diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,9386 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 7.0, + "eval_steps": 500, + "global_step": 4249, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.008237232289950576, + "grad_norm": 18.041129986855786, + "learning_rate": 3.764705882352941e-07, + "loss": 0.7519, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.8261098861694336, + "step": 5, + "valid_targets_mean": 2174.1, + "valid_targets_min": 693 + }, + { + "epoch": 0.016474464579901153, + "grad_norm": 14.34905148121762, + "learning_rate": 8.470588235294118e-07, + "loss": 0.7712, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.7180605530738831, + "step": 10, + "valid_targets_mean": 2644.4, + "valid_targets_min": 287 + }, + { + "epoch": 0.02471169686985173, + "grad_norm": 15.970319180575778, + "learning_rate": 1.3176470588235296e-06, + "loss": 0.7464, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.7657804489135742, + "step": 15, + "valid_targets_mean": 2121.8, + "valid_targets_min": 1030 + }, + { + "epoch": 0.032948929159802305, + "grad_norm": 10.573979142790474, + "learning_rate": 1.7882352941176474e-06, + "loss": 0.6642, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.5902353525161743, + "step": 20, + "valid_targets_mean": 3019.1, + "valid_targets_min": 388 + }, + { + "epoch": 0.04118616144975288, + "grad_norm": 6.5189844283078076, + "learning_rate": 2.258823529411765e-06, + "loss": 0.676, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.6557721495628357, + "step": 25, + "valid_targets_mean": 3073.6, + "valid_targets_min": 1036 + }, + { + "epoch": 0.04942339373970346, + "grad_norm": 4.427343793861997, + "learning_rate": 2.7294117647058825e-06, + "loss": 0.6411, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.6275306344032288, + "step": 30, + "valid_targets_mean": 3181.9, + "valid_targets_min": 797 + }, + { + "epoch": 0.057660626029654036, + "grad_norm": 3.0263978789805233, + "learning_rate": 3.2000000000000003e-06, + "loss": 0.5953, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.5235786437988281, + "step": 35, + "valid_targets_mean": 2863.2, + "valid_targets_min": 849 + }, + { + "epoch": 0.06589785831960461, + "grad_norm": 1.8462361669327314, + "learning_rate": 3.670588235294118e-06, + "loss": 0.5154, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.5156185030937195, + "step": 40, + "valid_targets_mean": 2872.0, + "valid_targets_min": 748 + }, + { + "epoch": 0.07413509060955518, + "grad_norm": 1.4886243049823848, + "learning_rate": 4.141176470588235e-06, + "loss": 0.4722, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4820077419281006, + "step": 45, + "valid_targets_mean": 3115.6, + "valid_targets_min": 1525 + }, + { + "epoch": 0.08237232289950576, + "grad_norm": 1.0186916863672602, + "learning_rate": 4.611764705882353e-06, + "loss": 0.4663, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.44742530584335327, + "step": 50, + "valid_targets_mean": 2791.6, + "valid_targets_min": 323 + }, + { + "epoch": 0.09060955518945635, + "grad_norm": 1.0555979955301744, + "learning_rate": 5.08235294117647e-06, + "loss": 0.5045, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.5457557439804077, + "step": 55, + "valid_targets_mean": 2666.8, + "valid_targets_min": 636 + }, + { + "epoch": 0.09884678747940692, + "grad_norm": 0.9431942413892103, + "learning_rate": 5.552941176470589e-06, + "loss": 0.4561, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.45544564723968506, + "step": 60, + "valid_targets_mean": 2670.2, + "valid_targets_min": 248 + }, + { + "epoch": 0.1070840197693575, + "grad_norm": 0.9827011262758822, + "learning_rate": 6.023529411764706e-06, + "loss": 0.5112, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4421389400959015, + "step": 65, + "valid_targets_mean": 2145.6, + "valid_targets_min": 763 + }, + { + "epoch": 0.11532125205930807, + "grad_norm": 0.864994814301313, + "learning_rate": 6.494117647058824e-06, + "loss": 0.4383, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.44878387451171875, + "step": 70, + "valid_targets_mean": 2175.3, + "valid_targets_min": 251 + }, + { + "epoch": 0.12355848434925865, + "grad_norm": 0.8853263389472636, + "learning_rate": 6.964705882352941e-06, + "loss": 0.497, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.5232568383216858, + "step": 75, + "valid_targets_mean": 2546.2, + "valid_targets_min": 265 + }, + { + "epoch": 0.13179571663920922, + "grad_norm": 0.8051192678757433, + "learning_rate": 7.43529411764706e-06, + "loss": 0.4507, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.45036208629608154, + "step": 80, + "valid_targets_mean": 2913.5, + "valid_targets_min": 735 + }, + { + "epoch": 0.1400329489291598, + "grad_norm": 0.7677204251168979, + "learning_rate": 7.905882352941176e-06, + "loss": 0.4379, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.40930110216140747, + "step": 85, + "valid_targets_mean": 2679.2, + "valid_targets_min": 768 + }, + { + "epoch": 0.14827018121911037, + "grad_norm": 0.7800094242006477, + "learning_rate": 8.376470588235295e-06, + "loss": 0.4128, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.44566720724105835, + "step": 90, + "valid_targets_mean": 3371.6, + "valid_targets_min": 1439 + }, + { + "epoch": 0.15650741350906094, + "grad_norm": 0.6767591438862864, + "learning_rate": 8.847058823529413e-06, + "loss": 0.4233, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.38728904724121094, + "step": 95, + "valid_targets_mean": 2819.8, + "valid_targets_min": 585 + }, + { + "epoch": 0.16474464579901152, + "grad_norm": 0.8131291107995611, + "learning_rate": 9.31764705882353e-06, + "loss": 0.4063, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.48519647121429443, + "step": 100, + "valid_targets_mean": 2543.0, + "valid_targets_min": 1046 + }, + { + "epoch": 0.17298187808896212, + "grad_norm": 0.516958608307116, + "learning_rate": 9.788235294117649e-06, + "loss": 0.3894, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3352322280406952, + "step": 105, + "valid_targets_mean": 3974.6, + "valid_targets_min": 1106 + }, + { + "epoch": 0.1812191103789127, + "grad_norm": 0.7099200398129607, + "learning_rate": 1.0258823529411766e-05, + "loss": 0.3724, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.34124284982681274, + "step": 110, + "valid_targets_mean": 2296.8, + "valid_targets_min": 311 + }, + { + "epoch": 0.18945634266886327, + "grad_norm": 1.3084249200454359, + "learning_rate": 1.0729411764705884e-05, + "loss": 0.3945, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3974684476852417, + "step": 115, + "valid_targets_mean": 2726.6, + "valid_targets_min": 841 + }, + { + "epoch": 0.19769357495881384, + "grad_norm": 0.7452807970704501, + "learning_rate": 1.1200000000000001e-05, + "loss": 0.4309, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.42342647910118103, + "step": 120, + "valid_targets_mean": 2717.6, + "valid_targets_min": 955 + }, + { + "epoch": 0.20593080724876442, + "grad_norm": 0.6071220337631781, + "learning_rate": 1.1670588235294118e-05, + "loss": 0.4154, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.44350916147232056, + "step": 125, + "valid_targets_mean": 3765.1, + "valid_targets_min": 287 + }, + { + "epoch": 0.214168039538715, + "grad_norm": 0.6810266168182173, + "learning_rate": 1.2141176470588237e-05, + "loss": 0.423, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3277353048324585, + "step": 130, + "valid_targets_mean": 2465.6, + "valid_targets_min": 321 + }, + { + "epoch": 0.22240527182866557, + "grad_norm": 0.6694184649708005, + "learning_rate": 1.2611764705882354e-05, + "loss": 0.4057, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4071721136569977, + "step": 135, + "valid_targets_mean": 3101.9, + "valid_targets_min": 742 + }, + { + "epoch": 0.23064250411861614, + "grad_norm": 0.6133654529819055, + "learning_rate": 1.3082352941176472e-05, + "loss": 0.4067, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32720082998275757, + "step": 140, + "valid_targets_mean": 3069.0, + "valid_targets_min": 1035 + }, + { + "epoch": 0.23887973640856672, + "grad_norm": 0.7161672984877867, + "learning_rate": 1.355294117647059e-05, + "loss": 0.3678, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4013705849647522, + "step": 145, + "valid_targets_mean": 2725.5, + "valid_targets_min": 319 + }, + { + "epoch": 0.2471169686985173, + "grad_norm": 0.6969489545726383, + "learning_rate": 1.4023529411764706e-05, + "loss": 0.3994, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.38001126050949097, + "step": 150, + "valid_targets_mean": 3276.0, + "valid_targets_min": 1031 + }, + { + "epoch": 0.2553542009884679, + "grad_norm": 0.6463131347430592, + "learning_rate": 1.4494117647058825e-05, + "loss": 0.3694, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4384726583957672, + "step": 155, + "valid_targets_mean": 3591.8, + "valid_targets_min": 912 + }, + { + "epoch": 0.26359143327841844, + "grad_norm": 0.6590571572375514, + "learning_rate": 1.4964705882352943e-05, + "loss": 0.3424, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31954580545425415, + "step": 160, + "valid_targets_mean": 2965.2, + "valid_targets_min": 839 + }, + { + "epoch": 0.27182866556836904, + "grad_norm": 0.6582067773489519, + "learning_rate": 1.543529411764706e-05, + "loss": 0.4073, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3766615390777588, + "step": 165, + "valid_targets_mean": 3487.7, + "valid_targets_min": 869 + }, + { + "epoch": 0.2800658978583196, + "grad_norm": 0.6816959560863262, + "learning_rate": 1.5905882352941177e-05, + "loss": 0.3678, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.38047826290130615, + "step": 170, + "valid_targets_mean": 2549.7, + "valid_targets_min": 821 + }, + { + "epoch": 0.2883031301482702, + "grad_norm": 0.7251450154291886, + "learning_rate": 1.6376470588235298e-05, + "loss": 0.3865, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4638799726963043, + "step": 175, + "valid_targets_mean": 2832.6, + "valid_targets_min": 382 + }, + { + "epoch": 0.29654036243822074, + "grad_norm": 0.6299448981935337, + "learning_rate": 1.6847058823529414e-05, + "loss": 0.3631, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3318817913532257, + "step": 180, + "valid_targets_mean": 2848.6, + "valid_targets_min": 944 + }, + { + "epoch": 0.30477759472817134, + "grad_norm": 0.6977606582667799, + "learning_rate": 1.731764705882353e-05, + "loss": 0.3772, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3715594410896301, + "step": 185, + "valid_targets_mean": 2431.6, + "valid_targets_min": 391 + }, + { + "epoch": 0.3130148270181219, + "grad_norm": 0.7061245411260707, + "learning_rate": 1.778823529411765e-05, + "loss": 0.3431, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.33297649025917053, + "step": 190, + "valid_targets_mean": 2197.2, + "valid_targets_min": 685 + }, + { + "epoch": 0.3212520593080725, + "grad_norm": 0.6859289408630844, + "learning_rate": 1.8258823529411765e-05, + "loss": 0.3479, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3794998824596405, + "step": 195, + "valid_targets_mean": 2878.8, + "valid_targets_min": 304 + }, + { + "epoch": 0.32948929159802304, + "grad_norm": 0.6130651983492513, + "learning_rate": 1.8729411764705886e-05, + "loss": 0.3828, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.38770943880081177, + "step": 200, + "valid_targets_mean": 3607.1, + "valid_targets_min": 230 + }, + { + "epoch": 0.33772652388797364, + "grad_norm": 0.7234411187513314, + "learning_rate": 1.9200000000000003e-05, + "loss": 0.3789, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4269868731498718, + "step": 205, + "valid_targets_mean": 2572.0, + "valid_targets_min": 246 + }, + { + "epoch": 0.34596375617792424, + "grad_norm": 0.5735815243504443, + "learning_rate": 1.967058823529412e-05, + "loss": 0.3501, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31644099950790405, + "step": 210, + "valid_targets_mean": 3513.8, + "valid_targets_min": 876 + }, + { + "epoch": 0.3542009884678748, + "grad_norm": 0.8651887451217208, + "learning_rate": 2.0141176470588236e-05, + "loss": 0.3411, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3497444987297058, + "step": 215, + "valid_targets_mean": 1987.6, + "valid_targets_min": 297 + }, + { + "epoch": 0.3624382207578254, + "grad_norm": 0.6522031387177083, + "learning_rate": 2.0611764705882353e-05, + "loss": 0.3654, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.38645434379577637, + "step": 220, + "valid_targets_mean": 3107.1, + "valid_targets_min": 1099 + }, + { + "epoch": 0.37067545304777594, + "grad_norm": 0.7920489560748225, + "learning_rate": 2.1082352941176474e-05, + "loss": 0.3672, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.36695098876953125, + "step": 225, + "valid_targets_mean": 1991.4, + "valid_targets_min": 295 + }, + { + "epoch": 0.37891268533772654, + "grad_norm": 0.6409036862164285, + "learning_rate": 2.155294117647059e-05, + "loss": 0.3494, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2995736598968506, + "step": 230, + "valid_targets_mean": 2558.9, + "valid_targets_min": 1074 + }, + { + "epoch": 0.3871499176276771, + "grad_norm": 0.6032322601598289, + "learning_rate": 2.2023529411764707e-05, + "loss": 0.3675, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.39067399501800537, + "step": 235, + "valid_targets_mean": 3427.1, + "valid_targets_min": 1229 + }, + { + "epoch": 0.3953871499176277, + "grad_norm": 0.640108850824614, + "learning_rate": 2.2494117647058828e-05, + "loss": 0.3517, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3539199233055115, + "step": 240, + "valid_targets_mean": 3306.6, + "valid_targets_min": 219 + }, + { + "epoch": 0.40362438220757824, + "grad_norm": 0.6923591104747878, + "learning_rate": 2.296470588235294e-05, + "loss": 0.3339, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25825247168540955, + "step": 245, + "valid_targets_mean": 2549.9, + "valid_targets_min": 271 + }, + { + "epoch": 0.41186161449752884, + "grad_norm": 0.8803432349351571, + "learning_rate": 2.343529411764706e-05, + "loss": 0.346, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.40536123514175415, + "step": 250, + "valid_targets_mean": 1919.9, + "valid_targets_min": 226 + }, + { + "epoch": 0.4200988467874794, + "grad_norm": 0.641574096444431, + "learning_rate": 2.390588235294118e-05, + "loss": 0.3585, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.35488161444664, + "step": 255, + "valid_targets_mean": 3300.2, + "valid_targets_min": 568 + }, + { + "epoch": 0.42833607907743, + "grad_norm": 0.6648054089698943, + "learning_rate": 2.4376470588235296e-05, + "loss": 0.3572, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3950795531272888, + "step": 260, + "valid_targets_mean": 3831.9, + "valid_targets_min": 563 + }, + { + "epoch": 0.43657331136738053, + "grad_norm": 0.7333690453284659, + "learning_rate": 2.4847058823529416e-05, + "loss": 0.316, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32087090611457825, + "step": 265, + "valid_targets_mean": 2422.1, + "valid_targets_min": 334 + }, + { + "epoch": 0.44481054365733114, + "grad_norm": 1.0576342544488297, + "learning_rate": 2.5317647058823533e-05, + "loss": 0.356, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4411831200122833, + "step": 270, + "valid_targets_mean": 3020.0, + "valid_targets_min": 1122 + }, + { + "epoch": 0.45304777594728174, + "grad_norm": 0.5404063898551336, + "learning_rate": 2.5788235294117646e-05, + "loss": 0.349, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2559725344181061, + "step": 275, + "valid_targets_mean": 3452.1, + "valid_targets_min": 287 + }, + { + "epoch": 0.4612850082372323, + "grad_norm": 0.6680784878314109, + "learning_rate": 2.6258823529411767e-05, + "loss": 0.3369, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3715960383415222, + "step": 280, + "valid_targets_mean": 3149.2, + "valid_targets_min": 1060 + }, + { + "epoch": 0.4695222405271829, + "grad_norm": 0.6163910100110204, + "learning_rate": 2.6729411764705884e-05, + "loss": 0.3221, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3177059292793274, + "step": 285, + "valid_targets_mean": 3021.4, + "valid_targets_min": 857 + }, + { + "epoch": 0.47775947281713343, + "grad_norm": 0.8667382342453931, + "learning_rate": 2.7200000000000004e-05, + "loss": 0.3793, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3662949502468109, + "step": 290, + "valid_targets_mean": 2007.4, + "valid_targets_min": 699 + }, + { + "epoch": 0.48599670510708404, + "grad_norm": 0.7018849627415584, + "learning_rate": 2.767058823529412e-05, + "loss": 0.3619, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3931698501110077, + "step": 295, + "valid_targets_mean": 2677.9, + "valid_targets_min": 968 + }, + { + "epoch": 0.4942339373970346, + "grad_norm": 0.6995387214546374, + "learning_rate": 2.8141176470588238e-05, + "loss": 0.3566, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3696545362472534, + "step": 300, + "valid_targets_mean": 2607.9, + "valid_targets_min": 270 + }, + { + "epoch": 0.5024711696869851, + "grad_norm": 0.6715688200029435, + "learning_rate": 2.8611764705882358e-05, + "loss": 0.3478, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27885138988494873, + "step": 305, + "valid_targets_mean": 2609.4, + "valid_targets_min": 785 + }, + { + "epoch": 0.5107084019769358, + "grad_norm": 0.7737861344674672, + "learning_rate": 2.908235294117647e-05, + "loss": 0.3742, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4706699550151825, + "step": 310, + "valid_targets_mean": 2705.9, + "valid_targets_min": 162 + }, + { + "epoch": 0.5189456342668863, + "grad_norm": 0.6334586406962706, + "learning_rate": 2.955294117647059e-05, + "loss": 0.3376, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3143928647041321, + "step": 315, + "valid_targets_mean": 3639.9, + "valid_targets_min": 1132 + }, + { + "epoch": 0.5271828665568369, + "grad_norm": 0.7026900238535507, + "learning_rate": 3.002352941176471e-05, + "loss": 0.3393, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3668755292892456, + "step": 320, + "valid_targets_mean": 3119.6, + "valid_targets_min": 896 + }, + { + "epoch": 0.5354200988467874, + "grad_norm": 0.70858594506907, + "learning_rate": 3.0494117647058826e-05, + "loss": 0.3343, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.42341750860214233, + "step": 325, + "valid_targets_mean": 2856.9, + "valid_targets_min": 716 + }, + { + "epoch": 0.5436573311367381, + "grad_norm": 0.9050380182770631, + "learning_rate": 3.0964705882352946e-05, + "loss": 0.3229, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28868210315704346, + "step": 330, + "valid_targets_mean": 2783.8, + "valid_targets_min": 1080 + }, + { + "epoch": 0.5518945634266886, + "grad_norm": 0.6793783253890467, + "learning_rate": 3.143529411764706e-05, + "loss": 0.358, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3081079125404358, + "step": 335, + "valid_targets_mean": 2909.4, + "valid_targets_min": 905 + }, + { + "epoch": 0.5601317957166392, + "grad_norm": 0.6531063772093401, + "learning_rate": 3.190588235294118e-05, + "loss": 0.3301, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3191760182380676, + "step": 340, + "valid_targets_mean": 2809.9, + "valid_targets_min": 829 + }, + { + "epoch": 0.5683690280065898, + "grad_norm": 0.8336208889400938, + "learning_rate": 3.23764705882353e-05, + "loss": 0.3222, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3880823254585266, + "step": 345, + "valid_targets_mean": 2561.3, + "valid_targets_min": 626 + }, + { + "epoch": 0.5766062602965404, + "grad_norm": 1.1368953862481592, + "learning_rate": 3.2847058823529414e-05, + "loss": 0.3541, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27717214822769165, + "step": 350, + "valid_targets_mean": 2409.9, + "valid_targets_min": 831 + }, + { + "epoch": 0.5848434925864909, + "grad_norm": 0.7026717555920957, + "learning_rate": 3.331764705882353e-05, + "loss": 0.3622, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3906007409095764, + "step": 355, + "valid_targets_mean": 2896.1, + "valid_targets_min": 385 + }, + { + "epoch": 0.5930807248764415, + "grad_norm": 0.6023606367373995, + "learning_rate": 3.378823529411765e-05, + "loss": 0.3704, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.37305983901023865, + "step": 360, + "valid_targets_mean": 3621.6, + "valid_targets_min": 1150 + }, + { + "epoch": 0.6013179571663921, + "grad_norm": 0.6514047229400881, + "learning_rate": 3.425882352941177e-05, + "loss": 0.3581, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29843461513519287, + "step": 365, + "valid_targets_mean": 3057.4, + "valid_targets_min": 1271 + }, + { + "epoch": 0.6095551894563427, + "grad_norm": 0.5626045279089444, + "learning_rate": 3.472941176470589e-05, + "loss": 0.3145, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27968597412109375, + "step": 370, + "valid_targets_mean": 3222.4, + "valid_targets_min": 374 + }, + { + "epoch": 0.6177924217462932, + "grad_norm": 0.5421510206663698, + "learning_rate": 3.52e-05, + "loss": 0.3319, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32853734493255615, + "step": 375, + "valid_targets_mean": 3641.3, + "valid_targets_min": 569 + }, + { + "epoch": 0.6260296540362438, + "grad_norm": 0.6144302274176291, + "learning_rate": 3.567058823529412e-05, + "loss": 0.3285, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.33135101199150085, + "step": 380, + "valid_targets_mean": 3392.9, + "valid_targets_min": 1176 + }, + { + "epoch": 0.6342668863261944, + "grad_norm": 0.5893086168232592, + "learning_rate": 3.614117647058824e-05, + "loss": 0.3337, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32672643661499023, + "step": 385, + "valid_targets_mean": 2965.9, + "valid_targets_min": 654 + }, + { + "epoch": 0.642504118616145, + "grad_norm": 0.6559702753858812, + "learning_rate": 3.6611764705882356e-05, + "loss": 0.3469, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30627205967903137, + "step": 390, + "valid_targets_mean": 2923.8, + "valid_targets_min": 737 + }, + { + "epoch": 0.6507413509060955, + "grad_norm": 0.7743858147760204, + "learning_rate": 3.708235294117647e-05, + "loss": 0.359, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4699430465698242, + "step": 395, + "valid_targets_mean": 2655.4, + "valid_targets_min": 857 + }, + { + "epoch": 0.6589785831960461, + "grad_norm": 0.6333244399993251, + "learning_rate": 3.755294117647059e-05, + "loss": 0.3314, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.38278141617774963, + "step": 400, + "valid_targets_mean": 2962.8, + "valid_targets_min": 256 + }, + { + "epoch": 0.6672158154859967, + "grad_norm": 0.6290385348921544, + "learning_rate": 3.802352941176471e-05, + "loss": 0.3252, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3128456175327301, + "step": 405, + "valid_targets_mean": 2868.9, + "valid_targets_min": 814 + }, + { + "epoch": 0.6754530477759473, + "grad_norm": 0.9556660046777031, + "learning_rate": 3.849411764705883e-05, + "loss": 0.3047, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23495997488498688, + "step": 410, + "valid_targets_mean": 3069.1, + "valid_targets_min": 719 + }, + { + "epoch": 0.6836902800658978, + "grad_norm": 0.6229890597206825, + "learning_rate": 3.8964705882352944e-05, + "loss": 0.385, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3396230638027191, + "step": 415, + "valid_targets_mean": 2948.0, + "valid_targets_min": 910 + }, + { + "epoch": 0.6919275123558485, + "grad_norm": 0.7604982982401985, + "learning_rate": 3.943529411764706e-05, + "loss": 0.3361, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.36150163412094116, + "step": 420, + "valid_targets_mean": 2440.2, + "valid_targets_min": 279 + }, + { + "epoch": 0.700164744645799, + "grad_norm": 0.6483929715914343, + "learning_rate": 3.990588235294118e-05, + "loss": 0.3206, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31857752799987793, + "step": 425, + "valid_targets_mean": 2489.6, + "valid_targets_min": 250 + }, + { + "epoch": 0.7084019769357496, + "grad_norm": 0.6077661449463507, + "learning_rate": 3.9999892009992835e-05, + "loss": 0.3347, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.35132089257240295, + "step": 430, + "valid_targets_mean": 3160.6, + "valid_targets_min": 711 + }, + { + "epoch": 0.7166392092257001, + "grad_norm": 0.6568155761335552, + "learning_rate": 3.999945330258739e-05, + "loss": 0.3485, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3767227828502655, + "step": 435, + "valid_targets_mean": 2791.2, + "valid_targets_min": 378 + }, + { + "epoch": 0.7248764415156508, + "grad_norm": 0.6549147402066008, + "learning_rate": 3.9998677135805034e-05, + "loss": 0.3356, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3399823307991028, + "step": 440, + "valid_targets_mean": 2741.8, + "valid_targets_min": 237 + }, + { + "epoch": 0.7331136738056013, + "grad_norm": 0.6266647101324666, + "learning_rate": 3.999756352274235e-05, + "loss": 0.3627, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2891290783882141, + "step": 445, + "valid_targets_mean": 2656.9, + "valid_targets_min": 766 + }, + { + "epoch": 0.7413509060955519, + "grad_norm": 0.7173443868039945, + "learning_rate": 3.9996112482189823e-05, + "loss": 0.3428, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.35799193382263184, + "step": 450, + "valid_targets_mean": 2232.7, + "valid_targets_min": 281 + }, + { + "epoch": 0.7495881383855024, + "grad_norm": 0.7475444722849752, + "learning_rate": 3.9994324038631475e-05, + "loss": 0.2933, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31564396619796753, + "step": 455, + "valid_targets_mean": 2473.4, + "valid_targets_min": 425 + }, + { + "epoch": 0.7578253706754531, + "grad_norm": 0.7565191294316428, + "learning_rate": 3.999219822224449e-05, + "loss": 0.3607, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3808445930480957, + "step": 460, + "valid_targets_mean": 2174.1, + "valid_targets_min": 260 + }, + { + "epoch": 0.7660626029654036, + "grad_norm": 0.6005283358301972, + "learning_rate": 3.998973506889867e-05, + "loss": 0.3102, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31527596712112427, + "step": 465, + "valid_targets_mean": 2908.4, + "valid_targets_min": 714 + }, + { + "epoch": 0.7742998352553542, + "grad_norm": 0.6631763471679214, + "learning_rate": 3.998693462015588e-05, + "loss": 0.3188, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2902739346027374, + "step": 470, + "valid_targets_mean": 2522.3, + "valid_targets_min": 440 + }, + { + "epoch": 0.7825370675453048, + "grad_norm": 0.7214500863472263, + "learning_rate": 3.9983796923269284e-05, + "loss": 0.3284, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3250043988227844, + "step": 475, + "valid_targets_mean": 2207.5, + "valid_targets_min": 629 + }, + { + "epoch": 0.7907742998352554, + "grad_norm": 0.6204766994564088, + "learning_rate": 3.99803220311826e-05, + "loss": 0.3399, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3031291365623474, + "step": 480, + "valid_targets_mean": 2860.6, + "valid_targets_min": 647 + }, + { + "epoch": 0.7990115321252059, + "grad_norm": 0.6251122638878225, + "learning_rate": 3.997651000252916e-05, + "loss": 0.3509, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3469051718711853, + "step": 485, + "valid_targets_mean": 3213.5, + "valid_targets_min": 317 + }, + { + "epoch": 0.8072487644151565, + "grad_norm": 0.6183054272635903, + "learning_rate": 3.997236090163097e-05, + "loss": 0.3618, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.36611422896385193, + "step": 490, + "valid_targets_mean": 3578.0, + "valid_targets_min": 1081 + }, + { + "epoch": 0.8154859967051071, + "grad_norm": 0.7643699670266144, + "learning_rate": 3.99678747984976e-05, + "loss": 0.3356, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4362257122993469, + "step": 495, + "valid_targets_mean": 2453.6, + "valid_targets_min": 542 + }, + { + "epoch": 0.8237232289950577, + "grad_norm": 0.8341652264979282, + "learning_rate": 3.9963051768824996e-05, + "loss": 0.2872, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26549232006073, + "step": 500, + "valid_targets_mean": 2642.1, + "valid_targets_min": 1008 + }, + { + "epoch": 0.8319604612850082, + "grad_norm": 0.6792332627863412, + "learning_rate": 3.995789189399421e-05, + "loss": 0.3312, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27049073576927185, + "step": 505, + "valid_targets_mean": 2643.8, + "valid_targets_min": 391 + }, + { + "epoch": 0.8401976935749588, + "grad_norm": 0.6315353366191043, + "learning_rate": 3.995239526107003e-05, + "loss": 0.3141, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29464757442474365, + "step": 510, + "valid_targets_mean": 2849.6, + "valid_targets_min": 330 + }, + { + "epoch": 0.8484349258649094, + "grad_norm": 0.7540062474541283, + "learning_rate": 3.99465619627995e-05, + "loss": 0.3232, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3380299508571625, + "step": 515, + "valid_targets_mean": 2119.2, + "valid_targets_min": 567 + }, + { + "epoch": 0.85667215815486, + "grad_norm": 0.613809876372666, + "learning_rate": 3.994039209761038e-05, + "loss": 0.3175, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2560957074165344, + "step": 520, + "valid_targets_mean": 3331.9, + "valid_targets_min": 748 + }, + { + "epoch": 0.8649093904448105, + "grad_norm": 0.6123292685393475, + "learning_rate": 3.993388576960947e-05, + "loss": 0.3065, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.33387625217437744, + "step": 525, + "valid_targets_mean": 3419.0, + "valid_targets_min": 326 + }, + { + "epoch": 0.8731466227347611, + "grad_norm": 0.5538651672396367, + "learning_rate": 3.9927043088580805e-05, + "loss": 0.3121, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.37178099155426025, + "step": 530, + "valid_targets_mean": 4541.1, + "valid_targets_min": 701 + }, + { + "epoch": 0.8813838550247117, + "grad_norm": 0.5909393191732646, + "learning_rate": 3.991986416998391e-05, + "loss": 0.3159, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3093166947364807, + "step": 535, + "valid_targets_mean": 2899.0, + "valid_targets_min": 397 + }, + { + "epoch": 0.8896210873146623, + "grad_norm": 0.6546363609842427, + "learning_rate": 3.991234913495177e-05, + "loss": 0.3403, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.296301007270813, + "step": 540, + "valid_targets_mean": 2484.8, + "valid_targets_min": 1012 + }, + { + "epoch": 0.8978583196046128, + "grad_norm": 0.6157315771578695, + "learning_rate": 3.99044981102888e-05, + "loss": 0.3454, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3564984202384949, + "step": 545, + "valid_targets_mean": 3432.4, + "valid_targets_min": 832 + }, + { + "epoch": 0.9060955518945635, + "grad_norm": 0.5853646079194815, + "learning_rate": 3.98963112284687e-05, + "loss": 0.2987, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27953511476516724, + "step": 550, + "valid_targets_mean": 2949.4, + "valid_targets_min": 677 + }, + { + "epoch": 0.914332784184514, + "grad_norm": 0.6929994545262912, + "learning_rate": 3.9887788627632266e-05, + "loss": 0.3304, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3079501986503601, + "step": 555, + "valid_targets_mean": 2296.4, + "valid_targets_min": 242 + }, + { + "epoch": 0.9225700164744646, + "grad_norm": 0.6175410466216474, + "learning_rate": 3.9878930451584995e-05, + "loss": 0.3331, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3442390561103821, + "step": 560, + "valid_targets_mean": 3016.9, + "valid_targets_min": 223 + }, + { + "epoch": 0.9308072487644151, + "grad_norm": 0.5759924285268283, + "learning_rate": 3.986973684979471e-05, + "loss": 0.339, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3262675404548645, + "step": 565, + "valid_targets_mean": 3156.9, + "valid_targets_min": 223 + }, + { + "epoch": 0.9390444810543658, + "grad_norm": 0.6354650452462268, + "learning_rate": 3.9860207977389e-05, + "loss": 0.3647, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4005314111709595, + "step": 570, + "valid_targets_mean": 3261.8, + "valid_targets_min": 718 + }, + { + "epoch": 0.9472817133443163, + "grad_norm": 0.816503500430313, + "learning_rate": 3.9850343995152634e-05, + "loss": 0.3038, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29235273599624634, + "step": 575, + "valid_targets_mean": 2347.9, + "valid_targets_min": 307 + }, + { + "epoch": 0.9555189456342669, + "grad_norm": 0.6708495614164758, + "learning_rate": 3.9840145069524825e-05, + "loss": 0.3157, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32615166902542114, + "step": 580, + "valid_targets_mean": 2426.5, + "valid_targets_min": 1424 + }, + { + "epoch": 0.9637561779242174, + "grad_norm": 0.6448102084828288, + "learning_rate": 3.982961137259643e-05, + "loss": 0.3464, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3473745882511139, + "step": 585, + "valid_targets_mean": 2862.9, + "valid_targets_min": 382 + }, + { + "epoch": 0.9719934102141681, + "grad_norm": 0.6252647889505737, + "learning_rate": 3.981874308210705e-05, + "loss": 0.3189, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31804078817367554, + "step": 590, + "valid_targets_mean": 2728.4, + "valid_targets_min": 647 + }, + { + "epoch": 0.9802306425041186, + "grad_norm": 0.5871449715572906, + "learning_rate": 3.980754038144202e-05, + "loss": 0.3016, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31086063385009766, + "step": 595, + "valid_targets_mean": 3214.8, + "valid_targets_min": 251 + }, + { + "epoch": 0.9884678747940692, + "grad_norm": 0.6461194424571096, + "learning_rate": 3.9796003459629335e-05, + "loss": 0.3205, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.39187324047088623, + "step": 600, + "valid_targets_mean": 2897.5, + "valid_targets_min": 702 + }, + { + "epoch": 0.9967051070840197, + "grad_norm": 0.6731453411587777, + "learning_rate": 3.978413251133643e-05, + "loss": 0.3463, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.304431676864624, + "step": 605, + "valid_targets_mean": 2816.6, + "valid_targets_min": 1209 + }, + { + "epoch": 1.0049423393739703, + "grad_norm": 0.6830394009205784, + "learning_rate": 3.9771927736866926e-05, + "loss": 0.2897, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29772090911865234, + "step": 610, + "valid_targets_mean": 2312.0, + "valid_targets_min": 847 + }, + { + "epoch": 1.013179571663921, + "grad_norm": 0.5982189648576913, + "learning_rate": 3.9759389342157225e-05, + "loss": 0.2912, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2953491508960724, + "step": 615, + "valid_targets_mean": 3731.6, + "valid_targets_min": 725 + }, + { + "epoch": 1.0214168039538716, + "grad_norm": 0.6464219651312553, + "learning_rate": 3.974651753877306e-05, + "loss": 0.339, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32838666439056396, + "step": 620, + "valid_targets_mean": 2931.8, + "valid_targets_min": 212 + }, + { + "epoch": 1.029654036243822, + "grad_norm": 0.6819518166567332, + "learning_rate": 3.9733312543905904e-05, + "loss": 0.2964, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28917109966278076, + "step": 625, + "valid_targets_mean": 2779.7, + "valid_targets_min": 614 + }, + { + "epoch": 1.0378912685337727, + "grad_norm": 0.6274036687821978, + "learning_rate": 3.971977458036931e-05, + "loss": 0.3108, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2930309772491455, + "step": 630, + "valid_targets_mean": 2777.7, + "valid_targets_min": 293 + }, + { + "epoch": 1.0461285008237233, + "grad_norm": 0.617268853064774, + "learning_rate": 3.9705903876595167e-05, + "loss": 0.333, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.34295734763145447, + "step": 635, + "valid_targets_mean": 3084.1, + "valid_targets_min": 938 + }, + { + "epoch": 1.0543657331136738, + "grad_norm": 0.6419145579444464, + "learning_rate": 3.969170066662982e-05, + "loss": 0.2914, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29465362429618835, + "step": 640, + "valid_targets_mean": 2784.9, + "valid_targets_min": 1046 + }, + { + "epoch": 1.0626029654036244, + "grad_norm": 0.5589047587447845, + "learning_rate": 3.967716519013014e-05, + "loss": 0.286, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.33022594451904297, + "step": 645, + "valid_targets_mean": 4427.2, + "valid_targets_min": 937 + }, + { + "epoch": 1.0708401976935749, + "grad_norm": 0.821250839806891, + "learning_rate": 3.9662297692359486e-05, + "loss": 0.3201, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3356301188468933, + "step": 650, + "valid_targets_mean": 3180.2, + "valid_targets_min": 886 + }, + { + "epoch": 1.0790774299835255, + "grad_norm": 0.5563507731732716, + "learning_rate": 3.964709842418351e-05, + "loss": 0.28, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2440907061100006, + "step": 655, + "valid_targets_mean": 3412.5, + "valid_targets_min": 780 + }, + { + "epoch": 1.0873146622734762, + "grad_norm": 0.6207510688705554, + "learning_rate": 3.9631567642066036e-05, + "loss": 0.3216, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30460572242736816, + "step": 660, + "valid_targets_mean": 3038.8, + "valid_targets_min": 322 + }, + { + "epoch": 1.0955518945634266, + "grad_norm": 0.7017906174899363, + "learning_rate": 3.961570560806461e-05, + "loss": 0.2867, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3034290075302124, + "step": 665, + "valid_targets_mean": 2448.2, + "valid_targets_min": 244 + }, + { + "epoch": 1.1037891268533773, + "grad_norm": 0.6601151517121949, + "learning_rate": 3.959951258982616e-05, + "loss": 0.3054, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31534379720687866, + "step": 670, + "valid_targets_mean": 2649.9, + "valid_targets_min": 790 + }, + { + "epoch": 1.112026359143328, + "grad_norm": 0.6564702714451534, + "learning_rate": 3.958298886058244e-05, + "loss": 0.2838, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2642975151538849, + "step": 675, + "valid_targets_mean": 2336.2, + "valid_targets_min": 915 + }, + { + "epoch": 1.1202635914332784, + "grad_norm": 0.7154383695421775, + "learning_rate": 3.956613469914546e-05, + "loss": 0.2983, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3255589008331299, + "step": 680, + "valid_targets_mean": 2366.6, + "valid_targets_min": 862 + }, + { + "epoch": 1.128500823723229, + "grad_norm": 0.8969246671748706, + "learning_rate": 3.95489503899027e-05, + "loss": 0.2985, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32357919216156006, + "step": 685, + "valid_targets_mean": 2685.2, + "valid_targets_min": 202 + }, + { + "epoch": 1.1367380560131797, + "grad_norm": 0.6519391757431306, + "learning_rate": 3.9531436222812434e-05, + "loss": 0.2979, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2531551718711853, + "step": 690, + "valid_targets_mean": 2837.1, + "valid_targets_min": 750 + }, + { + "epoch": 1.1449752883031301, + "grad_norm": 0.6259788308504319, + "learning_rate": 3.951359249339872e-05, + "loss": 0.2951, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27987170219421387, + "step": 695, + "valid_targets_mean": 2745.1, + "valid_targets_min": 866 + }, + { + "epoch": 1.1532125205930808, + "grad_norm": 0.6978771465302991, + "learning_rate": 3.9495419502746485e-05, + "loss": 0.2888, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3704025149345398, + "step": 700, + "valid_targets_mean": 2759.8, + "valid_targets_min": 846 + }, + { + "epoch": 1.1614497528830312, + "grad_norm": 0.5939150143068824, + "learning_rate": 3.9476917557496406e-05, + "loss": 0.3089, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28111281991004944, + "step": 705, + "valid_targets_mean": 3140.9, + "valid_targets_min": 831 + }, + { + "epoch": 1.1696869851729819, + "grad_norm": 0.6245181056230996, + "learning_rate": 3.945808696983977e-05, + "loss": 0.2915, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27186596393585205, + "step": 710, + "valid_targets_mean": 3023.4, + "valid_targets_min": 1342 + }, + { + "epoch": 1.1779242174629325, + "grad_norm": 0.6803023490819639, + "learning_rate": 3.943892805751318e-05, + "loss": 0.2826, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31058281660079956, + "step": 715, + "valid_targets_mean": 2626.4, + "valid_targets_min": 765 + }, + { + "epoch": 1.186161449752883, + "grad_norm": 0.6833347407466104, + "learning_rate": 3.941944114379321e-05, + "loss": 0.2898, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29879602789878845, + "step": 720, + "valid_targets_mean": 2617.1, + "valid_targets_min": 438 + }, + { + "epoch": 1.1943986820428336, + "grad_norm": 0.7184104527406685, + "learning_rate": 3.939962655749092e-05, + "loss": 0.3218, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28567400574684143, + "step": 725, + "valid_targets_mean": 3099.2, + "valid_targets_min": 817 + }, + { + "epoch": 1.2026359143327843, + "grad_norm": 0.5458746613899529, + "learning_rate": 3.937948463294636e-05, + "loss": 0.2879, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22731509804725647, + "step": 730, + "valid_targets_mean": 3388.2, + "valid_targets_min": 297 + }, + { + "epoch": 1.2108731466227347, + "grad_norm": 0.54504544091912, + "learning_rate": 3.935901571002288e-05, + "loss": 0.2819, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.33203086256980896, + "step": 735, + "valid_targets_mean": 4777.1, + "valid_targets_min": 1301 + }, + { + "epoch": 1.2191103789126854, + "grad_norm": 0.6106385198867497, + "learning_rate": 3.9338220134101426e-05, + "loss": 0.2879, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3434884548187256, + "step": 740, + "valid_targets_mean": 3176.8, + "valid_targets_min": 789 + }, + { + "epoch": 1.227347611202636, + "grad_norm": 0.632041382439529, + "learning_rate": 3.931709825607469e-05, + "loss": 0.3115, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28263598680496216, + "step": 745, + "valid_targets_mean": 2709.9, + "valid_targets_min": 749 + }, + { + "epoch": 1.2355848434925865, + "grad_norm": 0.6654709061054238, + "learning_rate": 3.9295650432341195e-05, + "loss": 0.3036, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2941563129425049, + "step": 750, + "valid_targets_mean": 2752.9, + "valid_targets_min": 569 + }, + { + "epoch": 1.2438220757825371, + "grad_norm": 0.5691752162475886, + "learning_rate": 3.927387702479931e-05, + "loss": 0.2433, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27349454164505005, + "step": 755, + "valid_targets_mean": 3335.9, + "valid_targets_min": 698 + }, + { + "epoch": 1.2520593080724876, + "grad_norm": 0.639705746497503, + "learning_rate": 3.925177840084109e-05, + "loss": 0.2977, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2904908061027527, + "step": 760, + "valid_targets_mean": 2948.4, + "valid_targets_min": 1389 + }, + { + "epoch": 1.2602965403624382, + "grad_norm": 0.6452781995342208, + "learning_rate": 3.922935493334614e-05, + "loss": 0.2733, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24950724840164185, + "step": 765, + "valid_targets_mean": 2633.5, + "valid_targets_min": 245 + }, + { + "epoch": 1.2685337726523889, + "grad_norm": 0.6556539492801983, + "learning_rate": 3.920660700067526e-05, + "loss": 0.2996, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3787687122821808, + "step": 770, + "valid_targets_mean": 3176.0, + "valid_targets_min": 1059 + }, + { + "epoch": 1.2767710049423393, + "grad_norm": 0.6614063472129723, + "learning_rate": 3.9183534986664094e-05, + "loss": 0.2788, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30761298537254333, + "step": 775, + "valid_targets_mean": 2789.6, + "valid_targets_min": 607 + }, + { + "epoch": 1.28500823723229, + "grad_norm": 0.5321065467089631, + "learning_rate": 3.9160139280616674e-05, + "loss": 0.3002, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3060571551322937, + "step": 780, + "valid_targets_mean": 4055.0, + "valid_targets_min": 1195 + }, + { + "epoch": 1.2932454695222404, + "grad_norm": 0.6354469805842081, + "learning_rate": 3.9136420277298795e-05, + "loss": 0.3141, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.268049955368042, + "step": 785, + "valid_targets_mean": 2501.8, + "valid_targets_min": 1260 + }, + { + "epoch": 1.301482701812191, + "grad_norm": 0.743079520777049, + "learning_rate": 3.911237837693141e-05, + "loss": 0.296, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2993124723434448, + "step": 790, + "valid_targets_mean": 2397.9, + "valid_targets_min": 702 + }, + { + "epoch": 1.3097199341021417, + "grad_norm": 0.6922482501855028, + "learning_rate": 3.9088013985183836e-05, + "loss": 0.3102, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3406824767589569, + "step": 795, + "valid_targets_mean": 2366.7, + "valid_targets_min": 334 + }, + { + "epoch": 1.3179571663920924, + "grad_norm": 0.6649210661228457, + "learning_rate": 3.9063327513166945e-05, + "loss": 0.2926, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3096727430820465, + "step": 800, + "valid_targets_mean": 2671.3, + "valid_targets_min": 331 + }, + { + "epoch": 1.3261943986820428, + "grad_norm": 0.7476716763557605, + "learning_rate": 3.9038319377426186e-05, + "loss": 0.3033, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2687912583351135, + "step": 805, + "valid_targets_mean": 2750.8, + "valid_targets_min": 1079 + }, + { + "epoch": 1.3344316309719935, + "grad_norm": 0.7202214301792093, + "learning_rate": 3.9012989999934595e-05, + "loss": 0.2848, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25632429122924805, + "step": 810, + "valid_targets_mean": 2071.9, + "valid_targets_min": 336 + }, + { + "epoch": 1.342668863261944, + "grad_norm": 0.7285006052829752, + "learning_rate": 3.898733980808565e-05, + "loss": 0.2714, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2646828293800354, + "step": 815, + "valid_targets_mean": 1977.5, + "valid_targets_min": 685 + }, + { + "epoch": 1.3509060955518946, + "grad_norm": 0.5637166344318452, + "learning_rate": 3.896136923468607e-05, + "loss": 0.2982, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26190030574798584, + "step": 820, + "valid_targets_mean": 3011.3, + "valid_targets_min": 321 + }, + { + "epoch": 1.3591433278418452, + "grad_norm": 0.6680895688573001, + "learning_rate": 3.8935078717948507e-05, + "loss": 0.3126, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.34075266122817993, + "step": 825, + "valid_targets_mean": 2643.4, + "valid_targets_min": 685 + }, + { + "epoch": 1.3673805601317957, + "grad_norm": 0.6343514611286315, + "learning_rate": 3.890846870148416e-05, + "loss": 0.2973, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2886373698711395, + "step": 830, + "valid_targets_mean": 2637.9, + "valid_targets_min": 915 + }, + { + "epoch": 1.3756177924217463, + "grad_norm": 0.6380139244277822, + "learning_rate": 3.8881539634295287e-05, + "loss": 0.2962, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.33058857917785645, + "step": 835, + "valid_targets_mean": 2910.9, + "valid_targets_min": 627 + }, + { + "epoch": 1.3838550247116967, + "grad_norm": 0.6309951441851109, + "learning_rate": 3.88542919707676e-05, + "loss": 0.3112, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25456634163856506, + "step": 840, + "valid_targets_mean": 3141.1, + "valid_targets_min": 1207 + }, + { + "epoch": 1.3920922570016474, + "grad_norm": 0.631875618821807, + "learning_rate": 3.882672617066266e-05, + "loss": 0.2905, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26138928532600403, + "step": 845, + "valid_targets_mean": 2551.6, + "valid_targets_min": 803 + }, + { + "epoch": 1.400329489291598, + "grad_norm": 0.6937940108661667, + "learning_rate": 3.8798842699110075e-05, + "loss": 0.2744, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2633447051048279, + "step": 850, + "valid_targets_mean": 2452.7, + "valid_targets_min": 858 + }, + { + "epoch": 1.4085667215815487, + "grad_norm": 0.550875012894034, + "learning_rate": 3.877064202659964e-05, + "loss": 0.3048, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30909478664398193, + "step": 855, + "valid_targets_mean": 3598.6, + "valid_targets_min": 867 + }, + { + "epoch": 1.4168039538714992, + "grad_norm": 0.7341957843695258, + "learning_rate": 3.874212462897343e-05, + "loss": 0.2858, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32587555050849915, + "step": 860, + "valid_targets_mean": 2998.2, + "valid_targets_min": 187 + }, + { + "epoch": 1.4250411861614498, + "grad_norm": 0.564651910132328, + "learning_rate": 3.871329098741776e-05, + "loss": 0.303, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27226191759109497, + "step": 865, + "valid_targets_mean": 2943.2, + "valid_targets_min": 1012 + }, + { + "epoch": 1.4332784184514002, + "grad_norm": 0.6486502156947155, + "learning_rate": 3.8684141588455076e-05, + "loss": 0.281, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2052830308675766, + "step": 870, + "valid_targets_mean": 2074.1, + "valid_targets_min": 290 + }, + { + "epoch": 1.441515650741351, + "grad_norm": 0.6013453446093197, + "learning_rate": 3.865467692393571e-05, + "loss": 0.2655, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24652577936649323, + "step": 875, + "valid_targets_mean": 2540.7, + "valid_targets_min": 746 + }, + { + "epoch": 1.4497528830313016, + "grad_norm": 0.5514571107088988, + "learning_rate": 3.862489749102963e-05, + "loss": 0.3127, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31356847286224365, + "step": 880, + "valid_targets_mean": 3385.4, + "valid_targets_min": 767 + }, + { + "epoch": 1.457990115321252, + "grad_norm": 0.7273008617384937, + "learning_rate": 3.8594803792218e-05, + "loss": 0.2913, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32229501008987427, + "step": 885, + "valid_targets_mean": 2529.9, + "valid_targets_min": 245 + }, + { + "epoch": 1.4662273476112027, + "grad_norm": 0.6091081729268533, + "learning_rate": 3.856439633528476e-05, + "loss": 0.3386, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31231796741485596, + "step": 890, + "valid_targets_mean": 2814.8, + "valid_targets_min": 1098 + }, + { + "epoch": 1.474464579901153, + "grad_norm": 0.5906549436799012, + "learning_rate": 3.853367563330799e-05, + "loss": 0.282, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24671581387519836, + "step": 895, + "valid_targets_mean": 3008.8, + "valid_targets_min": 768 + }, + { + "epoch": 1.4827018121911038, + "grad_norm": 0.6097661246585641, + "learning_rate": 3.850264220465134e-05, + "loss": 0.2837, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2906344532966614, + "step": 900, + "valid_targets_mean": 2819.9, + "valid_targets_min": 857 + }, + { + "epoch": 1.4909390444810544, + "grad_norm": 0.6471614592146676, + "learning_rate": 3.847129657295516e-05, + "loss": 0.2804, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26892954111099243, + "step": 905, + "valid_targets_mean": 3368.9, + "valid_targets_min": 806 + }, + { + "epoch": 1.499176276771005, + "grad_norm": 0.6035423144036351, + "learning_rate": 3.843963926712778e-05, + "loss": 0.2751, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25377434492111206, + "step": 910, + "valid_targets_mean": 2901.8, + "valid_targets_min": 633 + }, + { + "epoch": 1.5074135090609555, + "grad_norm": 0.7343329864064728, + "learning_rate": 3.840767082133656e-05, + "loss": 0.2875, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28473225235939026, + "step": 915, + "valid_targets_mean": 2271.8, + "valid_targets_min": 845 + }, + { + "epoch": 1.515650741350906, + "grad_norm": 0.7136048929578265, + "learning_rate": 3.837539177499882e-05, + "loss": 0.3064, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3092586398124695, + "step": 920, + "valid_targets_mean": 3439.1, + "valid_targets_min": 811 + }, + { + "epoch": 1.5238879736408566, + "grad_norm": 0.5393319708825398, + "learning_rate": 3.834280267277279e-05, + "loss": 0.326, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.33659571409225464, + "step": 925, + "valid_targets_mean": 3570.9, + "valid_targets_min": 1369 + }, + { + "epoch": 1.5321252059308073, + "grad_norm": 0.5596333567807844, + "learning_rate": 3.830990406454842e-05, + "loss": 0.2971, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25617244839668274, + "step": 930, + "valid_targets_mean": 2960.6, + "valid_targets_min": 312 + }, + { + "epoch": 1.540362438220758, + "grad_norm": 0.5510656760962294, + "learning_rate": 3.827669650543805e-05, + "loss": 0.2882, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.256226509809494, + "step": 935, + "valid_targets_mean": 2793.9, + "valid_targets_min": 248 + }, + { + "epoch": 1.5485996705107083, + "grad_norm": 0.6314296620321947, + "learning_rate": 3.8243180555767135e-05, + "loss": 0.2822, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26010096073150635, + "step": 940, + "valid_targets_mean": 2288.9, + "valid_targets_min": 626 + }, + { + "epoch": 1.556836902800659, + "grad_norm": 0.633916371794186, + "learning_rate": 3.8209356781064686e-05, + "loss": 0.2693, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26327386498451233, + "step": 945, + "valid_targets_mean": 2440.3, + "valid_targets_min": 253 + }, + { + "epoch": 1.5650741350906094, + "grad_norm": 0.5989790640609941, + "learning_rate": 3.8175225752053816e-05, + "loss": 0.316, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.305331826210022, + "step": 950, + "valid_targets_mean": 2859.9, + "valid_targets_min": 405 + }, + { + "epoch": 1.57331136738056, + "grad_norm": 0.5603727241833217, + "learning_rate": 3.814078804464205e-05, + "loss": 0.3024, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24717459082603455, + "step": 955, + "valid_targets_mean": 2985.9, + "valid_targets_min": 836 + }, + { + "epoch": 1.5815485996705108, + "grad_norm": 0.6330587421810094, + "learning_rate": 3.8106044239911645e-05, + "loss": 0.2827, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.293704628944397, + "step": 960, + "valid_targets_mean": 2647.1, + "valid_targets_min": 256 + }, + { + "epoch": 1.5897858319604614, + "grad_norm": 0.7241910589825704, + "learning_rate": 3.8070994924109746e-05, + "loss": 0.2833, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26694560050964355, + "step": 965, + "valid_targets_mean": 3463.4, + "valid_targets_min": 799 + }, + { + "epoch": 1.5980230642504119, + "grad_norm": 0.6839216284178794, + "learning_rate": 3.803564068863854e-05, + "loss": 0.2974, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3023781180381775, + "step": 970, + "valid_targets_mean": 2531.4, + "valid_targets_min": 271 + }, + { + "epoch": 1.6062602965403623, + "grad_norm": 0.5965662328078362, + "learning_rate": 3.7999982130045246e-05, + "loss": 0.305, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28980326652526855, + "step": 975, + "valid_targets_mean": 3020.4, + "valid_targets_min": 869 + }, + { + "epoch": 1.614497528830313, + "grad_norm": 0.6306332846050356, + "learning_rate": 3.796401985001206e-05, + "loss": 0.2763, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3018638491630554, + "step": 980, + "valid_targets_mean": 2907.6, + "valid_targets_min": 1108 + }, + { + "epoch": 1.6227347611202636, + "grad_norm": 0.6196308352831952, + "learning_rate": 3.792775445534601e-05, + "loss": 0.2874, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28585222363471985, + "step": 985, + "valid_targets_mean": 2755.2, + "valid_targets_min": 1110 + }, + { + "epoch": 1.6309719934102143, + "grad_norm": 0.5061334101918796, + "learning_rate": 3.7891186557968665e-05, + "loss": 0.2796, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2653998136520386, + "step": 990, + "valid_targets_mean": 3828.8, + "valid_targets_min": 380 + }, + { + "epoch": 1.6392092257001647, + "grad_norm": 0.5548171900791329, + "learning_rate": 3.785431677490592e-05, + "loss": 0.276, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25885725021362305, + "step": 995, + "valid_targets_mean": 3003.2, + "valid_targets_min": 716 + }, + { + "epoch": 1.6474464579901154, + "grad_norm": 0.544647990594083, + "learning_rate": 3.781714572827746e-05, + "loss": 0.3231, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3178400993347168, + "step": 1000, + "valid_targets_mean": 3262.6, + "valid_targets_min": 575 + }, + { + "epoch": 1.6556836902800658, + "grad_norm": 0.5957620379095658, + "learning_rate": 3.777967404528636e-05, + "loss": 0.315, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3391578793525696, + "step": 1005, + "valid_targets_mean": 3305.6, + "valid_targets_min": 224 + }, + { + "epoch": 1.6639209225700164, + "grad_norm": 0.5831365067144099, + "learning_rate": 3.774190235820843e-05, + "loss": 0.2871, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27633339166641235, + "step": 1010, + "valid_targets_mean": 3138.8, + "valid_targets_min": 1102 + }, + { + "epoch": 1.672158154859967, + "grad_norm": 0.5866526617411725, + "learning_rate": 3.770383130438159e-05, + "loss": 0.2656, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2521780729293823, + "step": 1015, + "valid_targets_mean": 2772.1, + "valid_targets_min": 899 + }, + { + "epoch": 1.6803953871499178, + "grad_norm": 0.6617850729419059, + "learning_rate": 3.76654615261951e-05, + "loss": 0.2942, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2704291045665741, + "step": 1020, + "valid_targets_mean": 2489.5, + "valid_targets_min": 800 + }, + { + "epoch": 1.6886326194398682, + "grad_norm": 0.6758228599415805, + "learning_rate": 3.7626793671078743e-05, + "loss": 0.2955, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29394012689590454, + "step": 1025, + "valid_targets_mean": 2479.1, + "valid_targets_min": 475 + }, + { + "epoch": 1.6968698517298186, + "grad_norm": 0.6302344983972948, + "learning_rate": 3.758782839149187e-05, + "loss": 0.3047, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2621959447860718, + "step": 1030, + "valid_targets_mean": 2202.9, + "valid_targets_min": 841 + }, + { + "epoch": 1.7051070840197693, + "grad_norm": 0.7505883002451076, + "learning_rate": 3.754856634491238e-05, + "loss": 0.3074, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3061571717262268, + "step": 1035, + "valid_targets_mean": 2011.4, + "valid_targets_min": 552 + }, + { + "epoch": 1.71334431630972, + "grad_norm": 0.6982098895422223, + "learning_rate": 3.75090081938257e-05, + "loss": 0.3261, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32887348532676697, + "step": 1040, + "valid_targets_mean": 2548.1, + "valid_targets_min": 264 + }, + { + "epoch": 1.7215815485996706, + "grad_norm": 0.6826562094159441, + "learning_rate": 3.7469154605713496e-05, + "loss": 0.2725, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2549431622028351, + "step": 1045, + "valid_targets_mean": 2234.9, + "valid_targets_min": 311 + }, + { + "epoch": 1.729818780889621, + "grad_norm": 0.6436168602201539, + "learning_rate": 3.742900625304252e-05, + "loss": 0.3193, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26707977056503296, + "step": 1050, + "valid_targets_mean": 2868.2, + "valid_targets_min": 730 + }, + { + "epoch": 1.7380560131795717, + "grad_norm": 0.6004930059899968, + "learning_rate": 3.7388563813253194e-05, + "loss": 0.2925, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.33015143871307373, + "step": 1055, + "valid_targets_mean": 3110.4, + "valid_targets_min": 728 + }, + { + "epoch": 1.7462932454695221, + "grad_norm": 0.5441236234486593, + "learning_rate": 3.7347827968748196e-05, + "loss": 0.3109, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2715633511543274, + "step": 1060, + "valid_targets_mean": 3396.2, + "valid_targets_min": 935 + }, + { + "epoch": 1.7545304777594728, + "grad_norm": 0.6675223889236319, + "learning_rate": 3.730679940688094e-05, + "loss": 0.2677, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3189924657344818, + "step": 1065, + "valid_targets_mean": 2620.6, + "valid_targets_min": 308 + }, + { + "epoch": 1.7627677100494235, + "grad_norm": 0.6168969110960196, + "learning_rate": 3.7265478819944004e-05, + "loss": 0.2884, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3395520746707916, + "step": 1070, + "valid_targets_mean": 2912.7, + "valid_targets_min": 985 + }, + { + "epoch": 1.771004942339374, + "grad_norm": 0.6046464594479612, + "learning_rate": 3.722386690515742e-05, + "loss": 0.2646, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24619776010513306, + "step": 1075, + "valid_targets_mean": 2565.3, + "valid_targets_min": 936 + }, + { + "epoch": 1.7792421746293245, + "grad_norm": 0.5733766197797237, + "learning_rate": 3.718196436465693e-05, + "loss": 0.3004, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24001248180866241, + "step": 1080, + "valid_targets_mean": 3276.8, + "valid_targets_min": 1391 + }, + { + "epoch": 1.787479406919275, + "grad_norm": 0.5359416776650924, + "learning_rate": 3.713977190548211e-05, + "loss": 0.2791, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31075796484947205, + "step": 1085, + "valid_targets_mean": 3900.6, + "valid_targets_min": 1288 + }, + { + "epoch": 1.7957166392092256, + "grad_norm": 0.7891318947723374, + "learning_rate": 3.709729023956448e-05, + "loss": 0.2452, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24248431622982025, + "step": 1090, + "valid_targets_mean": 2105.8, + "valid_targets_min": 525 + }, + { + "epoch": 1.8039538714991763, + "grad_norm": 0.4963674547261414, + "learning_rate": 3.705452008371545e-05, + "loss": 0.2889, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2084948718547821, + "step": 1095, + "valid_targets_mean": 3216.0, + "valid_targets_min": 539 + }, + { + "epoch": 1.812191103789127, + "grad_norm": 0.6700038145214892, + "learning_rate": 3.701146215961426e-05, + "loss": 0.2789, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2652949392795563, + "step": 1100, + "valid_targets_mean": 2111.5, + "valid_targets_min": 832 + }, + { + "epoch": 1.8204283360790774, + "grad_norm": 0.5716232564162521, + "learning_rate": 3.696811719379578e-05, + "loss": 0.27, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3448944389820099, + "step": 1105, + "valid_targets_mean": 3898.4, + "valid_targets_min": 287 + }, + { + "epoch": 1.828665568369028, + "grad_norm": 0.5998885452948804, + "learning_rate": 3.692448591763827e-05, + "loss": 0.266, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2568257451057434, + "step": 1110, + "valid_targets_mean": 2799.3, + "valid_targets_min": 751 + }, + { + "epoch": 1.8369028006589785, + "grad_norm": 0.7726688835753368, + "learning_rate": 3.6880569067350994e-05, + "loss": 0.298, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2870601415634155, + "step": 1115, + "valid_targets_mean": 2597.6, + "valid_targets_min": 275 + }, + { + "epoch": 1.8451400329489291, + "grad_norm": 0.6739523370370597, + "learning_rate": 3.683636738396187e-05, + "loss": 0.2979, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.295218288898468, + "step": 1120, + "valid_targets_mean": 2633.4, + "valid_targets_min": 323 + }, + { + "epoch": 1.8533772652388798, + "grad_norm": 0.5074395074268739, + "learning_rate": 3.6791881613304916e-05, + "loss": 0.2703, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2503700256347656, + "step": 1125, + "valid_targets_mean": 3885.8, + "valid_targets_min": 861 + }, + { + "epoch": 1.8616144975288305, + "grad_norm": 0.6419133897350886, + "learning_rate": 3.674711250600766e-05, + "loss": 0.3142, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30119603872299194, + "step": 1130, + "valid_targets_mean": 2609.1, + "valid_targets_min": 1010 + }, + { + "epoch": 1.869851729818781, + "grad_norm": 0.6074389296511881, + "learning_rate": 3.6702060817478524e-05, + "loss": 0.264, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2686932682991028, + "step": 1135, + "valid_targets_mean": 2378.4, + "valid_targets_min": 776 + }, + { + "epoch": 1.8780889621087313, + "grad_norm": 0.6457310520573254, + "learning_rate": 3.6656727307894e-05, + "loss": 0.3014, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2615449130535126, + "step": 1140, + "valid_targets_mean": 2275.4, + "valid_targets_min": 792 + }, + { + "epoch": 1.886326194398682, + "grad_norm": 0.6191056984207056, + "learning_rate": 3.6611112742185913e-05, + "loss": 0.2806, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25483569502830505, + "step": 1145, + "valid_targets_mean": 2271.7, + "valid_targets_min": 352 + }, + { + "epoch": 1.8945634266886326, + "grad_norm": 0.6246965459645147, + "learning_rate": 3.6565217890028435e-05, + "loss": 0.3055, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2788812816143036, + "step": 1150, + "valid_targets_mean": 2767.0, + "valid_targets_min": 679 + }, + { + "epoch": 1.9028006589785833, + "grad_norm": 0.656487573533247, + "learning_rate": 3.651904352582514e-05, + "loss": 0.3078, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3155135214328766, + "step": 1155, + "valid_targets_mean": 2277.2, + "valid_targets_min": 216 + }, + { + "epoch": 1.9110378912685337, + "grad_norm": 0.6272798767220624, + "learning_rate": 3.647259042869595e-05, + "loss": 0.301, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3550041615962982, + "step": 1160, + "valid_targets_mean": 2828.5, + "valid_targets_min": 356 + }, + { + "epoch": 1.9192751235584844, + "grad_norm": 0.6089019795513212, + "learning_rate": 3.642585938246394e-05, + "loss": 0.2961, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3001123070716858, + "step": 1165, + "valid_targets_mean": 2918.1, + "valid_targets_min": 1368 + }, + { + "epoch": 1.9275123558484348, + "grad_norm": 0.7155478877487227, + "learning_rate": 3.6378851175642166e-05, + "loss": 0.2658, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31469109654426575, + "step": 1170, + "valid_targets_mean": 2766.4, + "valid_targets_min": 764 + }, + { + "epoch": 1.9357495881383855, + "grad_norm": 0.5900071227360756, + "learning_rate": 3.63315666014203e-05, + "loss": 0.284, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28159189224243164, + "step": 1175, + "valid_targets_mean": 2722.8, + "valid_targets_min": 1564 + }, + { + "epoch": 1.9439868204283361, + "grad_norm": 0.5720042983430078, + "learning_rate": 3.6284006457651295e-05, + "loss": 0.271, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.306449294090271, + "step": 1180, + "valid_targets_mean": 3149.6, + "valid_targets_min": 821 + }, + { + "epoch": 1.9522240527182868, + "grad_norm": 0.6518952324340549, + "learning_rate": 3.6236171546837936e-05, + "loss": 0.2864, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28307563066482544, + "step": 1185, + "valid_targets_mean": 2688.5, + "valid_targets_min": 654 + }, + { + "epoch": 1.9604612850082372, + "grad_norm": 0.5643073151697097, + "learning_rate": 3.618806267611923e-05, + "loss": 0.3133, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32395491003990173, + "step": 1190, + "valid_targets_mean": 3477.4, + "valid_targets_min": 600 + }, + { + "epoch": 1.9686985172981877, + "grad_norm": 0.6448164120369934, + "learning_rate": 3.613968065725685e-05, + "loss": 0.2715, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2536458373069763, + "step": 1195, + "valid_targets_mean": 2676.1, + "valid_targets_min": 942 + }, + { + "epoch": 1.9769357495881383, + "grad_norm": 2.2836948181723016, + "learning_rate": 3.60910263066214e-05, + "loss": 0.2934, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24172762036323547, + "step": 1200, + "valid_targets_mean": 2737.7, + "valid_targets_min": 1061 + }, + { + "epoch": 1.985172981878089, + "grad_norm": 0.5707078354838783, + "learning_rate": 3.604210044517867e-05, + "loss": 0.3117, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31775957345962524, + "step": 1205, + "valid_targets_mean": 3227.9, + "valid_targets_min": 992 + }, + { + "epoch": 1.9934102141680397, + "grad_norm": 0.7989744132116524, + "learning_rate": 3.599290389847576e-05, + "loss": 0.3281, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4108525514602661, + "step": 1210, + "valid_targets_mean": 3928.9, + "valid_targets_min": 869 + }, + { + "epoch": 2.0016474464579903, + "grad_norm": 0.6264226403015248, + "learning_rate": 3.594343749662716e-05, + "loss": 0.2849, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27158790826797485, + "step": 1215, + "valid_targets_mean": 2425.3, + "valid_targets_min": 234 + }, + { + "epoch": 2.0098846787479405, + "grad_norm": 0.5942699391272229, + "learning_rate": 3.5893702074300745e-05, + "loss": 0.2483, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23825138807296753, + "step": 1220, + "valid_targets_mean": 2739.1, + "valid_targets_min": 233 + }, + { + "epoch": 2.018121911037891, + "grad_norm": 0.7236717591594628, + "learning_rate": 3.584369847070369e-05, + "loss": 0.259, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27776312828063965, + "step": 1225, + "valid_targets_mean": 2409.8, + "valid_targets_min": 795 + }, + { + "epoch": 2.026359143327842, + "grad_norm": 0.5761714008930021, + "learning_rate": 3.579342752956829e-05, + "loss": 0.2391, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2385309338569641, + "step": 1230, + "valid_targets_mean": 3224.9, + "valid_targets_min": 835 + }, + { + "epoch": 2.0345963756177925, + "grad_norm": 0.5691413404913723, + "learning_rate": 3.574289009913776e-05, + "loss": 0.2302, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18973645567893982, + "step": 1235, + "valid_targets_mean": 3347.5, + "valid_targets_min": 785 + }, + { + "epoch": 2.042833607907743, + "grad_norm": 0.6341526990742906, + "learning_rate": 3.5692087032151915e-05, + "loss": 0.2612, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24696478247642517, + "step": 1240, + "valid_targets_mean": 2765.2, + "valid_targets_min": 334 + }, + { + "epoch": 2.0510708401976934, + "grad_norm": 0.5063130857062907, + "learning_rate": 3.564101918583274e-05, + "loss": 0.2404, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24281460046768188, + "step": 1245, + "valid_targets_mean": 4392.6, + "valid_targets_min": 1107 + }, + { + "epoch": 2.059308072487644, + "grad_norm": 0.6823195765956594, + "learning_rate": 3.5589687421869964e-05, + "loss": 0.2555, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24046598374843597, + "step": 1250, + "valid_targets_mean": 2668.4, + "valid_targets_min": 879 + }, + { + "epoch": 2.0675453047775947, + "grad_norm": 0.5891801634089261, + "learning_rate": 3.553809260640653e-05, + "loss": 0.2589, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24028754234313965, + "step": 1255, + "valid_targets_mean": 3647.4, + "valid_targets_min": 878 + }, + { + "epoch": 2.0757825370675453, + "grad_norm": 0.560587648872072, + "learning_rate": 3.5486235610023935e-05, + "loss": 0.2715, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25208956003189087, + "step": 1260, + "valid_targets_mean": 3228.0, + "valid_targets_min": 1006 + }, + { + "epoch": 2.084019769357496, + "grad_norm": 0.5431257163133473, + "learning_rate": 3.54341173077276e-05, + "loss": 0.2669, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2619468569755554, + "step": 1265, + "valid_targets_mean": 3691.9, + "valid_targets_min": 1020 + }, + { + "epoch": 2.0922570016474467, + "grad_norm": 0.6876181195287832, + "learning_rate": 3.5381738578932035e-05, + "loss": 0.2478, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2359652817249298, + "step": 1270, + "valid_targets_mean": 2196.4, + "valid_targets_min": 896 + }, + { + "epoch": 2.100494233937397, + "grad_norm": 0.7408152195912641, + "learning_rate": 3.532910030744607e-05, + "loss": 0.241, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2573264539241791, + "step": 1275, + "valid_targets_mean": 2346.4, + "valid_targets_min": 297 + }, + { + "epoch": 2.1087314662273475, + "grad_norm": 0.6103530118299919, + "learning_rate": 3.5276203381457894e-05, + "loss": 0.2616, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2797989845275879, + "step": 1280, + "valid_targets_mean": 3206.1, + "valid_targets_min": 674 + }, + { + "epoch": 2.116968698517298, + "grad_norm": 0.5891608135099099, + "learning_rate": 3.5223048693520094e-05, + "loss": 0.2476, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2533876597881317, + "step": 1285, + "valid_targets_mean": 3108.0, + "valid_targets_min": 591 + }, + { + "epoch": 2.125205930807249, + "grad_norm": 0.6208888605283639, + "learning_rate": 3.5169637140534574e-05, + "loss": 0.2491, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2187802642583847, + "step": 1290, + "valid_targets_mean": 2592.9, + "valid_targets_min": 281 + }, + { + "epoch": 2.1334431630971995, + "grad_norm": 0.6676297767164048, + "learning_rate": 3.5115969623737447e-05, + "loss": 0.2604, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24208733439445496, + "step": 1295, + "valid_targets_mean": 2669.2, + "valid_targets_min": 330 + }, + { + "epoch": 2.1416803953871497, + "grad_norm": 0.7229124706726481, + "learning_rate": 3.506204704868379e-05, + "loss": 0.2634, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2806577980518341, + "step": 1300, + "valid_targets_mean": 2626.9, + "valid_targets_min": 349 + }, + { + "epoch": 2.1499176276771004, + "grad_norm": 0.5658374542999091, + "learning_rate": 3.500787032523242e-05, + "loss": 0.2518, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21356990933418274, + "step": 1305, + "valid_targets_mean": 3147.4, + "valid_targets_min": 275 + }, + { + "epoch": 2.158154859967051, + "grad_norm": 0.6089410486353419, + "learning_rate": 3.495344036753048e-05, + "loss": 0.2571, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27112677693367004, + "step": 1310, + "valid_targets_mean": 3331.5, + "valid_targets_min": 394 + }, + { + "epoch": 2.1663920922570017, + "grad_norm": 0.6145684158379482, + "learning_rate": 3.489875809399806e-05, + "loss": 0.2787, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2448844313621521, + "step": 1315, + "valid_targets_mean": 2984.2, + "valid_targets_min": 265 + }, + { + "epoch": 2.1746293245469523, + "grad_norm": 0.6846498864389476, + "learning_rate": 3.484382442731268e-05, + "loss": 0.2639, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2414667010307312, + "step": 1320, + "valid_targets_mean": 2606.9, + "valid_targets_min": 644 + }, + { + "epoch": 2.182866556836903, + "grad_norm": 0.73318481182418, + "learning_rate": 3.4788640294393725e-05, + "loss": 0.2743, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2895458936691284, + "step": 1325, + "valid_targets_mean": 2585.8, + "valid_targets_min": 216 + }, + { + "epoch": 2.191103789126853, + "grad_norm": 0.6942685225765722, + "learning_rate": 3.47332066263868e-05, + "loss": 0.265, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2770744562149048, + "step": 1330, + "valid_targets_mean": 3022.6, + "valid_targets_min": 242 + }, + { + "epoch": 2.199341021416804, + "grad_norm": 0.520985364981708, + "learning_rate": 3.467752435864803e-05, + "loss": 0.232, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20206497609615326, + "step": 1335, + "valid_targets_mean": 3354.1, + "valid_targets_min": 809 + }, + { + "epoch": 2.2075782537067545, + "grad_norm": 0.5832089119364511, + "learning_rate": 3.462159443072826e-05, + "loss": 0.2641, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2927054166793823, + "step": 1340, + "valid_targets_mean": 3694.7, + "valid_targets_min": 948 + }, + { + "epoch": 2.215815485996705, + "grad_norm": 0.58897748237891, + "learning_rate": 3.456541778635722e-05, + "loss": 0.2676, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19613176584243774, + "step": 1345, + "valid_targets_mean": 2776.6, + "valid_targets_min": 1254 + }, + { + "epoch": 2.224052718286656, + "grad_norm": 0.6484887362652936, + "learning_rate": 3.45089953734276e-05, + "loss": 0.2539, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20931027829647064, + "step": 1350, + "valid_targets_mean": 2288.8, + "valid_targets_min": 297 + }, + { + "epoch": 2.232289950576606, + "grad_norm": 0.5929321738275963, + "learning_rate": 3.445232814397902e-05, + "loss": 0.2367, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19637475907802582, + "step": 1355, + "valid_targets_mean": 2361.2, + "valid_targets_min": 912 + }, + { + "epoch": 2.2405271828665567, + "grad_norm": 0.6012994089232492, + "learning_rate": 3.4395417054182034e-05, + "loss": 0.2481, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27980607748031616, + "step": 1360, + "valid_targets_mean": 3480.1, + "valid_targets_min": 402 + }, + { + "epoch": 2.2487644151565074, + "grad_norm": 0.5960949440370955, + "learning_rate": 3.433826306432193e-05, + "loss": 0.26, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25292590260505676, + "step": 1365, + "valid_targets_mean": 2758.2, + "valid_targets_min": 886 + }, + { + "epoch": 2.257001647446458, + "grad_norm": 0.5481540263656652, + "learning_rate": 3.428086713878255e-05, + "loss": 0.2143, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18284395337104797, + "step": 1370, + "valid_targets_mean": 3020.9, + "valid_targets_min": 380 + }, + { + "epoch": 2.2652388797364087, + "grad_norm": 0.6107676338306715, + "learning_rate": 3.422323024603004e-05, + "loss": 0.2414, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2406243532896042, + "step": 1375, + "valid_targets_mean": 3207.1, + "valid_targets_min": 1181 + }, + { + "epoch": 2.2734761120263594, + "grad_norm": 0.7818816948424279, + "learning_rate": 3.4165353358596475e-05, + "loss": 0.3273, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.35243645310401917, + "step": 1380, + "valid_targets_mean": 3005.8, + "valid_targets_min": 688 + }, + { + "epoch": 2.2817133443163096, + "grad_norm": 0.6375768102457982, + "learning_rate": 3.410723745306348e-05, + "loss": 0.2583, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23199833929538727, + "step": 1385, + "valid_targets_mean": 2873.8, + "valid_targets_min": 735 + }, + { + "epoch": 2.2899505766062602, + "grad_norm": 0.7095261625145105, + "learning_rate": 3.404888351004571e-05, + "loss": 0.2554, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.275643527507782, + "step": 1390, + "valid_targets_mean": 2629.0, + "valid_targets_min": 930 + }, + { + "epoch": 2.298187808896211, + "grad_norm": 0.5955591755415022, + "learning_rate": 3.3990292514174346e-05, + "loss": 0.2624, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2331334948539734, + "step": 1395, + "valid_targets_mean": 2833.2, + "valid_targets_min": 776 + }, + { + "epoch": 2.3064250411861615, + "grad_norm": 0.5761183125798899, + "learning_rate": 3.393146545408047e-05, + "loss": 0.2595, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2591421902179718, + "step": 1400, + "valid_targets_mean": 3131.5, + "valid_targets_min": 830 + }, + { + "epoch": 2.314662273476112, + "grad_norm": 0.646758804736525, + "learning_rate": 3.3872403322378354e-05, + "loss": 0.2769, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23602330684661865, + "step": 1405, + "valid_targets_mean": 2488.8, + "valid_targets_min": 226 + }, + { + "epoch": 2.3228995057660624, + "grad_norm": 0.6866011148331042, + "learning_rate": 3.3813107115648745e-05, + "loss": 0.2592, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3126336634159088, + "step": 1410, + "valid_targets_mean": 2544.7, + "valid_targets_min": 723 + }, + { + "epoch": 2.331136738056013, + "grad_norm": 0.6346999798467989, + "learning_rate": 3.375357783442205e-05, + "loss": 0.2882, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2111891508102417, + "step": 1415, + "valid_targets_mean": 2556.5, + "valid_targets_min": 763 + }, + { + "epoch": 2.3393739703459637, + "grad_norm": 0.6309803885063188, + "learning_rate": 3.369381648316144e-05, + "loss": 0.2521, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24223890900611877, + "step": 1420, + "valid_targets_mean": 2757.2, + "valid_targets_min": 704 + }, + { + "epoch": 2.3476112026359144, + "grad_norm": 0.6326584329828758, + "learning_rate": 3.363382407024588e-05, + "loss": 0.2404, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26198041439056396, + "step": 1425, + "valid_targets_mean": 2648.4, + "valid_targets_min": 1051 + }, + { + "epoch": 2.355848434925865, + "grad_norm": 0.6361946094436111, + "learning_rate": 3.357360160795317e-05, + "loss": 0.2738, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2635315954685211, + "step": 1430, + "valid_targets_mean": 2999.0, + "valid_targets_min": 1031 + }, + { + "epoch": 2.3640856672158153, + "grad_norm": 0.6154578078423868, + "learning_rate": 3.35131501124428e-05, + "loss": 0.2799, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19209545850753784, + "step": 1435, + "valid_targets_mean": 2452.1, + "valid_targets_min": 679 + }, + { + "epoch": 2.372322899505766, + "grad_norm": 0.5871616148619758, + "learning_rate": 3.345247060373887e-05, + "loss": 0.2543, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23240095376968384, + "step": 1440, + "valid_targets_mean": 3175.5, + "valid_targets_min": 700 + }, + { + "epoch": 2.3805601317957166, + "grad_norm": 0.6841219210951218, + "learning_rate": 3.3391564105712824e-05, + "loss": 0.2715, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.33838337659835815, + "step": 1445, + "valid_targets_mean": 2693.4, + "valid_targets_min": 899 + }, + { + "epoch": 2.3887973640856672, + "grad_norm": 0.5614836697402698, + "learning_rate": 3.333043164606618e-05, + "loss": 0.2488, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2738487124443054, + "step": 1450, + "valid_targets_mean": 3429.0, + "valid_targets_min": 951 + }, + { + "epoch": 2.397034596375618, + "grad_norm": 0.53864317436498, + "learning_rate": 3.3269074256313246e-05, + "loss": 0.244, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20876136422157288, + "step": 1455, + "valid_targets_mean": 3511.2, + "valid_targets_min": 751 + }, + { + "epoch": 2.4052718286655685, + "grad_norm": 0.655242664968237, + "learning_rate": 3.320749297176363e-05, + "loss": 0.2368, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22833922505378723, + "step": 1460, + "valid_targets_mean": 2380.5, + "valid_targets_min": 346 + }, + { + "epoch": 2.4135090609555188, + "grad_norm": 0.8232236295806061, + "learning_rate": 3.3145688831504845e-05, + "loss": 0.2796, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2448880970478058, + "step": 1465, + "valid_targets_mean": 3051.5, + "valid_targets_min": 712 + }, + { + "epoch": 2.4217462932454694, + "grad_norm": 0.638531424640081, + "learning_rate": 3.308366287838472e-05, + "loss": 0.2578, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23629572987556458, + "step": 1470, + "valid_targets_mean": 2371.6, + "valid_targets_min": 675 + }, + { + "epoch": 2.42998352553542, + "grad_norm": 0.6526933180553507, + "learning_rate": 3.302141615899384e-05, + "loss": 0.2452, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2188447117805481, + "step": 1475, + "valid_targets_mean": 2430.1, + "valid_targets_min": 750 + }, + { + "epoch": 2.4382207578253707, + "grad_norm": 0.6236160880845308, + "learning_rate": 3.2958949723647875e-05, + "loss": 0.2475, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29466351866722107, + "step": 1480, + "valid_targets_mean": 2820.9, + "valid_targets_min": 615 + }, + { + "epoch": 2.4464579901153214, + "grad_norm": 0.6349685298066351, + "learning_rate": 3.289626462636985e-05, + "loss": 0.2461, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22705788910388947, + "step": 1485, + "valid_targets_mean": 2800.2, + "valid_targets_min": 638 + }, + { + "epoch": 2.454695222405272, + "grad_norm": 0.6592423626712279, + "learning_rate": 3.283336192487238e-05, + "loss": 0.2424, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22058206796646118, + "step": 1490, + "valid_targets_mean": 2350.7, + "valid_targets_min": 743 + }, + { + "epoch": 2.4629324546952223, + "grad_norm": 0.6228303104982533, + "learning_rate": 3.277024268053978e-05, + "loss": 0.2722, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24452120065689087, + "step": 1495, + "valid_targets_mean": 3045.2, + "valid_targets_min": 312 + }, + { + "epoch": 2.471169686985173, + "grad_norm": 0.6786946251861339, + "learning_rate": 3.2706907958410234e-05, + "loss": 0.2379, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24040010571479797, + "step": 1500, + "valid_targets_mean": 2688.6, + "valid_targets_min": 1069 + }, + { + "epoch": 2.4794069192751236, + "grad_norm": 0.56494805323628, + "learning_rate": 3.2643358827157724e-05, + "loss": 0.2111, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2005000114440918, + "step": 1505, + "valid_targets_mean": 3374.3, + "valid_targets_min": 1096 + }, + { + "epoch": 2.4876441515650742, + "grad_norm": 0.597724401455926, + "learning_rate": 3.257959635907409e-05, + "loss": 0.2325, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22339114546775818, + "step": 1510, + "valid_targets_mean": 2986.3, + "valid_targets_min": 920 + }, + { + "epoch": 2.495881383855025, + "grad_norm": 0.6561352881615815, + "learning_rate": 3.251562163005087e-05, + "loss": 0.2505, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22298067808151245, + "step": 1515, + "valid_targets_mean": 2253.1, + "valid_targets_min": 925 + }, + { + "epoch": 2.504118616144975, + "grad_norm": 0.6162020968409021, + "learning_rate": 3.245143571956117e-05, + "loss": 0.2692, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27440690994262695, + "step": 1520, + "valid_targets_mean": 2877.1, + "valid_targets_min": 821 + }, + { + "epoch": 2.5123558484349258, + "grad_norm": 0.6082712508042523, + "learning_rate": 3.2387039710641486e-05, + "loss": 0.2442, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22499428689479828, + "step": 1525, + "valid_targets_mean": 2861.8, + "valid_targets_min": 289 + }, + { + "epoch": 2.5205930807248764, + "grad_norm": 0.7384333926998505, + "learning_rate": 3.232243468987336e-05, + "loss": 0.2411, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21579819917678833, + "step": 1530, + "valid_targets_mean": 3231.4, + "valid_targets_min": 729 + }, + { + "epoch": 2.528830313014827, + "grad_norm": 0.6877047193253706, + "learning_rate": 3.2257621747365086e-05, + "loss": 0.2731, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3097109794616699, + "step": 1535, + "valid_targets_mean": 2337.8, + "valid_targets_min": 755 + }, + { + "epoch": 2.5370675453047777, + "grad_norm": 0.543065421568093, + "learning_rate": 3.219260197673333e-05, + "loss": 0.2434, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24384282529354095, + "step": 1540, + "valid_targets_mean": 3961.9, + "valid_targets_min": 1428 + }, + { + "epoch": 2.545304777594728, + "grad_norm": 0.6248347745562823, + "learning_rate": 3.212737647508464e-05, + "loss": 0.2458, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28097227215766907, + "step": 1545, + "valid_targets_mean": 3362.5, + "valid_targets_min": 821 + }, + { + "epoch": 2.5535420098846786, + "grad_norm": 0.5691317720774456, + "learning_rate": 3.206194634299696e-05, + "loss": 0.2317, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23817376792430878, + "step": 1550, + "valid_targets_mean": 3836.8, + "valid_targets_min": 1197 + }, + { + "epoch": 2.5617792421746293, + "grad_norm": 0.684915654232255, + "learning_rate": 3.199631268450104e-05, + "loss": 0.2387, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2511272132396698, + "step": 1555, + "valid_targets_mean": 2801.4, + "valid_targets_min": 471 + }, + { + "epoch": 2.57001647446458, + "grad_norm": 0.7982907744163267, + "learning_rate": 3.1930476607061835e-05, + "loss": 0.2806, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.38522768020629883, + "step": 1560, + "valid_targets_mean": 3067.9, + "valid_targets_min": 743 + }, + { + "epoch": 2.5782537067545306, + "grad_norm": 0.6053095568591514, + "learning_rate": 3.1864439221559797e-05, + "loss": 0.2573, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25679144263267517, + "step": 1565, + "valid_targets_mean": 2767.1, + "valid_targets_min": 829 + }, + { + "epoch": 2.586490939044481, + "grad_norm": 0.49767493638955396, + "learning_rate": 3.179820164227213e-05, + "loss": 0.2726, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3335208296775818, + "step": 1570, + "valid_targets_mean": 4638.4, + "valid_targets_min": 783 + }, + { + "epoch": 2.594728171334432, + "grad_norm": 0.6569402555128312, + "learning_rate": 3.1731764986853985e-05, + "loss": 0.2459, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24302177131175995, + "step": 1575, + "valid_targets_mean": 2489.9, + "valid_targets_min": 939 + }, + { + "epoch": 2.602965403624382, + "grad_norm": 0.5921594644458538, + "learning_rate": 3.166513037631965e-05, + "loss": 0.2638, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26702389121055603, + "step": 1580, + "valid_targets_mean": 2795.8, + "valid_targets_min": 744 + }, + { + "epoch": 2.6112026359143328, + "grad_norm": 0.6407404206933615, + "learning_rate": 3.1598298935023536e-05, + "loss": 0.2545, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2785078287124634, + "step": 1585, + "valid_targets_mean": 2990.9, + "valid_targets_min": 630 + }, + { + "epoch": 2.6194398682042834, + "grad_norm": 0.6833324929674904, + "learning_rate": 3.153127179064131e-05, + "loss": 0.2453, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.251248300075531, + "step": 1590, + "valid_targets_mean": 2696.2, + "valid_targets_min": 287 + }, + { + "epoch": 2.627677100494234, + "grad_norm": 0.6339558018061027, + "learning_rate": 3.146405007415081e-05, + "loss": 0.2575, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.281749427318573, + "step": 1595, + "valid_targets_mean": 2749.6, + "valid_targets_min": 849 + }, + { + "epoch": 2.6359143327841847, + "grad_norm": 0.6273664397725301, + "learning_rate": 3.1396634919812955e-05, + "loss": 0.2487, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2563917636871338, + "step": 1600, + "valid_targets_mean": 2623.4, + "valid_targets_min": 902 + }, + { + "epoch": 2.644151565074135, + "grad_norm": 0.6622434597804578, + "learning_rate": 3.1329027465152654e-05, + "loss": 0.2536, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3303392827510834, + "step": 1605, + "valid_targets_mean": 2999.7, + "valid_targets_min": 250 + }, + { + "epoch": 2.6523887973640856, + "grad_norm": 0.6020506711799892, + "learning_rate": 3.126122885093955e-05, + "loss": 0.2354, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22898972034454346, + "step": 1610, + "valid_targets_mean": 2827.9, + "valid_targets_min": 216 + }, + { + "epoch": 2.6606260296540363, + "grad_norm": 0.5656507553130597, + "learning_rate": 3.1193240221168835e-05, + "loss": 0.2258, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24721187353134155, + "step": 1615, + "valid_targets_mean": 3554.2, + "valid_targets_min": 824 + }, + { + "epoch": 2.668863261943987, + "grad_norm": 0.644282073051523, + "learning_rate": 3.112506272304189e-05, + "loss": 0.2411, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2700679898262024, + "step": 1620, + "valid_targets_mean": 2953.6, + "valid_targets_min": 1192 + }, + { + "epoch": 2.6771004942339376, + "grad_norm": 0.6841700983785681, + "learning_rate": 3.1056697506946965e-05, + "loss": 0.2558, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23583650588989258, + "step": 1625, + "valid_targets_mean": 2720.4, + "valid_targets_min": 523 + }, + { + "epoch": 2.685337726523888, + "grad_norm": 0.7340836959990856, + "learning_rate": 3.0988145726439754e-05, + "loss": 0.2696, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.307033509016037, + "step": 1630, + "valid_targets_mean": 2578.5, + "valid_targets_min": 382 + }, + { + "epoch": 2.6935749588138385, + "grad_norm": 0.6673998137709523, + "learning_rate": 3.091940853822393e-05, + "loss": 0.2537, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29108643531799316, + "step": 1635, + "valid_targets_mean": 2832.3, + "valid_targets_min": 1206 + }, + { + "epoch": 2.701812191103789, + "grad_norm": 0.6467186852795253, + "learning_rate": 3.0850487102131624e-05, + "loss": 0.2502, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21308313310146332, + "step": 1640, + "valid_targets_mean": 2594.9, + "valid_targets_min": 360 + }, + { + "epoch": 2.7100494233937398, + "grad_norm": 0.6944087919516522, + "learning_rate": 3.078138258110389e-05, + "loss": 0.2717, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23937004804611206, + "step": 1645, + "valid_targets_mean": 2263.4, + "valid_targets_min": 245 + }, + { + "epoch": 2.7182866556836904, + "grad_norm": 0.5326350886269784, + "learning_rate": 3.071209614117103e-05, + "loss": 0.2712, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3428668975830078, + "step": 1650, + "valid_targets_mean": 4543.8, + "valid_targets_min": 900 + }, + { + "epoch": 2.7265238879736406, + "grad_norm": 0.5078601743477384, + "learning_rate": 3.0642628951432916e-05, + "loss": 0.2629, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22137928009033203, + "step": 1655, + "valid_targets_mean": 3942.6, + "valid_targets_min": 769 + }, + { + "epoch": 2.7347611202635913, + "grad_norm": 0.601538874178317, + "learning_rate": 3.057298218403934e-05, + "loss": 0.2781, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2587862014770508, + "step": 1660, + "valid_targets_mean": 2621.9, + "valid_targets_min": 730 + }, + { + "epoch": 2.742998352553542, + "grad_norm": 0.5100732784028359, + "learning_rate": 3.050315701417016e-05, + "loss": 0.2444, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22114631533622742, + "step": 1665, + "valid_targets_mean": 3594.6, + "valid_targets_min": 1032 + }, + { + "epoch": 2.7512355848434926, + "grad_norm": 0.6403837824931091, + "learning_rate": 3.043315462001549e-05, + "loss": 0.2419, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2376575469970703, + "step": 1670, + "valid_targets_mean": 2361.1, + "valid_targets_min": 1050 + }, + { + "epoch": 2.7594728171334433, + "grad_norm": 0.6630544642728583, + "learning_rate": 3.0362976182755837e-05, + "loss": 0.2392, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2816362679004669, + "step": 1675, + "valid_targets_mean": 2989.6, + "valid_targets_min": 324 + }, + { + "epoch": 2.7677100494233935, + "grad_norm": 0.6196334831818463, + "learning_rate": 3.0292622886542154e-05, + "loss": 0.2662, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24809390306472778, + "step": 1680, + "valid_targets_mean": 2796.4, + "valid_targets_min": 884 + }, + { + "epoch": 2.775947281713344, + "grad_norm": 0.5970044349725444, + "learning_rate": 3.0222095918475873e-05, + "loss": 0.2771, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2813631594181061, + "step": 1685, + "valid_targets_mean": 3382.0, + "valid_targets_min": 303 + }, + { + "epoch": 2.784184514003295, + "grad_norm": 0.6241679803491641, + "learning_rate": 3.0151396468588848e-05, + "loss": 0.2588, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3226885199546814, + "step": 1690, + "valid_targets_mean": 3349.4, + "valid_targets_min": 629 + }, + { + "epoch": 2.7924217462932455, + "grad_norm": 0.6758678539427506, + "learning_rate": 3.0080525729823306e-05, + "loss": 0.256, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2838185131549835, + "step": 1695, + "valid_targets_mean": 2828.4, + "valid_targets_min": 414 + }, + { + "epoch": 2.800658978583196, + "grad_norm": 0.6626285467688362, + "learning_rate": 3.0009484898011707e-05, + "loss": 0.2682, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20693442225456238, + "step": 1700, + "valid_targets_mean": 2387.9, + "valid_targets_min": 330 + }, + { + "epoch": 2.808896210873147, + "grad_norm": 0.6348234491622945, + "learning_rate": 2.993827517185656e-05, + "loss": 0.2425, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1983412504196167, + "step": 1705, + "valid_targets_mean": 2287.1, + "valid_targets_min": 344 + }, + { + "epoch": 2.8171334431630974, + "grad_norm": 0.61643554024194, + "learning_rate": 2.9866897752910194e-05, + "loss": 0.2511, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.283508837223053, + "step": 1710, + "valid_targets_mean": 3377.9, + "valid_targets_min": 296 + }, + { + "epoch": 2.8253706754530477, + "grad_norm": 0.6199082976797735, + "learning_rate": 2.979535384555449e-05, + "loss": 0.2445, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2590184211730957, + "step": 1715, + "valid_targets_mean": 2785.8, + "valid_targets_min": 525 + }, + { + "epoch": 2.8336079077429983, + "grad_norm": 0.6125979185177308, + "learning_rate": 2.972364465698058e-05, + "loss": 0.2517, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20899072289466858, + "step": 1720, + "valid_targets_mean": 2579.9, + "valid_targets_min": 1082 + }, + { + "epoch": 2.841845140032949, + "grad_norm": 0.5290588033442578, + "learning_rate": 2.9651771397168443e-05, + "loss": 0.2386, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22369131445884705, + "step": 1725, + "valid_targets_mean": 3426.8, + "valid_targets_min": 293 + }, + { + "epoch": 2.8500823723228996, + "grad_norm": 0.6539333797819626, + "learning_rate": 2.95797352788665e-05, + "loss": 0.2378, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23585286736488342, + "step": 1730, + "valid_targets_mean": 2215.5, + "valid_targets_min": 995 + }, + { + "epoch": 2.8583196046128503, + "grad_norm": 0.5667125641727214, + "learning_rate": 2.9507537517571144e-05, + "loss": 0.2524, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26343971490859985, + "step": 1735, + "valid_targets_mean": 3617.1, + "valid_targets_min": 698 + }, + { + "epoch": 2.8665568369028005, + "grad_norm": 0.6487727126984469, + "learning_rate": 2.943517933150627e-05, + "loss": 0.23, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18966974318027496, + "step": 1740, + "valid_targets_mean": 2072.4, + "valid_targets_min": 620 + }, + { + "epoch": 2.874794069192751, + "grad_norm": 0.6607579482836561, + "learning_rate": 2.936266194160266e-05, + "loss": 0.2776, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2583504319190979, + "step": 1745, + "valid_targets_mean": 2565.9, + "valid_targets_min": 271 + }, + { + "epoch": 2.883031301482702, + "grad_norm": 0.5329687913430899, + "learning_rate": 2.9289986571477437e-05, + "loss": 0.2192, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20519870519638062, + "step": 1750, + "valid_targets_mean": 3771.8, + "valid_targets_min": 287 + }, + { + "epoch": 2.8912685337726525, + "grad_norm": 0.6309469478492378, + "learning_rate": 2.9217154447413353e-05, + "loss": 0.2593, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2913402020931244, + "step": 1755, + "valid_targets_mean": 3011.4, + "valid_targets_min": 440 + }, + { + "epoch": 2.899505766062603, + "grad_norm": 0.6181431157029986, + "learning_rate": 2.9144166798338183e-05, + "loss": 0.2763, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2582964301109314, + "step": 1760, + "valid_targets_mean": 2926.4, + "valid_targets_min": 904 + }, + { + "epoch": 2.9077429983525533, + "grad_norm": 0.561800631902138, + "learning_rate": 2.9071024855803915e-05, + "loss": 0.2283, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24287906289100647, + "step": 1765, + "valid_targets_mean": 3191.9, + "valid_targets_min": 725 + }, + { + "epoch": 2.915980230642504, + "grad_norm": 0.5450888844326778, + "learning_rate": 2.899772985396599e-05, + "loss": 0.2641, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.29268404841423035, + "step": 1770, + "valid_targets_mean": 4125.9, + "valid_targets_min": 219 + }, + { + "epoch": 2.9242174629324547, + "grad_norm": 0.6136775430077496, + "learning_rate": 2.8924283029562507e-05, + "loss": 0.2618, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2912731170654297, + "step": 1775, + "valid_targets_mean": 2938.9, + "valid_targets_min": 696 + }, + { + "epoch": 2.9324546952224053, + "grad_norm": 0.5978265935100938, + "learning_rate": 2.8850685621893325e-05, + "loss": 0.2561, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.252366840839386, + "step": 1780, + "valid_targets_mean": 2998.1, + "valid_targets_min": 1471 + }, + { + "epoch": 2.940691927512356, + "grad_norm": 0.7550055402423995, + "learning_rate": 2.8776938872799154e-05, + "loss": 0.2482, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3611225187778473, + "step": 1785, + "valid_targets_mean": 2618.1, + "valid_targets_min": 425 + }, + { + "epoch": 2.948929159802306, + "grad_norm": 0.7818665512831672, + "learning_rate": 2.8703044026640606e-05, + "loss": 0.2418, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23194675147533417, + "step": 1790, + "valid_targets_mean": 2316.6, + "valid_targets_min": 737 + }, + { + "epoch": 2.957166392092257, + "grad_norm": 0.5892658804250798, + "learning_rate": 2.8629002330277193e-05, + "loss": 0.2588, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2277798354625702, + "step": 1795, + "valid_targets_mean": 3049.9, + "valid_targets_min": 1228 + }, + { + "epoch": 2.9654036243822075, + "grad_norm": 0.5471885661828383, + "learning_rate": 2.8554815033046314e-05, + "loss": 0.2685, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2878086566925049, + "step": 1800, + "valid_targets_mean": 3484.8, + "valid_targets_min": 711 + }, + { + "epoch": 2.973640856672158, + "grad_norm": 0.6505108723919814, + "learning_rate": 2.8480483386742138e-05, + "loss": 0.2766, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3757530748844147, + "step": 1805, + "valid_targets_mean": 3585.3, + "valid_targets_min": 685 + }, + { + "epoch": 2.981878088962109, + "grad_norm": 0.7103264195165173, + "learning_rate": 2.8406008645594503e-05, + "loss": 0.2765, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2637771964073181, + "step": 1810, + "valid_targets_mean": 2895.6, + "valid_targets_min": 658 + }, + { + "epoch": 2.990115321252059, + "grad_norm": 0.5946550823036287, + "learning_rate": 2.8331392066247737e-05, + "loss": 0.2638, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.252275288105011, + "step": 1815, + "valid_targets_mean": 3185.4, + "valid_targets_min": 917 + }, + { + "epoch": 2.99835255354201, + "grad_norm": 0.6152199859425457, + "learning_rate": 2.825663490773949e-05, + "loss": 0.2372, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2030968815088272, + "step": 1820, + "valid_targets_mean": 2650.6, + "valid_targets_min": 1287 + }, + { + "epoch": 3.0065897858319603, + "grad_norm": 0.6258707256547769, + "learning_rate": 2.8181738431479436e-05, + "loss": 0.2275, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27519214153289795, + "step": 1825, + "valid_targets_mean": 3554.0, + "valid_targets_min": 242 + }, + { + "epoch": 3.014827018121911, + "grad_norm": 0.6979619035300266, + "learning_rate": 2.8106703901228038e-05, + "loss": 0.2217, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24069149792194366, + "step": 1830, + "valid_targets_mean": 2847.0, + "valid_targets_min": 793 + }, + { + "epoch": 3.0230642504118617, + "grad_norm": 0.7103856015735733, + "learning_rate": 2.803153258307519e-05, + "loss": 0.2357, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22712934017181396, + "step": 1835, + "valid_targets_mean": 3009.5, + "valid_targets_min": 1145 + }, + { + "epoch": 3.0313014827018123, + "grad_norm": 0.6606325974128326, + "learning_rate": 2.795622574541887e-05, + "loss": 0.2047, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18506592512130737, + "step": 1840, + "valid_targets_mean": 2418.6, + "valid_targets_min": 265 + }, + { + "epoch": 3.039538714991763, + "grad_norm": 0.6850431486225936, + "learning_rate": 2.7880784658943754e-05, + "loss": 0.2236, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18309485912322998, + "step": 1845, + "valid_targets_mean": 2489.2, + "valid_targets_min": 223 + }, + { + "epoch": 3.047775947281713, + "grad_norm": 0.5538682132728715, + "learning_rate": 2.780521059659972e-05, + "loss": 0.2287, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23113849759101868, + "step": 1850, + "valid_targets_mean": 3999.1, + "valid_targets_min": 202 + }, + { + "epoch": 3.056013179571664, + "grad_norm": 0.6796686047382778, + "learning_rate": 2.7729504833580435e-05, + "loss": 0.2261, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25922173261642456, + "step": 1855, + "valid_targets_mean": 3340.5, + "valid_targets_min": 296 + }, + { + "epoch": 3.0642504118616145, + "grad_norm": 0.6304747169185941, + "learning_rate": 2.7653668647301797e-05, + "loss": 0.2183, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16006682813167572, + "step": 1860, + "valid_targets_mean": 2294.0, + "valid_targets_min": 810 + }, + { + "epoch": 3.072487644151565, + "grad_norm": 0.7093598321151464, + "learning_rate": 2.7577703317380396e-05, + "loss": 0.2295, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18509194254875183, + "step": 1865, + "valid_targets_mean": 2160.2, + "valid_targets_min": 630 + }, + { + "epoch": 3.080724876441516, + "grad_norm": 0.7844752556313203, + "learning_rate": 2.7501610125611908e-05, + "loss": 0.2181, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27326297760009766, + "step": 1870, + "valid_targets_mean": 2360.1, + "valid_targets_min": 640 + }, + { + "epoch": 3.088962108731466, + "grad_norm": 0.6203395631329998, + "learning_rate": 2.7425390355949485e-05, + "loss": 0.2047, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20063695311546326, + "step": 1875, + "valid_targets_mean": 3214.6, + "valid_targets_min": 803 + }, + { + "epoch": 3.0971993410214167, + "grad_norm": 0.5972430112657969, + "learning_rate": 2.7349045294482085e-05, + "loss": 0.2151, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1896412968635559, + "step": 1880, + "valid_targets_mean": 3004.9, + "valid_targets_min": 858 + }, + { + "epoch": 3.1054365733113674, + "grad_norm": 0.6109690382269436, + "learning_rate": 2.727257622941276e-05, + "loss": 0.2056, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1823769062757492, + "step": 1885, + "valid_targets_mean": 3246.3, + "valid_targets_min": 869 + }, + { + "epoch": 3.113673805601318, + "grad_norm": 0.6680979030364619, + "learning_rate": 2.719598445103693e-05, + "loss": 0.2384, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30736616253852844, + "step": 1890, + "valid_targets_mean": 3581.5, + "valid_targets_min": 891 + }, + { + "epoch": 3.1219110378912687, + "grad_norm": 0.656674091764851, + "learning_rate": 2.7119271251720628e-05, + "loss": 0.2359, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24723771214485168, + "step": 1895, + "valid_targets_mean": 3369.2, + "valid_targets_min": 281 + }, + { + "epoch": 3.130148270181219, + "grad_norm": 0.7728854258315987, + "learning_rate": 2.7042437925878638e-05, + "loss": 0.2581, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2554164528846741, + "step": 1900, + "valid_targets_mean": 2117.9, + "valid_targets_min": 347 + }, + { + "epoch": 3.1383855024711695, + "grad_norm": 0.6642247085562812, + "learning_rate": 2.696548576995273e-05, + "loss": 0.2209, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1960965394973755, + "step": 1905, + "valid_targets_mean": 2602.1, + "valid_targets_min": 315 + }, + { + "epoch": 3.14662273476112, + "grad_norm": 0.6414494783887725, + "learning_rate": 2.688841608238971e-05, + "loss": 0.2409, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22181352972984314, + "step": 1910, + "valid_targets_mean": 3207.8, + "valid_targets_min": 372 + }, + { + "epoch": 3.154859967051071, + "grad_norm": 0.6164066662964309, + "learning_rate": 2.6811230163619573e-05, + "loss": 0.2032, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2142753303050995, + "step": 1915, + "valid_targets_mean": 3368.1, + "valid_targets_min": 508 + }, + { + "epoch": 3.1630971993410215, + "grad_norm": 0.6512147028551658, + "learning_rate": 2.6733929316033516e-05, + "loss": 0.2265, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2733863592147827, + "step": 1920, + "valid_targets_mean": 3103.7, + "valid_targets_min": 696 + }, + { + "epoch": 3.171334431630972, + "grad_norm": 0.7299868596833275, + "learning_rate": 2.6656514843961995e-05, + "loss": 0.2256, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24958714842796326, + "step": 1925, + "valid_targets_mean": 2279.0, + "valid_targets_min": 330 + }, + { + "epoch": 3.1795716639209224, + "grad_norm": 0.8142044039966199, + "learning_rate": 2.6578988053652674e-05, + "loss": 0.2191, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18350908160209656, + "step": 1930, + "valid_targets_mean": 2545.4, + "valid_targets_min": 311 + }, + { + "epoch": 3.187808896210873, + "grad_norm": 0.562898892452807, + "learning_rate": 2.650135025324845e-05, + "loss": 0.1971, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17142733931541443, + "step": 1935, + "valid_targets_mean": 3330.9, + "valid_targets_min": 1394 + }, + { + "epoch": 3.1960461285008237, + "grad_norm": 0.6582097758062362, + "learning_rate": 2.6423602752765313e-05, + "loss": 0.2304, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20524603128433228, + "step": 1940, + "valid_targets_mean": 2706.2, + "valid_targets_min": 762 + }, + { + "epoch": 3.2042833607907744, + "grad_norm": 0.6657980269588102, + "learning_rate": 2.6345746864070266e-05, + "loss": 0.2319, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18109837174415588, + "step": 1945, + "valid_targets_mean": 3052.4, + "valid_targets_min": 1259 + }, + { + "epoch": 3.212520593080725, + "grad_norm": 0.5233060688239787, + "learning_rate": 2.626778390085921e-05, + "loss": 0.2132, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23430457711219788, + "step": 1950, + "valid_targets_mean": 3784.8, + "valid_targets_min": 1133 + }, + { + "epoch": 3.2207578253706757, + "grad_norm": 0.5640310501280067, + "learning_rate": 2.618971517863475e-05, + "loss": 0.2298, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2508048415184021, + "step": 1955, + "valid_targets_mean": 3848.9, + "valid_targets_min": 829 + }, + { + "epoch": 3.228995057660626, + "grad_norm": 0.6757649235613026, + "learning_rate": 2.611154201468402e-05, + "loss": 0.2048, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19972142577171326, + "step": 1960, + "valid_targets_mean": 2526.3, + "valid_targets_min": 287 + }, + { + "epoch": 3.2372322899505765, + "grad_norm": 0.6327380969439266, + "learning_rate": 2.6033265728056437e-05, + "loss": 0.2411, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27250388264656067, + "step": 1965, + "valid_targets_mean": 3225.0, + "valid_targets_min": 162 + }, + { + "epoch": 3.245469522240527, + "grad_norm": 0.6804914744524357, + "learning_rate": 2.5954887639541435e-05, + "loss": 0.2172, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24783380329608917, + "step": 1970, + "valid_targets_mean": 3115.6, + "valid_targets_min": 1233 + }, + { + "epoch": 3.253706754530478, + "grad_norm": 0.6893147997495879, + "learning_rate": 2.5876409071646224e-05, + "loss": 0.21, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19912023842334747, + "step": 1975, + "valid_targets_mean": 2359.8, + "valid_targets_min": 704 + }, + { + "epoch": 3.2619439868204285, + "grad_norm": 0.7483268034445905, + "learning_rate": 2.5797831348573425e-05, + "loss": 0.2175, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24190068244934082, + "step": 1980, + "valid_targets_mean": 2175.4, + "valid_targets_min": 287 + }, + { + "epoch": 3.2701812191103787, + "grad_norm": 0.7575224906224661, + "learning_rate": 2.571915579619876e-05, + "loss": 0.2346, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20588669180870056, + "step": 1985, + "valid_targets_mean": 2110.9, + "valid_targets_min": 459 + }, + { + "epoch": 3.2784184514003294, + "grad_norm": 0.7392556685136732, + "learning_rate": 2.5640383742048654e-05, + "loss": 0.2322, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2323339730501175, + "step": 1990, + "valid_targets_mean": 3014.4, + "valid_targets_min": 617 + }, + { + "epoch": 3.28665568369028, + "grad_norm": 0.6831064439961768, + "learning_rate": 2.5561516515277863e-05, + "loss": 0.2233, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21883323788642883, + "step": 1995, + "valid_targets_mean": 2510.6, + "valid_targets_min": 302 + }, + { + "epoch": 3.2948929159802307, + "grad_norm": 0.8007206900810597, + "learning_rate": 2.5482555446647018e-05, + "loss": 0.2191, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22404079139232635, + "step": 2000, + "valid_targets_mean": 2083.4, + "valid_targets_min": 246 + }, + { + "epoch": 3.3031301482701814, + "grad_norm": 0.5429592839510521, + "learning_rate": 2.5403501868500185e-05, + "loss": 0.2114, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2400265783071518, + "step": 2005, + "valid_targets_mean": 4130.8, + "valid_targets_min": 1160 + }, + { + "epoch": 3.3113673805601316, + "grad_norm": 0.5660803486075174, + "learning_rate": 2.5324357114742397e-05, + "loss": 0.2144, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2689647376537323, + "step": 2010, + "valid_targets_mean": 3778.0, + "valid_targets_min": 1193 + }, + { + "epoch": 3.3196046128500822, + "grad_norm": 0.7767761209530037, + "learning_rate": 2.5245122520817117e-05, + "loss": 0.2081, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2215704619884491, + "step": 2015, + "valid_targets_mean": 2003.6, + "valid_targets_min": 1036 + }, + { + "epoch": 3.327841845140033, + "grad_norm": 0.6855725876862521, + "learning_rate": 2.516579942368373e-05, + "loss": 0.2346, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2817282974720001, + "step": 2020, + "valid_targets_mean": 2737.0, + "valid_targets_min": 888 + }, + { + "epoch": 3.3360790774299836, + "grad_norm": 0.6814741906770814, + "learning_rate": 2.508638916179497e-05, + "loss": 0.2064, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21324288845062256, + "step": 2025, + "valid_targets_mean": 2609.9, + "valid_targets_min": 702 + }, + { + "epoch": 3.344316309719934, + "grad_norm": 0.6327378315223203, + "learning_rate": 2.5006893075074326e-05, + "loss": 0.2102, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19734251499176025, + "step": 2030, + "valid_targets_mean": 2890.1, + "valid_targets_min": 1511 + }, + { + "epoch": 3.352553542009885, + "grad_norm": 0.6132165565742003, + "learning_rate": 2.4927312504893467e-05, + "loss": 0.2318, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18627655506134033, + "step": 2035, + "valid_targets_mean": 2605.4, + "valid_targets_min": 1161 + }, + { + "epoch": 3.360790774299835, + "grad_norm": 0.5659647262055583, + "learning_rate": 2.4847648794049587e-05, + "loss": 0.2332, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27079278230667114, + "step": 2040, + "valid_targets_mean": 3818.2, + "valid_targets_min": 307 + }, + { + "epoch": 3.3690280065897857, + "grad_norm": 0.7525363828497744, + "learning_rate": 2.4767903286742732e-05, + "loss": 0.2275, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2642737030982971, + "step": 2045, + "valid_targets_mean": 2758.6, + "valid_targets_min": 915 + }, + { + "epoch": 3.3772652388797364, + "grad_norm": 0.643032011061072, + "learning_rate": 2.468807732855314e-05, + "loss": 0.2318, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2521781921386719, + "step": 2050, + "valid_targets_mean": 2820.3, + "valid_targets_min": 869 + }, + { + "epoch": 3.385502471169687, + "grad_norm": 0.5925300737235305, + "learning_rate": 2.4608172266418543e-05, + "loss": 0.2179, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20423221588134766, + "step": 2055, + "valid_targets_mean": 3420.7, + "valid_targets_min": 1010 + }, + { + "epoch": 3.3937397034596377, + "grad_norm": 0.6235488121604282, + "learning_rate": 2.4528189448611417e-05, + "loss": 0.2131, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18752437829971313, + "step": 2060, + "valid_targets_mean": 2721.8, + "valid_targets_min": 592 + }, + { + "epoch": 3.4019769357495884, + "grad_norm": 0.6870040109780278, + "learning_rate": 2.4448130224716248e-05, + "loss": 0.238, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22506539523601532, + "step": 2065, + "valid_targets_mean": 2590.9, + "valid_targets_min": 264 + }, + { + "epoch": 3.4102141680395386, + "grad_norm": 0.7094267432239266, + "learning_rate": 2.436799594560675e-05, + "loss": 0.2211, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2151351273059845, + "step": 2070, + "valid_targets_mean": 2135.9, + "valid_targets_min": 556 + }, + { + "epoch": 3.4184514003294892, + "grad_norm": 0.6314415541358995, + "learning_rate": 2.4287787963423076e-05, + "loss": 0.2078, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21119928359985352, + "step": 2075, + "valid_targets_mean": 2751.8, + "valid_targets_min": 964 + }, + { + "epoch": 3.42668863261944, + "grad_norm": 0.6546436274570712, + "learning_rate": 2.4207507631549013e-05, + "loss": 0.272, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2236832231283188, + "step": 2080, + "valid_targets_mean": 2722.8, + "valid_targets_min": 438 + }, + { + "epoch": 3.4349258649093906, + "grad_norm": 0.7372518274682155, + "learning_rate": 2.412715630458912e-05, + "loss": 0.2134, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20843760669231415, + "step": 2085, + "valid_targets_mean": 1994.7, + "valid_targets_min": 289 + }, + { + "epoch": 3.443163097199341, + "grad_norm": 0.608006914662658, + "learning_rate": 2.4046735338345895e-05, + "loss": 0.1987, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18041086196899414, + "step": 2090, + "valid_targets_mean": 3166.0, + "valid_targets_min": 252 + }, + { + "epoch": 3.4514003294892914, + "grad_norm": 0.6151046163508607, + "learning_rate": 2.396624608979689e-05, + "loss": 0.205, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19882315397262573, + "step": 2095, + "valid_targets_mean": 3122.8, + "valid_targets_min": 565 + }, + { + "epoch": 3.459637561779242, + "grad_norm": 0.6562773996483675, + "learning_rate": 2.3885689917071822e-05, + "loss": 0.2141, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.235906183719635, + "step": 2100, + "valid_targets_mean": 2765.6, + "valid_targets_min": 686 + }, + { + "epoch": 3.4678747940691927, + "grad_norm": 0.6592007545756166, + "learning_rate": 2.3805068179429628e-05, + "loss": 0.2236, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1776544600725174, + "step": 2105, + "valid_targets_mean": 2520.6, + "valid_targets_min": 1343 + }, + { + "epoch": 3.4761120263591434, + "grad_norm": 0.6590580088665366, + "learning_rate": 2.372438223723555e-05, + "loss": 0.2463, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21588583290576935, + "step": 2110, + "valid_targets_mean": 2554.1, + "valid_targets_min": 380 + }, + { + "epoch": 3.484349258649094, + "grad_norm": 0.8076922098904575, + "learning_rate": 2.3643633451938216e-05, + "loss": 0.2256, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25158339738845825, + "step": 2115, + "valid_targets_mean": 2090.4, + "valid_targets_min": 303 + }, + { + "epoch": 3.4925864909390443, + "grad_norm": 0.6332341820976258, + "learning_rate": 2.35628231860466e-05, + "loss": 0.2277, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24031797051429749, + "step": 2120, + "valid_targets_mean": 2920.2, + "valid_targets_min": 354 + }, + { + "epoch": 3.500823723228995, + "grad_norm": 0.577407770793513, + "learning_rate": 2.3481952803107067e-05, + "loss": 0.2218, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22233733534812927, + "step": 2125, + "valid_targets_mean": 3850.7, + "valid_targets_min": 1515 + }, + { + "epoch": 3.5090609555189456, + "grad_norm": 0.6567622204640255, + "learning_rate": 2.3401023667680373e-05, + "loss": 0.2168, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19248053431510925, + "step": 2130, + "valid_targets_mean": 2304.1, + "valid_targets_min": 237 + }, + { + "epoch": 3.5172981878088962, + "grad_norm": 0.7141705443386381, + "learning_rate": 2.332003714531862e-05, + "loss": 0.2327, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1976984441280365, + "step": 2135, + "valid_targets_mean": 2048.7, + "valid_targets_min": 694 + }, + { + "epoch": 3.525535420098847, + "grad_norm": 0.607697425345425, + "learning_rate": 2.3238994602542246e-05, + "loss": 0.2115, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21744032204151154, + "step": 2140, + "valid_targets_mean": 3213.8, + "valid_targets_min": 413 + }, + { + "epoch": 3.533772652388797, + "grad_norm": 0.7021370663152642, + "learning_rate": 2.3157897406816917e-05, + "loss": 0.2237, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28610166907310486, + "step": 2145, + "valid_targets_mean": 2821.2, + "valid_targets_min": 230 + }, + { + "epoch": 3.5420098846787478, + "grad_norm": 0.5461733110954154, + "learning_rate": 2.3076746926530497e-05, + "loss": 0.2151, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2287541776895523, + "step": 2150, + "valid_targets_mean": 3553.9, + "valid_targets_min": 650 + }, + { + "epoch": 3.5502471169686984, + "grad_norm": 0.5850319326104395, + "learning_rate": 2.2995544530969946e-05, + "loss": 0.22, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22641529142856598, + "step": 2155, + "valid_targets_mean": 3302.9, + "valid_targets_min": 781 + }, + { + "epoch": 3.558484349258649, + "grad_norm": 0.582193722764864, + "learning_rate": 2.2914291590298197e-05, + "loss": 0.2079, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1886036992073059, + "step": 2160, + "valid_targets_mean": 2990.2, + "valid_targets_min": 1004 + }, + { + "epoch": 3.5667215815485998, + "grad_norm": 0.6473717041137795, + "learning_rate": 2.2832989475531073e-05, + "loss": 0.2245, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25963377952575684, + "step": 2165, + "valid_targets_mean": 3190.9, + "valid_targets_min": 769 + }, + { + "epoch": 3.5749588138385504, + "grad_norm": 0.6615310004983426, + "learning_rate": 2.2751639558514116e-05, + "loss": 0.2213, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2162645161151886, + "step": 2170, + "valid_targets_mean": 2861.2, + "valid_targets_min": 613 + }, + { + "epoch": 3.583196046128501, + "grad_norm": 0.6338711397955663, + "learning_rate": 2.267024321189947e-05, + "loss": 0.2201, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.251259446144104, + "step": 2175, + "valid_targets_mean": 3085.6, + "valid_targets_min": 941 + }, + { + "epoch": 3.5914332784184513, + "grad_norm": 0.783642027585893, + "learning_rate": 2.2588801809122692e-05, + "loss": 0.2052, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2465619146823883, + "step": 2180, + "valid_targets_mean": 2471.6, + "valid_targets_min": 1463 + }, + { + "epoch": 3.599670510708402, + "grad_norm": 0.586321202432999, + "learning_rate": 2.2507316724379603e-05, + "loss": 0.215, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1637313961982727, + "step": 2185, + "valid_targets_mean": 3156.4, + "valid_targets_min": 910 + }, + { + "epoch": 3.6079077429983526, + "grad_norm": 0.6379027755615722, + "learning_rate": 2.2425789332603073e-05, + "loss": 0.2315, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23997169733047485, + "step": 2190, + "valid_targets_mean": 2858.4, + "valid_targets_min": 847 + }, + { + "epoch": 3.6161449752883033, + "grad_norm": 0.6047690528170985, + "learning_rate": 2.2344221009439863e-05, + "loss": 0.2249, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24346941709518433, + "step": 2195, + "valid_targets_mean": 3866.5, + "valid_targets_min": 813 + }, + { + "epoch": 3.624382207578254, + "grad_norm": 0.6313067950028487, + "learning_rate": 2.2262613131227357e-05, + "loss": 0.2205, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22747395932674408, + "step": 2200, + "valid_targets_mean": 3051.4, + "valid_targets_min": 940 + }, + { + "epoch": 3.632619439868204, + "grad_norm": 0.6656343358447956, + "learning_rate": 2.2180967074970384e-05, + "loss": 0.2021, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19725120067596436, + "step": 2205, + "valid_targets_mean": 2394.8, + "valid_targets_min": 1078 + }, + { + "epoch": 3.640856672158155, + "grad_norm": 0.7130890060910889, + "learning_rate": 2.2099284218317965e-05, + "loss": 0.2356, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25655680894851685, + "step": 2210, + "valid_targets_mean": 3127.7, + "valid_targets_min": 1049 + }, + { + "epoch": 3.6490939044481054, + "grad_norm": 0.610819423734245, + "learning_rate": 2.201756593954007e-05, + "loss": 0.2174, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23688563704490662, + "step": 2215, + "valid_targets_mean": 2912.0, + "valid_targets_min": 591 + }, + { + "epoch": 3.657331136738056, + "grad_norm": 0.6314338546235789, + "learning_rate": 2.1935813617504372e-05, + "loss": 0.2894, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23836040496826172, + "step": 2220, + "valid_targets_mean": 3433.9, + "valid_targets_min": 1226 + }, + { + "epoch": 3.6655683690280068, + "grad_norm": 0.6913445443394045, + "learning_rate": 2.1854028631652958e-05, + "loss": 0.2016, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20062170922756195, + "step": 2225, + "valid_targets_mean": 3753.8, + "valid_targets_min": 1806 + }, + { + "epoch": 3.673805601317957, + "grad_norm": 0.6508619258749753, + "learning_rate": 2.1772212361979053e-05, + "loss": 0.224, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2572256624698639, + "step": 2230, + "valid_targets_mean": 2761.0, + "valid_targets_min": 261 + }, + { + "epoch": 3.6820428336079076, + "grad_norm": 0.5753873608912875, + "learning_rate": 2.1690366189003776e-05, + "loss": 0.231, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2219148576259613, + "step": 2235, + "valid_targets_mean": 3162.1, + "valid_targets_min": 871 + }, + { + "epoch": 3.6902800658978583, + "grad_norm": 0.6527308666941445, + "learning_rate": 2.16084914937528e-05, + "loss": 0.2263, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24218809604644775, + "step": 2240, + "valid_targets_mean": 3107.4, + "valid_targets_min": 1077 + }, + { + "epoch": 3.698517298187809, + "grad_norm": 1.124267119734622, + "learning_rate": 2.1526589657733076e-05, + "loss": 0.2477, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22716014087200165, + "step": 2245, + "valid_targets_mean": 2182.6, + "valid_targets_min": 245 + }, + { + "epoch": 3.7067545304777596, + "grad_norm": 0.5693915019753482, + "learning_rate": 2.1444662062909497e-05, + "loss": 0.2249, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18227519094944, + "step": 2250, + "valid_targets_mean": 3361.0, + "valid_targets_min": 377 + }, + { + "epoch": 3.71499176276771, + "grad_norm": 0.6891371173963926, + "learning_rate": 2.136271009168162e-05, + "loss": 0.212, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1876133233308792, + "step": 2255, + "valid_targets_mean": 2128.9, + "valid_targets_min": 719 + }, + { + "epoch": 3.7232289950576605, + "grad_norm": 0.6567207894440227, + "learning_rate": 2.128073512686029e-05, + "loss": 0.208, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2048790454864502, + "step": 2260, + "valid_targets_mean": 2656.5, + "valid_targets_min": 1183 + }, + { + "epoch": 3.731466227347611, + "grad_norm": 0.7458147326733252, + "learning_rate": 2.1198738551644344e-05, + "loss": 0.2398, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2422378957271576, + "step": 2265, + "valid_targets_mean": 2192.9, + "valid_targets_min": 251 + }, + { + "epoch": 3.739703459637562, + "grad_norm": 0.6442647879430062, + "learning_rate": 2.1116721749597265e-05, + "loss": 0.2049, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20766223967075348, + "step": 2270, + "valid_targets_mean": 2699.9, + "valid_targets_min": 251 + }, + { + "epoch": 3.7479406919275124, + "grad_norm": 0.6516890191749407, + "learning_rate": 2.1034686104623828e-05, + "loss": 0.2082, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24694222211837769, + "step": 2275, + "valid_targets_mean": 2857.2, + "valid_targets_min": 1068 + }, + { + "epoch": 3.7561779242174627, + "grad_norm": 0.7329917353431759, + "learning_rate": 2.0952633000946753e-05, + "loss": 0.2239, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23302693665027618, + "step": 2280, + "valid_targets_mean": 2236.2, + "valid_targets_min": 251 + }, + { + "epoch": 3.7644151565074138, + "grad_norm": 0.7031038589727836, + "learning_rate": 2.0870563823083347e-05, + "loss": 0.2235, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21459734439849854, + "step": 2285, + "valid_targets_mean": 2116.3, + "valid_targets_min": 683 + }, + { + "epoch": 3.772652388797364, + "grad_norm": 0.6857035428769572, + "learning_rate": 2.0788479955822137e-05, + "loss": 0.2095, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20176962018013, + "step": 2290, + "valid_targets_mean": 2616.1, + "valid_targets_min": 1007 + }, + { + "epoch": 3.7808896210873146, + "grad_norm": 0.8073211974656119, + "learning_rate": 2.0706382784199524e-05, + "loss": 0.2152, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27309340238571167, + "step": 2295, + "valid_targets_mean": 2650.2, + "valid_targets_min": 753 + }, + { + "epoch": 3.7891268533772653, + "grad_norm": 0.7210086182232971, + "learning_rate": 2.06242736934764e-05, + "loss": 0.207, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21143299341201782, + "step": 2300, + "valid_targets_mean": 2429.8, + "valid_targets_min": 693 + }, + { + "epoch": 3.797364085667216, + "grad_norm": 0.6610150115701522, + "learning_rate": 2.0542154069114753e-05, + "loss": 0.2355, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20834533870220184, + "step": 2305, + "valid_targets_mean": 2753.0, + "valid_targets_min": 317 + }, + { + "epoch": 3.8056013179571666, + "grad_norm": 0.8143475265522775, + "learning_rate": 2.0460025296754326e-05, + "loss": 0.2336, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2677311897277832, + "step": 2310, + "valid_targets_mean": 1949.5, + "valid_targets_min": 311 + }, + { + "epoch": 3.813838550247117, + "grad_norm": 0.6674017870698288, + "learning_rate": 2.037788876218923e-05, + "loss": 0.215, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17726172506809235, + "step": 2315, + "valid_targets_mean": 2284.8, + "valid_targets_min": 324 + }, + { + "epoch": 3.8220757825370675, + "grad_norm": 0.5838081717362156, + "learning_rate": 2.029574585134452e-05, + "loss": 0.2059, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22934329509735107, + "step": 2320, + "valid_targets_mean": 3687.3, + "valid_targets_min": 791 + }, + { + "epoch": 3.830313014827018, + "grad_norm": 0.6778643643059814, + "learning_rate": 2.0213597950252864e-05, + "loss": 0.2087, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21608680486679077, + "step": 2325, + "valid_targets_mean": 2646.7, + "valid_targets_min": 374 + }, + { + "epoch": 3.838550247116969, + "grad_norm": 0.5949453895813899, + "learning_rate": 2.0131446445031132e-05, + "loss": 0.217, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23367094993591309, + "step": 2330, + "valid_targets_mean": 3113.8, + "valid_targets_min": 1029 + }, + { + "epoch": 3.8467874794069195, + "grad_norm": 0.6215489423579595, + "learning_rate": 2.0049292721857005e-05, + "loss": 0.2403, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2354554384946823, + "step": 2335, + "valid_targets_mean": 2976.0, + "valid_targets_min": 1576 + }, + { + "epoch": 3.8550247116968697, + "grad_norm": 0.6525560252314341, + "learning_rate": 1.9967138166945584e-05, + "loss": 0.2151, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21559584140777588, + "step": 2340, + "valid_targets_mean": 2901.3, + "valid_targets_min": 1446 + }, + { + "epoch": 3.8632619439868203, + "grad_norm": 0.6584339432357748, + "learning_rate": 1.9884984166525996e-05, + "loss": 0.2242, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2388259917497635, + "step": 2345, + "valid_targets_mean": 2916.9, + "valid_targets_min": 809 + }, + { + "epoch": 3.871499176276771, + "grad_norm": 0.6625292350672669, + "learning_rate": 1.9802832106818038e-05, + "loss": 0.2294, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20597676932811737, + "step": 2350, + "valid_targets_mean": 2825.5, + "valid_targets_min": 425 + }, + { + "epoch": 3.8797364085667216, + "grad_norm": 0.6845618073993892, + "learning_rate": 1.9720683374008742e-05, + "loss": 0.2439, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2833724617958069, + "step": 2355, + "valid_targets_mean": 2759.7, + "valid_targets_min": 811 + }, + { + "epoch": 3.8879736408566723, + "grad_norm": 0.7583414594121382, + "learning_rate": 1.9638539354228994e-05, + "loss": 0.2353, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23047031462192535, + "step": 2360, + "valid_targets_mean": 1862.4, + "valid_targets_min": 728 + }, + { + "epoch": 3.8962108731466225, + "grad_norm": 0.6863919509729713, + "learning_rate": 1.9556401433530186e-05, + "loss": 0.1978, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19445081055164337, + "step": 2365, + "valid_targets_mean": 2312.0, + "valid_targets_min": 236 + }, + { + "epoch": 3.904448105436573, + "grad_norm": 0.5702207511533826, + "learning_rate": 1.947427099786077e-05, + "loss": 0.2032, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1521320939064026, + "step": 2370, + "valid_targets_mean": 2876.6, + "valid_targets_min": 574 + }, + { + "epoch": 3.912685337726524, + "grad_norm": 0.725652791371573, + "learning_rate": 1.9392149433042915e-05, + "loss": 0.222, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21470411121845245, + "step": 2375, + "valid_targets_mean": 2422.2, + "valid_targets_min": 1293 + }, + { + "epoch": 3.9209225700164745, + "grad_norm": 0.7521039314890973, + "learning_rate": 1.9310038124749102e-05, + "loss": 0.2146, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2105662226676941, + "step": 2380, + "valid_targets_mean": 2092.1, + "valid_targets_min": 914 + }, + { + "epoch": 3.929159802306425, + "grad_norm": 0.7221609491167119, + "learning_rate": 1.9227938458478744e-05, + "loss": 0.2032, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1993308663368225, + "step": 2385, + "valid_targets_mean": 2114.9, + "valid_targets_min": 286 + }, + { + "epoch": 3.9373970345963754, + "grad_norm": 0.6765877567980294, + "learning_rate": 1.9145851819534824e-05, + "loss": 0.2167, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19120612740516663, + "step": 2390, + "valid_targets_mean": 2558.6, + "valid_targets_min": 817 + }, + { + "epoch": 3.9456342668863265, + "grad_norm": 0.5868278571600912, + "learning_rate": 1.9063779593000507e-05, + "loss": 0.2148, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1894645094871521, + "step": 2395, + "valid_targets_mean": 3188.8, + "valid_targets_min": 847 + }, + { + "epoch": 3.9538714991762767, + "grad_norm": 0.5382630298422423, + "learning_rate": 1.8981723163715756e-05, + "loss": 0.217, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20977070927619934, + "step": 2400, + "valid_targets_mean": 3654.6, + "valid_targets_min": 708 + }, + { + "epoch": 3.9621087314662273, + "grad_norm": 0.6247881187672814, + "learning_rate": 1.8899683916254013e-05, + "loss": 0.2319, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19764485955238342, + "step": 2405, + "valid_targets_mean": 2881.1, + "valid_targets_min": 849 + }, + { + "epoch": 3.970345963756178, + "grad_norm": 0.7327354122670338, + "learning_rate": 1.881766323489877e-05, + "loss": 0.2108, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23579062521457672, + "step": 2410, + "valid_targets_mean": 2472.6, + "valid_targets_min": 746 + }, + { + "epoch": 3.9785831960461286, + "grad_norm": 0.6595175545075271, + "learning_rate": 1.8735662503620268e-05, + "loss": 0.227, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2350047081708908, + "step": 2415, + "valid_targets_mean": 3148.6, + "valid_targets_min": 309 + }, + { + "epoch": 3.9868204283360793, + "grad_norm": 0.510007753878104, + "learning_rate": 1.8653683106052097e-05, + "loss": 0.2015, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20294642448425293, + "step": 2420, + "valid_targets_mean": 3626.1, + "valid_targets_min": 341 + }, + { + "epoch": 3.9950576606260295, + "grad_norm": 0.6659822075544012, + "learning_rate": 1.857172642546791e-05, + "loss": 0.2659, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24123075604438782, + "step": 2425, + "valid_targets_mean": 2767.6, + "valid_targets_min": 866 + }, + { + "epoch": 4.003294892915981, + "grad_norm": 0.6848919535763092, + "learning_rate": 1.8489793844758015e-05, + "loss": 0.2112, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.196708083152771, + "step": 2430, + "valid_targets_mean": 2348.8, + "valid_targets_min": 780 + }, + { + "epoch": 4.011532125205931, + "grad_norm": 0.8126335054852767, + "learning_rate": 1.8407886746406082e-05, + "loss": 0.196, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18026752769947052, + "step": 2435, + "valid_targets_mean": 1642.7, + "valid_targets_min": 524 + }, + { + "epoch": 4.019769357495881, + "grad_norm": 0.6954560411576156, + "learning_rate": 1.8326006512465806e-05, + "loss": 0.2156, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1572655588388443, + "step": 2440, + "valid_targets_mean": 2657.9, + "valid_targets_min": 304 + }, + { + "epoch": 4.028006589785832, + "grad_norm": 0.6681376975506721, + "learning_rate": 1.8244154524537602e-05, + "loss": 0.2041, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1940767467021942, + "step": 2445, + "valid_targets_mean": 3489.4, + "valid_targets_min": 674 + }, + { + "epoch": 4.036243822075782, + "grad_norm": 0.675763921438596, + "learning_rate": 1.8162332163745253e-05, + "loss": 0.1837, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18810434639453888, + "step": 2450, + "valid_targets_mean": 2845.8, + "valid_targets_min": 245 + }, + { + "epoch": 4.0444810543657335, + "grad_norm": 0.6559161150800585, + "learning_rate": 1.8080540810712652e-05, + "loss": 0.1836, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1783156543970108, + "step": 2455, + "valid_targets_mean": 3050.2, + "valid_targets_min": 747 + }, + { + "epoch": 4.052718286655684, + "grad_norm": 0.7053939038206681, + "learning_rate": 1.799878184554047e-05, + "loss": 0.1702, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18840208649635315, + "step": 2460, + "valid_targets_mean": 2819.3, + "valid_targets_min": 1529 + }, + { + "epoch": 4.060955518945634, + "grad_norm": 0.5768463220547291, + "learning_rate": 1.7917056647782888e-05, + "loss": 0.2008, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2617749571800232, + "step": 2465, + "valid_targets_mean": 4129.8, + "valid_targets_min": 245 + }, + { + "epoch": 4.069192751235585, + "grad_norm": 0.6610629470321063, + "learning_rate": 1.7835366596424317e-05, + "loss": 0.1952, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20493917167186737, + "step": 2470, + "valid_targets_mean": 3070.0, + "valid_targets_min": 1007 + }, + { + "epoch": 4.077429983525535, + "grad_norm": 0.6908004026096404, + "learning_rate": 1.775371306985613e-05, + "loss": 0.2033, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2220386117696762, + "step": 2475, + "valid_targets_mean": 2883.8, + "valid_targets_min": 829 + }, + { + "epoch": 4.085667215815486, + "grad_norm": 0.679048587976391, + "learning_rate": 1.7672097445853388e-05, + "loss": 0.1979, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17979687452316284, + "step": 2480, + "valid_targets_mean": 2942.2, + "valid_targets_min": 1084 + }, + { + "epoch": 4.0939044481054365, + "grad_norm": 0.7517226829701907, + "learning_rate": 1.7590521101551635e-05, + "loss": 0.1936, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20365673303604126, + "step": 2485, + "valid_targets_mean": 2352.6, + "valid_targets_min": 250 + }, + { + "epoch": 4.102141680395387, + "grad_norm": 0.7091041235957187, + "learning_rate": 1.7508985413423595e-05, + "loss": 0.1844, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2280346006155014, + "step": 2490, + "valid_targets_mean": 2677.7, + "valid_targets_min": 1029 + }, + { + "epoch": 4.110378912685338, + "grad_norm": 0.7089396573462521, + "learning_rate": 1.742749175725601e-05, + "loss": 0.1962, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18262189626693726, + "step": 2495, + "valid_targets_mean": 2452.9, + "valid_targets_min": 224 + }, + { + "epoch": 4.118616144975288, + "grad_norm": 0.6892080263313686, + "learning_rate": 1.7346041508126386e-05, + "loss": 0.2154, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1696496605873108, + "step": 2500, + "valid_targets_mean": 2710.0, + "valid_targets_min": 891 + }, + { + "epoch": 4.126853377265239, + "grad_norm": 0.7552024083884684, + "learning_rate": 1.7264636040379808e-05, + "loss": 0.1897, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2135401964187622, + "step": 2505, + "valid_targets_mean": 2567.6, + "valid_targets_min": 307 + }, + { + "epoch": 4.135090609555189, + "grad_norm": 0.6846684162918165, + "learning_rate": 1.7183276727605743e-05, + "loss": 0.1984, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17049439251422882, + "step": 2510, + "valid_targets_mean": 3185.9, + "valid_targets_min": 1214 + }, + { + "epoch": 4.1433278418451405, + "grad_norm": 0.6726148670882198, + "learning_rate": 1.7101964942614864e-05, + "loss": 0.1904, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18531067669391632, + "step": 2515, + "valid_targets_mean": 2673.3, + "valid_targets_min": 241 + }, + { + "epoch": 4.151565074135091, + "grad_norm": 0.7050748175603723, + "learning_rate": 1.7020702057415884e-05, + "loss": 0.2148, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.192526176571846, + "step": 2520, + "valid_targets_mean": 3261.9, + "valid_targets_min": 890 + }, + { + "epoch": 4.159802306425041, + "grad_norm": 0.7166479950826307, + "learning_rate": 1.6939489443192435e-05, + "loss": 0.1992, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17323213815689087, + "step": 2525, + "valid_targets_mean": 2580.0, + "valid_targets_min": 279 + }, + { + "epoch": 4.168039538714992, + "grad_norm": 0.6307866731323056, + "learning_rate": 1.6858328470279862e-05, + "loss": 0.216, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18545058369636536, + "step": 2530, + "valid_targets_mean": 3305.8, + "valid_targets_min": 809 + }, + { + "epoch": 4.176276771004942, + "grad_norm": 0.6262445327354411, + "learning_rate": 1.6777220508142184e-05, + "loss": 0.1893, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2297663688659668, + "step": 2535, + "valid_targets_mean": 3642.7, + "valid_targets_min": 1657 + }, + { + "epoch": 4.184514003294893, + "grad_norm": 1.123430237088534, + "learning_rate": 1.6696166925348924e-05, + "loss": 0.2048, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15774188935756683, + "step": 2540, + "valid_targets_mean": 4065.2, + "valid_targets_min": 696 + }, + { + "epoch": 4.1927512355848435, + "grad_norm": 0.6851282095150327, + "learning_rate": 1.6615169089552038e-05, + "loss": 0.2046, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3130187392234802, + "step": 2545, + "valid_targets_mean": 3510.0, + "valid_targets_min": 346 + }, + { + "epoch": 4.200988467874794, + "grad_norm": 0.6776890152032327, + "learning_rate": 1.6534228367462843e-05, + "loss": 0.1728, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18178880214691162, + "step": 2550, + "valid_targets_mean": 2698.2, + "valid_targets_min": 380 + }, + { + "epoch": 4.209225700164745, + "grad_norm": 0.5858915064486262, + "learning_rate": 1.6453346124828966e-05, + "loss": 0.2023, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18752583861351013, + "step": 2555, + "valid_targets_mean": 4041.8, + "valid_targets_min": 876 + }, + { + "epoch": 4.217462932454695, + "grad_norm": 0.6603867459341165, + "learning_rate": 1.6372523726411253e-05, + "loss": 0.2195, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18429623544216156, + "step": 2560, + "valid_targets_mean": 3103.4, + "valid_targets_min": 817 + }, + { + "epoch": 4.225700164744646, + "grad_norm": 0.6627234827005007, + "learning_rate": 1.6291762535960817e-05, + "loss": 0.1996, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14199604094028473, + "step": 2565, + "valid_targets_mean": 2462.8, + "valid_targets_min": 320 + }, + { + "epoch": 4.233937397034596, + "grad_norm": 0.6553355284722763, + "learning_rate": 1.6211063916195945e-05, + "loss": 0.2317, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1949518620967865, + "step": 2570, + "valid_targets_mean": 3206.4, + "valid_targets_min": 667 + }, + { + "epoch": 4.242174629324547, + "grad_norm": 0.6980909727917753, + "learning_rate": 1.6130429228779154e-05, + "loss": 0.1874, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1674255132675171, + "step": 2575, + "valid_targets_mean": 2631.8, + "valid_targets_min": 817 + }, + { + "epoch": 4.250411861614498, + "grad_norm": 1.0730056513064912, + "learning_rate": 1.604985983429421e-05, + "loss": 0.1904, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18854576349258423, + "step": 2580, + "valid_targets_mean": 3124.4, + "valid_targets_min": 569 + }, + { + "epoch": 4.258649093904448, + "grad_norm": 0.7520222031647412, + "learning_rate": 1.596935709222315e-05, + "loss": 0.2015, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21556442975997925, + "step": 2585, + "valid_targets_mean": 2409.2, + "valid_targets_min": 708 + }, + { + "epoch": 4.266886326194399, + "grad_norm": 0.6583239060045123, + "learning_rate": 1.5888922360923375e-05, + "loss": 0.21, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19209733605384827, + "step": 2590, + "valid_targets_mean": 2888.0, + "valid_targets_min": 658 + }, + { + "epoch": 4.275123558484349, + "grad_norm": 0.643988352278543, + "learning_rate": 1.5808556997604684e-05, + "loss": 0.1972, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1911228597164154, + "step": 2595, + "valid_targets_mean": 3091.2, + "valid_targets_min": 649 + }, + { + "epoch": 4.283360790774299, + "grad_norm": 0.7313653605878269, + "learning_rate": 1.5728262358306418e-05, + "loss": 0.2037, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18181253969669342, + "step": 2600, + "valid_targets_mean": 2933.1, + "valid_targets_min": 866 + }, + { + "epoch": 4.2915980230642505, + "grad_norm": 0.7255332306218216, + "learning_rate": 1.564803979787457e-05, + "loss": 0.1782, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20717540383338928, + "step": 2605, + "valid_targets_mean": 2623.2, + "valid_targets_min": 1101 + }, + { + "epoch": 4.299835255354201, + "grad_norm": 0.7531476679995408, + "learning_rate": 1.5567890669938903e-05, + "loss": 0.1721, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1496737003326416, + "step": 2610, + "valid_targets_mean": 2502.5, + "valid_targets_min": 754 + }, + { + "epoch": 4.308072487644152, + "grad_norm": 0.6487606715317696, + "learning_rate": 1.5487816326890118e-05, + "loss": 0.2199, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19765833020210266, + "step": 2615, + "valid_targets_mean": 2939.9, + "valid_targets_min": 744 + }, + { + "epoch": 4.316309719934102, + "grad_norm": 0.7240518260494907, + "learning_rate": 1.540781811985704e-05, + "loss": 0.1901, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20093931257724762, + "step": 2620, + "valid_targets_mean": 3010.4, + "valid_targets_min": 237 + }, + { + "epoch": 4.324546952224052, + "grad_norm": 0.7571300929723414, + "learning_rate": 1.5327897398683827e-05, + "loss": 0.2019, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22667065262794495, + "step": 2625, + "valid_targets_mean": 2461.3, + "valid_targets_min": 320 + }, + { + "epoch": 4.332784184514003, + "grad_norm": 0.6867670673018871, + "learning_rate": 1.5248055511907183e-05, + "loss": 0.2204, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2281215339899063, + "step": 2630, + "valid_targets_mean": 2981.8, + "valid_targets_min": 340 + }, + { + "epoch": 4.341021416803954, + "grad_norm": 0.8256870657949581, + "learning_rate": 1.5168293806733594e-05, + "loss": 0.199, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1914563775062561, + "step": 2635, + "valid_targets_mean": 1873.8, + "valid_targets_min": 595 + }, + { + "epoch": 4.349258649093905, + "grad_norm": 0.6775685498781048, + "learning_rate": 1.5088613629016613e-05, + "loss": 0.1942, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19783690571784973, + "step": 2640, + "valid_targets_mean": 3498.0, + "valid_targets_min": 585 + }, + { + "epoch": 4.357495881383855, + "grad_norm": 0.6103793140474083, + "learning_rate": 1.5009016323234158e-05, + "loss": 0.1693, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14736308157444, + "step": 2645, + "valid_targets_mean": 3104.6, + "valid_targets_min": 809 + }, + { + "epoch": 4.365733113673806, + "grad_norm": 0.6444833412719732, + "learning_rate": 1.4929503232465804e-05, + "loss": 0.1914, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17908361554145813, + "step": 2650, + "valid_targets_mean": 2714.5, + "valid_targets_min": 896 + }, + { + "epoch": 4.373970345963756, + "grad_norm": 0.6125674016153442, + "learning_rate": 1.485007569837011e-05, + "loss": 0.187, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18148434162139893, + "step": 2655, + "valid_targets_mean": 3145.8, + "valid_targets_min": 324 + }, + { + "epoch": 4.382207578253706, + "grad_norm": 0.6907068990363627, + "learning_rate": 1.477073506116204e-05, + "loss": 0.2107, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2808583080768585, + "step": 2660, + "valid_targets_mean": 3140.5, + "valid_targets_min": 270 + }, + { + "epoch": 4.3904448105436575, + "grad_norm": 0.6513537135662765, + "learning_rate": 1.469148265959028e-05, + "loss": 0.1826, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20921581983566284, + "step": 2665, + "valid_targets_mean": 3432.5, + "valid_targets_min": 1145 + }, + { + "epoch": 4.398682042833608, + "grad_norm": 0.7190243287207262, + "learning_rate": 1.4612319830914684e-05, + "loss": 0.1742, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19826185703277588, + "step": 2670, + "valid_targets_mean": 2480.8, + "valid_targets_min": 959 + }, + { + "epoch": 4.406919275123559, + "grad_norm": 0.6374744860830139, + "learning_rate": 1.4533247910883704e-05, + "loss": 0.189, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17492321133613586, + "step": 2675, + "valid_targets_mean": 3242.9, + "valid_targets_min": 542 + }, + { + "epoch": 4.415156507413509, + "grad_norm": 0.736114349197071, + "learning_rate": 1.445426823371186e-05, + "loss": 0.1918, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15052483975887299, + "step": 2680, + "valid_targets_mean": 2257.1, + "valid_targets_min": 260 + }, + { + "epoch": 4.423393739703459, + "grad_norm": 0.6964652002410009, + "learning_rate": 1.4375382132057207e-05, + "loss": 0.195, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18424688279628754, + "step": 2685, + "valid_targets_mean": 2731.6, + "valid_targets_min": 751 + }, + { + "epoch": 4.43163097199341, + "grad_norm": 0.6646723911201924, + "learning_rate": 1.4296590936998875e-05, + "loss": 0.1962, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1773110330104828, + "step": 2690, + "valid_targets_mean": 2841.5, + "valid_targets_min": 786 + }, + { + "epoch": 4.439868204283361, + "grad_norm": 0.7043596578211233, + "learning_rate": 1.4217895978014583e-05, + "loss": 0.1856, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15232154726982117, + "step": 2695, + "valid_targets_mean": 2430.2, + "valid_targets_min": 738 + }, + { + "epoch": 4.448105436573312, + "grad_norm": 0.8743360493221132, + "learning_rate": 1.4139298582958208e-05, + "loss": 0.2248, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20513039827346802, + "step": 2700, + "valid_targets_mean": 3039.7, + "valid_targets_min": 729 + }, + { + "epoch": 4.456342668863262, + "grad_norm": 0.7367043918610189, + "learning_rate": 1.4060800078037407e-05, + "loss": 0.1997, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24137605726718903, + "step": 2705, + "valid_targets_mean": 3009.7, + "valid_targets_min": 539 + }, + { + "epoch": 4.464579901153212, + "grad_norm": 0.6735512497948377, + "learning_rate": 1.3982401787791215e-05, + "loss": 0.2222, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19957660138607025, + "step": 2710, + "valid_targets_mean": 3105.6, + "valid_targets_min": 1394 + }, + { + "epoch": 4.472817133443163, + "grad_norm": 0.6987608708119356, + "learning_rate": 1.3904105035067701e-05, + "loss": 0.1668, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18072089552879333, + "step": 2715, + "valid_targets_mean": 2651.4, + "valid_targets_min": 864 + }, + { + "epoch": 4.481054365733113, + "grad_norm": 0.718751412393381, + "learning_rate": 1.3825911141001637e-05, + "loss": 0.1764, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16804978251457214, + "step": 2720, + "valid_targets_mean": 2340.1, + "valid_targets_min": 567 + }, + { + "epoch": 4.4892915980230645, + "grad_norm": 0.7064126385360051, + "learning_rate": 1.3747821424992245e-05, + "loss": 0.1798, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2070191204547882, + "step": 2725, + "valid_targets_mean": 2916.1, + "valid_targets_min": 925 + }, + { + "epoch": 4.497528830313015, + "grad_norm": 0.6972537885141782, + "learning_rate": 1.3669837204680879e-05, + "loss": 0.1972, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23842665553092957, + "step": 2730, + "valid_targets_mean": 2726.9, + "valid_targets_min": 638 + }, + { + "epoch": 4.505766062602966, + "grad_norm": 0.67473224143467, + "learning_rate": 1.3591959795928821e-05, + "loss": 0.2145, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2541242241859436, + "step": 2735, + "valid_targets_mean": 3227.6, + "valid_targets_min": 278 + }, + { + "epoch": 4.514003294892916, + "grad_norm": 0.6398227567414478, + "learning_rate": 1.3514190512795096e-05, + "loss": 0.1798, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18463942408561707, + "step": 2740, + "valid_targets_mean": 3369.2, + "valid_targets_min": 768 + }, + { + "epoch": 4.522240527182866, + "grad_norm": 0.914208935807342, + "learning_rate": 1.3436530667514258e-05, + "loss": 0.2184, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25415050983428955, + "step": 2745, + "valid_targets_mean": 2069.6, + "valid_targets_min": 260 + }, + { + "epoch": 4.530477759472817, + "grad_norm": 0.666232807251606, + "learning_rate": 1.3358981570474276e-05, + "loss": 0.1799, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19541533291339874, + "step": 2750, + "valid_targets_mean": 3547.4, + "valid_targets_min": 440 + }, + { + "epoch": 4.538714991762768, + "grad_norm": 1.1834154877573295, + "learning_rate": 1.3281544530194412e-05, + "loss": 0.2104, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22194775938987732, + "step": 2755, + "valid_targets_mean": 2390.3, + "valid_targets_min": 352 + }, + { + "epoch": 4.546952224052719, + "grad_norm": 0.6988385348664411, + "learning_rate": 1.3204220853303157e-05, + "loss": 0.1899, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1807519942522049, + "step": 2760, + "valid_targets_mean": 2474.5, + "valid_targets_min": 930 + }, + { + "epoch": 4.555189456342669, + "grad_norm": 0.6268602624217262, + "learning_rate": 1.3127011844516165e-05, + "loss": 0.1719, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17390185594558716, + "step": 2765, + "valid_targets_mean": 2855.5, + "valid_targets_min": 364 + }, + { + "epoch": 4.563426688632619, + "grad_norm": 0.698300348357958, + "learning_rate": 1.304991880661425e-05, + "loss": 0.1813, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18733061850070953, + "step": 2770, + "valid_targets_mean": 2874.1, + "valid_targets_min": 1214 + }, + { + "epoch": 4.57166392092257, + "grad_norm": 0.904490419197706, + "learning_rate": 1.2972943040421398e-05, + "loss": 0.197, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1777077168226242, + "step": 2775, + "valid_targets_mean": 2739.2, + "valid_targets_min": 1111 + }, + { + "epoch": 4.5799011532125204, + "grad_norm": 0.7738385301230464, + "learning_rate": 1.2896085844782808e-05, + "loss": 0.1908, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19284678995609283, + "step": 2780, + "valid_targets_mean": 2336.1, + "valid_targets_min": 812 + }, + { + "epoch": 4.5881383855024716, + "grad_norm": 1.2378287916149417, + "learning_rate": 1.2819348516543003e-05, + "loss": 0.1621, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1683550775051117, + "step": 2785, + "valid_targets_mean": 2737.6, + "valid_targets_min": 696 + }, + { + "epoch": 4.596375617792422, + "grad_norm": 0.6483004452802794, + "learning_rate": 1.2742732350523928e-05, + "loss": 0.2077, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20438411831855774, + "step": 2790, + "valid_targets_mean": 3092.1, + "valid_targets_min": 356 + }, + { + "epoch": 4.604612850082372, + "grad_norm": 0.6880072116270369, + "learning_rate": 1.2666238639503099e-05, + "loss": 0.1884, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21004796028137207, + "step": 2795, + "valid_targets_mean": 2588.2, + "valid_targets_min": 866 + }, + { + "epoch": 4.612850082372323, + "grad_norm": 0.7996980024277847, + "learning_rate": 1.2589868674191794e-05, + "loss": 0.1857, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19884052872657776, + "step": 2800, + "valid_targets_mean": 2046.0, + "valid_targets_min": 979 + }, + { + "epoch": 4.621087314662273, + "grad_norm": 0.6228896426672722, + "learning_rate": 1.2513623743213297e-05, + "loss": 0.1706, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1847865730524063, + "step": 2805, + "valid_targets_mean": 3417.0, + "valid_targets_min": 252 + }, + { + "epoch": 4.629324546952224, + "grad_norm": 0.714956059080542, + "learning_rate": 1.243750513308111e-05, + "loss": 0.2117, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20548203587532043, + "step": 2810, + "valid_targets_mean": 2715.6, + "valid_targets_min": 594 + }, + { + "epoch": 4.637561779242175, + "grad_norm": 0.746190840563027, + "learning_rate": 1.2361514128177286e-05, + "loss": 0.21, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19804972410202026, + "step": 2815, + "valid_targets_mean": 2636.1, + "valid_targets_min": 711 + }, + { + "epoch": 4.645799011532125, + "grad_norm": 0.6236556080477057, + "learning_rate": 1.2285652010730727e-05, + "loss": 0.2088, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16948853433132172, + "step": 2820, + "valid_targets_mean": 3223.8, + "valid_targets_min": 246 + }, + { + "epoch": 4.654036243822076, + "grad_norm": 0.7487803781790527, + "learning_rate": 1.220992006079557e-05, + "loss": 0.1912, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18032564222812653, + "step": 2825, + "valid_targets_mean": 2885.8, + "valid_targets_min": 1173 + }, + { + "epoch": 4.662273476112026, + "grad_norm": 0.6953634890626381, + "learning_rate": 1.2134319556229577e-05, + "loss": 0.1936, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17370088398456573, + "step": 2830, + "valid_targets_mean": 2609.1, + "valid_targets_min": 636 + }, + { + "epoch": 4.670510708401977, + "grad_norm": 0.6207408692644977, + "learning_rate": 1.2058851772672569e-05, + "loss": 0.1837, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2368783950805664, + "step": 2835, + "valid_targets_mean": 4058.8, + "valid_targets_min": 718 + }, + { + "epoch": 4.6787479406919275, + "grad_norm": 0.7159596910910173, + "learning_rate": 1.1983517983524917e-05, + "loss": 0.2383, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2987104654312134, + "step": 2840, + "valid_targets_mean": 4173.1, + "valid_targets_min": 1281 + }, + { + "epoch": 4.686985172981878, + "grad_norm": 0.7698653286065977, + "learning_rate": 1.1908319459926059e-05, + "loss": 0.2009, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2789364457130432, + "step": 2845, + "valid_targets_mean": 2637.2, + "valid_targets_min": 341 + }, + { + "epoch": 4.695222405271829, + "grad_norm": 0.6885585604727101, + "learning_rate": 1.183325747073301e-05, + "loss": 0.1772, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16655626893043518, + "step": 2850, + "valid_targets_mean": 2233.1, + "valid_targets_min": 762 + }, + { + "epoch": 4.703459637561779, + "grad_norm": 0.7000465338140921, + "learning_rate": 1.1758333282499004e-05, + "loss": 0.179, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18717628717422485, + "step": 2855, + "valid_targets_mean": 2459.1, + "valid_targets_min": 251 + }, + { + "epoch": 4.71169686985173, + "grad_norm": 0.649981251933767, + "learning_rate": 1.1683548159452083e-05, + "loss": 0.185, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19186502695083618, + "step": 2860, + "valid_targets_mean": 3500.6, + "valid_targets_min": 1184 + }, + { + "epoch": 4.71993410214168, + "grad_norm": 0.7074888663516279, + "learning_rate": 1.1608903363473804e-05, + "loss": 0.1636, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14978522062301636, + "step": 2865, + "valid_targets_mean": 2693.1, + "valid_targets_min": 1092 + }, + { + "epoch": 4.7281713344316305, + "grad_norm": 0.5890780017197069, + "learning_rate": 1.1534400154077901e-05, + "loss": 0.1773, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18731525540351868, + "step": 2870, + "valid_targets_mean": 3308.1, + "valid_targets_min": 752 + }, + { + "epoch": 4.736408566721582, + "grad_norm": 0.62302716913029, + "learning_rate": 1.146003978838908e-05, + "loss": 0.2053, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20663988590240479, + "step": 2875, + "valid_targets_mean": 3739.2, + "valid_targets_min": 877 + }, + { + "epoch": 4.744645799011532, + "grad_norm": 0.7831927582297329, + "learning_rate": 1.1385823521121755e-05, + "loss": 0.1847, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18367904424667358, + "step": 2880, + "valid_targets_mean": 2389.8, + "valid_targets_min": 646 + }, + { + "epoch": 4.752883031301483, + "grad_norm": 0.7136796015209697, + "learning_rate": 1.1311752604558945e-05, + "loss": 0.1758, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18837086856365204, + "step": 2885, + "valid_targets_mean": 2352.7, + "valid_targets_min": 867 + }, + { + "epoch": 4.761120263591433, + "grad_norm": 0.7417881865230119, + "learning_rate": 1.1237828288531065e-05, + "loss": 0.1893, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16705411672592163, + "step": 2890, + "valid_targets_mean": 2411.9, + "valid_targets_min": 295 + }, + { + "epoch": 4.769357495881383, + "grad_norm": 0.7278603551951753, + "learning_rate": 1.1164051820394903e-05, + "loss": 0.1959, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18239541351795197, + "step": 2895, + "valid_targets_mean": 2383.0, + "valid_targets_min": 377 + }, + { + "epoch": 4.7775947281713345, + "grad_norm": 0.7684800575498447, + "learning_rate": 1.1090424445012524e-05, + "loss": 0.1973, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2619876265525818, + "step": 2900, + "valid_targets_mean": 2578.1, + "valid_targets_min": 753 + }, + { + "epoch": 4.785831960461285, + "grad_norm": 0.7762954518335069, + "learning_rate": 1.1016947404730302e-05, + "loss": 0.1827, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2098005712032318, + "step": 2905, + "valid_targets_mean": 2469.4, + "valid_targets_min": 280 + }, + { + "epoch": 4.794069192751236, + "grad_norm": 0.6972231214453025, + "learning_rate": 1.0943621939357952e-05, + "loss": 0.2173, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2021644413471222, + "step": 2910, + "valid_targets_mean": 2612.5, + "valid_targets_min": 799 + }, + { + "epoch": 4.802306425041186, + "grad_norm": 0.7647388529572376, + "learning_rate": 1.0870449286147565e-05, + "loss": 0.2031, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24047201871871948, + "step": 2915, + "valid_targets_mean": 2461.9, + "valid_targets_min": 219 + }, + { + "epoch": 4.810543657331137, + "grad_norm": 0.6738602004759084, + "learning_rate": 1.07974306797728e-05, + "loss": 0.2029, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17990803718566895, + "step": 2920, + "valid_targets_mean": 3235.8, + "valid_targets_min": 1435 + }, + { + "epoch": 4.818780889621087, + "grad_norm": 0.6592140334086931, + "learning_rate": 1.0724567352308008e-05, + "loss": 0.1886, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20495711266994476, + "step": 2925, + "valid_targets_mean": 3125.3, + "valid_targets_min": 226 + }, + { + "epoch": 4.8270181219110375, + "grad_norm": 0.7407499909683418, + "learning_rate": 1.0651860533207448e-05, + "loss": 0.1874, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1599995195865631, + "step": 2930, + "valid_targets_mean": 3042.9, + "valid_targets_min": 935 + }, + { + "epoch": 4.835255354200989, + "grad_norm": 0.716775005898679, + "learning_rate": 1.0579311449284555e-05, + "loss": 0.173, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17676642537117004, + "step": 2935, + "valid_targets_mean": 2293.6, + "valid_targets_min": 777 + }, + { + "epoch": 4.843492586490939, + "grad_norm": 0.8291279498158952, + "learning_rate": 1.0506921324691209e-05, + "loss": 0.2147, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19918978214263916, + "step": 2940, + "valid_targets_mean": 2218.4, + "valid_targets_min": 1162 + }, + { + "epoch": 4.85172981878089, + "grad_norm": 0.6111907605345152, + "learning_rate": 1.0434691380897132e-05, + "loss": 0.1909, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18697769939899445, + "step": 2945, + "valid_targets_mean": 3431.4, + "valid_targets_min": 895 + }, + { + "epoch": 4.85996705107084, + "grad_norm": 0.7998141165292254, + "learning_rate": 1.0362622836669216e-05, + "loss": 0.193, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1752634048461914, + "step": 2950, + "valid_targets_mean": 2012.6, + "valid_targets_min": 257 + }, + { + "epoch": 4.868204283360791, + "grad_norm": 0.7214959209341665, + "learning_rate": 1.0290716908051009e-05, + "loss": 0.1903, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23504498600959778, + "step": 2955, + "valid_targets_mean": 3218.4, + "valid_targets_min": 1304 + }, + { + "epoch": 4.8764415156507415, + "grad_norm": 0.6902196978381809, + "learning_rate": 1.0218974808342168e-05, + "loss": 0.202, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1906399428844452, + "step": 2960, + "valid_targets_mean": 2743.1, + "valid_targets_min": 322 + }, + { + "epoch": 4.884678747940692, + "grad_norm": 0.6282845815110746, + "learning_rate": 1.0147397748077999e-05, + "loss": 0.2122, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19511428475379944, + "step": 2965, + "valid_targets_mean": 3286.5, + "valid_targets_min": 385 + }, + { + "epoch": 4.892915980230643, + "grad_norm": 0.7362780284461559, + "learning_rate": 1.007598693500903e-05, + "loss": 0.1908, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19690576195716858, + "step": 2970, + "valid_targets_mean": 2372.4, + "valid_targets_min": 265 + }, + { + "epoch": 4.901153212520593, + "grad_norm": 0.6683978371796541, + "learning_rate": 1.0004743574080607e-05, + "loss": 0.1862, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1450890302658081, + "step": 2975, + "valid_targets_mean": 2457.7, + "valid_targets_min": 858 + }, + { + "epoch": 4.909390444810544, + "grad_norm": 0.6596995747936142, + "learning_rate": 9.93366886741261e-06, + "loss": 0.1956, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22884449362754822, + "step": 2980, + "valid_targets_mean": 3525.4, + "valid_targets_min": 187 + }, + { + "epoch": 4.917627677100494, + "grad_norm": 0.7659921918500595, + "learning_rate": 9.862764014279134e-06, + "loss": 0.2104, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2457534223794937, + "step": 2985, + "valid_targets_mean": 2558.6, + "valid_targets_min": 799 + }, + { + "epoch": 4.9258649093904445, + "grad_norm": 0.7799286605012021, + "learning_rate": 9.792030211088248e-06, + "loss": 0.2358, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22971343994140625, + "step": 2990, + "valid_targets_mean": 2437.7, + "valid_targets_min": 330 + }, + { + "epoch": 4.934102141680396, + "grad_norm": 0.6320931491436543, + "learning_rate": 9.721468651361852e-06, + "loss": 0.1975, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18385854363441467, + "step": 2995, + "valid_targets_mean": 3284.8, + "valid_targets_min": 696 + }, + { + "epoch": 4.942339373970346, + "grad_norm": 0.6523206752281229, + "learning_rate": 9.651080525715464e-06, + "loss": 0.1951, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16642795503139496, + "step": 3000, + "valid_targets_mean": 3221.4, + "valid_targets_min": 1577 + }, + { + "epoch": 4.950576606260297, + "grad_norm": 0.6643564942274582, + "learning_rate": 9.580867021838234e-06, + "loss": 0.208, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2085125744342804, + "step": 3005, + "valid_targets_mean": 3377.6, + "valid_targets_min": 297 + }, + { + "epoch": 4.958813838550247, + "grad_norm": 0.6413957328149001, + "learning_rate": 9.510829324472784e-06, + "loss": 0.2275, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25765886902809143, + "step": 3010, + "valid_targets_mean": 4052.6, + "valid_targets_min": 1031 + }, + { + "epoch": 4.967051070840197, + "grad_norm": 0.6990492677300354, + "learning_rate": 9.440968615395314e-06, + "loss": 0.1809, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18069782853126526, + "step": 3015, + "valid_targets_mean": 2553.2, + "valid_targets_min": 1073 + }, + { + "epoch": 4.9752883031301485, + "grad_norm": 0.6736246492056246, + "learning_rate": 9.37128607339562e-06, + "loss": 0.1932, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1594797819852829, + "step": 3020, + "valid_targets_mean": 2709.8, + "valid_targets_min": 290 + }, + { + "epoch": 4.983525535420099, + "grad_norm": 0.7071383567645622, + "learning_rate": 9.301782874257186e-06, + "loss": 0.1875, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15425880253314972, + "step": 3025, + "valid_targets_mean": 2314.9, + "valid_targets_min": 345 + }, + { + "epoch": 4.99176276771005, + "grad_norm": 0.6562276890652816, + "learning_rate": 9.232460190737394e-06, + "loss": 0.1819, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21255633234977722, + "step": 3030, + "valid_targets_mean": 4083.8, + "valid_targets_min": 1116 + }, + { + "epoch": 5.0, + "grad_norm": 0.7135456083913048, + "learning_rate": 9.163319192547702e-06, + "loss": 0.198, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18922117352485657, + "step": 3035, + "valid_targets_mean": 2817.9, + "valid_targets_min": 766 + }, + { + "epoch": 5.00823723228995, + "grad_norm": 0.6064180034994974, + "learning_rate": 9.094361046333887e-06, + "loss": 0.1614, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17224377393722534, + "step": 3040, + "valid_targets_mean": 3418.8, + "valid_targets_min": 1679 + }, + { + "epoch": 5.016474464579901, + "grad_norm": 0.6250290364668916, + "learning_rate": 9.025586915656435e-06, + "loss": 0.2112, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19228821992874146, + "step": 3045, + "valid_targets_mean": 3741.3, + "valid_targets_min": 886 + }, + { + "epoch": 5.0247116968698515, + "grad_norm": 0.6608009238868802, + "learning_rate": 8.956997960970811e-06, + "loss": 0.173, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19262053072452545, + "step": 3050, + "valid_targets_mean": 3611.7, + "valid_targets_min": 839 + }, + { + "epoch": 5.032948929159803, + "grad_norm": 0.7772909122140956, + "learning_rate": 8.888595339607961e-06, + "loss": 0.1874, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18313942849636078, + "step": 3055, + "valid_targets_mean": 2424.2, + "valid_targets_min": 356 + }, + { + "epoch": 5.041186161449753, + "grad_norm": 0.7689441961709437, + "learning_rate": 8.82038020575472e-06, + "loss": 0.1735, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2497469037771225, + "step": 3060, + "valid_targets_mean": 2754.6, + "valid_targets_min": 290 + }, + { + "epoch": 5.049423393739703, + "grad_norm": 0.6503950444326292, + "learning_rate": 8.752353710434384e-06, + "loss": 0.1762, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2629661560058594, + "step": 3065, + "valid_targets_mean": 4315.6, + "valid_targets_min": 980 + }, + { + "epoch": 5.057660626029654, + "grad_norm": 0.6604107548804747, + "learning_rate": 8.684517001487276e-06, + "loss": 0.1813, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20543421804904938, + "step": 3070, + "valid_targets_mean": 3637.5, + "valid_targets_min": 287 + }, + { + "epoch": 5.065897858319604, + "grad_norm": 0.65130591346722, + "learning_rate": 8.616871223551335e-06, + "loss": 0.1712, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1576567441225052, + "step": 3075, + "valid_targets_mean": 3172.4, + "valid_targets_min": 1218 + }, + { + "epoch": 5.0741350906095555, + "grad_norm": 0.7123187129434798, + "learning_rate": 8.549417518042899e-06, + "loss": 0.1789, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19344539940357208, + "step": 3080, + "valid_targets_mean": 3020.7, + "valid_targets_min": 702 + }, + { + "epoch": 5.082372322899506, + "grad_norm": 0.684466499187168, + "learning_rate": 8.482157023137326e-06, + "loss": 0.1638, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15878722071647644, + "step": 3085, + "valid_targets_mean": 3033.1, + "valid_targets_min": 1433 + }, + { + "epoch": 5.090609555189456, + "grad_norm": 0.6591878834846665, + "learning_rate": 8.415090873749882e-06, + "loss": 0.1535, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14512422680854797, + "step": 3090, + "valid_targets_mean": 2839.8, + "valid_targets_min": 326 + }, + { + "epoch": 5.098846787479407, + "grad_norm": 0.7689559250431048, + "learning_rate": 8.348220201516559e-06, + "loss": 0.1837, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2209683060646057, + "step": 3095, + "valid_targets_mean": 2806.2, + "valid_targets_min": 539 + }, + { + "epoch": 5.107084019769357, + "grad_norm": 0.6724369983360083, + "learning_rate": 8.281546134774952e-06, + "loss": 0.1622, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1659250110387802, + "step": 3100, + "valid_targets_mean": 3220.8, + "valid_targets_min": 586 + }, + { + "epoch": 5.115321252059308, + "grad_norm": 0.6592543385028731, + "learning_rate": 8.215069798545272e-06, + "loss": 0.1786, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23743778467178345, + "step": 3105, + "valid_targets_mean": 3924.8, + "valid_targets_min": 1061 + }, + { + "epoch": 5.1235584843492585, + "grad_norm": 0.7140903108295529, + "learning_rate": 8.148792314511338e-06, + "loss": 0.1739, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18408019840717316, + "step": 3110, + "valid_targets_mean": 2868.4, + "valid_targets_min": 279 + }, + { + "epoch": 5.13179571663921, + "grad_norm": 0.899945522646544, + "learning_rate": 8.082714801001619e-06, + "loss": 0.1758, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1668897271156311, + "step": 3115, + "valid_targets_mean": 1955.2, + "valid_targets_min": 809 + }, + { + "epoch": 5.14003294892916, + "grad_norm": 0.7903726889847759, + "learning_rate": 8.016838372970453e-06, + "loss": 0.1767, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19560575485229492, + "step": 3120, + "valid_targets_mean": 2365.5, + "valid_targets_min": 850 + }, + { + "epoch": 5.14827018121911, + "grad_norm": 0.6437043381500877, + "learning_rate": 7.951164141979115e-06, + "loss": 0.1798, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22203288972377777, + "step": 3125, + "valid_targets_mean": 3675.4, + "valid_targets_min": 1346 + }, + { + "epoch": 5.156507413509061, + "grad_norm": 0.6355970111306471, + "learning_rate": 7.885693216177166e-06, + "loss": 0.1967, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17794297635555267, + "step": 3130, + "valid_targets_mean": 3428.3, + "valid_targets_min": 993 + }, + { + "epoch": 5.164744645799011, + "grad_norm": 0.6602780647849495, + "learning_rate": 7.820426700283681e-06, + "loss": 0.1653, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1728067845106125, + "step": 3135, + "valid_targets_mean": 3440.2, + "valid_targets_min": 988 + }, + { + "epoch": 5.1729818780889625, + "grad_norm": 0.7558597560486282, + "learning_rate": 7.75536569556866e-06, + "loss": 0.1834, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20588910579681396, + "step": 3140, + "valid_targets_mean": 2655.5, + "valid_targets_min": 814 + }, + { + "epoch": 5.181219110378913, + "grad_norm": 0.7254810135039633, + "learning_rate": 7.690511299834422e-06, + "loss": 0.1736, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1662125587463379, + "step": 3145, + "valid_targets_mean": 2955.9, + "valid_targets_min": 311 + }, + { + "epoch": 5.189456342668863, + "grad_norm": 0.8088311377677032, + "learning_rate": 7.625864607397066e-06, + "loss": 0.1799, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25056391954421997, + "step": 3150, + "valid_targets_mean": 2665.3, + "valid_targets_min": 565 + }, + { + "epoch": 5.197693574958814, + "grad_norm": 0.6600076683827095, + "learning_rate": 7.561426709068045e-06, + "loss": 0.176, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18403570353984833, + "step": 3155, + "valid_targets_mean": 3254.5, + "valid_targets_min": 728 + }, + { + "epoch": 5.205930807248764, + "grad_norm": 0.5798390366412892, + "learning_rate": 7.497198692135728e-06, + "loss": 0.1845, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14264634251594543, + "step": 3160, + "valid_targets_mean": 4274.8, + "valid_targets_min": 1430 + }, + { + "epoch": 5.214168039538715, + "grad_norm": 0.8075445062618977, + "learning_rate": 7.433181640347069e-06, + "loss": 0.1801, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17545044422149658, + "step": 3165, + "valid_targets_mean": 3412.2, + "valid_targets_min": 219 + }, + { + "epoch": 5.2224052718286655, + "grad_norm": 0.8096547199240914, + "learning_rate": 7.369376633889325e-06, + "loss": 0.1744, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15541712939739227, + "step": 3170, + "valid_targets_mean": 2235.3, + "valid_targets_min": 204 + }, + { + "epoch": 5.230642504118616, + "grad_norm": 0.7262956103025833, + "learning_rate": 7.305784749371792e-06, + "loss": 0.1891, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.218002587556839, + "step": 3175, + "valid_targets_mean": 3621.4, + "valid_targets_min": 1416 + }, + { + "epoch": 5.238879736408567, + "grad_norm": 0.7538387339240314, + "learning_rate": 7.242407059807703e-06, + "loss": 0.174, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20847590267658234, + "step": 3180, + "valid_targets_mean": 2906.4, + "valid_targets_min": 900 + }, + { + "epoch": 5.247116968698517, + "grad_norm": 0.6712012846940215, + "learning_rate": 7.1792446345960585e-06, + "loss": 0.1557, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1522168517112732, + "step": 3185, + "valid_targets_mean": 2999.8, + "valid_targets_min": 335 + }, + { + "epoch": 5.255354200988468, + "grad_norm": 0.7008582088098334, + "learning_rate": 7.116298539503632e-06, + "loss": 0.1935, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15230123698711395, + "step": 3190, + "valid_targets_mean": 2597.4, + "valid_targets_min": 212 + }, + { + "epoch": 5.263591433278418, + "grad_norm": 0.6913752679862281, + "learning_rate": 7.053569836646954e-06, + "loss": 0.1624, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16717877984046936, + "step": 3195, + "valid_targets_mean": 3209.5, + "valid_targets_min": 445 + }, + { + "epoch": 5.2718286655683695, + "grad_norm": 0.6501280031713065, + "learning_rate": 6.991059584474411e-06, + "loss": 0.1594, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15155775845050812, + "step": 3200, + "valid_targets_mean": 3640.1, + "valid_targets_min": 244 + }, + { + "epoch": 5.28006589785832, + "grad_norm": 0.7765572752139934, + "learning_rate": 6.928768837748374e-06, + "loss": 0.17, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1907849758863449, + "step": 3205, + "valid_targets_mean": 2407.2, + "valid_targets_min": 1252 + }, + { + "epoch": 5.28830313014827, + "grad_norm": 0.7092172752068605, + "learning_rate": 6.866698647527392e-06, + "loss": 0.1899, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19017310440540314, + "step": 3210, + "valid_targets_mean": 2805.6, + "valid_targets_min": 776 + }, + { + "epoch": 5.296540362438221, + "grad_norm": 0.8973485963057156, + "learning_rate": 6.804850061148485e-06, + "loss": 0.2021, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1752794086933136, + "step": 3215, + "valid_targets_mean": 2147.9, + "valid_targets_min": 677 + }, + { + "epoch": 5.304777594728171, + "grad_norm": 0.707169418301808, + "learning_rate": 6.7432241222094595e-06, + "loss": 0.1752, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20905683934688568, + "step": 3220, + "valid_targets_mean": 3281.1, + "valid_targets_min": 724 + }, + { + "epoch": 5.313014827018122, + "grad_norm": 0.632264893798177, + "learning_rate": 6.68182187055127e-06, + "loss": 0.1543, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13711237907409668, + "step": 3225, + "valid_targets_mean": 3425.3, + "valid_targets_min": 847 + }, + { + "epoch": 5.3212520593080725, + "grad_norm": 0.6416477054286873, + "learning_rate": 6.620644342240534e-06, + "loss": 0.1708, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15817886590957642, + "step": 3230, + "valid_targets_mean": 3488.2, + "valid_targets_min": 585 + }, + { + "epoch": 5.329489291598023, + "grad_norm": 0.7253505491366616, + "learning_rate": 6.559692569551976e-06, + "loss": 0.1651, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18176016211509705, + "step": 3235, + "valid_targets_mean": 2424.8, + "valid_targets_min": 965 + }, + { + "epoch": 5.337726523887974, + "grad_norm": 0.7467293463976239, + "learning_rate": 6.498967580951099e-06, + "loss": 0.1723, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1569557785987854, + "step": 3240, + "valid_targets_mean": 2483.9, + "valid_targets_min": 1182 + }, + { + "epoch": 5.345963756177924, + "grad_norm": 0.8565193699113087, + "learning_rate": 6.4384704010767284e-06, + "loss": 0.1792, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18094724416732788, + "step": 3245, + "valid_targets_mean": 2141.0, + "valid_targets_min": 260 + }, + { + "epoch": 5.354200988467875, + "grad_norm": 0.7269022982597675, + "learning_rate": 6.37820205072381e-06, + "loss": 0.14, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.121427983045578, + "step": 3250, + "valid_targets_mean": 2299.0, + "valid_targets_min": 1207 + }, + { + "epoch": 5.362438220757825, + "grad_norm": 0.7414210784723774, + "learning_rate": 6.318163546826141e-06, + "loss": 0.1518, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17040827870368958, + "step": 3255, + "valid_targets_mean": 2556.3, + "valid_targets_min": 265 + }, + { + "epoch": 5.370675453047776, + "grad_norm": 0.7458160033756344, + "learning_rate": 6.258355902439211e-06, + "loss": 0.1589, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14077043533325195, + "step": 3260, + "valid_targets_mean": 2237.4, + "valid_targets_min": 1025 + }, + { + "epoch": 5.378912685337727, + "grad_norm": 0.8342015753026745, + "learning_rate": 6.198780126723127e-06, + "loss": 0.1746, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14754045009613037, + "step": 3265, + "valid_targets_mean": 1824.2, + "valid_targets_min": 402 + }, + { + "epoch": 5.387149917627677, + "grad_norm": 0.7694290384969541, + "learning_rate": 6.139437224925586e-06, + "loss": 0.2016, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1629246175289154, + "step": 3270, + "valid_targets_mean": 2445.3, + "valid_targets_min": 845 + }, + { + "epoch": 5.395387149917628, + "grad_norm": 0.8433928744995592, + "learning_rate": 6.080328198364871e-06, + "loss": 0.188, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2084483504295349, + "step": 3275, + "valid_targets_mean": 2179.2, + "valid_targets_min": 633 + }, + { + "epoch": 5.403624382207578, + "grad_norm": 0.6765682861675392, + "learning_rate": 6.021454044413036e-06, + "loss": 0.1779, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2180410623550415, + "step": 3280, + "valid_targets_mean": 3277.3, + "valid_targets_min": 1020 + }, + { + "epoch": 5.4118616144975284, + "grad_norm": 0.8357652595647097, + "learning_rate": 5.962815756478979e-06, + "loss": 0.1943, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1691618412733078, + "step": 3285, + "valid_targets_mean": 2519.4, + "valid_targets_min": 878 + }, + { + "epoch": 5.4200988467874796, + "grad_norm": 0.7365400222373201, + "learning_rate": 5.904414323991758e-06, + "loss": 0.186, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17638184130191803, + "step": 3290, + "valid_targets_mean": 2686.8, + "valid_targets_min": 358 + }, + { + "epoch": 5.42833607907743, + "grad_norm": 0.7247240550977059, + "learning_rate": 5.846250732383869e-06, + "loss": 0.2068, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24847513437271118, + "step": 3295, + "valid_targets_mean": 3847.8, + "valid_targets_min": 698 + }, + { + "epoch": 5.436573311367381, + "grad_norm": 0.7198196895329133, + "learning_rate": 5.788325963074594e-06, + "loss": 0.1738, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17607322335243225, + "step": 3300, + "valid_targets_mean": 3049.2, + "valid_targets_min": 260 + }, + { + "epoch": 5.444810543657331, + "grad_norm": 0.8022615837890243, + "learning_rate": 5.7306409934534936e-06, + "loss": 0.1856, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18957354128360748, + "step": 3305, + "valid_targets_mean": 2815.9, + "valid_targets_min": 796 + }, + { + "epoch": 5.453047775947281, + "grad_norm": 0.7703287318099119, + "learning_rate": 5.673196796863852e-06, + "loss": 0.1944, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19865599274635315, + "step": 3310, + "valid_targets_mean": 3163.5, + "valid_targets_min": 1041 + }, + { + "epoch": 5.461285008237232, + "grad_norm": 0.6861868123941506, + "learning_rate": 5.615994342586313e-06, + "loss": 0.1658, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.160249724984169, + "step": 3315, + "valid_targets_mean": 2706.8, + "valid_targets_min": 1026 + }, + { + "epoch": 5.469522240527183, + "grad_norm": 0.7519375725202005, + "learning_rate": 5.5590345958224875e-06, + "loss": 0.1747, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19434326887130737, + "step": 3320, + "valid_targets_mean": 2473.9, + "valid_targets_min": 357 + }, + { + "epoch": 5.477759472817134, + "grad_norm": 0.7034008504710153, + "learning_rate": 5.502318517678682e-06, + "loss": 0.1598, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13975057005882263, + "step": 3325, + "valid_targets_mean": 2624.3, + "valid_targets_min": 380 + }, + { + "epoch": 5.485996705107084, + "grad_norm": 0.8591217062074403, + "learning_rate": 5.445847065149679e-06, + "loss": 0.2001, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19654160737991333, + "step": 3330, + "valid_targets_mean": 2530.1, + "valid_targets_min": 647 + }, + { + "epoch": 5.494233937397034, + "grad_norm": 0.8338393901784925, + "learning_rate": 5.389621191102574e-06, + "loss": 0.1784, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19803670048713684, + "step": 3335, + "valid_targets_mean": 3171.4, + "valid_targets_min": 925 + }, + { + "epoch": 5.502471169686985, + "grad_norm": 0.8232013140154525, + "learning_rate": 5.333641844260726e-06, + "loss": 0.1935, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21081340312957764, + "step": 3340, + "valid_targets_mean": 2620.4, + "valid_targets_min": 607 + }, + { + "epoch": 5.5107084019769355, + "grad_norm": 0.7263973632736029, + "learning_rate": 5.277909969187738e-06, + "loss": 0.1745, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.11975730210542679, + "step": 3345, + "valid_targets_mean": 2141.7, + "valid_targets_min": 552 + }, + { + "epoch": 5.518945634266887, + "grad_norm": 0.7175609162042647, + "learning_rate": 5.222426506271492e-06, + "loss": 0.1814, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15939165651798248, + "step": 3350, + "valid_targets_mean": 3050.1, + "valid_targets_min": 746 + }, + { + "epoch": 5.527182866556837, + "grad_norm": 0.748834509252294, + "learning_rate": 5.1671923917083265e-06, + "loss": 0.1922, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18665283918380737, + "step": 3355, + "valid_targets_mean": 2947.6, + "valid_targets_min": 927 + }, + { + "epoch": 5.535420098846787, + "grad_norm": 0.915756061799089, + "learning_rate": 5.112208557487215e-06, + "loss": 0.2002, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17622342705726624, + "step": 3360, + "valid_targets_mean": 3114.1, + "valid_targets_min": 776 + }, + { + "epoch": 5.543657331136738, + "grad_norm": 0.6861044823588834, + "learning_rate": 5.057475931374043e-06, + "loss": 0.174, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15702880918979645, + "step": 3365, + "valid_targets_mean": 3333.8, + "valid_targets_min": 283 + }, + { + "epoch": 5.551894563426688, + "grad_norm": 0.5651714611404062, + "learning_rate": 5.002995436895939e-06, + "loss": 0.1846, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19566282629966736, + "step": 3370, + "valid_targets_mean": 4441.9, + "valid_targets_min": 1039 + }, + { + "epoch": 5.560131795716639, + "grad_norm": 0.6967484805438395, + "learning_rate": 4.9487679933257225e-06, + "loss": 0.1706, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17878486216068268, + "step": 3375, + "valid_targets_mean": 3512.9, + "valid_targets_min": 977 + }, + { + "epoch": 5.56836902800659, + "grad_norm": 0.8309410288700217, + "learning_rate": 4.894794515666379e-06, + "loss": 0.1515, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1670929193496704, + "step": 3380, + "valid_targets_mean": 2985.6, + "valid_targets_min": 791 + }, + { + "epoch": 5.576606260296541, + "grad_norm": 0.8303611692984448, + "learning_rate": 4.8410759146355955e-06, + "loss": 0.2015, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16505856812000275, + "step": 3385, + "valid_targets_mean": 2491.2, + "valid_targets_min": 940 + }, + { + "epoch": 5.584843492586491, + "grad_norm": 0.7021211702845122, + "learning_rate": 4.787613096650436e-06, + "loss": 0.1809, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12547199428081512, + "step": 3390, + "valid_targets_mean": 3197.2, + "valid_targets_min": 706 + }, + { + "epoch": 5.593080724876441, + "grad_norm": 0.6982593083876117, + "learning_rate": 4.734406963812024e-06, + "loss": 0.146, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16671398282051086, + "step": 3395, + "valid_targets_mean": 3031.1, + "valid_targets_min": 651 + }, + { + "epoch": 5.601317957166392, + "grad_norm": 0.7983795110267299, + "learning_rate": 4.68145841389032e-06, + "loss": 0.1836, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2452123761177063, + "step": 3400, + "valid_targets_mean": 3621.4, + "valid_targets_min": 868 + }, + { + "epoch": 5.6095551894563425, + "grad_norm": 0.6939388117607412, + "learning_rate": 4.6287683403089845e-06, + "loss": 0.1965, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1813758760690689, + "step": 3405, + "valid_targets_mean": 3380.9, + "valid_targets_min": 1276 + }, + { + "epoch": 5.617792421746294, + "grad_norm": 0.7353689335564925, + "learning_rate": 4.576337632130281e-06, + "loss": 0.1804, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14547865092754364, + "step": 3410, + "valid_targets_mean": 2548.1, + "valid_targets_min": 750 + }, + { + "epoch": 5.626029654036244, + "grad_norm": 0.6926011113099905, + "learning_rate": 4.524167174040106e-06, + "loss": 0.1756, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14587116241455078, + "step": 3415, + "valid_targets_mean": 2785.8, + "valid_targets_min": 288 + }, + { + "epoch": 5.634266886326195, + "grad_norm": 0.7100090253616618, + "learning_rate": 4.472257846333028e-06, + "loss": 0.1419, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12672248482704163, + "step": 3420, + "valid_targets_mean": 2911.2, + "valid_targets_min": 1017 + }, + { + "epoch": 5.642504118616145, + "grad_norm": 0.8387847090896776, + "learning_rate": 4.4206105248974595e-06, + "loss": 0.1817, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17093172669410706, + "step": 3425, + "valid_targets_mean": 2210.1, + "valid_targets_min": 633 + }, + { + "epoch": 5.650741350906095, + "grad_norm": 0.8135873943654238, + "learning_rate": 4.3692260812008795e-06, + "loss": 0.1709, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1717970371246338, + "step": 3430, + "valid_targets_mean": 2176.8, + "valid_targets_min": 866 + }, + { + "epoch": 5.658978583196046, + "grad_norm": 0.7923907688849422, + "learning_rate": 4.318105382275084e-06, + "loss": 0.1734, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2087642252445221, + "step": 3435, + "valid_targets_mean": 2451.8, + "valid_targets_min": 666 + }, + { + "epoch": 5.667215815485997, + "grad_norm": 0.6638888534023216, + "learning_rate": 4.267249290701636e-06, + "loss": 0.1683, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17858381569385529, + "step": 3440, + "valid_targets_mean": 3964.4, + "valid_targets_min": 307 + }, + { + "epoch": 5.675453047775948, + "grad_norm": 0.8617516970063258, + "learning_rate": 4.216658664597224e-06, + "loss": 0.1982, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18910962343215942, + "step": 3445, + "valid_targets_mean": 2019.4, + "valid_targets_min": 506 + }, + { + "epoch": 5.683690280065898, + "grad_norm": 0.7634977977727311, + "learning_rate": 4.166334357599246e-06, + "loss": 0.1683, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1956837773323059, + "step": 3450, + "valid_targets_mean": 2368.8, + "valid_targets_min": 562 + }, + { + "epoch": 5.691927512355848, + "grad_norm": 0.7838029896349801, + "learning_rate": 4.116277218851383e-06, + "loss": 0.1824, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17025712132453918, + "step": 3455, + "valid_targets_mean": 2937.7, + "valid_targets_min": 668 + }, + { + "epoch": 5.700164744645799, + "grad_norm": 0.7044949991777757, + "learning_rate": 4.066488092989256e-06, + "loss": 0.1821, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2160659283399582, + "step": 3460, + "valid_targets_mean": 3333.1, + "valid_targets_min": 1031 + }, + { + "epoch": 5.7084019769357495, + "grad_norm": 0.9783577658970115, + "learning_rate": 4.016967820126205e-06, + "loss": 0.1704, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18528926372528076, + "step": 3465, + "valid_targets_mean": 2410.6, + "valid_targets_min": 683 + }, + { + "epoch": 5.716639209225701, + "grad_norm": 0.7036357285381335, + "learning_rate": 3.967717235839086e-06, + "loss": 0.1577, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1655019372701645, + "step": 3470, + "valid_targets_mean": 2925.2, + "valid_targets_min": 834 + }, + { + "epoch": 5.724876441515651, + "grad_norm": 0.6713767526809996, + "learning_rate": 3.918737171154192e-06, + "loss": 0.1608, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13649466633796692, + "step": 3475, + "valid_targets_mean": 2739.8, + "valid_targets_min": 230 + }, + { + "epoch": 5.733113673805601, + "grad_norm": 0.7714880924709924, + "learning_rate": 3.870028452533221e-06, + "loss": 0.1673, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1947619915008545, + "step": 3480, + "valid_targets_mean": 2570.9, + "valid_targets_min": 708 + }, + { + "epoch": 5.741350906095552, + "grad_norm": 0.6732219111117853, + "learning_rate": 3.8215919018593295e-06, + "loss": 0.1519, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13189196586608887, + "step": 3485, + "valid_targets_mean": 3156.6, + "valid_targets_min": 1248 + }, + { + "epoch": 5.749588138385502, + "grad_norm": 0.7828985393778445, + "learning_rate": 3.7734283364232747e-06, + "loss": 0.159, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14875520765781403, + "step": 3490, + "valid_targets_mean": 2533.2, + "valid_targets_min": 685 + }, + { + "epoch": 5.757825370675453, + "grad_norm": 0.6734972968776775, + "learning_rate": 3.725538568909599e-06, + "loss": 0.1691, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20598509907722473, + "step": 3495, + "valid_targets_mean": 3320.4, + "valid_targets_min": 327 + }, + { + "epoch": 5.766062602965404, + "grad_norm": 0.6820313312676656, + "learning_rate": 3.6779234073829528e-06, + "loss": 0.1671, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16222935914993286, + "step": 3500, + "valid_targets_mean": 3013.1, + "valid_targets_min": 414 + }, + { + "epoch": 5.774299835255354, + "grad_norm": 0.6881508771610351, + "learning_rate": 3.63058365527444e-06, + "loss": 0.1799, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17598870396614075, + "step": 3505, + "valid_targets_mean": 3021.1, + "valid_targets_min": 260 + }, + { + "epoch": 5.782537067545305, + "grad_norm": 0.9256567228616236, + "learning_rate": 3.5835201113680463e-06, + "loss": 0.1563, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18399620056152344, + "step": 3510, + "valid_targets_mean": 3049.4, + "valid_targets_min": 780 + }, + { + "epoch": 5.790774299835255, + "grad_norm": 0.7127835516858468, + "learning_rate": 3.536733569787194e-06, + "loss": 0.1779, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14168041944503784, + "step": 3515, + "valid_targets_mean": 2699.2, + "valid_targets_min": 1104 + }, + { + "epoch": 5.799011532125206, + "grad_norm": 0.7360005045380545, + "learning_rate": 3.4902248199813203e-06, + "loss": 0.1731, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13892826437950134, + "step": 3520, + "valid_targets_mean": 2244.3, + "valid_targets_min": 602 + }, + { + "epoch": 5.8072487644151565, + "grad_norm": 0.7455487123063135, + "learning_rate": 3.443994646712567e-06, + "loss": 0.1871, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16907833516597748, + "step": 3525, + "valid_targets_mean": 2390.0, + "valid_targets_min": 781 + }, + { + "epoch": 5.815485996705107, + "grad_norm": 0.7953720238270837, + "learning_rate": 3.3980438300425324e-06, + "loss": 0.1631, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1514473408460617, + "step": 3530, + "valid_targets_mean": 2531.9, + "valid_targets_min": 838 + }, + { + "epoch": 5.823723228995058, + "grad_norm": 0.8642656461553999, + "learning_rate": 3.3523731453190987e-06, + "loss": 0.155, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16404001414775848, + "step": 3535, + "valid_targets_mean": 1984.9, + "valid_targets_min": 808 + }, + { + "epoch": 5.831960461285008, + "grad_norm": 0.7256800913138621, + "learning_rate": 3.306983363163381e-06, + "loss": 0.1811, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12058289349079132, + "step": 3540, + "valid_targets_mean": 2846.2, + "valid_targets_min": 786 + }, + { + "epoch": 5.840197693574959, + "grad_norm": 0.8398559139673432, + "learning_rate": 3.2618752494566854e-06, + "loss": 0.1863, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21678978204727173, + "step": 3545, + "valid_targets_mean": 2182.9, + "valid_targets_min": 807 + }, + { + "epoch": 5.848434925864909, + "grad_norm": 0.6879487573765043, + "learning_rate": 3.217049565327619e-06, + "loss": 0.1964, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13484476506710052, + "step": 3550, + "valid_targets_mean": 2719.2, + "valid_targets_min": 688 + }, + { + "epoch": 5.8566721581548595, + "grad_norm": 0.8490732841663783, + "learning_rate": 3.17250706713923e-06, + "loss": 0.2014, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21391984820365906, + "step": 3555, + "valid_targets_mean": 2457.4, + "valid_targets_min": 241 + }, + { + "epoch": 5.864909390444811, + "grad_norm": 0.7426460128321304, + "learning_rate": 3.128248506476246e-06, + "loss": 0.1823, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14091576635837555, + "step": 3560, + "valid_targets_mean": 2596.6, + "valid_targets_min": 245 + }, + { + "epoch": 5.873146622734761, + "grad_norm": 0.7853668676403339, + "learning_rate": 3.0842746301324e-06, + "loss": 0.1621, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17900870740413666, + "step": 3565, + "valid_targets_mean": 2629.8, + "valid_targets_min": 848 + }, + { + "epoch": 5.881383855024712, + "grad_norm": 0.6294725390125606, + "learning_rate": 3.040586180097811e-06, + "loss": 0.1697, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14573122560977936, + "step": 3570, + "valid_targets_mean": 3388.9, + "valid_targets_min": 980 + }, + { + "epoch": 5.889621087314662, + "grad_norm": 0.635131125967507, + "learning_rate": 2.9971838935464915e-06, + "loss": 0.1918, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18714481592178345, + "step": 3575, + "valid_targets_mean": 3639.6, + "valid_targets_min": 1601 + }, + { + "epoch": 5.897858319604612, + "grad_norm": 0.6935148132102796, + "learning_rate": 2.9540685028238925e-06, + "loss": 0.1735, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18102405965328217, + "step": 3580, + "valid_targets_mean": 2775.3, + "valid_targets_min": 318 + }, + { + "epoch": 5.9060955518945635, + "grad_norm": 0.8574201407311213, + "learning_rate": 2.9112407354345375e-06, + "loss": 0.1923, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2605670988559723, + "step": 3585, + "valid_targets_mean": 2246.1, + "valid_targets_min": 327 + }, + { + "epoch": 5.914332784184514, + "grad_norm": 0.9115993171771825, + "learning_rate": 2.8687013140297695e-06, + "loss": 0.1804, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1853502094745636, + "step": 3590, + "valid_targets_mean": 2516.3, + "valid_targets_min": 1178 + }, + { + "epoch": 5.922570016474465, + "grad_norm": 0.711616625135048, + "learning_rate": 2.826450956395532e-06, + "loss": 0.1854, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21265952289104462, + "step": 3595, + "valid_targets_mean": 3306.9, + "valid_targets_min": 1654 + }, + { + "epoch": 5.930807248764415, + "grad_norm": 0.6932061364220065, + "learning_rate": 2.7844903754402963e-06, + "loss": 0.1873, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19808021187782288, + "step": 3600, + "valid_targets_mean": 3391.8, + "valid_targets_min": 346 + }, + { + "epoch": 5.939044481054366, + "grad_norm": 0.7566426671217025, + "learning_rate": 2.7428202791829763e-06, + "loss": 0.1571, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16674354672431946, + "step": 3605, + "valid_targets_mean": 2808.9, + "valid_targets_min": 287 + }, + { + "epoch": 5.947281713344316, + "grad_norm": 0.8418388087549749, + "learning_rate": 2.7014413707410357e-06, + "loss": 0.1709, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16913406550884247, + "step": 3610, + "valid_targets_mean": 2207.9, + "valid_targets_min": 644 + }, + { + "epoch": 5.9555189456342665, + "grad_norm": 0.7988681269945298, + "learning_rate": 2.6603543483185966e-06, + "loss": 0.1865, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.159542053937912, + "step": 3615, + "valid_targets_mean": 2564.2, + "valid_targets_min": 687 + }, + { + "epoch": 5.963756177924218, + "grad_norm": 1.2032980721393542, + "learning_rate": 2.6195599051946465e-06, + "loss": 0.1868, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19994597136974335, + "step": 3620, + "valid_targets_mean": 2516.3, + "valid_targets_min": 987 + }, + { + "epoch": 5.971993410214168, + "grad_norm": 0.7714977813745794, + "learning_rate": 2.579058729711379e-06, + "loss": 0.1609, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1823139488697052, + "step": 3625, + "valid_targets_mean": 2836.2, + "valid_targets_min": 263 + }, + { + "epoch": 5.980230642504119, + "grad_norm": 0.8014540368577482, + "learning_rate": 2.538851505262545e-06, + "loss": 0.1763, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15929174423217773, + "step": 3630, + "valid_targets_mean": 2114.4, + "valid_targets_min": 320 + }, + { + "epoch": 5.988467874794069, + "grad_norm": 0.7461862435721606, + "learning_rate": 2.498938910281927e-06, + "loss": 0.1665, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1749926060438156, + "step": 3635, + "valid_targets_mean": 2477.5, + "valid_targets_min": 770 + }, + { + "epoch": 5.996705107084019, + "grad_norm": 0.7368517970426673, + "learning_rate": 2.4593216182319203e-06, + "loss": 0.1994, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15466895699501038, + "step": 3640, + "valid_targets_mean": 2621.8, + "valid_targets_min": 245 + }, + { + "epoch": 6.0049423393739705, + "grad_norm": 1.1139077443534142, + "learning_rate": 2.42000029759212e-06, + "loss": 0.173, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14994430541992188, + "step": 3645, + "valid_targets_mean": 2619.9, + "valid_targets_min": 636 + }, + { + "epoch": 6.013179571663921, + "grad_norm": 0.7501461411955563, + "learning_rate": 2.3809756118480887e-06, + "loss": 0.1773, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1623890995979309, + "step": 3650, + "valid_targets_mean": 2354.2, + "valid_targets_min": 614 + }, + { + "epoch": 6.021416803953872, + "grad_norm": 0.6157972732215693, + "learning_rate": 2.3422482194801257e-06, + "loss": 0.154, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15174467861652374, + "step": 3655, + "valid_targets_mean": 3239.4, + "valid_targets_min": 764 + }, + { + "epoch": 6.029654036243822, + "grad_norm": 0.7350556411709983, + "learning_rate": 2.30381877395218e-06, + "loss": 0.1741, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17065928876399994, + "step": 3660, + "valid_targets_mean": 2507.8, + "valid_targets_min": 650 + }, + { + "epoch": 6.037891268533772, + "grad_norm": 0.6901704684540658, + "learning_rate": 2.265687923700821e-06, + "loss": 0.1649, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16900062561035156, + "step": 3665, + "valid_targets_mean": 2752.2, + "valid_targets_min": 1188 + }, + { + "epoch": 6.046128500823723, + "grad_norm": 0.7086734284825671, + "learning_rate": 2.2278563121242722e-06, + "loss": 0.2035, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20763811469078064, + "step": 3670, + "valid_targets_mean": 3084.0, + "valid_targets_min": 307 + }, + { + "epoch": 6.0543657331136735, + "grad_norm": 0.6385356293662908, + "learning_rate": 2.190324577571592e-06, + "loss": 0.1468, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14509770274162292, + "step": 3675, + "valid_targets_mean": 3518.7, + "valid_targets_min": 527 + }, + { + "epoch": 6.062602965403625, + "grad_norm": 0.7028174328512578, + "learning_rate": 2.153093353331881e-06, + "loss": 0.2103, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15137967467308044, + "step": 3680, + "valid_targets_mean": 3283.1, + "valid_targets_min": 246 + }, + { + "epoch": 6.070840197693575, + "grad_norm": 0.7222091137953284, + "learning_rate": 2.1161632676235967e-06, + "loss": 0.1757, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20794406533241272, + "step": 3685, + "valid_targets_mean": 3220.8, + "valid_targets_min": 219 + }, + { + "epoch": 6.079077429983526, + "grad_norm": 0.6855245337047342, + "learning_rate": 2.0795349435839606e-06, + "loss": 0.1656, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12052720040082932, + "step": 3690, + "valid_targets_mean": 2771.8, + "valid_targets_min": 281 + }, + { + "epoch": 6.087314662273476, + "grad_norm": 0.707804572172083, + "learning_rate": 2.0432089992584326e-06, + "loss": 0.1587, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15237155556678772, + "step": 3695, + "valid_targets_mean": 2842.4, + "valid_targets_min": 1123 + }, + { + "epoch": 6.095551894563426, + "grad_norm": 0.6707345478391269, + "learning_rate": 2.007186047590297e-06, + "loss": 0.1741, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14419394731521606, + "step": 3700, + "valid_targets_mean": 3254.1, + "valid_targets_min": 849 + }, + { + "epoch": 6.1037891268533775, + "grad_norm": 0.7003141972567551, + "learning_rate": 1.971466696410309e-06, + "loss": 0.1423, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1416216790676117, + "step": 3705, + "valid_targets_mean": 2684.4, + "valid_targets_min": 1045 + }, + { + "epoch": 6.112026359143328, + "grad_norm": 0.7825738433775659, + "learning_rate": 1.936051548426443e-06, + "loss": 0.1518, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1420452892780304, + "step": 3710, + "valid_targets_mean": 2350.9, + "valid_targets_min": 338 + }, + { + "epoch": 6.120263591433279, + "grad_norm": 0.7115145095325718, + "learning_rate": 1.9009412012137218e-06, + "loss": 0.1951, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1335878074169159, + "step": 3715, + "valid_targets_mean": 2756.2, + "valid_targets_min": 251 + }, + { + "epoch": 6.128500823723229, + "grad_norm": 0.6743967837978332, + "learning_rate": 1.8661362472041355e-06, + "loss": 0.1632, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1492428481578827, + "step": 3720, + "valid_targets_mean": 2904.1, + "valid_targets_min": 725 + }, + { + "epoch": 6.136738056013179, + "grad_norm": 0.7703129690601694, + "learning_rate": 1.8316372736766452e-06, + "loss": 0.1638, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20545321702957153, + "step": 3725, + "valid_targets_mean": 3010.5, + "valid_targets_min": 278 + }, + { + "epoch": 6.14497528830313, + "grad_norm": 0.7888666022596672, + "learning_rate": 1.7974448627472618e-06, + "loss": 0.1484, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17078402638435364, + "step": 3730, + "valid_targets_mean": 2981.3, + "valid_targets_min": 766 + }, + { + "epoch": 6.1532125205930805, + "grad_norm": 0.7723578107116098, + "learning_rate": 1.763559591359243e-06, + "loss": 0.1542, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1601790189743042, + "step": 3735, + "valid_targets_mean": 2846.0, + "valid_targets_min": 298 + }, + { + "epoch": 6.161449752883032, + "grad_norm": 0.6714966512844651, + "learning_rate": 1.7299820312733518e-06, + "loss": 0.1756, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19277028739452362, + "step": 3740, + "valid_targets_mean": 4139.1, + "valid_targets_min": 334 + }, + { + "epoch": 6.169686985172982, + "grad_norm": 0.6556105712174505, + "learning_rate": 1.6967127490581892e-06, + "loss": 0.1744, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1665448248386383, + "step": 3745, + "valid_targets_mean": 3238.6, + "valid_targets_min": 747 + }, + { + "epoch": 6.177924217462932, + "grad_norm": 0.7965390505057974, + "learning_rate": 1.6637523060806749e-06, + "loss": 0.1695, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16333314776420593, + "step": 3750, + "valid_targets_mean": 2388.4, + "valid_targets_min": 783 + }, + { + "epoch": 6.186161449752883, + "grad_norm": 0.66375569184018, + "learning_rate": 1.6311012584965257e-06, + "loss": 0.177, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21853406727313995, + "step": 3755, + "valid_targets_mean": 3801.9, + "valid_targets_min": 317 + }, + { + "epoch": 6.194398682042833, + "grad_norm": 0.669378716443229, + "learning_rate": 1.5987601572409195e-06, + "loss": 0.1604, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1353456974029541, + "step": 3760, + "valid_targets_mean": 3566.6, + "valid_targets_min": 275 + }, + { + "epoch": 6.2026359143327845, + "grad_norm": 0.7658647113291661, + "learning_rate": 1.566729548019168e-06, + "loss": 0.1677, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1602291613817215, + "step": 3765, + "valid_targets_mean": 2423.1, + "valid_targets_min": 937 + }, + { + "epoch": 6.210873146622735, + "grad_norm": 0.6590955988866615, + "learning_rate": 1.5350099712975097e-06, + "loss": 0.1688, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12900327146053314, + "step": 3770, + "valid_targets_mean": 3467.8, + "valid_targets_min": 789 + }, + { + "epoch": 6.219110378912685, + "grad_norm": 0.6919074917450025, + "learning_rate": 1.50360196229401e-06, + "loss": 0.1634, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2072930634021759, + "step": 3775, + "valid_targets_mean": 3171.7, + "valid_targets_min": 969 + }, + { + "epoch": 6.227347611202636, + "grad_norm": 0.874989495696985, + "learning_rate": 1.4725060509695088e-06, + "loss": 0.1589, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21609821915626526, + "step": 3780, + "valid_targets_mean": 2349.0, + "valid_targets_min": 716 + }, + { + "epoch": 6.235584843492586, + "grad_norm": 0.7463878847371747, + "learning_rate": 1.4417227620186935e-06, + "loss": 0.1848, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16198115050792694, + "step": 3785, + "valid_targets_mean": 2775.0, + "valid_targets_min": 299 + }, + { + "epoch": 6.243822075782537, + "grad_norm": 0.6615073957168196, + "learning_rate": 1.4112526148612381e-06, + "loss": 0.1648, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12445035576820374, + "step": 3790, + "valid_targets_mean": 2758.4, + "valid_targets_min": 586 + }, + { + "epoch": 6.2520593080724876, + "grad_norm": 0.6921577907753451, + "learning_rate": 1.3810961236330345e-06, + "loss": 0.1834, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1393837332725525, + "step": 3795, + "valid_targets_mean": 3353.4, + "valid_targets_min": 1125 + }, + { + "epoch": 6.260296540362438, + "grad_norm": 0.8382225209020714, + "learning_rate": 1.3512537971775407e-06, + "loss": 0.1522, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19350308179855347, + "step": 3800, + "valid_targets_mean": 2176.9, + "valid_targets_min": 731 + }, + { + "epoch": 6.268533772652389, + "grad_norm": 0.6084781353758029, + "learning_rate": 1.321726139037156e-06, + "loss": 0.1665, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1659453809261322, + "step": 3805, + "valid_targets_mean": 3997.4, + "valid_targets_min": 797 + }, + { + "epoch": 6.276771004942339, + "grad_norm": 0.7719108277276304, + "learning_rate": 1.2925136474447575e-06, + "loss": 0.1777, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.155721515417099, + "step": 3810, + "valid_targets_mean": 2487.1, + "valid_targets_min": 219 + }, + { + "epoch": 6.28500823723229, + "grad_norm": 0.6385197418328769, + "learning_rate": 1.2636168153152828e-06, + "loss": 0.158, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13569995760917664, + "step": 3815, + "valid_targets_mean": 3099.1, + "valid_targets_min": 391 + }, + { + "epoch": 6.29324546952224, + "grad_norm": 0.7174793871853767, + "learning_rate": 1.2350361302374037e-06, + "loss": 0.1881, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1641596257686615, + "step": 3820, + "valid_targets_mean": 2929.2, + "valid_targets_min": 760 + }, + { + "epoch": 6.3014827018121915, + "grad_norm": 0.651538874031102, + "learning_rate": 1.2067720744653167e-06, + "loss": 0.1416, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12000978738069534, + "step": 3825, + "valid_targets_mean": 2685.2, + "valid_targets_min": 939 + }, + { + "epoch": 6.309719934102142, + "grad_norm": 0.8511093562682773, + "learning_rate": 1.178825124910583e-06, + "loss": 0.1761, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23789116740226746, + "step": 3830, + "valid_targets_mean": 2482.9, + "valid_targets_min": 896 + }, + { + "epoch": 6.317957166392092, + "grad_norm": 0.6180405933314328, + "learning_rate": 1.1511957531341067e-06, + "loss": 0.1749, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.11477728188037872, + "step": 3835, + "valid_targets_mean": 3123.0, + "valid_targets_min": 811 + }, + { + "epoch": 6.326194398682043, + "grad_norm": 0.6803850559923774, + "learning_rate": 1.123884425338162e-06, + "loss": 0.1652, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1466941237449646, + "step": 3840, + "valid_targets_mean": 3088.1, + "valid_targets_min": 944 + }, + { + "epoch": 6.334431630971993, + "grad_norm": 0.7546973039836182, + "learning_rate": 1.0968916023585252e-06, + "loss": 0.1916, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14515511691570282, + "step": 3845, + "valid_targets_mean": 2963.4, + "valid_targets_min": 941 + }, + { + "epoch": 6.342668863261944, + "grad_norm": 0.8849963117408646, + "learning_rate": 1.0702177396567114e-06, + "loss": 0.1813, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2112627923488617, + "step": 3850, + "valid_targets_mean": 1949.1, + "valid_targets_min": 345 + }, + { + "epoch": 6.350906095551895, + "grad_norm": 0.7459355712021345, + "learning_rate": 1.0438632873122746e-06, + "loss": 0.1523, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15826204419136047, + "step": 3855, + "valid_targets_mean": 2619.4, + "valid_targets_min": 677 + }, + { + "epoch": 6.359143327841845, + "grad_norm": 0.7772343391993064, + "learning_rate": 1.0178286900152234e-06, + "loss": 0.1745, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23661720752716064, + "step": 3860, + "valid_targets_mean": 2736.1, + "valid_targets_min": 836 + }, + { + "epoch": 6.367380560131796, + "grad_norm": 0.6357761439560725, + "learning_rate": 9.921143870585204e-07, + "loss": 0.168, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19213977456092834, + "step": 3865, + "valid_targets_mean": 3818.3, + "valid_targets_min": 618 + }, + { + "epoch": 6.375617792421746, + "grad_norm": 0.7752815261134633, + "learning_rate": 9.667208123306504e-07, + "loss": 0.1516, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1300448477268219, + "step": 3870, + "valid_targets_mean": 2181.9, + "valid_targets_min": 335 + }, + { + "epoch": 6.383855024711697, + "grad_norm": 1.0569445961287538, + "learning_rate": 9.416483943083254e-07, + "loss": 0.1819, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16549789905548096, + "step": 3875, + "valid_targets_mean": 3206.5, + "valid_targets_min": 1031 + }, + { + "epoch": 6.392092257001647, + "grad_norm": 0.7208101885756489, + "learning_rate": 9.168975560492365e-07, + "loss": 0.1703, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14495733380317688, + "step": 3880, + "valid_targets_mean": 3358.2, + "valid_targets_min": 1197 + }, + { + "epoch": 6.400329489291598, + "grad_norm": 0.7429014255052777, + "learning_rate": 8.924687151849287e-07, + "loss": 0.1713, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14917151629924774, + "step": 3885, + "valid_targets_mean": 2492.2, + "valid_targets_min": 890 + }, + { + "epoch": 6.408566721581549, + "grad_norm": 0.7462560187876585, + "learning_rate": 8.683622839137306e-07, + "loss": 0.1904, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23635666072368622, + "step": 3890, + "valid_targets_mean": 2909.4, + "valid_targets_min": 923 + }, + { + "epoch": 6.416803953871499, + "grad_norm": 0.8102243606807108, + "learning_rate": 8.445786689938295e-07, + "loss": 0.1844, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12910988926887512, + "step": 3895, + "valid_targets_mean": 2163.8, + "valid_targets_min": 429 + }, + { + "epoch": 6.42504118616145, + "grad_norm": 0.80398663184621, + "learning_rate": 8.211182717363897e-07, + "loss": 0.1616, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19596856832504272, + "step": 3900, + "valid_targets_mean": 2539.6, + "valid_targets_min": 857 + }, + { + "epoch": 6.4332784184514, + "grad_norm": 0.7481982084445109, + "learning_rate": 7.979814879987824e-07, + "loss": 0.1557, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17043043673038483, + "step": 3905, + "valid_targets_mean": 2694.1, + "valid_targets_min": 799 + }, + { + "epoch": 6.441515650741351, + "grad_norm": 0.6789946169423258, + "learning_rate": 7.751687081779091e-07, + "loss": 0.1938, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19388365745544434, + "step": 3910, + "valid_targets_mean": 3303.7, + "valid_targets_min": 347 + }, + { + "epoch": 6.449752883031302, + "grad_norm": 0.6422614423127603, + "learning_rate": 7.526803172036157e-07, + "loss": 0.1415, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.11945605278015137, + "step": 3915, + "valid_targets_mean": 2707.4, + "valid_targets_min": 184 + }, + { + "epoch": 6.457990115321252, + "grad_norm": 0.7156633329528237, + "learning_rate": 7.305166945321973e-07, + "loss": 0.1809, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19679643213748932, + "step": 3920, + "valid_targets_mean": 3079.4, + "valid_targets_min": 875 + }, + { + "epoch": 6.466227347611203, + "grad_norm": 0.7680662464415903, + "learning_rate": 7.086782141399906e-07, + "loss": 0.1605, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1780366599559784, + "step": 3925, + "valid_targets_mean": 2651.0, + "valid_targets_min": 460 + }, + { + "epoch": 6.474464579901153, + "grad_norm": 1.0095018002402179, + "learning_rate": 6.871652445170651e-07, + "loss": 0.1659, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1527692675590515, + "step": 3930, + "valid_targets_mean": 3270.0, + "valid_targets_min": 258 + }, + { + "epoch": 6.482701812191104, + "grad_norm": 0.8515545672589472, + "learning_rate": 6.659781486610129e-07, + "loss": 0.1759, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22965921461582184, + "step": 3935, + "valid_targets_mean": 3471.1, + "valid_targets_min": 1364 + }, + { + "epoch": 6.490939044481054, + "grad_norm": 0.712226673890132, + "learning_rate": 6.451172840708087e-07, + "loss": 0.1556, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16279277205467224, + "step": 3940, + "valid_targets_mean": 3104.0, + "valid_targets_min": 460 + }, + { + "epoch": 6.499176276771005, + "grad_norm": 0.7787226581803016, + "learning_rate": 6.245830027407973e-07, + "loss": 0.1738, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16567203402519226, + "step": 3945, + "valid_targets_mean": 2723.2, + "valid_targets_min": 649 + }, + { + "epoch": 6.507413509060956, + "grad_norm": 0.7502856008513566, + "learning_rate": 6.043756511547427e-07, + "loss": 0.1673, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16041092574596405, + "step": 3950, + "valid_targets_mean": 3132.1, + "valid_targets_min": 248 + }, + { + "epoch": 6.515650741350906, + "grad_norm": 0.675892844554783, + "learning_rate": 5.84495570279966e-07, + "loss": 0.1602, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15754902362823486, + "step": 3955, + "valid_targets_mean": 3227.8, + "valid_targets_min": 304 + }, + { + "epoch": 6.523887973640857, + "grad_norm": 0.7664735528399342, + "learning_rate": 5.649430955616365e-07, + "loss": 0.1609, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18401801586151123, + "step": 3960, + "valid_targets_mean": 2677.6, + "valid_targets_min": 821 + }, + { + "epoch": 6.532125205930807, + "grad_norm": 0.8538927050950035, + "learning_rate": 5.457185569170609e-07, + "loss": 0.148, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1640760451555252, + "step": 3965, + "valid_targets_mean": 2824.4, + "valid_targets_min": 319 + }, + { + "epoch": 6.5403624382207575, + "grad_norm": 0.7882400286252053, + "learning_rate": 5.268222787301502e-07, + "loss": 0.1549, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17009207606315613, + "step": 3970, + "valid_targets_mean": 2632.1, + "valid_targets_min": 783 + }, + { + "epoch": 6.548599670510709, + "grad_norm": 0.7311404280782681, + "learning_rate": 5.082545798459348e-07, + "loss": 0.1766, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18268424272537231, + "step": 3975, + "valid_targets_mean": 3112.3, + "valid_targets_min": 702 + }, + { + "epoch": 6.556836902800659, + "grad_norm": 0.7146800124090497, + "learning_rate": 4.900157735651779e-07, + "loss": 0.1566, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.163301020860672, + "step": 3980, + "valid_targets_mean": 3048.5, + "valid_targets_min": 964 + }, + { + "epoch": 6.56507413509061, + "grad_norm": 0.7405738502525695, + "learning_rate": 4.721061676391081e-07, + "loss": 0.1707, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14739716053009033, + "step": 3985, + "valid_targets_mean": 2616.9, + "valid_targets_min": 321 + }, + { + "epoch": 6.57331136738056, + "grad_norm": 0.7244028411140117, + "learning_rate": 4.5452606426420466e-07, + "loss": 0.1739, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1695755273103714, + "step": 3990, + "valid_targets_mean": 3027.8, + "valid_targets_min": 257 + }, + { + "epoch": 6.58154859967051, + "grad_norm": 0.7322525481149665, + "learning_rate": 4.372757600771116e-07, + "loss": 0.1684, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2116568684577942, + "step": 3995, + "valid_targets_mean": 3028.4, + "valid_targets_min": 917 + }, + { + "epoch": 6.589785831960461, + "grad_norm": 0.7470670386639713, + "learning_rate": 4.203555461496378e-07, + "loss": 0.1627, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15893974900245667, + "step": 4000, + "valid_targets_mean": 3174.8, + "valid_targets_min": 731 + }, + { + "epoch": 6.598023064250412, + "grad_norm": 0.8075055424213986, + "learning_rate": 4.0376570798382084e-07, + "loss": 0.1621, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15712273120880127, + "step": 4005, + "valid_targets_mean": 2127.1, + "valid_targets_min": 259 + }, + { + "epoch": 6.606260296540363, + "grad_norm": 0.7269307165713284, + "learning_rate": 3.875065255071442e-07, + "loss": 0.1707, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24262076616287231, + "step": 4010, + "valid_targets_mean": 3033.4, + "valid_targets_min": 989 + }, + { + "epoch": 6.614497528830313, + "grad_norm": 0.9343679862401862, + "learning_rate": 3.715782730677786e-07, + "loss": 0.1627, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1790003478527069, + "step": 4015, + "valid_targets_mean": 1810.2, + "valid_targets_min": 620 + }, + { + "epoch": 6.622734761120263, + "grad_norm": 0.7559108888626762, + "learning_rate": 3.559812194299794e-07, + "loss": 0.1541, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2013910859823227, + "step": 4020, + "valid_targets_mean": 3094.5, + "valid_targets_min": 1049 + }, + { + "epoch": 6.630971993410214, + "grad_norm": 0.7245273592521206, + "learning_rate": 3.407156277695456e-07, + "loss": 0.1727, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1565759778022766, + "step": 4025, + "valid_targets_mean": 4298.2, + "valid_targets_min": 935 + }, + { + "epoch": 6.6392092257001645, + "grad_norm": 0.6500329190512484, + "learning_rate": 3.257817556693654e-07, + "loss": 0.1616, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.11484387516975403, + "step": 4030, + "valid_targets_mean": 2797.9, + "valid_targets_min": 779 + }, + { + "epoch": 6.647446457990116, + "grad_norm": 0.7052895181792681, + "learning_rate": 3.1117985511508686e-07, + "loss": 0.1517, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.132914736866951, + "step": 4035, + "valid_targets_mean": 2802.5, + "valid_targets_min": 438 + }, + { + "epoch": 6.655683690280066, + "grad_norm": 0.7468505902354071, + "learning_rate": 2.969101724908585e-07, + "loss": 0.1759, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2299904227256775, + "step": 4040, + "valid_targets_mean": 3584.6, + "valid_targets_min": 978 + }, + { + "epoch": 6.663920922570016, + "grad_norm": 0.7150895108441544, + "learning_rate": 2.829729485751731e-07, + "loss": 0.1399, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14426125586032867, + "step": 4045, + "valid_targets_mean": 2814.2, + "valid_targets_min": 382 + }, + { + "epoch": 6.672158154859967, + "grad_norm": 0.5991220098063653, + "learning_rate": 2.6936841853680397e-07, + "loss": 0.1542, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12172479182481766, + "step": 4050, + "valid_targets_mean": 3342.2, + "valid_targets_min": 405 + }, + { + "epoch": 6.680395387149917, + "grad_norm": 0.6822342950062912, + "learning_rate": 2.560968119308349e-07, + "loss": 0.1745, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19796040654182434, + "step": 4055, + "valid_targets_mean": 3490.1, + "valid_targets_min": 838 + }, + { + "epoch": 6.688632619439868, + "grad_norm": 0.6514055359914713, + "learning_rate": 2.4315835269479446e-07, + "loss": 0.1798, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1311188042163849, + "step": 4060, + "valid_targets_mean": 3191.2, + "valid_targets_min": 380 + }, + { + "epoch": 6.696869851729819, + "grad_norm": 0.7294000714400222, + "learning_rate": 2.305532591448678e-07, + "loss": 0.1544, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14557018876075745, + "step": 4065, + "valid_targets_mean": 2508.8, + "valid_targets_min": 724 + }, + { + "epoch": 6.70510708401977, + "grad_norm": 0.5934845642808244, + "learning_rate": 2.1828174397222179e-07, + "loss": 0.1666, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12286630272865295, + "step": 4070, + "valid_targets_mean": 3624.8, + "valid_targets_min": 843 + }, + { + "epoch": 6.71334431630972, + "grad_norm": 0.7921124334514694, + "learning_rate": 2.0634401423940797e-07, + "loss": 0.1738, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18661408126354218, + "step": 4075, + "valid_targets_mean": 2451.5, + "valid_targets_min": 717 + }, + { + "epoch": 6.72158154859967, + "grad_norm": 0.5932236802274212, + "learning_rate": 1.9474027137687425e-07, + "loss": 0.1728, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.183675616979599, + "step": 4080, + "valid_targets_mean": 4072.7, + "valid_targets_min": 808 + }, + { + "epoch": 6.729818780889621, + "grad_norm": 0.793939849451655, + "learning_rate": 1.834707111795675e-07, + "loss": 0.1724, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18774361908435822, + "step": 4085, + "valid_targets_mean": 2594.6, + "valid_targets_min": 743 + }, + { + "epoch": 6.7380560131795715, + "grad_norm": 0.7977767421315517, + "learning_rate": 1.725355238036208e-07, + "loss": 0.182, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23956944048404694, + "step": 4090, + "valid_targets_mean": 2484.7, + "valid_targets_min": 271 + }, + { + "epoch": 6.746293245469523, + "grad_norm": 0.8987888128978652, + "learning_rate": 1.6193489376315375e-07, + "loss": 0.1782, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18341073393821716, + "step": 4095, + "valid_targets_mean": 1867.1, + "valid_targets_min": 346 + }, + { + "epoch": 6.754530477759473, + "grad_norm": 0.7775999067743287, + "learning_rate": 1.5166899992715922e-07, + "loss": 0.1603, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15121757984161377, + "step": 4100, + "valid_targets_mean": 2914.3, + "valid_targets_min": 287 + }, + { + "epoch": 6.762767710049423, + "grad_norm": 0.8019192495560047, + "learning_rate": 1.417380155164727e-07, + "loss": 0.1867, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22095879912376404, + "step": 4105, + "valid_targets_mean": 2675.8, + "valid_targets_min": 239 + }, + { + "epoch": 6.771004942339374, + "grad_norm": 0.8608991084693377, + "learning_rate": 1.3214210810087224e-07, + "loss": 0.1799, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19388195872306824, + "step": 4110, + "valid_targets_mean": 2775.4, + "valid_targets_min": 915 + }, + { + "epoch": 6.779242174629324, + "grad_norm": 0.6496355441738153, + "learning_rate": 1.2288143959622523e-07, + "loss": 0.1441, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1336159110069275, + "step": 4115, + "valid_targets_mean": 3077.2, + "valid_targets_min": 935 + }, + { + "epoch": 6.787479406919275, + "grad_norm": 0.9623189464394726, + "learning_rate": 1.1395616626177718e-07, + "loss": 0.1649, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18686145544052124, + "step": 4120, + "valid_targets_mean": 1904.4, + "valid_targets_min": 377 + }, + { + "epoch": 6.795716639209226, + "grad_norm": 0.8506497654414124, + "learning_rate": 1.0536643869750508e-07, + "loss": 0.1842, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23170790076255798, + "step": 4125, + "valid_targets_mean": 2162.9, + "valid_targets_min": 1039 + }, + { + "epoch": 6.803953871499177, + "grad_norm": 0.647365705379033, + "learning_rate": 9.711240184157255e-08, + "loss": 0.1744, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19694972038269043, + "step": 4130, + "valid_targets_mean": 4015.7, + "valid_targets_min": 1299 + }, + { + "epoch": 6.812191103789127, + "grad_norm": 0.6899146547611903, + "learning_rate": 8.919419496789428e-08, + "loss": 0.1343, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13986879587173462, + "step": 4135, + "valid_targets_mean": 3022.9, + "valid_targets_min": 1184 + }, + { + "epoch": 6.820428336079077, + "grad_norm": 0.6826690079265932, + "learning_rate": 8.161195168378211e-08, + "loss": 0.1541, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13281051814556122, + "step": 4140, + "valid_targets_mean": 3272.3, + "valid_targets_min": 232 + }, + { + "epoch": 6.828665568369028, + "grad_norm": 0.6641391234585438, + "learning_rate": 7.4365799927687e-08, + "loss": 0.1595, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14258205890655518, + "step": 4145, + "valid_targets_mean": 3228.8, + "valid_targets_min": 1676 + }, + { + "epoch": 6.8369028006589785, + "grad_norm": 0.6748162422562181, + "learning_rate": 6.745586196704512e-08, + "loss": 0.1633, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14117348194122314, + "step": 4150, + "valid_targets_mean": 3223.8, + "valid_targets_min": 377 + }, + { + "epoch": 6.84514003294893, + "grad_norm": 0.8649156647091603, + "learning_rate": 6.088225439620621e-08, + "loss": 0.1906, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.216533362865448, + "step": 4155, + "valid_targets_mean": 2923.2, + "valid_targets_min": 610 + }, + { + "epoch": 6.85337726523888, + "grad_norm": 0.9134316166153748, + "learning_rate": 5.464508813448621e-08, + "loss": 0.1934, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17572906613349915, + "step": 4160, + "valid_targets_mean": 2606.5, + "valid_targets_min": 290 + }, + { + "epoch": 6.86161449752883, + "grad_norm": 0.6317158342753643, + "learning_rate": 4.874446842426883e-08, + "loss": 0.1664, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13150852918624878, + "step": 4165, + "valid_targets_mean": 3455.3, + "valid_targets_min": 1075 + }, + { + "epoch": 6.869851729818781, + "grad_norm": 0.78925360032166, + "learning_rate": 4.318049482924913e-08, + "loss": 0.1567, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1530526578426361, + "step": 4170, + "valid_targets_mean": 2389.6, + "valid_targets_min": 708 + }, + { + "epoch": 6.878088962108731, + "grad_norm": 0.6158559873310975, + "learning_rate": 3.7953261232743785e-08, + "loss": 0.1698, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1563345193862915, + "step": 4175, + "valid_targets_mean": 3645.4, + "valid_targets_min": 848 + }, + { + "epoch": 6.886326194398682, + "grad_norm": 0.8168240860097061, + "learning_rate": 3.3062855836119014e-08, + "loss": 0.1666, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15081787109375, + "step": 4180, + "valid_targets_mean": 2565.6, + "valid_targets_min": 1303 + }, + { + "epoch": 6.894563426688633, + "grad_norm": 0.6402154892970332, + "learning_rate": 2.8509361157285088e-08, + "loss": 0.1535, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2066039741039276, + "step": 4185, + "valid_targets_mean": 4142.2, + "valid_targets_min": 326 + }, + { + "epoch": 6.902800658978583, + "grad_norm": 0.6139396948124894, + "learning_rate": 2.429285402931525e-08, + "loss": 0.1424, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13339337706565857, + "step": 4190, + "valid_targets_mean": 3607.2, + "valid_targets_min": 776 + }, + { + "epoch": 6.911037891268534, + "grad_norm": 0.8049219522828062, + "learning_rate": 2.041340559914895e-08, + "loss": 0.1747, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16743309795856476, + "step": 4195, + "valid_targets_mean": 2342.8, + "valid_targets_min": 611 + }, + { + "epoch": 6.919275123558484, + "grad_norm": 0.7235236484755135, + "learning_rate": 1.6871081326388372e-08, + "loss": 0.1475, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1539749652147293, + "step": 4200, + "valid_targets_mean": 2721.6, + "valid_targets_min": 255 + }, + { + "epoch": 6.927512355848435, + "grad_norm": 0.6996590910363392, + "learning_rate": 1.366594098219487e-08, + "loss": 0.1698, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14047235250473022, + "step": 4205, + "valid_targets_mean": 3126.0, + "valid_targets_min": 340 + }, + { + "epoch": 6.9357495881383855, + "grad_norm": 0.9081681836064551, + "learning_rate": 1.0798038648278664e-08, + "loss": 0.1805, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15091955661773682, + "step": 4210, + "valid_targets_mean": 2809.3, + "valid_targets_min": 829 + }, + { + "epoch": 6.943986820428336, + "grad_norm": 0.6676226069756183, + "learning_rate": 8.26742271598846e-09, + "loss": 0.1621, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17465578019618988, + "step": 4215, + "valid_targets_mean": 3463.6, + "valid_targets_min": 915 + }, + { + "epoch": 6.952224052718287, + "grad_norm": 0.7700986208930122, + "learning_rate": 6.074135885498767e-09, + "loss": 0.1522, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17234516143798828, + "step": 4220, + "valid_targets_mean": 2481.6, + "valid_targets_min": 943 + }, + { + "epoch": 6.960461285008237, + "grad_norm": 0.679973114052915, + "learning_rate": 4.218215165077144e-09, + "loss": 0.1781, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1480589061975479, + "step": 4225, + "valid_targets_mean": 3207.2, + "valid_targets_min": 287 + }, + { + "epoch": 6.968698517298188, + "grad_norm": 1.0661673612936513, + "learning_rate": 2.699691870473586e-09, + "loss": 0.1647, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14552760124206543, + "step": 4230, + "valid_targets_mean": 2714.8, + "valid_targets_min": 832 + }, + { + "epoch": 6.976935749588138, + "grad_norm": 0.7275517602358317, + "learning_rate": 1.5185916243876108e-09, + "loss": 0.1599, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17081902921199799, + "step": 4235, + "valid_targets_mean": 2979.4, + "valid_targets_min": 1228 + }, + { + "epoch": 6.9851729818780885, + "grad_norm": 0.8670521747754626, + "learning_rate": 6.749343560286114e-10, + "loss": 0.1846, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18648138642311096, + "step": 4240, + "valid_targets_mean": 2214.1, + "valid_targets_min": 311 + }, + { + "epoch": 6.99341021416804, + "grad_norm": 0.8106614308590394, + "learning_rate": 1.6873430078945264e-10, + "loss": 0.1749, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2821575999259949, + "step": 4245, + "valid_targets_mean": 3544.4, + "valid_targets_min": 980 + }, + { + "epoch": 7.0, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17179973423480988, + "step": 4249, + "total_flos": 835464024555520.0, + "train_loss": 0.2412474821612088, + "train_runtime": 18396.3609, + "train_samples_per_second": 3.695, + "train_steps_per_second": 0.231, + "valid_targets_mean": 2603.9, + "valid_targets_min": 249 + } + ], + "logging_steps": 5, + "max_steps": 4249, + "num_input_tokens_seen": 0, + "num_train_epochs": 7, + "save_steps": 1500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 835464024555520.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +}