": 0.1088, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014040106907486916, + "step": 1216, + "valid_targets_mean": 5526.5, + "valid_targets_min": 1119 + }, + { + "epoch": 4.927594936708861, + "grad_norm": 0.243821254252586, + "learning_rate": 2.8858540237641786e-08, + "loss": 0.1023, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01630534417927265, + "step": 1217, + "valid_targets_mean": 4622.2, + "valid_targets_min": 1320 + }, + { + "epoch": 4.931645569620253, + "grad_norm": 0.23514977499476192, + "learning_rate": 2.5901378315458247e-08, + "loss": 0.1164, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015973731875419617, + "step": 1218, + "valid_targets_mean": 5489.3, + "valid_targets_min": 941 + }, + { + "epoch": 4.935696202531646, + "grad_norm": 0.23066847016906347, + "learning_rate": 2.3103928753598347e-08, + "loss": 0.1082, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.014494219794869423, + "step": 1219, + "valid_targets_mean": 4435.1, + "valid_targets_min": 1143 + }, + { + "epoch": 4.939746835443038, + "grad_norm": 0.23599418339991424, + "learning_rate": 2.0466213920393718e-08, + "loss": 0.1134, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.009911555796861649, + "step": 1220, + "valid_targets_mean": 4003.4, + "valid_targets_min": 1057 + }, + { + "epoch": 4.943797468354431, + "grad_norm": 0.23839899590308894, + "learning_rate": 1.7988254906942115e-08, + "loss": 0.1136, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.016164466738700867, + "step": 1221, + "valid_targets_mean": 5077.0, + "valid_targets_min": 1211 + }, + { + "epoch": 4.947848101265823, + "grad_norm": 0.23916470486920938, + "learning_rate": 1.5670071526936447e-08, + "loss": 0.1126, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.01405949704349041, + "step": 1222, + "valid_targets_mean": 4898.8, + "valid_targets_min": 1266 + }, + { + "epoch": 4.951898734177215, + "grad_norm": 0.21741608069325183, + "learning_rate": 1.3511682316509344e-08, + "loss": 0.1014, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011473024263978004, + "step": 1223, + "valid_targets_mean": 4689.2, + "valid_targets_min": 1033 + }, + { + "epoch": 4.955949367088608, + "grad_norm": 0.2292728694270739, + "learning_rate": 1.1513104534086605e-08, + "loss": 0.1031, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012841494753956795, + "step": 1224, + "valid_targets_mean": 4275.0, + "valid_targets_min": 1314 + }, + { + "epoch": 4.96, + "grad_norm": 0.25041866378749583, + "learning_rate": 9.67435416023843e-09, + "loss": 0.1033, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018508005887269974, + "step": 1225, + "valid_targets_mean": 4550.4, + "valid_targets_min": 1057 + }, + { + "epoch": 4.964050632911392, + "grad_norm": 0.22277492122517895, + "learning_rate": 7.995445897566178e-09, + "loss": 0.1046, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.010599160566926003, + "step": 1226, + "valid_targets_mean": 4327.2, + "valid_targets_min": 1313 + }, + { + "epoch": 4.968101265822785, + "grad_norm": 0.2367509043949318, + "learning_rate": 6.476393170578021e-09, + "loss": 0.1001, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.011691408231854439, + "step": 1227, + "valid_targets_mean": 4577.2, + "valid_targets_min": 936 + }, + { + "epoch": 4.972151898734177, + "grad_norm": 0.22551297933814682, + "learning_rate": 5.11720812557348e-09, + "loss": 0.1036, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013229941949248314, + "step": 1228, + "valid_targets_mean": 5726.6, + "valid_targets_min": 869 + }, + { + "epoch": 4.976202531645569, + "grad_norm": 0.23734494557382904, + "learning_rate": 3.917901630563492e-09, + "loss": 0.1112, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.022276364266872406, + "step": 1229, + "valid_targets_mean": 6153.0, + "valid_targets_min": 1226 + }, + { + "epoch": 4.980253164556962, + "grad_norm": 0.23363612704383607, + "learning_rate": 2.878483275170485e-09, + "loss": 0.1163, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.015504706650972366, + "step": 1230, + "valid_targets_mean": 5061.0, + "valid_targets_min": 1139 }, { - "epoch": 4.970126582278481, - "grad_norm": 0.21237574487414587, - "learning_rate": 7.248437353468695e-10, + "epoch": 4.984303797468354, + "grad_norm": 0.2305284497602762, + "learning_rate": 1.998961370555108e-09, "loss": 0.1054, - "step": 1227 - }, - { - "epoch": 4.974177215189873, - "grad_norm": 0.19762341059700764, - "learning_rate": 3.221538523412449e-10, - "loss": 0.1168, - "step": 1228 + "loss_nan_ranks": 0, + "loss_rank_avg": 0.007713980041444302, + "step": 1231, + "valid_targets_mean": 3291.6, + "valid_targets_min": 930 }, { - "epoch": 4.978227848101266, - "grad_norm": 0.20565669764660643, - "learning_rate": 8.053862524670663e-11, - "loss": 0.1198, - "step": 1229 - }, - { - "epoch": 4.982278481012658, - "grad_norm": 0.20290470136362806, - "learning_rate": 0.0, - "loss": 0.1357, - "step": 1230 - }, - { - "epoch": 4.982278481012658, - "step": 1230, - "total_flos": 2.090332464318251e+18, - "train_loss": 0.24650180314977962, - "train_runtime": 35345.5491, - "train_samples_per_second": 4.47, - "train_steps_per_second": 0.035 + "epoch": 4.988354430379747, + "grad_norm": 0.22672850532808966, + "learning_rate": 1.2793429493518361e-09, + "loss": 0.1142, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.013615965843200684, + "step": 1232, + "valid_targets_mean": 5076.8, + "valid_targets_min": 1208 + }, + { + "epoch": 4.992405063291139, + "grad_norm": 0.2394791229613049, + "learning_rate": 7.196337656112384e-10, + "loss": 0.103, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.012944923713803291, + "step": 1233, + "valid_targets_mean": 3943.1, + "valid_targets_min": 1136 + }, + { + "epoch": 4.996455696202531, + "grad_norm": 0.24313900796577423, + "learning_rate": 3.198382947511292e-10, + "loss": 0.1081, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018745126202702522, + "step": 1234, + "valid_targets_mean": 5020.2, + "valid_targets_min": 1195 + }, + { + "epoch": 5.0, + "grad_norm": 0.2701508029879809, + "learning_rate": 7.995973352770137e-11, + "loss": 0.1094, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018037570640444756, + "step": 1235, + "valid_targets_mean": 4350.2, + "valid_targets_min": 878 + }, + { + "epoch": 5.0, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.018037570640444756, + "step": 1235, + "total_flos": 2.10289015818027e+18, + "train_loss": 0.2609484052790804, + "train_runtime": 28006.4544, + "train_samples_per_second": 5.642, + "train_steps_per_second": 0.044, + "valid_targets_mean": 4350.2, + "valid_targets_min": 878 } ], "logging_steps": 1, - "max_steps": 1230, + "max_steps": 1235, "num_input_tokens_seen": 0, "num_train_epochs": 5, "save_steps": 500, @@ -8645,7 +13625,7 @@ "attributes": {} } }, - "total_flos": 2.090332464318251e+18, + "total_flos": 2.10289015818027e+18, "train_batch_size": 1, "trial_name": null, "trial_params": null