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": 13.838484688249258, + "learning_rate": 3.764705882352941e-07, + "loss": 0.6118, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.6064709424972534, + "step": 5, + "valid_targets_mean": 5784.8, + "valid_targets_min": 330 + }, + { + "epoch": 0.016474464579901153, + "grad_norm": 13.32901770197041, + "learning_rate": 8.470588235294118e-07, + "loss": 0.5804, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.5605111718177795, + "step": 10, + "valid_targets_mean": 5472.2, + "valid_targets_min": 302 + }, + { + "epoch": 0.02471169686985173, + "grad_norm": 11.580100121107343, + "learning_rate": 1.3176470588235296e-06, + "loss": 0.5825, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.521429181098938, + "step": 15, + "valid_targets_mean": 7353.7, + "valid_targets_min": 1985 + }, + { + "epoch": 0.032948929159802305, + "grad_norm": 10.103393446647656, + "learning_rate": 1.7882352941176474e-06, + "loss": 0.531, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.5192767381668091, + "step": 20, + "valid_targets_mean": 6811.0, + "valid_targets_min": 478 + }, + { + "epoch": 0.04118616144975288, + "grad_norm": 6.257789545103483, + "learning_rate": 2.258823529411765e-06, + "loss": 0.4974, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.48259904980659485, + "step": 25, + "valid_targets_mean": 4482.8, + "valid_targets_min": 696 + }, + { + "epoch": 0.04942339373970346, + "grad_norm": 3.4046228381087698, + "learning_rate": 2.7294117647058825e-06, + "loss": 0.4402, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.41031402349472046, + "step": 30, + "valid_targets_mean": 6257.0, + "valid_targets_min": 920 + }, + { + "epoch": 0.057660626029654036, + "grad_norm": 2.4965600200587112, + "learning_rate": 3.2000000000000003e-06, + "loss": 0.4411, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.4100528955459595, + "step": 35, + "valid_targets_mean": 4688.1, + "valid_targets_min": 281 + }, + { + "epoch": 0.06589785831960461, + "grad_norm": 1.3768965526806027, + "learning_rate": 3.670588235294118e-06, + "loss": 0.4167, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3804216980934143, + "step": 40, + "valid_targets_mean": 4690.6, + "valid_targets_min": 254 + }, + { + "epoch": 0.07413509060955518, + "grad_norm": 0.9454923709313997, + "learning_rate": 4.141176470588235e-06, + "loss": 0.3692, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.36766085028648376, + "step": 45, + "valid_targets_mean": 5414.4, + "valid_targets_min": 312 + }, + { + "epoch": 0.08237232289950576, + "grad_norm": 0.895620303556488, + "learning_rate": 4.611764705882353e-06, + "loss": 0.3545, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.374552845954895, + "step": 50, + "valid_targets_mean": 4861.9, + "valid_targets_min": 325 + }, + { + "epoch": 0.09060955518945635, + "grad_norm": 0.6828779095753463, + "learning_rate": 5.08235294117647e-06, + "loss": 0.3537, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.320254385471344, + "step": 55, + "valid_targets_mean": 4765.4, + "valid_targets_min": 273 + }, + { + "epoch": 0.09884678747940692, + "grad_norm": 0.6524094091597135, + "learning_rate": 5.552941176470589e-06, + "loss": 0.3345, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.34802478551864624, + "step": 60, + "valid_targets_mean": 5150.5, + "valid_targets_min": 333 + }, + { + "epoch": 0.1070840197693575, + "grad_norm": 0.5319046745810463, + "learning_rate": 6.023529411764706e-06, + "loss": 0.3405, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32357484102249146, + "step": 65, + "valid_targets_mean": 6576.6, + "valid_targets_min": 2014 + }, + { + "epoch": 0.11532125205930807, + "grad_norm": 0.5431891570006655, + "learning_rate": 6.494117647058824e-06, + "loss": 0.3479, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.34960079193115234, + "step": 70, + "valid_targets_mean": 6236.7, + "valid_targets_min": 712 + }, + { + "epoch": 0.12355848434925865, + "grad_norm": 0.5215208361615085, + "learning_rate": 6.964705882352941e-06, + "loss": 0.3292, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3283542990684509, + "step": 75, + "valid_targets_mean": 5510.8, + "valid_targets_min": 383 + }, + { + "epoch": 0.13179571663920922, + "grad_norm": 0.5335500745911854, + "learning_rate": 7.43529411764706e-06, + "loss": 0.3182, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.330269992351532, + "step": 80, + "valid_targets_mean": 5953.8, + "valid_targets_min": 525 + }, + { + "epoch": 0.1400329489291598, + "grad_norm": 0.4791132982310386, + "learning_rate": 7.905882352941176e-06, + "loss": 0.3313, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31799042224884033, + "step": 85, + "valid_targets_mean": 6383.9, + "valid_targets_min": 331 + }, + { + "epoch": 0.14827018121911037, + "grad_norm": 0.6014083377497861, + "learning_rate": 8.376470588235295e-06, + "loss": 0.324, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.32026463747024536, + "step": 90, + "valid_targets_mean": 4742.5, + "valid_targets_min": 430 + }, + { + "epoch": 0.15650741350906094, + "grad_norm": 0.4491945455497733, + "learning_rate": 8.847058823529413e-06, + "loss": 0.3073, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3021693229675293, + "step": 95, + "valid_targets_mean": 5802.7, + "valid_targets_min": 298 + }, + { + "epoch": 0.16474464579901152, + "grad_norm": 0.48340705186331245, + "learning_rate": 9.31764705882353e-06, + "loss": 0.3077, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3046356439590454, + "step": 100, + "valid_targets_mean": 5698.9, + "valid_targets_min": 2110 + }, + { + "epoch": 0.17298187808896212, + "grad_norm": 0.45376506719321025, + "learning_rate": 9.788235294117649e-06, + "loss": 0.2947, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3012458384037018, + "step": 105, + "valid_targets_mean": 5975.9, + "valid_targets_min": 392 + }, + { + "epoch": 0.1812191103789127, + "grad_norm": 0.48884539184282266, + "learning_rate": 1.0258823529411766e-05, + "loss": 0.2952, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30425509810447693, + "step": 110, + "valid_targets_mean": 5205.1, + "valid_targets_min": 416 + }, + { + "epoch": 0.18945634266886327, + "grad_norm": 0.545402654783909, + "learning_rate": 1.0729411764705884e-05, + "loss": 0.302, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31671977043151855, + "step": 115, + "valid_targets_mean": 5585.2, + "valid_targets_min": 1766 + }, + { + "epoch": 0.19769357495881384, + "grad_norm": 0.5044877249055142, + "learning_rate": 1.1200000000000001e-05, + "loss": 0.3002, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.31186243891716003, + "step": 120, + "valid_targets_mean": 4735.4, + "valid_targets_min": 350 + }, + { + "epoch": 0.20593080724876442, + "grad_norm": 0.4917352363375157, + "learning_rate": 1.1670588235294118e-05, + "loss": 0.2917, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2831946909427643, + "step": 125, + "valid_targets_mean": 5007.1, + "valid_targets_min": 333 + }, + { + "epoch": 0.214168039538715, + "grad_norm": 0.40165976638911155, + "learning_rate": 1.2141176470588237e-05, + "loss": 0.2954, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2792291045188904, + "step": 130, + "valid_targets_mean": 6581.2, + "valid_targets_min": 1047 + }, + { + "epoch": 0.22240527182866557, + "grad_norm": 0.5413117187263645, + "learning_rate": 1.2611764705882354e-05, + "loss": 0.299, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3235633075237274, + "step": 135, + "valid_targets_mean": 5528.8, + "valid_targets_min": 361 + }, + { + "epoch": 0.23064250411861614, + "grad_norm": 0.5673717120036079, + "learning_rate": 1.3082352941176472e-05, + "loss": 0.2965, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.30562540888786316, + "step": 140, + "valid_targets_mean": 5469.2, + "valid_targets_min": 474 + }, + { + "epoch": 0.23887973640856672, + "grad_norm": 0.49097902543645666, + "learning_rate": 1.355294117647059e-05, + "loss": 0.2885, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2939978241920471, + "step": 145, + "valid_targets_mean": 5334.3, + "valid_targets_min": 260 + }, + { + "epoch": 0.2471169686985173, + "grad_norm": 0.5093548478479683, + "learning_rate": 1.4023529411764706e-05, + "loss": 0.2835, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2908037006855011, + "step": 150, + "valid_targets_mean": 6239.1, + "valid_targets_min": 336 + }, + { + "epoch": 0.2553542009884679, + "grad_norm": 0.4620067588632813, + "learning_rate": 1.4494117647058825e-05, + "loss": 0.2833, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2623613476753235, + "step": 155, + "valid_targets_mean": 4657.4, + "valid_targets_min": 319 + }, + { + "epoch": 0.26359143327841844, + "grad_norm": 0.46233214702239567, + "learning_rate": 1.4964705882352943e-05, + "loss": 0.2818, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25934308767318726, + "step": 160, + "valid_targets_mean": 5620.9, + "valid_targets_min": 365 + }, + { + "epoch": 0.27182866556836904, + "grad_norm": 0.48550710356868004, + "learning_rate": 1.543529411764706e-05, + "loss": 0.2915, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28682294487953186, + "step": 165, + "valid_targets_mean": 5591.1, + "valid_targets_min": 390 + }, + { + "epoch": 0.2800658978583196, + "grad_norm": 0.4466982509739381, + "learning_rate": 1.5905882352941177e-05, + "loss": 0.2763, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23975607752799988, + "step": 170, + "valid_targets_mean": 5620.1, + "valid_targets_min": 440 + }, + { + "epoch": 0.2883031301482702, + "grad_norm": 0.3962321896769396, + "learning_rate": 1.6376470588235298e-05, + "loss": 0.2743, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2576720416545868, + "step": 175, + "valid_targets_mean": 7407.4, + "valid_targets_min": 1483 + }, + { + "epoch": 0.29654036243822074, + "grad_norm": 0.46736148924856563, + "learning_rate": 1.6847058823529414e-05, + "loss": 0.2725, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27849751710891724, + "step": 180, + "valid_targets_mean": 6247.8, + "valid_targets_min": 373 + }, + { + "epoch": 0.30477759472817134, + "grad_norm": 0.4598062289821109, + "learning_rate": 1.731764705882353e-05, + "loss": 0.2667, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2737147808074951, + "step": 185, + "valid_targets_mean": 7175.9, + "valid_targets_min": 2045 + }, + { + "epoch": 0.3130148270181219, + "grad_norm": 0.578819497206239, + "learning_rate": 1.778823529411765e-05, + "loss": 0.2878, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2716781198978424, + "step": 190, + "valid_targets_mean": 4155.1, + "valid_targets_min": 224 + }, + { + "epoch": 0.3212520593080725, + "grad_norm": 0.9767672691031627, + "learning_rate": 1.8258823529411765e-05, + "loss": 0.2734, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2786101698875427, + "step": 195, + "valid_targets_mean": 5027.1, + "valid_targets_min": 310 + }, + { + "epoch": 0.32948929159802304, + "grad_norm": 0.4264957766087033, + "learning_rate": 1.8729411764705886e-05, + "loss": 0.2782, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25898993015289307, + "step": 200, + "valid_targets_mean": 6737.0, + "valid_targets_min": 3647 + }, + { + "epoch": 0.33772652388797364, + "grad_norm": 0.44189888957694456, + "learning_rate": 1.9200000000000003e-05, + "loss": 0.262, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24469071626663208, + "step": 205, + "valid_targets_mean": 5740.4, + "valid_targets_min": 635 + }, + { + "epoch": 0.34596375617792424, + "grad_norm": 0.5733106926699414, + "learning_rate": 1.967058823529412e-05, + "loss": 0.2695, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2947193682193756, + "step": 210, + "valid_targets_mean": 4685.6, + "valid_targets_min": 344 + }, + { + "epoch": 0.3542009884678748, + "grad_norm": 0.48099444976450817, + "learning_rate": 2.0141176470588236e-05, + "loss": 0.2745, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2802830636501312, + "step": 215, + "valid_targets_mean": 5317.8, + "valid_targets_min": 376 + }, + { + "epoch": 0.3624382207578254, + "grad_norm": 0.584239794292724, + "learning_rate": 2.0611764705882353e-05, + "loss": 0.2643, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25567081570625305, + "step": 220, + "valid_targets_mean": 5466.9, + "valid_targets_min": 1282 + }, + { + "epoch": 0.37067545304777594, + "grad_norm": 0.4745121059458066, + "learning_rate": 2.1082352941176474e-05, + "loss": 0.2671, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26192885637283325, + "step": 225, + "valid_targets_mean": 5270.5, + "valid_targets_min": 296 + }, + { + "epoch": 0.37891268533772654, + "grad_norm": 0.41969400360486403, + "learning_rate": 2.155294117647059e-05, + "loss": 0.2573, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2373993843793869, + "step": 230, + "valid_targets_mean": 6760.0, + "valid_targets_min": 326 + }, + { + "epoch": 0.3871499176276771, + "grad_norm": 0.5483205019486727, + "learning_rate": 2.2023529411764707e-05, + "loss": 0.2684, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2549372911453247, + "step": 235, + "valid_targets_mean": 5682.2, + "valid_targets_min": 308 + }, + { + "epoch": 0.3953871499176277, + "grad_norm": 0.6168666569029855, + "learning_rate": 2.2494117647058828e-05, + "loss": 0.276, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.3015628457069397, + "step": 240, + "valid_targets_mean": 3946.9, + "valid_targets_min": 414 + }, + { + "epoch": 0.40362438220757824, + "grad_norm": 0.5365792559906071, + "learning_rate": 2.296470588235294e-05, + "loss": 0.2717, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26749783754348755, + "step": 245, + "valid_targets_mean": 3893.8, + "valid_targets_min": 542 + }, + { + "epoch": 0.41186161449752884, + "grad_norm": 0.6167717661808814, + "learning_rate": 2.343529411764706e-05, + "loss": 0.25, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2404215931892395, + "step": 250, + "valid_targets_mean": 5115.6, + "valid_targets_min": 423 + }, + { + "epoch": 0.4200988467874794, + "grad_norm": 0.4913706628525373, + "learning_rate": 2.390588235294118e-05, + "loss": 0.2655, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.26305335760116577, + "step": 255, + "valid_targets_mean": 5529.3, + "valid_targets_min": 557 + }, + { + "epoch": 0.42833607907743, + "grad_norm": 0.442082406594195, + "learning_rate": 2.4376470588235296e-05, + "loss": 0.2719, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2634507715702057, + "step": 260, + "valid_targets_mean": 5923.1, + "valid_targets_min": 1430 + }, + { + "epoch": 0.43657331136738053, + "grad_norm": 0.49009125726576164, + "learning_rate": 2.4847058823529416e-05, + "loss": 0.2583, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25144073367118835, + "step": 265, + "valid_targets_mean": 5996.3, + "valid_targets_min": 3044 + }, + { + "epoch": 0.44481054365733114, + "grad_norm": 0.47334592009364207, + "learning_rate": 2.5317647058823533e-05, + "loss": 0.2556, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2636913061141968, + "step": 270, + "valid_targets_mean": 5090.6, + "valid_targets_min": 3108 + }, + { + "epoch": 0.45304777594728174, + "grad_norm": 0.8202441780411607, + "learning_rate": 2.5788235294117646e-05, + "loss": 0.2653, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2844095528125763, + "step": 275, + "valid_targets_mean": 5099.9, + "valid_targets_min": 384 + }, + { + "epoch": 0.4612850082372323, + "grad_norm": 0.44215328777494795, + "learning_rate": 2.6258823529411767e-05, + "loss": 0.2594, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2599104940891266, + "step": 280, + "valid_targets_mean": 5995.9, + "valid_targets_min": 1275 + }, + { + "epoch": 0.4695222405271829, + "grad_norm": 0.518261753355759, + "learning_rate": 2.6729411764705884e-05, + "loss": 0.2402, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23824268579483032, + "step": 285, + "valid_targets_mean": 6041.9, + "valid_targets_min": 780 + }, + { + "epoch": 0.47775947281713343, + "grad_norm": 0.5600242998987862, + "learning_rate": 2.7200000000000004e-05, + "loss": 0.2499, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24535280466079712, + "step": 290, + "valid_targets_mean": 5678.4, + "valid_targets_min": 1038 + }, + { + "epoch": 0.48599670510708404, + "grad_norm": 0.5046606384417934, + "learning_rate": 2.767058823529412e-05, + "loss": 0.2467, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23094016313552856, + "step": 295, + "valid_targets_mean": 6001.8, + "valid_targets_min": 881 + }, + { + "epoch": 0.4942339373970346, + "grad_norm": 0.4919795163540572, + "learning_rate": 2.8141176470588238e-05, + "loss": 0.2538, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2650337815284729, + "step": 300, + "valid_targets_mean": 6865.7, + "valid_targets_min": 3232 + }, + { + "epoch": 0.5024711696869851, + "grad_norm": 0.5272382038625899, + "learning_rate": 2.8611764705882358e-05, + "loss": 0.2606, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25129497051239014, + "step": 305, + "valid_targets_mean": 5451.0, + "valid_targets_min": 756 + }, + { + "epoch": 0.5107084019769358, + "grad_norm": 0.6093269737474022, + "learning_rate": 2.908235294117647e-05, + "loss": 0.2444, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25131070613861084, + "step": 310, + "valid_targets_mean": 4781.3, + "valid_targets_min": 327 + }, + { + "epoch": 0.5189456342668863, + "grad_norm": 0.4752202448157912, + "learning_rate": 2.955294117647059e-05, + "loss": 0.2489, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23877941071987152, + "step": 315, + "valid_targets_mean": 5204.8, + "valid_targets_min": 414 + }, + { + "epoch": 0.5271828665568369, + "grad_norm": 2.473505356827449, + "learning_rate": 3.002352941176471e-05, + "loss": 0.2396, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22138109803199768, + "step": 320, + "valid_targets_mean": 6812.0, + "valid_targets_min": 402 + }, + { + "epoch": 0.5354200988467874, + "grad_norm": 0.4227222126444928, + "learning_rate": 3.0494117647058826e-05, + "loss": 0.2581, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2810346782207489, + "step": 325, + "valid_targets_mean": 6907.1, + "valid_targets_min": 1919 + }, + { + "epoch": 0.5436573311367381, + "grad_norm": 0.5641877950067977, + "learning_rate": 3.0964705882352946e-05, + "loss": 0.2831, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23556017875671387, + "step": 330, + "valid_targets_mean": 3956.6, + "valid_targets_min": 284 + }, + { + "epoch": 0.5518945634266886, + "grad_norm": 0.5731856664818545, + "learning_rate": 3.143529411764706e-05, + "loss": 0.2607, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2894389033317566, + "step": 335, + "valid_targets_mean": 4187.3, + "valid_targets_min": 372 + }, + { + "epoch": 0.5601317957166392, + "grad_norm": 0.599774786273948, + "learning_rate": 3.190588235294118e-05, + "loss": 0.2503, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27301275730133057, + "step": 340, + "valid_targets_mean": 4816.8, + "valid_targets_min": 590 + }, + { + "epoch": 0.5683690280065898, + "grad_norm": 0.4861564202382807, + "learning_rate": 3.23764705882353e-05, + "loss": 0.2497, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24534930288791656, + "step": 345, + "valid_targets_mean": 5835.0, + "valid_targets_min": 273 + }, + { + "epoch": 0.5766062602965404, + "grad_norm": 0.46826059651931723, + "learning_rate": 3.2847058823529414e-05, + "loss": 0.256, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23597204685211182, + "step": 350, + "valid_targets_mean": 5569.6, + "valid_targets_min": 299 + }, + { + "epoch": 0.5848434925864909, + "grad_norm": 0.468741254368515, + "learning_rate": 3.331764705882353e-05, + "loss": 0.2447, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2362544685602188, + "step": 355, + "valid_targets_mean": 5644.9, + "valid_targets_min": 955 + }, + { + "epoch": 0.5930807248764415, + "grad_norm": 0.5115784291313441, + "learning_rate": 3.378823529411765e-05, + "loss": 0.2549, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24134746193885803, + "step": 360, + "valid_targets_mean": 4920.2, + "valid_targets_min": 549 + }, + { + "epoch": 0.6013179571663921, + "grad_norm": 0.46121075656681765, + "learning_rate": 3.425882352941177e-05, + "loss": 0.2525, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2458174228668213, + "step": 365, + "valid_targets_mean": 5673.4, + "valid_targets_min": 349 + }, + { + "epoch": 0.6095551894563427, + "grad_norm": 0.5766200034536373, + "learning_rate": 3.472941176470589e-05, + "loss": 0.2466, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2716304361820221, + "step": 370, + "valid_targets_mean": 3649.0, + "valid_targets_min": 337 + }, + { + "epoch": 0.6177924217462932, + "grad_norm": 0.43965910915542156, + "learning_rate": 3.52e-05, + "loss": 0.2451, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24214795231819153, + "step": 375, + "valid_targets_mean": 5985.5, + "valid_targets_min": 2331 + }, + { + "epoch": 0.6260296540362438, + "grad_norm": 0.5445523156203298, + "learning_rate": 3.567058823529412e-05, + "loss": 0.2497, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.27419784665107727, + "step": 380, + "valid_targets_mean": 5857.4, + "valid_targets_min": 270 + }, + { + "epoch": 0.6342668863261944, + "grad_norm": 0.45076566676047086, + "learning_rate": 3.614117647058824e-05, + "loss": 0.2359, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2390020340681076, + "step": 385, + "valid_targets_mean": 5877.6, + "valid_targets_min": 379 + }, + { + "epoch": 0.642504118616145, + "grad_norm": 0.5392434391586065, + "learning_rate": 3.6611764705882356e-05, + "loss": 0.2579, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24283994734287262, + "step": 390, + "valid_targets_mean": 4931.6, + "valid_targets_min": 315 + }, + { + "epoch": 0.6507413509060955, + "grad_norm": 0.5128687151504063, + "learning_rate": 3.708235294117647e-05, + "loss": 0.2398, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2664172649383545, + "step": 395, + "valid_targets_mean": 6280.8, + "valid_targets_min": 256 + }, + { + "epoch": 0.6589785831960461, + "grad_norm": 0.48861927361947727, + "learning_rate": 3.755294117647059e-05, + "loss": 0.2361, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22510594129562378, + "step": 400, + "valid_targets_mean": 5486.4, + "valid_targets_min": 230 + }, + { + "epoch": 0.6672158154859967, + "grad_norm": 0.4578245672892225, + "learning_rate": 3.802352941176471e-05, + "loss": 0.2488, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25764355063438416, + "step": 405, + "valid_targets_mean": 6231.9, + "valid_targets_min": 460 + }, + { + "epoch": 0.6754530477759473, + "grad_norm": 0.8301830725579878, + "learning_rate": 3.849411764705883e-05, + "loss": 0.2406, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2402649223804474, + "step": 410, + "valid_targets_mean": 3948.3, + "valid_targets_min": 242 + }, + { + "epoch": 0.6836902800658978, + "grad_norm": 0.44648575578094224, + "learning_rate": 3.8964705882352944e-05, + "loss": 0.2417, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23881413042545319, + "step": 415, + "valid_targets_mean": 5817.6, + "valid_targets_min": 323 + }, + { + "epoch": 0.6919275123558485, + "grad_norm": 0.49886875102186296, + "learning_rate": 3.943529411764706e-05, + "loss": 0.2373, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.28194451332092285, + "step": 420, + "valid_targets_mean": 6352.9, + "valid_targets_min": 1581 + }, + { + "epoch": 0.700164744645799, + "grad_norm": 0.4891568260174497, + "learning_rate": 3.990588235294118e-05, + "loss": 0.2424, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24988554418087006, + "step": 425, + "valid_targets_mean": 6102.6, + "valid_targets_min": 2769 + }, + { + "epoch": 0.7084019769357496, + "grad_norm": 0.4540233317989794, + "learning_rate": 3.9999892009992835e-05, + "loss": 0.2359, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2257465124130249, + "step": 430, + "valid_targets_mean": 6351.2, + "valid_targets_min": 671 + }, + { + "epoch": 0.7166392092257001, + "grad_norm": 0.5060649229158644, + "learning_rate": 3.999945330258739e-05, + "loss": 0.2359, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2537679672241211, + "step": 435, + "valid_targets_mean": 4834.8, + "valid_targets_min": 309 + }, + { + "epoch": 0.7248764415156508, + "grad_norm": 0.5382641331253157, + "learning_rate": 3.9998677135805034e-05, + "loss": 0.2347, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.250946968793869, + "step": 440, + "valid_targets_mean": 4850.4, + "valid_targets_min": 274 + }, + { + "epoch": 0.7331136738056013, + "grad_norm": 0.48149113904020235, + "learning_rate": 3.999756352274235e-05, + "loss": 0.2619, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24606043100357056, + "step": 445, + "valid_targets_mean": 4880.9, + "valid_targets_min": 410 + }, + { + "epoch": 0.7413509060955519, + "grad_norm": 0.40703407464849967, + "learning_rate": 3.9996112482189823e-05, + "loss": 0.2331, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2441137731075287, + "step": 450, + "valid_targets_mean": 7023.7, + "valid_targets_min": 500 + }, + { + "epoch": 0.7495881383855024, + "grad_norm": 0.567651261036972, + "learning_rate": 3.9994324038631475e-05, + "loss": 0.2333, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25954297184944153, + "step": 455, + "valid_targets_mean": 5304.1, + "valid_targets_min": 733 + }, + { + "epoch": 0.7578253706754531, + "grad_norm": 0.5007540500147702, + "learning_rate": 3.999219822224449e-05, + "loss": 0.2351, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21681588888168335, + "step": 460, + "valid_targets_mean": 4537.3, + "valid_targets_min": 319 + }, + { + "epoch": 0.7660626029654036, + "grad_norm": 0.48312602178484315, + "learning_rate": 3.998973506889867e-05, + "loss": 0.2371, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24203741550445557, + "step": 465, + "valid_targets_mean": 5523.3, + "valid_targets_min": 712 + }, + { + "epoch": 0.7742998352553542, + "grad_norm": 0.4993949797656458, + "learning_rate": 3.998693462015588e-05, + "loss": 0.2261, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23800945281982422, + "step": 470, + "valid_targets_mean": 5283.7, + "valid_targets_min": 405 + }, + { + "epoch": 0.7825370675453048, + "grad_norm": 0.6911936462179921, + "learning_rate": 3.9983796923269284e-05, + "loss": 0.2526, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2643522322177887, + "step": 475, + "valid_targets_mean": 4526.2, + "valid_targets_min": 397 + }, + { + "epoch": 0.7907742998352554, + "grad_norm": 0.46237825781467196, + "learning_rate": 3.99803220311826e-05, + "loss": 0.2321, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20834973454475403, + "step": 480, + "valid_targets_mean": 6325.9, + "valid_targets_min": 1607 + }, + { + "epoch": 0.7990115321252059, + "grad_norm": 0.5178848730466026, + "learning_rate": 3.997651000252916e-05, + "loss": 0.2325, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24416354298591614, + "step": 485, + "valid_targets_mean": 5612.9, + "valid_targets_min": 471 + }, + { + "epoch": 0.8072487644151565, + "grad_norm": 0.4804988255352975, + "learning_rate": 3.997236090163097e-05, + "loss": 0.2403, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2480747252702713, + "step": 490, + "valid_targets_mean": 5062.2, + "valid_targets_min": 278 + }, + { + "epoch": 0.8154859967051071, + "grad_norm": 0.4601940493818994, + "learning_rate": 3.99678747984976e-05, + "loss": 0.2285, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22695352137088776, + "step": 495, + "valid_targets_mean": 6248.0, + "valid_targets_min": 326 + }, + { + "epoch": 0.8237232289950577, + "grad_norm": 0.6489913836517276, + "learning_rate": 3.9963051768824996e-05, + "loss": 0.2413, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2015495002269745, + "step": 500, + "valid_targets_mean": 4947.4, + "valid_targets_min": 514 + }, + { + "epoch": 0.8319604612850082, + "grad_norm": 0.47186518041594383, + "learning_rate": 3.995789189399421e-05, + "loss": 0.2412, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2775544226169586, + "step": 505, + "valid_targets_mean": 5327.4, + "valid_targets_min": 494 + }, + { + "epoch": 0.8401976935749588, + "grad_norm": 0.49441063849501865, + "learning_rate": 3.995239526107003e-05, + "loss": 0.236, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19490137696266174, + "step": 510, + "valid_targets_mean": 5805.4, + "valid_targets_min": 375 + }, + { + "epoch": 0.8484349258649094, + "grad_norm": 0.49399739098758555, + "learning_rate": 3.99465619627995e-05, + "loss": 0.2348, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22286713123321533, + "step": 515, + "valid_targets_mean": 5033.8, + "valid_targets_min": 395 + }, + { + "epoch": 0.85667215815486, + "grad_norm": 0.47527825732717, + "learning_rate": 3.994039209761038e-05, + "loss": 0.2295, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22310678660869598, + "step": 520, + "valid_targets_mean": 4940.3, + "valid_targets_min": 363 + }, + { + "epoch": 0.8649093904448105, + "grad_norm": 0.43851822473266283, + "learning_rate": 3.993388576960947e-05, + "loss": 0.2347, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22508376836776733, + "step": 525, + "valid_targets_mean": 5517.1, + "valid_targets_min": 2534 + }, + { + "epoch": 0.8731466227347611, + "grad_norm": 0.5181505361314813, + "learning_rate": 3.9927043088580805e-05, + "loss": 0.2347, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22217413783073425, + "step": 530, + "valid_targets_mean": 5870.6, + "valid_targets_min": 2742 + }, + { + "epoch": 0.8813838550247117, + "grad_norm": 0.5172044137985335, + "learning_rate": 3.991986416998391e-05, + "loss": 0.2351, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2242281138896942, + "step": 535, + "valid_targets_mean": 4839.9, + "valid_targets_min": 383 + }, + { + "epoch": 0.8896210873146623, + "grad_norm": 0.4915178095143843, + "learning_rate": 3.991234913495177e-05, + "loss": 0.2345, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2577744722366333, + "step": 540, + "valid_targets_mean": 5669.3, + "valid_targets_min": 265 + }, + { + "epoch": 0.8978583196046128, + "grad_norm": 0.517453189663422, + "learning_rate": 3.99044981102888e-05, + "loss": 0.2409, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24107156693935394, + "step": 545, + "valid_targets_mean": 4511.5, + "valid_targets_min": 336 + }, + { + "epoch": 0.9060955518945635, + "grad_norm": 0.4249092247184588, + "learning_rate": 3.98963112284687e-05, + "loss": 0.2279, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22153885662555695, + "step": 550, + "valid_targets_mean": 5979.3, + "valid_targets_min": 1764 + }, + { + "epoch": 0.914332784184514, + "grad_norm": 0.4231497851737803, + "learning_rate": 3.9887788627632266e-05, + "loss": 0.2285, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23912326991558075, + "step": 555, + "valid_targets_mean": 6845.6, + "valid_targets_min": 501 + }, + { + "epoch": 0.9225700164744646, + "grad_norm": 0.5827210887236848, + "learning_rate": 3.9878930451584995e-05, + "loss": 0.2357, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.25541239976882935, + "step": 560, + "valid_targets_mean": 5153.3, + "valid_targets_min": 417 + }, + { + "epoch": 0.9308072487644151, + "grad_norm": 0.447268034989374, + "learning_rate": 3.986973684979471e-05, + "loss": 0.2335, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20884302258491516, + "step": 565, + "valid_targets_mean": 6352.9, + "valid_targets_min": 1201 + }, + { + "epoch": 0.9390444810543658, + "grad_norm": 0.410089047384293, + "learning_rate": 3.9860207977389e-05, + "loss": 0.2254, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.207575261592865, + "step": 570, + "valid_targets_mean": 6985.4, + "valid_targets_min": 790 + }, + { + "epoch": 0.9472817133443163, + "grad_norm": 0.7473346503345262, + "learning_rate": 3.9850343995152634e-05, + "loss": 0.2314, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23254147171974182, + "step": 575, + "valid_targets_mean": 5084.3, + "valid_targets_min": 269 + }, + { + "epoch": 0.9555189456342669, + "grad_norm": 0.48067485765275164, + "learning_rate": 3.9840145069524825e-05, + "loss": 0.2323, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22275948524475098, + "step": 580, + "valid_targets_mean": 5225.8, + "valid_targets_min": 252 + }, + { + "epoch": 0.9637561779242174, + "grad_norm": 0.5036323546289907, + "learning_rate": 3.982961137259643e-05, + "loss": 0.2259, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23761439323425293, + "step": 585, + "valid_targets_mean": 5808.1, + "valid_targets_min": 426 + }, + { + "epoch": 0.9719934102141681, + "grad_norm": 0.38418952548881075, + "learning_rate": 3.981874308210705e-05, + "loss": 0.2168, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21421602368354797, + "step": 590, + "valid_targets_mean": 6400.6, + "valid_targets_min": 335 + }, + { + "epoch": 0.9802306425041186, + "grad_norm": 0.4957357691924066, + "learning_rate": 3.980754038144202e-05, + "loss": 0.2314, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23045432567596436, + "step": 595, + "valid_targets_mean": 5057.3, + "valid_targets_min": 304 + }, + { + "epoch": 0.9884678747940692, + "grad_norm": 0.5349521308655141, + "learning_rate": 3.9796003459629335e-05, + "loss": 0.2327, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23907701671123505, + "step": 600, + "valid_targets_mean": 5696.1, + "valid_targets_min": 265 + }, + { + "epoch": 0.9967051070840197, + "grad_norm": 0.4707893596360662, + "learning_rate": 3.978413251133643e-05, + "loss": 0.2261, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22950200736522675, + "step": 605, + "valid_targets_mean": 4733.3, + "valid_targets_min": 350 + }, + { + "epoch": 1.0049423393739703, + "grad_norm": 0.41194237268820666, + "learning_rate": 3.9771927736866926e-05, + "loss": 0.234, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21946033835411072, + "step": 610, + "valid_targets_mean": 6868.2, + "valid_targets_min": 1850 + }, + { + "epoch": 1.013179571663921, + "grad_norm": 0.4825240221007433, + "learning_rate": 3.9759389342157225e-05, + "loss": 0.2187, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21633988618850708, + "step": 615, + "valid_targets_mean": 6282.6, + "valid_targets_min": 331 + }, + { + "epoch": 1.0214168039538716, + "grad_norm": 0.4864765772803356, + "learning_rate": 3.974651753877306e-05, + "loss": 0.2214, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23486509919166565, + "step": 620, + "valid_targets_mean": 5138.8, + "valid_targets_min": 362 + }, + { + "epoch": 1.029654036243822, + "grad_norm": 0.5005826843944357, + "learning_rate": 3.9733312543905904e-05, + "loss": 0.2154, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21022102236747742, + "step": 625, + "valid_targets_mean": 4148.2, + "valid_targets_min": 224 + }, + { + "epoch": 1.0378912685337727, + "grad_norm": 0.4864604603417918, + "learning_rate": 3.971977458036931e-05, + "loss": 0.2234, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22325535118579865, + "step": 630, + "valid_targets_mean": 4923.6, + "valid_targets_min": 347 + }, + { + "epoch": 1.0461285008237233, + "grad_norm": 0.4688017802978337, + "learning_rate": 3.9705903876595167e-05, + "loss": 0.2191, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2212015837430954, + "step": 635, + "valid_targets_mean": 5238.6, + "valid_targets_min": 425 + }, + { + "epoch": 1.0543657331136738, + "grad_norm": 0.428267438257124, + "learning_rate": 3.969170066662982e-05, + "loss": 0.2164, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21326056122779846, + "step": 640, + "valid_targets_mean": 6401.2, + "valid_targets_min": 2001 + }, + { + "epoch": 1.0626029654036244, + "grad_norm": 0.492908870611759, + "learning_rate": 3.967716519013014e-05, + "loss": 0.2071, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19257579743862152, + "step": 645, + "valid_targets_mean": 5080.1, + "valid_targets_min": 308 + }, + { + "epoch": 1.0708401976935749, + "grad_norm": 0.5149843564039698, + "learning_rate": 3.9662297692359486e-05, + "loss": 0.219, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2227625548839569, + "step": 650, + "valid_targets_mean": 4573.6, + "valid_targets_min": 352 + }, + { + "epoch": 1.0790774299835255, + "grad_norm": 0.7996486231537374, + "learning_rate": 3.964709842418351e-05, + "loss": 0.2201, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2152567356824875, + "step": 655, + "valid_targets_mean": 4025.2, + "valid_targets_min": 305 + }, + { + "epoch": 1.0873146622734762, + "grad_norm": 0.47476102749486415, + "learning_rate": 3.9631567642066036e-05, + "loss": 0.2062, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19000795483589172, + "step": 660, + "valid_targets_mean": 5094.8, + "valid_targets_min": 333 + }, + { + "epoch": 1.0955518945634266, + "grad_norm": 0.47703850464320063, + "learning_rate": 3.961570560806461e-05, + "loss": 0.2221, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.24766787886619568, + "step": 665, + "valid_targets_mean": 5477.7, + "valid_targets_min": 382 + }, + { + "epoch": 1.1037891268533773, + "grad_norm": 0.5069242817566683, + "learning_rate": 3.959951258982616e-05, + "loss": 0.2256, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23409639298915863, + "step": 670, + "valid_targets_mean": 5145.1, + "valid_targets_min": 325 + }, + { + "epoch": 1.112026359143328, + "grad_norm": 0.4919381443213265, + "learning_rate": 3.958298886058244e-05, + "loss": 0.2184, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.226350337266922, + "step": 675, + "valid_targets_mean": 5394.1, + "valid_targets_min": 408 + }, + { + "epoch": 1.1202635914332784, + "grad_norm": 0.5303991621418709, + "learning_rate": 3.956613469914546e-05, + "loss": 0.2236, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22548964619636536, + "step": 680, + "valid_targets_mean": 6668.1, + "valid_targets_min": 1192 + }, + { + "epoch": 1.128500823723229, + "grad_norm": 0.42339841271119033, + "learning_rate": 3.95489503899027e-05, + "loss": 0.2161, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19513583183288574, + "step": 685, + "valid_targets_mean": 5445.3, + "valid_targets_min": 325 + }, + { + "epoch": 1.1367380560131797, + "grad_norm": 0.4454529502666242, + "learning_rate": 3.9531436222812434e-05, + "loss": 0.2285, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22786834836006165, + "step": 690, + "valid_targets_mean": 6334.8, + "valid_targets_min": 385 + }, + { + "epoch": 1.1449752883031301, + "grad_norm": 0.47279587682132945, + "learning_rate": 3.951359249339872e-05, + "loss": 0.2232, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23073595762252808, + "step": 695, + "valid_targets_mean": 5592.6, + "valid_targets_min": 2403 + }, + { + "epoch": 1.1532125205930808, + "grad_norm": 0.4956575408878937, + "learning_rate": 3.9495419502746485e-05, + "loss": 0.2234, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2247496247291565, + "step": 700, + "valid_targets_mean": 4740.6, + "valid_targets_min": 265 + }, + { + "epoch": 1.1614497528830312, + "grad_norm": 0.39388974668957244, + "learning_rate": 3.9476917557496406e-05, + "loss": 0.2045, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1931227445602417, + "step": 705, + "valid_targets_mean": 6726.2, + "valid_targets_min": 3456 + }, + { + "epoch": 1.1696869851729819, + "grad_norm": 0.5233021564155736, + "learning_rate": 3.945808696983977e-05, + "loss": 0.2151, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20530879497528076, + "step": 710, + "valid_targets_mean": 4515.9, + "valid_targets_min": 368 + }, + { + "epoch": 1.1779242174629325, + "grad_norm": 0.44802130443163707, + "learning_rate": 3.943892805751318e-05, + "loss": 0.2113, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.221025288105011, + "step": 715, + "valid_targets_mean": 5391.6, + "valid_targets_min": 359 + }, + { + "epoch": 1.186161449752883, + "grad_norm": 0.5190302515381963, + "learning_rate": 3.941944114379321e-05, + "loss": 0.217, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.238490492105484, + "step": 720, + "valid_targets_mean": 4902.8, + "valid_targets_min": 320 + }, + { + "epoch": 1.1943986820428336, + "grad_norm": 0.4587014194986882, + "learning_rate": 3.939962655749092e-05, + "loss": 0.2215, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22766120731830597, + "step": 725, + "valid_targets_mean": 6048.8, + "valid_targets_min": 953 + }, + { + "epoch": 1.2026359143327843, + "grad_norm": 0.7269077468395682, + "learning_rate": 3.937948463294636e-05, + "loss": 0.2081, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20253407955169678, + "step": 730, + "valid_targets_mean": 5467.1, + "valid_targets_min": 307 + }, + { + "epoch": 1.2108731466227347, + "grad_norm": 0.47128651735720306, + "learning_rate": 3.935901571002288e-05, + "loss": 0.2086, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21129494905471802, + "step": 735, + "valid_targets_mean": 4862.2, + "valid_targets_min": 451 + }, + { + "epoch": 1.2191103789126854, + "grad_norm": 0.48356488331820596, + "learning_rate": 3.9338220134101426e-05, + "loss": 0.2156, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23600584268569946, + "step": 740, + "valid_targets_mean": 5907.7, + "valid_targets_min": 392 + }, + { + "epoch": 1.227347611202636, + "grad_norm": 0.4054882234317549, + "learning_rate": 3.931709825607469e-05, + "loss": 0.2161, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23984688520431519, + "step": 745, + "valid_targets_mean": 6646.1, + "valid_targets_min": 270 + }, + { + "epoch": 1.2355848434925865, + "grad_norm": 0.5529526522103348, + "learning_rate": 3.9295650432341195e-05, + "loss": 0.2164, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2136891484260559, + "step": 750, + "valid_targets_mean": 4384.1, + "valid_targets_min": 281 + }, + { + "epoch": 1.2438220757825371, + "grad_norm": 0.44931791363596874, + "learning_rate": 3.927387702479931e-05, + "loss": 0.22, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2048332244157791, + "step": 755, + "valid_targets_mean": 4977.4, + "valid_targets_min": 260 + }, + { + "epoch": 1.2520593080724876, + "grad_norm": 0.5100825061887327, + "learning_rate": 3.925177840084109e-05, + "loss": 0.2163, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2232269048690796, + "step": 760, + "valid_targets_mean": 5746.7, + "valid_targets_min": 766 + }, + { + "epoch": 1.2602965403624382, + "grad_norm": 0.46978923825758434, + "learning_rate": 3.922935493334614e-05, + "loss": 0.2057, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20063738524913788, + "step": 765, + "valid_targets_mean": 5882.4, + "valid_targets_min": 805 + }, + { + "epoch": 1.2685337726523889, + "grad_norm": 0.4468454526508097, + "learning_rate": 3.920660700067526e-05, + "loss": 0.2093, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21785524487495422, + "step": 770, + "valid_targets_mean": 5499.2, + "valid_targets_min": 2014 + }, + { + "epoch": 1.2767710049423393, + "grad_norm": 0.40158310809679576, + "learning_rate": 3.9183534986664094e-05, + "loss": 0.212, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22869275510311127, + "step": 775, + "valid_targets_mean": 6712.8, + "valid_targets_min": 332 + }, + { + "epoch": 1.28500823723229, + "grad_norm": 0.45415318965640566, + "learning_rate": 3.9160139280616674e-05, + "loss": 0.2209, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21420571208000183, + "step": 780, + "valid_targets_mean": 4906.4, + "valid_targets_min": 2072 + }, + { + "epoch": 1.2932454695222404, + "grad_norm": 0.5581765063544478, + "learning_rate": 3.9136420277298795e-05, + "loss": 0.2207, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2164240926504135, + "step": 785, + "valid_targets_mean": 4428.5, + "valid_targets_min": 370 + }, + { + "epoch": 1.301482701812191, + "grad_norm": 0.49806923271667086, + "learning_rate": 3.911237837693141e-05, + "loss": 0.2039, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21554507315158844, + "step": 790, + "valid_targets_mean": 4793.6, + "valid_targets_min": 230 + }, + { + "epoch": 1.3097199341021417, + "grad_norm": 0.45786660001198226, + "learning_rate": 3.9088013985183836e-05, + "loss": 0.2152, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22300955653190613, + "step": 795, + "valid_targets_mean": 5414.1, + "valid_targets_min": 2898 + }, + { + "epoch": 1.3179571663920924, + "grad_norm": 0.4395544205039379, + "learning_rate": 3.9063327513166945e-05, + "loss": 0.2196, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22037920355796814, + "step": 800, + "valid_targets_mean": 4776.6, + "valid_targets_min": 338 + }, + { + "epoch": 1.3261943986820428, + "grad_norm": 0.3795581984916046, + "learning_rate": 3.9038319377426186e-05, + "loss": 0.2172, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22801026701927185, + "step": 805, + "valid_targets_mean": 7366.5, + "valid_targets_min": 3706 + }, + { + "epoch": 1.3344316309719935, + "grad_norm": 0.3818588596696933, + "learning_rate": 3.9012989999934595e-05, + "loss": 0.2062, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18441705405712128, + "step": 810, + "valid_targets_mean": 6655.9, + "valid_targets_min": 269 + }, + { + "epoch": 1.342668863261944, + "grad_norm": 0.42620581429826004, + "learning_rate": 3.898733980808565e-05, + "loss": 0.2149, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19475027918815613, + "step": 815, + "valid_targets_mean": 5158.4, + "valid_targets_min": 716 + }, + { + "epoch": 1.3509060955518946, + "grad_norm": 0.4984329967854055, + "learning_rate": 3.896136923468607e-05, + "loss": 0.2187, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20715723931789398, + "step": 820, + "valid_targets_mean": 7209.1, + "valid_targets_min": 2177 + }, + { + "epoch": 1.3591433278418452, + "grad_norm": 0.41716974384175576, + "learning_rate": 3.8935078717948507e-05, + "loss": 0.2088, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20283842086791992, + "step": 825, + "valid_targets_mean": 6637.9, + "valid_targets_min": 321 + }, + { + "epoch": 1.3673805601317957, + "grad_norm": 0.44915758841416215, + "learning_rate": 3.890846870148416e-05, + "loss": 0.2047, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20052731037139893, + "step": 830, + "valid_targets_mean": 4960.1, + "valid_targets_min": 381 + }, + { + "epoch": 1.3756177924217463, + "grad_norm": 0.5013943897191635, + "learning_rate": 3.8881539634295287e-05, + "loss": 0.2236, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20464956760406494, + "step": 835, + "valid_targets_mean": 4348.9, + "valid_targets_min": 274 + }, + { + "epoch": 1.3838550247116967, + "grad_norm": 0.5068964263720004, + "learning_rate": 3.88542919707676e-05, + "loss": 0.212, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20616479218006134, + "step": 840, + "valid_targets_mean": 4562.9, + "valid_targets_min": 405 + }, + { + "epoch": 1.3920922570016474, + "grad_norm": 0.47570489286531875, + "learning_rate": 3.882672617066266e-05, + "loss": 0.2145, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20461995899677277, + "step": 845, + "valid_targets_mean": 4907.6, + "valid_targets_min": 476 + }, + { + "epoch": 1.400329489291598, + "grad_norm": 0.4025399969142752, + "learning_rate": 3.8798842699110075e-05, + "loss": 0.2118, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20368187129497528, + "step": 850, + "valid_targets_mean": 6078.8, + "valid_targets_min": 253 + }, + { + "epoch": 1.4085667215815487, + "grad_norm": 0.6119611882715695, + "learning_rate": 3.877064202659964e-05, + "loss": 0.223, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2284880429506302, + "step": 855, + "valid_targets_mean": 5368.0, + "valid_targets_min": 310 + }, + { + "epoch": 1.4168039538714992, + "grad_norm": 0.48313598417744963, + "learning_rate": 3.874212462897343e-05, + "loss": 0.2012, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21637935936450958, + "step": 860, + "valid_targets_mean": 5340.4, + "valid_targets_min": 385 + }, + { + "epoch": 1.4250411861614498, + "grad_norm": 0.5025920361955611, + "learning_rate": 3.871329098741776e-05, + "loss": 0.2073, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23110291361808777, + "step": 865, + "valid_targets_mean": 3986.9, + "valid_targets_min": 319 + }, + { + "epoch": 1.4332784184514002, + "grad_norm": 0.44545873367031735, + "learning_rate": 3.8684141588455076e-05, + "loss": 0.2147, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2106075882911682, + "step": 870, + "valid_targets_mean": 5611.4, + "valid_targets_min": 493 + }, + { + "epoch": 1.441515650741351, + "grad_norm": 0.4107282913016226, + "learning_rate": 3.865467692393571e-05, + "loss": 0.2066, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21611574292182922, + "step": 875, + "valid_targets_mean": 5921.5, + "valid_targets_min": 964 + }, + { + "epoch": 1.4497528830313016, + "grad_norm": 0.4790434208811441, + "learning_rate": 3.862489749102963e-05, + "loss": 0.209, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20723897218704224, + "step": 880, + "valid_targets_mean": 4724.5, + "valid_targets_min": 309 + }, + { + "epoch": 1.457990115321252, + "grad_norm": 0.4091069480001445, + "learning_rate": 3.8594803792218e-05, + "loss": 0.2095, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20824208855628967, + "step": 885, + "valid_targets_mean": 5924.8, + "valid_targets_min": 1634 + }, + { + "epoch": 1.4662273476112027, + "grad_norm": 0.5379923390169834, + "learning_rate": 3.856439633528476e-05, + "loss": 0.2069, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2114851176738739, + "step": 890, + "valid_targets_mean": 5122.6, + "valid_targets_min": 362 + }, + { + "epoch": 1.474464579901153, + "grad_norm": 0.4108382282606267, + "learning_rate": 3.853367563330799e-05, + "loss": 0.2146, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21595284342765808, + "step": 895, + "valid_targets_mean": 6272.7, + "valid_targets_min": 1989 + }, + { + "epoch": 1.4827018121911038, + "grad_norm": 0.4556536946526364, + "learning_rate": 3.850264220465134e-05, + "loss": 0.1994, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21706634759902954, + "step": 900, + "valid_targets_mean": 5730.1, + "valid_targets_min": 752 + }, + { + "epoch": 1.4909390444810544, + "grad_norm": 0.4301435195930705, + "learning_rate": 3.847129657295516e-05, + "loss": 0.202, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20228904485702515, + "step": 905, + "valid_targets_mean": 5140.6, + "valid_targets_min": 388 + }, + { + "epoch": 1.499176276771005, + "grad_norm": 0.40466531702536584, + "learning_rate": 3.843963926712778e-05, + "loss": 0.2018, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20213758945465088, + "step": 910, + "valid_targets_mean": 6237.1, + "valid_targets_min": 2286 + }, + { + "epoch": 1.5074135090609555, + "grad_norm": 0.4252404187833234, + "learning_rate": 3.840767082133656e-05, + "loss": 0.2204, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1813409924507141, + "step": 915, + "valid_targets_mean": 5218.1, + "valid_targets_min": 311 + }, + { + "epoch": 1.515650741350906, + "grad_norm": 0.4544865201227028, + "learning_rate": 3.837539177499882e-05, + "loss": 0.208, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21020415425300598, + "step": 920, + "valid_targets_mean": 5514.9, + "valid_targets_min": 303 + }, + { + "epoch": 1.5238879736408566, + "grad_norm": 0.4220823729123365, + "learning_rate": 3.834280267277279e-05, + "loss": 0.2143, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18695414066314697, + "step": 925, + "valid_targets_mean": 4791.7, + "valid_targets_min": 369 + }, + { + "epoch": 1.5321252059308073, + "grad_norm": 0.44549970172652165, + "learning_rate": 3.830990406454842e-05, + "loss": 0.2059, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2096899449825287, + "step": 930, + "valid_targets_mean": 5779.6, + "valid_targets_min": 357 + }, + { + "epoch": 1.540362438220758, + "grad_norm": 0.4509202561250749, + "learning_rate": 3.827669650543805e-05, + "loss": 0.2134, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20951557159423828, + "step": 935, + "valid_targets_mean": 5422.6, + "valid_targets_min": 442 + }, + { + "epoch": 1.5485996705107083, + "grad_norm": 0.43475972027953813, + "learning_rate": 3.8243180555767135e-05, + "loss": 0.2196, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22084136307239532, + "step": 940, + "valid_targets_mean": 5615.9, + "valid_targets_min": 1337 + }, + { + "epoch": 1.556836902800659, + "grad_norm": 0.4231693225032134, + "learning_rate": 3.8209356781064686e-05, + "loss": 0.2185, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22242343425750732, + "step": 945, + "valid_targets_mean": 6372.5, + "valid_targets_min": 326 + }, + { + "epoch": 1.5650741350906094, + "grad_norm": 0.4315275861397971, + "learning_rate": 3.8175225752053816e-05, + "loss": 0.2191, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2107992321252823, + "step": 950, + "valid_targets_mean": 5179.2, + "valid_targets_min": 379 + }, + { + "epoch": 1.57331136738056, + "grad_norm": 0.4255890164918819, + "learning_rate": 3.814078804464205e-05, + "loss": 0.2138, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2127227485179901, + "step": 955, + "valid_targets_mean": 5411.9, + "valid_targets_min": 1032 + }, + { + "epoch": 1.5815485996705108, + "grad_norm": 0.48605730492655386, + "learning_rate": 3.8106044239911645e-05, + "loss": 0.2139, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20564338564872742, + "step": 960, + "valid_targets_mean": 5263.7, + "valid_targets_min": 384 + }, + { + "epoch": 1.5897858319604614, + "grad_norm": 0.40316591610709385, + "learning_rate": 3.8070994924109746e-05, + "loss": 0.2043, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20632685720920563, + "step": 965, + "valid_targets_mean": 6254.6, + "valid_targets_min": 364 + }, + { + "epoch": 1.5980230642504119, + "grad_norm": 0.4086555388604155, + "learning_rate": 3.803564068863854e-05, + "loss": 0.1907, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19364306330680847, + "step": 970, + "valid_targets_mean": 6633.9, + "valid_targets_min": 3712 + }, + { + "epoch": 1.6062602965403623, + "grad_norm": 0.38782328074430267, + "learning_rate": 3.7999982130045246e-05, + "loss": 0.2016, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2168004810810089, + "step": 975, + "valid_targets_mean": 6655.2, + "valid_targets_min": 3691 + }, + { + "epoch": 1.614497528830313, + "grad_norm": 0.4330948997917468, + "learning_rate": 3.796401985001206e-05, + "loss": 0.215, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2149375081062317, + "step": 980, + "valid_targets_mean": 6287.2, + "valid_targets_min": 318 + }, + { + "epoch": 1.6227347611202636, + "grad_norm": 0.4733788567450916, + "learning_rate": 3.792775445534601e-05, + "loss": 0.2095, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19296729564666748, + "step": 985, + "valid_targets_mean": 5054.3, + "valid_targets_min": 335 + }, + { + "epoch": 1.6309719934102143, + "grad_norm": 0.458391192134762, + "learning_rate": 3.7891186557968665e-05, + "loss": 0.2034, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21453091502189636, + "step": 990, + "valid_targets_mean": 6577.2, + "valid_targets_min": 4384 + }, + { + "epoch": 1.6392092257001647, + "grad_norm": 0.4743187663655409, + "learning_rate": 3.785431677490592e-05, + "loss": 0.1971, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17525094747543335, + "step": 995, + "valid_targets_mean": 6421.0, + "valid_targets_min": 1150 + }, + { + "epoch": 1.6474464579901154, + "grad_norm": 0.5108222541349242, + "learning_rate": 3.781714572827746e-05, + "loss": 0.2161, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20602372288703918, + "step": 1000, + "valid_targets_mean": 4063.1, + "valid_targets_min": 345 + }, + { + "epoch": 1.6556836902800658, + "grad_norm": 0.42592240451237107, + "learning_rate": 3.777967404528636e-05, + "loss": 0.2089, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21424970030784607, + "step": 1005, + "valid_targets_mean": 6034.2, + "valid_targets_min": 407 + }, + { + "epoch": 1.6639209225700164, + "grad_norm": 0.40989244289770793, + "learning_rate": 3.774190235820843e-05, + "loss": 0.2194, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19179269671440125, + "step": 1010, + "valid_targets_mean": 5793.4, + "valid_targets_min": 373 + }, + { + "epoch": 1.672158154859967, + "grad_norm": 0.42689685330038074, + "learning_rate": 3.770383130438159e-05, + "loss": 0.2129, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19655553996562958, + "step": 1015, + "valid_targets_mean": 5861.7, + "valid_targets_min": 309 + }, + { + "epoch": 1.6803953871499178, + "grad_norm": 0.5214662975794156, + "learning_rate": 3.76654615261951e-05, + "loss": 0.2079, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22449663281440735, + "step": 1020, + "valid_targets_mean": 5074.8, + "valid_targets_min": 298 + }, + { + "epoch": 1.6886326194398682, + "grad_norm": 0.394367346520587, + "learning_rate": 3.7626793671078743e-05, + "loss": 0.2154, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18750452995300293, + "step": 1025, + "valid_targets_mean": 6294.2, + "valid_targets_min": 301 + }, + { + "epoch": 1.6968698517298186, + "grad_norm": 0.6127740131059339, + "learning_rate": 3.758782839149187e-05, + "loss": 0.2, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18330290913581848, + "step": 1030, + "valid_targets_mean": 4697.9, + "valid_targets_min": 315 + }, + { + "epoch": 1.7051070840197693, + "grad_norm": 0.4847112036985494, + "learning_rate": 3.754856634491238e-05, + "loss": 0.2018, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1978096067905426, + "step": 1035, + "valid_targets_mean": 5872.2, + "valid_targets_min": 233 + }, + { + "epoch": 1.71334431630972, + "grad_norm": 0.5058791546645084, + "learning_rate": 3.75090081938257e-05, + "loss": 0.2188, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23929041624069214, + "step": 1040, + "valid_targets_mean": 5497.1, + "valid_targets_min": 459 + }, + { + "epoch": 1.7215815485996706, + "grad_norm": 0.41712531401074715, + "learning_rate": 3.7469154605713496e-05, + "loss": 0.204, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18083135783672333, + "step": 1045, + "valid_targets_mean": 5281.0, + "valid_targets_min": 372 + }, + { + "epoch": 1.729818780889621, + "grad_norm": 0.45224754824628705, + "learning_rate": 3.742900625304252e-05, + "loss": 0.2046, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20000770688056946, + "step": 1050, + "valid_targets_mean": 5335.8, + "valid_targets_min": 442 + }, + { + "epoch": 1.7380560131795717, + "grad_norm": 0.4276669151075953, + "learning_rate": 3.7388563813253194e-05, + "loss": 0.2026, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21081122756004333, + "step": 1055, + "valid_targets_mean": 5431.8, + "valid_targets_min": 357 + }, + { + "epoch": 1.7462932454695221, + "grad_norm": 0.42270391290398784, + "learning_rate": 3.7347827968748196e-05, + "loss": 0.2142, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21518349647521973, + "step": 1060, + "valid_targets_mean": 4905.4, + "valid_targets_min": 361 + }, + { + "epoch": 1.7545304777594728, + "grad_norm": 0.4938586648526237, + "learning_rate": 3.730679940688094e-05, + "loss": 0.213, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2053917944431305, + "step": 1065, + "valid_targets_mean": 5753.6, + "valid_targets_min": 549 + }, + { + "epoch": 1.7627677100494235, + "grad_norm": 0.41970694178562035, + "learning_rate": 3.7265478819944004e-05, + "loss": 0.211, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2033730149269104, + "step": 1070, + "valid_targets_mean": 5415.0, + "valid_targets_min": 331 + }, + { + "epoch": 1.771004942339374, + "grad_norm": 0.5513001892847159, + "learning_rate": 3.722386690515742e-05, + "loss": 0.2144, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2219332456588745, + "step": 1075, + "valid_targets_mean": 4620.9, + "valid_targets_min": 443 + }, + { + "epoch": 1.7792421746293245, + "grad_norm": 0.5136327419552082, + "learning_rate": 3.718196436465693e-05, + "loss": 0.206, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19483298063278198, + "step": 1080, + "valid_targets_mean": 5695.8, + "valid_targets_min": 301 + }, + { + "epoch": 1.787479406919275, + "grad_norm": 0.46368358882142574, + "learning_rate": 3.713977190548211e-05, + "loss": 0.2152, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2157488614320755, + "step": 1085, + "valid_targets_mean": 5854.6, + "valid_targets_min": 1771 + }, + { + "epoch": 1.7957166392092256, + "grad_norm": 0.4376783891180776, + "learning_rate": 3.709729023956448e-05, + "loss": 0.208, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20793405175209045, + "step": 1090, + "valid_targets_mean": 5121.4, + "valid_targets_min": 306 + }, + { + "epoch": 1.8039538714991763, + "grad_norm": 0.6020766343191308, + "learning_rate": 3.705452008371545e-05, + "loss": 0.2072, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20232078433036804, + "step": 1095, + "valid_targets_mean": 5209.9, + "valid_targets_min": 594 + }, + { + "epoch": 1.812191103789127, + "grad_norm": 0.3929612039377728, + "learning_rate": 3.701146215961426e-05, + "loss": 0.2007, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1929633766412735, + "step": 1100, + "valid_targets_mean": 6037.1, + "valid_targets_min": 2363 + }, + { + "epoch": 1.8204283360790774, + "grad_norm": 0.47583432200204007, + "learning_rate": 3.696811719379578e-05, + "loss": 0.1964, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19097734987735748, + "step": 1105, + "valid_targets_mean": 5827.8, + "valid_targets_min": 337 + }, + { + "epoch": 1.828665568369028, + "grad_norm": 0.42084059788905986, + "learning_rate": 3.692448591763827e-05, + "loss": 0.2096, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20220178365707397, + "step": 1110, + "valid_targets_mean": 5310.2, + "valid_targets_min": 414 + }, + { + "epoch": 1.8369028006589785, + "grad_norm": 0.621033541298011, + "learning_rate": 3.6880569067350994e-05, + "loss": 0.2113, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2220468968153, + "step": 1115, + "valid_targets_mean": 5355.2, + "valid_targets_min": 967 + }, + { + "epoch": 1.8451400329489291, + "grad_norm": 0.5417180016003424, + "learning_rate": 3.683636738396187e-05, + "loss": 0.215, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22816117107868195, + "step": 1120, + "valid_targets_mean": 3872.6, + "valid_targets_min": 298 + }, + { + "epoch": 1.8533772652388798, + "grad_norm": 0.39419195825057896, + "learning_rate": 3.6791881613304916e-05, + "loss": 0.199, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1901552528142929, + "step": 1125, + "valid_targets_mean": 6075.9, + "valid_targets_min": 364 + }, + { + "epoch": 1.8616144975288305, + "grad_norm": 0.48439504999155003, + "learning_rate": 3.674711250600766e-05, + "loss": 0.2356, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23211458325386047, + "step": 1130, + "valid_targets_mean": 4931.1, + "valid_targets_min": 309 + }, + { + "epoch": 1.869851729818781, + "grad_norm": 0.4566773154514815, + "learning_rate": 3.6702060817478524e-05, + "loss": 0.2082, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20753763616085052, + "step": 1135, + "valid_targets_mean": 5041.4, + "valid_targets_min": 576 + }, + { + "epoch": 1.8780889621087313, + "grad_norm": 0.4206786715682531, + "learning_rate": 3.6656727307894e-05, + "loss": 0.2028, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2010621726512909, + "step": 1140, + "valid_targets_mean": 5559.8, + "valid_targets_min": 344 + }, + { + "epoch": 1.886326194398682, + "grad_norm": 0.4398821973871485, + "learning_rate": 3.6611112742185913e-05, + "loss": 0.2045, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21115025877952576, + "step": 1145, + "valid_targets_mean": 4765.1, + "valid_targets_min": 755 + }, + { + "epoch": 1.8945634266886326, + "grad_norm": 0.446473805886963, + "learning_rate": 3.6565217890028435e-05, + "loss": 0.2197, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19354012608528137, + "step": 1150, + "valid_targets_mean": 5087.8, + "valid_targets_min": 295 + }, + { + "epoch": 1.9028006589785833, + "grad_norm": 0.37804036389153517, + "learning_rate": 3.651904352582514e-05, + "loss": 0.2072, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19648414850234985, + "step": 1155, + "valid_targets_mean": 6540.2, + "valid_targets_min": 356 + }, + { + "epoch": 1.9110378912685337, + "grad_norm": 0.6203253075668428, + "learning_rate": 3.647259042869595e-05, + "loss": 0.2186, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23002168536186218, + "step": 1160, + "valid_targets_mean": 4733.4, + "valid_targets_min": 323 + }, + { + "epoch": 1.9192751235584844, + "grad_norm": 0.38880529466871483, + "learning_rate": 3.642585938246394e-05, + "loss": 0.1957, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19138900935649872, + "step": 1165, + "valid_targets_mean": 6324.9, + "valid_targets_min": 373 + }, + { + "epoch": 1.9275123558484348, + "grad_norm": 0.42675443770878757, + "learning_rate": 3.6378851175642166e-05, + "loss": 0.2018, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20937415957450867, + "step": 1170, + "valid_targets_mean": 5846.4, + "valid_targets_min": 360 + }, + { + "epoch": 1.9357495881383855, + "grad_norm": 0.4296389983820003, + "learning_rate": 3.63315666014203e-05, + "loss": 0.2071, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19081468880176544, + "step": 1175, + "valid_targets_mean": 5246.9, + "valid_targets_min": 336 + }, + { + "epoch": 1.9439868204283361, + "grad_norm": 0.5039473970395317, + "learning_rate": 3.6284006457651295e-05, + "loss": 0.214, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19345268607139587, + "step": 1180, + "valid_targets_mean": 5206.6, + "valid_targets_min": 760 + }, + { + "epoch": 1.9522240527182868, + "grad_norm": 0.4820224223402811, + "learning_rate": 3.6236171546837936e-05, + "loss": 0.2022, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.22416003048419952, + "step": 1185, + "valid_targets_mean": 4467.2, + "valid_targets_min": 1428 + }, + { + "epoch": 1.9604612850082372, + "grad_norm": 0.4511616532068528, + "learning_rate": 3.618806267611923e-05, + "loss": 0.211, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21213218569755554, + "step": 1190, + "valid_targets_mean": 5811.9, + "valid_targets_min": 414 + }, + { + "epoch": 1.9686985172981877, + "grad_norm": 0.4572207471838012, + "learning_rate": 3.613968065725685e-05, + "loss": 0.2094, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23261676728725433, + "step": 1195, + "valid_targets_mean": 6750.7, + "valid_targets_min": 3184 + }, + { + "epoch": 1.9769357495881383, + "grad_norm": 0.37778737740814883, + "learning_rate": 3.60910263066214e-05, + "loss": 0.2042, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2077624648809433, + "step": 1200, + "valid_targets_mean": 6281.2, + "valid_targets_min": 309 + }, + { + "epoch": 1.985172981878089, + "grad_norm": 0.44767855166484793, + "learning_rate": 3.604210044517867e-05, + "loss": 0.1944, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19422310590744019, + "step": 1205, + "valid_targets_mean": 4698.5, + "valid_targets_min": 366 + }, + { + "epoch": 1.9934102141680397, + "grad_norm": 0.4453341741888798, + "learning_rate": 3.599290389847576e-05, + "loss": 0.2063, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2169584333896637, + "step": 1210, + "valid_targets_mean": 6427.6, + "valid_targets_min": 1029 + }, + { + "epoch": 2.0016474464579903, + "grad_norm": 0.4691655308778174, + "learning_rate": 3.594343749662716e-05, + "loss": 0.1958, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16686594486236572, + "step": 1215, + "valid_targets_mean": 5029.6, + "valid_targets_min": 250 + }, + { + "epoch": 2.0098846787479405, + "grad_norm": 0.49854351226994864, + "learning_rate": 3.5893702074300745e-05, + "loss": 0.1902, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19464629888534546, + "step": 1220, + "valid_targets_mean": 5672.7, + "valid_targets_min": 2131 + }, + { + "epoch": 2.018121911037891, + "grad_norm": 0.4964318584547705, + "learning_rate": 3.584369847070369e-05, + "loss": 0.1925, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21309590339660645, + "step": 1225, + "valid_targets_mean": 5729.2, + "valid_targets_min": 393 + }, + { + "epoch": 2.026359143327842, + "grad_norm": 0.4023341001155962, + "learning_rate": 3.579342752956829e-05, + "loss": 0.1915, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1911289393901825, + "step": 1230, + "valid_targets_mean": 6125.6, + "valid_targets_min": 2194 + }, + { + "epoch": 2.0345963756177925, + "grad_norm": 0.46394946737605847, + "learning_rate": 3.574289009913776e-05, + "loss": 0.1993, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1845487356185913, + "step": 1235, + "valid_targets_mean": 5679.7, + "valid_targets_min": 355 + }, + { + "epoch": 2.042833607907743, + "grad_norm": 0.4215099447257407, + "learning_rate": 3.5692087032151915e-05, + "loss": 0.1806, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1543101668357849, + "step": 1240, + "valid_targets_mean": 4842.9, + "valid_targets_min": 453 + }, + { + "epoch": 2.0510708401976934, + "grad_norm": 0.4781452383124886, + "learning_rate": 3.564101918583274e-05, + "loss": 0.183, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18705658614635468, + "step": 1245, + "valid_targets_mean": 4635.4, + "valid_targets_min": 366 + }, + { + "epoch": 2.059308072487644, + "grad_norm": 0.4242016530032853, + "learning_rate": 3.5589687421869964e-05, + "loss": 0.1811, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1837356686592102, + "step": 1250, + "valid_targets_mean": 6262.1, + "valid_targets_min": 1453 + }, + { + "epoch": 2.0675453047775947, + "grad_norm": 0.47101868005036085, + "learning_rate": 3.553809260640653e-05, + "loss": 0.1855, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17792844772338867, + "step": 1255, + "valid_targets_mean": 5523.5, + "valid_targets_min": 307 + }, + { + "epoch": 2.0757825370675453, + "grad_norm": 0.44058326488327854, + "learning_rate": 3.5486235610023935e-05, + "loss": 0.1768, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17145931720733643, + "step": 1260, + "valid_targets_mean": 5019.3, + "valid_targets_min": 320 + }, + { + "epoch": 2.084019769357496, + "grad_norm": 0.4720538729006096, + "learning_rate": 3.54341173077276e-05, + "loss": 0.2022, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19867974519729614, + "step": 1265, + "valid_targets_mean": 5480.0, + "valid_targets_min": 386 + }, + { + "epoch": 2.0922570016474467, + "grad_norm": 0.49979798033678424, + "learning_rate": 3.5381738578932035e-05, + "loss": 0.1881, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19164550304412842, + "step": 1270, + "valid_targets_mean": 5641.0, + "valid_targets_min": 344 + }, + { + "epoch": 2.100494233937397, + "grad_norm": 0.5126675244916216, + "learning_rate": 3.532910030744607e-05, + "loss": 0.1838, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17244023084640503, + "step": 1275, + "valid_targets_mean": 5319.9, + "valid_targets_min": 335 + }, + { + "epoch": 2.1087314662273475, + "grad_norm": 0.4510430101034234, + "learning_rate": 3.5276203381457894e-05, + "loss": 0.1832, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18565404415130615, + "step": 1280, + "valid_targets_mean": 6039.1, + "valid_targets_min": 672 + }, + { + "epoch": 2.116968698517298, + "grad_norm": 0.4308787285559032, + "learning_rate": 3.5223048693520094e-05, + "loss": 0.185, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18874096870422363, + "step": 1285, + "valid_targets_mean": 5633.6, + "valid_targets_min": 484 + }, + { + "epoch": 2.125205930807249, + "grad_norm": 0.468459853247736, + "learning_rate": 3.5169637140534574e-05, + "loss": 0.1858, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19510608911514282, + "step": 1290, + "valid_targets_mean": 5250.2, + "valid_targets_min": 589 + }, + { + "epoch": 2.1334431630971995, + "grad_norm": 0.4658684095500768, + "learning_rate": 3.5115969623737447e-05, + "loss": 0.1957, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18428108096122742, + "step": 1295, + "valid_targets_mean": 5244.8, + "valid_targets_min": 390 + }, + { + "epoch": 2.1416803953871497, + "grad_norm": 0.4137733677060094, + "learning_rate": 3.506204704868379e-05, + "loss": 0.1905, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1802988052368164, + "step": 1300, + "valid_targets_mean": 5942.2, + "valid_targets_min": 2248 + }, + { + "epoch": 2.1499176276771004, + "grad_norm": 0.4398392014709175, + "learning_rate": 3.500787032523242e-05, + "loss": 0.1945, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20117981731891632, + "step": 1305, + "valid_targets_mean": 5333.5, + "valid_targets_min": 351 + }, + { + "epoch": 2.158154859967051, + "grad_norm": 0.41011198319086234, + "learning_rate": 3.495344036753048e-05, + "loss": 0.204, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18782642483711243, + "step": 1310, + "valid_targets_mean": 6204.0, + "valid_targets_min": 332 + }, + { + "epoch": 2.1663920922570017, + "grad_norm": 0.4747376443014503, + "learning_rate": 3.489875809399806e-05, + "loss": 0.1801, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18644480407238007, + "step": 1315, + "valid_targets_mean": 5435.4, + "valid_targets_min": 341 + }, + { + "epoch": 2.1746293245469523, + "grad_norm": 0.41595362322660934, + "learning_rate": 3.484382442731268e-05, + "loss": 0.1878, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1779285967350006, + "step": 1320, + "valid_targets_mean": 6901.5, + "valid_targets_min": 3389 + }, + { + "epoch": 2.182866556836903, + "grad_norm": 0.40494315889864735, + "learning_rate": 3.4788640294393725e-05, + "loss": 0.1843, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19018739461898804, + "step": 1325, + "valid_targets_mean": 7549.8, + "valid_targets_min": 348 + }, + { + "epoch": 2.191103789126853, + "grad_norm": 0.4280136085385474, + "learning_rate": 3.47332066263868e-05, + "loss": 0.1872, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18496966361999512, + "step": 1330, + "valid_targets_mean": 5089.6, + "valid_targets_min": 390 + }, + { + "epoch": 2.199341021416804, + "grad_norm": 0.5025570901923755, + "learning_rate": 3.467752435864803e-05, + "loss": 0.1818, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16634967923164368, + "step": 1335, + "valid_targets_mean": 4767.1, + "valid_targets_min": 468 + }, + { + "epoch": 2.2075782537067545, + "grad_norm": 0.4025390072839639, + "learning_rate": 3.462159443072826e-05, + "loss": 0.187, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17967519164085388, + "step": 1340, + "valid_targets_mean": 6592.0, + "valid_targets_min": 294 + }, + { + "epoch": 2.215815485996705, + "grad_norm": 0.5084259282496368, + "learning_rate": 3.456541778635722e-05, + "loss": 0.2037, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2011231929063797, + "step": 1345, + "valid_targets_mean": 5602.5, + "valid_targets_min": 338 + }, + { + "epoch": 2.224052718286656, + "grad_norm": 0.6204597597615517, + "learning_rate": 3.45089953734276e-05, + "loss": 0.1904, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.187906876206398, + "step": 1350, + "valid_targets_mean": 5300.2, + "valid_targets_min": 502 + }, + { + "epoch": 2.232289950576606, + "grad_norm": 0.43339460655750894, + "learning_rate": 3.445232814397902e-05, + "loss": 0.1934, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18730773031711578, + "step": 1355, + "valid_targets_mean": 5800.6, + "valid_targets_min": 2008 + }, + { + "epoch": 2.2405271828665567, + "grad_norm": 0.49437914997453064, + "learning_rate": 3.4395417054182034e-05, + "loss": 0.1815, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1944451928138733, + "step": 1360, + "valid_targets_mean": 5698.8, + "valid_targets_min": 650 + }, + { + "epoch": 2.2487644151565074, + "grad_norm": 0.4679757922035915, + "learning_rate": 3.433826306432193e-05, + "loss": 0.1955, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1887608915567398, + "step": 1365, + "valid_targets_mean": 5416.2, + "valid_targets_min": 451 + }, + { + "epoch": 2.257001647446458, + "grad_norm": 0.4833050563367706, + "learning_rate": 3.428086713878255e-05, + "loss": 0.1916, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18888002634048462, + "step": 1370, + "valid_targets_mean": 5162.8, + "valid_targets_min": 561 + }, + { + "epoch": 2.2652388797364087, + "grad_norm": 0.9103976886906952, + "learning_rate": 3.422323024603004e-05, + "loss": 0.1872, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1994047462940216, + "step": 1375, + "valid_targets_mean": 4683.3, + "valid_targets_min": 836 + }, + { + "epoch": 2.2734761120263594, + "grad_norm": 0.5014385943250483, + "learning_rate": 3.4165353358596475e-05, + "loss": 0.1864, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21568092703819275, + "step": 1380, + "valid_targets_mean": 4668.3, + "valid_targets_min": 375 + }, + { + "epoch": 2.2817133443163096, + "grad_norm": 0.4840041536578911, + "learning_rate": 3.410723745306348e-05, + "loss": 0.185, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19082829356193542, + "step": 1385, + "valid_targets_mean": 4344.3, + "valid_targets_min": 622 + }, + { + "epoch": 2.2899505766062602, + "grad_norm": 0.49307871381195234, + "learning_rate": 3.404888351004571e-05, + "loss": 0.1815, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18089547753334045, + "step": 1390, + "valid_targets_mean": 4720.2, + "valid_targets_min": 374 + }, + { + "epoch": 2.298187808896211, + "grad_norm": 0.49935101497553586, + "learning_rate": 3.3990292514174346e-05, + "loss": 0.1838, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1871870756149292, + "step": 1395, + "valid_targets_mean": 5122.2, + "valid_targets_min": 382 + }, + { + "epoch": 2.3064250411861615, + "grad_norm": 0.4076277361465614, + "learning_rate": 3.393146545408047e-05, + "loss": 0.1815, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18360033631324768, + "step": 1400, + "valid_targets_mean": 6602.8, + "valid_targets_min": 491 + }, + { + "epoch": 2.314662273476112, + "grad_norm": 0.4223706656199542, + "learning_rate": 3.3872403322378354e-05, + "loss": 0.1908, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18353493511676788, + "step": 1405, + "valid_targets_mean": 6359.8, + "valid_targets_min": 525 + }, + { + "epoch": 2.3228995057660624, + "grad_norm": 0.4849400359217284, + "learning_rate": 3.3813107115648745e-05, + "loss": 0.1924, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2019674926996231, + "step": 1410, + "valid_targets_mean": 5309.1, + "valid_targets_min": 328 + }, + { + "epoch": 2.331136738056013, + "grad_norm": 0.4575537443988003, + "learning_rate": 3.375357783442205e-05, + "loss": 0.1938, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20391184091567993, + "step": 1415, + "valid_targets_mean": 5717.2, + "valid_targets_min": 1413 + }, + { + "epoch": 2.3393739703459637, + "grad_norm": 0.5731198968413408, + "learning_rate": 3.369381648316144e-05, + "loss": 0.1859, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18056750297546387, + "step": 1420, + "valid_targets_mean": 4796.6, + "valid_targets_min": 330 + }, + { + "epoch": 2.3476112026359144, + "grad_norm": 0.4055911598397016, + "learning_rate": 3.363382407024588e-05, + "loss": 0.1933, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19828687608242035, + "step": 1425, + "valid_targets_mean": 6127.6, + "valid_targets_min": 305 + }, + { + "epoch": 2.355848434925865, + "grad_norm": 0.514361900038238, + "learning_rate": 3.357360160795317e-05, + "loss": 0.1816, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1833609938621521, + "step": 1430, + "valid_targets_mean": 5110.0, + "valid_targets_min": 635 + }, + { + "epoch": 2.3640856672158153, + "grad_norm": 0.4939780199608309, + "learning_rate": 3.35131501124428e-05, + "loss": 0.1828, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18957363069057465, + "step": 1435, + "valid_targets_mean": 5170.4, + "valid_targets_min": 344 + }, + { + "epoch": 2.372322899505766, + "grad_norm": 0.4366021181201065, + "learning_rate": 3.345247060373887e-05, + "loss": 0.1873, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19346314668655396, + "step": 1440, + "valid_targets_mean": 5597.6, + "valid_targets_min": 342 + }, + { + "epoch": 2.3805601317957166, + "grad_norm": 0.5616766081872978, + "learning_rate": 3.3391564105712824e-05, + "loss": 0.1877, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19493432343006134, + "step": 1445, + "valid_targets_mean": 3618.6, + "valid_targets_min": 318 + }, + { + "epoch": 2.3887973640856672, + "grad_norm": 0.4983461122768153, + "learning_rate": 3.333043164606618e-05, + "loss": 0.2063, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.199821338057518, + "step": 1450, + "valid_targets_mean": 4254.8, + "valid_targets_min": 315 + }, + { + "epoch": 2.397034596375618, + "grad_norm": 0.4498876894639847, + "learning_rate": 3.3269074256313246e-05, + "loss": 0.1877, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20630428194999695, + "step": 1455, + "valid_targets_mean": 5124.8, + "valid_targets_min": 293 + }, + { + "epoch": 2.4052718286655685, + "grad_norm": 0.46335332374164745, + "learning_rate": 3.320749297176363e-05, + "loss": 0.1804, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19408290088176727, + "step": 1460, + "valid_targets_mean": 6146.5, + "valid_targets_min": 233 + }, + { + "epoch": 2.4135090609555188, + "grad_norm": 0.432494507185893, + "learning_rate": 3.3145688831504845e-05, + "loss": 0.168, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1638782024383545, + "step": 1465, + "valid_targets_mean": 6058.1, + "valid_targets_min": 303 + }, + { + "epoch": 2.4217462932454694, + "grad_norm": 0.4233371763444396, + "learning_rate": 3.308366287838472e-05, + "loss": 0.1832, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17348337173461914, + "step": 1470, + "valid_targets_mean": 6430.6, + "valid_targets_min": 1391 + }, + { + "epoch": 2.42998352553542, + "grad_norm": 0.5173084245980966, + "learning_rate": 3.302141615899384e-05, + "loss": 0.1845, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16355247795581818, + "step": 1475, + "valid_targets_mean": 5664.9, + "valid_targets_min": 336 + }, + { + "epoch": 2.4382207578253707, + "grad_norm": 0.5007091972827029, + "learning_rate": 3.2958949723647875e-05, + "loss": 0.1883, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18168076872825623, + "step": 1480, + "valid_targets_mean": 4153.4, + "valid_targets_min": 327 + }, + { + "epoch": 2.4464579901153214, + "grad_norm": 0.4540610488829723, + "learning_rate": 3.289626462636985e-05, + "loss": 0.1882, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19426754117012024, + "step": 1485, + "valid_targets_mean": 5505.4, + "valid_targets_min": 1120 + }, + { + "epoch": 2.454695222405272, + "grad_norm": 0.3850445234018962, + "learning_rate": 3.283336192487238e-05, + "loss": 0.1776, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17522047460079193, + "step": 1490, + "valid_targets_mean": 7046.2, + "valid_targets_min": 760 + }, + { + "epoch": 2.4629324546952223, + "grad_norm": 0.49847271228259965, + "learning_rate": 3.277024268053978e-05, + "loss": 0.193, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19626614451408386, + "step": 1495, + "valid_targets_mean": 4122.3, + "valid_targets_min": 357 + }, + { + "epoch": 2.471169686985173, + "grad_norm": 0.4299543359076253, + "learning_rate": 3.2706907958410234e-05, + "loss": 0.1892, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21099016070365906, + "step": 1500, + "valid_targets_mean": 5716.4, + "valid_targets_min": 230 + }, + { + "epoch": 2.4794069192751236, + "grad_norm": 0.5168100523876604, + "learning_rate": 3.2643358827157724e-05, + "loss": 0.1774, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1860761046409607, + "step": 1505, + "valid_targets_mean": 4338.4, + "valid_targets_min": 418 + }, + { + "epoch": 2.4876441515650742, + "grad_norm": 0.4324694230224118, + "learning_rate": 3.257959635907409e-05, + "loss": 0.1919, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21066716313362122, + "step": 1510, + "valid_targets_mean": 6845.1, + "valid_targets_min": 3145 + }, + { + "epoch": 2.495881383855025, + "grad_norm": 0.4426728632174381, + "learning_rate": 3.251562163005087e-05, + "loss": 0.1794, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17823879420757294, + "step": 1515, + "valid_targets_mean": 4796.9, + "valid_targets_min": 317 + }, + { + "epoch": 2.504118616144975, + "grad_norm": 0.4615007706928971, + "learning_rate": 3.245143571956117e-05, + "loss": 0.1834, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17330557107925415, + "step": 1520, + "valid_targets_mean": 5856.2, + "valid_targets_min": 299 + }, + { + "epoch": 2.5123558484349258, + "grad_norm": 0.4929827190489361, + "learning_rate": 3.2387039710641486e-05, + "loss": 0.1983, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.23035567998886108, + "step": 1525, + "valid_targets_mean": 5336.1, + "valid_targets_min": 314 + }, + { + "epoch": 2.5205930807248764, + "grad_norm": 0.47838149878259656, + "learning_rate": 3.232243468987336e-05, + "loss": 0.1815, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18333536386489868, + "step": 1530, + "valid_targets_mean": 4696.2, + "valid_targets_min": 363 + }, + { + "epoch": 2.528830313014827, + "grad_norm": 0.4774794225296381, + "learning_rate": 3.2257621747365086e-05, + "loss": 0.1825, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18388643860816956, + "step": 1535, + "valid_targets_mean": 4851.4, + "valid_targets_min": 315 + }, + { + "epoch": 2.5370675453047777, + "grad_norm": 0.5784682298913758, + "learning_rate": 3.219260197673333e-05, + "loss": 0.1912, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1999967098236084, + "step": 1540, + "valid_targets_mean": 5049.6, + "valid_targets_min": 339 + }, + { + "epoch": 2.545304777594728, + "grad_norm": 0.4870283917003109, + "learning_rate": 3.212737647508464e-05, + "loss": 0.191, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1813567727804184, + "step": 1545, + "valid_targets_mean": 4931.9, + "valid_targets_min": 460 + }, + { + "epoch": 2.5535420098846786, + "grad_norm": 0.4912766743384124, + "learning_rate": 3.206194634299696e-05, + "loss": 0.1835, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18001456558704376, + "step": 1550, + "valid_targets_mean": 4067.6, + "valid_targets_min": 354 + }, + { + "epoch": 2.5617792421746293, + "grad_norm": 0.4177379128740769, + "learning_rate": 3.199631268450104e-05, + "loss": 0.1882, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1830926239490509, + "step": 1555, + "valid_targets_mean": 5638.4, + "valid_targets_min": 452 + }, + { + "epoch": 2.57001647446458, + "grad_norm": 0.4569501634208448, + "learning_rate": 3.1930476607061835e-05, + "loss": 0.187, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1815715730190277, + "step": 1560, + "valid_targets_mean": 6495.7, + "valid_targets_min": 1668 + }, + { + "epoch": 2.5782537067545306, + "grad_norm": 0.4755155809990493, + "learning_rate": 3.1864439221559797e-05, + "loss": 0.1921, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2156897485256195, + "step": 1565, + "valid_targets_mean": 5195.4, + "valid_targets_min": 367 + }, + { + "epoch": 2.586490939044481, + "grad_norm": 0.4352493070918896, + "learning_rate": 3.179820164227213e-05, + "loss": 0.1874, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1801442950963974, + "step": 1570, + "valid_targets_mean": 5301.5, + "valid_targets_min": 298 + }, + { + "epoch": 2.594728171334432, + "grad_norm": 0.5418117949683932, + "learning_rate": 3.1731764986853985e-05, + "loss": 0.193, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20953936874866486, + "step": 1575, + "valid_targets_mean": 4256.2, + "valid_targets_min": 273 + }, + { + "epoch": 2.602965403624382, + "grad_norm": 0.4712036121831422, + "learning_rate": 3.166513037631965e-05, + "loss": 0.1785, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18240496516227722, + "step": 1580, + "valid_targets_mean": 5122.9, + "valid_targets_min": 386 + }, + { + "epoch": 2.6112026359143328, + "grad_norm": 0.4479622079225454, + "learning_rate": 3.1598298935023536e-05, + "loss": 0.1833, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17743943631649017, + "step": 1585, + "valid_targets_mean": 4750.2, + "valid_targets_min": 323 + }, + { + "epoch": 2.6194398682042834, + "grad_norm": 0.4363238458372522, + "learning_rate": 3.153127179064131e-05, + "loss": 0.19, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1790778636932373, + "step": 1590, + "valid_targets_mean": 5834.9, + "valid_targets_min": 1060 + }, + { + "epoch": 2.627677100494234, + "grad_norm": 0.4144716349542137, + "learning_rate": 3.146405007415081e-05, + "loss": 0.1838, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18715772032737732, + "step": 1595, + "valid_targets_mean": 5915.6, + "valid_targets_min": 2732 + }, + { + "epoch": 2.6359143327841847, + "grad_norm": 0.402511904543039, + "learning_rate": 3.1396634919812955e-05, + "loss": 0.1798, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18299850821495056, + "step": 1600, + "valid_targets_mean": 6042.9, + "valid_targets_min": 1751 + }, + { + "epoch": 2.644151565074135, + "grad_norm": 0.42949378682217304, + "learning_rate": 3.1329027465152654e-05, + "loss": 0.1797, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19716359674930573, + "step": 1605, + "valid_targets_mean": 5723.6, + "valid_targets_min": 1438 + }, + { + "epoch": 2.6523887973640856, + "grad_norm": 0.41523450789292266, + "learning_rate": 3.126122885093955e-05, + "loss": 0.1846, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1815474033355713, + "step": 1610, + "valid_targets_mean": 6400.8, + "valid_targets_min": 290 + }, + { + "epoch": 2.6606260296540363, + "grad_norm": 0.42400960287790107, + "learning_rate": 3.1193240221168835e-05, + "loss": 0.1912, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19176776707172394, + "step": 1615, + "valid_targets_mean": 7038.9, + "valid_targets_min": 790 + }, + { + "epoch": 2.668863261943987, + "grad_norm": 0.4519946593514569, + "learning_rate": 3.112506272304189e-05, + "loss": 0.1875, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17423108220100403, + "step": 1620, + "valid_targets_mean": 4906.9, + "valid_targets_min": 326 + }, + { + "epoch": 2.6771004942339376, + "grad_norm": 0.40628518983943557, + "learning_rate": 3.1056697506946965e-05, + "loss": 0.1858, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19080524146556854, + "step": 1625, + "valid_targets_mean": 6026.1, + "valid_targets_min": 2394 + }, + { + "epoch": 2.685337726523888, + "grad_norm": 0.43758527040491757, + "learning_rate": 3.0988145726439754e-05, + "loss": 0.1894, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19035451114177704, + "step": 1630, + "valid_targets_mean": 5603.6, + "valid_targets_min": 600 + }, + { + "epoch": 2.6935749588138385, + "grad_norm": 0.4213759976520408, + "learning_rate": 3.091940853822393e-05, + "loss": 0.1904, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17566537857055664, + "step": 1635, + "valid_targets_mean": 5556.9, + "valid_targets_min": 692 + }, + { + "epoch": 2.701812191103789, + "grad_norm": 0.41064334724669393, + "learning_rate": 3.0850487102131624e-05, + "loss": 0.1802, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17263707518577576, + "step": 1640, + "valid_targets_mean": 6448.4, + "valid_targets_min": 365 + }, + { + "epoch": 2.7100494233937398, + "grad_norm": 0.40647920955642636, + "learning_rate": 3.078138258110389e-05, + "loss": 0.1925, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1742803454399109, + "step": 1645, + "valid_targets_mean": 7037.6, + "valid_targets_min": 600 + }, + { + "epoch": 2.7182866556836904, + "grad_norm": 0.4439437502158048, + "learning_rate": 3.071209614117103e-05, + "loss": 0.1869, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18262037634849548, + "step": 1650, + "valid_targets_mean": 5216.6, + "valid_targets_min": 386 + }, + { + "epoch": 2.7265238879736406, + "grad_norm": 0.4272024997497859, + "learning_rate": 3.0642628951432916e-05, + "loss": 0.1811, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18403027951717377, + "step": 1655, + "valid_targets_mean": 5560.3, + "valid_targets_min": 384 + }, + { + "epoch": 2.7347611202635913, + "grad_norm": 0.4106960597797865, + "learning_rate": 3.057298218403934e-05, + "loss": 0.1764, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16496650874614716, + "step": 1660, + "valid_targets_mean": 5558.3, + "valid_targets_min": 515 + }, + { + "epoch": 2.742998352553542, + "grad_norm": 0.4573999255888336, + "learning_rate": 3.050315701417016e-05, + "loss": 0.1854, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1997423619031906, + "step": 1665, + "valid_targets_mean": 5737.6, + "valid_targets_min": 556 + }, + { + "epoch": 2.7512355848434926, + "grad_norm": 0.4588007386145714, + "learning_rate": 3.043315462001549e-05, + "loss": 0.1779, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1824597418308258, + "step": 1670, + "valid_targets_mean": 6000.8, + "valid_targets_min": 291 + }, + { + "epoch": 2.7594728171334433, + "grad_norm": 0.45545623228765714, + "learning_rate": 3.0362976182755837e-05, + "loss": 0.1922, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1609131544828415, + "step": 1675, + "valid_targets_mean": 4831.3, + "valid_targets_min": 365 + }, + { + "epoch": 2.7677100494233935, + "grad_norm": 0.43744918321348536, + "learning_rate": 3.0292622886542154e-05, + "loss": 0.1934, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19479726254940033, + "step": 1680, + "valid_targets_mean": 5040.9, + "valid_targets_min": 290 + }, + { + "epoch": 2.775947281713344, + "grad_norm": 0.3723836457441067, + "learning_rate": 3.0222095918475873e-05, + "loss": 0.1836, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16930866241455078, + "step": 1685, + "valid_targets_mean": 7237.4, + "valid_targets_min": 3984 + }, + { + "epoch": 2.784184514003295, + "grad_norm": 0.4397650823089542, + "learning_rate": 3.0151396468588848e-05, + "loss": 0.1955, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2224178910255432, + "step": 1690, + "valid_targets_mean": 5372.0, + "valid_targets_min": 716 + }, + { + "epoch": 2.7924217462932455, + "grad_norm": 0.4766636021138799, + "learning_rate": 3.0080525729823306e-05, + "loss": 0.1988, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19022653996944427, + "step": 1695, + "valid_targets_mean": 5674.8, + "valid_targets_min": 372 + }, + { + "epoch": 2.800658978583196, + "grad_norm": 0.5783854027781857, + "learning_rate": 3.0009484898011707e-05, + "loss": 0.1826, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18798254430294037, + "step": 1700, + "valid_targets_mean": 6211.0, + "valid_targets_min": 384 + }, + { + "epoch": 2.808896210873147, + "grad_norm": 0.6044879964158381, + "learning_rate": 2.993827517185656e-05, + "loss": 0.1976, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.20792728662490845, + "step": 1705, + "valid_targets_mean": 2618.7, + "valid_targets_min": 336 + }, + { + "epoch": 2.8171334431630974, + "grad_norm": 0.48777693314470333, + "learning_rate": 2.9866897752910194e-05, + "loss": 0.1781, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18926022946834564, + "step": 1710, + "valid_targets_mean": 4935.6, + "valid_targets_min": 383 + }, + { + "epoch": 2.8253706754530477, + "grad_norm": 0.4355724130572252, + "learning_rate": 2.979535384555449e-05, + "loss": 0.1895, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18864750862121582, + "step": 1715, + "valid_targets_mean": 6106.8, + "valid_targets_min": 452 + }, + { + "epoch": 2.8336079077429983, + "grad_norm": 0.5005269538168059, + "learning_rate": 2.972364465698058e-05, + "loss": 0.1718, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17418703436851501, + "step": 1720, + "valid_targets_mean": 4066.6, + "valid_targets_min": 343 + }, + { + "epoch": 2.841845140032949, + "grad_norm": 0.35405912220850505, + "learning_rate": 2.9651771397168443e-05, + "loss": 0.1864, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16310462355613708, + "step": 1725, + "valid_targets_mean": 7079.8, + "valid_targets_min": 3151 + }, + { + "epoch": 2.8500823723228996, + "grad_norm": 0.4118485814337778, + "learning_rate": 2.95797352788665e-05, + "loss": 0.1815, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19650161266326904, + "step": 1730, + "valid_targets_mean": 6517.4, + "valid_targets_min": 1213 + }, + { + "epoch": 2.8583196046128503, + "grad_norm": 0.3985995816730429, + "learning_rate": 2.9507537517571144e-05, + "loss": 0.1812, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16181087493896484, + "step": 1735, + "valid_targets_mean": 5469.2, + "valid_targets_min": 329 + }, + { + "epoch": 2.8665568369028005, + "grad_norm": 0.41174127791281484, + "learning_rate": 2.943517933150627e-05, + "loss": 0.1792, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16833308339118958, + "step": 1740, + "valid_targets_mean": 6566.2, + "valid_targets_min": 697 + }, + { + "epoch": 2.874794069192751, + "grad_norm": 0.49914804119853634, + "learning_rate": 2.936266194160266e-05, + "loss": 0.1934, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19393935799598694, + "step": 1745, + "valid_targets_mean": 4929.6, + "valid_targets_min": 310 + }, + { + "epoch": 2.883031301482702, + "grad_norm": 0.45669196924100525, + "learning_rate": 2.9289986571477437e-05, + "loss": 0.1893, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18913184106349945, + "step": 1750, + "valid_targets_mean": 5159.7, + "valid_targets_min": 345 + }, + { + "epoch": 2.8912685337726525, + "grad_norm": 0.4050992240314355, + "learning_rate": 2.9217154447413353e-05, + "loss": 0.1856, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19301308691501617, + "step": 1755, + "valid_targets_mean": 5732.8, + "valid_targets_min": 643 + }, + { + "epoch": 2.899505766062603, + "grad_norm": 0.4166433021084068, + "learning_rate": 2.9144166798338183e-05, + "loss": 0.1813, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16880372166633606, + "step": 1760, + "valid_targets_mean": 5482.8, + "valid_targets_min": 346 + }, + { + "epoch": 2.9077429983525533, + "grad_norm": 0.4876653800212749, + "learning_rate": 2.9071024855803915e-05, + "loss": 0.175, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16207365691661835, + "step": 1765, + "valid_targets_mean": 5472.4, + "valid_targets_min": 313 + }, + { + "epoch": 2.915980230642504, + "grad_norm": 0.4432418577836535, + "learning_rate": 2.899772985396599e-05, + "loss": 0.18, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17681077122688293, + "step": 1770, + "valid_targets_mean": 5663.9, + "valid_targets_min": 2714 + }, + { + "epoch": 2.9242174629324547, + "grad_norm": 0.4057865576825728, + "learning_rate": 2.8924283029562507e-05, + "loss": 0.1753, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17672967910766602, + "step": 1775, + "valid_targets_mean": 6175.4, + "valid_targets_min": 1138 + }, + { + "epoch": 2.9324546952224053, + "grad_norm": 0.5907926016816272, + "learning_rate": 2.8850685621893325e-05, + "loss": 0.1944, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1969795525074005, + "step": 1780, + "valid_targets_mean": 4073.1, + "valid_targets_min": 306 + }, + { + "epoch": 2.940691927512356, + "grad_norm": 0.44725505789253106, + "learning_rate": 2.8776938872799154e-05, + "loss": 0.1916, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1989360749721527, + "step": 1785, + "valid_targets_mean": 5965.4, + "valid_targets_min": 282 + }, + { + "epoch": 2.948929159802306, + "grad_norm": 0.4760043999500545, + "learning_rate": 2.8703044026640606e-05, + "loss": 0.1756, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18851858377456665, + "step": 1790, + "valid_targets_mean": 4529.5, + "valid_targets_min": 408 + }, + { + "epoch": 2.957166392092257, + "grad_norm": 0.5146855051453981, + "learning_rate": 2.8629002330277193e-05, + "loss": 0.1809, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.19129858911037445, + "step": 1795, + "valid_targets_mean": 4191.7, + "valid_targets_min": 313 + }, + { + "epoch": 2.9654036243822075, + "grad_norm": 0.5551959217056265, + "learning_rate": 2.8554815033046314e-05, + "loss": 0.1859, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.21176767349243164, + "step": 1800, + "valid_targets_mean": 5302.6, + "valid_targets_min": 318 + }, + { + "epoch": 2.973640856672158, + "grad_norm": 0.43494351702575534, + "learning_rate": 2.8480483386742138e-05, + "loss": 0.1778, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1887683868408203, + "step": 1805, + "valid_targets_mean": 5780.3, + "valid_targets_min": 3622 + }, + { + "epoch": 2.981878088962109, + "grad_norm": 0.6709852335599017, + "learning_rate": 2.8406008645594503e-05, + "loss": 0.1755, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18137386441230774, + "step": 1810, + "valid_targets_mean": 6075.0, + "valid_targets_min": 280 + }, + { + "epoch": 2.990115321252059, + "grad_norm": 0.4533452043383325, + "learning_rate": 2.8331392066247737e-05, + "loss": 0.1829, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1966426968574524, + "step": 1815, + "valid_targets_mean": 6176.1, + "valid_targets_min": 2014 + }, + { + "epoch": 2.99835255354201, + "grad_norm": 0.4872843448253986, + "learning_rate": 2.825663490773949e-05, + "loss": 0.1921, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2191101312637329, + "step": 1820, + "valid_targets_mean": 4317.1, + "valid_targets_min": 366 + }, + { + "epoch": 3.0065897858319603, + "grad_norm": 0.4925134741814334, + "learning_rate": 2.8181738431479436e-05, + "loss": 0.1654, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17591401934623718, + "step": 1825, + "valid_targets_mean": 5124.8, + "valid_targets_min": 300 + }, + { + "epoch": 3.014827018121911, + "grad_norm": 0.46951633331962356, + "learning_rate": 2.8106703901228038e-05, + "loss": 0.1724, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16612932085990906, + "step": 1830, + "valid_targets_mean": 5064.6, + "valid_targets_min": 382 + }, + { + "epoch": 3.0230642504118617, + "grad_norm": 0.4201136329542275, + "learning_rate": 2.803153258307519e-05, + "loss": 0.1627, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1488945186138153, + "step": 1835, + "valid_targets_mean": 5890.2, + "valid_targets_min": 327 + }, + { + "epoch": 3.0313014827018123, + "grad_norm": 0.4070141782866309, + "learning_rate": 2.795622574541887e-05, + "loss": 0.1545, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13417211174964905, + "step": 1840, + "valid_targets_mean": 6251.1, + "valid_targets_min": 400 + }, + { + "epoch": 3.039538714991763, + "grad_norm": 0.4474445147917244, + "learning_rate": 2.7880784658943754e-05, + "loss": 0.1673, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.162965327501297, + "step": 1845, + "valid_targets_mean": 5795.4, + "valid_targets_min": 2153 + }, + { + "epoch": 3.047775947281713, + "grad_norm": 0.5042184100386404, + "learning_rate": 2.780521059659972e-05, + "loss": 0.1735, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17654483020305634, + "step": 1850, + "valid_targets_mean": 4581.5, + "valid_targets_min": 763 + }, + { + "epoch": 3.056013179571664, + "grad_norm": 0.4690284250626636, + "learning_rate": 2.7729504833580435e-05, + "loss": 0.1723, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17030039429664612, + "step": 1855, + "valid_targets_mean": 5360.2, + "valid_targets_min": 515 + }, + { + "epoch": 3.0642504118616145, + "grad_norm": 0.4457242021093399, + "learning_rate": 2.7653668647301797e-05, + "loss": 0.1598, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15688768029212952, + "step": 1860, + "valid_targets_mean": 5233.9, + "valid_targets_min": 270 + }, + { + "epoch": 3.072487644151565, + "grad_norm": 0.5262957209265869, + "learning_rate": 2.7577703317380396e-05, + "loss": 0.1596, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.162797212600708, + "step": 1865, + "valid_targets_mean": 5007.5, + "valid_targets_min": 318 + }, + { + "epoch": 3.080724876441516, + "grad_norm": 0.435607664368139, + "learning_rate": 2.7501610125611908e-05, + "loss": 0.1639, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16544128954410553, + "step": 1870, + "valid_targets_mean": 5627.2, + "valid_targets_min": 336 + }, + { + "epoch": 3.088962108731466, + "grad_norm": 0.48754061786282266, + "learning_rate": 2.7425390355949485e-05, + "loss": 0.1718, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17840203642845154, + "step": 1875, + "valid_targets_mean": 5176.1, + "valid_targets_min": 664 + }, + { + "epoch": 3.0971993410214167, + "grad_norm": 0.4713253285078091, + "learning_rate": 2.7349045294482085e-05, + "loss": 0.1593, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16155745089054108, + "step": 1880, + "valid_targets_mean": 5086.4, + "valid_targets_min": 357 + }, + { + "epoch": 3.1054365733113674, + "grad_norm": 0.43308469551129186, + "learning_rate": 2.727257622941276e-05, + "loss": 0.1629, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1810024082660675, + "step": 1885, + "valid_targets_mean": 6242.1, + "valid_targets_min": 2910 + }, + { + "epoch": 3.113673805601318, + "grad_norm": 0.4364624354543134, + "learning_rate": 2.719598445103693e-05, + "loss": 0.1561, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1708287000656128, + "step": 1890, + "valid_targets_mean": 5678.8, + "valid_targets_min": 3190 + }, + { + "epoch": 3.1219110378912687, + "grad_norm": 0.4737469378096958, + "learning_rate": 2.7119271251720628e-05, + "loss": 0.1647, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18933191895484924, + "step": 1895, + "valid_targets_mean": 5111.1, + "valid_targets_min": 556 + }, + { + "epoch": 3.130148270181219, + "grad_norm": 0.41676344263541504, + "learning_rate": 2.7042437925878638e-05, + "loss": 0.1673, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15466684103012085, + "step": 1900, + "valid_targets_mean": 5804.4, + "valid_targets_min": 649 + }, + { + "epoch": 3.1383855024711695, + "grad_norm": 0.5653553337325623, + "learning_rate": 2.696548576995273e-05, + "loss": 0.1783, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18597179651260376, + "step": 1905, + "valid_targets_mean": 5242.3, + "valid_targets_min": 359 + }, + { + "epoch": 3.14662273476112, + "grad_norm": 0.4879099377031038, + "learning_rate": 2.688841608238971e-05, + "loss": 0.1683, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16499802470207214, + "step": 1910, + "valid_targets_mean": 4917.2, + "valid_targets_min": 311 + }, + { + "epoch": 3.154859967051071, + "grad_norm": 0.4667216223779826, + "learning_rate": 2.6811230163619573e-05, + "loss": 0.1684, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16547727584838867, + "step": 1915, + "valid_targets_mean": 4575.3, + "valid_targets_min": 372 + }, + { + "epoch": 3.1630971993410215, + "grad_norm": 0.42534284414973267, + "learning_rate": 2.6733929316033516e-05, + "loss": 0.1547, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1600956916809082, + "step": 1920, + "valid_targets_mean": 5993.5, + "valid_targets_min": 336 + }, + { + "epoch": 3.171334431630972, + "grad_norm": 0.4682657752517238, + "learning_rate": 2.6656514843961995e-05, + "loss": 0.1667, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17428819835186005, + "step": 1925, + "valid_targets_mean": 5779.7, + "valid_targets_min": 385 + }, + { + "epoch": 3.1795716639209224, + "grad_norm": 0.47746321265894026, + "learning_rate": 2.6578988053652674e-05, + "loss": 0.1715, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15723752975463867, + "step": 1930, + "valid_targets_mean": 4613.3, + "valid_targets_min": 286 + }, + { + "epoch": 3.187808896210873, + "grad_norm": 0.4951364632733134, + "learning_rate": 2.650135025324845e-05, + "loss": 0.1651, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1764582395553589, + "step": 1935, + "valid_targets_mean": 4870.2, + "valid_targets_min": 373 + }, + { + "epoch": 3.1960461285008237, + "grad_norm": 0.40725037546610426, + "learning_rate": 2.6423602752765313e-05, + "loss": 0.1553, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13663309812545776, + "step": 1940, + "valid_targets_mean": 5494.2, + "valid_targets_min": 315 + }, + { + "epoch": 3.2042833607907744, + "grad_norm": 0.42521249997987376, + "learning_rate": 2.6345746864070266e-05, + "loss": 0.1652, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15042027831077576, + "step": 1945, + "valid_targets_mean": 6342.6, + "valid_targets_min": 484 + }, + { + "epoch": 3.212520593080725, + "grad_norm": 0.5413285488120383, + "learning_rate": 2.626778390085921e-05, + "loss": 0.1686, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16320699453353882, + "step": 1950, + "valid_targets_mean": 5744.4, + "valid_targets_min": 429 + }, + { + "epoch": 3.2207578253706757, + "grad_norm": 0.43089481661574836, + "learning_rate": 2.618971517863475e-05, + "loss": 0.163, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12983350455760956, + "step": 1955, + "valid_targets_mean": 5079.2, + "valid_targets_min": 971 + }, + { + "epoch": 3.228995057660626, + "grad_norm": 0.4911902168052223, + "learning_rate": 2.611154201468402e-05, + "loss": 0.1676, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16104954481124878, + "step": 1960, + "valid_targets_mean": 5500.1, + "valid_targets_min": 347 + }, + { + "epoch": 3.2372322899505765, + "grad_norm": 0.4133285626323947, + "learning_rate": 2.6033265728056437e-05, + "loss": 0.1694, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14631503820419312, + "step": 1965, + "valid_targets_mean": 5502.2, + "valid_targets_min": 1724 + }, + { + "epoch": 3.245469522240527, + "grad_norm": 0.5213460978823949, + "learning_rate": 2.5954887639541435e-05, + "loss": 0.1733, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17967018485069275, + "step": 1970, + "valid_targets_mean": 5081.2, + "valid_targets_min": 452 + }, + { + "epoch": 3.253706754530478, + "grad_norm": 0.4681430241946772, + "learning_rate": 2.5876409071646224e-05, + "loss": 0.1626, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1624462753534317, + "step": 1975, + "valid_targets_mean": 5415.8, + "valid_targets_min": 252 + }, + { + "epoch": 3.2619439868204285, + "grad_norm": 0.46303449443059874, + "learning_rate": 2.5797831348573425e-05, + "loss": 0.1665, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17155566811561584, + "step": 1980, + "valid_targets_mean": 5258.4, + "valid_targets_min": 2373 + }, + { + "epoch": 3.2701812191103787, + "grad_norm": 0.532316430057948, + "learning_rate": 2.571915579619876e-05, + "loss": 0.1683, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17313048243522644, + "step": 1985, + "valid_targets_mean": 4552.9, + "valid_targets_min": 459 + }, + { + "epoch": 3.2784184514003294, + "grad_norm": 0.4197831861072301, + "learning_rate": 2.5640383742048654e-05, + "loss": 0.1794, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18682700395584106, + "step": 1990, + "valid_targets_mean": 6343.2, + "valid_targets_min": 311 + }, + { + "epoch": 3.28665568369028, + "grad_norm": 0.47975068369647395, + "learning_rate": 2.5561516515277863e-05, + "loss": 0.1635, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1698731780052185, + "step": 1995, + "valid_targets_mean": 4894.1, + "valid_targets_min": 429 + }, + { + "epoch": 3.2948929159802307, + "grad_norm": 0.44143749997803633, + "learning_rate": 2.5482555446647018e-05, + "loss": 0.1712, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15826047956943512, + "step": 2000, + "valid_targets_mean": 4921.8, + "valid_targets_min": 385 + }, + { + "epoch": 3.3031301482701814, + "grad_norm": 0.47872780397929104, + "learning_rate": 2.5403501868500185e-05, + "loss": 0.1711, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1793244183063507, + "step": 2005, + "valid_targets_mean": 5263.2, + "valid_targets_min": 366 + }, + { + "epoch": 3.3113673805601316, + "grad_norm": 0.43795563431108253, + "learning_rate": 2.5324357114742397e-05, + "loss": 0.162, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1651221364736557, + "step": 2010, + "valid_targets_mean": 5599.2, + "valid_targets_min": 999 + }, + { + "epoch": 3.3196046128500822, + "grad_norm": 0.4194171717207259, + "learning_rate": 2.5245122520817117e-05, + "loss": 0.1685, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16536174714565277, + "step": 2015, + "valid_targets_mean": 5968.4, + "valid_targets_min": 253 + }, + { + "epoch": 3.327841845140033, + "grad_norm": 0.630395759920917, + "learning_rate": 2.516579942368373e-05, + "loss": 0.1726, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16959768533706665, + "step": 2020, + "valid_targets_mean": 4148.6, + "valid_targets_min": 295 + }, + { + "epoch": 3.3360790774299836, + "grad_norm": 0.4597314779792269, + "learning_rate": 2.508638916179497e-05, + "loss": 0.1592, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16048160195350647, + "step": 2025, + "valid_targets_mean": 5179.0, + "valid_targets_min": 416 + }, + { + "epoch": 3.344316309719934, + "grad_norm": 0.40076416546655863, + "learning_rate": 2.5006893075074326e-05, + "loss": 0.1638, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14166682958602905, + "step": 2030, + "valid_targets_mean": 5947.9, + "valid_targets_min": 408 + }, + { + "epoch": 3.352553542009885, + "grad_norm": 0.6022307050548966, + "learning_rate": 2.4927312504893467e-05, + "loss": 0.1577, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14992672204971313, + "step": 2035, + "valid_targets_mean": 6018.8, + "valid_targets_min": 357 + }, + { + "epoch": 3.360790774299835, + "grad_norm": 0.48131640597198694, + "learning_rate": 2.4847648794049587e-05, + "loss": 0.1592, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.177500382065773, + "step": 2040, + "valid_targets_mean": 4628.6, + "valid_targets_min": 270 + }, + { + "epoch": 3.3690280065897857, + "grad_norm": 0.40253163403627884, + "learning_rate": 2.4767903286742732e-05, + "loss": 0.1682, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.170414999127388, + "step": 2045, + "valid_targets_mean": 6258.0, + "valid_targets_min": 2733 + }, + { + "epoch": 3.3772652388797364, + "grad_norm": 0.429610208965216, + "learning_rate": 2.468807732855314e-05, + "loss": 0.1666, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15542641282081604, + "step": 2050, + "valid_targets_mean": 5290.5, + "valid_targets_min": 239 + }, + { + "epoch": 3.385502471169687, + "grad_norm": 0.5228662176128691, + "learning_rate": 2.4608172266418543e-05, + "loss": 0.1939, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.2773077189922333, + "step": 2055, + "valid_targets_mean": 6195.2, + "valid_targets_min": 307 + }, + { + "epoch": 3.3937397034596377, + "grad_norm": 0.4679719439792988, + "learning_rate": 2.4528189448611417e-05, + "loss": 0.1711, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15937143564224243, + "step": 2060, + "valid_targets_mean": 5585.6, + "valid_targets_min": 1274 + }, + { + "epoch": 3.4019769357495884, + "grad_norm": 0.6292251061768565, + "learning_rate": 2.4448130224716248e-05, + "loss": 0.168, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17720317840576172, + "step": 2065, + "valid_targets_mean": 4486.8, + "valid_targets_min": 440 + }, + { + "epoch": 3.4102141680395386, + "grad_norm": 0.41773728976670876, + "learning_rate": 2.436799594560675e-05, + "loss": 0.1747, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16367585957050323, + "step": 2070, + "valid_targets_mean": 5920.7, + "valid_targets_min": 282 + }, + { + "epoch": 3.4184514003294892, + "grad_norm": 0.5569444713805546, + "learning_rate": 2.4287787963423076e-05, + "loss": 0.164, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.172664612531662, + "step": 2075, + "valid_targets_mean": 4224.7, + "valid_targets_min": 269 + }, + { + "epoch": 3.42668863261944, + "grad_norm": 0.4136416394016962, + "learning_rate": 2.4207507631549013e-05, + "loss": 0.1664, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17453311383724213, + "step": 2080, + "valid_targets_mean": 6280.9, + "valid_targets_min": 469 + }, + { + "epoch": 3.4349258649093906, + "grad_norm": 0.5492874775817022, + "learning_rate": 2.412715630458912e-05, + "loss": 0.1691, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18640339374542236, + "step": 2085, + "valid_targets_mean": 5082.9, + "valid_targets_min": 315 + }, + { + "epoch": 3.443163097199341, + "grad_norm": 0.4846899712538489, + "learning_rate": 2.4046735338345895e-05, + "loss": 0.1688, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16321972012519836, + "step": 2090, + "valid_targets_mean": 4117.7, + "valid_targets_min": 381 + }, + { + "epoch": 3.4514003294892914, + "grad_norm": 0.43078882336824376, + "learning_rate": 2.396624608979689e-05, + "loss": 0.1637, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1680295765399933, + "step": 2095, + "valid_targets_mean": 5800.4, + "valid_targets_min": 486 + }, + { + "epoch": 3.459637561779242, + "grad_norm": 0.48957765167848494, + "learning_rate": 2.3885689917071822e-05, + "loss": 0.1686, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1648954600095749, + "step": 2100, + "valid_targets_mean": 4547.6, + "valid_targets_min": 433 + }, + { + "epoch": 3.4678747940691927, + "grad_norm": 0.4024882547668721, + "learning_rate": 2.3805068179429628e-05, + "loss": 0.1605, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15916281938552856, + "step": 2105, + "valid_targets_mean": 5993.3, + "valid_targets_min": 362 + }, + { + "epoch": 3.4761120263591434, + "grad_norm": 0.45971252491424164, + "learning_rate": 2.372438223723555e-05, + "loss": 0.1666, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14879193902015686, + "step": 2110, + "valid_targets_mean": 5487.4, + "valid_targets_min": 826 + }, + { + "epoch": 3.484349258649094, + "grad_norm": 0.5282714874848389, + "learning_rate": 2.3643633451938216e-05, + "loss": 0.1773, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18927377462387085, + "step": 2115, + "valid_targets_mean": 4617.0, + "valid_targets_min": 404 + }, + { + "epoch": 3.4925864909390443, + "grad_norm": 0.4385501002050119, + "learning_rate": 2.35628231860466e-05, + "loss": 0.1648, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15995869040489197, + "step": 2120, + "valid_targets_mean": 7026.8, + "valid_targets_min": 3996 + }, + { + "epoch": 3.500823723228995, + "grad_norm": 0.4226288385578095, + "learning_rate": 2.3481952803107067e-05, + "loss": 0.1601, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16267460584640503, + "step": 2125, + "valid_targets_mean": 5791.6, + "valid_targets_min": 526 + }, + { + "epoch": 3.5090609555189456, + "grad_norm": 0.3999368378824118, + "learning_rate": 2.3401023667680373e-05, + "loss": 0.1753, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1635211855173111, + "step": 2130, + "valid_targets_mean": 7097.8, + "valid_targets_min": 2006 + }, + { + "epoch": 3.5172981878088962, + "grad_norm": 0.4298215892191779, + "learning_rate": 2.332003714531862e-05, + "loss": 0.1636, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16600899398326874, + "step": 2135, + "valid_targets_mean": 6014.1, + "valid_targets_min": 327 + }, + { + "epoch": 3.525535420098847, + "grad_norm": 0.39900638433270885, + "learning_rate": 2.3238994602542246e-05, + "loss": 0.1597, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15637260675430298, + "step": 2140, + "valid_targets_mean": 6231.6, + "valid_targets_min": 467 + }, + { + "epoch": 3.533772652388797, + "grad_norm": 0.45498071505271753, + "learning_rate": 2.3157897406816917e-05, + "loss": 0.1656, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1908491551876068, + "step": 2145, + "valid_targets_mean": 5242.5, + "valid_targets_min": 343 + }, + { + "epoch": 3.5420098846787478, + "grad_norm": 0.46961507934645447, + "learning_rate": 2.3076746926530497e-05, + "loss": 0.1764, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17604225873947144, + "step": 2150, + "valid_targets_mean": 5637.7, + "valid_targets_min": 951 + }, + { + "epoch": 3.5502471169686984, + "grad_norm": 0.4373049137664838, + "learning_rate": 2.2995544530969946e-05, + "loss": 0.1715, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16728343069553375, + "step": 2155, + "valid_targets_mean": 5702.9, + "valid_targets_min": 407 + }, + { + "epoch": 3.558484349258649, + "grad_norm": 0.4384327378195429, + "learning_rate": 2.2914291590298197e-05, + "loss": 0.1634, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16371147334575653, + "step": 2160, + "valid_targets_mean": 5260.1, + "valid_targets_min": 300 + }, + { + "epoch": 3.5667215815485998, + "grad_norm": 0.44590623551665487, + "learning_rate": 2.2832989475531073e-05, + "loss": 0.1705, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17798402905464172, + "step": 2165, + "valid_targets_mean": 5838.9, + "valid_targets_min": 1807 + }, + { + "epoch": 3.5749588138385504, + "grad_norm": 0.4441625485059226, + "learning_rate": 2.2751639558514116e-05, + "loss": 0.1728, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17515239119529724, + "step": 2170, + "valid_targets_mean": 5587.0, + "valid_targets_min": 2698 + }, + { + "epoch": 3.583196046128501, + "grad_norm": 0.477649896519292, + "learning_rate": 2.267024321189947e-05, + "loss": 0.1725, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18034707009792328, + "step": 2175, + "valid_targets_mean": 4963.6, + "valid_targets_min": 325 + }, + { + "epoch": 3.5914332784184513, + "grad_norm": 0.4663224086535974, + "learning_rate": 2.2588801809122692e-05, + "loss": 0.1719, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18077439069747925, + "step": 2180, + "valid_targets_mean": 5901.2, + "valid_targets_min": 396 + }, + { + "epoch": 3.599670510708402, + "grad_norm": 0.4046716315074143, + "learning_rate": 2.2507316724379603e-05, + "loss": 0.1643, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16994984447956085, + "step": 2185, + "valid_targets_mean": 6699.7, + "valid_targets_min": 3961 + }, + { + "epoch": 3.6079077429983526, + "grad_norm": 0.46538609652857194, + "learning_rate": 2.2425789332603073e-05, + "loss": 0.1687, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15804654359817505, + "step": 2190, + "valid_targets_mean": 5469.8, + "valid_targets_min": 852 + }, + { + "epoch": 3.6161449752883033, + "grad_norm": 0.39597613600689074, + "learning_rate": 2.2344221009439863e-05, + "loss": 0.1664, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1535954773426056, + "step": 2195, + "valid_targets_mean": 6177.7, + "valid_targets_min": 290 + }, + { + "epoch": 3.624382207578254, + "grad_norm": 0.432923735193132, + "learning_rate": 2.2262613131227357e-05, + "loss": 0.1625, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15266185998916626, + "step": 2200, + "valid_targets_mean": 5965.2, + "valid_targets_min": 290 + }, + { + "epoch": 3.632619439868204, + "grad_norm": 0.40127577438372597, + "learning_rate": 2.2180967074970384e-05, + "loss": 0.1695, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16095589101314545, + "step": 2205, + "valid_targets_mean": 6422.1, + "valid_targets_min": 3765 + }, + { + "epoch": 3.640856672158155, + "grad_norm": 0.4650779716931073, + "learning_rate": 2.2099284218317965e-05, + "loss": 0.1726, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18171751499176025, + "step": 2210, + "valid_targets_mean": 5056.1, + "valid_targets_min": 904 + }, + { + "epoch": 3.6490939044481054, + "grad_norm": 0.43130673319289825, + "learning_rate": 2.201756593954007e-05, + "loss": 0.1583, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15492397546768188, + "step": 2215, + "valid_targets_mean": 5113.2, + "valid_targets_min": 393 + }, + { + "epoch": 3.657331136738056, + "grad_norm": 0.5295744702809257, + "learning_rate": 2.1935813617504372e-05, + "loss": 0.1756, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16201339662075043, + "step": 2220, + "valid_targets_mean": 4780.9, + "valid_targets_min": 277 + }, + { + "epoch": 3.6655683690280068, + "grad_norm": 0.5946369400747427, + "learning_rate": 2.1854028631652958e-05, + "loss": 0.1702, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16037127375602722, + "step": 2225, + "valid_targets_mean": 3275.8, + "valid_targets_min": 279 + }, + { + "epoch": 3.673805601317957, + "grad_norm": 0.44952551623528814, + "learning_rate": 2.1772212361979053e-05, + "loss": 0.1648, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16338971257209778, + "step": 2230, + "valid_targets_mean": 5095.8, + "valid_targets_min": 346 + }, + { + "epoch": 3.6820428336079076, + "grad_norm": 0.43791872886873806, + "learning_rate": 2.1690366189003776e-05, + "loss": 0.1586, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15391787886619568, + "step": 2235, + "valid_targets_mean": 5727.9, + "valid_targets_min": 1342 + }, + { + "epoch": 3.6902800658978583, + "grad_norm": 0.3970801737083101, + "learning_rate": 2.16084914937528e-05, + "loss": 0.1678, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16625526547431946, + "step": 2240, + "valid_targets_mean": 6496.0, + "valid_targets_min": 325 + }, + { + "epoch": 3.698517298187809, + "grad_norm": 0.484519359275333, + "learning_rate": 2.1526589657733076e-05, + "loss": 0.1663, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14906881749629974, + "step": 2245, + "valid_targets_mean": 5056.9, + "valid_targets_min": 303 + }, + { + "epoch": 3.7067545304777596, + "grad_norm": 0.543054449266146, + "learning_rate": 2.1444662062909497e-05, + "loss": 0.1765, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18880906701087952, + "step": 2250, + "valid_targets_mean": 5087.6, + "valid_targets_min": 386 + }, + { + "epoch": 3.71499176276771, + "grad_norm": 0.412373847155115, + "learning_rate": 2.136271009168162e-05, + "loss": 0.1843, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16423764824867249, + "step": 2255, + "valid_targets_mean": 5792.0, + "valid_targets_min": 1805 + }, + { + "epoch": 3.7232289950576605, + "grad_norm": 0.4126065374679235, + "learning_rate": 2.128073512686029e-05, + "loss": 0.1561, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15496188402175903, + "step": 2260, + "valid_targets_mean": 6312.9, + "valid_targets_min": 331 + }, + { + "epoch": 3.731466227347611, + "grad_norm": 0.4737629695491757, + "learning_rate": 2.1198738551644344e-05, + "loss": 0.1623, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18088474869728088, + "step": 2265, + "valid_targets_mean": 4541.9, + "valid_targets_min": 331 + }, + { + "epoch": 3.739703459637562, + "grad_norm": 0.43047213132792306, + "learning_rate": 2.1116721749597265e-05, + "loss": 0.1655, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1655322015285492, + "step": 2270, + "valid_targets_mean": 5827.6, + "valid_targets_min": 316 + }, + { + "epoch": 3.7479406919275124, + "grad_norm": 0.3918342153717676, + "learning_rate": 2.1034686104623828e-05, + "loss": 0.1624, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16483767330646515, + "step": 2275, + "valid_targets_mean": 6364.0, + "valid_targets_min": 335 + }, + { + "epoch": 3.7561779242174627, + "grad_norm": 0.4478424801826196, + "learning_rate": 2.0952633000946753e-05, + "loss": 0.1623, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16138796508312225, + "step": 2280, + "valid_targets_mean": 5547.7, + "valid_targets_min": 374 + }, + { + "epoch": 3.7644151565074138, + "grad_norm": 0.4736058075123787, + "learning_rate": 2.0870563823083347e-05, + "loss": 0.1682, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1633884608745575, + "step": 2285, + "valid_targets_mean": 4713.3, + "valid_targets_min": 315 + }, + { + "epoch": 3.772652388797364, + "grad_norm": 0.4076171431317024, + "learning_rate": 2.0788479955822137e-05, + "loss": 0.167, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15652526915073395, + "step": 2290, + "valid_targets_mean": 6205.0, + "valid_targets_min": 393 + }, + { + "epoch": 3.7808896210873146, + "grad_norm": 0.5176507339253252, + "learning_rate": 2.0706382784199524e-05, + "loss": 0.16, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16365516185760498, + "step": 2295, + "valid_targets_mean": 6140.5, + "valid_targets_min": 335 + }, + { + "epoch": 3.7891268533772653, + "grad_norm": 0.45103078449937783, + "learning_rate": 2.06242736934764e-05, + "loss": 0.1706, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17881439626216888, + "step": 2300, + "valid_targets_mean": 4976.1, + "valid_targets_min": 1110 + }, + { + "epoch": 3.797364085667216, + "grad_norm": 0.4068376487085923, + "learning_rate": 2.0542154069114753e-05, + "loss": 0.1612, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16194987297058105, + "step": 2305, + "valid_targets_mean": 6501.2, + "valid_targets_min": 434 + }, + { + "epoch": 3.8056013179571666, + "grad_norm": 0.5485960350199107, + "learning_rate": 2.0460025296754326e-05, + "loss": 0.1686, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1692267209291458, + "step": 2310, + "valid_targets_mean": 4711.4, + "valid_targets_min": 405 + }, + { + "epoch": 3.813838550247117, + "grad_norm": 0.5349054412710811, + "learning_rate": 2.037788876218923e-05, + "loss": 0.1671, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1499655842781067, + "step": 2315, + "valid_targets_mean": 5649.0, + "valid_targets_min": 390 + }, + { + "epoch": 3.8220757825370675, + "grad_norm": 0.40935562611532483, + "learning_rate": 2.029574585134452e-05, + "loss": 0.1613, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1631348431110382, + "step": 2320, + "valid_targets_mean": 6357.0, + "valid_targets_min": 482 + }, + { + "epoch": 3.830313014827018, + "grad_norm": 0.4645878611326604, + "learning_rate": 2.0213597950252864e-05, + "loss": 0.1697, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17020228505134583, + "step": 2325, + "valid_targets_mean": 4671.6, + "valid_targets_min": 484 + }, + { + "epoch": 3.838550247116969, + "grad_norm": 0.4871420384446106, + "learning_rate": 2.0131446445031132e-05, + "loss": 0.1654, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16444051265716553, + "step": 2330, + "valid_targets_mean": 4968.4, + "valid_targets_min": 310 + }, + { + "epoch": 3.8467874794069195, + "grad_norm": 0.4813413395155231, + "learning_rate": 2.0049292721857005e-05, + "loss": 0.1658, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17408815026283264, + "step": 2335, + "valid_targets_mean": 5293.2, + "valid_targets_min": 768 + }, + { + "epoch": 3.8550247116968697, + "grad_norm": 0.4742090254671904, + "learning_rate": 1.9967138166945584e-05, + "loss": 0.165, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17291584610939026, + "step": 2340, + "valid_targets_mean": 6698.8, + "valid_targets_min": 4143 + }, + { + "epoch": 3.8632619439868203, + "grad_norm": 0.4163498895697308, + "learning_rate": 1.9884984166525996e-05, + "loss": 0.1741, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1733531653881073, + "step": 2345, + "valid_targets_mean": 6222.6, + "valid_targets_min": 858 + }, + { + "epoch": 3.871499176276771, + "grad_norm": 0.46658867545999017, + "learning_rate": 1.9802832106818038e-05, + "loss": 0.163, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16182400286197662, + "step": 2350, + "valid_targets_mean": 5577.4, + "valid_targets_min": 405 + }, + { + "epoch": 3.8797364085667216, + "grad_norm": 0.6884250760473662, + "learning_rate": 1.9720683374008742e-05, + "loss": 0.1738, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1815515160560608, + "step": 2355, + "valid_targets_mean": 4938.5, + "valid_targets_min": 474 + }, + { + "epoch": 3.8879736408566723, + "grad_norm": 0.44499718070883826, + "learning_rate": 1.9638539354228994e-05, + "loss": 0.1618, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1717439889907837, + "step": 2360, + "valid_targets_mean": 5651.2, + "valid_targets_min": 423 + }, + { + "epoch": 3.8962108731466225, + "grad_norm": 0.40457261570684494, + "learning_rate": 1.9556401433530186e-05, + "loss": 0.1622, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15134787559509277, + "step": 2365, + "valid_targets_mean": 5634.9, + "valid_targets_min": 351 + }, + { + "epoch": 3.904448105436573, + "grad_norm": 0.6553583489153849, + "learning_rate": 1.947427099786077e-05, + "loss": 0.1706, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16197733581066132, + "step": 2370, + "valid_targets_mean": 4561.9, + "valid_targets_min": 301 + }, + { + "epoch": 3.912685337726524, + "grad_norm": 0.4157227379076624, + "learning_rate": 1.9392149433042915e-05, + "loss": 0.1626, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16081485152244568, + "step": 2375, + "valid_targets_mean": 5540.6, + "valid_targets_min": 329 + }, + { + "epoch": 3.9209225700164745, + "grad_norm": 0.4359729610143708, + "learning_rate": 1.9310038124749102e-05, + "loss": 0.1651, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15543285012245178, + "step": 2380, + "valid_targets_mean": 5001.1, + "valid_targets_min": 418 + }, + { + "epoch": 3.929159802306425, + "grad_norm": 0.7705180986032697, + "learning_rate": 1.9227938458478744e-05, + "loss": 0.167, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1696222722530365, + "step": 2385, + "valid_targets_mean": 4766.1, + "valid_targets_min": 303 + }, + { + "epoch": 3.9373970345963754, + "grad_norm": 0.4047261335899435, + "learning_rate": 1.9145851819534824e-05, + "loss": 0.1757, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15419232845306396, + "step": 2390, + "valid_targets_mean": 6061.1, + "valid_targets_min": 447 + }, + { + "epoch": 3.9456342668863265, + "grad_norm": 0.41724450128225465, + "learning_rate": 1.9063779593000507e-05, + "loss": 0.1603, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1801353394985199, + "step": 2395, + "valid_targets_mean": 6642.6, + "valid_targets_min": 396 + }, + { + "epoch": 3.9538714991762767, + "grad_norm": 0.4423034009298218, + "learning_rate": 1.8981723163715756e-05, + "loss": 0.1638, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17328378558158875, + "step": 2400, + "valid_targets_mean": 5702.5, + "valid_targets_min": 432 + }, + { + "epoch": 3.9621087314662273, + "grad_norm": 0.444583031013224, + "learning_rate": 1.8899683916254013e-05, + "loss": 0.1671, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16314169764518738, + "step": 2405, + "valid_targets_mean": 5586.6, + "valid_targets_min": 465 + }, + { + "epoch": 3.970345963756178, + "grad_norm": 0.5567518529012295, + "learning_rate": 1.881766323489877e-05, + "loss": 0.1515, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15508978068828583, + "step": 2410, + "valid_targets_mean": 5686.6, + "valid_targets_min": 842 + }, + { + "epoch": 3.9785831960461286, + "grad_norm": 0.4528089798600689, + "learning_rate": 1.8735662503620268e-05, + "loss": 0.1616, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1737988293170929, + "step": 2415, + "valid_targets_mean": 5447.9, + "valid_targets_min": 270 + }, + { + "epoch": 3.9868204283360793, + "grad_norm": 0.4437374282788194, + "learning_rate": 1.8653683106052097e-05, + "loss": 0.1737, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16056564450263977, + "step": 2420, + "valid_targets_mean": 5152.9, + "valid_targets_min": 1012 + }, + { + "epoch": 3.9950576606260295, + "grad_norm": 0.4629669597591696, + "learning_rate": 1.857172642546791e-05, + "loss": 0.1631, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16307508945465088, + "step": 2425, + "valid_targets_mean": 4988.5, + "valid_targets_min": 327 + }, + { + "epoch": 4.003294892915981, + "grad_norm": 0.362102397580454, + "learning_rate": 1.8489793844758015e-05, + "loss": 0.149, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12046860158443451, + "step": 2430, + "valid_targets_mean": 7373.6, + "valid_targets_min": 2163 + }, + { + "epoch": 4.011532125205931, + "grad_norm": 0.4588004757668333, + "learning_rate": 1.8407886746406082e-05, + "loss": 0.1504, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12914225459098816, + "step": 2435, + "valid_targets_mean": 4912.9, + "valid_targets_min": 243 + }, + { + "epoch": 4.019769357495881, + "grad_norm": 0.46936874186243444, + "learning_rate": 1.8326006512465806e-05, + "loss": 0.1515, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15502551198005676, + "step": 2440, + "valid_targets_mean": 5579.7, + "valid_targets_min": 366 + }, + { + "epoch": 4.028006589785832, + "grad_norm": 0.42358155648588003, + "learning_rate": 1.8244154524537602e-05, + "loss": 0.1506, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1456337869167328, + "step": 2445, + "valid_targets_mean": 6027.8, + "valid_targets_min": 315 + }, + { + "epoch": 4.036243822075782, + "grad_norm": 0.411694955371839, + "learning_rate": 1.8162332163745253e-05, + "loss": 0.1535, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15597039461135864, + "step": 2450, + "valid_targets_mean": 6774.1, + "valid_targets_min": 2283 + }, + { + "epoch": 4.0444810543657335, + "grad_norm": 0.47459473453439693, + "learning_rate": 1.8080540810712652e-05, + "loss": 0.1474, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17040134966373444, + "step": 2455, + "valid_targets_mean": 5131.1, + "valid_targets_min": 346 + }, + { + "epoch": 4.052718286655684, + "grad_norm": 0.41826585402865163, + "learning_rate": 1.799878184554047e-05, + "loss": 0.1506, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1321791410446167, + "step": 2460, + "valid_targets_mean": 6608.9, + "valid_targets_min": 3516 + }, + { + "epoch": 4.060955518945634, + "grad_norm": 0.4155868317445694, + "learning_rate": 1.7917056647782888e-05, + "loss": 0.1443, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13027681410312653, + "step": 2465, + "valid_targets_mean": 5578.6, + "valid_targets_min": 303 + }, + { + "epoch": 4.069192751235585, + "grad_norm": 0.46080020104599295, + "learning_rate": 1.7835366596424317e-05, + "loss": 0.1553, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14736926555633545, + "step": 2470, + "valid_targets_mean": 5789.2, + "valid_targets_min": 2207 + }, + { + "epoch": 4.077429983525535, + "grad_norm": 0.7024450667359199, + "learning_rate": 1.775371306985613e-05, + "loss": 0.1473, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.141681507229805, + "step": 2475, + "valid_targets_mean": 5083.5, + "valid_targets_min": 338 + }, + { + "epoch": 4.085667215815486, + "grad_norm": 0.4645130391173883, + "learning_rate": 1.7672097445853388e-05, + "loss": 0.1515, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15968500077724457, + "step": 2480, + "valid_targets_mean": 5763.9, + "valid_targets_min": 426 + }, + { + "epoch": 4.0939044481054365, + "grad_norm": 0.4784694238147179, + "learning_rate": 1.7590521101551635e-05, + "loss": 0.1577, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16522473096847534, + "step": 2485, + "valid_targets_mean": 5397.2, + "valid_targets_min": 379 + }, + { + "epoch": 4.102141680395387, + "grad_norm": 0.46501246791017226, + "learning_rate": 1.7508985413423595e-05, + "loss": 0.1444, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13691683113574982, + "step": 2490, + "valid_targets_mean": 4509.5, + "valid_targets_min": 306 + }, + { + "epoch": 4.110378912685338, + "grad_norm": 0.5700981481383305, + "learning_rate": 1.742749175725601e-05, + "loss": 0.148, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1452605426311493, + "step": 2495, + "valid_targets_mean": 3580.9, + "valid_targets_min": 336 + }, + { + "epoch": 4.118616144975288, + "grad_norm": 0.5705186798816394, + "learning_rate": 1.7346041508126386e-05, + "loss": 0.1472, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17359477281570435, + "step": 2500, + "valid_targets_mean": 4987.8, + "valid_targets_min": 295 + }, + { + "epoch": 4.126853377265239, + "grad_norm": 0.4400809895847404, + "learning_rate": 1.7264636040379808e-05, + "loss": 0.1484, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13463068008422852, + "step": 2505, + "valid_targets_mean": 5214.5, + "valid_targets_min": 722 + }, + { + "epoch": 4.135090609555189, + "grad_norm": 0.4877814194690551, + "learning_rate": 1.7183276727605743e-05, + "loss": 0.1444, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15212027728557587, + "step": 2510, + "valid_targets_mean": 4686.6, + "valid_targets_min": 424 + }, + { + "epoch": 4.1433278418451405, + "grad_norm": 0.5285967708806336, + "learning_rate": 1.7101964942614864e-05, + "loss": 0.1576, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14906981587409973, + "step": 2515, + "valid_targets_mean": 4191.4, + "valid_targets_min": 362 + }, + { + "epoch": 4.151565074135091, + "grad_norm": 0.5783330832204944, + "learning_rate": 1.7020702057415884e-05, + "loss": 0.1494, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14080877602100372, + "step": 2520, + "valid_targets_mean": 5295.0, + "valid_targets_min": 347 + }, + { + "epoch": 4.159802306425041, + "grad_norm": 0.6973570205068965, + "learning_rate": 1.6939489443192435e-05, + "loss": 0.1589, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1889844387769699, + "step": 2525, + "valid_targets_mean": 3328.6, + "valid_targets_min": 303 + }, + { + "epoch": 4.168039538714992, + "grad_norm": 0.44997217802378553, + "learning_rate": 1.6858328470279862e-05, + "loss": 0.1532, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15209095180034637, + "step": 2530, + "valid_targets_mean": 5899.5, + "valid_targets_min": 3436 + }, + { + "epoch": 4.176276771004942, + "grad_norm": 0.4625339471080827, + "learning_rate": 1.6777220508142184e-05, + "loss": 0.1511, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14391621947288513, + "step": 2535, + "valid_targets_mean": 5531.8, + "valid_targets_min": 656 + }, + { + "epoch": 4.184514003294893, + "grad_norm": 0.4957253313719207, + "learning_rate": 1.6696166925348924e-05, + "loss": 0.148, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13907882571220398, + "step": 2540, + "valid_targets_mean": 4948.8, + "valid_targets_min": 310 + }, + { + "epoch": 4.1927512355848435, + "grad_norm": 0.7565853150262708, + "learning_rate": 1.6615169089552038e-05, + "loss": 0.1506, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1515434831380844, + "step": 2545, + "valid_targets_mean": 5125.2, + "valid_targets_min": 347 + }, + { + "epoch": 4.200988467874794, + "grad_norm": 0.49753975963313524, + "learning_rate": 1.6534228367462843e-05, + "loss": 0.1476, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15490424633026123, + "step": 2550, + "valid_targets_mean": 5573.6, + "valid_targets_min": 315 + }, + { + "epoch": 4.209225700164745, + "grad_norm": 0.5345554614335993, + "learning_rate": 1.6453346124828966e-05, + "loss": 0.1529, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18045078217983246, + "step": 2555, + "valid_targets_mean": 4684.8, + "valid_targets_min": 268 + }, + { + "epoch": 4.217462932454695, + "grad_norm": 0.4613752387425423, + "learning_rate": 1.6372523726411253e-05, + "loss": 0.1491, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14130446314811707, + "step": 2560, + "valid_targets_mean": 5415.4, + "valid_targets_min": 343 + }, + { + "epoch": 4.225700164744646, + "grad_norm": 0.4525908163882441, + "learning_rate": 1.6291762535960817e-05, + "loss": 0.1524, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16135191917419434, + "step": 2565, + "valid_targets_mean": 5391.4, + "valid_targets_min": 399 + }, + { + "epoch": 4.233937397034596, + "grad_norm": 0.4596645755169701, + "learning_rate": 1.6211063916195945e-05, + "loss": 0.1437, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15719518065452576, + "step": 2570, + "valid_targets_mean": 5872.5, + "valid_targets_min": 403 + }, + { + "epoch": 4.242174629324547, + "grad_norm": 0.5069424876548778, + "learning_rate": 1.6130429228779154e-05, + "loss": 0.1503, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16834935545921326, + "step": 2575, + "valid_targets_mean": 5199.9, + "valid_targets_min": 344 + }, + { + "epoch": 4.250411861614498, + "grad_norm": 0.48516012282655313, + "learning_rate": 1.604985983429421e-05, + "loss": 0.1495, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1605086624622345, + "step": 2580, + "valid_targets_mean": 5155.4, + "valid_targets_min": 407 + }, + { + "epoch": 4.258649093904448, + "grad_norm": 0.4468201225192397, + "learning_rate": 1.596935709222315e-05, + "loss": 0.1498, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1490262746810913, + "step": 2585, + "valid_targets_mean": 6065.2, + "valid_targets_min": 1807 + }, + { + "epoch": 4.266886326194399, + "grad_norm": 0.41002444382532305, + "learning_rate": 1.5888922360923375e-05, + "loss": 0.1488, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13134261965751648, + "step": 2590, + "valid_targets_mean": 6304.5, + "valid_targets_min": 303 + }, + { + "epoch": 4.275123558484349, + "grad_norm": 0.5102813058004463, + "learning_rate": 1.5808556997604684e-05, + "loss": 0.1477, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14674484729766846, + "step": 2595, + "valid_targets_mean": 4860.8, + "valid_targets_min": 364 + }, + { + "epoch": 4.283360790774299, + "grad_norm": 0.4368439014835675, + "learning_rate": 1.5728262358306418e-05, + "loss": 0.1464, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1445227563381195, + "step": 2600, + "valid_targets_mean": 5476.2, + "valid_targets_min": 1056 + }, + { + "epoch": 4.2915980230642505, + "grad_norm": 0.5676665382040276, + "learning_rate": 1.564803979787457e-05, + "loss": 0.1531, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15031370520591736, + "step": 2605, + "valid_targets_mean": 3444.8, + "valid_targets_min": 253 + }, + { + "epoch": 4.299835255354201, + "grad_norm": 0.5109094701134649, + "learning_rate": 1.5567890669938903e-05, + "loss": 0.1509, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15295559167861938, + "step": 2610, + "valid_targets_mean": 5027.6, + "valid_targets_min": 719 + }, + { + "epoch": 4.308072487644152, + "grad_norm": 0.45013114123730513, + "learning_rate": 1.5487816326890118e-05, + "loss": 0.1629, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14927126467227936, + "step": 2615, + "valid_targets_mean": 6415.9, + "valid_targets_min": 1665 + }, + { + "epoch": 4.316309719934102, + "grad_norm": 0.43711700962396716, + "learning_rate": 1.540781811985704e-05, + "loss": 0.1452, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14677034318447113, + "step": 2620, + "valid_targets_mean": 5701.2, + "valid_targets_min": 403 + }, + { + "epoch": 4.324546952224052, + "grad_norm": 0.4567319205617331, + "learning_rate": 1.5327897398683827e-05, + "loss": 0.1482, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1432798206806183, + "step": 2625, + "valid_targets_mean": 6331.5, + "valid_targets_min": 1060 + }, + { + "epoch": 4.332784184514003, + "grad_norm": 0.6150417219974588, + "learning_rate": 1.5248055511907183e-05, + "loss": 0.1532, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1534081995487213, + "step": 2630, + "valid_targets_mean": 5722.9, + "valid_targets_min": 895 + }, + { + "epoch": 4.341021416803954, + "grad_norm": 0.5439641469775286, + "learning_rate": 1.5168293806733594e-05, + "loss": 0.1612, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15055815875530243, + "step": 2635, + "valid_targets_mean": 3872.1, + "valid_targets_min": 301 + }, + { + "epoch": 4.349258649093905, + "grad_norm": 0.49432098862551554, + "learning_rate": 1.5088613629016613e-05, + "loss": 0.158, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16186079382896423, + "step": 2640, + "valid_targets_mean": 5156.1, + "valid_targets_min": 295 + }, + { + "epoch": 4.357495881383855, + "grad_norm": 0.4377662832647051, + "learning_rate": 1.5009016323234158e-05, + "loss": 0.159, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1668049693107605, + "step": 2645, + "valid_targets_mean": 6354.6, + "valid_targets_min": 383 + }, + { + "epoch": 4.365733113673806, + "grad_norm": 0.4969241114877333, + "learning_rate": 1.4929503232465804e-05, + "loss": 0.1484, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15549027919769287, + "step": 2650, + "valid_targets_mean": 4336.9, + "valid_targets_min": 325 + }, + { + "epoch": 4.373970345963756, + "grad_norm": 0.39112878825838426, + "learning_rate": 1.485007569837011e-05, + "loss": 0.1444, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13279537856578827, + "step": 2655, + "valid_targets_mean": 6286.1, + "valid_targets_min": 3182 + }, + { + "epoch": 4.382207578253706, + "grad_norm": 0.4899567925864192, + "learning_rate": 1.477073506116204e-05, + "loss": 0.1515, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17375296354293823, + "step": 2660, + "valid_targets_mean": 5219.9, + "valid_targets_min": 211 + }, + { + "epoch": 4.3904448105436575, + "grad_norm": 0.45711905561871397, + "learning_rate": 1.469148265959028e-05, + "loss": 0.1534, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1331019401550293, + "step": 2665, + "valid_targets_mean": 5110.2, + "valid_targets_min": 382 + }, + { + "epoch": 4.398682042833608, + "grad_norm": 0.5069456092396808, + "learning_rate": 1.4612319830914684e-05, + "loss": 0.1516, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14841151237487793, + "step": 2670, + "valid_targets_mean": 4724.4, + "valid_targets_min": 487 + }, + { + "epoch": 4.406919275123559, + "grad_norm": 0.44034448930841963, + "learning_rate": 1.4533247910883704e-05, + "loss": 0.1446, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1460045576095581, + "step": 2675, + "valid_targets_mean": 5382.2, + "valid_targets_min": 252 + }, + { + "epoch": 4.415156507413509, + "grad_norm": 0.4085357778933215, + "learning_rate": 1.445426823371186e-05, + "loss": 0.1468, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14165529608726501, + "step": 2680, + "valid_targets_mean": 6587.8, + "valid_targets_min": 3807 + }, + { + "epoch": 4.423393739703459, + "grad_norm": 0.4985432420723232, + "learning_rate": 1.4375382132057207e-05, + "loss": 0.1541, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14564910531044006, + "step": 2685, + "valid_targets_mean": 7100.1, + "valid_targets_min": 1402 + }, + { + "epoch": 4.43163097199341, + "grad_norm": 0.4558487525400957, + "learning_rate": 1.4296590936998875e-05, + "loss": 0.1527, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13396936655044556, + "step": 2690, + "valid_targets_mean": 5222.4, + "valid_targets_min": 1873 + }, + { + "epoch": 4.439868204283361, + "grad_norm": 0.5212832580946739, + "learning_rate": 1.4217895978014583e-05, + "loss": 0.1475, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16112340986728668, + "step": 2695, + "valid_targets_mean": 6251.2, + "valid_targets_min": 2514 + }, + { + "epoch": 4.448105436573312, + "grad_norm": 0.511176800268245, + "learning_rate": 1.4139298582958208e-05, + "loss": 0.1586, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17118722200393677, + "step": 2700, + "valid_targets_mean": 5283.8, + "valid_targets_min": 475 + }, + { + "epoch": 4.456342668863262, + "grad_norm": 0.4133253589641243, + "learning_rate": 1.4060800078037407e-05, + "loss": 0.1453, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13945919275283813, + "step": 2705, + "valid_targets_mean": 6163.3, + "valid_targets_min": 348 + }, + { + "epoch": 4.464579901153212, + "grad_norm": 0.5355709069173332, + "learning_rate": 1.3982401787791215e-05, + "loss": 0.1589, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16837137937545776, + "step": 2710, + "valid_targets_mean": 4398.6, + "valid_targets_min": 307 + }, + { + "epoch": 4.472817133443163, + "grad_norm": 0.47608650460084356, + "learning_rate": 1.3904105035067701e-05, + "loss": 0.1492, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14236906170845032, + "step": 2715, + "valid_targets_mean": 5176.2, + "valid_targets_min": 357 + }, + { + "epoch": 4.481054365733113, + "grad_norm": 0.4680478602386323, + "learning_rate": 1.3825911141001637e-05, + "loss": 0.1504, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14760847389698029, + "step": 2720, + "valid_targets_mean": 5412.5, + "valid_targets_min": 282 + }, + { + "epoch": 4.4892915980230645, + "grad_norm": 0.619587835664657, + "learning_rate": 1.3747821424992245e-05, + "loss": 0.1506, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15162773430347443, + "step": 2725, + "valid_targets_mean": 5752.3, + "valid_targets_min": 400 + }, + { + "epoch": 4.497528830313015, + "grad_norm": 0.4661528683504612, + "learning_rate": 1.3669837204680879e-05, + "loss": 0.1488, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14053663611412048, + "step": 2730, + "valid_targets_mean": 5016.6, + "valid_targets_min": 410 + }, + { + "epoch": 4.505766062602966, + "grad_norm": 0.42047460260378905, + "learning_rate": 1.3591959795928821e-05, + "loss": 0.1482, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15329483151435852, + "step": 2735, + "valid_targets_mean": 6674.5, + "valid_targets_min": 3457 + }, + { + "epoch": 4.514003294892916, + "grad_norm": 0.5636482204398094, + "learning_rate": 1.3514190512795096e-05, + "loss": 0.1437, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14222252368927002, + "step": 2740, + "valid_targets_mean": 5046.1, + "valid_targets_min": 281 + }, + { + "epoch": 4.522240527182866, + "grad_norm": 0.45583530674550665, + "learning_rate": 1.3436530667514258e-05, + "loss": 0.1517, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15300938487052917, + "step": 2745, + "valid_targets_mean": 5698.6, + "valid_targets_min": 323 + }, + { + "epoch": 4.530477759472817, + "grad_norm": 0.4261953081142523, + "learning_rate": 1.3358981570474276e-05, + "loss": 0.1465, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13072721660137177, + "step": 2750, + "valid_targets_mean": 5420.9, + "valid_targets_min": 366 + }, + { + "epoch": 4.538714991762768, + "grad_norm": 0.5880573918987582, + "learning_rate": 1.3281544530194412e-05, + "loss": 0.1463, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14867545664310455, + "step": 2755, + "valid_targets_mean": 5139.9, + "valid_targets_min": 1453 + }, + { + "epoch": 4.546952224052719, + "grad_norm": 0.4689692885338706, + "learning_rate": 1.3204220853303157e-05, + "loss": 0.1497, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14390037953853607, + "step": 2760, + "valid_targets_mean": 4890.1, + "valid_targets_min": 336 + }, + { + "epoch": 4.555189456342669, + "grad_norm": 0.5368370242051151, + "learning_rate": 1.3127011844516165e-05, + "loss": 0.1539, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14073890447616577, + "step": 2765, + "valid_targets_mean": 5944.6, + "valid_targets_min": 841 + }, + { + "epoch": 4.563426688632619, + "grad_norm": 0.4342038972954832, + "learning_rate": 1.304991880661425e-05, + "loss": 0.1486, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14130136370658875, + "step": 2770, + "valid_targets_mean": 5363.7, + "valid_targets_min": 360 + }, + { + "epoch": 4.57166392092257, + "grad_norm": 0.5021035499032942, + "learning_rate": 1.2972943040421398e-05, + "loss": 0.1524, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.146042600274086, + "step": 2775, + "valid_targets_mean": 7223.3, + "valid_targets_min": 374 + }, + { + "epoch": 4.5799011532125204, + "grad_norm": 0.4739904329110086, + "learning_rate": 1.2896085844782808e-05, + "loss": 0.1505, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14053794741630554, + "step": 2780, + "valid_targets_mean": 6500.1, + "valid_targets_min": 2785 + }, + { + "epoch": 4.5881383855024716, + "grad_norm": 0.43782595703510596, + "learning_rate": 1.2819348516543003e-05, + "loss": 0.1489, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1457478106021881, + "step": 2785, + "valid_targets_mean": 5669.9, + "valid_targets_min": 350 + }, + { + "epoch": 4.596375617792422, + "grad_norm": 0.5041051641997347, + "learning_rate": 1.2742732350523928e-05, + "loss": 0.1582, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14434212446212769, + "step": 2790, + "valid_targets_mean": 4667.4, + "valid_targets_min": 348 + }, + { + "epoch": 4.604612850082372, + "grad_norm": 0.5876482351362182, + "learning_rate": 1.2666238639503099e-05, + "loss": 0.1648, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1538345068693161, + "step": 2795, + "valid_targets_mean": 5997.9, + "valid_targets_min": 306 + }, + { + "epoch": 4.612850082372323, + "grad_norm": 0.4801368026632177, + "learning_rate": 1.2589868674191794e-05, + "loss": 0.1535, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15362213551998138, + "step": 2800, + "valid_targets_mean": 5321.2, + "valid_targets_min": 435 + }, + { + "epoch": 4.621087314662273, + "grad_norm": 0.48008674858622163, + "learning_rate": 1.2513623743213297e-05, + "loss": 0.1647, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16761736571788788, + "step": 2805, + "valid_targets_mean": 5240.1, + "valid_targets_min": 327 + }, + { + "epoch": 4.629324546952224, + "grad_norm": 0.39861573361053565, + "learning_rate": 1.243750513308111e-05, + "loss": 0.1423, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14428246021270752, + "step": 2810, + "valid_targets_mean": 6673.4, + "valid_targets_min": 2183 + }, + { + "epoch": 4.637561779242175, + "grad_norm": 0.4416153244849873, + "learning_rate": 1.2361514128177286e-05, + "loss": 0.1503, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1608775556087494, + "step": 2815, + "valid_targets_mean": 5941.4, + "valid_targets_min": 353 + }, + { + "epoch": 4.645799011532125, + "grad_norm": 0.5094541742636675, + "learning_rate": 1.2285652010730727e-05, + "loss": 0.1509, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15818873047828674, + "step": 2820, + "valid_targets_mean": 4775.7, + "valid_targets_min": 709 + }, + { + "epoch": 4.654036243822076, + "grad_norm": 0.5138894968734843, + "learning_rate": 1.220992006079557e-05, + "loss": 0.1508, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17340649664402008, + "step": 2825, + "valid_targets_mean": 5142.4, + "valid_targets_min": 780 + }, + { + "epoch": 4.662273476112026, + "grad_norm": 0.4349646733764812, + "learning_rate": 1.2134319556229577e-05, + "loss": 0.1562, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17126768827438354, + "step": 2830, + "valid_targets_mean": 6356.2, + "valid_targets_min": 335 + }, + { + "epoch": 4.670510708401977, + "grad_norm": 0.4590448852323375, + "learning_rate": 1.2058851772672569e-05, + "loss": 0.1595, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.18234387040138245, + "step": 2835, + "valid_targets_mean": 5573.4, + "valid_targets_min": 292 + }, + { + "epoch": 4.6787479406919275, + "grad_norm": 0.4682703319290092, + "learning_rate": 1.1983517983524917e-05, + "loss": 0.1583, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15168213844299316, + "step": 2840, + "valid_targets_mean": 5342.8, + "valid_targets_min": 498 + }, + { + "epoch": 4.686985172981878, + "grad_norm": 0.46709472071321484, + "learning_rate": 1.1908319459926059e-05, + "loss": 0.1497, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15055695176124573, + "step": 2845, + "valid_targets_mean": 5698.6, + "valid_targets_min": 430 + }, + { + "epoch": 4.695222405271829, + "grad_norm": 0.41530709694840034, + "learning_rate": 1.183325747073301e-05, + "loss": 0.1504, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1470167487859726, + "step": 2850, + "valid_targets_mean": 6569.5, + "valid_targets_min": 2168 + }, + { + "epoch": 4.703459637561779, + "grad_norm": 0.4986031558793924, + "learning_rate": 1.1758333282499004e-05, + "loss": 0.1616, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15256115794181824, + "step": 2855, + "valid_targets_mean": 5291.1, + "valid_targets_min": 356 + }, + { + "epoch": 4.71169686985173, + "grad_norm": 0.4644899790045432, + "learning_rate": 1.1683548159452083e-05, + "loss": 0.144, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13906921446323395, + "step": 2860, + "valid_targets_mean": 5042.7, + "valid_targets_min": 298 + }, + { + "epoch": 4.71993410214168, + "grad_norm": 0.4628750656923671, + "learning_rate": 1.1608903363473804e-05, + "loss": 0.1447, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15038900077342987, + "step": 2865, + "valid_targets_mean": 5170.4, + "valid_targets_min": 392 + }, + { + "epoch": 4.7281713344316305, + "grad_norm": 0.46077501269852494, + "learning_rate": 1.1534400154077901e-05, + "loss": 0.1557, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14422756433486938, + "step": 2870, + "valid_targets_mean": 5135.5, + "valid_targets_min": 562 + }, + { + "epoch": 4.736408566721582, + "grad_norm": 0.44080772013963004, + "learning_rate": 1.146003978838908e-05, + "loss": 0.1468, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15541505813598633, + "step": 2875, + "valid_targets_mean": 5824.4, + "valid_targets_min": 2243 + }, + { + "epoch": 4.744645799011532, + "grad_norm": 0.5154411270196606, + "learning_rate": 1.1385823521121755e-05, + "loss": 0.1556, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1654052883386612, + "step": 2880, + "valid_targets_mean": 5482.2, + "valid_targets_min": 390 + }, + { + "epoch": 4.752883031301483, + "grad_norm": 0.48283476082428284, + "learning_rate": 1.1311752604558945e-05, + "loss": 0.1458, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1402990221977234, + "step": 2885, + "valid_targets_mean": 5145.9, + "valid_targets_min": 385 + }, + { + "epoch": 4.761120263591433, + "grad_norm": 0.4694411295793455, + "learning_rate": 1.1237828288531065e-05, + "loss": 0.1492, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13541021943092346, + "step": 2890, + "valid_targets_mean": 5297.4, + "valid_targets_min": 397 + }, + { + "epoch": 4.769357495881383, + "grad_norm": 0.42611432241362635, + "learning_rate": 1.1164051820394903e-05, + "loss": 0.1558, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1598133146762848, + "step": 2895, + "valid_targets_mean": 6070.9, + "valid_targets_min": 309 + }, + { + "epoch": 4.7775947281713345, + "grad_norm": 0.416503846836609, + "learning_rate": 1.1090424445012524e-05, + "loss": 0.153, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14090408384799957, + "step": 2900, + "valid_targets_mean": 6663.6, + "valid_targets_min": 1886 + }, + { + "epoch": 4.785831960461285, + "grad_norm": 2.5787771100055266, + "learning_rate": 1.1016947404730302e-05, + "loss": 0.1579, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14577308297157288, + "step": 2905, + "valid_targets_mean": 6727.4, + "valid_targets_min": 2421 + }, + { + "epoch": 4.794069192751236, + "grad_norm": 0.40795595011087066, + "learning_rate": 1.0943621939357952e-05, + "loss": 0.152, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1484503597021103, + "step": 2910, + "valid_targets_mean": 6500.4, + "valid_targets_min": 997 + }, + { + "epoch": 4.802306425041186, + "grad_norm": 0.43831358879625426, + "learning_rate": 1.0870449286147565e-05, + "loss": 0.1491, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14645683765411377, + "step": 2915, + "valid_targets_mean": 6608.6, + "valid_targets_min": 771 + }, + { + "epoch": 4.810543657331137, + "grad_norm": 0.5171802770026964, + "learning_rate": 1.07974306797728e-05, + "loss": 0.1546, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14750051498413086, + "step": 2920, + "valid_targets_mean": 4631.4, + "valid_targets_min": 323 + }, + { + "epoch": 4.818780889621087, + "grad_norm": 0.4227242008790658, + "learning_rate": 1.0724567352308008e-05, + "loss": 0.1424, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14488349854946136, + "step": 2925, + "valid_targets_mean": 6282.1, + "valid_targets_min": 642 + }, + { + "epoch": 4.8270181219110375, + "grad_norm": 0.4478606504649285, + "learning_rate": 1.0651860533207448e-05, + "loss": 0.1502, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1508277952671051, + "step": 2930, + "valid_targets_mean": 5273.1, + "valid_targets_min": 615 + }, + { + "epoch": 4.835255354200989, + "grad_norm": 0.42744288780436157, + "learning_rate": 1.0579311449284555e-05, + "loss": 0.1389, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14862512052059174, + "step": 2935, + "valid_targets_mean": 5654.1, + "valid_targets_min": 330 + }, + { + "epoch": 4.843492586490939, + "grad_norm": 0.511057530937132, + "learning_rate": 1.0506921324691209e-05, + "loss": 0.1471, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13991223275661469, + "step": 2940, + "valid_targets_mean": 5849.8, + "valid_targets_min": 239 + }, + { + "epoch": 4.85172981878089, + "grad_norm": 0.43402347813103515, + "learning_rate": 1.0434691380897132e-05, + "loss": 0.1422, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14862912893295288, + "step": 2945, + "valid_targets_mean": 7194.9, + "valid_targets_min": 366 + }, + { + "epoch": 4.85996705107084, + "grad_norm": 0.5276452752375149, + "learning_rate": 1.0362622836669216e-05, + "loss": 0.1486, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15401889383792877, + "step": 2950, + "valid_targets_mean": 5620.6, + "valid_targets_min": 711 + }, + { + "epoch": 4.868204283360791, + "grad_norm": 0.4492778144090402, + "learning_rate": 1.0290716908051009e-05, + "loss": 0.1502, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16334179043769836, + "step": 2955, + "valid_targets_mean": 6076.8, + "valid_targets_min": 365 + }, + { + "epoch": 4.8764415156507415, + "grad_norm": 0.47655472357038325, + "learning_rate": 1.0218974808342168e-05, + "loss": 0.1696, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17806747555732727, + "step": 2960, + "valid_targets_mean": 5522.4, + "valid_targets_min": 344 + }, + { + "epoch": 4.884678747940692, + "grad_norm": 0.45349469543761634, + "learning_rate": 1.0147397748077999e-05, + "loss": 0.1457, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1454918086528778, + "step": 2965, + "valid_targets_mean": 5321.0, + "valid_targets_min": 224 + }, + { + "epoch": 4.892915980230643, + "grad_norm": 0.43886529178564526, + "learning_rate": 1.007598693500903e-05, + "loss": 0.1546, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14286306500434875, + "step": 2970, + "valid_targets_mean": 5753.2, + "valid_targets_min": 802 + }, + { + "epoch": 4.901153212520593, + "grad_norm": 0.518966892778548, + "learning_rate": 1.0004743574080607e-05, + "loss": 0.1578, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1655164361000061, + "step": 2975, + "valid_targets_mean": 4567.6, + "valid_targets_min": 312 + }, + { + "epoch": 4.909390444810544, + "grad_norm": 0.4440850818507626, + "learning_rate": 9.93366886741261e-06, + "loss": 0.1481, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15223294496536255, + "step": 2980, + "valid_targets_mean": 6050.8, + "valid_targets_min": 1823 + }, + { + "epoch": 4.917627677100494, + "grad_norm": 0.4514254890627531, + "learning_rate": 9.862764014279134e-06, + "loss": 0.1544, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14469844102859497, + "step": 2985, + "valid_targets_mean": 6442.1, + "valid_targets_min": 846 + }, + { + "epoch": 4.9258649093904445, + "grad_norm": 0.43580792454868544, + "learning_rate": 9.792030211088248e-06, + "loss": 0.145, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1455616056919098, + "step": 2990, + "valid_targets_mean": 6012.9, + "valid_targets_min": 1013 + }, + { + "epoch": 4.934102141680396, + "grad_norm": 0.43958810455468345, + "learning_rate": 9.721468651361852e-06, + "loss": 0.1495, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17629413306713104, + "step": 2995, + "valid_targets_mean": 6032.8, + "valid_targets_min": 934 + }, + { + "epoch": 4.942339373970346, + "grad_norm": 0.5343160717700296, + "learning_rate": 9.651080525715464e-06, + "loss": 0.149, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15880396962165833, + "step": 3000, + "valid_targets_mean": 4276.4, + "valid_targets_min": 576 + }, + { + "epoch": 4.950576606260297, + "grad_norm": 0.4662434799894812, + "learning_rate": 9.580867021838234e-06, + "loss": 0.1432, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13793358206748962, + "step": 3005, + "valid_targets_mean": 5537.8, + "valid_targets_min": 1519 + }, + { + "epoch": 4.958813838550247, + "grad_norm": 0.47078971320012264, + "learning_rate": 9.510829324472784e-06, + "loss": 0.1494, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1519417017698288, + "step": 3010, + "valid_targets_mean": 5382.9, + "valid_targets_min": 351 + }, + { + "epoch": 4.967051070840197, + "grad_norm": 0.4478704939612275, + "learning_rate": 9.440968615395314e-06, + "loss": 0.1489, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14127302169799805, + "step": 3015, + "valid_targets_mean": 6064.4, + "valid_targets_min": 1989 + }, + { + "epoch": 4.9752883031301485, + "grad_norm": 0.4750655640254115, + "learning_rate": 9.37128607339562e-06, + "loss": 0.1572, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14737387001514435, + "step": 3020, + "valid_targets_mean": 6327.5, + "valid_targets_min": 821 + }, + { + "epoch": 4.983525535420099, + "grad_norm": 0.4555400005042597, + "learning_rate": 9.301782874257186e-06, + "loss": 0.1688, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15319132804870605, + "step": 3025, + "valid_targets_mean": 5603.2, + "valid_targets_min": 419 + }, + { + "epoch": 4.99176276771005, + "grad_norm": 0.4695221755893214, + "learning_rate": 9.232460190737394e-06, + "loss": 0.1551, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1426851451396942, + "step": 3030, + "valid_targets_mean": 5278.2, + "valid_targets_min": 294 + }, + { + "epoch": 5.0, + "grad_norm": 0.478596890936015, + "learning_rate": 9.163319192547702e-06, + "loss": 0.1465, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.142055481672287, + "step": 3035, + "valid_targets_mean": 5582.5, + "valid_targets_min": 531 + }, + { + "epoch": 5.00823723228995, + "grad_norm": 0.49069009485190446, + "learning_rate": 9.094361046333887e-06, + "loss": 0.1409, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1303725689649582, + "step": 3040, + "valid_targets_mean": 6621.2, + "valid_targets_min": 911 + }, + { + "epoch": 5.016474464579901, + "grad_norm": 0.4821661219915955, + "learning_rate": 9.025586915656435e-06, + "loss": 0.1458, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1344459503889084, + "step": 3045, + "valid_targets_mean": 5295.7, + "valid_targets_min": 325 + }, + { + "epoch": 5.0247116968698515, + "grad_norm": 0.5505894006961336, + "learning_rate": 8.956997960970811e-06, + "loss": 0.1548, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17692875862121582, + "step": 3050, + "valid_targets_mean": 4894.8, + "valid_targets_min": 310 + }, + { + "epoch": 5.032948929159803, + "grad_norm": 0.46750433795789914, + "learning_rate": 8.888595339607961e-06, + "loss": 0.1373, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13959461450576782, + "step": 3055, + "valid_targets_mean": 6171.8, + "valid_targets_min": 587 + }, + { + "epoch": 5.041186161449753, + "grad_norm": 0.5070608295694433, + "learning_rate": 8.82038020575472e-06, + "loss": 0.1417, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14376729726791382, + "step": 3060, + "valid_targets_mean": 4882.1, + "valid_targets_min": 555 + }, + { + "epoch": 5.049423393739703, + "grad_norm": 0.6060377491564969, + "learning_rate": 8.752353710434384e-06, + "loss": 0.1419, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15076524019241333, + "step": 3065, + "valid_targets_mean": 3471.9, + "valid_targets_min": 270 + }, + { + "epoch": 5.057660626029654, + "grad_norm": 0.469440093985586, + "learning_rate": 8.684517001487276e-06, + "loss": 0.1425, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14725562930107117, + "step": 3070, + "valid_targets_mean": 5937.0, + "valid_targets_min": 331 + }, + { + "epoch": 5.065897858319604, + "grad_norm": 0.4900110272276438, + "learning_rate": 8.616871223551335e-06, + "loss": 0.1409, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13916555047035217, + "step": 3075, + "valid_targets_mean": 5039.7, + "valid_targets_min": 726 + }, + { + "epoch": 5.0741350906095555, + "grad_norm": 0.41725739160888586, + "learning_rate": 8.549417518042899e-06, + "loss": 0.1351, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1271580159664154, + "step": 3080, + "valid_targets_mean": 6236.5, + "valid_targets_min": 422 + }, + { + "epoch": 5.082372322899506, + "grad_norm": 0.4605522955017288, + "learning_rate": 8.482157023137326e-06, + "loss": 0.1385, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.11955632269382477, + "step": 3085, + "valid_targets_mean": 5906.8, + "valid_targets_min": 600 + }, + { + "epoch": 5.090609555189456, + "grad_norm": 0.502778885048082, + "learning_rate": 8.415090873749882e-06, + "loss": 0.1439, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14969505369663239, + "step": 3090, + "valid_targets_mean": 5029.8, + "valid_targets_min": 211 + }, + { + "epoch": 5.098846787479407, + "grad_norm": 0.5053536000161294, + "learning_rate": 8.348220201516559e-06, + "loss": 0.1469, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13605719804763794, + "step": 3095, + "valid_targets_mean": 6112.5, + "valid_targets_min": 303 + }, + { + "epoch": 5.107084019769357, + "grad_norm": 0.4598478330032498, + "learning_rate": 8.281546134774952e-06, + "loss": 0.1494, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15661340951919556, + "step": 3100, + "valid_targets_mean": 5920.4, + "valid_targets_min": 1189 + }, + { + "epoch": 5.115321252059308, + "grad_norm": 0.5547180056529626, + "learning_rate": 8.215069798545272e-06, + "loss": 0.1366, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13134953379631042, + "step": 3105, + "valid_targets_mean": 4233.9, + "valid_targets_min": 265 + }, + { + "epoch": 5.1235584843492585, + "grad_norm": 0.44622191419482493, + "learning_rate": 8.148792314511338e-06, + "loss": 0.1402, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14064301550388336, + "step": 3110, + "valid_targets_mean": 5863.4, + "valid_targets_min": 486 + }, + { + "epoch": 5.13179571663921, + "grad_norm": 0.47904865237329824, + "learning_rate": 8.082714801001619e-06, + "loss": 0.138, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13501349091529846, + "step": 3115, + "valid_targets_mean": 5212.8, + "valid_targets_min": 395 + }, + { + "epoch": 5.14003294892916, + "grad_norm": 0.46284849540082074, + "learning_rate": 8.016838372970453e-06, + "loss": 0.1392, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13676853477954865, + "step": 3120, + "valid_targets_mean": 6098.6, + "valid_targets_min": 1111 + }, + { + "epoch": 5.14827018121911, + "grad_norm": 0.4994068688425364, + "learning_rate": 7.951164141979115e-06, + "loss": 0.1306, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12372538447380066, + "step": 3125, + "valid_targets_mean": 6541.1, + "valid_targets_min": 1550 + }, + { + "epoch": 5.156507413509061, + "grad_norm": 0.5022833293912967, + "learning_rate": 7.885693216177166e-06, + "loss": 0.1437, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13442087173461914, + "step": 3130, + "valid_targets_mean": 5174.1, + "valid_targets_min": 318 + }, + { + "epoch": 5.164744645799011, + "grad_norm": 0.4775571872447023, + "learning_rate": 7.820426700283681e-06, + "loss": 0.1453, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13596433401107788, + "step": 3135, + "valid_targets_mean": 4972.1, + "valid_targets_min": 298 + }, + { + "epoch": 5.1729818780889625, + "grad_norm": 0.49050982979527447, + "learning_rate": 7.75536569556866e-06, + "loss": 0.1383, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1351958066225052, + "step": 3140, + "valid_targets_mean": 5699.8, + "valid_targets_min": 333 + }, + { + "epoch": 5.181219110378913, + "grad_norm": 0.5461539540093063, + "learning_rate": 7.690511299834422e-06, + "loss": 0.1374, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1379501223564148, + "step": 3145, + "valid_targets_mean": 5473.7, + "valid_targets_min": 298 + }, + { + "epoch": 5.189456342668863, + "grad_norm": 0.4501236120342665, + "learning_rate": 7.625864607397066e-06, + "loss": 0.138, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.146168053150177, + "step": 3150, + "valid_targets_mean": 6165.9, + "valid_targets_min": 385 + }, + { + "epoch": 5.197693574958814, + "grad_norm": 0.5244359282299395, + "learning_rate": 7.561426709068045e-06, + "loss": 0.1483, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1434154212474823, + "step": 3155, + "valid_targets_mean": 4548.9, + "valid_targets_min": 293 + }, + { + "epoch": 5.205930807248764, + "grad_norm": 0.46684517074949977, + "learning_rate": 7.497198692135728e-06, + "loss": 0.1381, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13306406140327454, + "step": 3160, + "valid_targets_mean": 5791.3, + "valid_targets_min": 2554 + }, + { + "epoch": 5.214168039538715, + "grad_norm": 0.5243868937758703, + "learning_rate": 7.433181640347069e-06, + "loss": 0.1427, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15170177817344666, + "step": 3165, + "valid_targets_mean": 6504.1, + "valid_targets_min": 1501 + }, + { + "epoch": 5.2224052718286655, + "grad_norm": 0.4643698500548393, + "learning_rate": 7.369376633889325e-06, + "loss": 0.1373, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.140183687210083, + "step": 3170, + "valid_targets_mean": 5628.1, + "valid_targets_min": 349 + }, + { + "epoch": 5.230642504118616, + "grad_norm": 0.4965467591482577, + "learning_rate": 7.305784749371792e-06, + "loss": 0.1417, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1593542993068695, + "step": 3175, + "valid_targets_mean": 5535.8, + "valid_targets_min": 753 + }, + { + "epoch": 5.238879736408567, + "grad_norm": 0.48661738008906197, + "learning_rate": 7.242407059807703e-06, + "loss": 0.1357, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14572195708751678, + "step": 3180, + "valid_targets_mean": 5237.9, + "valid_targets_min": 549 + }, + { + "epoch": 5.247116968698517, + "grad_norm": 0.4806448175923749, + "learning_rate": 7.1792446345960585e-06, + "loss": 0.151, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1412971317768097, + "step": 3185, + "valid_targets_mean": 5538.9, + "valid_targets_min": 423 + }, + { + "epoch": 5.255354200988468, + "grad_norm": 0.5155778651423555, + "learning_rate": 7.116298539503632e-06, + "loss": 0.141, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17066286504268646, + "step": 3190, + "valid_targets_mean": 5100.2, + "valid_targets_min": 872 + }, + { + "epoch": 5.263591433278418, + "grad_norm": 0.48430509588775345, + "learning_rate": 7.053569836646954e-06, + "loss": 0.1446, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1346225142478943, + "step": 3195, + "valid_targets_mean": 5719.1, + "valid_targets_min": 2796 + }, + { + "epoch": 5.2718286655683695, + "grad_norm": 0.46910515647052425, + "learning_rate": 6.991059584474411e-06, + "loss": 0.1389, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13558635115623474, + "step": 3200, + "valid_targets_mean": 5281.7, + "valid_targets_min": 312 + }, + { + "epoch": 5.28006589785832, + "grad_norm": 0.5480174850491485, + "learning_rate": 6.928768837748374e-06, + "loss": 0.1442, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1344684362411499, + "step": 3205, + "valid_targets_mean": 6858.4, + "valid_targets_min": 320 + }, + { + "epoch": 5.28830313014827, + "grad_norm": 0.4918052522785785, + "learning_rate": 6.866698647527392e-06, + "loss": 0.1369, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1560150682926178, + "step": 3210, + "valid_targets_mean": 5481.4, + "valid_targets_min": 3738 + }, + { + "epoch": 5.296540362438221, + "grad_norm": 0.4490381704275577, + "learning_rate": 6.804850061148485e-06, + "loss": 0.1393, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1290242075920105, + "step": 3215, + "valid_targets_mean": 5527.1, + "valid_targets_min": 339 + }, + { + "epoch": 5.304777594728171, + "grad_norm": 0.8224095339940158, + "learning_rate": 6.7432241222094595e-06, + "loss": 0.1473, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.17443184554576874, + "step": 3220, + "valid_targets_mean": 4949.4, + "valid_targets_min": 334 + }, + { + "epoch": 5.313014827018122, + "grad_norm": 0.47003033255636556, + "learning_rate": 6.68182187055127e-06, + "loss": 0.1416, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1260230839252472, + "step": 3225, + "valid_targets_mean": 5310.6, + "valid_targets_min": 375 + }, + { + "epoch": 5.3212520593080725, + "grad_norm": 0.4908187369808388, + "learning_rate": 6.620644342240534e-06, + "loss": 0.1393, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14359594881534576, + "step": 3230, + "valid_targets_mean": 5568.1, + "valid_targets_min": 1129 + }, + { + "epoch": 5.329489291598023, + "grad_norm": 0.551319463728614, + "learning_rate": 6.559692569551976e-06, + "loss": 0.1411, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15381722152233124, + "step": 3235, + "valid_targets_mean": 5568.9, + "valid_targets_min": 815 + }, + { + "epoch": 5.337726523887974, + "grad_norm": 0.4450876804926074, + "learning_rate": 6.498967580951099e-06, + "loss": 0.1335, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12999829649925232, + "step": 3240, + "valid_targets_mean": 5879.9, + "valid_targets_min": 361 + }, + { + "epoch": 5.345963756177924, + "grad_norm": 0.4711121467460195, + "learning_rate": 6.4384704010767284e-06, + "loss": 0.1508, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14210301637649536, + "step": 3245, + "valid_targets_mean": 5384.6, + "valid_targets_min": 281 + }, + { + "epoch": 5.354200988467875, + "grad_norm": 0.5603577329282675, + "learning_rate": 6.37820205072381e-06, + "loss": 0.1346, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13441210985183716, + "step": 3250, + "valid_targets_mean": 3903.1, + "valid_targets_min": 269 + }, + { + "epoch": 5.362438220757825, + "grad_norm": 0.581302886704017, + "learning_rate": 6.318163546826141e-06, + "loss": 0.1357, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1303476095199585, + "step": 3255, + "valid_targets_mean": 5103.0, + "valid_targets_min": 365 + }, + { + "epoch": 5.370675453047776, + "grad_norm": 0.5001476639868807, + "learning_rate": 6.258355902439211e-06, + "loss": 0.1411, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.149338960647583, + "step": 3260, + "valid_targets_mean": 5191.6, + "valid_targets_min": 364 + }, + { + "epoch": 5.378912685337727, + "grad_norm": 0.6039641844272668, + "learning_rate": 6.198780126723127e-06, + "loss": 0.1404, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15267781913280487, + "step": 3265, + "valid_targets_mean": 3882.5, + "valid_targets_min": 309 + }, + { + "epoch": 5.387149917627677, + "grad_norm": 0.4942844367813607, + "learning_rate": 6.139437224925586e-06, + "loss": 0.1414, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12675516307353973, + "step": 3270, + "valid_targets_mean": 5851.5, + "valid_targets_min": 1922 + }, + { + "epoch": 5.395387149917628, + "grad_norm": 0.5310086973536533, + "learning_rate": 6.080328198364871e-06, + "loss": 0.1478, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14578582346439362, + "step": 3275, + "valid_targets_mean": 4682.0, + "valid_targets_min": 274 + }, + { + "epoch": 5.403624382207578, + "grad_norm": 0.510770120262347, + "learning_rate": 6.021454044413036e-06, + "loss": 0.1324, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14449571073055267, + "step": 3280, + "valid_targets_mean": 4886.8, + "valid_targets_min": 357 + }, + { + "epoch": 5.4118616144975284, + "grad_norm": 0.6719528824690161, + "learning_rate": 5.962815756478979e-06, + "loss": 0.1408, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.146399587392807, + "step": 3285, + "valid_targets_mean": 4660.1, + "valid_targets_min": 282 + }, + { + "epoch": 5.4200988467874796, + "grad_norm": 0.5505355745551022, + "learning_rate": 5.904414323991758e-06, + "loss": 0.1388, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14041973650455475, + "step": 3290, + "valid_targets_mean": 5069.0, + "valid_targets_min": 372 + }, + { + "epoch": 5.42833607907743, + "grad_norm": 0.5162576529384314, + "learning_rate": 5.846250732383869e-06, + "loss": 0.1335, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13150526583194733, + "step": 3295, + "valid_targets_mean": 5460.1, + "valid_targets_min": 293 + }, + { + "epoch": 5.436573311367381, + "grad_norm": 0.6352297078680663, + "learning_rate": 5.788325963074594e-06, + "loss": 0.1504, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16142921149730682, + "step": 3300, + "valid_targets_mean": 4705.2, + "valid_targets_min": 291 + }, + { + "epoch": 5.444810543657331, + "grad_norm": 0.580160236040931, + "learning_rate": 5.7306409934534936e-06, + "loss": 0.143, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15034732222557068, + "step": 3305, + "valid_targets_mean": 4449.2, + "valid_targets_min": 360 + }, + { + "epoch": 5.453047775947281, + "grad_norm": 0.5256262308325671, + "learning_rate": 5.673196796863852e-06, + "loss": 0.1398, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14116880297660828, + "step": 3310, + "valid_targets_mean": 5438.1, + "valid_targets_min": 300 + }, + { + "epoch": 5.461285008237232, + "grad_norm": 0.4206557921596052, + "learning_rate": 5.615994342586313e-06, + "loss": 0.1317, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1337016373872757, + "step": 3315, + "valid_targets_mean": 6748.2, + "valid_targets_min": 2293 + }, + { + "epoch": 5.469522240527183, + "grad_norm": 0.5268814730787679, + "learning_rate": 5.5590345958224875e-06, + "loss": 0.1263, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12679679691791534, + "step": 3320, + "valid_targets_mean": 4569.9, + "valid_targets_min": 317 + }, + { + "epoch": 5.477759472817134, + "grad_norm": 0.5252752092181787, + "learning_rate": 5.502318517678682e-06, + "loss": 0.1446, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14012542366981506, + "step": 3325, + "valid_targets_mean": 4685.8, + "valid_targets_min": 307 + }, + { + "epoch": 5.485996705107084, + "grad_norm": 0.5623820612844119, + "learning_rate": 5.445847065149679e-06, + "loss": 0.1452, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15179842710494995, + "step": 3330, + "valid_targets_mean": 4619.6, + "valid_targets_min": 383 + }, + { + "epoch": 5.494233937397034, + "grad_norm": 0.45684783663012746, + "learning_rate": 5.389621191102574e-06, + "loss": 0.1418, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1381663978099823, + "step": 3335, + "valid_targets_mean": 5621.9, + "valid_targets_min": 2010 + }, + { + "epoch": 5.502471169686985, + "grad_norm": 0.4806800362003253, + "learning_rate": 5.333641844260726e-06, + "loss": 0.1429, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14656078815460205, + "step": 3340, + "valid_targets_mean": 5493.8, + "valid_targets_min": 780 + }, + { + "epoch": 5.5107084019769355, + "grad_norm": 0.4542630067117667, + "learning_rate": 5.277909969187738e-06, + "loss": 0.1337, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12890984117984772, + "step": 3345, + "valid_targets_mean": 4942.9, + "valid_targets_min": 1065 + }, + { + "epoch": 5.518945634266887, + "grad_norm": 0.48692358345346237, + "learning_rate": 5.222426506271492e-06, + "loss": 0.1341, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13926216959953308, + "step": 3350, + "valid_targets_mean": 5788.2, + "valid_targets_min": 255 + }, + { + "epoch": 5.527182866556837, + "grad_norm": 0.41013954032758926, + "learning_rate": 5.1671923917083265e-06, + "loss": 0.1423, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1503438502550125, + "step": 3355, + "valid_targets_mean": 6804.8, + "valid_targets_min": 5049 + }, + { + "epoch": 5.535420098846787, + "grad_norm": 0.45924517683091326, + "learning_rate": 5.112208557487215e-06, + "loss": 0.131, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1323554813861847, + "step": 3360, + "valid_targets_mean": 6357.6, + "valid_targets_min": 377 + }, + { + "epoch": 5.543657331136738, + "grad_norm": 0.527172213063129, + "learning_rate": 5.057475931374043e-06, + "loss": 0.1416, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15223285555839539, + "step": 3365, + "valid_targets_mean": 4769.6, + "valid_targets_min": 333 + }, + { + "epoch": 5.551894563426688, + "grad_norm": 0.5024208642551655, + "learning_rate": 5.002995436895939e-06, + "loss": 0.1445, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14082445204257965, + "step": 3370, + "valid_targets_mean": 6298.3, + "valid_targets_min": 2535 + }, + { + "epoch": 5.560131795716639, + "grad_norm": 0.4833609888224746, + "learning_rate": 4.9487679933257225e-06, + "loss": 0.1469, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15281173586845398, + "step": 3375, + "valid_targets_mean": 5999.8, + "valid_targets_min": 708 + }, + { + "epoch": 5.56836902800659, + "grad_norm": 0.45535217794864863, + "learning_rate": 4.894794515666379e-06, + "loss": 0.1425, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13284778594970703, + "step": 3380, + "valid_targets_mean": 5627.4, + "valid_targets_min": 615 + }, + { + "epoch": 5.576606260296541, + "grad_norm": 0.4170755788775851, + "learning_rate": 4.8410759146355955e-06, + "loss": 0.1414, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12243291735649109, + "step": 3385, + "valid_targets_mean": 6536.8, + "valid_targets_min": 402 + }, + { + "epoch": 5.584843492586491, + "grad_norm": 0.5003624466674033, + "learning_rate": 4.787613096650436e-06, + "loss": 0.1377, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1479077935218811, + "step": 3390, + "valid_targets_mean": 5083.2, + "valid_targets_min": 393 + }, + { + "epoch": 5.593080724876441, + "grad_norm": 0.44526841862204153, + "learning_rate": 4.734406963812024e-06, + "loss": 0.1461, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1423017829656601, + "step": 3395, + "valid_targets_mean": 6407.9, + "valid_targets_min": 4165 + }, + { + "epoch": 5.601317957166392, + "grad_norm": 0.42750551337562315, + "learning_rate": 4.68145841389032e-06, + "loss": 0.1408, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12783539295196533, + "step": 3400, + "valid_targets_mean": 5636.9, + "valid_targets_min": 369 + }, + { + "epoch": 5.6095551894563425, + "grad_norm": 0.4236352461981568, + "learning_rate": 4.6287683403089845e-06, + "loss": 0.1324, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1300227791070938, + "step": 3405, + "valid_targets_mean": 6075.1, + "valid_targets_min": 1246 + }, + { + "epoch": 5.617792421746294, + "grad_norm": 0.5163705704902521, + "learning_rate": 4.576337632130281e-06, + "loss": 0.1428, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14247193932533264, + "step": 3410, + "valid_targets_mean": 4827.1, + "valid_targets_min": 310 + }, + { + "epoch": 5.626029654036244, + "grad_norm": 0.43379156718976253, + "learning_rate": 4.524167174040106e-06, + "loss": 0.1377, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1339322179555893, + "step": 3415, + "valid_targets_mean": 6573.1, + "valid_targets_min": 424 + }, + { + "epoch": 5.634266886326195, + "grad_norm": 0.4707230248503362, + "learning_rate": 4.472257846333028e-06, + "loss": 0.1391, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1403801143169403, + "step": 3420, + "valid_targets_mean": 5142.9, + "valid_targets_min": 298 + }, + { + "epoch": 5.642504118616145, + "grad_norm": 0.5319126431444242, + "learning_rate": 4.4206105248974595e-06, + "loss": 0.139, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1340303272008896, + "step": 3425, + "valid_targets_mean": 6185.4, + "valid_targets_min": 381 + }, + { + "epoch": 5.650741350906095, + "grad_norm": 0.440524877940003, + "learning_rate": 4.3692260812008795e-06, + "loss": 0.1394, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12920840084552765, + "step": 3430, + "valid_targets_mean": 5609.5, + "valid_targets_min": 250 + }, + { + "epoch": 5.658978583196046, + "grad_norm": 0.4609698993867005, + "learning_rate": 4.318105382275084e-06, + "loss": 0.1358, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14189115166664124, + "step": 3435, + "valid_targets_mean": 5514.4, + "valid_targets_min": 305 + }, + { + "epoch": 5.667215815485997, + "grad_norm": 0.43144316493853585, + "learning_rate": 4.267249290701636e-06, + "loss": 0.1383, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12442096322774887, + "step": 3440, + "valid_targets_mean": 5884.9, + "valid_targets_min": 375 + }, + { + "epoch": 5.675453047775948, + "grad_norm": 0.4997124831898971, + "learning_rate": 4.216658664597224e-06, + "loss": 0.1364, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13440588116645813, + "step": 3445, + "valid_targets_mean": 5936.8, + "valid_targets_min": 425 + }, + { + "epoch": 5.683690280065898, + "grad_norm": 0.4542471741740084, + "learning_rate": 4.166334357599246e-06, + "loss": 0.1392, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12336796522140503, + "step": 3450, + "valid_targets_mean": 5466.2, + "valid_targets_min": 242 + }, + { + "epoch": 5.691927512355848, + "grad_norm": 0.7475283328556905, + "learning_rate": 4.116277218851383e-06, + "loss": 0.1343, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1225927472114563, + "step": 3455, + "valid_targets_mean": 5669.9, + "valid_targets_min": 465 + }, + { + "epoch": 5.700164744645799, + "grad_norm": 0.502370666703315, + "learning_rate": 4.066488092989256e-06, + "loss": 0.1652, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13243255019187927, + "step": 3460, + "valid_targets_mean": 4953.4, + "valid_targets_min": 395 + }, + { + "epoch": 5.7084019769357495, + "grad_norm": 0.5522762130532978, + "learning_rate": 4.016967820126205e-06, + "loss": 0.1507, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15482047200202942, + "step": 3465, + "valid_targets_mean": 4570.9, + "valid_targets_min": 298 + }, + { + "epoch": 5.716639209225701, + "grad_norm": 0.5267311706871121, + "learning_rate": 3.967717235839086e-06, + "loss": 0.1343, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13417063653469086, + "step": 3470, + "valid_targets_mean": 4840.9, + "valid_targets_min": 345 + }, + { + "epoch": 5.724876441515651, + "grad_norm": 0.5287502682061341, + "learning_rate": 3.918737171154192e-06, + "loss": 0.1335, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14333754777908325, + "step": 3475, + "valid_targets_mean": 4829.8, + "valid_targets_min": 286 + }, + { + "epoch": 5.733113673805601, + "grad_norm": 0.48784650075697344, + "learning_rate": 3.870028452533221e-06, + "loss": 0.1411, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14025351405143738, + "step": 3480, + "valid_targets_mean": 5756.6, + "valid_targets_min": 265 + }, + { + "epoch": 5.741350906095552, + "grad_norm": 0.5403889463433232, + "learning_rate": 3.8215919018593295e-06, + "loss": 0.1422, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.146037757396698, + "step": 3485, + "valid_targets_mean": 4279.8, + "valid_targets_min": 320 + }, + { + "epoch": 5.749588138385502, + "grad_norm": 0.4592507398840409, + "learning_rate": 3.7734283364232747e-06, + "loss": 0.1367, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14514651894569397, + "step": 3490, + "valid_targets_mean": 5427.7, + "valid_targets_min": 732 + }, + { + "epoch": 5.757825370675453, + "grad_norm": 0.4887221571373193, + "learning_rate": 3.725538568909599e-06, + "loss": 0.14, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14044776558876038, + "step": 3495, + "valid_targets_mean": 5423.7, + "valid_targets_min": 362 + }, + { + "epoch": 5.766062602965404, + "grad_norm": 0.5486820425373832, + "learning_rate": 3.6779234073829528e-06, + "loss": 0.1404, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13362336158752441, + "step": 3500, + "valid_targets_mean": 5640.1, + "valid_targets_min": 301 + }, + { + "epoch": 5.774299835255354, + "grad_norm": 0.44478317936989153, + "learning_rate": 3.63058365527444e-06, + "loss": 0.1323, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14148974418640137, + "step": 3505, + "valid_targets_mean": 5923.2, + "valid_targets_min": 369 + }, + { + "epoch": 5.782537067545305, + "grad_norm": 0.4929393880606377, + "learning_rate": 3.5835201113680463e-06, + "loss": 0.1402, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13513973355293274, + "step": 3510, + "valid_targets_mean": 4818.0, + "valid_targets_min": 282 + }, + { + "epoch": 5.790774299835255, + "grad_norm": 0.4174321213852269, + "learning_rate": 3.536733569787194e-06, + "loss": 0.1379, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14224773645401, + "step": 3515, + "valid_targets_mean": 7219.0, + "valid_targets_min": 3108 + }, + { + "epoch": 5.799011532125206, + "grad_norm": 0.48728034697704387, + "learning_rate": 3.4902248199813203e-06, + "loss": 0.1449, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.137048602104187, + "step": 3520, + "valid_targets_mean": 5947.2, + "valid_targets_min": 416 + }, + { + "epoch": 5.8072487644151565, + "grad_norm": 0.4262984193290179, + "learning_rate": 3.443994646712567e-06, + "loss": 0.145, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14041151106357574, + "step": 3525, + "valid_targets_mean": 7137.3, + "valid_targets_min": 997 + }, + { + "epoch": 5.815485996705107, + "grad_norm": 0.41903023686254465, + "learning_rate": 3.3980438300425324e-06, + "loss": 0.147, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13303492963314056, + "step": 3530, + "valid_targets_mean": 6324.1, + "valid_targets_min": 379 + }, + { + "epoch": 5.823723228995058, + "grad_norm": 0.48396886689123936, + "learning_rate": 3.3523731453190987e-06, + "loss": 0.1437, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14490509033203125, + "step": 3535, + "valid_targets_mean": 5659.3, + "valid_targets_min": 407 + }, + { + "epoch": 5.831960461285008, + "grad_norm": 0.5634002824321679, + "learning_rate": 3.306983363163381e-06, + "loss": 0.139, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.149214506149292, + "step": 3540, + "valid_targets_mean": 4438.4, + "valid_targets_min": 224 + }, + { + "epoch": 5.840197693574959, + "grad_norm": 0.43591445348583646, + "learning_rate": 3.2618752494566854e-06, + "loss": 0.1394, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13709425926208496, + "step": 3545, + "valid_targets_mean": 6600.5, + "valid_targets_min": 4529 + }, + { + "epoch": 5.848434925864909, + "grad_norm": 0.45792537128808686, + "learning_rate": 3.217049565327619e-06, + "loss": 0.1366, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15016064047813416, + "step": 3550, + "valid_targets_mean": 5791.0, + "valid_targets_min": 414 + }, + { + "epoch": 5.8566721581548595, + "grad_norm": 0.41811339718156515, + "learning_rate": 3.17250706713923e-06, + "loss": 0.137, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.11696778237819672, + "step": 3555, + "valid_targets_mean": 6163.2, + "valid_targets_min": 768 + }, + { + "epoch": 5.864909390444811, + "grad_norm": 0.5171929946947682, + "learning_rate": 3.128248506476246e-06, + "loss": 0.1328, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13196325302124023, + "step": 3560, + "valid_targets_mean": 5356.0, + "valid_targets_min": 474 + }, + { + "epoch": 5.873146622734761, + "grad_norm": 0.496618184466259, + "learning_rate": 3.0842746301324e-06, + "loss": 0.1412, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15133965015411377, + "step": 3565, + "valid_targets_mean": 5518.6, + "valid_targets_min": 1579 + }, + { + "epoch": 5.881383855024712, + "grad_norm": 0.47452647506786294, + "learning_rate": 3.040586180097811e-06, + "loss": 0.1399, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13840806484222412, + "step": 3570, + "valid_targets_mean": 5662.7, + "valid_targets_min": 471 + }, + { + "epoch": 5.889621087314662, + "grad_norm": 0.45494258722616504, + "learning_rate": 2.9971838935464915e-06, + "loss": 0.1393, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1390582174062729, + "step": 3575, + "valid_targets_mean": 6040.2, + "valid_targets_min": 501 + }, + { + "epoch": 5.897858319604612, + "grad_norm": 0.5316573344410656, + "learning_rate": 2.9540685028238925e-06, + "loss": 0.1369, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1437787562608719, + "step": 3580, + "valid_targets_mean": 4216.3, + "valid_targets_min": 303 + }, + { + "epoch": 5.9060955518945635, + "grad_norm": 0.5021377198403308, + "learning_rate": 2.9112407354345375e-06, + "loss": 0.1376, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1306840479373932, + "step": 3585, + "valid_targets_mean": 4726.6, + "valid_targets_min": 336 + }, + { + "epoch": 5.914332784184514, + "grad_norm": 0.5214586963224782, + "learning_rate": 2.8687013140297695e-06, + "loss": 0.1319, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1259566992521286, + "step": 3590, + "valid_targets_mean": 6115.4, + "valid_targets_min": 536 + }, + { + "epoch": 5.922570016474465, + "grad_norm": 0.5277563358185039, + "learning_rate": 2.826450956395532e-06, + "loss": 0.1401, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1559084802865982, + "step": 3595, + "valid_targets_mean": 4852.1, + "valid_targets_min": 348 + }, + { + "epoch": 5.930807248764415, + "grad_norm": 0.4598808489728509, + "learning_rate": 2.7844903754402963e-06, + "loss": 0.1487, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1452733725309372, + "step": 3600, + "valid_targets_mean": 5961.7, + "valid_targets_min": 303 + }, + { + "epoch": 5.939044481054366, + "grad_norm": 0.44420598869056455, + "learning_rate": 2.7428202791829763e-06, + "loss": 0.1403, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.11765453219413757, + "step": 3605, + "valid_targets_mean": 5541.5, + "valid_targets_min": 576 + }, + { + "epoch": 5.947281713344316, + "grad_norm": 0.5733465605447501, + "learning_rate": 2.7014413707410357e-06, + "loss": 0.145, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13703860342502594, + "step": 3610, + "valid_targets_mean": 5567.2, + "valid_targets_min": 639 + }, + { + "epoch": 5.9555189456342665, + "grad_norm": 0.4757594325785292, + "learning_rate": 2.6603543483185966e-06, + "loss": 0.1328, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1293085515499115, + "step": 3615, + "valid_targets_mean": 5337.7, + "valid_targets_min": 422 + }, + { + "epoch": 5.963756177924218, + "grad_norm": 0.5172925361057765, + "learning_rate": 2.6195599051946465e-06, + "loss": 0.1401, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13227033615112305, + "step": 3620, + "valid_targets_mean": 4333.9, + "valid_targets_min": 353 + }, + { + "epoch": 5.971993410214168, + "grad_norm": 0.4453302162097895, + "learning_rate": 2.579058729711379e-06, + "loss": 0.1346, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13876956701278687, + "step": 3625, + "valid_targets_mean": 5969.4, + "valid_targets_min": 290 + }, + { + "epoch": 5.980230642504119, + "grad_norm": 0.5283908360552565, + "learning_rate": 2.538851505262545e-06, + "loss": 0.1337, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13446350395679474, + "step": 3630, + "valid_targets_mean": 4775.6, + "valid_targets_min": 281 + }, + { + "epoch": 5.988467874794069, + "grad_norm": 0.47399508195474715, + "learning_rate": 2.498938910281927e-06, + "loss": 0.1392, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13291141390800476, + "step": 3635, + "valid_targets_mean": 5101.6, + "valid_targets_min": 298 + }, + { + "epoch": 5.996705107084019, + "grad_norm": 0.4700837808956898, + "learning_rate": 2.4593216182319203e-06, + "loss": 0.1315, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13764753937721252, + "step": 3640, + "valid_targets_mean": 5740.4, + "valid_targets_min": 820 + }, + { + "epoch": 6.0049423393739705, + "grad_norm": 0.4282286520739445, + "learning_rate": 2.42000029759212e-06, + "loss": 0.1384, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13631629943847656, + "step": 3645, + "valid_targets_mean": 6149.7, + "valid_targets_min": 582 + }, + { + "epoch": 6.013179571663921, + "grad_norm": 0.4663616514291284, + "learning_rate": 2.3809756118480887e-06, + "loss": 0.1323, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12601891160011292, + "step": 3650, + "valid_targets_mean": 6049.3, + "valid_targets_min": 364 + }, + { + "epoch": 6.021416803953872, + "grad_norm": 0.4517526044118583, + "learning_rate": 2.3422482194801257e-06, + "loss": 0.1303, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.11501075327396393, + "step": 3655, + "valid_targets_mean": 6240.2, + "valid_targets_min": 242 + }, + { + "epoch": 6.029654036243822, + "grad_norm": 0.45773566091644025, + "learning_rate": 2.30381877395218e-06, + "loss": 0.1443, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14801597595214844, + "step": 3660, + "valid_targets_mean": 6157.1, + "valid_targets_min": 442 + }, + { + "epoch": 6.037891268533772, + "grad_norm": 0.4657053658680516, + "learning_rate": 2.265687923700821e-06, + "loss": 0.1245, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.11897988617420197, + "step": 3665, + "valid_targets_mean": 5020.0, + "valid_targets_min": 279 + }, + { + "epoch": 6.046128500823723, + "grad_norm": 0.49670506821102967, + "learning_rate": 2.2278563121242722e-06, + "loss": 0.1307, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14285153150558472, + "step": 3670, + "valid_targets_mean": 6166.0, + "valid_targets_min": 469 + }, + { + "epoch": 6.0543657331136735, + "grad_norm": 0.544682632896979, + "learning_rate": 2.190324577571592e-06, + "loss": 0.1349, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14309510588645935, + "step": 3675, + "valid_targets_mean": 4997.1, + "valid_targets_min": 367 + }, + { + "epoch": 6.062602965403625, + "grad_norm": 0.46759468317431707, + "learning_rate": 2.153093353331881e-06, + "loss": 0.1325, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12411434948444366, + "step": 3680, + "valid_targets_mean": 5710.9, + "valid_targets_min": 300 + }, + { + "epoch": 6.070840197693575, + "grad_norm": 0.4818458379154973, + "learning_rate": 2.1161632676235967e-06, + "loss": 0.1388, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1334155797958374, + "step": 3685, + "valid_targets_mean": 6071.1, + "valid_targets_min": 3393 + }, + { + "epoch": 6.079077429983526, + "grad_norm": 0.4961800974677331, + "learning_rate": 2.0795349435839606e-06, + "loss": 0.1326, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14636382460594177, + "step": 3690, + "valid_targets_mean": 4735.6, + "valid_targets_min": 310 + }, + { + "epoch": 6.087314662273476, + "grad_norm": 0.47523702383058075, + "learning_rate": 2.0432089992584326e-06, + "loss": 0.13, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12490198016166687, + "step": 3695, + "valid_targets_mean": 6293.3, + "valid_targets_min": 621 + }, + { + "epoch": 6.095551894563426, + "grad_norm": 0.4473024162840745, + "learning_rate": 2.007186047590297e-06, + "loss": 0.1314, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13148343563079834, + "step": 3700, + "valid_targets_mean": 5969.4, + "valid_targets_min": 332 + }, + { + "epoch": 6.1037891268533775, + "grad_norm": 0.4660267208167183, + "learning_rate": 1.971466696410309e-06, + "loss": 0.1277, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.10753028839826584, + "step": 3705, + "valid_targets_mean": 5304.6, + "valid_targets_min": 1081 + }, + { + "epoch": 6.112026359143328, + "grad_norm": 0.5422612688689189, + "learning_rate": 1.936051548426443e-06, + "loss": 0.134, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15001899003982544, + "step": 3710, + "valid_targets_mean": 5986.7, + "valid_targets_min": 423 + }, + { + "epoch": 6.120263591433279, + "grad_norm": 0.5274266888800593, + "learning_rate": 1.9009412012137218e-06, + "loss": 0.1291, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13106083869934082, + "step": 3715, + "valid_targets_mean": 5678.3, + "valid_targets_min": 406 + }, + { + "epoch": 6.128500823723229, + "grad_norm": 0.4576755101043015, + "learning_rate": 1.8661362472041355e-06, + "loss": 0.1309, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13249346613883972, + "step": 3720, + "valid_targets_mean": 5423.2, + "valid_targets_min": 600 + }, + { + "epoch": 6.136738056013179, + "grad_norm": 0.5133862834647515, + "learning_rate": 1.8316372736766452e-06, + "loss": 0.1335, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13682904839515686, + "step": 3725, + "valid_targets_mean": 5098.6, + "valid_targets_min": 1764 + }, + { + "epoch": 6.14497528830313, + "grad_norm": 0.5069868484461164, + "learning_rate": 1.7974448627472618e-06, + "loss": 0.138, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1550980806350708, + "step": 3730, + "valid_targets_mean": 5112.9, + "valid_targets_min": 579 + }, + { + "epoch": 6.1532125205930805, + "grad_norm": 0.41399824699276144, + "learning_rate": 1.763559591359243e-06, + "loss": 0.133, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13211315870285034, + "step": 3735, + "valid_targets_mean": 7211.7, + "valid_targets_min": 336 + }, + { + "epoch": 6.161449752883032, + "grad_norm": 0.5310567988388422, + "learning_rate": 1.7299820312733518e-06, + "loss": 0.1394, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1566026508808136, + "step": 3740, + "valid_targets_mean": 4989.5, + "valid_targets_min": 359 + }, + { + "epoch": 6.169686985172982, + "grad_norm": 0.5124474888168294, + "learning_rate": 1.6967127490581892e-06, + "loss": 0.1379, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14107312262058258, + "step": 3745, + "valid_targets_mean": 5098.3, + "valid_targets_min": 664 + }, + { + "epoch": 6.177924217462932, + "grad_norm": 0.4382769185305687, + "learning_rate": 1.6637523060806749e-06, + "loss": 0.1306, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.11474081128835678, + "step": 3750, + "valid_targets_mean": 6166.8, + "valid_targets_min": 1918 + }, + { + "epoch": 6.186161449752883, + "grad_norm": 0.43472143369355076, + "learning_rate": 1.6311012584965257e-06, + "loss": 0.1399, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1271776556968689, + "step": 3755, + "valid_targets_mean": 6410.5, + "valid_targets_min": 303 + }, + { + "epoch": 6.194398682042833, + "grad_norm": 0.45645793927866285, + "learning_rate": 1.5987601572409195e-06, + "loss": 0.1275, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13115490972995758, + "step": 3760, + "valid_targets_mean": 6528.2, + "valid_targets_min": 460 + }, + { + "epoch": 6.2026359143327845, + "grad_norm": 0.5159376108603132, + "learning_rate": 1.566729548019168e-06, + "loss": 0.1378, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14629538357257843, + "step": 3765, + "valid_targets_mean": 4745.2, + "valid_targets_min": 315 + }, + { + "epoch": 6.210873146622735, + "grad_norm": 0.4815968005801268, + "learning_rate": 1.5350099712975097e-06, + "loss": 0.139, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12448601424694061, + "step": 3770, + "valid_targets_mean": 5488.1, + "valid_targets_min": 329 + }, + { + "epoch": 6.219110378912685, + "grad_norm": 0.4832899827342596, + "learning_rate": 1.50360196229401e-06, + "loss": 0.1378, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.16062882542610168, + "step": 3775, + "valid_targets_mean": 5798.6, + "valid_targets_min": 852 + }, + { + "epoch": 6.227347611202636, + "grad_norm": 0.4849698048363265, + "learning_rate": 1.4725060509695088e-06, + "loss": 0.1372, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13764359056949615, + "step": 3780, + "valid_targets_mean": 6223.4, + "valid_targets_min": 346 + }, + { + "epoch": 6.235584843492586, + "grad_norm": 0.5424460364962002, + "learning_rate": 1.4417227620186935e-06, + "loss": 0.1387, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14788150787353516, + "step": 3785, + "valid_targets_mean": 4785.4, + "valid_targets_min": 303 + }, + { + "epoch": 6.243822075782537, + "grad_norm": 0.46125398686289026, + "learning_rate": 1.4112526148612381e-06, + "loss": 0.1314, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1359138935804367, + "step": 3790, + "valid_targets_mean": 5888.5, + "valid_targets_min": 360 + }, + { + "epoch": 6.2520593080724876, + "grad_norm": 0.7772213081626521, + "learning_rate": 1.3810961236330345e-06, + "loss": 0.1396, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12322203814983368, + "step": 3795, + "valid_targets_mean": 4800.7, + "valid_targets_min": 281 + }, + { + "epoch": 6.260296540362438, + "grad_norm": 0.41414462497435917, + "learning_rate": 1.3512537971775407e-06, + "loss": 0.1358, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13013538718223572, + "step": 3800, + "valid_targets_mean": 7567.3, + "valid_targets_min": 1657 + }, + { + "epoch": 6.268533772652389, + "grad_norm": 0.4316675872766028, + "learning_rate": 1.321726139037156e-06, + "loss": 0.1324, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12924161553382874, + "step": 3805, + "valid_targets_mean": 7021.1, + "valid_targets_min": 704 + }, + { + "epoch": 6.276771004942339, + "grad_norm": 0.5125954945406457, + "learning_rate": 1.2925136474447575e-06, + "loss": 0.1359, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13661538064479828, + "step": 3810, + "valid_targets_mean": 4975.6, + "valid_targets_min": 1201 + }, + { + "epoch": 6.28500823723229, + "grad_norm": 0.49772458500840794, + "learning_rate": 1.2636168153152828e-06, + "loss": 0.1365, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13991546630859375, + "step": 3815, + "valid_targets_mean": 5599.8, + "valid_targets_min": 472 + }, + { + "epoch": 6.29324546952224, + "grad_norm": 0.49945615845513786, + "learning_rate": 1.2350361302374037e-06, + "loss": 0.1372, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12564127147197723, + "step": 3820, + "valid_targets_mean": 4814.4, + "valid_targets_min": 362 + }, + { + "epoch": 6.3014827018121915, + "grad_norm": 0.45827931098556673, + "learning_rate": 1.2067720744653167e-06, + "loss": 0.1315, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13158249855041504, + "step": 3825, + "valid_targets_mean": 5550.9, + "valid_targets_min": 333 + }, + { + "epoch": 6.309719934102142, + "grad_norm": 0.46714676648447434, + "learning_rate": 1.178825124910583e-06, + "loss": 0.1338, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1324673593044281, + "step": 3830, + "valid_targets_mean": 5685.1, + "valid_targets_min": 269 + }, + { + "epoch": 6.317957166392092, + "grad_norm": 0.4896652597587443, + "learning_rate": 1.1511957531341067e-06, + "loss": 0.1374, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.11876912415027618, + "step": 3835, + "valid_targets_mean": 5610.4, + "valid_targets_min": 1521 + }, + { + "epoch": 6.326194398682043, + "grad_norm": 0.5371200728560797, + "learning_rate": 1.123884425338162e-06, + "loss": 0.1304, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12302712351083755, + "step": 3840, + "valid_targets_mean": 4245.2, + "valid_targets_min": 341 + }, + { + "epoch": 6.334431630971993, + "grad_norm": 0.6041879366529833, + "learning_rate": 1.0968916023585252e-06, + "loss": 0.142, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1337546855211258, + "step": 3845, + "valid_targets_mean": 5333.9, + "valid_targets_min": 397 + }, + { + "epoch": 6.342668863261944, + "grad_norm": 0.7693854601953327, + "learning_rate": 1.0702177396567114e-06, + "loss": 0.1286, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13358263671398163, + "step": 3850, + "valid_targets_mean": 5477.5, + "valid_targets_min": 370 + }, + { + "epoch": 6.350906095551895, + "grad_norm": 0.5217168667394281, + "learning_rate": 1.0438632873122746e-06, + "loss": 0.1348, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13919466733932495, + "step": 3855, + "valid_targets_mean": 4802.3, + "valid_targets_min": 278 + }, + { + "epoch": 6.359143327841845, + "grad_norm": 0.49627335521623195, + "learning_rate": 1.0178286900152234e-06, + "loss": 0.1347, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1277734637260437, + "step": 3860, + "valid_targets_mean": 5456.2, + "valid_targets_min": 347 + }, + { + "epoch": 6.367380560131796, + "grad_norm": 0.4640966805166447, + "learning_rate": 9.921143870585204e-07, + "loss": 0.1339, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1288243532180786, + "step": 3865, + "valid_targets_mean": 7001.1, + "valid_targets_min": 3741 + }, + { + "epoch": 6.375617792421746, + "grad_norm": 0.4493431529213666, + "learning_rate": 9.667208123306504e-07, + "loss": 0.1308, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12215395271778107, + "step": 3870, + "valid_targets_mean": 6469.6, + "valid_targets_min": 409 + }, + { + "epoch": 6.383855024711697, + "grad_norm": 0.5223283250804539, + "learning_rate": 9.416483943083254e-07, + "loss": 0.1385, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14019393920898438, + "step": 3875, + "valid_targets_mean": 4799.6, + "valid_targets_min": 356 + }, + { + "epoch": 6.392092257001647, + "grad_norm": 0.5839320585301707, + "learning_rate": 9.168975560492365e-07, + "loss": 0.137, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1326146423816681, + "step": 3880, + "valid_targets_mean": 6029.2, + "valid_targets_min": 468 + }, + { + "epoch": 6.400329489291598, + "grad_norm": 0.45084115574340156, + "learning_rate": 8.924687151849287e-07, + "loss": 0.1437, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13785238564014435, + "step": 3885, + "valid_targets_mean": 5878.5, + "valid_targets_min": 2455 + }, + { + "epoch": 6.408566721581549, + "grad_norm": 0.5297128307959439, + "learning_rate": 8.683622839137306e-07, + "loss": 0.1355, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12668374180793762, + "step": 3890, + "valid_targets_mean": 5506.2, + "valid_targets_min": 2193 + }, + { + "epoch": 6.416803953871499, + "grad_norm": 0.48322783971651356, + "learning_rate": 8.445786689938295e-07, + "loss": 0.1346, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1286458671092987, + "step": 3895, + "valid_targets_mean": 5458.9, + "valid_targets_min": 301 + }, + { + "epoch": 6.42504118616145, + "grad_norm": 0.48395434493993517, + "learning_rate": 8.211182717363897e-07, + "loss": 0.1432, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15668228268623352, + "step": 3900, + "valid_targets_mean": 5767.8, + "valid_targets_min": 3561 + }, + { + "epoch": 6.4332784184514, + "grad_norm": 0.4433161669875194, + "learning_rate": 7.979814879987824e-07, + "loss": 0.1408, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.138276606798172, + "step": 3905, + "valid_targets_mean": 5992.8, + "valid_targets_min": 364 + }, + { + "epoch": 6.441515650741351, + "grad_norm": 0.45862585322452476, + "learning_rate": 7.751687081779091e-07, + "loss": 0.1322, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1268194019794464, + "step": 3910, + "valid_targets_mean": 5809.1, + "valid_targets_min": 304 + }, + { + "epoch": 6.449752883031302, + "grad_norm": 0.5178320041186394, + "learning_rate": 7.526803172036157e-07, + "loss": 0.1342, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1460873931646347, + "step": 3915, + "valid_targets_mean": 5288.8, + "valid_targets_min": 373 + }, + { + "epoch": 6.457990115321252, + "grad_norm": 0.5881511480130761, + "learning_rate": 7.305166945321973e-07, + "loss": 0.1284, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14629772305488586, + "step": 3920, + "valid_targets_mean": 5360.9, + "valid_targets_min": 281 + }, + { + "epoch": 6.466227347611203, + "grad_norm": 0.45789052921456197, + "learning_rate": 7.086782141399906e-07, + "loss": 0.1328, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13530060648918152, + "step": 3925, + "valid_targets_mean": 6599.0, + "valid_targets_min": 295 + }, + { + "epoch": 6.474464579901153, + "grad_norm": 0.5108194896271699, + "learning_rate": 6.871652445170651e-07, + "loss": 0.1324, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13670974969863892, + "step": 3930, + "valid_targets_mean": 5607.4, + "valid_targets_min": 284 + }, + { + "epoch": 6.482701812191104, + "grad_norm": 0.4986363661068449, + "learning_rate": 6.659781486610129e-07, + "loss": 0.1316, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14160466194152832, + "step": 3935, + "valid_targets_mean": 5862.1, + "valid_targets_min": 336 + }, + { + "epoch": 6.490939044481054, + "grad_norm": 0.486747793891971, + "learning_rate": 6.451172840708087e-07, + "loss": 0.1327, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1396481692790985, + "step": 3940, + "valid_targets_mean": 5292.6, + "valid_targets_min": 746 + }, + { + "epoch": 6.499176276771005, + "grad_norm": 0.47788821872124543, + "learning_rate": 6.245830027407973e-07, + "loss": 0.1271, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12838926911354065, + "step": 3945, + "valid_targets_mean": 5533.8, + "valid_targets_min": 360 + }, + { + "epoch": 6.507413509060956, + "grad_norm": 0.49062328683604933, + "learning_rate": 6.043756511547427e-07, + "loss": 0.1405, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13557396829128265, + "step": 3950, + "valid_targets_mean": 4935.4, + "valid_targets_min": 305 + }, + { + "epoch": 6.515650741350906, + "grad_norm": 0.6375889676833415, + "learning_rate": 5.84495570279966e-07, + "loss": 0.1413, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14044895768165588, + "step": 3955, + "valid_targets_mean": 3416.4, + "valid_targets_min": 342 + }, + { + "epoch": 6.523887973640857, + "grad_norm": 0.430404832431965, + "learning_rate": 5.649430955616365e-07, + "loss": 0.127, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.117364302277565, + "step": 3960, + "valid_targets_mean": 6041.8, + "valid_targets_min": 1574 + }, + { + "epoch": 6.532125205930807, + "grad_norm": 0.6302140342224846, + "learning_rate": 5.457185569170609e-07, + "loss": 0.1279, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14673450589179993, + "step": 3965, + "valid_targets_mean": 5401.9, + "valid_targets_min": 533 + }, + { + "epoch": 6.5403624382207575, + "grad_norm": 0.5003641376463149, + "learning_rate": 5.268222787301502e-07, + "loss": 0.1377, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1596682220697403, + "step": 3970, + "valid_targets_mean": 6112.6, + "valid_targets_min": 258 + }, + { + "epoch": 6.548599670510709, + "grad_norm": 0.5506918588497802, + "learning_rate": 5.082545798459348e-07, + "loss": 0.131, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12708808481693268, + "step": 3975, + "valid_targets_mean": 4950.1, + "valid_targets_min": 365 + }, + { + "epoch": 6.556836902800659, + "grad_norm": 0.5124195626339795, + "learning_rate": 4.900157735651779e-07, + "loss": 0.1357, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1341022253036499, + "step": 3980, + "valid_targets_mean": 4943.3, + "valid_targets_min": 388 + }, + { + "epoch": 6.56507413509061, + "grad_norm": 0.5172186903087186, + "learning_rate": 4.721061676391081e-07, + "loss": 0.1292, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14156344532966614, + "step": 3985, + "valid_targets_mean": 5344.0, + "valid_targets_min": 1391 + }, + { + "epoch": 6.57331136738056, + "grad_norm": 0.4351269973360783, + "learning_rate": 4.5452606426420466e-07, + "loss": 0.1247, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13022242486476898, + "step": 3990, + "valid_targets_mean": 5943.5, + "valid_targets_min": 1326 + }, + { + "epoch": 6.58154859967051, + "grad_norm": 0.5132235188431938, + "learning_rate": 4.372757600771116e-07, + "loss": 0.1364, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12379980832338333, + "step": 3995, + "valid_targets_mean": 4823.9, + "valid_targets_min": 292 + }, + { + "epoch": 6.589785831960461, + "grad_norm": 0.5489329155887852, + "learning_rate": 4.203555461496378e-07, + "loss": 0.1562, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14551113545894623, + "step": 4000, + "valid_targets_mean": 4537.8, + "valid_targets_min": 440 + }, + { + "epoch": 6.598023064250412, + "grad_norm": 0.47238912777246345, + "learning_rate": 4.0376570798382084e-07, + "loss": 0.1356, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.11419009417295456, + "step": 4005, + "valid_targets_mean": 5518.3, + "valid_targets_min": 1469 + }, + { + "epoch": 6.606260296540363, + "grad_norm": 0.5439108468408408, + "learning_rate": 3.875065255071442e-07, + "loss": 0.1319, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13088250160217285, + "step": 4010, + "valid_targets_mean": 5277.8, + "valid_targets_min": 777 + }, + { + "epoch": 6.614497528830313, + "grad_norm": 0.44567096265885786, + "learning_rate": 3.715782730677786e-07, + "loss": 0.1358, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12983649969100952, + "step": 4015, + "valid_targets_mean": 5881.4, + "valid_targets_min": 366 + }, + { + "epoch": 6.622734761120263, + "grad_norm": 0.4941823682417998, + "learning_rate": 3.559812194299794e-07, + "loss": 0.1389, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13084876537322998, + "step": 4020, + "valid_targets_mean": 5684.1, + "valid_targets_min": 2730 + }, + { + "epoch": 6.630971993410214, + "grad_norm": 0.4513343311045982, + "learning_rate": 3.407156277695456e-07, + "loss": 0.1287, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.136585995554924, + "step": 4025, + "valid_targets_mean": 6158.8, + "valid_targets_min": 1771 + }, + { + "epoch": 6.6392092257001645, + "grad_norm": 0.4841668577651671, + "learning_rate": 3.257817556693654e-07, + "loss": 0.1365, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13883385062217712, + "step": 4030, + "valid_targets_mean": 5912.1, + "valid_targets_min": 432 + }, + { + "epoch": 6.647446457990116, + "grad_norm": 0.5305211565307272, + "learning_rate": 3.1117985511508686e-07, + "loss": 0.1397, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13103139400482178, + "step": 4035, + "valid_targets_mean": 5152.6, + "valid_targets_min": 315 + }, + { + "epoch": 6.655683690280066, + "grad_norm": 0.4580349337969464, + "learning_rate": 2.969101724908585e-07, + "loss": 0.1335, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13851921260356903, + "step": 4040, + "valid_targets_mean": 6540.2, + "valid_targets_min": 2527 + }, + { + "epoch": 6.663920922570016, + "grad_norm": 0.5059365029484628, + "learning_rate": 2.829729485751731e-07, + "loss": 0.1373, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13825827836990356, + "step": 4045, + "valid_targets_mean": 5032.9, + "valid_targets_min": 414 + }, + { + "epoch": 6.672158154859967, + "grad_norm": 0.5024959144928178, + "learning_rate": 2.6936841853680397e-07, + "loss": 0.1331, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13369962573051453, + "step": 4050, + "valid_targets_mean": 5008.1, + "valid_targets_min": 327 + }, + { + "epoch": 6.680395387149917, + "grad_norm": 0.6871281294342154, + "learning_rate": 2.560968119308349e-07, + "loss": 0.1411, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14997810125350952, + "step": 4055, + "valid_targets_mean": 5776.6, + "valid_targets_min": 353 + }, + { + "epoch": 6.688632619439868, + "grad_norm": 0.4735902654948438, + "learning_rate": 2.4315835269479446e-07, + "loss": 0.1306, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12766751646995544, + "step": 4060, + "valid_targets_mean": 5684.6, + "valid_targets_min": 294 + }, + { + "epoch": 6.696869851729819, + "grad_norm": 1.2645434899601486, + "learning_rate": 2.305532591448678e-07, + "loss": 0.1405, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1371457278728485, + "step": 4065, + "valid_targets_mean": 6512.0, + "valid_targets_min": 345 + }, + { + "epoch": 6.70510708401977, + "grad_norm": 0.5114470671638929, + "learning_rate": 2.1828174397222179e-07, + "loss": 0.1367, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.15383315086364746, + "step": 4070, + "valid_targets_mean": 5463.8, + "valid_targets_min": 549 + }, + { + "epoch": 6.71334431630972, + "grad_norm": 0.5588565234687837, + "learning_rate": 2.0634401423940797e-07, + "loss": 0.139, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12472236156463623, + "step": 4075, + "valid_targets_mean": 6355.6, + "valid_targets_min": 291 + }, + { + "epoch": 6.72158154859967, + "grad_norm": 0.47512870482477276, + "learning_rate": 1.9474027137687425e-07, + "loss": 0.124, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1210930347442627, + "step": 4080, + "valid_targets_mean": 5650.8, + "valid_targets_min": 777 + }, + { + "epoch": 6.729818780889621, + "grad_norm": 0.5516275865205322, + "learning_rate": 1.834707111795675e-07, + "loss": 0.1305, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12554940581321716, + "step": 4085, + "valid_targets_mean": 6649.1, + "valid_targets_min": 343 + }, + { + "epoch": 6.7380560131795715, + "grad_norm": 0.48757816526391146, + "learning_rate": 1.725355238036208e-07, + "loss": 0.131, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13415946066379547, + "step": 4090, + "valid_targets_mean": 5752.8, + "valid_targets_min": 830 + }, + { + "epoch": 6.746293245469523, + "grad_norm": 0.4747972140397399, + "learning_rate": 1.6193489376315375e-07, + "loss": 0.1395, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14257818460464478, + "step": 4095, + "valid_targets_mean": 5895.2, + "valid_targets_min": 1811 + }, + { + "epoch": 6.754530477759473, + "grad_norm": 0.45751329514365424, + "learning_rate": 1.5166899992715922e-07, + "loss": 0.129, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13291612267494202, + "step": 4100, + "valid_targets_mean": 6247.6, + "valid_targets_min": 323 + }, + { + "epoch": 6.762767710049423, + "grad_norm": 0.48075546248084383, + "learning_rate": 1.417380155164727e-07, + "loss": 0.1358, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1332646906375885, + "step": 4105, + "valid_targets_mean": 5221.9, + "valid_targets_min": 738 + }, + { + "epoch": 6.771004942339374, + "grad_norm": 0.5601515010199739, + "learning_rate": 1.3214210810087224e-07, + "loss": 0.1434, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13819454610347748, + "step": 4110, + "valid_targets_mean": 4406.5, + "valid_targets_min": 419 + }, + { + "epoch": 6.779242174629324, + "grad_norm": 0.6961080541032358, + "learning_rate": 1.2288143959622523e-07, + "loss": 0.1461, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13622626662254333, + "step": 4115, + "valid_targets_mean": 3236.4, + "valid_targets_min": 346 + }, + { + "epoch": 6.787479406919275, + "grad_norm": 0.45949694176316913, + "learning_rate": 1.1395616626177718e-07, + "loss": 0.137, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1468203365802765, + "step": 4120, + "valid_targets_mean": 6123.7, + "valid_targets_min": 744 + }, + { + "epoch": 6.795716639209226, + "grad_norm": 0.4805711737263706, + "learning_rate": 1.0536643869750508e-07, + "loss": 0.1334, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14250558614730835, + "step": 4125, + "valid_targets_mean": 5812.6, + "valid_targets_min": 293 + }, + { + "epoch": 6.803953871499177, + "grad_norm": 0.46887014456462534, + "learning_rate": 9.711240184157255e-08, + "loss": 0.1302, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13092397153377533, + "step": 4130, + "valid_targets_mean": 5349.0, + "valid_targets_min": 250 + }, + { + "epoch": 6.812191103789127, + "grad_norm": 0.5313920324876659, + "learning_rate": 8.919419496789428e-08, + "loss": 0.1332, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1342686414718628, + "step": 4135, + "valid_targets_mean": 4346.2, + "valid_targets_min": 360 + }, + { + "epoch": 6.820428336079077, + "grad_norm": 0.4859272674478183, + "learning_rate": 8.161195168378211e-08, + "loss": 0.1323, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.11980287730693817, + "step": 4140, + "valid_targets_mean": 5261.4, + "valid_targets_min": 352 + }, + { + "epoch": 6.828665568369028, + "grad_norm": 0.4423077861344137, + "learning_rate": 7.4365799927687e-08, + "loss": 0.1452, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13322511315345764, + "step": 4145, + "valid_targets_mean": 6199.2, + "valid_targets_min": 392 + }, + { + "epoch": 6.8369028006589785, + "grad_norm": 0.44086143214512336, + "learning_rate": 6.745586196704512e-08, + "loss": 0.1336, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12674112617969513, + "step": 4150, + "valid_targets_mean": 6344.1, + "valid_targets_min": 2470 + }, + { + "epoch": 6.84514003294893, + "grad_norm": 0.4598893989182098, + "learning_rate": 6.088225439620621e-08, + "loss": 0.1391, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12376689910888672, + "step": 4155, + "valid_targets_mean": 5618.9, + "valid_targets_min": 375 + }, + { + "epoch": 6.85337726523888, + "grad_norm": 0.4387084582808184, + "learning_rate": 5.464508813448621e-08, + "loss": 0.1258, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13196054100990295, + "step": 4160, + "valid_targets_mean": 6483.1, + "valid_targets_min": 1453 + }, + { + "epoch": 6.86161449752883, + "grad_norm": 0.5826307617034738, + "learning_rate": 4.874446842426883e-08, + "loss": 0.1393, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13274163007736206, + "step": 4165, + "valid_targets_mean": 3858.2, + "valid_targets_min": 265 + }, + { + "epoch": 6.869851729818781, + "grad_norm": 0.4902863000025557, + "learning_rate": 4.318049482924913e-08, + "loss": 0.1372, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14127911627292633, + "step": 4170, + "valid_targets_mean": 5120.2, + "valid_targets_min": 365 + }, + { + "epoch": 6.878088962108731, + "grad_norm": 0.4716241280190188, + "learning_rate": 3.7953261232743785e-08, + "loss": 0.1354, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13096028566360474, + "step": 4175, + "valid_targets_mean": 5707.4, + "valid_targets_min": 2026 + }, + { + "epoch": 6.886326194398682, + "grad_norm": 0.48274750236462316, + "learning_rate": 3.3062855836119014e-08, + "loss": 0.1297, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12582214176654816, + "step": 4180, + "valid_targets_mean": 6500.5, + "valid_targets_min": 304 + }, + { + "epoch": 6.894563426688633, + "grad_norm": 0.520762828050495, + "learning_rate": 2.8509361157285088e-08, + "loss": 0.1249, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12486450374126434, + "step": 4185, + "valid_targets_mean": 5814.1, + "valid_targets_min": 315 + }, + { + "epoch": 6.902800658978583, + "grad_norm": 0.5051747976357015, + "learning_rate": 2.429285402931525e-08, + "loss": 0.138, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13390396535396576, + "step": 4190, + "valid_targets_mean": 4789.9, + "valid_targets_min": 299 + }, + { + "epoch": 6.911037891268534, + "grad_norm": 0.46815700178456404, + "learning_rate": 2.041340559914895e-08, + "loss": 0.1377, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1404329240322113, + "step": 4195, + "valid_targets_mean": 6338.1, + "valid_targets_min": 429 + }, + { + "epoch": 6.919275123558484, + "grad_norm": 0.4512663494555607, + "learning_rate": 1.6871081326388372e-08, + "loss": 0.1353, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12632620334625244, + "step": 4200, + "valid_targets_mean": 5904.5, + "valid_targets_min": 270 + }, + { + "epoch": 6.927512355848435, + "grad_norm": 0.5289087445432837, + "learning_rate": 1.366594098219487e-08, + "loss": 0.1367, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1431260108947754, + "step": 4205, + "valid_targets_mean": 4917.8, + "valid_targets_min": 402 + }, + { + "epoch": 6.9357495881383855, + "grad_norm": 0.576493514416454, + "learning_rate": 1.0798038648278664e-08, + "loss": 0.1303, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1551305204629898, + "step": 4210, + "valid_targets_mean": 4842.8, + "valid_targets_min": 307 + }, + { + "epoch": 6.943986820428336, + "grad_norm": 0.48286976995059333, + "learning_rate": 8.26742271598846e-09, + "loss": 0.1321, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1334967464208603, + "step": 4215, + "valid_targets_mean": 5332.4, + "valid_targets_min": 266 + }, + { + "epoch": 6.952224052718287, + "grad_norm": 0.4832915302519884, + "learning_rate": 6.074135885498767e-09, + "loss": 0.1345, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13673850893974304, + "step": 4220, + "valid_targets_mean": 5995.8, + "valid_targets_min": 1239 + }, + { + "epoch": 6.960461285008237, + "grad_norm": 0.4966652005135941, + "learning_rate": 4.218215165077144e-09, + "loss": 0.1357, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13495373725891113, + "step": 4225, + "valid_targets_mean": 5102.9, + "valid_targets_min": 350 + }, + { + "epoch": 6.968698517298188, + "grad_norm": 0.437645161494385, + "learning_rate": 2.699691870473586e-09, + "loss": 0.131, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.12133049219846725, + "step": 4230, + "valid_targets_mean": 6069.5, + "valid_targets_min": 988 + }, + { + "epoch": 6.976935749588138, + "grad_norm": 0.43585901256643306, + "learning_rate": 1.5185916243876108e-09, + "loss": 0.1461, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14331373572349548, + "step": 4235, + "valid_targets_mean": 6261.4, + "valid_targets_min": 1324 + }, + { + "epoch": 6.9851729818780885, + "grad_norm": 0.49008244310573496, + "learning_rate": 6.749343560286114e-10, + "loss": 0.1387, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.14540137350559235, + "step": 4240, + "valid_targets_mean": 5976.2, + "valid_targets_min": 1274 + }, + { + "epoch": 6.99341021416804, + "grad_norm": 0.5246405834495542, + "learning_rate": 1.6873430078945264e-10, + "loss": 0.1408, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.13585354387760162, + "step": 4245, + "valid_targets_mean": 5064.4, + "valid_targets_min": 832 + }, + { + "epoch": 7.0, + "loss_nan_ranks": 0, + "loss_rank_avg": 0.1299251914024353, + "step": 4249, + "total_flos": 1654874798292992.0, + "train_loss": 0.18129448925475283, + "train_runtime": 26249.1152, + "train_samples_per_second": 2.586, + "train_steps_per_second": 0.162, + "valid_targets_mean": 5076.1, + "valid_targets_min": 783 + } + ], + "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": 1654874798292992.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +}