diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10299 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 7.0,
+  "eval_steps": 500,
+  "global_step": 4662,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0075075075075075074,
+      "grad_norm": 11.018023139973177,
+      "learning_rate": 3.426124197002142e-07,
+      "loss": 0.6962,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.666359007358551,
+      "step": 5,
+      "valid_targets_mean": 3406.2,
+      "valid_targets_min": 590
+    },
+    {
+      "epoch": 0.015015015015015015,
+      "grad_norm": 14.300498234778246,
+      "learning_rate": 7.708779443254819e-07,
+      "loss": 0.6812,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.7260710597038269,
+      "step": 10,
+      "valid_targets_mean": 2929.3,
+      "valid_targets_min": 540
+    },
+    {
+      "epoch": 0.02252252252252252,
+      "grad_norm": 10.308334513580103,
+      "learning_rate": 1.1991434689507496e-06,
+      "loss": 0.694,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.7622815370559692,
+      "step": 15,
+      "valid_targets_mean": 4942.4,
+      "valid_targets_min": 474
+    },
+    {
+      "epoch": 0.03003003003003003,
+      "grad_norm": 9.480467405179247,
+      "learning_rate": 1.6274089935760173e-06,
+      "loss": 0.63,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.5915040969848633,
+      "step": 20,
+      "valid_targets_mean": 4988.7,
+      "valid_targets_min": 506
+    },
+    {
+      "epoch": 0.03753753753753754,
+      "grad_norm": 6.98322327787682,
+      "learning_rate": 2.055674518201285e-06,
+      "loss": 0.634,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.6035254001617432,
+      "step": 25,
+      "valid_targets_mean": 4144.8,
+      "valid_targets_min": 406
+    },
+    {
+      "epoch": 0.04504504504504504,
+      "grad_norm": 4.519545202299929,
+      "learning_rate": 2.4839400428265524e-06,
+      "loss": 0.6108,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.6693063974380493,
+      "step": 30,
+      "valid_targets_mean": 2645.9,
+      "valid_targets_min": 811
+    },
+    {
+      "epoch": 0.052552552552552555,
+      "grad_norm": 2.254075439509915,
+      "learning_rate": 2.9122055674518203e-06,
+      "loss": 0.583,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.6287423968315125,
+      "step": 35,
+      "valid_targets_mean": 5337.9,
+      "valid_targets_min": 424
+    },
+    {
+      "epoch": 0.06006006006006006,
+      "grad_norm": 1.5869006029617074,
+      "learning_rate": 3.3404710920770882e-06,
+      "loss": 0.4981,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.5136668682098389,
+      "step": 40,
+      "valid_targets_mean": 4675.1,
+      "valid_targets_min": 683
+    },
+    {
+      "epoch": 0.06756756756756757,
+      "grad_norm": 1.2871585570908246,
+      "learning_rate": 3.7687366167023558e-06,
+      "loss": 0.498,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.4761298894882202,
+      "step": 45,
+      "valid_targets_mean": 4580.2,
+      "valid_targets_min": 695
+    },
+    {
+      "epoch": 0.07507507507507508,
+      "grad_norm": 1.1073292407135484,
+      "learning_rate": 4.197002141327624e-06,
+      "loss": 0.4618,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.4800295829772949,
+      "step": 50,
+      "valid_targets_mean": 3829.7,
+      "valid_targets_min": 408
+    },
+    {
+      "epoch": 0.08258258258258258,
+      "grad_norm": 0.8919504665868666,
+      "learning_rate": 4.625267665952891e-06,
+      "loss": 0.5237,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.525561511516571,
+      "step": 55,
+      "valid_targets_mean": 5515.5,
+      "valid_targets_min": 571
+    },
+    {
+      "epoch": 0.09009009009009009,
+      "grad_norm": 0.865780829661676,
+      "learning_rate": 5.05353319057816e-06,
+      "loss": 0.5163,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.5415364503860474,
+      "step": 60,
+      "valid_targets_mean": 3433.2,
+      "valid_targets_min": 511
+    },
+    {
+      "epoch": 0.09759759759759759,
+      "grad_norm": 0.5872913252913201,
+      "learning_rate": 5.481798715203427e-06,
+      "loss": 0.4707,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3888802230358124,
+      "step": 65,
+      "valid_targets_mean": 5232.6,
+      "valid_targets_min": 664
+    },
+    {
+      "epoch": 0.10510510510510511,
+      "grad_norm": 0.5994653626627001,
+      "learning_rate": 5.910064239828695e-06,
+      "loss": 0.4821,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.43450668454170227,
+      "step": 70,
+      "valid_targets_mean": 5187.1,
+      "valid_targets_min": 460
+    },
+    {
+      "epoch": 0.11261261261261261,
+      "grad_norm": 0.6662883071321841,
+      "learning_rate": 6.338329764453962e-06,
+      "loss": 0.4331,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3964645564556122,
+      "step": 75,
+      "valid_targets_mean": 3182.0,
+      "valid_targets_min": 880
+    },
+    {
+      "epoch": 0.12012012012012012,
+      "grad_norm": 0.6301518751447736,
+      "learning_rate": 6.76659528907923e-06,
+      "loss": 0.4374,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.37995022535324097,
+      "step": 80,
+      "valid_targets_mean": 3907.1,
+      "valid_targets_min": 773
+    },
+    {
+      "epoch": 0.12762762762762764,
+      "grad_norm": 0.6479909629866574,
+      "learning_rate": 7.194860813704497e-06,
+      "loss": 0.4749,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.43532025814056396,
+      "step": 85,
+      "valid_targets_mean": 3381.8,
+      "valid_targets_min": 727
+    },
+    {
+      "epoch": 0.13513513513513514,
+      "grad_norm": 0.601954910384077,
+      "learning_rate": 7.623126338329765e-06,
+      "loss": 0.449,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.37232670187950134,
+      "step": 90,
+      "valid_targets_mean": 3859.4,
+      "valid_targets_min": 681
+    },
+    {
+      "epoch": 0.14264264264264265,
+      "grad_norm": 0.6010699211352588,
+      "learning_rate": 8.051391862955033e-06,
+      "loss": 0.4036,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3912065625190735,
+      "step": 95,
+      "valid_targets_mean": 4532.2,
+      "valid_targets_min": 525
+    },
+    {
+      "epoch": 0.15015015015015015,
+      "grad_norm": 0.6776934870921543,
+      "learning_rate": 8.4796573875803e-06,
+      "loss": 0.4009,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3918367922306061,
+      "step": 100,
+      "valid_targets_mean": 3923.7,
+      "valid_targets_min": 631
+    },
+    {
+      "epoch": 0.15765765765765766,
+      "grad_norm": 0.5815397148408697,
+      "learning_rate": 8.907922912205568e-06,
+      "loss": 0.422,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.40705981850624084,
+      "step": 105,
+      "valid_targets_mean": 4907.4,
+      "valid_targets_min": 624
+    },
+    {
+      "epoch": 0.16516516516516516,
+      "grad_norm": 0.6207231229898003,
+      "learning_rate": 9.336188436830836e-06,
+      "loss": 0.4197,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.35407790541648865,
+      "step": 110,
+      "valid_targets_mean": 4749.1,
+      "valid_targets_min": 357
+    },
+    {
+      "epoch": 0.17267267267267267,
+      "grad_norm": 0.7161182478604338,
+      "learning_rate": 9.764453961456105e-06,
+      "loss": 0.4271,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3748543858528137,
+      "step": 115,
+      "valid_targets_mean": 2599.0,
+      "valid_targets_min": 495
+    },
+    {
+      "epoch": 0.18018018018018017,
+      "grad_norm": 0.6433782885571062,
+      "learning_rate": 1.019271948608137e-05,
+      "loss": 0.4129,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.44952958822250366,
+      "step": 120,
+      "valid_targets_mean": 3456.8,
+      "valid_targets_min": 521
+    },
+    {
+      "epoch": 0.18768768768768768,
+      "grad_norm": 0.608760987576289,
+      "learning_rate": 1.062098501070664e-05,
+      "loss": 0.396,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.4421440064907074,
+      "step": 125,
+      "valid_targets_mean": 4304.8,
+      "valid_targets_min": 676
+    },
+    {
+      "epoch": 0.19519519519519518,
+      "grad_norm": 0.6056247540086898,
+      "learning_rate": 1.1049250535331907e-05,
+      "loss": 0.356,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3524633049964905,
+      "step": 130,
+      "valid_targets_mean": 4378.4,
+      "valid_targets_min": 699
+    },
+    {
+      "epoch": 0.20270270270270271,
+      "grad_norm": 0.7470542876873747,
+      "learning_rate": 1.1477516059957175e-05,
+      "loss": 0.3928,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3509212136268616,
+      "step": 135,
+      "valid_targets_mean": 3348.1,
+      "valid_targets_min": 580
+    },
+    {
+      "epoch": 0.21021021021021022,
+      "grad_norm": 0.5297517332166896,
+      "learning_rate": 1.1905781584582443e-05,
+      "loss": 0.3971,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3493044376373291,
+      "step": 140,
+      "valid_targets_mean": 4820.6,
+      "valid_targets_min": 588
+    },
+    {
+      "epoch": 0.21771771771771772,
+      "grad_norm": 0.8236972811131019,
+      "learning_rate": 1.233404710920771e-05,
+      "loss": 0.4229,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.4382990002632141,
+      "step": 145,
+      "valid_targets_mean": 4910.1,
+      "valid_targets_min": 704
+    },
+    {
+      "epoch": 0.22522522522522523,
+      "grad_norm": 0.5641188867988419,
+      "learning_rate": 1.2762312633832978e-05,
+      "loss": 0.4053,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.4189315438270569,
+      "step": 150,
+      "valid_targets_mean": 4847.2,
+      "valid_targets_min": 479
+    },
+    {
+      "epoch": 0.23273273273273273,
+      "grad_norm": 0.759436996305121,
+      "learning_rate": 1.3190578158458245e-05,
+      "loss": 0.3995,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.4512358009815216,
+      "step": 155,
+      "valid_targets_mean": 3609.1,
+      "valid_targets_min": 660
+    },
+    {
+      "epoch": 0.24024024024024024,
+      "grad_norm": 0.5776616654260357,
+      "learning_rate": 1.3618843683083513e-05,
+      "loss": 0.3982,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.366862416267395,
+      "step": 160,
+      "valid_targets_mean": 4489.5,
+      "valid_targets_min": 709
+    },
+    {
+      "epoch": 0.24774774774774774,
+      "grad_norm": 0.5621030834623711,
+      "learning_rate": 1.404710920770878e-05,
+      "loss": 0.3562,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3245847225189209,
+      "step": 165,
+      "valid_targets_mean": 3611.0,
+      "valid_targets_min": 857
+    },
+    {
+      "epoch": 0.2552552552552553,
+      "grad_norm": 0.6383475705856398,
+      "learning_rate": 1.4475374732334048e-05,
+      "loss": 0.3633,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3641152083873749,
+      "step": 170,
+      "valid_targets_mean": 3776.7,
+      "valid_targets_min": 273
+    },
+    {
+      "epoch": 0.2627627627627628,
+      "grad_norm": 0.6387883870614447,
+      "learning_rate": 1.4903640256959315e-05,
+      "loss": 0.3892,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.4122806787490845,
+      "step": 175,
+      "valid_targets_mean": 4106.8,
+      "valid_targets_min": 367
+    },
+    {
+      "epoch": 0.2702702702702703,
+      "grad_norm": 0.5859730062305724,
+      "learning_rate": 1.5331905781584583e-05,
+      "loss": 0.4145,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.4545392692089081,
+      "step": 180,
+      "valid_targets_mean": 5412.8,
+      "valid_targets_min": 485
+    },
+    {
+      "epoch": 0.2777777777777778,
+      "grad_norm": 0.6291566957212851,
+      "learning_rate": 1.576017130620985e-05,
+      "loss": 0.3586,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.34801965951919556,
+      "step": 185,
+      "valid_targets_mean": 3683.8,
+      "valid_targets_min": 388
+    },
+    {
+      "epoch": 0.2852852852852853,
+      "grad_norm": 0.7714952807971188,
+      "learning_rate": 1.618843683083512e-05,
+      "loss": 0.4031,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.33357149362564087,
+      "step": 190,
+      "valid_targets_mean": 3992.8,
+      "valid_targets_min": 645
+    },
+    {
+      "epoch": 0.2927927927927928,
+      "grad_norm": 0.6022688124054402,
+      "learning_rate": 1.6616702355460387e-05,
+      "loss": 0.3632,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.32963526248931885,
+      "step": 195,
+      "valid_targets_mean": 4479.6,
+      "valid_targets_min": 376
+    },
+    {
+      "epoch": 0.3003003003003003,
+      "grad_norm": 0.5461125797479052,
+      "learning_rate": 1.7044967880085653e-05,
+      "loss": 0.3609,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3471255898475647,
+      "step": 200,
+      "valid_targets_mean": 5124.2,
+      "valid_targets_min": 2440
+    },
+    {
+      "epoch": 0.3078078078078078,
+      "grad_norm": 0.6479231652954496,
+      "learning_rate": 1.7473233404710924e-05,
+      "loss": 0.3709,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.36097437143325806,
+      "step": 205,
+      "valid_targets_mean": 3874.2,
+      "valid_targets_min": 536
+    },
+    {
+      "epoch": 0.3153153153153153,
+      "grad_norm": 0.6307321400936947,
+      "learning_rate": 1.790149892933619e-05,
+      "loss": 0.356,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.32168492674827576,
+      "step": 210,
+      "valid_targets_mean": 4215.9,
+      "valid_targets_min": 750
+    },
+    {
+      "epoch": 0.3228228228228228,
+      "grad_norm": 0.7342733098843338,
+      "learning_rate": 1.8329764453961457e-05,
+      "loss": 0.3716,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3192724585533142,
+      "step": 215,
+      "valid_targets_mean": 3300.5,
+      "valid_targets_min": 465
+    },
+    {
+      "epoch": 0.3303303303303303,
+      "grad_norm": 0.6269217295173055,
+      "learning_rate": 1.8758029978586724e-05,
+      "loss": 0.382,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.393771767616272,
+      "step": 220,
+      "valid_targets_mean": 3780.5,
+      "valid_targets_min": 520
+    },
+    {
+      "epoch": 0.33783783783783783,
+      "grad_norm": 0.6699586606597604,
+      "learning_rate": 1.9186295503211994e-05,
+      "loss": 0.3367,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3441130518913269,
+      "step": 225,
+      "valid_targets_mean": 3570.4,
+      "valid_targets_min": 367
+    },
+    {
+      "epoch": 0.34534534534534533,
+      "grad_norm": 0.6212030440388314,
+      "learning_rate": 1.961456102783726e-05,
+      "loss": 0.3249,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28860020637512207,
+      "step": 230,
+      "valid_targets_mean": 4043.8,
+      "valid_targets_min": 797
+    },
+    {
+      "epoch": 0.35285285285285284,
+      "grad_norm": 0.5933247879185318,
+      "learning_rate": 2.004282655246253e-05,
+      "loss": 0.3663,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2963605523109436,
+      "step": 235,
+      "valid_targets_mean": 4519.4,
+      "valid_targets_min": 679
+    },
+    {
+      "epoch": 0.36036036036036034,
+      "grad_norm": 0.5915857140141431,
+      "learning_rate": 2.0471092077087794e-05,
+      "loss": 0.3707,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3590105175971985,
+      "step": 240,
+      "valid_targets_mean": 4801.9,
+      "valid_targets_min": 786
+    },
+    {
+      "epoch": 0.36786786786786785,
+      "grad_norm": 0.6926624493340766,
+      "learning_rate": 2.089935760171306e-05,
+      "loss": 0.3699,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3375079035758972,
+      "step": 245,
+      "valid_targets_mean": 3064.8,
+      "valid_targets_min": 636
+    },
+    {
+      "epoch": 0.37537537537537535,
+      "grad_norm": 0.5625060900455121,
+      "learning_rate": 2.1327623126338334e-05,
+      "loss": 0.3717,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2878052592277527,
+      "step": 250,
+      "valid_targets_mean": 4984.9,
+      "valid_targets_min": 389
+    },
+    {
+      "epoch": 0.38288288288288286,
+      "grad_norm": 0.4895346484151352,
+      "learning_rate": 2.17558886509636e-05,
+      "loss": 0.3319,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3224838972091675,
+      "step": 255,
+      "valid_targets_mean": 5919.2,
+      "valid_targets_min": 631
+    },
+    {
+      "epoch": 0.39039039039039036,
+      "grad_norm": 0.6186907667129198,
+      "learning_rate": 2.2184154175588867e-05,
+      "loss": 0.3133,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3299029767513275,
+      "step": 260,
+      "valid_targets_mean": 4151.9,
+      "valid_targets_min": 694
+    },
+    {
+      "epoch": 0.3978978978978979,
+      "grad_norm": 0.637540257330486,
+      "learning_rate": 2.2612419700214137e-05,
+      "loss": 0.3034,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.31127655506134033,
+      "step": 265,
+      "valid_targets_mean": 4565.4,
+      "valid_targets_min": 667
+    },
+    {
+      "epoch": 0.40540540540540543,
+      "grad_norm": 0.6482997719326944,
+      "learning_rate": 2.3040685224839404e-05,
+      "loss": 0.371,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3371477723121643,
+      "step": 270,
+      "valid_targets_mean": 4799.7,
+      "valid_targets_min": 565
+    },
+    {
+      "epoch": 0.41291291291291293,
+      "grad_norm": 0.630240697831865,
+      "learning_rate": 2.346895074946467e-05,
+      "loss": 0.4042,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3549198508262634,
+      "step": 275,
+      "valid_targets_mean": 3973.2,
+      "valid_targets_min": 571
+    },
+    {
+      "epoch": 0.42042042042042044,
+      "grad_norm": 0.5378005948550142,
+      "learning_rate": 2.3897216274089937e-05,
+      "loss": 0.3568,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3140275776386261,
+      "step": 280,
+      "valid_targets_mean": 4524.4,
+      "valid_targets_min": 466
+    },
+    {
+      "epoch": 0.42792792792792794,
+      "grad_norm": 0.5496707584360776,
+      "learning_rate": 2.4325481798715207e-05,
+      "loss": 0.3439,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27984702587127686,
+      "step": 285,
+      "valid_targets_mean": 5251.4,
+      "valid_targets_min": 871
+    },
+    {
+      "epoch": 0.43543543543543545,
+      "grad_norm": 0.6054529261217243,
+      "learning_rate": 2.4753747323340474e-05,
+      "loss": 0.3399,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2717384994029999,
+      "step": 290,
+      "valid_targets_mean": 5978.3,
+      "valid_targets_min": 665
+    },
+    {
+      "epoch": 0.44294294294294295,
+      "grad_norm": 0.60946853866193,
+      "learning_rate": 2.518201284796574e-05,
+      "loss": 0.3324,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.31978553533554077,
+      "step": 295,
+      "valid_targets_mean": 3610.3,
+      "valid_targets_min": 729
+    },
+    {
+      "epoch": 0.45045045045045046,
+      "grad_norm": 0.7112237553385803,
+      "learning_rate": 2.5610278372591007e-05,
+      "loss": 0.365,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.36318349838256836,
+      "step": 300,
+      "valid_targets_mean": 2997.6,
+      "valid_targets_min": 398
+    },
+    {
+      "epoch": 0.45795795795795796,
+      "grad_norm": 0.5754561363093669,
+      "learning_rate": 2.6038543897216277e-05,
+      "loss": 0.3475,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3775876760482788,
+      "step": 305,
+      "valid_targets_mean": 5485.2,
+      "valid_targets_min": 291
+    },
+    {
+      "epoch": 0.46546546546546547,
+      "grad_norm": 0.5225836627231101,
+      "learning_rate": 2.6466809421841544e-05,
+      "loss": 0.3327,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2630726099014282,
+      "step": 310,
+      "valid_targets_mean": 5477.4,
+      "valid_targets_min": 1953
+    },
+    {
+      "epoch": 0.47297297297297297,
+      "grad_norm": 0.6631066150226813,
+      "learning_rate": 2.689507494646681e-05,
+      "loss": 0.3341,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.4078267812728882,
+      "step": 315,
+      "valid_targets_mean": 4227.4,
+      "valid_targets_min": 691
+    },
+    {
+      "epoch": 0.4804804804804805,
+      "grad_norm": 0.5625056894327084,
+      "learning_rate": 2.732334047109208e-05,
+      "loss": 0.3307,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.34400713443756104,
+      "step": 320,
+      "valid_targets_mean": 6082.6,
+      "valid_targets_min": 756
+    },
+    {
+      "epoch": 0.487987987987988,
+      "grad_norm": 0.5720940827291631,
+      "learning_rate": 2.7751605995717347e-05,
+      "loss": 0.3544,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.364154189825058,
+      "step": 325,
+      "valid_targets_mean": 5058.0,
+      "valid_targets_min": 1324
+    },
+    {
+      "epoch": 0.4954954954954955,
+      "grad_norm": 0.6142764560167397,
+      "learning_rate": 2.8179871520342614e-05,
+      "loss": 0.3567,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3340011239051819,
+      "step": 330,
+      "valid_targets_mean": 3809.8,
+      "valid_targets_min": 606
+    },
+    {
+      "epoch": 0.503003003003003,
+      "grad_norm": 0.6862489554890606,
+      "learning_rate": 2.860813704496788e-05,
+      "loss": 0.3377,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.347086638212204,
+      "step": 335,
+      "valid_targets_mean": 3381.4,
+      "valid_targets_min": 529
+    },
+    {
+      "epoch": 0.5105105105105106,
+      "grad_norm": 0.5884162448896875,
+      "learning_rate": 2.903640256959315e-05,
+      "loss": 0.3156,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3551441431045532,
+      "step": 340,
+      "valid_targets_mean": 4280.0,
+      "valid_targets_min": 656
+    },
+    {
+      "epoch": 0.5180180180180181,
+      "grad_norm": 0.5985519809481195,
+      "learning_rate": 2.9464668094218417e-05,
+      "loss": 0.3718,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3538270592689514,
+      "step": 345,
+      "valid_targets_mean": 4933.6,
+      "valid_targets_min": 679
+    },
+    {
+      "epoch": 0.5255255255255256,
+      "grad_norm": 0.6608489868770677,
+      "learning_rate": 2.9892933618843684e-05,
+      "loss": 0.3821,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3977230191230774,
+      "step": 350,
+      "valid_targets_mean": 4619.8,
+      "valid_targets_min": 617
+    },
+    {
+      "epoch": 0.5330330330330331,
+      "grad_norm": 0.6666435654761541,
+      "learning_rate": 3.032119914346895e-05,
+      "loss": 0.3205,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3276807963848114,
+      "step": 355,
+      "valid_targets_mean": 3228.0,
+      "valid_targets_min": 396
+    },
+    {
+      "epoch": 0.5405405405405406,
+      "grad_norm": 0.5700169845584041,
+      "learning_rate": 3.0749464668094224e-05,
+      "loss": 0.3373,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.252112478017807,
+      "step": 360,
+      "valid_targets_mean": 4043.2,
+      "valid_targets_min": 470
+    },
+    {
+      "epoch": 0.5480480480480481,
+      "grad_norm": 0.5969317196396643,
+      "learning_rate": 3.117773019271949e-05,
+      "loss": 0.3344,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.257426917552948,
+      "step": 365,
+      "valid_targets_mean": 3501.2,
+      "valid_targets_min": 637
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 0.5338582932799799,
+      "learning_rate": 3.160599571734476e-05,
+      "loss": 0.3469,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.4118390679359436,
+      "step": 370,
+      "valid_targets_mean": 6022.6,
+      "valid_targets_min": 668
+    },
+    {
+      "epoch": 0.5630630630630631,
+      "grad_norm": 0.7156868770693144,
+      "learning_rate": 3.2034261241970024e-05,
+      "loss": 0.342,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.33580997586250305,
+      "step": 375,
+      "valid_targets_mean": 3146.6,
+      "valid_targets_min": 580
+    },
+    {
+      "epoch": 0.5705705705705706,
+      "grad_norm": 0.6105442252909425,
+      "learning_rate": 3.246252676659529e-05,
+      "loss": 0.3246,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.29976391792297363,
+      "step": 380,
+      "valid_targets_mean": 4423.6,
+      "valid_targets_min": 579
+    },
+    {
+      "epoch": 0.5780780780780781,
+      "grad_norm": 0.48535869781119606,
+      "learning_rate": 3.289079229122056e-05,
+      "loss": 0.3211,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2958953380584717,
+      "step": 385,
+      "valid_targets_mean": 5096.5,
+      "valid_targets_min": 1111
+    },
+    {
+      "epoch": 0.5855855855855856,
+      "grad_norm": 0.5293285996606892,
+      "learning_rate": 3.3319057815845824e-05,
+      "loss": 0.3012,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.29280033707618713,
+      "step": 390,
+      "valid_targets_mean": 4133.6,
+      "valid_targets_min": 664
+    },
+    {
+      "epoch": 0.5930930930930931,
+      "grad_norm": 0.6364074446274751,
+      "learning_rate": 3.37473233404711e-05,
+      "loss": 0.3546,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3336215019226074,
+      "step": 395,
+      "valid_targets_mean": 3474.1,
+      "valid_targets_min": 277
+    },
+    {
+      "epoch": 0.6006006006006006,
+      "grad_norm": 0.7064277061832516,
+      "learning_rate": 3.4175588865096364e-05,
+      "loss": 0.3238,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.38183677196502686,
+      "step": 400,
+      "valid_targets_mean": 4343.8,
+      "valid_targets_min": 831
+    },
+    {
+      "epoch": 0.6081081081081081,
+      "grad_norm": 0.5234255969114723,
+      "learning_rate": 3.460385438972163e-05,
+      "loss": 0.3259,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3263946771621704,
+      "step": 405,
+      "valid_targets_mean": 5912.4,
+      "valid_targets_min": 525
+    },
+    {
+      "epoch": 0.6156156156156156,
+      "grad_norm": 0.6859056535544392,
+      "learning_rate": 3.50321199143469e-05,
+      "loss": 0.3613,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.4518062472343445,
+      "step": 410,
+      "valid_targets_mean": 4294.0,
+      "valid_targets_min": 672
+    },
+    {
+      "epoch": 0.6231231231231231,
+      "grad_norm": 0.593812126409495,
+      "learning_rate": 3.5460385438972164e-05,
+      "loss": 0.3236,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.284976601600647,
+      "step": 415,
+      "valid_targets_mean": 4542.8,
+      "valid_targets_min": 732
+    },
+    {
+      "epoch": 0.6306306306306306,
+      "grad_norm": 0.5542761631226861,
+      "learning_rate": 3.588865096359743e-05,
+      "loss": 0.3575,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3369911313056946,
+      "step": 420,
+      "valid_targets_mean": 4173.8,
+      "valid_targets_min": 623
+    },
+    {
+      "epoch": 0.6381381381381381,
+      "grad_norm": 0.694482077890855,
+      "learning_rate": 3.63169164882227e-05,
+      "loss": 0.331,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.35457199811935425,
+      "step": 425,
+      "valid_targets_mean": 3347.3,
+      "valid_targets_min": 323
+    },
+    {
+      "epoch": 0.6456456456456456,
+      "grad_norm": 0.5087776401105766,
+      "learning_rate": 3.674518201284797e-05,
+      "loss": 0.3077,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2870054244995117,
+      "step": 430,
+      "valid_targets_mean": 5496.8,
+      "valid_targets_min": 995
+    },
+    {
+      "epoch": 0.6531531531531531,
+      "grad_norm": 0.5710934865190025,
+      "learning_rate": 3.717344753747324e-05,
+      "loss": 0.3449,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3406960964202881,
+      "step": 435,
+      "valid_targets_mean": 4272.1,
+      "valid_targets_min": 578
+    },
+    {
+      "epoch": 0.6606606606606606,
+      "grad_norm": 0.6532417856194169,
+      "learning_rate": 3.7601713062098504e-05,
+      "loss": 0.3358,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3105970621109009,
+      "step": 440,
+      "valid_targets_mean": 4052.7,
+      "valid_targets_min": 469
+    },
+    {
+      "epoch": 0.6681681681681682,
+      "grad_norm": 0.5810179731496691,
+      "learning_rate": 3.802997858672377e-05,
+      "loss": 0.3534,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2967357933521271,
+      "step": 445,
+      "valid_targets_mean": 5399.5,
+      "valid_targets_min": 481
+    },
+    {
+      "epoch": 0.6756756756756757,
+      "grad_norm": 0.4933051723343659,
+      "learning_rate": 3.8458244111349045e-05,
+      "loss": 0.307,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2656002640724182,
+      "step": 450,
+      "valid_targets_mean": 5050.2,
+      "valid_targets_min": 536
+    },
+    {
+      "epoch": 0.6831831831831832,
+      "grad_norm": 0.5117823141215931,
+      "learning_rate": 3.8886509635974305e-05,
+      "loss": 0.3345,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3394744396209717,
+      "step": 455,
+      "valid_targets_mean": 4780.2,
+      "valid_targets_min": 698
+    },
+    {
+      "epoch": 0.6906906906906907,
+      "grad_norm": 0.5513150421712305,
+      "learning_rate": 3.931477516059957e-05,
+      "loss": 0.3344,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.38678789138793945,
+      "step": 460,
+      "valid_targets_mean": 4995.6,
+      "valid_targets_min": 787
+    },
+    {
+      "epoch": 0.6981981981981982,
+      "grad_norm": 0.5223578821175074,
+      "learning_rate": 3.974304068522484e-05,
+      "loss": 0.3301,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3217771053314209,
+      "step": 465,
+      "valid_targets_mean": 5331.4,
+      "valid_targets_min": 1007
+    },
+    {
+      "epoch": 0.7057057057057057,
+      "grad_norm": 0.5938737922502441,
+      "learning_rate": 3.999997756656772e-05,
+      "loss": 0.3306,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.31824183464050293,
+      "step": 470,
+      "valid_targets_mean": 4223.0,
+      "valid_targets_min": 452
+    },
+    {
+      "epoch": 0.7132132132132132,
+      "grad_norm": 0.49626310595338613,
+      "learning_rate": 3.9999725191032436e-05,
+      "loss": 0.2875,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26925772428512573,
+      "step": 475,
+      "valid_targets_mean": 5763.9,
+      "valid_targets_min": 761
+    },
+    {
+      "epoch": 0.7207207207207207,
+      "grad_norm": 0.6276161448251935,
+      "learning_rate": 3.999919240172184e-05,
+      "loss": 0.3228,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.325122594833374,
+      "step": 480,
+      "valid_targets_mean": 4395.4,
+      "valid_targets_min": 662
+    },
+    {
+      "epoch": 0.7282282282282282,
+      "grad_norm": 0.598897987227093,
+      "learning_rate": 3.9998379206106104e-05,
+      "loss": 0.2941,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2764396369457245,
+      "step": 485,
+      "valid_targets_mean": 5031.5,
+      "valid_targets_min": 836
+    },
+    {
+      "epoch": 0.7357357357357357,
+      "grad_norm": 0.5920738798268582,
+      "learning_rate": 3.9997285615586944e-05,
+      "loss": 0.3449,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.33547401428222656,
+      "step": 490,
+      "valid_targets_mean": 4218.7,
+      "valid_targets_min": 416
+    },
+    {
+      "epoch": 0.7432432432432432,
+      "grad_norm": 0.5388292737396405,
+      "learning_rate": 3.999591164549747e-05,
+      "loss": 0.3216,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.30026373267173767,
+      "step": 495,
+      "valid_targets_mean": 5046.2,
+      "valid_targets_min": 885
+    },
+    {
+      "epoch": 0.7507507507507507,
+      "grad_norm": 0.7138413019632255,
+      "learning_rate": 3.999425731510195e-05,
+      "loss": 0.3161,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3095625340938568,
+      "step": 500,
+      "valid_targets_mean": 3115.5,
+      "valid_targets_min": 696
+    },
+    {
+      "epoch": 0.7582582582582582,
+      "grad_norm": 0.5065067605354636,
+      "learning_rate": 3.9992322647595543e-05,
+      "loss": 0.3132,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2668853998184204,
+      "step": 505,
+      "valid_targets_mean": 4890.2,
+      "valid_targets_min": 583
+    },
+    {
+      "epoch": 0.7657657657657657,
+      "grad_norm": 0.48772270569522996,
+      "learning_rate": 3.999010767010401e-05,
+      "loss": 0.2792,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21127644181251526,
+      "step": 510,
+      "valid_targets_mean": 4030.5,
+      "valid_targets_min": 620
+    },
+    {
+      "epoch": 0.7732732732732732,
+      "grad_norm": 0.688910852981923,
+      "learning_rate": 3.998761241368328e-05,
+      "loss": 0.3356,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.31102752685546875,
+      "step": 515,
+      "valid_targets_mean": 2963.6,
+      "valid_targets_min": 662
+    },
+    {
+      "epoch": 0.7807807807807807,
+      "grad_norm": 0.590399156229966,
+      "learning_rate": 3.9984836913319044e-05,
+      "loss": 0.3022,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.295174777507782,
+      "step": 520,
+      "valid_targets_mean": 3820.8,
+      "valid_targets_min": 547
+    },
+    {
+      "epoch": 0.7882882882882883,
+      "grad_norm": 0.7498553089205983,
+      "learning_rate": 3.998178120792627e-05,
+      "loss": 0.3289,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.29107627272605896,
+      "step": 525,
+      "valid_targets_mean": 4623.9,
+      "valid_targets_min": 463
+    },
+    {
+      "epoch": 0.7957957957957958,
+      "grad_norm": 0.541430521248251,
+      "learning_rate": 3.9978445340348646e-05,
+      "loss": 0.3114,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.35057809948921204,
+      "step": 530,
+      "valid_targets_mean": 4895.9,
+      "valid_targets_min": 810
+    },
+    {
+      "epoch": 0.8033033033033034,
+      "grad_norm": 0.6145616694631061,
+      "learning_rate": 3.997482935735796e-05,
+      "loss": 0.2989,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2825467586517334,
+      "step": 535,
+      "valid_targets_mean": 4079.4,
+      "valid_targets_min": 593
+    },
+    {
+      "epoch": 0.8108108108108109,
+      "grad_norm": 0.6457216047110463,
+      "learning_rate": 3.997093330965349e-05,
+      "loss": 0.3289,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.37978658080101013,
+      "step": 540,
+      "valid_targets_mean": 3493.6,
+      "valid_targets_min": 513
+    },
+    {
+      "epoch": 0.8183183183183184,
+      "grad_norm": 0.5512024613609153,
+      "learning_rate": 3.9966757251861256e-05,
+      "loss": 0.3083,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.30473941564559937,
+      "step": 545,
+      "valid_targets_mean": 4363.6,
+      "valid_targets_min": 798
+    },
+    {
+      "epoch": 0.8258258258258259,
+      "grad_norm": 0.5404758251655593,
+      "learning_rate": 3.996230124253328e-05,
+      "loss": 0.2928,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2958792746067047,
+      "step": 550,
+      "valid_targets_mean": 4708.9,
+      "valid_targets_min": 480
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 0.5409326702314928,
+      "learning_rate": 3.995756534414672e-05,
+      "loss": 0.3015,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3055042028427124,
+      "step": 555,
+      "valid_targets_mean": 4889.0,
+      "valid_targets_min": 634
+    },
+    {
+      "epoch": 0.8408408408408409,
+      "grad_norm": 0.877523092977663,
+      "learning_rate": 3.9952549623103065e-05,
+      "loss": 0.3031,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27620965242385864,
+      "step": 560,
+      "valid_targets_mean": 3901.4,
+      "valid_targets_min": 782
+    },
+    {
+      "epoch": 0.8483483483483484,
+      "grad_norm": 0.5593566066914221,
+      "learning_rate": 3.994725414972713e-05,
+      "loss": 0.331,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.33034461736679077,
+      "step": 565,
+      "valid_targets_mean": 4464.5,
+      "valid_targets_min": 847
+    },
+    {
+      "epoch": 0.8558558558558559,
+      "grad_norm": 0.6652170124791761,
+      "learning_rate": 3.994167899826614e-05,
+      "loss": 0.2967,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.258162260055542,
+      "step": 570,
+      "valid_targets_mean": 3842.9,
+      "valid_targets_min": 410
+    },
+    {
+      "epoch": 0.8633633633633634,
+      "grad_norm": 0.5319260103081287,
+      "learning_rate": 3.993582424688861e-05,
+      "loss": 0.2819,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2436560094356537,
+      "step": 575,
+      "valid_targets_mean": 4503.2,
+      "valid_targets_min": 810
+    },
+    {
+      "epoch": 0.8708708708708709,
+      "grad_norm": 0.5435805707257723,
+      "learning_rate": 3.9929689977683315e-05,
+      "loss": 0.3088,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2837572991847992,
+      "step": 580,
+      "valid_targets_mean": 4475.9,
+      "valid_targets_min": 697
+    },
+    {
+      "epoch": 0.8783783783783784,
+      "grad_norm": 0.6279074659022925,
+      "learning_rate": 3.992327627665814e-05,
+      "loss": 0.3233,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3020919859409332,
+      "step": 585,
+      "valid_targets_mean": 3179.3,
+      "valid_targets_min": 602
+    },
+    {
+      "epoch": 0.8858858858858859,
+      "grad_norm": 0.6751035596060991,
+      "learning_rate": 3.99165832337388e-05,
+      "loss": 0.3098,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2893165349960327,
+      "step": 590,
+      "valid_targets_mean": 3005.4,
+      "valid_targets_min": 504
+    },
+    {
+      "epoch": 0.8933933933933934,
+      "grad_norm": 0.9253018906213728,
+      "learning_rate": 3.990961094276767e-05,
+      "loss": 0.3072,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3459416627883911,
+      "step": 595,
+      "valid_targets_mean": 2077.6,
+      "valid_targets_min": 485
+    },
+    {
+      "epoch": 0.9009009009009009,
+      "grad_norm": 0.6687225112938137,
+      "learning_rate": 3.990235950150242e-05,
+      "loss": 0.3079,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.36818212270736694,
+      "step": 600,
+      "valid_targets_mean": 3371.7,
+      "valid_targets_min": 715
+    },
+    {
+      "epoch": 0.9084084084084084,
+      "grad_norm": 0.5328749806757945,
+      "learning_rate": 3.9894829011614627e-05,
+      "loss": 0.3135,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25350844860076904,
+      "step": 605,
+      "valid_targets_mean": 3857.8,
+      "valid_targets_min": 521
+    },
+    {
+      "epoch": 0.9159159159159159,
+      "grad_norm": 0.675741595187441,
+      "learning_rate": 3.988701957868842e-05,
+      "loss": 0.3172,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3360707759857178,
+      "step": 610,
+      "valid_targets_mean": 2917.3,
+      "valid_targets_min": 605
+    },
+    {
+      "epoch": 0.9234234234234234,
+      "grad_norm": 0.6571910621752247,
+      "learning_rate": 3.987893131221892e-05,
+      "loss": 0.2958,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2482086569070816,
+      "step": 615,
+      "valid_targets_mean": 3313.1,
+      "valid_targets_min": 510
+    },
+    {
+      "epoch": 0.9309309309309309,
+      "grad_norm": 0.6979768745974733,
+      "learning_rate": 3.987056432561075e-05,
+      "loss": 0.3034,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3537478744983673,
+      "step": 620,
+      "valid_targets_mean": 3899.6,
+      "valid_targets_min": 471
+    },
+    {
+      "epoch": 0.9384384384384384,
+      "grad_norm": 0.46982739010374835,
+      "learning_rate": 3.986191873617645e-05,
+      "loss": 0.318,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27537375688552856,
+      "step": 625,
+      "valid_targets_mean": 5885.1,
+      "valid_targets_min": 593
+    },
+    {
+      "epoch": 0.9459459459459459,
+      "grad_norm": 0.5400448261626185,
+      "learning_rate": 3.985299466513479e-05,
+      "loss": 0.3106,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3220127820968628,
+      "step": 630,
+      "valid_targets_mean": 4269.6,
+      "valid_targets_min": 558
+    },
+    {
+      "epoch": 0.9534534534534534,
+      "grad_norm": 0.5242379048874888,
+      "learning_rate": 3.984379223760912e-05,
+      "loss": 0.3102,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.36460551619529724,
+      "step": 635,
+      "valid_targets_mean": 4742.2,
+      "valid_targets_min": 670
+    },
+    {
+      "epoch": 0.960960960960961,
+      "grad_norm": 0.5056910710384006,
+      "learning_rate": 3.983431158262558e-05,
+      "loss": 0.2783,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2684648334980011,
+      "step": 640,
+      "valid_targets_mean": 4608.9,
+      "valid_targets_min": 805
+    },
+    {
+      "epoch": 0.9684684684684685,
+      "grad_norm": 0.5297540154671703,
+      "learning_rate": 3.982455283311131e-05,
+      "loss": 0.325,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25975459814071655,
+      "step": 645,
+      "valid_targets_mean": 4382.8,
+      "valid_targets_min": 772
+    },
+    {
+      "epoch": 0.975975975975976,
+      "grad_norm": 0.555629136176801,
+      "learning_rate": 3.981451612589259e-05,
+      "loss": 0.28,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26162397861480713,
+      "step": 650,
+      "valid_targets_mean": 4827.0,
+      "valid_targets_min": 611
+    },
+    {
+      "epoch": 0.9834834834834835,
+      "grad_norm": 0.7355128171152777,
+      "learning_rate": 3.980420160169289e-05,
+      "loss": 0.2788,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28353631496429443,
+      "step": 655,
+      "valid_targets_mean": 4060.4,
+      "valid_targets_min": 766
+    },
+    {
+      "epoch": 0.990990990990991,
+      "grad_norm": 0.6098329219635357,
+      "learning_rate": 3.979360940513092e-05,
+      "loss": 0.3267,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2917211651802063,
+      "step": 660,
+      "valid_targets_mean": 3110.2,
+      "valid_targets_min": 536
+    },
+    {
+      "epoch": 0.9984984984984985,
+      "grad_norm": 0.6602685528415261,
+      "learning_rate": 3.978273968471865e-05,
+      "loss": 0.3291,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.37517815828323364,
+      "step": 665,
+      "valid_targets_mean": 3097.9,
+      "valid_targets_min": 623
+    },
+    {
+      "epoch": 1.006006006006006,
+      "grad_norm": 0.5990905965495634,
+      "learning_rate": 3.97715925928591e-05,
+      "loss": 0.3701,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.31573376059532166,
+      "step": 670,
+      "valid_targets_mean": 5335.8,
+      "valid_targets_min": 657
+    },
+    {
+      "epoch": 1.0135135135135136,
+      "grad_norm": 0.5051572310687712,
+      "learning_rate": 3.9760168285844345e-05,
+      "loss": 0.2747,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2896278500556946,
+      "step": 675,
+      "valid_targets_mean": 4987.1,
+      "valid_targets_min": 724
+    },
+    {
+      "epoch": 1.021021021021021,
+      "grad_norm": 0.5358689615840903,
+      "learning_rate": 3.974846692385323e-05,
+      "loss": 0.2765,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.29317814111709595,
+      "step": 680,
+      "valid_targets_mean": 5053.4,
+      "valid_targets_min": 723
+    },
+    {
+      "epoch": 1.0285285285285286,
+      "grad_norm": 0.7789287894144826,
+      "learning_rate": 3.973648867094917e-05,
+      "loss": 0.3181,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3238382935523987,
+      "step": 685,
+      "valid_targets_mean": 3482.4,
+      "valid_targets_min": 492
+    },
+    {
+      "epoch": 1.0360360360360361,
+      "grad_norm": 0.5563018194673951,
+      "learning_rate": 3.972423369507783e-05,
+      "loss": 0.3166,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3259349465370178,
+      "step": 690,
+      "valid_targets_mean": 5252.0,
+      "valid_targets_min": 850
+    },
+    {
+      "epoch": 1.0435435435435436,
+      "grad_norm": 0.7250068591194968,
+      "learning_rate": 3.971170216806477e-05,
+      "loss": 0.3198,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3443298935890198,
+      "step": 695,
+      "valid_targets_mean": 2799.4,
+      "valid_targets_min": 588
+    },
+    {
+      "epoch": 1.0510510510510511,
+      "grad_norm": 0.7089445149146085,
+      "learning_rate": 3.969889426561304e-05,
+      "loss": 0.3328,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.31399601697921753,
+      "step": 700,
+      "valid_targets_mean": 3693.1,
+      "valid_targets_min": 606
+    },
+    {
+      "epoch": 1.0585585585585586,
+      "grad_norm": 0.4834579345868247,
+      "learning_rate": 3.968581016730073e-05,
+      "loss": 0.2893,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28393715620040894,
+      "step": 705,
+      "valid_targets_mean": 5210.3,
+      "valid_targets_min": 1146
+    },
+    {
+      "epoch": 1.0660660660660661,
+      "grad_norm": 0.580488043726015,
+      "learning_rate": 3.9672450056578426e-05,
+      "loss": 0.3114,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3525095582008362,
+      "step": 710,
+      "valid_targets_mean": 4987.9,
+      "valid_targets_min": 872
+    },
+    {
+      "epoch": 1.0735735735735736,
+      "grad_norm": 0.5830757190666622,
+      "learning_rate": 3.965881412076665e-05,
+      "loss": 0.2927,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2518344223499298,
+      "step": 715,
+      "valid_targets_mean": 3772.8,
+      "valid_targets_min": 301
+    },
+    {
+      "epoch": 1.0810810810810811,
+      "grad_norm": 0.603911299798853,
+      "learning_rate": 3.964490255105325e-05,
+      "loss": 0.2897,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2885856330394745,
+      "step": 720,
+      "valid_targets_mean": 3888.7,
+      "valid_targets_min": 610
+    },
+    {
+      "epoch": 1.0885885885885886,
+      "grad_norm": 0.721271255619491,
+      "learning_rate": 3.96307155424907e-05,
+      "loss": 0.3009,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.33753496408462524,
+      "step": 725,
+      "valid_targets_mean": 3255.9,
+      "valid_targets_min": 323
+    },
+    {
+      "epoch": 1.0960960960960962,
+      "grad_norm": 0.5688287102135644,
+      "learning_rate": 3.961625329399335e-05,
+      "loss": 0.288,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24445214867591858,
+      "step": 730,
+      "valid_targets_mean": 4295.9,
+      "valid_targets_min": 322
+    },
+    {
+      "epoch": 1.1036036036036037,
+      "grad_norm": 0.5822780762193985,
+      "learning_rate": 3.960151600833469e-05,
+      "loss": 0.303,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3242020606994629,
+      "step": 735,
+      "valid_targets_mean": 4509.4,
+      "valid_targets_min": 749
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 0.6988383396255673,
+      "learning_rate": 3.958650389214444e-05,
+      "loss": 0.2908,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2766273021697998,
+      "step": 740,
+      "valid_targets_mean": 3003.1,
+      "valid_targets_min": 528
+    },
+    {
+      "epoch": 1.1186186186186187,
+      "grad_norm": 0.5530037303674098,
+      "learning_rate": 3.95712171559057e-05,
+      "loss": 0.2862,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2562218904495239,
+      "step": 745,
+      "valid_targets_mean": 4301.1,
+      "valid_targets_min": 761
+    },
+    {
+      "epoch": 1.1261261261261262,
+      "grad_norm": 0.5917466496490228,
+      "learning_rate": 3.955565601395199e-05,
+      "loss": 0.2995,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27801835536956787,
+      "step": 750,
+      "valid_targets_mean": 4016.3,
+      "valid_targets_min": 611
+    },
+    {
+      "epoch": 1.1336336336336337,
+      "grad_norm": 0.5314431651828969,
+      "learning_rate": 3.953982068446424e-05,
+      "loss": 0.274,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2895042896270752,
+      "step": 755,
+      "valid_targets_mean": 4829.9,
+      "valid_targets_min": 689
+    },
+    {
+      "epoch": 1.1411411411411412,
+      "grad_norm": 0.5403700692721292,
+      "learning_rate": 3.952371138946773e-05,
+      "loss": 0.311,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2900671660900116,
+      "step": 760,
+      "valid_targets_mean": 5102.2,
+      "valid_targets_min": 811
+    },
+    {
+      "epoch": 1.1486486486486487,
+      "grad_norm": 0.4466270912829498,
+      "learning_rate": 3.950732835482896e-05,
+      "loss": 0.2634,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23144987225532532,
+      "step": 765,
+      "valid_targets_mean": 4800.4,
+      "valid_targets_min": 566
+    },
+    {
+      "epoch": 1.1561561561561562,
+      "grad_norm": 0.5126441976083996,
+      "learning_rate": 3.9490671810252525e-05,
+      "loss": 0.2738,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2398991733789444,
+      "step": 770,
+      "valid_targets_mean": 3322.2,
+      "valid_targets_min": 260
+    },
+    {
+      "epoch": 1.1636636636636637,
+      "grad_norm": 0.6835376056729422,
+      "learning_rate": 3.947374198927787e-05,
+      "loss": 0.2818,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2788790464401245,
+      "step": 775,
+      "valid_targets_mean": 2824.4,
+      "valid_targets_min": 481
+    },
+    {
+      "epoch": 1.1711711711711712,
+      "grad_norm": 0.7480255450833108,
+      "learning_rate": 3.945653912927599e-05,
+      "loss": 0.3116,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.41352379322052,
+      "step": 780,
+      "valid_targets_mean": 3108.0,
+      "valid_targets_min": 525
+    },
+    {
+      "epoch": 1.1786786786786787,
+      "grad_norm": 0.7012989879636262,
+      "learning_rate": 3.943906347144615e-05,
+      "loss": 0.2908,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2697921395301819,
+      "step": 785,
+      "valid_targets_mean": 2765.4,
+      "valid_targets_min": 559
+    },
+    {
+      "epoch": 1.1861861861861862,
+      "grad_norm": 0.6294134200242447,
+      "learning_rate": 3.942131526081249e-05,
+      "loss": 0.2965,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28815174102783203,
+      "step": 790,
+      "valid_targets_mean": 5498.2,
+      "valid_targets_min": 760
+    },
+    {
+      "epoch": 1.1936936936936937,
+      "grad_norm": 0.5405122434808368,
+      "learning_rate": 3.940329474622055e-05,
+      "loss": 0.3025,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26721876859664917,
+      "step": 795,
+      "valid_targets_mean": 4779.1,
+      "valid_targets_min": 657
+    },
+    {
+      "epoch": 1.2012012012012012,
+      "grad_norm": 0.6027467491848216,
+      "learning_rate": 3.938500218033384e-05,
+      "loss": 0.2749,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28523164987564087,
+      "step": 800,
+      "valid_targets_mean": 3985.0,
+      "valid_targets_min": 673
+    },
+    {
+      "epoch": 1.2087087087087087,
+      "grad_norm": 0.49337094306984625,
+      "learning_rate": 3.936643781963024e-05,
+      "loss": 0.3508,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3622555136680603,
+      "step": 805,
+      "valid_targets_mean": 6322.6,
+      "valid_targets_min": 390
+    },
+    {
+      "epoch": 1.2162162162162162,
+      "grad_norm": 0.5183504295911457,
+      "learning_rate": 3.9347601924398475e-05,
+      "loss": 0.2833,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2589194178581238,
+      "step": 810,
+      "valid_targets_mean": 4678.0,
+      "valid_targets_min": 899
+    },
+    {
+      "epoch": 1.2237237237237237,
+      "grad_norm": 0.5342990678808706,
+      "learning_rate": 3.932849475873438e-05,
+      "loss": 0.3152,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2209431231021881,
+      "step": 815,
+      "valid_targets_mean": 4677.4,
+      "valid_targets_min": 815
+    },
+    {
+      "epoch": 1.2312312312312312,
+      "grad_norm": 0.4783352437662895,
+      "learning_rate": 3.9309116590537266e-05,
+      "loss": 0.2653,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2841101884841919,
+      "step": 820,
+      "valid_targets_mean": 4398.6,
+      "valid_targets_min": 874
+    },
+    {
+      "epoch": 1.2387387387387387,
+      "grad_norm": 0.5066260881718098,
+      "learning_rate": 3.9289467691506125e-05,
+      "loss": 0.3159,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2934754490852356,
+      "step": 825,
+      "valid_targets_mean": 4760.6,
+      "valid_targets_min": 670
+    },
+    {
+      "epoch": 1.2462462462462462,
+      "grad_norm": 0.6920092254384417,
+      "learning_rate": 3.926954833713585e-05,
+      "loss": 0.3049,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3148398697376251,
+      "step": 830,
+      "valid_targets_mean": 2882.6,
+      "valid_targets_min": 635
+    },
+    {
+      "epoch": 1.2537537537537538,
+      "grad_norm": 0.49074537646024496,
+      "learning_rate": 3.924935880671334e-05,
+      "loss": 0.3005,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2520464062690735,
+      "step": 835,
+      "valid_targets_mean": 5137.5,
+      "valid_targets_min": 1027
+    },
+    {
+      "epoch": 1.2612612612612613,
+      "grad_norm": 0.6626501857332863,
+      "learning_rate": 3.922889938331361e-05,
+      "loss": 0.3003,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.38677817583084106,
+      "step": 840,
+      "valid_targets_mean": 3420.8,
+      "valid_targets_min": 627
+    },
+    {
+      "epoch": 1.2687687687687688,
+      "grad_norm": 0.5332552957049799,
+      "learning_rate": 3.920817035379581e-05,
+      "loss": 0.2804,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.30235785245895386,
+      "step": 845,
+      "valid_targets_mean": 4300.7,
+      "valid_targets_min": 645
+    },
+    {
+      "epoch": 1.2762762762762763,
+      "grad_norm": 0.5542167310126439,
+      "learning_rate": 3.9187172008799195e-05,
+      "loss": 0.3007,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3331342339515686,
+      "step": 850,
+      "valid_targets_mean": 4431.8,
+      "valid_targets_min": 382
+    },
+    {
+      "epoch": 1.2837837837837838,
+      "grad_norm": 0.47602297520673265,
+      "learning_rate": 3.9165904642739085e-05,
+      "loss": 0.2825,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24128374457359314,
+      "step": 855,
+      "valid_targets_mean": 5224.0,
+      "valid_targets_min": 565
+    },
+    {
+      "epoch": 1.2912912912912913,
+      "grad_norm": 0.4907394738004159,
+      "learning_rate": 3.914436855380269e-05,
+      "loss": 0.2907,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2902330458164215,
+      "step": 860,
+      "valid_targets_mean": 4872.4,
+      "valid_targets_min": 469
+    },
+    {
+      "epoch": 1.2987987987987988,
+      "grad_norm": 0.5780103792011559,
+      "learning_rate": 3.9122564043944956e-05,
+      "loss": 0.3022,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.324124276638031,
+      "step": 865,
+      "valid_targets_mean": 4095.2,
+      "valid_targets_min": 836
+    },
+    {
+      "epoch": 1.3063063063063063,
+      "grad_norm": 1.4238637814086352,
+      "learning_rate": 3.910049141888433e-05,
+      "loss": 0.3061,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3050238788127899,
+      "step": 870,
+      "valid_targets_mean": 5166.2,
+      "valid_targets_min": 716
+    },
+    {
+      "epoch": 1.3138138138138138,
+      "grad_norm": 0.49302688153022367,
+      "learning_rate": 3.907815098809849e-05,
+      "loss": 0.268,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2744116187095642,
+      "step": 875,
+      "valid_targets_mean": 5715.9,
+      "valid_targets_min": 491
+    },
+    {
+      "epoch": 1.3213213213213213,
+      "grad_norm": 0.6362296442287568,
+      "learning_rate": 3.9055543064819945e-05,
+      "loss": 0.3095,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28804486989974976,
+      "step": 880,
+      "valid_targets_mean": 3622.8,
+      "valid_targets_min": 642
+    },
+    {
+      "epoch": 1.3288288288288288,
+      "grad_norm": 0.627623374325793,
+      "learning_rate": 3.9032667966031704e-05,
+      "loss": 0.2754,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3453376293182373,
+      "step": 885,
+      "valid_targets_mean": 3533.0,
+      "valid_targets_min": 600
+    },
+    {
+      "epoch": 1.3363363363363363,
+      "grad_norm": 0.6493174026676456,
+      "learning_rate": 3.9009526012462834e-05,
+      "loss": 0.292,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3329022228717804,
+      "step": 890,
+      "valid_targets_mean": 4443.5,
+      "valid_targets_min": 699
+    },
+    {
+      "epoch": 1.3438438438438438,
+      "grad_norm": 0.5328791757099103,
+      "learning_rate": 3.89861175285839e-05,
+      "loss": 0.2754,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2938082218170166,
+      "step": 895,
+      "valid_targets_mean": 4643.6,
+      "valid_targets_min": 674
+    },
+    {
+      "epoch": 1.3513513513513513,
+      "grad_norm": 0.5404971152718483,
+      "learning_rate": 3.89624428426025e-05,
+      "loss": 0.3149,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3295516073703766,
+      "step": 900,
+      "valid_targets_mean": 5400.6,
+      "valid_targets_min": 431
+    },
+    {
+      "epoch": 1.3588588588588588,
+      "grad_norm": 0.5148482135727713,
+      "learning_rate": 3.8938502286458586e-05,
+      "loss": 0.2965,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2672591209411621,
+      "step": 905,
+      "valid_targets_mean": 4434.6,
+      "valid_targets_min": 414
+    },
+    {
+      "epoch": 1.3663663663663663,
+      "grad_norm": 0.46940700753823117,
+      "learning_rate": 3.891429619581986e-05,
+      "loss": 0.2612,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25686946511268616,
+      "step": 910,
+      "valid_targets_mean": 5001.9,
+      "valid_targets_min": 750
+    },
+    {
+      "epoch": 1.3738738738738738,
+      "grad_norm": 0.5588622944159107,
+      "learning_rate": 3.888982491007705e-05,
+      "loss": 0.3063,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.37934252619743347,
+      "step": 915,
+      "valid_targets_mean": 4426.3,
+      "valid_targets_min": 480
+    },
+    {
+      "epoch": 1.3813813813813813,
+      "grad_norm": 0.5032151155279457,
+      "learning_rate": 3.886508877233914e-05,
+      "loss": 0.3002,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24294015765190125,
+      "step": 920,
+      "valid_targets_mean": 4203.2,
+      "valid_targets_min": 537
+    },
+    {
+      "epoch": 1.3888888888888888,
+      "grad_norm": 0.6375386060346272,
+      "learning_rate": 3.8840088129428594e-05,
+      "loss": 0.2822,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2843095660209656,
+      "step": 925,
+      "valid_targets_mean": 2762.1,
+      "valid_targets_min": 628
+    },
+    {
+      "epoch": 1.3963963963963963,
+      "grad_norm": 0.42954687639507877,
+      "learning_rate": 3.881482333187647e-05,
+      "loss": 0.2934,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24300193786621094,
+      "step": 930,
+      "valid_targets_mean": 6139.7,
+      "valid_targets_min": 1420
+    },
+    {
+      "epoch": 1.4039039039039038,
+      "grad_norm": 0.5288152668995841,
+      "learning_rate": 3.8789294733917486e-05,
+      "loss": 0.2809,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27250343561172485,
+      "step": 935,
+      "valid_targets_mean": 5227.9,
+      "valid_targets_min": 601
+    },
+    {
+      "epoch": 1.4114114114114114,
+      "grad_norm": 0.4018203487416013,
+      "learning_rate": 3.876350269348509e-05,
+      "loss": 0.2843,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2769063115119934,
+      "step": 940,
+      "valid_targets_mean": 7001.6,
+      "valid_targets_min": 791
+    },
+    {
+      "epoch": 1.4189189189189189,
+      "grad_norm": 0.4733738366398111,
+      "learning_rate": 3.873744757220642e-05,
+      "loss": 0.3156,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2413673847913742,
+      "step": 945,
+      "valid_targets_mean": 5955.6,
+      "valid_targets_min": 723
+    },
+    {
+      "epoch": 1.4264264264264264,
+      "grad_norm": 0.40178048986196624,
+      "learning_rate": 3.871112973539725e-05,
+      "loss": 0.2818,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22361788153648376,
+      "step": 950,
+      "valid_targets_mean": 5932.0,
+      "valid_targets_min": 684
+    },
+    {
+      "epoch": 1.4339339339339339,
+      "grad_norm": 0.5858661329948348,
+      "learning_rate": 3.868454955205685e-05,
+      "loss": 0.2967,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26762083172798157,
+      "step": 955,
+      "valid_targets_mean": 4612.1,
+      "valid_targets_min": 638
+    },
+    {
+      "epoch": 1.4414414414414414,
+      "grad_norm": 0.59370232296713,
+      "learning_rate": 3.86577073948628e-05,
+      "loss": 0.3087,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.245170459151268,
+      "step": 960,
+      "valid_targets_mean": 4544.8,
+      "valid_targets_min": 700
+    },
+    {
+      "epoch": 1.4489489489489489,
+      "grad_norm": 0.5776834809155071,
+      "learning_rate": 3.8630603640165815e-05,
+      "loss": 0.2795,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3211957812309265,
+      "step": 965,
+      "valid_targets_mean": 3560.0,
+      "valid_targets_min": 462
+    },
+    {
+      "epoch": 1.4564564564564564,
+      "grad_norm": 0.5619424459014478,
+      "learning_rate": 3.860323866798443e-05,
+      "loss": 0.2582,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21985271573066711,
+      "step": 970,
+      "valid_targets_mean": 4147.0,
+      "valid_targets_min": 964
+    },
+    {
+      "epoch": 1.4639639639639639,
+      "grad_norm": 0.5783533174246083,
+      "learning_rate": 3.857561286199968e-05,
+      "loss": 0.2969,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.29089707136154175,
+      "step": 975,
+      "valid_targets_mean": 3580.6,
+      "valid_targets_min": 489
+    },
+    {
+      "epoch": 1.4714714714714714,
+      "grad_norm": 0.4955776875207588,
+      "learning_rate": 3.85477266095497e-05,
+      "loss": 0.2825,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2684110999107361,
+      "step": 980,
+      "valid_targets_mean": 4121.5,
+      "valid_targets_min": 565
+    },
+    {
+      "epoch": 1.478978978978979,
+      "grad_norm": 0.5707278559095864,
+      "learning_rate": 3.851958030162433e-05,
+      "loss": 0.274,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3071454167366028,
+      "step": 985,
+      "valid_targets_mean": 4583.5,
+      "valid_targets_min": 711
+    },
+    {
+      "epoch": 1.4864864864864864,
+      "grad_norm": 0.5349960494922307,
+      "learning_rate": 3.8491174332859625e-05,
+      "loss": 0.2644,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2876233160495758,
+      "step": 990,
+      "valid_targets_mean": 3913.2,
+      "valid_targets_min": 652
+    },
+    {
+      "epoch": 1.493993993993994,
+      "grad_norm": 0.7242433186973595,
+      "learning_rate": 3.84625091015323e-05,
+      "loss": 0.2832,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2717621624469757,
+      "step": 995,
+      "valid_targets_mean": 3196.1,
+      "valid_targets_min": 613
+    },
+    {
+      "epoch": 1.5015015015015014,
+      "grad_norm": 0.559821205756259,
+      "learning_rate": 3.843358500955416e-05,
+      "loss": 0.277,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27973076701164246,
+      "step": 1000,
+      "valid_targets_mean": 3837.9,
+      "valid_targets_min": 875
+    },
+    {
+      "epoch": 1.509009009009009,
+      "grad_norm": 0.5377924424199211,
+      "learning_rate": 3.840440246246648e-05,
+      "loss": 0.2818,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2526509165763855,
+      "step": 1005,
+      "valid_targets_mean": 3364.9,
+      "valid_targets_min": 732
+    },
+    {
+      "epoch": 1.5165165165165164,
+      "grad_norm": 0.6037241854694765,
+      "learning_rate": 3.837496186943428e-05,
+      "loss": 0.2838,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3358380198478699,
+      "step": 1010,
+      "valid_targets_mean": 4278.9,
+      "valid_targets_min": 638
+    },
+    {
+      "epoch": 1.524024024024024,
+      "grad_norm": 0.4711675437570829,
+      "learning_rate": 3.8345263643240634e-05,
+      "loss": 0.2535,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26968130469322205,
+      "step": 1015,
+      "valid_targets_mean": 5544.2,
+      "valid_targets_min": 648
+    },
+    {
+      "epoch": 1.5315315315315314,
+      "grad_norm": 0.5958428128963504,
+      "learning_rate": 3.831530820028084e-05,
+      "loss": 0.3192,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.32176512479782104,
+      "step": 1020,
+      "valid_targets_mean": 3855.1,
+      "valid_targets_min": 674
+    },
+    {
+      "epoch": 1.539039039039039,
+      "grad_norm": 0.4865056298850766,
+      "learning_rate": 3.828509596055662e-05,
+      "loss": 0.2712,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2525503635406494,
+      "step": 1025,
+      "valid_targets_mean": 6991.5,
+      "valid_targets_min": 567
+    },
+    {
+      "epoch": 1.5465465465465464,
+      "grad_norm": 0.5566299314673824,
+      "learning_rate": 3.82546273476702e-05,
+      "loss": 0.3577,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2959470748901367,
+      "step": 1030,
+      "valid_targets_mean": 4056.7,
+      "valid_targets_min": 528
+    },
+    {
+      "epoch": 1.554054054054054,
+      "grad_norm": 0.6266122143638768,
+      "learning_rate": 3.822390278881839e-05,
+      "loss": 0.2924,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.31994956731796265,
+      "step": 1035,
+      "valid_targets_mean": 3313.5,
+      "valid_targets_min": 870
+    },
+    {
+      "epoch": 1.5615615615615615,
+      "grad_norm": 0.5379754675133966,
+      "learning_rate": 3.819292271478659e-05,
+      "loss": 0.2853,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26793667674064636,
+      "step": 1040,
+      "valid_targets_mean": 3280.4,
+      "valid_targets_min": 540
+    },
+    {
+      "epoch": 1.569069069069069,
+      "grad_norm": 0.42701565978810035,
+      "learning_rate": 3.816168755994274e-05,
+      "loss": 0.2752,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21840263903141022,
+      "step": 1045,
+      "valid_targets_mean": 4529.6,
+      "valid_targets_min": 904
+    },
+    {
+      "epoch": 1.5765765765765765,
+      "grad_norm": 0.5000336725002723,
+      "learning_rate": 3.813019776223124e-05,
+      "loss": 0.286,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2662447690963745,
+      "step": 1050,
+      "valid_targets_mean": 4867.2,
+      "valid_targets_min": 988
+    },
+    {
+      "epoch": 1.584084084084084,
+      "grad_norm": 0.5648658240320987,
+      "learning_rate": 3.809845376316681e-05,
+      "loss": 0.3156,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.29331260919570923,
+      "step": 1055,
+      "valid_targets_mean": 4341.2,
+      "valid_targets_min": 328
+    },
+    {
+      "epoch": 1.5915915915915915,
+      "grad_norm": 0.5284383529439995,
+      "learning_rate": 3.8066456007828304e-05,
+      "loss": 0.2835,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23197105526924133,
+      "step": 1060,
+      "valid_targets_mean": 3968.8,
+      "valid_targets_min": 389
+    },
+    {
+      "epoch": 1.599099099099099,
+      "grad_norm": 0.5216262376393843,
+      "learning_rate": 3.803420494485244e-05,
+      "loss": 0.273,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2597469687461853,
+      "step": 1065,
+      "valid_targets_mean": 4115.9,
+      "valid_targets_min": 579
+    },
+    {
+      "epoch": 1.6066066066066065,
+      "grad_norm": 0.4939695307516829,
+      "learning_rate": 3.800170102642757e-05,
+      "loss": 0.2835,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23144802451133728,
+      "step": 1070,
+      "valid_targets_mean": 3866.6,
+      "valid_targets_min": 486
+    },
+    {
+      "epoch": 1.614114114114114,
+      "grad_norm": 0.5829421644769698,
+      "learning_rate": 3.796894470828726e-05,
+      "loss": 0.267,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.222642183303833,
+      "step": 1075,
+      "valid_targets_mean": 4570.8,
+      "valid_targets_min": 844
+    },
+    {
+      "epoch": 1.6216216216216215,
+      "grad_norm": 0.5335734411442947,
+      "learning_rate": 3.793593644970397e-05,
+      "loss": 0.2881,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2972537577152252,
+      "step": 1080,
+      "valid_targets_mean": 4192.4,
+      "valid_targets_min": 648
+    },
+    {
+      "epoch": 1.629129129129129,
+      "grad_norm": 0.7509523435082093,
+      "learning_rate": 3.7902676713482584e-05,
+      "loss": 0.2991,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3354663848876953,
+      "step": 1085,
+      "valid_targets_mean": 3534.2,
+      "valid_targets_min": 875
+    },
+    {
+      "epoch": 1.6366366366366365,
+      "grad_norm": 0.45680598867761874,
+      "learning_rate": 3.786916596595391e-05,
+      "loss": 0.2835,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2275020182132721,
+      "step": 1090,
+      "valid_targets_mean": 4664.6,
+      "valid_targets_min": 768
+    },
+    {
+      "epoch": 1.644144144144144,
+      "grad_norm": 0.614422278972368,
+      "learning_rate": 3.783540467696817e-05,
+      "loss": 0.2845,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28407660126686096,
+      "step": 1095,
+      "valid_targets_mean": 3863.8,
+      "valid_targets_min": 409
+    },
+    {
+      "epoch": 1.6516516516516515,
+      "grad_norm": 0.5074836388259119,
+      "learning_rate": 3.78013933198884e-05,
+      "loss": 0.2611,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23754659295082092,
+      "step": 1100,
+      "valid_targets_mean": 3914.6,
+      "valid_targets_min": 1027
+    },
+    {
+      "epoch": 1.659159159159159,
+      "grad_norm": 0.45212979425084815,
+      "learning_rate": 3.776713237158379e-05,
+      "loss": 0.2728,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2909110188484192,
+      "step": 1105,
+      "valid_targets_mean": 5534.1,
+      "valid_targets_min": 655
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.6316240705989837,
+      "learning_rate": 3.773262231242305e-05,
+      "loss": 0.3053,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2717238664627075,
+      "step": 1110,
+      "valid_targets_mean": 3312.1,
+      "valid_targets_min": 367
+    },
+    {
+      "epoch": 1.674174174174174,
+      "grad_norm": 0.5332254751800973,
+      "learning_rate": 3.7697863626267615e-05,
+      "loss": 0.3138,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2853066623210907,
+      "step": 1115,
+      "valid_targets_mean": 3955.4,
+      "valid_targets_min": 731
+    },
+    {
+      "epoch": 1.6816816816816815,
+      "grad_norm": 0.6436721167931566,
+      "learning_rate": 3.7662856800464907e-05,
+      "loss": 0.2876,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.33219900727272034,
+      "step": 1120,
+      "valid_targets_mean": 4150.2,
+      "valid_targets_min": 402
+    },
+    {
+      "epoch": 1.689189189189189,
+      "grad_norm": 0.41517090061996553,
+      "learning_rate": 3.762760232584148e-05,
+      "loss": 0.2544,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2333643138408661,
+      "step": 1125,
+      "valid_targets_mean": 5115.1,
+      "valid_targets_min": 841
+    },
+    {
+      "epoch": 1.6966966966966965,
+      "grad_norm": 0.5176992190887993,
+      "learning_rate": 3.7592100696696134e-05,
+      "loss": 0.2732,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2570965886116028,
+      "step": 1130,
+      "valid_targets_mean": 3668.6,
+      "valid_targets_min": 656
+    },
+    {
+      "epoch": 1.704204204204204,
+      "grad_norm": 0.511279913702578,
+      "learning_rate": 3.755635241079301e-05,
+      "loss": 0.2719,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22817838191986084,
+      "step": 1135,
+      "valid_targets_mean": 3872.6,
+      "valid_targets_min": 475
+    },
+    {
+      "epoch": 1.7117117117117115,
+      "grad_norm": 0.6164706177166612,
+      "learning_rate": 3.7520357969354584e-05,
+      "loss": 0.2907,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3176250457763672,
+      "step": 1140,
+      "valid_targets_mean": 3526.4,
+      "valid_targets_min": 693
+    },
+    {
+      "epoch": 1.7192192192192193,
+      "grad_norm": 0.4109077325218178,
+      "learning_rate": 3.748411787705466e-05,
+      "loss": 0.3218,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2506415843963623,
+      "step": 1145,
+      "valid_targets_mean": 5783.5,
+      "valid_targets_min": 524
+    },
+    {
+      "epoch": 1.7267267267267268,
+      "grad_norm": 0.47340788265525957,
+      "learning_rate": 3.744763264201125e-05,
+      "loss": 0.259,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28487300872802734,
+      "step": 1150,
+      "valid_targets_mean": 5147.9,
+      "valid_targets_min": 311
+    },
+    {
+      "epoch": 1.7342342342342343,
+      "grad_norm": 0.6361764376110007,
+      "learning_rate": 3.741090277577953e-05,
+      "loss": 0.2747,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.33571434020996094,
+      "step": 1155,
+      "valid_targets_mean": 3613.9,
+      "valid_targets_min": 633
+    },
+    {
+      "epoch": 1.7417417417417418,
+      "grad_norm": 0.7263029802880552,
+      "learning_rate": 3.73739287933446e-05,
+      "loss": 0.3013,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.30988189578056335,
+      "step": 1160,
+      "valid_targets_mean": 2780.0,
+      "valid_targets_min": 388
+    },
+    {
+      "epoch": 1.7492492492492493,
+      "grad_norm": 0.5312249305901402,
+      "learning_rate": 3.733671121311426e-05,
+      "loss": 0.2691,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25757136940956116,
+      "step": 1165,
+      "valid_targets_mean": 4026.9,
+      "valid_targets_min": 605
+    },
+    {
+      "epoch": 1.7567567567567568,
+      "grad_norm": 0.5476708426420898,
+      "learning_rate": 3.729925055691182e-05,
+      "loss": 0.2875,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.30963361263275146,
+      "step": 1170,
+      "valid_targets_mean": 4677.7,
+      "valid_targets_min": 434
+    },
+    {
+      "epoch": 1.7642642642642643,
+      "grad_norm": 0.613322007834183,
+      "learning_rate": 3.7261547349968675e-05,
+      "loss": 0.2902,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.34018972516059875,
+      "step": 1175,
+      "valid_targets_mean": 3271.2,
+      "valid_targets_min": 407
+    },
+    {
+      "epoch": 1.7717717717717718,
+      "grad_norm": 0.6162258596946204,
+      "learning_rate": 3.722360212091703e-05,
+      "loss": 0.3093,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.32241833209991455,
+      "step": 1180,
+      "valid_targets_mean": 3718.6,
+      "valid_targets_min": 466
+    },
+    {
+      "epoch": 1.7792792792792793,
+      "grad_norm": 0.664703409023323,
+      "learning_rate": 3.7185415401782434e-05,
+      "loss": 0.2791,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.33108675479888916,
+      "step": 1185,
+      "valid_targets_mean": 5692.8,
+      "valid_targets_min": 702
+    },
+    {
+      "epoch": 1.7867867867867868,
+      "grad_norm": 0.5138638922358946,
+      "learning_rate": 3.7146987727976354e-05,
+      "loss": 0.2819,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.30338913202285767,
+      "step": 1190,
+      "valid_targets_mean": 5333.4,
+      "valid_targets_min": 752
+    },
+    {
+      "epoch": 1.7942942942942943,
+      "grad_norm": 0.4986654846084407,
+      "learning_rate": 3.7108319638288654e-05,
+      "loss": 0.2899,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2863334119319916,
+      "step": 1195,
+      "valid_targets_mean": 4185.2,
+      "valid_targets_min": 683
+    },
+    {
+      "epoch": 1.8018018018018018,
+      "grad_norm": 0.514824462976426,
+      "learning_rate": 3.706941167488002e-05,
+      "loss": 0.305,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3966107666492462,
+      "step": 1200,
+      "valid_targets_mean": 4997.3,
+      "valid_targets_min": 583
+    },
+    {
+      "epoch": 1.8093093093093093,
+      "grad_norm": 0.5002026479937768,
+      "learning_rate": 3.70302643832744e-05,
+      "loss": 0.3046,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3028540015220642,
+      "step": 1205,
+      "valid_targets_mean": 4384.4,
+      "valid_targets_min": 646
+    },
+    {
+      "epoch": 1.8168168168168168,
+      "grad_norm": 0.5733662063497278,
+      "learning_rate": 3.699087831235133e-05,
+      "loss": 0.271,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28234004974365234,
+      "step": 1210,
+      "valid_targets_mean": 3882.9,
+      "valid_targets_min": 781
+    },
+    {
+      "epoch": 1.8243243243243243,
+      "grad_norm": 0.49934816295444895,
+      "learning_rate": 3.6951254014338236e-05,
+      "loss": 0.2656,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27083009481430054,
+      "step": 1215,
+      "valid_targets_mean": 4725.4,
+      "valid_targets_min": 497
+    },
+    {
+      "epoch": 1.8318318318318318,
+      "grad_norm": 0.732103811366184,
+      "learning_rate": 3.69113920448027e-05,
+      "loss": 0.2718,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3054293394088745,
+      "step": 1220,
+      "valid_targets_mean": 2746.9,
+      "valid_targets_min": 478
+    },
+    {
+      "epoch": 1.8393393393393394,
+      "grad_norm": 0.6337417424539031,
+      "learning_rate": 3.687129296264468e-05,
+      "loss": 0.2591,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26185324788093567,
+      "step": 1225,
+      "valid_targets_mean": 4788.4,
+      "valid_targets_min": 695
+    },
+    {
+      "epoch": 1.8468468468468469,
+      "grad_norm": 0.5621683788362308,
+      "learning_rate": 3.683095733008864e-05,
+      "loss": 0.2693,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2606724202632904,
+      "step": 1230,
+      "valid_targets_mean": 3237.6,
+      "valid_targets_min": 534
+    },
+    {
+      "epoch": 1.8543543543543544,
+      "grad_norm": 0.5288214426455534,
+      "learning_rate": 3.6790385712675705e-05,
+      "loss": 0.2837,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3149155378341675,
+      "step": 1235,
+      "valid_targets_mean": 3973.2,
+      "valid_targets_min": 572
+    },
+    {
+      "epoch": 1.8618618618618619,
+      "grad_norm": 0.4813442560130786,
+      "learning_rate": 3.6749578679255715e-05,
+      "loss": 0.2996,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2811282277107239,
+      "step": 1240,
+      "valid_targets_mean": 5313.8,
+      "valid_targets_min": 420
+    },
+    {
+      "epoch": 1.8693693693693694,
+      "grad_norm": 0.6170835431330234,
+      "learning_rate": 3.6708536801979246e-05,
+      "loss": 0.2905,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.32795318961143494,
+      "step": 1245,
+      "valid_targets_mean": 3692.1,
+      "valid_targets_min": 724
+    },
+    {
+      "epoch": 1.8768768768768769,
+      "grad_norm": 0.5375710532679772,
+      "learning_rate": 3.666726065628959e-05,
+      "loss": 0.2688,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23410853743553162,
+      "step": 1250,
+      "valid_targets_mean": 3626.6,
+      "valid_targets_min": 663
+    },
+    {
+      "epoch": 1.8843843843843844,
+      "grad_norm": 0.5400535513286914,
+      "learning_rate": 3.6625750820914694e-05,
+      "loss": 0.2569,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3105897903442383,
+      "step": 1255,
+      "valid_targets_mean": 3530.0,
+      "valid_targets_min": 400
+    },
+    {
+      "epoch": 1.8918918918918919,
+      "grad_norm": 0.7192704328062035,
+      "learning_rate": 3.658400787785903e-05,
+      "loss": 0.2711,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.31259477138519287,
+      "step": 1260,
+      "valid_targets_mean": 2492.8,
+      "valid_targets_min": 367
+    },
+    {
+      "epoch": 1.8993993993993994,
+      "grad_norm": 0.5689420833987145,
+      "learning_rate": 3.654203241239546e-05,
+      "loss": 0.2752,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2669737935066223,
+      "step": 1265,
+      "valid_targets_mean": 5125.1,
+      "valid_targets_min": 766
+    },
+    {
+      "epoch": 1.906906906906907,
+      "grad_norm": 0.6991606704570577,
+      "learning_rate": 3.649982501305701e-05,
+      "loss": 0.2681,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3605828285217285,
+      "step": 1270,
+      "valid_targets_mean": 2681.3,
+      "valid_targets_min": 603
+    },
+    {
+      "epoch": 1.9144144144144144,
+      "grad_norm": 0.5136360235806671,
+      "learning_rate": 3.6457386271628615e-05,
+      "loss": 0.2815,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2559846043586731,
+      "step": 1275,
+      "valid_targets_mean": 4230.4,
+      "valid_targets_min": 467
+    },
+    {
+      "epoch": 1.921921921921922,
+      "grad_norm": 0.5755425636039267,
+      "learning_rate": 3.641471678313883e-05,
+      "loss": 0.2629,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.32143378257751465,
+      "step": 1280,
+      "valid_targets_mean": 4531.5,
+      "valid_targets_min": 723
+    },
+    {
+      "epoch": 1.9294294294294294,
+      "grad_norm": 0.5172980766220905,
+      "learning_rate": 3.6371817145851514e-05,
+      "loss": 0.2902,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3124349117279053,
+      "step": 1285,
+      "valid_targets_mean": 5154.6,
+      "valid_targets_min": 690
+    },
+    {
+      "epoch": 1.936936936936937,
+      "grad_norm": 0.5579160547269432,
+      "learning_rate": 3.6328687961257374e-05,
+      "loss": 0.2863,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23261310160160065,
+      "step": 1290,
+      "valid_targets_mean": 3315.2,
+      "valid_targets_min": 432
+    },
+    {
+      "epoch": 1.9444444444444444,
+      "grad_norm": 0.471485760499144,
+      "learning_rate": 3.62853298340656e-05,
+      "loss": 0.2875,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23489312827587128,
+      "step": 1295,
+      "valid_targets_mean": 4736.7,
+      "valid_targets_min": 689
+    },
+    {
+      "epoch": 1.951951951951952,
+      "grad_norm": 0.45892772148849675,
+      "learning_rate": 3.624174337219536e-05,
+      "loss": 0.2919,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23488348722457886,
+      "step": 1300,
+      "valid_targets_mean": 4925.0,
+      "valid_targets_min": 602
+    },
+    {
+      "epoch": 1.9594594594594594,
+      "grad_norm": 0.6569352257148999,
+      "learning_rate": 3.6197929186767255e-05,
+      "loss": 0.2827,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26401716470718384,
+      "step": 1305,
+      "valid_targets_mean": 3648.6,
+      "valid_targets_min": 358
+    },
+    {
+      "epoch": 1.966966966966967,
+      "grad_norm": 0.47233655136122743,
+      "learning_rate": 3.6153887892094786e-05,
+      "loss": 0.292,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2689140737056732,
+      "step": 1310,
+      "valid_targets_mean": 4251.4,
+      "valid_targets_min": 521
+    },
+    {
+      "epoch": 1.9744744744744744,
+      "grad_norm": 0.4937344312534394,
+      "learning_rate": 3.6109620105675734e-05,
+      "loss": 0.2502,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2564142346382141,
+      "step": 1315,
+      "valid_targets_mean": 4050.1,
+      "valid_targets_min": 846
+    },
+    {
+      "epoch": 1.981981981981982,
+      "grad_norm": 0.5040985465764987,
+      "learning_rate": 3.606512644818347e-05,
+      "loss": 0.2959,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27598607540130615,
+      "step": 1320,
+      "valid_targets_mean": 5606.7,
+      "valid_targets_min": 561
+    },
+    {
+      "epoch": 1.9894894894894894,
+      "grad_norm": 0.503656514625708,
+      "learning_rate": 3.6020407543458295e-05,
+      "loss": 0.2656,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24646782875061035,
+      "step": 1325,
+      "valid_targets_mean": 4084.1,
+      "valid_targets_min": 542
+    },
+    {
+      "epoch": 1.996996996996997,
+      "grad_norm": 0.48663621599509704,
+      "learning_rate": 3.597546401849866e-05,
+      "loss": 0.2877,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3501949608325958,
+      "step": 1330,
+      "valid_targets_mean": 6161.8,
+      "valid_targets_min": 829
+    },
+    {
+      "epoch": 2.0045045045045047,
+      "grad_norm": 0.4886145150612323,
+      "learning_rate": 3.593029650345242e-05,
+      "loss": 0.248,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2267410159111023,
+      "step": 1335,
+      "valid_targets_mean": 4201.8,
+      "valid_targets_min": 585
+    },
+    {
+      "epoch": 2.012012012012012,
+      "grad_norm": 0.918223183885553,
+      "learning_rate": 3.588490563160793e-05,
+      "loss": 0.263,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20230059325695038,
+      "step": 1340,
+      "valid_targets_mean": 3759.8,
+      "valid_targets_min": 668
+    },
+    {
+      "epoch": 2.0195195195195197,
+      "grad_norm": 0.41635246230968925,
+      "learning_rate": 3.583929203938523e-05,
+      "loss": 0.2647,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20971567928791046,
+      "step": 1345,
+      "valid_targets_mean": 6513.2,
+      "valid_targets_min": 506
+    },
+    {
+      "epoch": 2.027027027027027,
+      "grad_norm": 0.5324224076323132,
+      "learning_rate": 3.5793456366327074e-05,
+      "loss": 0.2579,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2577601969242096,
+      "step": 1350,
+      "valid_targets_mean": 4199.0,
+      "valid_targets_min": 611
+    },
+    {
+      "epoch": 2.0345345345345347,
+      "grad_norm": 0.666399334701293,
+      "learning_rate": 3.5747399255090016e-05,
+      "loss": 0.2532,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25944650173187256,
+      "step": 1355,
+      "valid_targets_mean": 3050.4,
+      "valid_targets_min": 513
+    },
+    {
+      "epoch": 2.042042042042042,
+      "grad_norm": 0.5560230204936001,
+      "learning_rate": 3.5701121351435346e-05,
+      "loss": 0.2565,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.30907580256462097,
+      "step": 1360,
+      "valid_targets_mean": 4342.1,
+      "valid_targets_min": 483
+    },
+    {
+      "epoch": 2.0495495495495497,
+      "grad_norm": 0.5093455715924559,
+      "learning_rate": 3.565462330422006e-05,
+      "loss": 0.2366,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23613181710243225,
+      "step": 1365,
+      "valid_targets_mean": 4160.1,
+      "valid_targets_min": 624
+    },
+    {
+      "epoch": 2.057057057057057,
+      "grad_norm": 0.5143252526017971,
+      "learning_rate": 3.560790576538777e-05,
+      "loss": 0.2441,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28200429677963257,
+      "step": 1370,
+      "valid_targets_mean": 5243.5,
+      "valid_targets_min": 529
+    },
+    {
+      "epoch": 2.0645645645645647,
+      "grad_norm": 0.6420496487933931,
+      "learning_rate": 3.556096938995953e-05,
+      "loss": 0.2644,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3083629906177521,
+      "step": 1375,
+      "valid_targets_mean": 3444.8,
+      "valid_targets_min": 474
+    },
+    {
+      "epoch": 2.0720720720720722,
+      "grad_norm": 0.6041025909633767,
+      "learning_rate": 3.551381483602472e-05,
+      "loss": 0.3043,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.34277862310409546,
+      "step": 1380,
+      "valid_targets_mean": 5168.2,
+      "valid_targets_min": 656
+    },
+    {
+      "epoch": 2.0795795795795797,
+      "grad_norm": 0.5065151180208464,
+      "learning_rate": 3.546644276473174e-05,
+      "loss": 0.2703,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22405189275741577,
+      "step": 1385,
+      "valid_targets_mean": 4451.1,
+      "valid_targets_min": 653
+    },
+    {
+      "epoch": 2.0870870870870872,
+      "grad_norm": 0.4653600091882295,
+      "learning_rate": 3.541885384027877e-05,
+      "loss": 0.2867,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23494037985801697,
+      "step": 1390,
+      "valid_targets_mean": 4773.1,
+      "valid_targets_min": 515
+    },
+    {
+      "epoch": 2.0945945945945947,
+      "grad_norm": 0.4737552542155004,
+      "learning_rate": 3.53710487299045e-05,
+      "loss": 0.2301,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20185862481594086,
+      "step": 1395,
+      "valid_targets_mean": 4372.3,
+      "valid_targets_min": 656
+    },
+    {
+      "epoch": 2.1021021021021022,
+      "grad_norm": 0.618244462622168,
+      "learning_rate": 3.5323028103878695e-05,
+      "loss": 0.2647,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28137439489364624,
+      "step": 1400,
+      "valid_targets_mean": 3066.6,
+      "valid_targets_min": 558
+    },
+    {
+      "epoch": 2.1096096096096097,
+      "grad_norm": 0.6276854706539091,
+      "learning_rate": 3.527479263549287e-05,
+      "loss": 0.2567,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24222517013549805,
+      "step": 1405,
+      "valid_targets_mean": 3623.4,
+      "valid_targets_min": 562
+    },
+    {
+      "epoch": 2.1171171171171173,
+      "grad_norm": 0.5119888962756797,
+      "learning_rate": 3.52263430010508e-05,
+      "loss": 0.2456,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24120667576789856,
+      "step": 1410,
+      "valid_targets_mean": 4184.2,
+      "valid_targets_min": 571
+    },
+    {
+      "epoch": 2.1246246246246248,
+      "grad_norm": 0.40193152670779986,
+      "learning_rate": 3.517767987985906e-05,
+      "loss": 0.2543,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21013514697551727,
+      "step": 1415,
+      "valid_targets_mean": 7112.9,
+      "valid_targets_min": 958
+    },
+    {
+      "epoch": 2.1321321321321323,
+      "grad_norm": 0.6042469303834141,
+      "learning_rate": 3.5128803954217516e-05,
+      "loss": 0.245,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24615558981895447,
+      "step": 1420,
+      "valid_targets_mean": 3546.8,
+      "valid_targets_min": 647
+    },
+    {
+      "epoch": 2.1396396396396398,
+      "grad_norm": 0.517598650535331,
+      "learning_rate": 3.507971590940972e-05,
+      "loss": 0.2361,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2518162727355957,
+      "step": 1425,
+      "valid_targets_mean": 4095.2,
+      "valid_targets_min": 700
+    },
+    {
+      "epoch": 2.1471471471471473,
+      "grad_norm": 0.5314103976572915,
+      "learning_rate": 3.503041643369332e-05,
+      "loss": 0.2657,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.29391518235206604,
+      "step": 1430,
+      "valid_targets_mean": 4662.9,
+      "valid_targets_min": 673
+    },
+    {
+      "epoch": 2.1546546546546548,
+      "grad_norm": 0.7994874645886633,
+      "learning_rate": 3.4980906218290425e-05,
+      "loss": 0.2778,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24668291211128235,
+      "step": 1435,
+      "valid_targets_mean": 4393.5,
+      "valid_targets_min": 639
+    },
+    {
+      "epoch": 2.1621621621621623,
+      "grad_norm": 0.5508650913511237,
+      "learning_rate": 3.493118595737789e-05,
+      "loss": 0.2655,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26784124970436096,
+      "step": 1440,
+      "valid_targets_mean": 3928.2,
+      "valid_targets_min": 482
+    },
+    {
+      "epoch": 2.16966966966967,
+      "grad_norm": 0.48256368285307105,
+      "learning_rate": 3.4881256348077606e-05,
+      "loss": 0.2493,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2078143209218979,
+      "step": 1445,
+      "valid_targets_mean": 4895.9,
+      "valid_targets_min": 605
+    },
+    {
+      "epoch": 2.1771771771771773,
+      "grad_norm": 0.5518092824723564,
+      "learning_rate": 3.4831118090446695e-05,
+      "loss": 0.2614,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27934592962265015,
+      "step": 1450,
+      "valid_targets_mean": 4175.9,
+      "valid_targets_min": 557
+    },
+    {
+      "epoch": 2.184684684684685,
+      "grad_norm": 0.6510442042007786,
+      "learning_rate": 3.478077188746774e-05,
+      "loss": 0.2825,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25130707025527954,
+      "step": 1455,
+      "valid_targets_mean": 2940.2,
+      "valid_targets_min": 746
+    },
+    {
+      "epoch": 2.1921921921921923,
+      "grad_norm": 0.6653821572090174,
+      "learning_rate": 3.473021844503885e-05,
+      "loss": 0.2915,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.29144030809402466,
+      "step": 1460,
+      "valid_targets_mean": 3176.5,
+      "valid_targets_min": 466
+    },
+    {
+      "epoch": 2.1996996996997,
+      "grad_norm": 0.5835363387672817,
+      "learning_rate": 3.4679458471963876e-05,
+      "loss": 0.2609,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2668296694755554,
+      "step": 1465,
+      "valid_targets_mean": 3661.9,
+      "valid_targets_min": 657
+    },
+    {
+      "epoch": 2.2072072072072073,
+      "grad_norm": 0.473510630048058,
+      "learning_rate": 3.462849267994237e-05,
+      "loss": 0.2391,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2349366545677185,
+      "step": 1470,
+      "valid_targets_mean": 4610.1,
+      "valid_targets_min": 583
+    },
+    {
+      "epoch": 2.214714714714715,
+      "grad_norm": 0.5034796149297331,
+      "learning_rate": 3.457732178355966e-05,
+      "loss": 0.2786,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22843249142169952,
+      "step": 1475,
+      "valid_targets_mean": 4560.3,
+      "valid_targets_min": 870
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 0.6628701647407635,
+      "learning_rate": 3.4525946500276815e-05,
+      "loss": 0.2524,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.33600878715515137,
+      "step": 1480,
+      "valid_targets_mean": 4715.0,
+      "valid_targets_min": 409
+    },
+    {
+      "epoch": 2.22972972972973,
+      "grad_norm": 0.4438069409661387,
+      "learning_rate": 3.447436755042059e-05,
+      "loss": 0.2579,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23582454025745392,
+      "step": 1485,
+      "valid_targets_mean": 6454.2,
+      "valid_targets_min": 720
+    },
+    {
+      "epoch": 2.2372372372372373,
+      "grad_norm": 0.5589297614395898,
+      "learning_rate": 3.442258565717333e-05,
+      "loss": 0.2906,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3400152325630188,
+      "step": 1490,
+      "valid_targets_mean": 5422.4,
+      "valid_targets_min": 928
+    },
+    {
+      "epoch": 2.244744744744745,
+      "grad_norm": 0.5331121771902878,
+      "learning_rate": 3.437060154656281e-05,
+      "loss": 0.2911,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.31767523288726807,
+      "step": 1495,
+      "valid_targets_mean": 4458.2,
+      "valid_targets_min": 291
+    },
+    {
+      "epoch": 2.2522522522522523,
+      "grad_norm": 0.40712934057436334,
+      "learning_rate": 3.4318415947452085e-05,
+      "loss": 0.252,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28288692235946655,
+      "step": 1500,
+      "valid_targets_mean": 6532.0,
+      "valid_targets_min": 648
+    },
+    {
+      "epoch": 2.25975975975976,
+      "grad_norm": 0.45673386040732317,
+      "learning_rate": 3.4266029591529246e-05,
+      "loss": 0.2864,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2521343529224396,
+      "step": 1505,
+      "valid_targets_mean": 5607.8,
+      "valid_targets_min": 725
+    },
+    {
+      "epoch": 2.2672672672672673,
+      "grad_norm": 0.6013211253935284,
+      "learning_rate": 3.421344321329719e-05,
+      "loss": 0.2529,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28024786710739136,
+      "step": 1510,
+      "valid_targets_mean": 3546.8,
+      "valid_targets_min": 515
+    },
+    {
+      "epoch": 2.274774774774775,
+      "grad_norm": 0.9210774515682121,
+      "learning_rate": 3.4160657550063286e-05,
+      "loss": 0.262,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2434246838092804,
+      "step": 1515,
+      "valid_targets_mean": 3957.6,
+      "valid_targets_min": 444
+    },
+    {
+      "epoch": 2.2822822822822824,
+      "grad_norm": 0.5034622925230656,
+      "learning_rate": 3.410767334192905e-05,
+      "loss": 0.2519,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2538553476333618,
+      "step": 1520,
+      "valid_targets_mean": 4692.2,
+      "valid_targets_min": 605
+    },
+    {
+      "epoch": 2.28978978978979,
+      "grad_norm": 0.6140827528489169,
+      "learning_rate": 3.40544913317798e-05,
+      "loss": 0.2476,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2580946087837219,
+      "step": 1525,
+      "valid_targets_mean": 3429.3,
+      "valid_targets_min": 549
+    },
+    {
+      "epoch": 2.2972972972972974,
+      "grad_norm": 0.4522744249612033,
+      "learning_rate": 3.400111226527417e-05,
+      "loss": 0.2326,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22649413347244263,
+      "step": 1530,
+      "valid_targets_mean": 5068.2,
+      "valid_targets_min": 731
+    },
+    {
+      "epoch": 2.304804804804805,
+      "grad_norm": 0.49640935192095975,
+      "learning_rate": 3.3947536890833744e-05,
+      "loss": 0.2507,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2515903115272522,
+      "step": 1535,
+      "valid_targets_mean": 4226.1,
+      "valid_targets_min": 726
+    },
+    {
+      "epoch": 2.3123123123123124,
+      "grad_norm": 1.016722378855143,
+      "learning_rate": 3.389376595963248e-05,
+      "loss": 0.2882,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.40263426303863525,
+      "step": 1540,
+      "valid_targets_mean": 3808.3,
+      "valid_targets_min": 756
+    },
+    {
+      "epoch": 2.31981981981982,
+      "grad_norm": 0.520047683970159,
+      "learning_rate": 3.3839800225586233e-05,
+      "loss": 0.2725,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24709469079971313,
+      "step": 1545,
+      "valid_targets_mean": 4321.6,
+      "valid_targets_min": 663
+    },
+    {
+      "epoch": 2.3273273273273274,
+      "grad_norm": 0.4955591390887456,
+      "learning_rate": 3.378564044534216e-05,
+      "loss": 0.2218,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24357634782791138,
+      "step": 1550,
+      "valid_targets_mean": 4470.6,
+      "valid_targets_min": 670
+    },
+    {
+      "epoch": 2.334834834834835,
+      "grad_norm": 0.5304638752162832,
+      "learning_rate": 3.373128737826812e-05,
+      "loss": 0.2625,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27869170904159546,
+      "step": 1555,
+      "valid_targets_mean": 3793.8,
+      "valid_targets_min": 252
+    },
+    {
+      "epoch": 2.3423423423423424,
+      "grad_norm": 0.6052044405674071,
+      "learning_rate": 3.367674178644201e-05,
+      "loss": 0.2578,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2955855429172516,
+      "step": 1560,
+      "valid_targets_mean": 3434.8,
+      "valid_targets_min": 662
+    },
+    {
+      "epoch": 2.34984984984985,
+      "grad_norm": 0.45641925043102494,
+      "learning_rate": 3.362200443464112e-05,
+      "loss": 0.2572,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25648972392082214,
+      "step": 1565,
+      "valid_targets_mean": 5664.2,
+      "valid_targets_min": 487
+    },
+    {
+      "epoch": 2.3573573573573574,
+      "grad_norm": 0.4104470454624804,
+      "learning_rate": 3.356707609033136e-05,
+      "loss": 0.2382,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1871146857738495,
+      "step": 1570,
+      "valid_targets_mean": 5469.4,
+      "valid_targets_min": 650
+    },
+    {
+      "epoch": 2.364864864864865,
+      "grad_norm": 0.5305663319530391,
+      "learning_rate": 3.351195752365654e-05,
+      "loss": 0.2546,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20778313279151917,
+      "step": 1575,
+      "valid_targets_mean": 3653.4,
+      "valid_targets_min": 548
+    },
+    {
+      "epoch": 2.3723723723723724,
+      "grad_norm": 0.952177731638365,
+      "learning_rate": 3.3456649507427537e-05,
+      "loss": 0.2785,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.30603039264678955,
+      "step": 1580,
+      "valid_targets_mean": 3391.8,
+      "valid_targets_min": 541
+    },
+    {
+      "epoch": 2.37987987987988,
+      "grad_norm": 0.625558644174225,
+      "learning_rate": 3.340115281711149e-05,
+      "loss": 0.2578,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2915250360965729,
+      "step": 1585,
+      "valid_targets_mean": 3354.9,
+      "valid_targets_min": 376
+    },
+    {
+      "epoch": 2.3873873873873874,
+      "grad_norm": 0.4383965531441897,
+      "learning_rate": 3.3345468230820924e-05,
+      "loss": 0.2545,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20519056916236877,
+      "step": 1590,
+      "valid_targets_mean": 5249.9,
+      "valid_targets_min": 576
+    },
+    {
+      "epoch": 2.394894894894895,
+      "grad_norm": 0.7263449171228276,
+      "learning_rate": 3.32895965293028e-05,
+      "loss": 0.2629,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24948550760746002,
+      "step": 1595,
+      "valid_targets_mean": 3813.8,
+      "valid_targets_min": 700
+    },
+    {
+      "epoch": 2.4024024024024024,
+      "grad_norm": 0.7704936140888641,
+      "learning_rate": 3.323353849592764e-05,
+      "loss": 0.27,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3838420808315277,
+      "step": 1600,
+      "valid_targets_mean": 4764.8,
+      "valid_targets_min": 930
+    },
+    {
+      "epoch": 2.40990990990991,
+      "grad_norm": 0.5262208670623167,
+      "learning_rate": 3.317729491667848e-05,
+      "loss": 0.2854,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26990848779678345,
+      "step": 1605,
+      "valid_targets_mean": 4314.4,
+      "valid_targets_min": 728
+    },
+    {
+      "epoch": 2.4174174174174174,
+      "grad_norm": 0.4905094743952616,
+      "learning_rate": 3.3120866580139856e-05,
+      "loss": 0.2678,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24697014689445496,
+      "step": 1610,
+      "valid_targets_mean": 5779.1,
+      "valid_targets_min": 863
+    },
+    {
+      "epoch": 2.424924924924925,
+      "grad_norm": 0.5795672522700869,
+      "learning_rate": 3.306425427748681e-05,
+      "loss": 0.2753,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3122454881668091,
+      "step": 1615,
+      "valid_targets_mean": 4678.4,
+      "valid_targets_min": 597
+    },
+    {
+      "epoch": 2.4324324324324325,
+      "grad_norm": 0.5127623781647497,
+      "learning_rate": 3.3007458802473705e-05,
+      "loss": 0.2418,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2780628502368927,
+      "step": 1620,
+      "valid_targets_mean": 5270.1,
+      "valid_targets_min": 408
+    },
+    {
+      "epoch": 2.43993993993994,
+      "grad_norm": 0.6000042814068866,
+      "learning_rate": 3.2950480951423166e-05,
+      "loss": 0.2492,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26539573073387146,
+      "step": 1625,
+      "valid_targets_mean": 2969.9,
+      "valid_targets_min": 292
+    },
+    {
+      "epoch": 2.4474474474474475,
+      "grad_norm": 0.5601820602459276,
+      "learning_rate": 3.289332152321489e-05,
+      "loss": 0.2744,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.34554967284202576,
+      "step": 1630,
+      "valid_targets_mean": 4328.4,
+      "valid_targets_min": 570
+    },
+    {
+      "epoch": 2.454954954954955,
+      "grad_norm": 0.5017335839364364,
+      "learning_rate": 3.2835981319274436e-05,
+      "loss": 0.3016,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.31424376368522644,
+      "step": 1635,
+      "valid_targets_mean": 4981.0,
+      "valid_targets_min": 513
+    },
+    {
+      "epoch": 2.4624624624624625,
+      "grad_norm": 0.5989444157971933,
+      "learning_rate": 3.277846114356201e-05,
+      "loss": 0.2812,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.29167452454566956,
+      "step": 1640,
+      "valid_targets_mean": 4280.9,
+      "valid_targets_min": 583
+    },
+    {
+      "epoch": 2.46996996996997,
+      "grad_norm": 0.4818252524686284,
+      "learning_rate": 3.272076180256117e-05,
+      "loss": 0.2503,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2451312243938446,
+      "step": 1645,
+      "valid_targets_mean": 5266.5,
+      "valid_targets_min": 456
+    },
+    {
+      "epoch": 2.4774774774774775,
+      "grad_norm": 0.5474083922971454,
+      "learning_rate": 3.2662884105267545e-05,
+      "loss": 0.2802,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28529974818229675,
+      "step": 1650,
+      "valid_targets_mean": 4656.7,
+      "valid_targets_min": 323
+    },
+    {
+      "epoch": 2.484984984984985,
+      "grad_norm": 0.3927761927981878,
+      "learning_rate": 3.260482886317745e-05,
+      "loss": 0.2586,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2354969084262848,
+      "step": 1655,
+      "valid_targets_mean": 6819.5,
+      "valid_targets_min": 769
+    },
+    {
+      "epoch": 2.4924924924924925,
+      "grad_norm": 0.5095139881585496,
+      "learning_rate": 3.254659689027656e-05,
+      "loss": 0.2822,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25443506240844727,
+      "step": 1660,
+      "valid_targets_mean": 4691.8,
+      "valid_targets_min": 754
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.48122470806496775,
+      "learning_rate": 3.2488189003028456e-05,
+      "loss": 0.2512,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25341033935546875,
+      "step": 1665,
+      "valid_targets_mean": 5473.4,
+      "valid_targets_min": 667
+    },
+    {
+      "epoch": 2.5075075075075075,
+      "grad_norm": 0.535101793490416,
+      "learning_rate": 3.24296060203632e-05,
+      "loss": 0.2725,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.29341810941696167,
+      "step": 1670,
+      "valid_targets_mean": 4545.0,
+      "valid_targets_min": 727
+    },
+    {
+      "epoch": 2.515015015015015,
+      "grad_norm": 0.5407510878775967,
+      "learning_rate": 3.2370848763665854e-05,
+      "loss": 0.2631,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.298563152551651,
+      "step": 1675,
+      "valid_targets_mean": 4343.1,
+      "valid_targets_min": 487
+    },
+    {
+      "epoch": 2.5225225225225225,
+      "grad_norm": 0.7328443730397846,
+      "learning_rate": 3.231191805676494e-05,
+      "loss": 0.2687,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24576136469841003,
+      "step": 1680,
+      "valid_targets_mean": 4246.7,
+      "valid_targets_min": 381
+    },
+    {
+      "epoch": 2.53003003003003,
+      "grad_norm": 0.4953608748768144,
+      "learning_rate": 3.22528147259209e-05,
+      "loss": 0.2894,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3240085542201996,
+      "step": 1685,
+      "valid_targets_mean": 5762.2,
+      "valid_targets_min": 734
+    },
+    {
+      "epoch": 2.5375375375375375,
+      "grad_norm": 0.6439990940182198,
+      "learning_rate": 3.2193539599814546e-05,
+      "loss": 0.2563,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2948322892189026,
+      "step": 1690,
+      "valid_targets_mean": 3211.7,
+      "valid_targets_min": 785
+    },
+    {
+      "epoch": 2.545045045045045,
+      "grad_norm": 0.5649997969615153,
+      "learning_rate": 3.213409350953539e-05,
+      "loss": 0.2763,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2883799374103546,
+      "step": 1695,
+      "valid_targets_mean": 3863.2,
+      "valid_targets_min": 655
+    },
+    {
+      "epoch": 2.5525525525525525,
+      "grad_norm": 0.6654178682497041,
+      "learning_rate": 3.207447728857e-05,
+      "loss": 0.2697,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2643584907054901,
+      "step": 1700,
+      "valid_targets_mean": 3119.7,
+      "valid_targets_min": 557
+    },
+    {
+      "epoch": 2.56006006006006,
+      "grad_norm": 0.6865303086852059,
+      "learning_rate": 3.201469177279036e-05,
+      "loss": 0.2489,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27196943759918213,
+      "step": 1705,
+      "valid_targets_mean": 2167.2,
+      "valid_targets_min": 570
+    },
+    {
+      "epoch": 2.5675675675675675,
+      "grad_norm": 0.5565838052443746,
+      "learning_rate": 3.195473780044208e-05,
+      "loss": 0.2622,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2557700276374817,
+      "step": 1710,
+      "valid_targets_mean": 3659.3,
+      "valid_targets_min": 446
+    },
+    {
+      "epoch": 2.575075075075075,
+      "grad_norm": 0.5197442676457078,
+      "learning_rate": 3.189461621213271e-05,
+      "loss": 0.2617,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2678113579750061,
+      "step": 1715,
+      "valid_targets_mean": 4663.4,
+      "valid_targets_min": 609
+    },
+    {
+      "epoch": 2.5825825825825826,
+      "grad_norm": 0.5166944613990696,
+      "learning_rate": 3.18343278508199e-05,
+      "loss": 0.2584,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24757936596870422,
+      "step": 1720,
+      "valid_targets_mean": 4511.7,
+      "valid_targets_min": 460
+    },
+    {
+      "epoch": 2.59009009009009,
+      "grad_norm": 0.4761985812024692,
+      "learning_rate": 3.177387356179962e-05,
+      "loss": 0.2292,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20293045043945312,
+      "step": 1725,
+      "valid_targets_mean": 4317.4,
+      "valid_targets_min": 477
+    },
+    {
+      "epoch": 2.5975975975975976,
+      "grad_norm": 0.5529954148422914,
+      "learning_rate": 3.1713254192694284e-05,
+      "loss": 0.2407,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24401406943798065,
+      "step": 1730,
+      "valid_targets_mean": 3834.1,
+      "valid_targets_min": 610
+    },
+    {
+      "epoch": 2.605105105105105,
+      "grad_norm": 0.4955364019562657,
+      "learning_rate": 3.1652470593440885e-05,
+      "loss": 0.2746,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22095078229904175,
+      "step": 1735,
+      "valid_targets_mean": 5501.8,
+      "valid_targets_min": 681
+    },
+    {
+      "epoch": 2.6126126126126126,
+      "grad_norm": 0.45641687982187235,
+      "learning_rate": 3.1591523616279055e-05,
+      "loss": 0.2507,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20338258147239685,
+      "step": 1740,
+      "valid_targets_mean": 4772.7,
+      "valid_targets_min": 700
+    },
+    {
+      "epoch": 2.62012012012012,
+      "grad_norm": 0.6226280659286444,
+      "learning_rate": 3.153041411573914e-05,
+      "loss": 0.2552,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3405519723892212,
+      "step": 1745,
+      "valid_targets_mean": 4217.1,
+      "valid_targets_min": 518
+    },
+    {
+      "epoch": 2.6276276276276276,
+      "grad_norm": 0.6357283334230661,
+      "learning_rate": 3.1469142948630194e-05,
+      "loss": 0.2552,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27663713693618774,
+      "step": 1750,
+      "valid_targets_mean": 3277.6,
+      "valid_targets_min": 416
+    },
+    {
+      "epoch": 2.635135135135135,
+      "grad_norm": 0.5459413685496709,
+      "learning_rate": 3.140771097402798e-05,
+      "loss": 0.2528,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.257800430059433,
+      "step": 1755,
+      "valid_targets_mean": 4285.2,
+      "valid_targets_min": 485
+    },
+    {
+      "epoch": 2.6426426426426426,
+      "grad_norm": 0.5121436251283435,
+      "learning_rate": 3.134611905326295e-05,
+      "loss": 0.2374,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2595219910144806,
+      "step": 1760,
+      "valid_targets_mean": 4998.1,
+      "valid_targets_min": 900
+    },
+    {
+      "epoch": 2.65015015015015,
+      "grad_norm": 0.5290785372037086,
+      "learning_rate": 3.12843680499081e-05,
+      "loss": 0.2823,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.380996972322464,
+      "step": 1765,
+      "valid_targets_mean": 5265.2,
+      "valid_targets_min": 668
+    },
+    {
+      "epoch": 2.6576576576576576,
+      "grad_norm": 0.7048181374115327,
+      "learning_rate": 3.122245882976693e-05,
+      "loss": 0.2989,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2908703684806824,
+      "step": 1770,
+      "valid_targets_mean": 2736.2,
+      "valid_targets_min": 909
+    },
+    {
+      "epoch": 2.665165165165165,
+      "grad_norm": 0.47236135012538977,
+      "learning_rate": 3.116039226086128e-05,
+      "loss": 0.2746,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.30878859758377075,
+      "step": 1775,
+      "valid_targets_mean": 5567.8,
+      "valid_targets_min": 635
+    },
+    {
+      "epoch": 2.6726726726726726,
+      "grad_norm": 0.5445102500932314,
+      "learning_rate": 3.1098169213419155e-05,
+      "loss": 0.2366,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2582714259624481,
+      "step": 1780,
+      "valid_targets_mean": 3773.7,
+      "valid_targets_min": 376
+    },
+    {
+      "epoch": 2.68018018018018,
+      "grad_norm": 0.4265129001988632,
+      "learning_rate": 3.103579055986251e-05,
+      "loss": 0.2629,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2143097072839737,
+      "step": 1785,
+      "valid_targets_mean": 5390.5,
+      "valid_targets_min": 704
+    },
+    {
+      "epoch": 2.6876876876876876,
+      "grad_norm": 0.5312743639290497,
+      "learning_rate": 3.0973257174795074e-05,
+      "loss": 0.2788,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3563080430030823,
+      "step": 1790,
+      "valid_targets_mean": 4896.1,
+      "valid_targets_min": 666
+    },
+    {
+      "epoch": 2.695195195195195,
+      "grad_norm": 0.46913349560934214,
+      "learning_rate": 3.0910569934990004e-05,
+      "loss": 0.2995,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22366014122962952,
+      "step": 1795,
+      "valid_targets_mean": 5595.9,
+      "valid_targets_min": 764
+    },
+    {
+      "epoch": 2.7027027027027026,
+      "grad_norm": 0.4542845059111592,
+      "learning_rate": 3.084772971937766e-05,
+      "loss": 0.2346,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.238237202167511,
+      "step": 1800,
+      "valid_targets_mean": 4653.6,
+      "valid_targets_min": 745
+    },
+    {
+      "epoch": 2.71021021021021,
+      "grad_norm": 0.5378230783334604,
+      "learning_rate": 3.078473740903325e-05,
+      "loss": 0.2585,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.30502867698669434,
+      "step": 1805,
+      "valid_targets_mean": 4299.5,
+      "valid_targets_min": 755
+    },
+    {
+      "epoch": 2.7177177177177176,
+      "grad_norm": 0.5591431243642122,
+      "learning_rate": 3.07215938871645e-05,
+      "loss": 0.2414,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.232755646109581,
+      "step": 1810,
+      "valid_targets_mean": 3461.2,
+      "valid_targets_min": 744
+    },
+    {
+      "epoch": 2.725225225225225,
+      "grad_norm": 0.6091187770028943,
+      "learning_rate": 3.0658300039099235e-05,
+      "loss": 0.2612,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2591820955276489,
+      "step": 1815,
+      "valid_targets_mean": 3384.4,
+      "valid_targets_min": 685
+    },
+    {
+      "epoch": 2.7327327327327327,
+      "grad_norm": 0.6447111768897391,
+      "learning_rate": 3.0594856752272987e-05,
+      "loss": 0.2546,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25599902868270874,
+      "step": 1820,
+      "valid_targets_mean": 3581.4,
+      "valid_targets_min": 657
+    },
+    {
+      "epoch": 2.74024024024024,
+      "grad_norm": 0.44894238539094306,
+      "learning_rate": 3.053126491621657e-05,
+      "loss": 0.2745,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2371259331703186,
+      "step": 1825,
+      "valid_targets_mean": 4889.4,
+      "valid_targets_min": 511
+    },
+    {
+      "epoch": 2.7477477477477477,
+      "grad_norm": 0.5214275290965238,
+      "learning_rate": 3.046752542254357e-05,
+      "loss": 0.2397,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22412139177322388,
+      "step": 1830,
+      "valid_targets_mean": 3863.1,
+      "valid_targets_min": 670
+    },
+    {
+      "epoch": 2.755255255255255,
+      "grad_norm": 0.5573497856924043,
+      "learning_rate": 3.0403639164937865e-05,
+      "loss": 0.2636,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24575021862983704,
+      "step": 1835,
+      "valid_targets_mean": 3198.3,
+      "valid_targets_min": 461
+    },
+    {
+      "epoch": 2.7627627627627627,
+      "grad_norm": 0.47880699639373114,
+      "learning_rate": 3.0339607039141107e-05,
+      "loss": 0.2303,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25173407793045044,
+      "step": 1840,
+      "valid_targets_mean": 4396.7,
+      "valid_targets_min": 277
+    },
+    {
+      "epoch": 2.77027027027027,
+      "grad_norm": 0.5293345660103224,
+      "learning_rate": 3.027542994294013e-05,
+      "loss": 0.2464,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24035535752773285,
+      "step": 1845,
+      "valid_targets_mean": 3506.6,
+      "valid_targets_min": 562
+    },
+    {
+      "epoch": 2.7777777777777777,
+      "grad_norm": 0.5030154656400301,
+      "learning_rate": 3.0211108776154394e-05,
+      "loss": 0.27,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24871063232421875,
+      "step": 1850,
+      "valid_targets_mean": 4511.1,
+      "valid_targets_min": 654
+    },
+    {
+      "epoch": 2.785285285285285,
+      "grad_norm": 0.4172382663816932,
+      "learning_rate": 3.0146644440623355e-05,
+      "loss": 0.2676,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22848474979400635,
+      "step": 1855,
+      "valid_targets_mean": 5617.8,
+      "valid_targets_min": 761
+    },
+    {
+      "epoch": 2.7927927927927927,
+      "grad_norm": 0.6045475468870822,
+      "learning_rate": 3.0082037840193812e-05,
+      "loss": 0.2434,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26820242404937744,
+      "step": 1860,
+      "valid_targets_mean": 3179.9,
+      "valid_targets_min": 612
+    },
+    {
+      "epoch": 2.8003003003003,
+      "grad_norm": 0.5418178665003754,
+      "learning_rate": 3.001728988070724e-05,
+      "loss": 0.3025,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3145066499710083,
+      "step": 1865,
+      "valid_targets_mean": 4837.1,
+      "valid_targets_min": 709
+    },
+    {
+      "epoch": 2.8078078078078077,
+      "grad_norm": 0.5134558903404736,
+      "learning_rate": 2.995240146998712e-05,
+      "loss": 0.2523,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21890783309936523,
+      "step": 1870,
+      "valid_targets_mean": 4246.1,
+      "valid_targets_min": 536
+    },
+    {
+      "epoch": 2.815315315315315,
+      "grad_norm": 0.583571397927643,
+      "learning_rate": 2.9887373517826153e-05,
+      "loss": 0.2692,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27330344915390015,
+      "step": 1875,
+      "valid_targets_mean": 3632.4,
+      "valid_targets_min": 630
+    },
+    {
+      "epoch": 2.8228228228228227,
+      "grad_norm": 0.47961940508462947,
+      "learning_rate": 2.9822206935973548e-05,
+      "loss": 0.2635,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2063194215297699,
+      "step": 1880,
+      "valid_targets_mean": 4621.7,
+      "valid_targets_min": 492
+    },
+    {
+      "epoch": 2.83033033033033,
+      "grad_norm": 0.5920736352732671,
+      "learning_rate": 2.9756902638122213e-05,
+      "loss": 0.2319,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23864233493804932,
+      "step": 1885,
+      "valid_targets_mean": 4123.9,
+      "valid_targets_min": 629
+    },
+    {
+      "epoch": 2.8378378378378377,
+      "grad_norm": 0.5565272788823533,
+      "learning_rate": 2.969146153989598e-05,
+      "loss": 0.2509,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22772765159606934,
+      "step": 1890,
+      "valid_targets_mean": 3277.6,
+      "valid_targets_min": 713
+    },
+    {
+      "epoch": 2.8453453453453452,
+      "grad_norm": 0.6239789579469722,
+      "learning_rate": 2.9625884558836716e-05,
+      "loss": 0.2457,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26187804341316223,
+      "step": 1895,
+      "valid_targets_mean": 2902.2,
+      "valid_targets_min": 503
+    },
+    {
+      "epoch": 2.8528528528528527,
+      "grad_norm": 0.5680461426104465,
+      "learning_rate": 2.95601726143915e-05,
+      "loss": 0.2528,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3167528510093689,
+      "step": 1900,
+      "valid_targets_mean": 4054.1,
+      "valid_targets_min": 697
+    },
+    {
+      "epoch": 2.8603603603603602,
+      "grad_norm": 0.45456838507650504,
+      "learning_rate": 2.949432662789971e-05,
+      "loss": 0.2483,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.29696983098983765,
+      "step": 1905,
+      "valid_targets_mean": 6072.6,
+      "valid_targets_min": 598
+    },
+    {
+      "epoch": 2.8678678678678677,
+      "grad_norm": 0.496183629571747,
+      "learning_rate": 2.942834752258012e-05,
+      "loss": 0.2392,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21913406252861023,
+      "step": 1910,
+      "valid_targets_mean": 4109.5,
+      "valid_targets_min": 879
+    },
+    {
+      "epoch": 2.8753753753753752,
+      "grad_norm": 0.5334630488557025,
+      "learning_rate": 2.936223622351794e-05,
+      "loss": 0.2325,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.31040453910827637,
+      "step": 1915,
+      "valid_targets_mean": 4864.2,
+      "valid_targets_min": 328
+    },
+    {
+      "epoch": 2.8828828828828827,
+      "grad_norm": 0.5005858977972606,
+      "learning_rate": 2.929599365765185e-05,
+      "loss": 0.2493,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.31291884183883667,
+      "step": 1920,
+      "valid_targets_mean": 5317.4,
+      "valid_targets_min": 547
+    },
+    {
+      "epoch": 2.8903903903903903,
+      "grad_norm": 0.4974443542961955,
+      "learning_rate": 2.9229620753761013e-05,
+      "loss": 0.2473,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.301303505897522,
+      "step": 1925,
+      "valid_targets_mean": 4499.8,
+      "valid_targets_min": 625
+    },
+    {
+      "epoch": 2.8978978978978978,
+      "grad_norm": 0.5691190427521686,
+      "learning_rate": 2.9163118442452046e-05,
+      "loss": 0.2449,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27089810371398926,
+      "step": 1930,
+      "valid_targets_mean": 4097.7,
+      "valid_targets_min": 667
+    },
+    {
+      "epoch": 2.9054054054054053,
+      "grad_norm": 0.4198119186033015,
+      "learning_rate": 2.9096487656145968e-05,
+      "loss": 0.2596,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24759484827518463,
+      "step": 1935,
+      "valid_targets_mean": 6335.4,
+      "valid_targets_min": 564
+    },
+    {
+      "epoch": 2.9129129129129128,
+      "grad_norm": 0.5531720093912547,
+      "learning_rate": 2.9029729329065134e-05,
+      "loss": 0.2598,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23836839199066162,
+      "step": 1940,
+      "valid_targets_mean": 3271.4,
+      "valid_targets_min": 510
+    },
+    {
+      "epoch": 2.9204204204204203,
+      "grad_norm": 0.4289864597777698,
+      "learning_rate": 2.896284439722013e-05,
+      "loss": 0.2767,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2916072607040405,
+      "step": 1945,
+      "valid_targets_mean": 6052.6,
+      "valid_targets_min": 466
+    },
+    {
+      "epoch": 2.9279279279279278,
+      "grad_norm": 0.47685929908340674,
+      "learning_rate": 2.8895833798396657e-05,
+      "loss": 0.2414,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22440025210380554,
+      "step": 1950,
+      "valid_targets_mean": 4654.8,
+      "valid_targets_min": 550
+    },
+    {
+      "epoch": 2.9354354354354353,
+      "grad_norm": 0.4511746235008368,
+      "learning_rate": 2.882869847214237e-05,
+      "loss": 0.2683,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2768084406852722,
+      "step": 1955,
+      "valid_targets_mean": 6411.6,
+      "valid_targets_min": 648
+    },
+    {
+      "epoch": 2.942942942942943,
+      "grad_norm": 0.3897834130570682,
+      "learning_rate": 2.876143935975373e-05,
+      "loss": 0.2327,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23787030577659607,
+      "step": 1960,
+      "valid_targets_mean": 6063.3,
+      "valid_targets_min": 668
+    },
+    {
+      "epoch": 2.9504504504504503,
+      "grad_norm": 0.5645744222530638,
+      "learning_rate": 2.8694057404262757e-05,
+      "loss": 0.2618,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26880699396133423,
+      "step": 1965,
+      "valid_targets_mean": 3741.8,
+      "valid_targets_min": 651
+    },
+    {
+      "epoch": 2.957957957957958,
+      "grad_norm": 0.48669690646418146,
+      "learning_rate": 2.862655355042387e-05,
+      "loss": 0.261,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.33517420291900635,
+      "step": 1970,
+      "valid_targets_mean": 5469.8,
+      "valid_targets_min": 723
+    },
+    {
+      "epoch": 2.9654654654654653,
+      "grad_norm": 0.4988811225771945,
+      "learning_rate": 2.85589287447006e-05,
+      "loss": 0.2573,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2532571256160736,
+      "step": 1975,
+      "valid_targets_mean": 4500.7,
+      "valid_targets_min": 824
+    },
+    {
+      "epoch": 2.972972972972973,
+      "grad_norm": 0.5067859361295777,
+      "learning_rate": 2.849118393525233e-05,
+      "loss": 0.2958,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22295460104942322,
+      "step": 1980,
+      "valid_targets_mean": 4019.2,
+      "valid_targets_min": 439
+    },
+    {
+      "epoch": 2.9804804804804803,
+      "grad_norm": 0.5614437078904748,
+      "learning_rate": 2.8423320071920986e-05,
+      "loss": 0.2496,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25288334488868713,
+      "step": 1985,
+      "valid_targets_mean": 3232.5,
+      "valid_targets_min": 790
+    },
+    {
+      "epoch": 2.987987987987988,
+      "grad_norm": 0.4074367928471507,
+      "learning_rate": 2.835533810621777e-05,
+      "loss": 0.2508,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22124746441841125,
+      "step": 1990,
+      "valid_targets_mean": 6511.8,
+      "valid_targets_min": 610
+    },
+    {
+      "epoch": 2.9954954954954953,
+      "grad_norm": 0.4746388334736856,
+      "learning_rate": 2.8287238991309746e-05,
+      "loss": 0.2591,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22991371154785156,
+      "step": 1995,
+      "valid_targets_mean": 4459.8,
+      "valid_targets_min": 742
+    },
+    {
+      "epoch": 3.003003003003003,
+      "grad_norm": 0.5433525185386173,
+      "learning_rate": 2.8219023682006533e-05,
+      "loss": 0.261,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2538215219974518,
+      "step": 2000,
+      "valid_targets_mean": 3868.6,
+      "valid_targets_min": 593
+    },
+    {
+      "epoch": 3.0105105105105103,
+      "grad_norm": 0.6522092709294501,
+      "learning_rate": 2.8150693134746895e-05,
+      "loss": 0.2505,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25473958253860474,
+      "step": 2005,
+      "valid_targets_mean": 4023.8,
+      "valid_targets_min": 939
+    },
+    {
+      "epoch": 3.018018018018018,
+      "grad_norm": 0.4893307325525604,
+      "learning_rate": 2.8082248307585332e-05,
+      "loss": 0.2284,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20237843692302704,
+      "step": 2010,
+      "valid_targets_mean": 4555.2,
+      "valid_targets_min": 590
+    },
+    {
+      "epoch": 3.0255255255255253,
+      "grad_norm": 0.4667369885254396,
+      "learning_rate": 2.801369016017865e-05,
+      "loss": 0.2271,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18705838918685913,
+      "step": 2015,
+      "valid_targets_mean": 4288.2,
+      "valid_targets_min": 690
+    },
+    {
+      "epoch": 3.033033033033033,
+      "grad_norm": 0.49033218061993816,
+      "learning_rate": 2.7945019653772504e-05,
+      "loss": 0.2433,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2408279925584793,
+      "step": 2020,
+      "valid_targets_mean": 5262.1,
+      "valid_targets_min": 515
+    },
+    {
+      "epoch": 3.0405405405405403,
+      "grad_norm": 0.5391107334148109,
+      "learning_rate": 2.7876237751187917e-05,
+      "loss": 0.2436,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21394947171211243,
+      "step": 2025,
+      "valid_targets_mean": 3897.1,
+      "valid_targets_min": 1733
+    },
+    {
+      "epoch": 3.048048048048048,
+      "grad_norm": 0.6193116901994394,
+      "learning_rate": 2.7807345416807793e-05,
+      "loss": 0.2442,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23730601370334625,
+      "step": 2030,
+      "valid_targets_mean": 3262.4,
+      "valid_targets_min": 439
+    },
+    {
+      "epoch": 3.0555555555555554,
+      "grad_norm": 0.5319064047511843,
+      "learning_rate": 2.773834361656339e-05,
+      "loss": 0.2363,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2560746669769287,
+      "step": 2035,
+      "valid_targets_mean": 4626.2,
+      "valid_targets_min": 550
+    },
+    {
+      "epoch": 3.063063063063063,
+      "grad_norm": 0.7633521135620636,
+      "learning_rate": 2.766923331792075e-05,
+      "loss": 0.2507,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.30914023518562317,
+      "step": 2040,
+      "valid_targets_mean": 3140.8,
+      "valid_targets_min": 595
+    },
+    {
+      "epoch": 3.0705705705705704,
+      "grad_norm": 1.0068010745411222,
+      "learning_rate": 2.7600015489867188e-05,
+      "loss": 0.2413,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2271232008934021,
+      "step": 2045,
+      "valid_targets_mean": 4060.6,
+      "valid_targets_min": 667
+    },
+    {
+      "epoch": 3.078078078078078,
+      "grad_norm": 0.5451000708994922,
+      "learning_rate": 2.753069110289766e-05,
+      "loss": 0.2459,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22742116451263428,
+      "step": 2050,
+      "valid_targets_mean": 3681.6,
+      "valid_targets_min": 668
+    },
+    {
+      "epoch": 3.0855855855855854,
+      "grad_norm": 0.5547733055241225,
+      "learning_rate": 2.7461261129001174e-05,
+      "loss": 0.2353,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24244071543216705,
+      "step": 2055,
+      "valid_targets_mean": 3975.5,
+      "valid_targets_min": 495
+    },
+    {
+      "epoch": 3.093093093093093,
+      "grad_norm": 0.42158312641460743,
+      "learning_rate": 2.7391726541647163e-05,
+      "loss": 0.2154,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1724860966205597,
+      "step": 2060,
+      "valid_targets_mean": 5298.3,
+      "valid_targets_min": 423
+    },
+    {
+      "epoch": 3.1006006006006004,
+      "grad_norm": 0.5499035304566007,
+      "learning_rate": 2.7322088315771834e-05,
+      "loss": 0.2427,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24446254968643188,
+      "step": 2065,
+      "valid_targets_mean": 3607.4,
+      "valid_targets_min": 478
+    },
+    {
+      "epoch": 3.108108108108108,
+      "grad_norm": 0.5590055107325951,
+      "learning_rate": 2.725234742776448e-05,
+      "loss": 0.2278,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23849673569202423,
+      "step": 2070,
+      "valid_targets_mean": 5030.8,
+      "valid_targets_min": 659
+    },
+    {
+      "epoch": 3.1156156156156154,
+      "grad_norm": 0.478630517596832,
+      "learning_rate": 2.7182504855453834e-05,
+      "loss": 0.2499,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20381265878677368,
+      "step": 2075,
+      "valid_targets_mean": 4702.2,
+      "valid_targets_min": 865
+    },
+    {
+      "epoch": 3.123123123123123,
+      "grad_norm": 0.5451665017950299,
+      "learning_rate": 2.7112561578094327e-05,
+      "loss": 0.224,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2420223355293274,
+      "step": 2080,
+      "valid_targets_mean": 5865.1,
+      "valid_targets_min": 611
+    },
+    {
+      "epoch": 3.1306306306306304,
+      "grad_norm": 0.5244203768707708,
+      "learning_rate": 2.704251857635234e-05,
+      "loss": 0.2323,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20124025642871857,
+      "step": 2085,
+      "valid_targets_mean": 3856.1,
+      "valid_targets_min": 622
+    },
+    {
+      "epoch": 3.138138138138138,
+      "grad_norm": 0.553596247360053,
+      "learning_rate": 2.69723768322925e-05,
+      "loss": 0.2429,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2544230818748474,
+      "step": 2090,
+      "valid_targets_mean": 3771.1,
+      "valid_targets_min": 479
+    },
+    {
+      "epoch": 3.1456456456456454,
+      "grad_norm": 0.5826774718951724,
+      "learning_rate": 2.6902137329363892e-05,
+      "loss": 0.2476,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27420639991760254,
+      "step": 2095,
+      "valid_targets_mean": 3996.2,
+      "valid_targets_min": 703
+    },
+    {
+      "epoch": 3.153153153153153,
+      "grad_norm": 0.6143348318690984,
+      "learning_rate": 2.683180105238625e-05,
+      "loss": 0.2307,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2823088765144348,
+      "step": 2100,
+      "valid_targets_mean": 3603.2,
+      "valid_targets_min": 664
+    },
+    {
+      "epoch": 3.1606606606606604,
+      "grad_norm": 0.5590576178147459,
+      "learning_rate": 2.676136898753617e-05,
+      "loss": 0.2448,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25955310463905334,
+      "step": 2105,
+      "valid_targets_mean": 3985.4,
+      "valid_targets_min": 773
+    },
+    {
+      "epoch": 3.1681681681681684,
+      "grad_norm": 0.596190398717275,
+      "learning_rate": 2.6690842122333286e-05,
+      "loss": 0.2255,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24386368691921234,
+      "step": 2110,
+      "valid_targets_mean": 3912.7,
+      "valid_targets_min": 728
+    },
+    {
+      "epoch": 3.175675675675676,
+      "grad_norm": 0.6071315295064825,
+      "learning_rate": 2.6620221445626416e-05,
+      "loss": 0.2096,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23695698380470276,
+      "step": 2115,
+      "valid_targets_mean": 3008.4,
+      "valid_targets_min": 656
+    },
+    {
+      "epoch": 3.1831831831831834,
+      "grad_norm": 0.6533097039267628,
+      "learning_rate": 2.6549507947579685e-05,
+      "loss": 0.2575,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2619466185569763,
+      "step": 2120,
+      "valid_targets_mean": 5412.9,
+      "valid_targets_min": 491
+    },
+    {
+      "epoch": 3.190690690690691,
+      "grad_norm": 0.49336937759151656,
+      "learning_rate": 2.6478702619658672e-05,
+      "loss": 0.2538,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23457828164100647,
+      "step": 2125,
+      "valid_targets_mean": 4639.4,
+      "valid_targets_min": 456
+    },
+    {
+      "epoch": 3.1981981981981984,
+      "grad_norm": 0.48643384497808734,
+      "learning_rate": 2.6407806454616472e-05,
+      "loss": 0.2368,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18629509210586548,
+      "step": 2130,
+      "valid_targets_mean": 3978.7,
+      "valid_targets_min": 815
+    },
+    {
+      "epoch": 3.205705705705706,
+      "grad_norm": 0.7137473060732604,
+      "learning_rate": 2.633682044647982e-05,
+      "loss": 0.2492,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.34111836552619934,
+      "step": 2135,
+      "valid_targets_mean": 2986.7,
+      "valid_targets_min": 211
+    },
+    {
+      "epoch": 3.2132132132132134,
+      "grad_norm": 0.4960650096175783,
+      "learning_rate": 2.626574559053512e-05,
+      "loss": 0.254,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2693920433521271,
+      "step": 2140,
+      "valid_targets_mean": 6290.2,
+      "valid_targets_min": 728
+    },
+    {
+      "epoch": 3.220720720720721,
+      "grad_norm": 0.543711754415262,
+      "learning_rate": 2.619458288331449e-05,
+      "loss": 0.2648,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.34854060411453247,
+      "step": 2145,
+      "valid_targets_mean": 4955.6,
+      "valid_targets_min": 273
+    },
+    {
+      "epoch": 3.2282282282282284,
+      "grad_norm": 0.4118913840003615,
+      "learning_rate": 2.6123333322581806e-05,
+      "loss": 0.2272,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22846825420856476,
+      "step": 2150,
+      "valid_targets_mean": 6945.4,
+      "valid_targets_min": 620
+    },
+    {
+      "epoch": 3.235735735735736,
+      "grad_norm": 0.6226838119580127,
+      "learning_rate": 2.6051997907318724e-05,
+      "loss": 0.2492,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2525981664657593,
+      "step": 2155,
+      "valid_targets_mean": 3692.9,
+      "valid_targets_min": 649
+    },
+    {
+      "epoch": 3.2432432432432434,
+      "grad_norm": 0.4814778903119386,
+      "learning_rate": 2.5980577637710632e-05,
+      "loss": 0.2397,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21874025464057922,
+      "step": 2160,
+      "valid_targets_mean": 5269.8,
+      "valid_targets_min": 814
+    },
+    {
+      "epoch": 3.250750750750751,
+      "grad_norm": 0.48643926272296106,
+      "learning_rate": 2.5909073515132667e-05,
+      "loss": 0.2508,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23266538977622986,
+      "step": 2165,
+      "valid_targets_mean": 4732.6,
+      "valid_targets_min": 898
+    },
+    {
+      "epoch": 3.2582582582582584,
+      "grad_norm": 0.5108891449221944,
+      "learning_rate": 2.5837486542135648e-05,
+      "loss": 0.2329,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2183685451745987,
+      "step": 2170,
+      "valid_targets_mean": 4184.6,
+      "valid_targets_min": 659
+    },
+    {
+      "epoch": 3.265765765765766,
+      "grad_norm": 0.5625875087588166,
+      "learning_rate": 2.576581772243204e-05,
+      "loss": 0.2766,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21907004714012146,
+      "step": 2175,
+      "valid_targets_mean": 5702.3,
+      "valid_targets_min": 447
+    },
+    {
+      "epoch": 3.2732732732732734,
+      "grad_norm": 0.44524330986567173,
+      "learning_rate": 2.5694068060881856e-05,
+      "loss": 0.2347,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2379521131515503,
+      "step": 2180,
+      "valid_targets_mean": 5613.1,
+      "valid_targets_min": 444
+    },
+    {
+      "epoch": 3.280780780780781,
+      "grad_norm": 0.41557044227218803,
+      "learning_rate": 2.5622238563478603e-05,
+      "loss": 0.2161,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1911526620388031,
+      "step": 2185,
+      "valid_targets_mean": 6414.1,
+      "valid_targets_min": 926
+    },
+    {
+      "epoch": 3.2882882882882885,
+      "grad_norm": 0.6461866697440014,
+      "learning_rate": 2.555033023733514e-05,
+      "loss": 0.2227,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2639209032058716,
+      "step": 2190,
+      "valid_targets_mean": 3629.4,
+      "valid_targets_min": 580
+    },
+    {
+      "epoch": 3.295795795795796,
+      "grad_norm": 0.5159050633898964,
+      "learning_rate": 2.547834409066958e-05,
+      "loss": 0.2363,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27777135372161865,
+      "step": 2195,
+      "valid_targets_mean": 5053.3,
+      "valid_targets_min": 906
+    },
+    {
+      "epoch": 3.3033033033033035,
+      "grad_norm": 0.477902570867691,
+      "learning_rate": 2.540628113279116e-05,
+      "loss": 0.2412,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3054026961326599,
+      "step": 2200,
+      "valid_targets_mean": 6047.9,
+      "valid_targets_min": 806
+    },
+    {
+      "epoch": 3.310810810810811,
+      "grad_norm": 0.45215264311766057,
+      "learning_rate": 2.5334142374086053e-05,
+      "loss": 0.2229,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22195221483707428,
+      "step": 2205,
+      "valid_targets_mean": 5648.2,
+      "valid_targets_min": 742
+    },
+    {
+      "epoch": 3.3183183183183185,
+      "grad_norm": 0.8158948988984059,
+      "learning_rate": 2.5261928826003246e-05,
+      "loss": 0.2694,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3665313720703125,
+      "step": 2210,
+      "valid_targets_mean": 4771.2,
+      "valid_targets_min": 663
+    },
+    {
+      "epoch": 3.325825825825826,
+      "grad_norm": 0.6202523003371706,
+      "learning_rate": 2.518964150104034e-05,
+      "loss": 0.242,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22199711203575134,
+      "step": 2215,
+      "valid_targets_mean": 2897.6,
+      "valid_targets_min": 698
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.6530333048449412,
+      "learning_rate": 2.511728141272934e-05,
+      "loss": 0.2304,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24529844522476196,
+      "step": 2220,
+      "valid_targets_mean": 3370.7,
+      "valid_targets_min": 503
+    },
+    {
+      "epoch": 3.340840840840841,
+      "grad_norm": 0.5905198582269967,
+      "learning_rate": 2.5044849575622458e-05,
+      "loss": 0.2272,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2153581976890564,
+      "step": 2225,
+      "valid_targets_mean": 4749.2,
+      "valid_targets_min": 668
+    },
+    {
+      "epoch": 3.3483483483483485,
+      "grad_norm": 0.4868966063934165,
+      "learning_rate": 2.4972347005277903e-05,
+      "loss": 0.2831,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2968922257423401,
+      "step": 2230,
+      "valid_targets_mean": 5965.1,
+      "valid_targets_min": 498
+    },
+    {
+      "epoch": 3.355855855855856,
+      "grad_norm": 0.8961039788278082,
+      "learning_rate": 2.489977471824561e-05,
+      "loss": 0.2348,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23392842710018158,
+      "step": 2235,
+      "valid_targets_mean": 3846.9,
+      "valid_targets_min": 368
+    },
+    {
+      "epoch": 3.3633633633633635,
+      "grad_norm": 0.6752840687920286,
+      "learning_rate": 2.4827133732053014e-05,
+      "loss": 0.2576,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25479578971862793,
+      "step": 2240,
+      "valid_targets_mean": 3281.4,
+      "valid_targets_min": 291
+    },
+    {
+      "epoch": 3.370870870870871,
+      "grad_norm": 0.5573523830693498,
+      "learning_rate": 2.475442506519077e-05,
+      "loss": 0.2353,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20670834183692932,
+      "step": 2245,
+      "valid_targets_mean": 3441.7,
+      "valid_targets_min": 637
+    },
+    {
+      "epoch": 3.3783783783783785,
+      "grad_norm": 0.47716647459061995,
+      "learning_rate": 2.468164973709847e-05,
+      "loss": 0.2249,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21283169090747833,
+      "step": 2250,
+      "valid_targets_mean": 4464.2,
+      "valid_targets_min": 910
+    },
+    {
+      "epoch": 3.385885885885886,
+      "grad_norm": 0.5064998483360107,
+      "learning_rate": 2.4608808768150357e-05,
+      "loss": 0.2216,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19695068895816803,
+      "step": 2255,
+      "valid_targets_mean": 3949.1,
+      "valid_targets_min": 738
+    },
+    {
+      "epoch": 3.3933933933933935,
+      "grad_norm": 0.5175599454597674,
+      "learning_rate": 2.4535903179641026e-05,
+      "loss": 0.2265,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2480076104402542,
+      "step": 2260,
+      "valid_targets_mean": 5609.1,
+      "valid_targets_min": 1410
+    },
+    {
+      "epoch": 3.400900900900901,
+      "grad_norm": 0.42240678217071037,
+      "learning_rate": 2.4462933993771088e-05,
+      "loss": 0.2399,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26719754934310913,
+      "step": 2265,
+      "valid_targets_mean": 6515.5,
+      "valid_targets_min": 726
+    },
+    {
+      "epoch": 3.4084084084084085,
+      "grad_norm": 0.506514863267876,
+      "learning_rate": 2.438990223363284e-05,
+      "loss": 0.2589,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25513604283332825,
+      "step": 2270,
+      "valid_targets_mean": 4471.0,
+      "valid_targets_min": 732
+    },
+    {
+      "epoch": 3.415915915915916,
+      "grad_norm": 0.5367632072440157,
+      "learning_rate": 2.4316808923195926e-05,
+      "loss": 0.2387,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22310830652713776,
+      "step": 2275,
+      "valid_targets_mean": 4361.2,
+      "valid_targets_min": 700
+    },
+    {
+      "epoch": 3.4234234234234235,
+      "grad_norm": 0.44852402801221397,
+      "learning_rate": 2.4243655087293e-05,
+      "loss": 0.2582,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26332229375839233,
+      "step": 2280,
+      "valid_targets_mean": 6653.6,
+      "valid_targets_min": 583
+    },
+    {
+      "epoch": 3.430930930930931,
+      "grad_norm": 0.48402583907149443,
+      "learning_rate": 2.4170441751605308e-05,
+      "loss": 0.2328,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24115243554115295,
+      "step": 2285,
+      "valid_targets_mean": 4474.8,
+      "valid_targets_min": 723
+    },
+    {
+      "epoch": 3.4384384384384385,
+      "grad_norm": 0.58593699094807,
+      "learning_rate": 2.4097169942648356e-05,
+      "loss": 0.2145,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16966642439365387,
+      "step": 2290,
+      "valid_targets_mean": 5481.7,
+      "valid_targets_min": 1177
+    },
+    {
+      "epoch": 3.445945945945946,
+      "grad_norm": 0.4710415176014163,
+      "learning_rate": 2.4023840687757476e-05,
+      "loss": 0.2432,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20201632380485535,
+      "step": 2295,
+      "valid_targets_mean": 5250.4,
+      "valid_targets_min": 717
+    },
+    {
+      "epoch": 3.4534534534534536,
+      "grad_norm": 0.5084005135734221,
+      "learning_rate": 2.395045501507347e-05,
+      "loss": 0.2254,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20893435180187225,
+      "step": 2300,
+      "valid_targets_mean": 4357.3,
+      "valid_targets_min": 489
+    },
+    {
+      "epoch": 3.460960960960961,
+      "grad_norm": 0.619899875868498,
+      "learning_rate": 2.387701395352815e-05,
+      "loss": 0.227,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21758192777633667,
+      "step": 2305,
+      "valid_targets_mean": 2881.0,
+      "valid_targets_min": 578
+    },
+    {
+      "epoch": 3.4684684684684686,
+      "grad_norm": 0.48471567383203296,
+      "learning_rate": 2.380351853282992e-05,
+      "loss": 0.2445,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24553707242012024,
+      "step": 2310,
+      "valid_targets_mean": 5598.6,
+      "valid_targets_min": 552
+    },
+    {
+      "epoch": 3.475975975975976,
+      "grad_norm": 0.6056696323887422,
+      "learning_rate": 2.372996978344937e-05,
+      "loss": 0.2255,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26694750785827637,
+      "step": 2315,
+      "valid_targets_mean": 3924.4,
+      "valid_targets_min": 523
+    },
+    {
+      "epoch": 3.4834834834834836,
+      "grad_norm": 0.601885827026815,
+      "learning_rate": 2.3656368736604786e-05,
+      "loss": 0.2568,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28267890214920044,
+      "step": 2320,
+      "valid_targets_mean": 3601.5,
+      "valid_targets_min": 724
+    },
+    {
+      "epoch": 3.490990990990991,
+      "grad_norm": 0.6254412393703039,
+      "learning_rate": 2.3582716424247728e-05,
+      "loss": 0.2711,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2540096044540405,
+      "step": 2325,
+      "valid_targets_mean": 2951.5,
+      "valid_targets_min": 537
+    },
+    {
+      "epoch": 3.4984984984984986,
+      "grad_norm": 0.5798449489473179,
+      "learning_rate": 2.3509013879048526e-05,
+      "loss": 0.2484,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.30426591634750366,
+      "step": 2330,
+      "valid_targets_mean": 4265.3,
+      "valid_targets_min": 712
+    },
+    {
+      "epoch": 3.506006006006006,
+      "grad_norm": 0.4923122631477329,
+      "learning_rate": 2.3435262134381823e-05,
+      "loss": 0.2068,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1547277569770813,
+      "step": 2335,
+      "valid_targets_mean": 4490.8,
+      "valid_targets_min": 620
+    },
+    {
+      "epoch": 3.5135135135135136,
+      "grad_norm": 0.49204973031276056,
+      "learning_rate": 2.3361462224312094e-05,
+      "loss": 0.2224,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1860155165195465,
+      "step": 2340,
+      "valid_targets_mean": 4172.9,
+      "valid_targets_min": 968
+    },
+    {
+      "epoch": 3.521021021021021,
+      "grad_norm": 0.5430832529381092,
+      "learning_rate": 2.3287615183579126e-05,
+      "loss": 0.2485,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23752620816230774,
+      "step": 2345,
+      "valid_targets_mean": 3755.6,
+      "valid_targets_min": 762
+    },
+    {
+      "epoch": 3.5285285285285286,
+      "grad_norm": 0.47304505139451253,
+      "learning_rate": 2.3213722047583517e-05,
+      "loss": 0.2238,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21495875716209412,
+      "step": 2350,
+      "valid_targets_mean": 5194.9,
+      "valid_targets_min": 781
+    },
+    {
+      "epoch": 3.536036036036036,
+      "grad_norm": 0.5593072985053018,
+      "learning_rate": 2.3139783852372157e-05,
+      "loss": 0.2415,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.32621026039123535,
+      "step": 2355,
+      "valid_targets_mean": 5330.6,
+      "valid_targets_min": 483
+    },
+    {
+      "epoch": 3.5435435435435436,
+      "grad_norm": 0.5133771865366825,
+      "learning_rate": 2.306580163462373e-05,
+      "loss": 0.2152,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24318471550941467,
+      "step": 2360,
+      "valid_targets_mean": 5257.0,
+      "valid_targets_min": 850
+    },
+    {
+      "epoch": 3.551051051051051,
+      "grad_norm": 0.7096881143231963,
+      "learning_rate": 2.299177643163413e-05,
+      "loss": 0.2412,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24741387367248535,
+      "step": 2365,
+      "valid_targets_mean": 2511.2,
+      "valid_targets_min": 508
+    },
+    {
+      "epoch": 3.5585585585585586,
+      "grad_norm": 0.5184545440577579,
+      "learning_rate": 2.2917709281301946e-05,
+      "loss": 0.2377,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28968262672424316,
+      "step": 2370,
+      "valid_targets_mean": 4808.0,
+      "valid_targets_min": 628
+    },
+    {
+      "epoch": 3.566066066066066,
+      "grad_norm": 0.5497412004652409,
+      "learning_rate": 2.2843601222113915e-05,
+      "loss": 0.2846,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24640163779258728,
+      "step": 2375,
+      "valid_targets_mean": 4700.1,
+      "valid_targets_min": 418
+    },
+    {
+      "epoch": 3.5735735735735736,
+      "grad_norm": 0.543092075993824,
+      "learning_rate": 2.2769453293130345e-05,
+      "loss": 0.2377,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2757692039012909,
+      "step": 2380,
+      "valid_targets_mean": 3818.9,
+      "valid_targets_min": 423
+    },
+    {
+      "epoch": 3.581081081081081,
+      "grad_norm": 0.4756311989207704,
+      "learning_rate": 2.2695266533970556e-05,
+      "loss": 0.242,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2674590051174164,
+      "step": 2385,
+      "valid_targets_mean": 5847.6,
+      "valid_targets_min": 794
+    },
+    {
+      "epoch": 3.5885885885885886,
+      "grad_norm": 0.7040802634858209,
+      "learning_rate": 2.2621041984798304e-05,
+      "loss": 0.2401,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27846670150756836,
+      "step": 2390,
+      "valid_targets_mean": 2509.4,
+      "valid_targets_min": 686
+    },
+    {
+      "epoch": 3.596096096096096,
+      "grad_norm": 0.6323033311574447,
+      "learning_rate": 2.2546780686307183e-05,
+      "loss": 0.2495,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23559826612472534,
+      "step": 2395,
+      "valid_targets_mean": 4216.6,
+      "valid_targets_min": 608
+    },
+    {
+      "epoch": 3.6036036036036037,
+      "grad_norm": 0.5439419831820919,
+      "learning_rate": 2.2472483679706067e-05,
+      "loss": 0.2578,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.271353542804718,
+      "step": 2400,
+      "valid_targets_mean": 4691.6,
+      "valid_targets_min": 755
+    },
+    {
+      "epoch": 3.611111111111111,
+      "grad_norm": 0.657338134532918,
+      "learning_rate": 2.2398152006704463e-05,
+      "loss": 0.2602,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26753100752830505,
+      "step": 2405,
+      "valid_targets_mean": 3173.4,
+      "valid_targets_min": 404
+    },
+    {
+      "epoch": 3.6186186186186187,
+      "grad_norm": 0.45053072622125706,
+      "learning_rate": 2.232378670949795e-05,
+      "loss": 0.2137,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18780098855495453,
+      "step": 2410,
+      "valid_targets_mean": 5203.4,
+      "valid_targets_min": 1334
+    },
+    {
+      "epoch": 3.626126126126126,
+      "grad_norm": 0.4023749713359982,
+      "learning_rate": 2.2249388830753534e-05,
+      "loss": 0.2321,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17731866240501404,
+      "step": 2415,
+      "valid_targets_mean": 5612.2,
+      "valid_targets_min": 480
+    },
+    {
+      "epoch": 3.6336336336336337,
+      "grad_norm": 0.5062285511706474,
+      "learning_rate": 2.217495941359506e-05,
+      "loss": 0.2514,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23712016642093658,
+      "step": 2420,
+      "valid_targets_mean": 4567.2,
+      "valid_targets_min": 567
+    },
+    {
+      "epoch": 3.641141141141141,
+      "grad_norm": 0.5407394227692897,
+      "learning_rate": 2.2100499501588558e-05,
+      "loss": 0.2402,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25140225887298584,
+      "step": 2425,
+      "valid_targets_mean": 4480.6,
+      "valid_targets_min": 801
+    },
+    {
+      "epoch": 3.6486486486486487,
+      "grad_norm": 0.5432137809783814,
+      "learning_rate": 2.2026010138727628e-05,
+      "loss": 0.2364,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23100072145462036,
+      "step": 2430,
+      "valid_targets_mean": 4535.2,
+      "valid_targets_min": 696
+    },
+    {
+      "epoch": 3.656156156156156,
+      "grad_norm": 0.5702741738198654,
+      "learning_rate": 2.1951492369418786e-05,
+      "loss": 0.2448,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24533988535404205,
+      "step": 2435,
+      "valid_targets_mean": 4315.2,
+      "valid_targets_min": 571
+    },
+    {
+      "epoch": 3.6636636636636637,
+      "grad_norm": 0.6650337977127149,
+      "learning_rate": 2.1876947238466838e-05,
+      "loss": 0.256,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2869791090488434,
+      "step": 2440,
+      "valid_targets_mean": 3385.6,
+      "valid_targets_min": 557
+    },
+    {
+      "epoch": 3.671171171171171,
+      "grad_norm": 0.473867504379004,
+      "learning_rate": 2.1802375791060232e-05,
+      "loss": 0.2315,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18947833776474,
+      "step": 2445,
+      "valid_targets_mean": 5216.6,
+      "valid_targets_min": 742
+    },
+    {
+      "epoch": 3.6786786786786787,
+      "grad_norm": 0.7199290550127364,
+      "learning_rate": 2.1727779072756388e-05,
+      "loss": 0.2197,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.30827030539512634,
+      "step": 2450,
+      "valid_targets_mean": 4398.8,
+      "valid_targets_min": 696
+    },
+    {
+      "epoch": 3.686186186186186,
+      "grad_norm": 0.6169439986122458,
+      "learning_rate": 2.1653158129467048e-05,
+      "loss": 0.281,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2743998169898987,
+      "step": 2455,
+      "valid_targets_mean": 3159.7,
+      "valid_targets_min": 683
+    },
+    {
+      "epoch": 3.6936936936936937,
+      "grad_norm": 0.630068337186312,
+      "learning_rate": 2.1578514007443602e-05,
+      "loss": 0.2446,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2756376266479492,
+      "step": 2460,
+      "valid_targets_mean": 3321.1,
+      "valid_targets_min": 663
+    },
+    {
+      "epoch": 3.701201201201201,
+      "grad_norm": 8.40963413100615,
+      "learning_rate": 2.1503847753262447e-05,
+      "loss": 0.2376,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21705785393714905,
+      "step": 2465,
+      "valid_targets_mean": 5065.1,
+      "valid_targets_min": 1345
+    },
+    {
+      "epoch": 3.7087087087087087,
+      "grad_norm": 0.8840964089151675,
+      "learning_rate": 2.1429160413810268e-05,
+      "loss": 0.2268,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2861403822898865,
+      "step": 2470,
+      "valid_targets_mean": 5522.8,
+      "valid_targets_min": 991
+    },
+    {
+      "epoch": 3.7162162162162162,
+      "grad_norm": 0.47904088821045393,
+      "learning_rate": 2.1354453036269397e-05,
+      "loss": 0.2283,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.182481050491333,
+      "step": 2475,
+      "valid_targets_mean": 3823.9,
+      "valid_targets_min": 657
+    },
+    {
+      "epoch": 3.7237237237237237,
+      "grad_norm": 0.6350177279756045,
+      "learning_rate": 2.1279726668103112e-05,
+      "loss": 0.2235,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.289498507976532,
+      "step": 2480,
+      "valid_targets_mean": 3174.9,
+      "valid_targets_min": 597
+    },
+    {
+      "epoch": 3.7312312312312312,
+      "grad_norm": 0.5475085951271096,
+      "learning_rate": 2.1204982357040974e-05,
+      "loss": 0.2321,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2232646644115448,
+      "step": 2485,
+      "valid_targets_mean": 3746.9,
+      "valid_targets_min": 684
+    },
+    {
+      "epoch": 3.7387387387387387,
+      "grad_norm": 0.5721638552098921,
+      "learning_rate": 2.1130221151064095e-05,
+      "loss": 0.2408,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2669849693775177,
+      "step": 2490,
+      "valid_targets_mean": 4861.1,
+      "valid_targets_min": 469
+    },
+    {
+      "epoch": 3.7462462462462462,
+      "grad_norm": 0.46167196232037994,
+      "learning_rate": 2.105544409839048e-05,
+      "loss": 0.2249,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1929030418395996,
+      "step": 2495,
+      "valid_targets_mean": 4768.8,
+      "valid_targets_min": 888
+    },
+    {
+      "epoch": 3.7537537537537538,
+      "grad_norm": 0.5709387953688929,
+      "learning_rate": 2.098065224746031e-05,
+      "loss": 0.2465,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26624026894569397,
+      "step": 2500,
+      "valid_targets_mean": 3260.7,
+      "valid_targets_min": 788
+    },
+    {
+      "epoch": 3.7612612612612613,
+      "grad_norm": 0.5636970256534266,
+      "learning_rate": 2.0905846646921266e-05,
+      "loss": 0.2595,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.29653650522232056,
+      "step": 2505,
+      "valid_targets_mean": 3569.1,
+      "valid_targets_min": 856
+    },
+    {
+      "epoch": 3.7687687687687688,
+      "grad_norm": 0.4293486036135388,
+      "learning_rate": 2.0831028345613807e-05,
+      "loss": 0.2521,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22826270759105682,
+      "step": 2510,
+      "valid_targets_mean": 5393.5,
+      "valid_targets_min": 404
+    },
+    {
+      "epoch": 3.7762762762762763,
+      "grad_norm": 0.5914611642027671,
+      "learning_rate": 2.0756198392556458e-05,
+      "loss": 0.2313,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22963778674602509,
+      "step": 2515,
+      "valid_targets_mean": 2932.8,
+      "valid_targets_min": 513
+    },
+    {
+      "epoch": 3.7837837837837838,
+      "grad_norm": 0.5163134230928269,
+      "learning_rate": 2.0681357836931114e-05,
+      "loss": 0.224,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.207438126206398,
+      "step": 2520,
+      "valid_targets_mean": 4660.4,
+      "valid_targets_min": 750
+    },
+    {
+      "epoch": 3.7912912912912913,
+      "grad_norm": 0.7190308106564102,
+      "learning_rate": 2.0606507728068338e-05,
+      "loss": 0.2276,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2699480652809143,
+      "step": 2525,
+      "valid_targets_mean": 2463.4,
+      "valid_targets_min": 475
+    },
+    {
+      "epoch": 3.798798798798799,
+      "grad_norm": 0.6346404271383566,
+      "learning_rate": 2.0531649115432626e-05,
+      "loss": 0.2426,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26241040229797363,
+      "step": 2530,
+      "valid_targets_mean": 3470.4,
+      "valid_targets_min": 613
+    },
+    {
+      "epoch": 3.8063063063063063,
+      "grad_norm": 0.5405574803276272,
+      "learning_rate": 2.0456783048607708e-05,
+      "loss": 0.234,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2565130591392517,
+      "step": 2535,
+      "valid_targets_mean": 4024.1,
+      "valid_targets_min": 669
+    },
+    {
+      "epoch": 3.813813813813814,
+      "grad_norm": 0.604890803793606,
+      "learning_rate": 2.038191057728183e-05,
+      "loss": 0.2225,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2380007952451706,
+      "step": 2540,
+      "valid_targets_mean": 3456.8,
+      "valid_targets_min": 662
+    },
+    {
+      "epoch": 3.8213213213213213,
+      "grad_norm": 0.6793469587756615,
+      "learning_rate": 2.0307032751233038e-05,
+      "loss": 0.2593,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2839970290660858,
+      "step": 2545,
+      "valid_targets_mean": 4329.7,
+      "valid_targets_min": 681
+    },
+    {
+      "epoch": 3.828828828828829,
+      "grad_norm": 0.5213037499939942,
+      "learning_rate": 2.023215062031445e-05,
+      "loss": 0.247,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2534673511981964,
+      "step": 2550,
+      "valid_targets_mean": 4234.5,
+      "valid_targets_min": 437
+    },
+    {
+      "epoch": 3.8363363363363363,
+      "grad_norm": 0.48716095297652656,
+      "learning_rate": 2.0157265234439545e-05,
+      "loss": 0.2399,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17926955223083496,
+      "step": 2555,
+      "valid_targets_mean": 4508.6,
+      "valid_targets_min": 404
+    },
+    {
+      "epoch": 3.843843843843844,
+      "grad_norm": 0.45658767125521904,
+      "learning_rate": 2.0082377643567427e-05,
+      "loss": 0.2135,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16734261810779572,
+      "step": 2560,
+      "valid_targets_mean": 4532.5,
+      "valid_targets_min": 484
+    },
+    {
+      "epoch": 3.8513513513513513,
+      "grad_norm": 0.5887721035383384,
+      "learning_rate": 2.0007488897688145e-05,
+      "loss": 0.2469,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2615935802459717,
+      "step": 2565,
+      "valid_targets_mean": 3602.4,
+      "valid_targets_min": 596
+    },
+    {
+      "epoch": 3.858858858858859,
+      "grad_norm": 0.5556930711337941,
+      "learning_rate": 1.9932600046807914e-05,
+      "loss": 0.2355,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1911911964416504,
+      "step": 2570,
+      "valid_targets_mean": 3493.8,
+      "valid_targets_min": 801
+    },
+    {
+      "epoch": 3.8663663663663663,
+      "grad_norm": 0.4918999956248638,
+      "learning_rate": 1.985771214093444e-05,
+      "loss": 0.2162,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19995930790901184,
+      "step": 2575,
+      "valid_targets_mean": 4204.2,
+      "valid_targets_min": 634
+    },
+    {
+      "epoch": 3.873873873873874,
+      "grad_norm": 0.52841578647817,
+      "learning_rate": 1.9782826230062155e-05,
+      "loss": 0.2705,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28531813621520996,
+      "step": 2580,
+      "valid_targets_mean": 4070.0,
+      "valid_targets_min": 731
+    },
+    {
+      "epoch": 3.8813813813813813,
+      "grad_norm": 0.5040983553316225,
+      "learning_rate": 1.9707943364157552e-05,
+      "loss": 0.2039,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18574920296669006,
+      "step": 2585,
+      "valid_targets_mean": 3998.8,
+      "valid_targets_min": 416
+    },
+    {
+      "epoch": 3.888888888888889,
+      "grad_norm": 0.49554889736984387,
+      "learning_rate": 1.9633064593144408e-05,
+      "loss": 0.217,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20392203330993652,
+      "step": 2590,
+      "valid_targets_mean": 4092.8,
+      "valid_targets_min": 638
+    },
+    {
+      "epoch": 3.8963963963963963,
+      "grad_norm": 0.4727254866083527,
+      "learning_rate": 1.9558190966889093e-05,
+      "loss": 0.2598,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21000804007053375,
+      "step": 2595,
+      "valid_targets_mean": 5088.9,
+      "valid_targets_min": 615
+    },
+    {
+      "epoch": 3.903903903903904,
+      "grad_norm": 0.43824564462204646,
+      "learning_rate": 1.9483323535185838e-05,
+      "loss": 0.2129,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1834757924079895,
+      "step": 2600,
+      "valid_targets_mean": 4795.2,
+      "valid_targets_min": 852
+    },
+    {
+      "epoch": 3.9114114114114114,
+      "grad_norm": 0.4414256886221553,
+      "learning_rate": 1.940846334774203e-05,
+      "loss": 0.2198,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1729404479265213,
+      "step": 2605,
+      "valid_targets_mean": 4699.6,
+      "valid_targets_min": 635
+    },
+    {
+      "epoch": 3.918918918918919,
+      "grad_norm": 0.5060621091961622,
+      "learning_rate": 1.933361145416348e-05,
+      "loss": 0.2215,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2322886735200882,
+      "step": 2610,
+      "valid_targets_mean": 6213.0,
+      "valid_targets_min": 639
+    },
+    {
+      "epoch": 3.9264264264264264,
+      "grad_norm": 0.593672866808795,
+      "learning_rate": 1.9258768903939706e-05,
+      "loss": 0.2432,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24251478910446167,
+      "step": 2615,
+      "valid_targets_mean": 3359.5,
+      "valid_targets_min": 591
+    },
+    {
+      "epoch": 3.933933933933934,
+      "grad_norm": 0.4404905443315122,
+      "learning_rate": 1.9183936746429234e-05,
+      "loss": 0.2666,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2132512629032135,
+      "step": 2620,
+      "valid_targets_mean": 5353.2,
+      "valid_targets_min": 322
+    },
+    {
+      "epoch": 3.9414414414414414,
+      "grad_norm": 0.5116438369388818,
+      "learning_rate": 1.9109116030844874e-05,
+      "loss": 0.2318,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27721619606018066,
+      "step": 2625,
+      "valid_targets_mean": 4330.2,
+      "valid_targets_min": 757
+    },
+    {
+      "epoch": 3.948948948948949,
+      "grad_norm": 0.5905044794050749,
+      "learning_rate": 1.9034307806239004e-05,
+      "loss": 0.2603,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.33162492513656616,
+      "step": 2630,
+      "valid_targets_mean": 4979.0,
+      "valid_targets_min": 434
+    },
+    {
+      "epoch": 3.9564564564564564,
+      "grad_norm": 0.5353633419703424,
+      "learning_rate": 1.8959513121488868e-05,
+      "loss": 0.2544,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23549598455429077,
+      "step": 2635,
+      "valid_targets_mean": 4418.6,
+      "valid_targets_min": 858
+    },
+    {
+      "epoch": 3.963963963963964,
+      "grad_norm": 0.47963900371966894,
+      "learning_rate": 1.8884733025281876e-05,
+      "loss": 0.244,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1948413848876953,
+      "step": 2640,
+      "valid_targets_mean": 4141.8,
+      "valid_targets_min": 726
+    },
+    {
+      "epoch": 3.9714714714714714,
+      "grad_norm": 0.44606918683200836,
+      "learning_rate": 1.880996856610088e-05,
+      "loss": 0.2417,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21382328867912292,
+      "step": 2645,
+      "valid_targets_mean": 5472.8,
+      "valid_targets_min": 745
+    },
+    {
+      "epoch": 3.978978978978979,
+      "grad_norm": 0.5149587332986721,
+      "learning_rate": 1.8735220792209513e-05,
+      "loss": 0.228,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19594892859458923,
+      "step": 2650,
+      "valid_targets_mean": 4153.9,
+      "valid_targets_min": 598
+    },
+    {
+      "epoch": 3.9864864864864864,
+      "grad_norm": 0.5874287195534135,
+      "learning_rate": 1.8660490751637435e-05,
+      "loss": 0.2352,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22586959600448608,
+      "step": 2655,
+      "valid_targets_mean": 3378.9,
+      "valid_targets_min": 506
+    },
+    {
+      "epoch": 3.993993993993994,
+      "grad_norm": 0.46189741004980345,
+      "learning_rate": 1.858577949216569e-05,
+      "loss": 0.2134,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2193407416343689,
+      "step": 2660,
+      "valid_targets_mean": 4348.1,
+      "valid_targets_min": 447
+    },
+    {
+      "epoch": 4.001501501501502,
+      "grad_norm": 0.5233378328639746,
+      "learning_rate": 1.8511088061311982e-05,
+      "loss": 0.2246,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2248275727033615,
+      "step": 2665,
+      "valid_targets_mean": 3952.7,
+      "valid_targets_min": 671
+    },
+    {
+      "epoch": 4.009009009009009,
+      "grad_norm": 0.6071971400069938,
+      "learning_rate": 1.8436417506316013e-05,
+      "loss": 0.2354,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23917962610721588,
+      "step": 2670,
+      "valid_targets_mean": 3420.5,
+      "valid_targets_min": 429
+    },
+    {
+      "epoch": 4.016516516516517,
+      "grad_norm": 0.5914162200350297,
+      "learning_rate": 1.8361768874124778e-05,
+      "loss": 0.2518,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2445971816778183,
+      "step": 2675,
+      "valid_targets_mean": 4843.9,
+      "valid_targets_min": 471
+    },
+    {
+      "epoch": 4.024024024024024,
+      "grad_norm": 0.547907620188072,
+      "learning_rate": 1.8287143211377893e-05,
+      "loss": 0.2157,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1973641812801361,
+      "step": 2680,
+      "valid_targets_mean": 4041.2,
+      "valid_targets_min": 418
+    },
+    {
+      "epoch": 4.031531531531532,
+      "grad_norm": 0.5133198173072647,
+      "learning_rate": 1.8212541564392924e-05,
+      "loss": 0.2281,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17949888110160828,
+      "step": 2685,
+      "valid_targets_mean": 3785.6,
+      "valid_targets_min": 600
+    },
+    {
+      "epoch": 4.039039039039039,
+      "grad_norm": 0.47846466241843566,
+      "learning_rate": 1.813796497915073e-05,
+      "loss": 0.2478,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2564462423324585,
+      "step": 2690,
+      "valid_targets_mean": 5844.4,
+      "valid_targets_min": 506
+    },
+    {
+      "epoch": 4.046546546546547,
+      "grad_norm": 0.569435583148903,
+      "learning_rate": 1.806341450128076e-05,
+      "loss": 0.2314,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22870904207229614,
+      "step": 2695,
+      "valid_targets_mean": 4163.9,
+      "valid_targets_min": 699
+    },
+    {
+      "epoch": 4.054054054054054,
+      "grad_norm": 0.5070243213845931,
+      "learning_rate": 1.798889117604643e-05,
+      "loss": 0.2164,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23002973198890686,
+      "step": 2700,
+      "valid_targets_mean": 4445.3,
+      "valid_targets_min": 705
+    },
+    {
+      "epoch": 4.061561561561562,
+      "grad_norm": 0.4867044644660382,
+      "learning_rate": 1.7914396048330428e-05,
+      "loss": 0.1903,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19053609669208527,
+      "step": 2705,
+      "valid_targets_mean": 4508.4,
+      "valid_targets_min": 569
+    },
+    {
+      "epoch": 4.069069069069069,
+      "grad_norm": 0.5805381784749756,
+      "learning_rate": 1.7839930162620128e-05,
+      "loss": 0.2136,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1698979139328003,
+      "step": 2710,
+      "valid_targets_mean": 3140.1,
+      "valid_targets_min": 798
+    },
+    {
+      "epoch": 4.076576576576577,
+      "grad_norm": 0.602504277734938,
+      "learning_rate": 1.7765494562992878e-05,
+      "loss": 0.2122,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18390074372291565,
+      "step": 2715,
+      "valid_targets_mean": 3297.9,
+      "valid_targets_min": 534
+    },
+    {
+      "epoch": 4.084084084084084,
+      "grad_norm": 0.6470125645401709,
+      "learning_rate": 1.7691090293101386e-05,
+      "loss": 0.2572,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28710392117500305,
+      "step": 2720,
+      "valid_targets_mean": 3918.4,
+      "valid_targets_min": 437
+    },
+    {
+      "epoch": 4.091591591591592,
+      "grad_norm": 0.5066063002958375,
+      "learning_rate": 1.76167183961591e-05,
+      "loss": 0.2266,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2291889488697052,
+      "step": 2725,
+      "valid_targets_mean": 5055.3,
+      "valid_targets_min": 388
+    },
+    {
+      "epoch": 4.099099099099099,
+      "grad_norm": 0.6455258191217119,
+      "learning_rate": 1.7542379914925575e-05,
+      "loss": 0.2079,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23442783951759338,
+      "step": 2730,
+      "valid_targets_mean": 2962.8,
+      "valid_targets_min": 666
+    },
+    {
+      "epoch": 4.106606606606607,
+      "grad_norm": 0.7386040382747444,
+      "learning_rate": 1.7468075891691832e-05,
+      "loss": 0.2052,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21252450346946716,
+      "step": 2735,
+      "valid_targets_mean": 3100.8,
+      "valid_targets_min": 518
+    },
+    {
+      "epoch": 4.114114114114114,
+      "grad_norm": 0.48445294014230816,
+      "learning_rate": 1.7393807368265777e-05,
+      "loss": 0.208,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22077061235904694,
+      "step": 2740,
+      "valid_targets_mean": 5577.0,
+      "valid_targets_min": 614
+    },
+    {
+      "epoch": 4.121621621621622,
+      "grad_norm": 0.48865023064949326,
+      "learning_rate": 1.731957538595756e-05,
+      "loss": 0.22,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19657564163208008,
+      "step": 2745,
+      "valid_targets_mean": 4633.6,
+      "valid_targets_min": 542
+    },
+    {
+      "epoch": 4.129129129129129,
+      "grad_norm": 0.6690952004993063,
+      "learning_rate": 1.7245380985565014e-05,
+      "loss": 0.2026,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18894386291503906,
+      "step": 2750,
+      "valid_targets_mean": 3604.8,
+      "valid_targets_min": 437
+    },
+    {
+      "epoch": 4.136636636636637,
+      "grad_norm": 0.7195870104883615,
+      "learning_rate": 1.7171225207359023e-05,
+      "loss": 0.2258,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27104708552360535,
+      "step": 2755,
+      "valid_targets_mean": 3011.1,
+      "valid_targets_min": 948
+    },
+    {
+      "epoch": 4.1441441441441444,
+      "grad_norm": 0.5379416037342353,
+      "learning_rate": 1.7097109091068965e-05,
+      "loss": 0.2531,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3290520906448364,
+      "step": 2760,
+      "valid_targets_mean": 6327.2,
+      "valid_targets_min": 875
+    },
+    {
+      "epoch": 4.151651651651652,
+      "grad_norm": 0.4819288545219707,
+      "learning_rate": 1.7023033675868107e-05,
+      "loss": 0.2337,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19341519474983215,
+      "step": 2765,
+      "valid_targets_mean": 5650.7,
+      "valid_targets_min": 1052
+    },
+    {
+      "epoch": 4.1591591591591595,
+      "grad_norm": 0.4339406422217775,
+      "learning_rate": 1.694900000035907e-05,
+      "loss": 0.1945,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18829095363616943,
+      "step": 2770,
+      "valid_targets_mean": 5408.4,
+      "valid_targets_min": 808
+    },
+    {
+      "epoch": 4.166666666666667,
+      "grad_norm": 0.5494279815313058,
+      "learning_rate": 1.687500910255924e-05,
+      "loss": 0.2345,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.29033520817756653,
+      "step": 2775,
+      "valid_targets_mean": 4524.4,
+      "valid_targets_min": 660
+    },
+    {
+      "epoch": 4.1741741741741745,
+      "grad_norm": 0.4489742319836925,
+      "learning_rate": 1.6801062019886216e-05,
+      "loss": 0.2443,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2034509927034378,
+      "step": 2780,
+      "valid_targets_mean": 4920.3,
+      "valid_targets_min": 427
+    },
+    {
+      "epoch": 4.181681681681682,
+      "grad_norm": 0.5280057369516138,
+      "learning_rate": 1.6727159789143276e-05,
+      "loss": 0.2251,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25000298023223877,
+      "step": 2785,
+      "valid_targets_mean": 6445.3,
+      "valid_targets_min": 460
+    },
+    {
+      "epoch": 4.1891891891891895,
+      "grad_norm": 0.570852623942594,
+      "learning_rate": 1.665330344650484e-05,
+      "loss": 0.2091,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23752200603485107,
+      "step": 2790,
+      "valid_targets_mean": 4025.2,
+      "valid_targets_min": 639
+    },
+    {
+      "epoch": 4.196696696696697,
+      "grad_norm": 0.5022860617467421,
+      "learning_rate": 1.6579494027501926e-05,
+      "loss": 0.2702,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25862377882003784,
+      "step": 2795,
+      "valid_targets_mean": 5591.9,
+      "valid_targets_min": 1079
+    },
+    {
+      "epoch": 4.2042042042042045,
+      "grad_norm": 0.5808638456409436,
+      "learning_rate": 1.6505732567007644e-05,
+      "loss": 0.2144,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2629520297050476,
+      "step": 2800,
+      "valid_targets_mean": 4279.3,
+      "valid_targets_min": 681
+    },
+    {
+      "epoch": 4.211711711711712,
+      "grad_norm": 0.5011677067782193,
+      "learning_rate": 1.6432020099222686e-05,
+      "loss": 0.2239,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17377246916294098,
+      "step": 2805,
+      "valid_targets_mean": 4540.3,
+      "valid_targets_min": 506
+    },
+    {
+      "epoch": 4.2192192192192195,
+      "grad_norm": 0.6312221103561748,
+      "learning_rate": 1.635835765766082e-05,
+      "loss": 0.1895,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1711842268705368,
+      "step": 2810,
+      "valid_targets_mean": 4152.6,
+      "valid_targets_min": 576
+    },
+    {
+      "epoch": 4.226726726726727,
+      "grad_norm": 0.5455241347866726,
+      "learning_rate": 1.6284746275134413e-05,
+      "loss": 0.2123,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2519665062427521,
+      "step": 2815,
+      "valid_targets_mean": 4392.9,
+      "valid_targets_min": 605
+    },
+    {
+      "epoch": 4.2342342342342345,
+      "grad_norm": 0.544839041978784,
+      "learning_rate": 1.621118698373992e-05,
+      "loss": 0.2643,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22317945957183838,
+      "step": 2820,
+      "valid_targets_mean": 4256.4,
+      "valid_targets_min": 508
+    },
+    {
+      "epoch": 4.241741741741742,
+      "grad_norm": 0.5207115855205976,
+      "learning_rate": 1.6137680814843447e-05,
+      "loss": 0.2287,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26101672649383545,
+      "step": 2825,
+      "valid_targets_mean": 4533.8,
+      "valid_targets_min": 702
+    },
+    {
+      "epoch": 4.2492492492492495,
+      "grad_norm": 0.7554547871541064,
+      "learning_rate": 1.6064228799066272e-05,
+      "loss": 0.2369,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2488705813884735,
+      "step": 2830,
+      "valid_targets_mean": 3717.3,
+      "valid_targets_min": 599
+    },
+    {
+      "epoch": 4.256756756756757,
+      "grad_norm": 0.5076882374389426,
+      "learning_rate": 1.5990831966270396e-05,
+      "loss": 0.2121,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20610791444778442,
+      "step": 2835,
+      "valid_targets_mean": 5082.6,
+      "valid_targets_min": 660
+    },
+    {
+      "epoch": 4.2642642642642645,
+      "grad_norm": 0.563842648058139,
+      "learning_rate": 1.5917491345544113e-05,
+      "loss": 0.2136,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18700063228607178,
+      "step": 2840,
+      "valid_targets_mean": 4268.7,
+      "valid_targets_min": 596
+    },
+    {
+      "epoch": 4.271771771771772,
+      "grad_norm": 0.5630065305626606,
+      "learning_rate": 1.584420796518756e-05,
+      "loss": 0.244,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2228190302848816,
+      "step": 2845,
+      "valid_targets_mean": 3847.0,
+      "valid_targets_min": 865
+    },
+    {
+      "epoch": 4.2792792792792795,
+      "grad_norm": 0.5990679311909958,
+      "learning_rate": 1.5770982852698337e-05,
+      "loss": 0.2485,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2809998095035553,
+      "step": 2850,
+      "valid_targets_mean": 3885.7,
+      "valid_targets_min": 634
+    },
+    {
+      "epoch": 4.286786786786787,
+      "grad_norm": 0.5512552937482088,
+      "learning_rate": 1.5697817034757052e-05,
+      "loss": 0.2027,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1838597059249878,
+      "step": 2855,
+      "valid_targets_mean": 3917.1,
+      "valid_targets_min": 550
+    },
+    {
+      "epoch": 4.2942942942942945,
+      "grad_norm": 0.560081521292376,
+      "learning_rate": 1.5624711537212967e-05,
+      "loss": 0.2265,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20176030695438385,
+      "step": 2860,
+      "valid_targets_mean": 3504.7,
+      "valid_targets_min": 694
+    },
+    {
+      "epoch": 4.301801801801802,
+      "grad_norm": 0.7258528593213625,
+      "learning_rate": 1.5551667385069593e-05,
+      "loss": 0.2005,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22372370958328247,
+      "step": 2865,
+      "valid_targets_mean": 2295.4,
+      "valid_targets_min": 423
+    },
+    {
+      "epoch": 4.3093093093093096,
+      "grad_norm": 0.7471444365279868,
+      "learning_rate": 1.5478685602470324e-05,
+      "loss": 0.2511,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24615955352783203,
+      "step": 2870,
+      "valid_targets_mean": 3811.6,
+      "valid_targets_min": 748
+    },
+    {
+      "epoch": 4.316816816816817,
+      "grad_norm": 0.6829203066694415,
+      "learning_rate": 1.5405767212684078e-05,
+      "loss": 0.2372,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21476216614246368,
+      "step": 2875,
+      "valid_targets_mean": 3288.6,
+      "valid_targets_min": 535
+    },
+    {
+      "epoch": 4.324324324324325,
+      "grad_norm": 0.5733535380374483,
+      "learning_rate": 1.5332913238090962e-05,
+      "loss": 0.2276,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2381066530942917,
+      "step": 2880,
+      "valid_targets_mean": 3251.5,
+      "valid_targets_min": 697
+    },
+    {
+      "epoch": 4.331831831831832,
+      "grad_norm": 0.6499639228286483,
+      "learning_rate": 1.52601247001679e-05,
+      "loss": 0.2267,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23366223275661469,
+      "step": 2885,
+      "valid_targets_mean": 2987.1,
+      "valid_targets_min": 539
+    },
+    {
+      "epoch": 4.33933933933934,
+      "grad_norm": 0.6285744664120035,
+      "learning_rate": 1.5187402619474361e-05,
+      "loss": 0.2083,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.195681631565094,
+      "step": 2890,
+      "valid_targets_mean": 4077.1,
+      "valid_targets_min": 678
+    },
+    {
+      "epoch": 4.346846846846847,
+      "grad_norm": 0.6210882117570284,
+      "learning_rate": 1.5114748015638013e-05,
+      "loss": 0.2127,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21708352863788605,
+      "step": 2895,
+      "valid_targets_mean": 4107.8,
+      "valid_targets_min": 760
+    },
+    {
+      "epoch": 4.354354354354355,
+      "grad_norm": 0.5855096649034774,
+      "learning_rate": 1.5042161907340438e-05,
+      "loss": 0.2292,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25563323497772217,
+      "step": 2900,
+      "valid_targets_mean": 3565.5,
+      "valid_targets_min": 562
+    },
+    {
+      "epoch": 4.361861861861862,
+      "grad_norm": 0.5300025815920703,
+      "learning_rate": 1.4969645312302851e-05,
+      "loss": 0.2197,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22752434015274048,
+      "step": 2905,
+      "valid_targets_mean": 4113.8,
+      "valid_targets_min": 381
+    },
+    {
+      "epoch": 4.36936936936937,
+      "grad_norm": 0.5804189010231626,
+      "learning_rate": 1.4897199247271842e-05,
+      "loss": 0.2025,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1874658167362213,
+      "step": 2910,
+      "valid_targets_mean": 3798.9,
+      "valid_targets_min": 452
+    },
+    {
+      "epoch": 4.376876876876877,
+      "grad_norm": 0.48760141515061334,
+      "learning_rate": 1.4824824728005092e-05,
+      "loss": 0.2434,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.257068395614624,
+      "step": 2915,
+      "valid_targets_mean": 5025.7,
+      "valid_targets_min": 584
+    },
+    {
+      "epoch": 4.384384384384385,
+      "grad_norm": 0.658477050968013,
+      "learning_rate": 1.4752522769257152e-05,
+      "loss": 0.2147,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22656235098838806,
+      "step": 2920,
+      "valid_targets_mean": 3549.4,
+      "valid_targets_min": 624
+    },
+    {
+      "epoch": 4.391891891891892,
+      "grad_norm": 0.7492347794498558,
+      "learning_rate": 1.468029438476521e-05,
+      "loss": 0.2305,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22716817259788513,
+      "step": 2925,
+      "valid_targets_mean": 2520.4,
+      "valid_targets_min": 536
+    },
+    {
+      "epoch": 4.3993993993994,
+      "grad_norm": 0.5532580940259121,
+      "learning_rate": 1.4608140587234887e-05,
+      "loss": 0.2064,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20708701014518738,
+      "step": 2930,
+      "valid_targets_mean": 3850.1,
+      "valid_targets_min": 761
+    },
+    {
+      "epoch": 4.406906906906907,
+      "grad_norm": 0.4546505507730152,
+      "learning_rate": 1.453606238832602e-05,
+      "loss": 0.2218,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21117320656776428,
+      "step": 2935,
+      "valid_targets_mean": 6005.3,
+      "valid_targets_min": 813
+    },
+    {
+      "epoch": 4.414414414414415,
+      "grad_norm": 0.7252302746614107,
+      "learning_rate": 1.4464060798638484e-05,
+      "loss": 0.2291,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18549787998199463,
+      "step": 2940,
+      "valid_targets_mean": 3605.3,
+      "valid_targets_min": 847
+    },
+    {
+      "epoch": 4.421921921921922,
+      "grad_norm": 0.43525814011469965,
+      "learning_rate": 1.4392136827698032e-05,
+      "loss": 0.1929,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16544996201992035,
+      "step": 2945,
+      "valid_targets_mean": 5133.1,
+      "valid_targets_min": 591
+    },
+    {
+      "epoch": 4.42942942942943,
+      "grad_norm": 0.6290052391762472,
+      "learning_rate": 1.4320291483942135e-05,
+      "loss": 0.2245,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21728135645389557,
+      "step": 2950,
+      "valid_targets_mean": 3071.6,
+      "valid_targets_min": 601
+    },
+    {
+      "epoch": 4.436936936936937,
+      "grad_norm": 0.568290272425125,
+      "learning_rate": 1.424852577470584e-05,
+      "loss": 0.2298,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18296922743320465,
+      "step": 2955,
+      "valid_targets_mean": 5362.2,
+      "valid_targets_min": 803
+    },
+    {
+      "epoch": 4.444444444444445,
+      "grad_norm": 0.603259783928807,
+      "learning_rate": 1.417684070620764e-05,
+      "loss": 0.2261,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23192979395389557,
+      "step": 2960,
+      "valid_targets_mean": 3260.2,
+      "valid_targets_min": 713
+    },
+    {
+      "epoch": 4.451951951951952,
+      "grad_norm": 0.5712898946994219,
+      "learning_rate": 1.4105237283535376e-05,
+      "loss": 0.207,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2266012728214264,
+      "step": 2965,
+      "valid_targets_mean": 4215.6,
+      "valid_targets_min": 691
+    },
+    {
+      "epoch": 4.45945945945946,
+      "grad_norm": 0.6445579585402789,
+      "learning_rate": 1.403371651063216e-05,
+      "loss": 0.2363,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24278521537780762,
+      "step": 2970,
+      "valid_targets_mean": 3313.9,
+      "valid_targets_min": 374
+    },
+    {
+      "epoch": 4.466966966966967,
+      "grad_norm": 0.5733027694106058,
+      "learning_rate": 1.3962279390282261e-05,
+      "loss": 0.243,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2066403329372406,
+      "step": 2975,
+      "valid_targets_mean": 4162.3,
+      "valid_targets_min": 764
+    },
+    {
+      "epoch": 4.474474474474475,
+      "grad_norm": 0.613924525173538,
+      "learning_rate": 1.3890926924097071e-05,
+      "loss": 0.1938,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20435193181037903,
+      "step": 2980,
+      "valid_targets_mean": 3513.4,
+      "valid_targets_min": 652
+    },
+    {
+      "epoch": 4.481981981981982,
+      "grad_norm": 0.5403888711125328,
+      "learning_rate": 1.3819660112501054e-05,
+      "loss": 0.2242,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20203976333141327,
+      "step": 2985,
+      "valid_targets_mean": 4108.9,
+      "valid_targets_min": 655
+    },
+    {
+      "epoch": 4.48948948948949,
+      "grad_norm": 0.5594472556122174,
+      "learning_rate": 1.3748479954717735e-05,
+      "loss": 0.2462,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20039498805999756,
+      "step": 2990,
+      "valid_targets_mean": 3769.4,
+      "valid_targets_min": 504
+    },
+    {
+      "epoch": 4.496996996996997,
+      "grad_norm": 0.5426786497323537,
+      "learning_rate": 1.3677387448755657e-05,
+      "loss": 0.2068,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21668657660484314,
+      "step": 2995,
+      "valid_targets_mean": 3945.2,
+      "valid_targets_min": 667
+    },
+    {
+      "epoch": 4.504504504504505,
+      "grad_norm": 0.4470248026304529,
+      "learning_rate": 1.360638359139442e-05,
+      "loss": 0.2203,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2280035763978958,
+      "step": 3000,
+      "valid_targets_mean": 5678.4,
+      "valid_targets_min": 1551
+    },
+    {
+      "epoch": 4.512012012012012,
+      "grad_norm": 0.609794314414809,
+      "learning_rate": 1.3535469378170683e-05,
+      "loss": 0.239,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2744963467121124,
+      "step": 3005,
+      "valid_targets_mean": 3750.4,
+      "valid_targets_min": 674
+    },
+    {
+      "epoch": 4.51951951951952,
+      "grad_norm": 0.4627448347366898,
+      "learning_rate": 1.3464645803364228e-05,
+      "loss": 0.2439,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21215581893920898,
+      "step": 3010,
+      "valid_targets_mean": 5590.1,
+      "valid_targets_min": 528
+    },
+    {
+      "epoch": 4.527027027027027,
+      "grad_norm": 0.6035357299819745,
+      "learning_rate": 1.3393913859983996e-05,
+      "loss": 0.2138,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2198573499917984,
+      "step": 3015,
+      "valid_targets_mean": 6302.9,
+      "valid_targets_min": 789
+    },
+    {
+      "epoch": 4.534534534534535,
+      "grad_norm": 0.5643676209847303,
+      "learning_rate": 1.3323274539754177e-05,
+      "loss": 0.2198,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25696855783462524,
+      "step": 3020,
+      "valid_targets_mean": 4303.8,
+      "valid_targets_min": 634
+    },
+    {
+      "epoch": 4.542042042042042,
+      "grad_norm": 0.570048922978165,
+      "learning_rate": 1.3252728833100296e-05,
+      "loss": 0.2203,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19746547937393188,
+      "step": 3025,
+      "valid_targets_mean": 3560.8,
+      "valid_targets_min": 598
+    },
+    {
+      "epoch": 4.54954954954955,
+      "grad_norm": 0.4464417263684137,
+      "learning_rate": 1.3182277729135358e-05,
+      "loss": 0.2287,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19879616796970367,
+      "step": 3030,
+      "valid_targets_mean": 7131.6,
+      "valid_targets_min": 460
+    },
+    {
+      "epoch": 4.557057057057057,
+      "grad_norm": 0.5089601591681008,
+      "learning_rate": 1.3111922215645922e-05,
+      "loss": 0.2126,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23360319435596466,
+      "step": 3035,
+      "valid_targets_mean": 4634.0,
+      "valid_targets_min": 654
+    },
+    {
+      "epoch": 4.564564564564565,
+      "grad_norm": 0.6823720809640251,
+      "learning_rate": 1.3041663279078311e-05,
+      "loss": 0.2104,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2097683995962143,
+      "step": 3040,
+      "valid_targets_mean": 2433.6,
+      "valid_targets_min": 562
+    },
+    {
+      "epoch": 4.572072072072072,
+      "grad_norm": 0.6700606322768459,
+      "learning_rate": 1.297150190452473e-05,
+      "loss": 0.2051,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2061670869588852,
+      "step": 3045,
+      "valid_targets_mean": 2720.6,
+      "valid_targets_min": 529
+    },
+    {
+      "epoch": 4.57957957957958,
+      "grad_norm": 0.6228866174244457,
+      "learning_rate": 1.2901439075709506e-05,
+      "loss": 0.2165,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18749915063381195,
+      "step": 3050,
+      "valid_targets_mean": 3271.4,
+      "valid_targets_min": 416
+    },
+    {
+      "epoch": 4.587087087087087,
+      "grad_norm": 0.600075319444778,
+      "learning_rate": 1.2831475774975247e-05,
+      "loss": 0.2088,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24314680695533752,
+      "step": 3055,
+      "valid_targets_mean": 4222.2,
+      "valid_targets_min": 520
+    },
+    {
+      "epoch": 4.594594594594595,
+      "grad_norm": 0.4219894648635082,
+      "learning_rate": 1.2761612983269094e-05,
+      "loss": 0.2195,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18051698803901672,
+      "step": 3060,
+      "valid_targets_mean": 6413.9,
+      "valid_targets_min": 703
+    },
+    {
+      "epoch": 4.602102102102102,
+      "grad_norm": 0.5430062852225149,
+      "learning_rate": 1.2691851680128966e-05,
+      "loss": 0.2119,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18320244550704956,
+      "step": 3065,
+      "valid_targets_mean": 3880.8,
+      "valid_targets_min": 727
+    },
+    {
+      "epoch": 4.60960960960961,
+      "grad_norm": 0.4910324433811435,
+      "learning_rate": 1.2622192843669826e-05,
+      "loss": 0.2111,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1959684193134308,
+      "step": 3070,
+      "valid_targets_mean": 5296.5,
+      "valid_targets_min": 512
+    },
+    {
+      "epoch": 4.617117117117117,
+      "grad_norm": 0.5134831494706088,
+      "learning_rate": 1.255263745056996e-05,
+      "loss": 0.2051,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21355494856834412,
+      "step": 3075,
+      "valid_targets_mean": 5030.8,
+      "valid_targets_min": 469
+    },
+    {
+      "epoch": 4.624624624624625,
+      "grad_norm": 0.5527873198315599,
+      "learning_rate": 1.2483186476057281e-05,
+      "loss": 0.2181,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1989469826221466,
+      "step": 3080,
+      "valid_targets_mean": 3602.9,
+      "valid_targets_min": 630
+    },
+    {
+      "epoch": 4.632132132132132,
+      "grad_norm": 0.7003658441849595,
+      "learning_rate": 1.2413840893895668e-05,
+      "loss": 0.1931,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20069898664951324,
+      "step": 3085,
+      "valid_targets_mean": 3747.1,
+      "valid_targets_min": 474
+    },
+    {
+      "epoch": 4.63963963963964,
+      "grad_norm": 0.5774616260959406,
+      "learning_rate": 1.2344601676371312e-05,
+      "loss": 0.2107,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2032512128353119,
+      "step": 3090,
+      "valid_targets_mean": 5143.7,
+      "valid_targets_min": 1085
+    },
+    {
+      "epoch": 4.647147147147147,
+      "grad_norm": 0.609625942880746,
+      "learning_rate": 1.2275469794279068e-05,
+      "loss": 0.2376,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24463210999965668,
+      "step": 3095,
+      "valid_targets_mean": 3302.1,
+      "valid_targets_min": 211
+    },
+    {
+      "epoch": 4.654654654654655,
+      "grad_norm": 0.5539889957775487,
+      "learning_rate": 1.220644621690885e-05,
+      "loss": 0.2184,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19467709958553314,
+      "step": 3100,
+      "valid_targets_mean": 3723.1,
+      "valid_targets_min": 637
+    },
+    {
+      "epoch": 4.662162162162162,
+      "grad_norm": 0.5324494392088087,
+      "learning_rate": 1.2137531912032058e-05,
+      "loss": 0.2179,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2482069581747055,
+      "step": 3105,
+      "valid_targets_mean": 4753.6,
+      "valid_targets_min": 634
+    },
+    {
+      "epoch": 4.66966966966967,
+      "grad_norm": 0.5048180559075982,
+      "learning_rate": 1.2068727845887995e-05,
+      "loss": 0.2118,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24020060896873474,
+      "step": 3110,
+      "valid_targets_mean": 5051.0,
+      "valid_targets_min": 462
+    },
+    {
+      "epoch": 4.677177177177177,
+      "grad_norm": 0.4832724873118443,
+      "learning_rate": 1.2000034983170305e-05,
+      "loss": 0.2357,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22151488065719604,
+      "step": 3115,
+      "valid_targets_mean": 5193.5,
+      "valid_targets_min": 674
+    },
+    {
+      "epoch": 4.684684684684685,
+      "grad_norm": 0.5679523589750985,
+      "learning_rate": 1.193145428701347e-05,
+      "loss": 0.1978,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2017378807067871,
+      "step": 3120,
+      "valid_targets_mean": 4752.9,
+      "valid_targets_min": 764
+    },
+    {
+      "epoch": 4.692192192192192,
+      "grad_norm": 0.8966754020165514,
+      "learning_rate": 1.18629867189793e-05,
+      "loss": 0.2302,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24184086918830872,
+      "step": 3125,
+      "valid_targets_mean": 2982.5,
+      "valid_targets_min": 515
+    },
+    {
+      "epoch": 4.6996996996997,
+      "grad_norm": 0.49411819508382815,
+      "learning_rate": 1.1794633239043449e-05,
+      "loss": 0.2028,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2202903926372528,
+      "step": 3130,
+      "valid_targets_mean": 4615.9,
+      "valid_targets_min": 738
+    },
+    {
+      "epoch": 4.707207207207207,
+      "grad_norm": 0.4928998645668594,
+      "learning_rate": 1.1726394805581957e-05,
+      "loss": 0.2398,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21853674948215485,
+      "step": 3135,
+      "valid_targets_mean": 5042.8,
+      "valid_targets_min": 656
+    },
+    {
+      "epoch": 4.714714714714715,
+      "grad_norm": 0.5803190883509034,
+      "learning_rate": 1.1658272375357797e-05,
+      "loss": 0.2426,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24375799298286438,
+      "step": 3140,
+      "valid_targets_mean": 4044.2,
+      "valid_targets_min": 525
+    },
+    {
+      "epoch": 4.722222222222222,
+      "grad_norm": 0.5795814998007034,
+      "learning_rate": 1.1590266903507499e-05,
+      "loss": 0.2371,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2230663299560547,
+      "step": 3145,
+      "valid_targets_mean": 4489.6,
+      "valid_targets_min": 930
+    },
+    {
+      "epoch": 4.72972972972973,
+      "grad_norm": 0.6742257430638947,
+      "learning_rate": 1.1522379343527708e-05,
+      "loss": 0.2535,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23786276578903198,
+      "step": 3150,
+      "valid_targets_mean": 4348.9,
+      "valid_targets_min": 534
+    },
+    {
+      "epoch": 4.737237237237237,
+      "grad_norm": 0.49627665219198364,
+      "learning_rate": 1.1454610647261859e-05,
+      "loss": 0.2174,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2170691192150116,
+      "step": 3155,
+      "valid_targets_mean": 4579.3,
+      "valid_targets_min": 701
+    },
+    {
+      "epoch": 4.744744744744745,
+      "grad_norm": 0.4965773011218609,
+      "learning_rate": 1.1386961764886806e-05,
+      "loss": 0.2299,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21680960059165955,
+      "step": 3160,
+      "valid_targets_mean": 4751.6,
+      "valid_targets_min": 970
+    },
+    {
+      "epoch": 4.752252252252252,
+      "grad_norm": 0.46564876478736006,
+      "learning_rate": 1.1319433644899496e-05,
+      "loss": 0.2346,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20024517178535461,
+      "step": 3165,
+      "valid_targets_mean": 5608.7,
+      "valid_targets_min": 565
+    },
+    {
+      "epoch": 4.75975975975976,
+      "grad_norm": 0.8219903727111237,
+      "learning_rate": 1.1252027234103709e-05,
+      "loss": 0.2175,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21334412693977356,
+      "step": 3170,
+      "valid_targets_mean": 3558.3,
+      "valid_targets_min": 684
+    },
+    {
+      "epoch": 4.767267267267267,
+      "grad_norm": 0.5331331638771257,
+      "learning_rate": 1.118474347759673e-05,
+      "loss": 0.2032,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23813585937023163,
+      "step": 3175,
+      "valid_targets_mean": 4562.6,
+      "valid_targets_min": 700
+    },
+    {
+      "epoch": 4.774774774774775,
+      "grad_norm": 0.7204879070632291,
+      "learning_rate": 1.1117583318756128e-05,
+      "loss": 0.2218,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21081534028053284,
+      "step": 3180,
+      "valid_targets_mean": 4559.9,
+      "valid_targets_min": 347
+    },
+    {
+      "epoch": 4.782282282282282,
+      "grad_norm": 0.6342862982793008,
+      "learning_rate": 1.1050547699226522e-05,
+      "loss": 0.2234,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21464157104492188,
+      "step": 3185,
+      "valid_targets_mean": 4187.8,
+      "valid_targets_min": 611
+    },
+    {
+      "epoch": 4.78978978978979,
+      "grad_norm": 0.5094911353991051,
+      "learning_rate": 1.09836375589064e-05,
+      "loss": 0.2126,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21020804345607758,
+      "step": 3190,
+      "valid_targets_mean": 5157.7,
+      "valid_targets_min": 1150
+    },
+    {
+      "epoch": 4.797297297297297,
+      "grad_norm": 0.47526721126268623,
+      "learning_rate": 1.0916853835934891e-05,
+      "loss": 0.2157,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22524023056030273,
+      "step": 3195,
+      "valid_targets_mean": 5497.2,
+      "valid_targets_min": 714
+    },
+    {
+      "epoch": 4.804804804804805,
+      "grad_norm": 0.7451217907444951,
+      "learning_rate": 1.0850197466678662e-05,
+      "loss": 0.232,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20764270424842834,
+      "step": 3200,
+      "valid_targets_mean": 4149.2,
+      "valid_targets_min": 628
+    },
+    {
+      "epoch": 4.812312312312312,
+      "grad_norm": 0.4957698184650846,
+      "learning_rate": 1.0783669385718762e-05,
+      "loss": 0.2371,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21627801656723022,
+      "step": 3205,
+      "valid_targets_mean": 5803.0,
+      "valid_targets_min": 790
+    },
+    {
+      "epoch": 4.81981981981982,
+      "grad_norm": 0.5309967929446427,
+      "learning_rate": 1.0717270525837523e-05,
+      "loss": 0.2449,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2549176514148712,
+      "step": 3210,
+      "valid_targets_mean": 5175.2,
+      "valid_targets_min": 558
+    },
+    {
+      "epoch": 4.827327327327327,
+      "grad_norm": 0.6675846311975975,
+      "learning_rate": 1.0651001818005487e-05,
+      "loss": 0.2157,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2339574694633484,
+      "step": 3215,
+      "valid_targets_mean": 3242.3,
+      "valid_targets_min": 499
+    },
+    {
+      "epoch": 4.834834834834835,
+      "grad_norm": 0.5348472304451167,
+      "learning_rate": 1.0584864191368345e-05,
+      "loss": 0.2287,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2503393888473511,
+      "step": 3220,
+      "valid_targets_mean": 5664.9,
+      "valid_targets_min": 487
+    },
+    {
+      "epoch": 4.842342342342342,
+      "grad_norm": 0.5338144959827715,
+      "learning_rate": 1.0518858573233911e-05,
+      "loss": 0.2171,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21501460671424866,
+      "step": 3225,
+      "valid_targets_mean": 4275.4,
+      "valid_targets_min": 631
+    },
+    {
+      "epoch": 4.84984984984985,
+      "grad_norm": 0.583867201837148,
+      "learning_rate": 1.0452985889059151e-05,
+      "loss": 0.2321,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2584874629974365,
+      "step": 3230,
+      "valid_targets_mean": 3758.3,
+      "valid_targets_min": 367
+    },
+    {
+      "epoch": 4.857357357357357,
+      "grad_norm": 0.4704381431905369,
+      "learning_rate": 1.0387247062437144e-05,
+      "loss": 0.2159,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2556239068508148,
+      "step": 3235,
+      "valid_targets_mean": 5887.3,
+      "valid_targets_min": 489
+    },
+    {
+      "epoch": 4.864864864864865,
+      "grad_norm": 0.6249528070324213,
+      "learning_rate": 1.0321643015084187e-05,
+      "loss": 0.2111,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23891428112983704,
+      "step": 3240,
+      "valid_targets_mean": 3309.9,
+      "valid_targets_min": 301
+    },
+    {
+      "epoch": 4.872372372372372,
+      "grad_norm": 0.44764920429718186,
+      "learning_rate": 1.0256174666826841e-05,
+      "loss": 0.213,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19511452317237854,
+      "step": 3245,
+      "valid_targets_mean": 5169.8,
+      "valid_targets_min": 980
+    },
+    {
+      "epoch": 4.87987987987988,
+      "grad_norm": 0.5340956174536189,
+      "learning_rate": 1.0190842935589065e-05,
+      "loss": 0.206,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18626420199871063,
+      "step": 3250,
+      "valid_targets_mean": 3967.4,
+      "valid_targets_min": 798
+    },
+    {
+      "epoch": 4.887387387387387,
+      "grad_norm": 0.56288269038886,
+      "learning_rate": 1.0125648737379307e-05,
+      "loss": 0.2185,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20922933518886566,
+      "step": 3255,
+      "valid_targets_mean": 4145.1,
+      "valid_targets_min": 511
+    },
+    {
+      "epoch": 4.894894894894895,
+      "grad_norm": 0.4659624202841574,
+      "learning_rate": 1.0060592986277693e-05,
+      "loss": 0.2216,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18871676921844482,
+      "step": 3260,
+      "valid_targets_mean": 5421.9,
+      "valid_targets_min": 2150
+    },
+    {
+      "epoch": 4.902402402402402,
+      "grad_norm": 0.5691277577002785,
+      "learning_rate": 9.995676594423186e-06,
+      "loss": 0.2241,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22213050723075867,
+      "step": 3265,
+      "valid_targets_mean": 4555.8,
+      "valid_targets_min": 839
+    },
+    {
+      "epoch": 4.90990990990991,
+      "grad_norm": 0.6445683801786387,
+      "learning_rate": 9.930900472000834e-06,
+      "loss": 0.1918,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20591846108436584,
+      "step": 3270,
+      "valid_targets_mean": 3039.4,
+      "valid_targets_min": 574
+    },
+    {
+      "epoch": 4.9174174174174174,
+      "grad_norm": 0.48632405588473493,
+      "learning_rate": 9.866265527228961e-06,
+      "loss": 0.2047,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.224864199757576,
+      "step": 3275,
+      "valid_targets_mean": 5041.8,
+      "valid_targets_min": 562
+    },
+    {
+      "epoch": 4.924924924924925,
+      "grad_norm": 0.4551299572592259,
+      "learning_rate": 9.801772666346462e-06,
+      "loss": 0.2383,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17636257410049438,
+      "step": 3280,
+      "valid_targets_mean": 5264.3,
+      "valid_targets_min": 260
+    },
+    {
+      "epoch": 4.9324324324324325,
+      "grad_norm": 0.5303503051433811,
+      "learning_rate": 9.737422793600092e-06,
+      "loss": 0.2529,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2605491876602173,
+      "step": 3285,
+      "valid_targets_mean": 4923.2,
+      "valid_targets_min": 551
+    },
+    {
+      "epoch": 4.93993993993994,
+      "grad_norm": 0.5619328215648045,
+      "learning_rate": 9.67321681123179e-06,
+      "loss": 0.2219,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21425651013851166,
+      "step": 3290,
+      "valid_targets_mean": 4545.6,
+      "valid_targets_min": 515
+    },
+    {
+      "epoch": 4.9474474474474475,
+      "grad_norm": 0.4969814195952852,
+      "learning_rate": 9.609155619466016e-06,
+      "loss": 0.202,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19319353997707367,
+      "step": 3295,
+      "valid_targets_mean": 4383.9,
+      "valid_targets_min": 711
+    },
+    {
+      "epoch": 4.954954954954955,
+      "grad_norm": 0.6244151395343142,
+      "learning_rate": 9.545240116497143e-06,
+      "loss": 0.2181,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18449707329273224,
+      "step": 3300,
+      "valid_targets_mean": 3211.4,
+      "valid_targets_min": 484
+    },
+    {
+      "epoch": 4.9624624624624625,
+      "grad_norm": 0.5893223235253725,
+      "learning_rate": 9.481471198476855e-06,
+      "loss": 0.2294,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22556568682193756,
+      "step": 3305,
+      "valid_targets_mean": 5031.6,
+      "valid_targets_min": 684
+    },
+    {
+      "epoch": 4.96996996996997,
+      "grad_norm": 0.4915902916242456,
+      "learning_rate": 9.417849759501603e-06,
+      "loss": 0.2226,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2357444167137146,
+      "step": 3310,
+      "valid_targets_mean": 5379.2,
+      "valid_targets_min": 860
+    },
+    {
+      "epoch": 4.9774774774774775,
+      "grad_norm": 0.5245483991646102,
+      "learning_rate": 9.354376691600034e-06,
+      "loss": 0.2091,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18305274844169617,
+      "step": 3315,
+      "valid_targets_mean": 4195.8,
+      "valid_targets_min": 388
+    },
+    {
+      "epoch": 4.984984984984985,
+      "grad_norm": 0.49053726456598334,
+      "learning_rate": 9.2910528847205e-06,
+      "loss": 0.2419,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23515072464942932,
+      "step": 3320,
+      "valid_targets_mean": 4802.1,
+      "valid_targets_min": 666
+    },
+    {
+      "epoch": 4.9924924924924925,
+      "grad_norm": 0.5458070078892359,
+      "learning_rate": 9.227879226718595e-06,
+      "loss": 0.2204,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2729540467262268,
+      "step": 3325,
+      "valid_targets_mean": 4229.2,
+      "valid_targets_min": 716
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.4617612878592164,
+      "learning_rate": 9.164856603344681e-06,
+      "loss": 0.2209,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19838783144950867,
+      "step": 3330,
+      "valid_targets_mean": 5383.6,
+      "valid_targets_min": 1005
+    },
+    {
+      "epoch": 5.0075075075075075,
+      "grad_norm": 0.5006372512661854,
+      "learning_rate": 9.10198589823149e-06,
+      "loss": 0.2043,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21871143579483032,
+      "step": 3335,
+      "valid_targets_mean": 4899.0,
+      "valid_targets_min": 478
+    },
+    {
+      "epoch": 5.015015015015015,
+      "grad_norm": 0.493071836759497,
+      "learning_rate": 9.039267992881724e-06,
+      "loss": 0.1867,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15860621631145477,
+      "step": 3340,
+      "valid_targets_mean": 4637.3,
+      "valid_targets_min": 539
+    },
+    {
+      "epoch": 5.0225225225225225,
+      "grad_norm": 0.5942000008682329,
+      "learning_rate": 8.97670376665569e-06,
+      "loss": 0.212,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21339187026023865,
+      "step": 3345,
+      "valid_targets_mean": 4111.1,
+      "valid_targets_min": 482
+    },
+    {
+      "epoch": 5.03003003003003,
+      "grad_norm": 0.5098653931776418,
+      "learning_rate": 8.914294096758995e-06,
+      "loss": 0.2524,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21369382739067078,
+      "step": 3350,
+      "valid_targets_mean": 5529.7,
+      "valid_targets_min": 571
+    },
+    {
+      "epoch": 5.0375375375375375,
+      "grad_norm": 0.48801798277011926,
+      "learning_rate": 8.852039858230217e-06,
+      "loss": 0.2076,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16710881888866425,
+      "step": 3355,
+      "valid_targets_mean": 4623.7,
+      "valid_targets_min": 694
+    },
+    {
+      "epoch": 5.045045045045045,
+      "grad_norm": 0.5406142033120225,
+      "learning_rate": 8.78994192392865e-06,
+      "loss": 0.1969,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18688131868839264,
+      "step": 3360,
+      "valid_targets_mean": 3961.8,
+      "valid_targets_min": 529
+    },
+    {
+      "epoch": 5.0525525525525525,
+      "grad_norm": 0.49064863606657333,
+      "learning_rate": 8.728001164522068e-06,
+      "loss": 0.2054,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19211602210998535,
+      "step": 3365,
+      "valid_targets_mean": 5132.5,
+      "valid_targets_min": 557
+    },
+    {
+      "epoch": 5.06006006006006,
+      "grad_norm": 0.5428647141950651,
+      "learning_rate": 8.666218448474517e-06,
+      "loss": 0.2235,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19043126702308655,
+      "step": 3370,
+      "valid_targets_mean": 4518.9,
+      "valid_targets_min": 623
+    },
+    {
+      "epoch": 5.0675675675675675,
+      "grad_norm": 0.5285735991494391,
+      "learning_rate": 8.604594642034126e-06,
+      "loss": 0.2093,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22566662728786469,
+      "step": 3375,
+      "valid_targets_mean": 4554.3,
+      "valid_targets_min": 628
+    },
+    {
+      "epoch": 5.075075075075075,
+      "grad_norm": 0.6397497283612681,
+      "learning_rate": 8.543130609220982e-06,
+      "loss": 0.1913,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17037567496299744,
+      "step": 3380,
+      "valid_targets_mean": 3197.0,
+      "valid_targets_min": 631
+    },
+    {
+      "epoch": 5.0825825825825826,
+      "grad_norm": 0.5624772140252701,
+      "learning_rate": 8.481827211814991e-06,
+      "loss": 0.2315,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2726784348487854,
+      "step": 3385,
+      "valid_targets_mean": 4160.6,
+      "valid_targets_min": 727
+    },
+    {
+      "epoch": 5.09009009009009,
+      "grad_norm": 0.5877366129284431,
+      "learning_rate": 8.420685309343835e-06,
+      "loss": 0.2131,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20283541083335876,
+      "step": 3390,
+      "valid_targets_mean": 3707.2,
+      "valid_targets_min": 663
+    },
+    {
+      "epoch": 5.097597597597598,
+      "grad_norm": 0.6489090565754443,
+      "learning_rate": 8.35970575907087e-06,
+      "loss": 0.2123,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2192349135875702,
+      "step": 3395,
+      "valid_targets_mean": 5407.1,
+      "valid_targets_min": 672
+    },
+    {
+      "epoch": 5.105105105105105,
+      "grad_norm": 0.5565202403018633,
+      "learning_rate": 8.298889415983142e-06,
+      "loss": 0.2057,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1590254306793213,
+      "step": 3400,
+      "valid_targets_mean": 3583.3,
+      "valid_targets_min": 735
+    },
+    {
+      "epoch": 5.112612612612613,
+      "grad_norm": 0.41960211523789304,
+      "learning_rate": 8.23823713277938e-06,
+      "loss": 0.2056,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.14446228742599487,
+      "step": 3405,
+      "valid_targets_mean": 5082.1,
+      "valid_targets_min": 276
+    },
+    {
+      "epoch": 5.12012012012012,
+      "grad_norm": 0.6517049110770983,
+      "learning_rate": 8.17774975985806e-06,
+      "loss": 0.207,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23023463785648346,
+      "step": 3410,
+      "valid_targets_mean": 4596.8,
+      "valid_targets_min": 786
+    },
+    {
+      "epoch": 5.127627627627628,
+      "grad_norm": 0.8256063611735115,
+      "learning_rate": 8.117428145305464e-06,
+      "loss": 0.1705,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19371360540390015,
+      "step": 3415,
+      "valid_targets_mean": 3733.1,
+      "valid_targets_min": 946
+    },
+    {
+      "epoch": 5.135135135135135,
+      "grad_norm": 0.47817908036094603,
+      "learning_rate": 8.057273134883794e-06,
+      "loss": 0.2065,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20559421181678772,
+      "step": 3420,
+      "valid_targets_mean": 5914.1,
+      "valid_targets_min": 746
+    },
+    {
+      "epoch": 5.142642642642643,
+      "grad_norm": 0.6404035798093046,
+      "learning_rate": 7.997285572019316e-06,
+      "loss": 0.2294,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19509994983673096,
+      "step": 3425,
+      "valid_targets_mean": 3609.6,
+      "valid_targets_min": 719
+    },
+    {
+      "epoch": 5.15015015015015,
+      "grad_norm": 0.7814615076086997,
+      "learning_rate": 7.937466297790542e-06,
+      "loss": 0.2033,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17219263315200806,
+      "step": 3430,
+      "valid_targets_mean": 4124.9,
+      "valid_targets_min": 593
+    },
+    {
+      "epoch": 5.157657657657658,
+      "grad_norm": 0.5711178140418124,
+      "learning_rate": 7.877816150916422e-06,
+      "loss": 0.2186,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20192334055900574,
+      "step": 3435,
+      "valid_targets_mean": 4497.1,
+      "valid_targets_min": 673
+    },
+    {
+      "epoch": 5.165165165165165,
+      "grad_norm": 0.6605000045012411,
+      "learning_rate": 7.81833596774459e-06,
+      "loss": 0.1922,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21957352757453918,
+      "step": 3440,
+      "valid_targets_mean": 3223.1,
+      "valid_targets_min": 617
+    },
+    {
+      "epoch": 5.172672672672673,
+      "grad_norm": 1.2228952471867771,
+      "learning_rate": 7.759026582239639e-06,
+      "loss": 0.1919,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2631041407585144,
+      "step": 3445,
+      "valid_targets_mean": 4348.8,
+      "valid_targets_min": 574
+    },
+    {
+      "epoch": 5.18018018018018,
+      "grad_norm": 0.858620752436695,
+      "learning_rate": 7.699888825971433e-06,
+      "loss": 0.2026,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.162127286195755,
+      "step": 3450,
+      "valid_targets_mean": 6938.4,
+      "valid_targets_min": 683
+    },
+    {
+      "epoch": 5.187687687687688,
+      "grad_norm": 0.6070759956430632,
+      "learning_rate": 7.640923528103441e-06,
+      "loss": 0.2168,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2934483289718628,
+      "step": 3455,
+      "valid_targets_mean": 4662.9,
+      "valid_targets_min": 739
+    },
+    {
+      "epoch": 5.195195195195195,
+      "grad_norm": 0.5186921298999412,
+      "learning_rate": 7.582131515381108e-06,
+      "loss": 0.203,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2093011438846588,
+      "step": 3460,
+      "valid_targets_mean": 4782.7,
+      "valid_targets_min": 492
+    },
+    {
+      "epoch": 5.202702702702703,
+      "grad_norm": 0.5661778349613429,
+      "learning_rate": 7.5235136121202675e-06,
+      "loss": 0.2171,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24795249104499817,
+      "step": 3465,
+      "valid_targets_mean": 4599.2,
+      "valid_targets_min": 324
+    },
+    {
+      "epoch": 5.21021021021021,
+      "grad_norm": 0.5360911975296877,
+      "learning_rate": 7.465070640195606e-06,
+      "loss": 0.2068,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26138007640838623,
+      "step": 3470,
+      "valid_targets_mean": 4960.6,
+      "valid_targets_min": 376
+    },
+    {
+      "epoch": 5.217717717717718,
+      "grad_norm": 0.5642393094183789,
+      "learning_rate": 7.406803419029094e-06,
+      "loss": 0.2323,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24130575358867645,
+      "step": 3475,
+      "valid_targets_mean": 4966.7,
+      "valid_targets_min": 821
+    },
+    {
+      "epoch": 5.225225225225225,
+      "grad_norm": 0.4550076790955841,
+      "learning_rate": 7.3487127655785295e-06,
+      "loss": 0.2106,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1952485740184784,
+      "step": 3480,
+      "valid_targets_mean": 5512.1,
+      "valid_targets_min": 622
+    },
+    {
+      "epoch": 5.232732732732733,
+      "grad_norm": 0.6885372734490723,
+      "learning_rate": 7.290799494326077e-06,
+      "loss": 0.2016,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16893789172172546,
+      "step": 3485,
+      "valid_targets_mean": 4656.0,
+      "valid_targets_min": 477
+    },
+    {
+      "epoch": 5.24024024024024,
+      "grad_norm": 0.5045869253443142,
+      "learning_rate": 7.233064417266846e-06,
+      "loss": 0.231,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16010183095932007,
+      "step": 3490,
+      "valid_targets_mean": 3909.6,
+      "valid_targets_min": 609
+    },
+    {
+      "epoch": 5.247747747747748,
+      "grad_norm": 0.6679277663347103,
+      "learning_rate": 7.1755083438975056e-06,
+      "loss": 0.22,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2579156756401062,
+      "step": 3495,
+      "valid_targets_mean": 3035.2,
+      "valid_targets_min": 410
+    },
+    {
+      "epoch": 5.255255255255255,
+      "grad_norm": 0.48166787337419537,
+      "learning_rate": 7.118132081204936e-06,
+      "loss": 0.2034,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17768923938274384,
+      "step": 3500,
+      "valid_targets_mean": 4611.5,
+      "valid_targets_min": 540
+    },
+    {
+      "epoch": 5.262762762762763,
+      "grad_norm": 0.6437356623831512,
+      "learning_rate": 7.060936433654901e-06,
+      "loss": 0.2106,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22873665392398834,
+      "step": 3505,
+      "valid_targets_mean": 4224.2,
+      "valid_targets_min": 1214
+    },
+    {
+      "epoch": 5.27027027027027,
+      "grad_norm": 0.5041184297812231,
+      "learning_rate": 7.003922203180813e-06,
+      "loss": 0.1829,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17131251096725464,
+      "step": 3510,
+      "valid_targets_mean": 4996.4,
+      "valid_targets_min": 762
+    },
+    {
+      "epoch": 5.277777777777778,
+      "grad_norm": 0.42636959404581054,
+      "learning_rate": 6.947090189172425e-06,
+      "loss": 0.1905,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17408744990825653,
+      "step": 3515,
+      "valid_targets_mean": 6548.1,
+      "valid_targets_min": 681
+    },
+    {
+      "epoch": 5.285285285285285,
+      "grad_norm": 0.7004634661839949,
+      "learning_rate": 6.89044118846467e-06,
+      "loss": 0.1871,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21478641033172607,
+      "step": 3520,
+      "valid_targets_mean": 3402.0,
+      "valid_targets_min": 534
+    },
+    {
+      "epoch": 5.292792792792793,
+      "grad_norm": 0.6584020869113968,
+      "learning_rate": 6.833975995326465e-06,
+      "loss": 0.2226,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17992106080055237,
+      "step": 3525,
+      "valid_targets_mean": 3046.6,
+      "valid_targets_min": 592
+    },
+    {
+      "epoch": 5.3003003003003,
+      "grad_norm": 0.505387458907387,
+      "learning_rate": 6.777695401449591e-06,
+      "loss": 0.1905,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20670431852340698,
+      "step": 3530,
+      "valid_targets_mean": 5285.2,
+      "valid_targets_min": 893
+    },
+    {
+      "epoch": 5.307807807807808,
+      "grad_norm": 0.5767720272580488,
+      "learning_rate": 6.721600195937578e-06,
+      "loss": 0.1898,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18273615837097168,
+      "step": 3535,
+      "valid_targets_mean": 3702.8,
+      "valid_targets_min": 877
+    },
+    {
+      "epoch": 5.315315315315315,
+      "grad_norm": 0.5468384364940457,
+      "learning_rate": 6.665691165294654e-06,
+      "loss": 0.2308,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2389727234840393,
+      "step": 3540,
+      "valid_targets_mean": 4842.5,
+      "valid_targets_min": 1056
+    },
+    {
+      "epoch": 5.322822822822823,
+      "grad_norm": 0.5417666029457835,
+      "learning_rate": 6.609969093414699e-06,
+      "loss": 0.2186,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2088838666677475,
+      "step": 3545,
+      "valid_targets_mean": 4254.4,
+      "valid_targets_min": 367
+    },
+    {
+      "epoch": 5.33033033033033,
+      "grad_norm": 0.5649058281181645,
+      "learning_rate": 6.55443476157029e-06,
+      "loss": 0.1991,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1812952756881714,
+      "step": 3550,
+      "valid_targets_mean": 4252.7,
+      "valid_targets_min": 673
+    },
+    {
+      "epoch": 5.337837837837838,
+      "grad_norm": 0.5452388205148444,
+      "learning_rate": 6.4990889484017e-06,
+      "loss": 0.2067,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1796906590461731,
+      "step": 3555,
+      "valid_targets_mean": 3792.6,
+      "valid_targets_min": 600
+    },
+    {
+      "epoch": 5.345345345345345,
+      "grad_norm": 0.5732225965401323,
+      "learning_rate": 6.443932429906013e-06,
+      "loss": 0.2323,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2930184006690979,
+      "step": 3560,
+      "valid_targets_mean": 4310.6,
+      "valid_targets_min": 541
+    },
+    {
+      "epoch": 5.352852852852853,
+      "grad_norm": 0.6028750589087655,
+      "learning_rate": 6.3889659794262334e-06,
+      "loss": 0.2129,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21735335886478424,
+      "step": 3565,
+      "valid_targets_mean": 3685.1,
+      "valid_targets_min": 583
+    },
+    {
+      "epoch": 5.36036036036036,
+      "grad_norm": 0.602478347893381,
+      "learning_rate": 6.334190367640449e-06,
+      "loss": 0.1793,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18190817534923553,
+      "step": 3570,
+      "valid_targets_mean": 4439.8,
+      "valid_targets_min": 656
+    },
+    {
+      "epoch": 5.367867867867868,
+      "grad_norm": 0.5557346424426345,
+      "learning_rate": 6.279606362551016e-06,
+      "loss": 0.2101,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1880781054496765,
+      "step": 3575,
+      "valid_targets_mean": 4198.0,
+      "valid_targets_min": 456
+    },
+    {
+      "epoch": 5.375375375375375,
+      "grad_norm": 0.4806991447402713,
+      "learning_rate": 6.225214729473794e-06,
+      "loss": 0.2637,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22571183741092682,
+      "step": 3580,
+      "valid_targets_mean": 5347.3,
+      "valid_targets_min": 652
+    },
+    {
+      "epoch": 5.382882882882883,
+      "grad_norm": 0.6189915178471305,
+      "learning_rate": 6.171016231027418e-06,
+      "loss": 0.2266,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22540658712387085,
+      "step": 3585,
+      "valid_targets_mean": 3396.4,
+      "valid_targets_min": 619
+    },
+    {
+      "epoch": 5.39039039039039,
+      "grad_norm": 0.4781145702083429,
+      "learning_rate": 6.117011627122617e-06,
+      "loss": 0.2178,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22122199833393097,
+      "step": 3590,
+      "valid_targets_mean": 5498.2,
+      "valid_targets_min": 407
+    },
+    {
+      "epoch": 5.397897897897898,
+      "grad_norm": 0.501236643073855,
+      "learning_rate": 6.063201674951535e-06,
+      "loss": 0.2113,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19432801008224487,
+      "step": 3595,
+      "valid_targets_mean": 4399.0,
+      "valid_targets_min": 748
+    },
+    {
+      "epoch": 5.405405405405405,
+      "grad_norm": 0.5083333895430631,
+      "learning_rate": 6.009587128977128e-06,
+      "loss": 0.1969,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21600385010242462,
+      "step": 3600,
+      "valid_targets_mean": 4527.8,
+      "valid_targets_min": 441
+    },
+    {
+      "epoch": 5.412912912912913,
+      "grad_norm": 0.43801311222355105,
+      "learning_rate": 5.956168740922592e-06,
+      "loss": 0.2232,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17686530947685242,
+      "step": 3605,
+      "valid_targets_mean": 5968.8,
+      "valid_targets_min": 650
+    },
+    {
+      "epoch": 5.42042042042042,
+      "grad_norm": 0.5142170347253675,
+      "learning_rate": 5.902947259760805e-06,
+      "loss": 0.2163,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16562317311763763,
+      "step": 3610,
+      "valid_targets_mean": 4170.2,
+      "valid_targets_min": 562
+    },
+    {
+      "epoch": 5.427927927927928,
+      "grad_norm": 0.57494822664475,
+      "learning_rate": 5.84992343170385e-06,
+      "loss": 0.2344,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19728896021842957,
+      "step": 3615,
+      "valid_targets_mean": 3665.9,
+      "valid_targets_min": 602
+    },
+    {
+      "epoch": 5.435435435435435,
+      "grad_norm": 0.6682604390565138,
+      "learning_rate": 5.797098000192527e-06,
+      "loss": 0.194,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18809694051742554,
+      "step": 3620,
+      "valid_targets_mean": 5038.7,
+      "valid_targets_min": 709
+    },
+    {
+      "epoch": 5.442942942942943,
+      "grad_norm": 0.5420242165199347,
+      "learning_rate": 5.744471705885946e-06,
+      "loss": 0.2203,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19964119791984558,
+      "step": 3625,
+      "valid_targets_mean": 4113.4,
+      "valid_targets_min": 400
+    },
+    {
+      "epoch": 5.45045045045045,
+      "grad_norm": 0.4897694331671357,
+      "learning_rate": 5.692045286651145e-06,
+      "loss": 0.2233,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21164345741271973,
+      "step": 3630,
+      "valid_targets_mean": 5133.8,
+      "valid_targets_min": 842
+    },
+    {
+      "epoch": 5.457957957957958,
+      "grad_norm": 0.5130796784439051,
+      "learning_rate": 5.6398194775527305e-06,
+      "loss": 0.2175,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1640782356262207,
+      "step": 3635,
+      "valid_targets_mean": 4620.8,
+      "valid_targets_min": 957
+    },
+    {
+      "epoch": 5.465465465465465,
+      "grad_norm": 0.7096256538606244,
+      "learning_rate": 5.587795010842578e-06,
+      "loss": 0.2188,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.229142963886261,
+      "step": 3640,
+      "valid_targets_mean": 3584.5,
+      "valid_targets_min": 291
+    },
+    {
+      "epoch": 5.472972972972973,
+      "grad_norm": 0.45320281126185963,
+      "learning_rate": 5.535972615949565e-06,
+      "loss": 0.1955,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15912821888923645,
+      "step": 3645,
+      "valid_targets_mean": 6046.6,
+      "valid_targets_min": 565
+    },
+    {
+      "epoch": 5.48048048048048,
+      "grad_norm": 0.6580766138686164,
+      "learning_rate": 5.484353019469344e-06,
+      "loss": 0.2385,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24230948090553284,
+      "step": 3650,
+      "valid_targets_mean": 3922.4,
+      "valid_targets_min": 618
+    },
+    {
+      "epoch": 5.487987987987988,
+      "grad_norm": 0.560301246319454,
+      "learning_rate": 5.432936945154159e-06,
+      "loss": 0.2035,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18417420983314514,
+      "step": 3655,
+      "valid_targets_mean": 4271.6,
+      "valid_targets_min": 590
+    },
+    {
+      "epoch": 5.495495495495495,
+      "grad_norm": 0.6206056063840636,
+      "learning_rate": 5.381725113902689e-06,
+      "loss": 0.1936,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17472167313098907,
+      "step": 3660,
+      "valid_targets_mean": 3848.2,
+      "valid_targets_min": 581
+    },
+    {
+      "epoch": 5.503003003003003,
+      "grad_norm": 0.4966662168616601,
+      "learning_rate": 5.330718243749942e-06,
+      "loss": 0.1978,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.191782146692276,
+      "step": 3665,
+      "valid_targets_mean": 4883.6,
+      "valid_targets_min": 648
+    },
+    {
+      "epoch": 5.51051051051051,
+      "grad_norm": 0.5471645413638431,
+      "learning_rate": 5.279917049857209e-06,
+      "loss": 0.1937,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17728394269943237,
+      "step": 3670,
+      "valid_targets_mean": 3883.9,
+      "valid_targets_min": 531
+    },
+    {
+      "epoch": 5.518018018018018,
+      "grad_norm": 0.49020284035840306,
+      "learning_rate": 5.229322244501996e-06,
+      "loss": 0.1986,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2151007056236267,
+      "step": 3675,
+      "valid_targets_mean": 5625.6,
+      "valid_targets_min": 710
+    },
+    {
+      "epoch": 5.525525525525525,
+      "grad_norm": 0.502623769306998,
+      "learning_rate": 5.1789345370680726e-06,
+      "loss": 0.2007,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18778328597545624,
+      "step": 3680,
+      "valid_targets_mean": 5465.0,
+      "valid_targets_min": 604
+    },
+    {
+      "epoch": 5.533033033033033,
+      "grad_norm": 0.6082538279267172,
+      "learning_rate": 5.128754634035509e-06,
+      "loss": 0.2006,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2307537943124771,
+      "step": 3685,
+      "valid_targets_mean": 3897.9,
+      "valid_targets_min": 475
+    },
+    {
+      "epoch": 5.54054054054054,
+      "grad_norm": 0.4677087975060976,
+      "learning_rate": 5.078783238970775e-06,
+      "loss": 0.1746,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1452668309211731,
+      "step": 3690,
+      "valid_targets_mean": 4697.0,
+      "valid_targets_min": 820
+    },
+    {
+      "epoch": 5.548048048048048,
+      "grad_norm": 0.5258809236597033,
+      "learning_rate": 5.029021052516874e-06,
+      "loss": 0.2364,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1686006635427475,
+      "step": 3695,
+      "valid_targets_mean": 5070.6,
+      "valid_targets_min": 874
+    },
+    {
+      "epoch": 5.555555555555555,
+      "grad_norm": 0.5418543854859791,
+      "learning_rate": 4.979468772383525e-06,
+      "loss": 0.2528,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23895879089832306,
+      "step": 3700,
+      "valid_targets_mean": 4703.9,
+      "valid_targets_min": 799
+    },
+    {
+      "epoch": 5.563063063063063,
+      "grad_norm": 0.5580957987554833,
+      "learning_rate": 4.93012709333736e-06,
+      "loss": 0.2182,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18779677152633667,
+      "step": 3705,
+      "valid_targets_mean": 3780.4,
+      "valid_targets_min": 700
+    },
+    {
+      "epoch": 5.57057057057057,
+      "grad_norm": 0.5796418427811311,
+      "learning_rate": 4.880996707192227e-06,
+      "loss": 0.2076,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18160055577754974,
+      "step": 3710,
+      "valid_targets_mean": 3617.5,
+      "valid_targets_min": 691
+    },
+    {
+      "epoch": 5.578078078078078,
+      "grad_norm": 0.625613489620117,
+      "learning_rate": 4.832078302799437e-06,
+      "loss": 0.231,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2596314549446106,
+      "step": 3715,
+      "valid_targets_mean": 4067.5,
+      "valid_targets_min": 928
+    },
+    {
+      "epoch": 5.585585585585585,
+      "grad_norm": 0.4752383217997911,
+      "learning_rate": 4.7833725660381356e-06,
+      "loss": 0.21,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15992337465286255,
+      "step": 3720,
+      "valid_targets_mean": 4770.9,
+      "valid_targets_min": 761
+    },
+    {
+      "epoch": 5.593093093093093,
+      "grad_norm": 0.5779722936684374,
+      "learning_rate": 4.734880179805685e-06,
+      "loss": 0.2091,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18457244336605072,
+      "step": 3725,
+      "valid_targets_mean": 3342.9,
+      "valid_targets_min": 691
+    },
+    {
+      "epoch": 5.6006006006006,
+      "grad_norm": 0.550075630002128,
+      "learning_rate": 4.686601824008079e-06,
+      "loss": 0.2177,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18044841289520264,
+      "step": 3730,
+      "valid_targets_mean": 4945.1,
+      "valid_targets_min": 906
+    },
+    {
+      "epoch": 5.608108108108108,
+      "grad_norm": 0.631353153629254,
+      "learning_rate": 4.638538175550422e-06,
+      "loss": 0.195,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21107108891010284,
+      "step": 3735,
+      "valid_targets_mean": 3289.5,
+      "valid_targets_min": 552
+    },
+    {
+      "epoch": 5.615615615615615,
+      "grad_norm": 0.7558092942754301,
+      "learning_rate": 4.590689908327428e-06,
+      "loss": 0.2214,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20774057507514954,
+      "step": 3740,
+      "valid_targets_mean": 2326.2,
+      "valid_targets_min": 684
+    },
+    {
+      "epoch": 5.623123123123123,
+      "grad_norm": 0.6050928606517935,
+      "learning_rate": 4.543057693213973e-06,
+      "loss": 0.2071,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.230237677693367,
+      "step": 3745,
+      "valid_targets_mean": 3875.2,
+      "valid_targets_min": 504
+    },
+    {
+      "epoch": 5.63063063063063,
+      "grad_norm": 0.5844617202863044,
+      "learning_rate": 4.495642198055707e-06,
+      "loss": 0.2215,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2029937505722046,
+      "step": 3750,
+      "valid_targets_mean": 4027.4,
+      "valid_targets_min": 536
+    },
+    {
+      "epoch": 5.638138138138138,
+      "grad_norm": 0.7169568097777915,
+      "learning_rate": 4.4484440876596625e-06,
+      "loss": 0.2364,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22157275676727295,
+      "step": 3755,
+      "valid_targets_mean": 2619.1,
+      "valid_targets_min": 657
+    },
+    {
+      "epoch": 5.645645645645645,
+      "grad_norm": 0.5563696214524937,
+      "learning_rate": 4.401464023784945e-06,
+      "loss": 0.2007,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1895512342453003,
+      "step": 3760,
+      "valid_targets_mean": 4033.1,
+      "valid_targets_min": 639
+    },
+    {
+      "epoch": 5.653153153153153,
+      "grad_norm": 0.7184497569373026,
+      "learning_rate": 4.354702665133457e-06,
+      "loss": 0.2035,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17378777265548706,
+      "step": 3765,
+      "valid_targets_mean": 4711.1,
+      "valid_targets_min": 460
+    },
+    {
+      "epoch": 5.66066066066066,
+      "grad_norm": 0.672562372546706,
+      "learning_rate": 4.308160667340671e-06,
+      "loss": 0.22,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23928751051425934,
+      "step": 3770,
+      "valid_targets_mean": 2964.8,
+      "valid_targets_min": 402
+    },
+    {
+      "epoch": 5.668168168168168,
+      "grad_norm": 0.4887664405970491,
+      "learning_rate": 4.261838682966413e-06,
+      "loss": 0.2001,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19159993529319763,
+      "step": 3775,
+      "valid_targets_mean": 5129.8,
+      "valid_targets_min": 752
+    },
+    {
+      "epoch": 5.675675675675675,
+      "grad_norm": 0.5457059516129726,
+      "learning_rate": 4.215737361485737e-06,
+      "loss": 0.2193,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21986320614814758,
+      "step": 3780,
+      "valid_targets_mean": 5641.1,
+      "valid_targets_min": 915
+    },
+    {
+      "epoch": 5.683183183183183,
+      "grad_norm": 0.5115604631418055,
+      "learning_rate": 4.1698573492798e-06,
+      "loss": 0.198,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1333058923482895,
+      "step": 3785,
+      "valid_targets_mean": 3278.0,
+      "valid_targets_min": 647
+    },
+    {
+      "epoch": 5.6906906906906904,
+      "grad_norm": 0.7200821505353998,
+      "learning_rate": 4.124199289626818e-06,
+      "loss": 0.2007,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22922024130821228,
+      "step": 3790,
+      "valid_targets_mean": 2683.2,
+      "valid_targets_min": 625
+    },
+    {
+      "epoch": 5.698198198198198,
+      "grad_norm": 0.42145237729701907,
+      "learning_rate": 4.078763822693021e-06,
+      "loss": 0.206,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18455691635608673,
+      "step": 3795,
+      "valid_targets_mean": 6690.8,
+      "valid_targets_min": 920
+    },
+    {
+      "epoch": 5.7057057057057055,
+      "grad_norm": 0.5325975126541347,
+      "learning_rate": 4.033551585523716e-06,
+      "loss": 0.2049,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19546617567539215,
+      "step": 3800,
+      "valid_targets_mean": 5228.1,
+      "valid_targets_min": 594
+    },
+    {
+      "epoch": 5.713213213213213,
+      "grad_norm": 0.6601846614023362,
+      "learning_rate": 3.988563212034315e-06,
+      "loss": 0.2511,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15158836543560028,
+      "step": 3805,
+      "valid_targets_mean": 3667.6,
+      "valid_targets_min": 358
+    },
+    {
+      "epoch": 5.7207207207207205,
+      "grad_norm": 0.6027115950409477,
+      "learning_rate": 3.943799333001466e-06,
+      "loss": 0.2164,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19812247157096863,
+      "step": 3810,
+      "valid_targets_mean": 3187.2,
+      "valid_targets_min": 542
+    },
+    {
+      "epoch": 5.728228228228228,
+      "grad_norm": 0.7999468435844861,
+      "learning_rate": 3.899260576054209e-06,
+      "loss": 0.2164,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24161022901535034,
+      "step": 3815,
+      "valid_targets_mean": 2608.4,
+      "valid_targets_min": 416
+    },
+    {
+      "epoch": 5.7357357357357355,
+      "grad_norm": 0.6197457192371475,
+      "learning_rate": 3.854947565665172e-06,
+      "loss": 0.2337,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20346400141716003,
+      "step": 3820,
+      "valid_targets_mean": 3887.9,
+      "valid_targets_min": 641
+    },
+    {
+      "epoch": 5.743243243243243,
+      "grad_norm": 0.5811299922570492,
+      "learning_rate": 3.810860923141824e-06,
+      "loss": 0.2173,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2590106129646301,
+      "step": 3825,
+      "valid_targets_mean": 4175.2,
+      "valid_targets_min": 750
+    },
+    {
+      "epoch": 5.7507507507507505,
+      "grad_norm": 0.6108735444142156,
+      "learning_rate": 3.767001266617747e-06,
+      "loss": 0.2198,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21409021317958832,
+      "step": 3830,
+      "valid_targets_mean": 3856.4,
+      "valid_targets_min": 504
+    },
+    {
+      "epoch": 5.758258258258258,
+      "grad_norm": 0.5427297235952976,
+      "learning_rate": 3.7233692110439814e-06,
+      "loss": 0.2104,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20367692410945892,
+      "step": 3835,
+      "valid_targets_mean": 4545.9,
+      "valid_targets_min": 666
+    },
+    {
+      "epoch": 5.7657657657657655,
+      "grad_norm": 0.4682965132065941,
+      "learning_rate": 3.679965368180416e-06,
+      "loss": 0.1959,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1938939094543457,
+      "step": 3840,
+      "valid_targets_mean": 5888.4,
+      "valid_targets_min": 915
+    },
+    {
+      "epoch": 5.773273273273273,
+      "grad_norm": 0.5399446918848115,
+      "learning_rate": 3.6367903465871755e-06,
+      "loss": 0.2268,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24548377096652985,
+      "step": 3845,
+      "valid_targets_mean": 5109.5,
+      "valid_targets_min": 602
+    },
+    {
+      "epoch": 5.7807807807807805,
+      "grad_norm": 0.4864834575204545,
+      "learning_rate": 3.59384475161612e-06,
+      "loss": 0.2163,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21027377247810364,
+      "step": 3850,
+      "valid_targets_mean": 4894.0,
+      "valid_targets_min": 636
+    },
+    {
+      "epoch": 5.788288288288288,
+      "grad_norm": 0.5438912695013488,
+      "learning_rate": 3.5511291854023466e-06,
+      "loss": 0.2059,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23461908102035522,
+      "step": 3855,
+      "valid_targets_mean": 4590.5,
+      "valid_targets_min": 666
+    },
+    {
+      "epoch": 5.7957957957957955,
+      "grad_norm": 0.5232434234880898,
+      "learning_rate": 3.508644246855739e-06,
+      "loss": 0.2097,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20156699419021606,
+      "step": 3860,
+      "valid_targets_mean": 4557.0,
+      "valid_targets_min": 694
+    },
+    {
+      "epoch": 5.803303303303303,
+      "grad_norm": 0.5868070563906084,
+      "learning_rate": 3.466390531652588e-06,
+      "loss": 0.2079,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19073697924613953,
+      "step": 3865,
+      "valid_targets_mean": 4407.1,
+      "valid_targets_min": 259
+    },
+    {
+      "epoch": 5.8108108108108105,
+      "grad_norm": 0.5169034481605863,
+      "learning_rate": 3.4243686322272195e-06,
+      "loss": 0.2133,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2452038675546646,
+      "step": 3870,
+      "valid_targets_mean": 5199.6,
+      "valid_targets_min": 564
+    },
+    {
+      "epoch": 5.818318318318318,
+      "grad_norm": 0.5384261747880087,
+      "learning_rate": 3.3825791377636998e-06,
+      "loss": 0.1954,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2101140320301056,
+      "step": 3875,
+      "valid_targets_mean": 4844.0,
+      "valid_targets_min": 427
+    },
+    {
+      "epoch": 5.8258258258258255,
+      "grad_norm": 0.7881322308404421,
+      "learning_rate": 3.341022634187585e-06,
+      "loss": 0.2041,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23327091336250305,
+      "step": 3880,
+      "valid_targets_mean": 3447.8,
+      "valid_targets_min": 485
+    },
+    {
+      "epoch": 5.833333333333333,
+      "grad_norm": 0.5514296552758701,
+      "learning_rate": 3.2996997041576795e-06,
+      "loss": 0.2105,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20250976085662842,
+      "step": 3885,
+      "valid_targets_mean": 4440.7,
+      "valid_targets_min": 434
+    },
+    {
+      "epoch": 5.8408408408408405,
+      "grad_norm": 0.651802460092001,
+      "learning_rate": 3.2586109270578836e-06,
+      "loss": 0.193,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2257906049489975,
+      "step": 3890,
+      "valid_targets_mean": 4516.6,
+      "valid_targets_min": 623
+    },
+    {
+      "epoch": 5.848348348348348,
+      "grad_norm": 0.5643346645467794,
+      "learning_rate": 3.217756878989069e-06,
+      "loss": 0.2237,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22649335861206055,
+      "step": 3895,
+      "valid_targets_mean": 4361.2,
+      "valid_targets_min": 764
+    },
+    {
+      "epoch": 5.8558558558558556,
+      "grad_norm": 0.6943626771374626,
+      "learning_rate": 3.1771381327609997e-06,
+      "loss": 0.2263,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23687607049942017,
+      "step": 3900,
+      "valid_targets_mean": 3878.1,
+      "valid_targets_min": 391
+    },
+    {
+      "epoch": 5.863363363363363,
+      "grad_norm": 0.5703396214687839,
+      "learning_rate": 3.1367552578842986e-06,
+      "loss": 0.1972,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22399486601352692,
+      "step": 3905,
+      "valid_targets_mean": 4644.7,
+      "valid_targets_min": 439
+    },
+    {
+      "epoch": 5.870870870870871,
+      "grad_norm": 3.5482019538966396,
+      "learning_rate": 3.096608820562468e-06,
+      "loss": 0.1959,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.14243656396865845,
+      "step": 3910,
+      "valid_targets_mean": 3601.3,
+      "valid_targets_min": 723
+    },
+    {
+      "epoch": 5.878378378378378,
+      "grad_norm": 0.6997352643905546,
+      "learning_rate": 3.0566993836839387e-06,
+      "loss": 0.2354,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22578692436218262,
+      "step": 3915,
+      "valid_targets_mean": 2871.8,
+      "valid_targets_min": 481
+    },
+    {
+      "epoch": 5.885885885885886,
+      "grad_norm": 0.61982038677274,
+      "learning_rate": 3.017027506814203e-06,
+      "loss": 0.2138,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.212412029504776,
+      "step": 3920,
+      "valid_targets_mean": 3693.6,
+      "valid_targets_min": 641
+    },
+    {
+      "epoch": 5.893393393393393,
+      "grad_norm": 0.502898749801898,
+      "learning_rate": 2.9775937461879368e-06,
+      "loss": 0.2034,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18976537883281708,
+      "step": 3925,
+      "valid_targets_mean": 5294.2,
+      "valid_targets_min": 649
+    },
+    {
+      "epoch": 5.900900900900901,
+      "grad_norm": 0.705380106461315,
+      "learning_rate": 2.9383986547012222e-06,
+      "loss": 0.2052,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21276956796646118,
+      "step": 3930,
+      "valid_targets_mean": 2696.2,
+      "valid_targets_min": 569
+    },
+    {
+      "epoch": 5.908408408408408,
+      "grad_norm": 0.6824495687596949,
+      "learning_rate": 2.8994427819037873e-06,
+      "loss": 0.2071,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27314475178718567,
+      "step": 3935,
+      "valid_targets_mean": 5253.1,
+      "valid_targets_min": 879
+    },
+    {
+      "epoch": 5.915915915915916,
+      "grad_norm": 0.6133652704497186,
+      "learning_rate": 2.8607266739913077e-06,
+      "loss": 0.2078,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24844960868358612,
+      "step": 3940,
+      "valid_targets_mean": 4655.8,
+      "valid_targets_min": 819
+    },
+    {
+      "epoch": 5.923423423423423,
+      "grad_norm": 0.5714988842182993,
+      "learning_rate": 2.822250873797743e-06,
+      "loss": 0.226,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19975441694259644,
+      "step": 3945,
+      "valid_targets_mean": 4395.9,
+      "valid_targets_min": 726
+    },
+    {
+      "epoch": 5.930930930930931,
+      "grad_norm": 0.8230004411475231,
+      "learning_rate": 2.784015920787719e-06,
+      "loss": 0.2254,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3142227828502655,
+      "step": 3950,
+      "valid_targets_mean": 4321.6,
+      "valid_targets_min": 853
+    },
+    {
+      "epoch": 5.938438438438438,
+      "grad_norm": 0.572231017896013,
+      "learning_rate": 2.746022351048978e-06,
+      "loss": 0.2267,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2292689085006714,
+      "step": 3955,
+      "valid_targets_mean": 4084.4,
+      "valid_targets_min": 615
+    },
+    {
+      "epoch": 5.945945945945946,
+      "grad_norm": 0.8188722407870543,
+      "learning_rate": 2.708270697284865e-06,
+      "loss": 0.2123,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24927443265914917,
+      "step": 3960,
+      "valid_targets_mean": 3767.1,
+      "valid_targets_min": 686
+    },
+    {
+      "epoch": 5.953453453453453,
+      "grad_norm": 0.5185251151107919,
+      "learning_rate": 2.6707614888068365e-06,
+      "loss": 0.209,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19668208062648773,
+      "step": 3965,
+      "valid_targets_mean": 4310.9,
+      "valid_targets_min": 423
+    },
+    {
+      "epoch": 5.960960960960961,
+      "grad_norm": 0.5542673950873109,
+      "learning_rate": 2.633495251527054e-06,
+      "loss": 0.1931,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24870452284812927,
+      "step": 3970,
+      "valid_targets_mean": 5467.7,
+      "valid_targets_min": 973
+    },
+    {
+      "epoch": 5.968468468468468,
+      "grad_norm": 0.6048295638583769,
+      "learning_rate": 2.5964725079510134e-06,
+      "loss": 0.2045,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17934688925743103,
+      "step": 3975,
+      "valid_targets_mean": 3691.8,
+      "valid_targets_min": 715
+    },
+    {
+      "epoch": 5.975975975975976,
+      "grad_norm": 0.5714679755869654,
+      "learning_rate": 2.5596937771702113e-06,
+      "loss": 0.1995,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18001726269721985,
+      "step": 3980,
+      "valid_targets_mean": 3830.1,
+      "valid_targets_min": 698
+    },
+    {
+      "epoch": 5.983483483483483,
+      "grad_norm": 0.5678385470019257,
+      "learning_rate": 2.523159574854872e-06,
+      "loss": 0.2096,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19507171213626862,
+      "step": 3985,
+      "valid_targets_mean": 3516.0,
+      "valid_targets_min": 311
+    },
+    {
+      "epoch": 5.990990990990991,
+      "grad_norm": 0.685934053595724,
+      "learning_rate": 2.4868704132467115e-06,
+      "loss": 0.2047,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2302461564540863,
+      "step": 3990,
+      "valid_targets_mean": 2869.4,
+      "valid_targets_min": 584
+    },
+    {
+      "epoch": 5.998498498498498,
+      "grad_norm": 0.6065925509118129,
+      "learning_rate": 2.45082680115176e-06,
+      "loss": 0.2199,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19993174076080322,
+      "step": 3995,
+      "valid_targets_mean": 3674.2,
+      "valid_targets_min": 414
+    },
+    {
+      "epoch": 6.006006006006006,
+      "grad_norm": 0.4490328583511514,
+      "learning_rate": 2.4150292439332355e-06,
+      "loss": 0.1979,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19256381690502167,
+      "step": 4000,
+      "valid_targets_mean": 5463.9,
+      "valid_targets_min": 623
+    },
+    {
+      "epoch": 6.013513513513513,
+      "grad_norm": 0.5503944239111109,
+      "learning_rate": 2.3794782435044384e-06,
+      "loss": 0.1788,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17218606173992157,
+      "step": 4005,
+      "valid_targets_mean": 3465.8,
+      "valid_targets_min": 930
+    },
+    {
+      "epoch": 6.021021021021021,
+      "grad_norm": 0.5533090724442471,
+      "learning_rate": 2.344174298321733e-06,
+      "loss": 0.2125,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.213973268866539,
+      "step": 4010,
+      "valid_targets_mean": 4543.8,
+      "valid_targets_min": 570
+    },
+    {
+      "epoch": 6.028528528528528,
+      "grad_norm": 0.6792685607327363,
+      "learning_rate": 2.3091179033775445e-06,
+      "loss": 0.2038,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2688198983669281,
+      "step": 4015,
+      "valid_targets_mean": 3235.7,
+      "valid_targets_min": 605
+    },
+    {
+      "epoch": 6.036036036036036,
+      "grad_norm": 0.46026148213907403,
+      "learning_rate": 2.27430955019343e-06,
+      "loss": 0.2157,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1564239263534546,
+      "step": 4020,
+      "valid_targets_mean": 5029.9,
+      "valid_targets_min": 1069
+    },
+    {
+      "epoch": 6.043543543543543,
+      "grad_norm": 0.5322076051668858,
+      "learning_rate": 2.239749726813183e-06,
+      "loss": 0.2239,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19674474000930786,
+      "step": 4025,
+      "valid_targets_mean": 4798.2,
+      "valid_targets_min": 478
+    },
+    {
+      "epoch": 6.051051051051051,
+      "grad_norm": 0.5891185848467975,
+      "learning_rate": 2.20543891779599e-06,
+      "loss": 0.199,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18149876594543457,
+      "step": 4030,
+      "valid_targets_mean": 4520.9,
+      "valid_targets_min": 648
+    },
+    {
+      "epoch": 6.058558558558558,
+      "grad_norm": 0.6755239064217278,
+      "learning_rate": 2.1713776042096302e-06,
+      "loss": 0.2046,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2210536003112793,
+      "step": 4035,
+      "valid_targets_mean": 3545.1,
+      "valid_targets_min": 665
+    },
+    {
+      "epoch": 6.066066066066066,
+      "grad_norm": 0.6088360110015828,
+      "learning_rate": 2.1375662636237493e-06,
+      "loss": 0.1992,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22115203738212585,
+      "step": 4040,
+      "valid_targets_mean": 5270.9,
+      "valid_targets_min": 656
+    },
+    {
+      "epoch": 6.073573573573573,
+      "grad_norm": 0.5822383908375098,
+      "learning_rate": 2.1040053701031414e-06,
+      "loss": 0.2067,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21398739516735077,
+      "step": 4045,
+      "valid_targets_mean": 4107.5,
+      "valid_targets_min": 483
+    },
+    {
+      "epoch": 6.081081081081081,
+      "grad_norm": 0.6863357601545187,
+      "learning_rate": 2.0706953942011165e-06,
+      "loss": 0.2227,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21303421258926392,
+      "step": 4050,
+      "valid_targets_mean": 3397.6,
+      "valid_targets_min": 691
+    },
+    {
+      "epoch": 6.088588588588588,
+      "grad_norm": 0.6835718993938039,
+      "learning_rate": 2.0376368029528935e-06,
+      "loss": 0.2104,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2734207510948181,
+      "step": 4055,
+      "valid_targets_mean": 3650.6,
+      "valid_targets_min": 652
+    },
+    {
+      "epoch": 6.096096096096096,
+      "grad_norm": 0.45120552286100196,
+      "learning_rate": 2.0048300598690627e-06,
+      "loss": 0.2164,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25283950567245483,
+      "step": 4060,
+      "valid_targets_mean": 7365.3,
+      "valid_targets_min": 993
+    },
+    {
+      "epoch": 6.103603603603603,
+      "grad_norm": 0.63959559159808,
+      "learning_rate": 1.9722756249290785e-06,
+      "loss": 0.2116,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2537308633327484,
+      "step": 4065,
+      "valid_targets_mean": 4346.6,
+      "valid_targets_min": 668
+    },
+    {
+      "epoch": 6.111111111111111,
+      "grad_norm": 0.6496824147481889,
+      "learning_rate": 1.939973954574812e-06,
+      "loss": 0.175,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15889085829257965,
+      "step": 4070,
+      "valid_targets_mean": 3991.0,
+      "valid_targets_min": 738
+    },
+    {
+      "epoch": 6.118618618618618,
+      "grad_norm": 0.5705409403401243,
+      "learning_rate": 1.9079255017041487e-06,
+      "loss": 0.2026,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17797492444515228,
+      "step": 4075,
+      "valid_targets_mean": 4241.4,
+      "valid_targets_min": 618
+    },
+    {
+      "epoch": 6.126126126126126,
+      "grad_norm": 0.4815875979801429,
+      "learning_rate": 1.8761307156646547e-06,
+      "loss": 0.2204,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15079014003276825,
+      "step": 4080,
+      "valid_targets_mean": 4996.1,
+      "valid_targets_min": 513
+    },
+    {
+      "epoch": 6.133633633633633,
+      "grad_norm": 0.5146233292308801,
+      "learning_rate": 1.8445900422472495e-06,
+      "loss": 0.2046,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18511003255844116,
+      "step": 4085,
+      "valid_targets_mean": 4634.7,
+      "valid_targets_min": 685
+    },
+    {
+      "epoch": 6.141141141141141,
+      "grad_norm": 0.5624700118014101,
+      "learning_rate": 1.8133039236799788e-06,
+      "loss": 0.2067,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2083277553319931,
+      "step": 4090,
+      "valid_targets_mean": 4171.3,
+      "valid_targets_min": 511
+    },
+    {
+      "epoch": 6.148648648648648,
+      "grad_norm": 0.7177782923244695,
+      "learning_rate": 1.7822727986217957e-06,
+      "loss": 0.2237,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18404030799865723,
+      "step": 4095,
+      "valid_targets_mean": 2409.4,
+      "valid_targets_min": 593
+    },
+    {
+      "epoch": 6.156156156156156,
+      "grad_norm": 0.5228873362141804,
+      "learning_rate": 1.7514971021564275e-06,
+      "loss": 0.2147,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1940818727016449,
+      "step": 4100,
+      "valid_targets_mean": 4779.9,
+      "valid_targets_min": 511
+    },
+    {
+      "epoch": 6.163663663663663,
+      "grad_norm": 0.4487406897661156,
+      "learning_rate": 1.7209772657862657e-06,
+      "loss": 0.1958,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1845383495092392,
+      "step": 4105,
+      "valid_targets_mean": 6135.4,
+      "valid_targets_min": 515
+    },
+    {
+      "epoch": 6.171171171171171,
+      "grad_norm": 0.5106337969240039,
+      "learning_rate": 1.6907137174263156e-06,
+      "loss": 0.2004,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17622476816177368,
+      "step": 4110,
+      "valid_targets_mean": 4512.0,
+      "valid_targets_min": 463
+    },
+    {
+      "epoch": 6.178678678678678,
+      "grad_norm": 0.6441468445611757,
+      "learning_rate": 1.6607068813981952e-06,
+      "loss": 0.1916,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2345384955406189,
+      "step": 4115,
+      "valid_targets_mean": 3844.5,
+      "valid_targets_min": 471
+    },
+    {
+      "epoch": 6.186186186186186,
+      "grad_norm": 0.5581311436354888,
+      "learning_rate": 1.6309571784242062e-06,
+      "loss": 0.2017,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17409540712833405,
+      "step": 4120,
+      "valid_targets_mean": 4888.1,
+      "valid_targets_min": 645
+    },
+    {
+      "epoch": 6.193693693693693,
+      "grad_norm": 0.5762764403446647,
+      "learning_rate": 1.601465025621398e-06,
+      "loss": 0.2212,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22839732468128204,
+      "step": 4125,
+      "valid_targets_mean": 4442.8,
+      "valid_targets_min": 669
+    },
+    {
+      "epoch": 6.201201201201201,
+      "grad_norm": 0.488635665892412,
+      "learning_rate": 1.5722308364957517e-06,
+      "loss": 0.1868,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19067010283470154,
+      "step": 4130,
+      "valid_targets_mean": 5109.4,
+      "valid_targets_min": 606
+    },
+    {
+      "epoch": 6.208708708708708,
+      "grad_norm": 0.5702695176534012,
+      "learning_rate": 1.5432550209363606e-06,
+      "loss": 0.2119,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19797605276107788,
+      "step": 4135,
+      "valid_targets_mean": 4942.1,
+      "valid_targets_min": 694
+    },
+    {
+      "epoch": 6.216216216216216,
+      "grad_norm": 0.5547765610808129,
+      "learning_rate": 1.5145379852097054e-06,
+      "loss": 0.1995,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21563218533992767,
+      "step": 4140,
+      "valid_targets_mean": 4537.9,
+      "valid_targets_min": 498
+    },
+    {
+      "epoch": 6.223723723723724,
+      "grad_norm": 0.6133297188420032,
+      "learning_rate": 1.4860801319539354e-06,
+      "loss": 0.2217,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2022094428539276,
+      "step": 4145,
+      "valid_targets_mean": 3766.9,
+      "valid_targets_min": 602
+    },
+    {
+      "epoch": 6.231231231231231,
+      "grad_norm": 0.5500242111051548,
+      "learning_rate": 1.457881860173236e-06,
+      "loss": 0.1928,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.14841407537460327,
+      "step": 4150,
+      "valid_targets_mean": 3554.3,
+      "valid_targets_min": 628
+    },
+    {
+      "epoch": 6.238738738738739,
+      "grad_norm": 0.6493235742255749,
+      "learning_rate": 1.429943565232228e-06,
+      "loss": 0.2165,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.29499688744544983,
+      "step": 4155,
+      "valid_targets_mean": 4105.4,
+      "valid_targets_min": 391
+    },
+    {
+      "epoch": 6.246246246246246,
+      "grad_norm": 0.46401350568259536,
+      "learning_rate": 1.4022656388504418e-06,
+      "loss": 0.2026,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1871238350868225,
+      "step": 4160,
+      "valid_targets_mean": 5385.7,
+      "valid_targets_min": 467
+    },
+    {
+      "epoch": 6.253753753753754,
+      "grad_norm": 0.5751734802750025,
+      "learning_rate": 1.3748484690967967e-06,
+      "loss": 0.1754,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17546819150447845,
+      "step": 4165,
+      "valid_targets_mean": 4171.8,
+      "valid_targets_min": 564
+    },
+    {
+      "epoch": 6.261261261261261,
+      "grad_norm": 0.824863745151526,
+      "learning_rate": 1.3476924403841762e-06,
+      "loss": 0.2352,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21639591455459595,
+      "step": 4170,
+      "valid_targets_mean": 2713.2,
+      "valid_targets_min": 473
+    },
+    {
+      "epoch": 6.268768768768769,
+      "grad_norm": 0.570544358689332,
+      "learning_rate": 1.3207979334640419e-06,
+      "loss": 0.2095,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20927639305591583,
+      "step": 4175,
+      "valid_targets_mean": 4503.6,
+      "valid_targets_min": 774
+    },
+    {
+      "epoch": 6.276276276276276,
+      "grad_norm": 0.5313264329238784,
+      "learning_rate": 1.294165325421084e-06,
+      "loss": 0.2398,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.29111388325691223,
+      "step": 4180,
+      "valid_targets_mean": 5414.2,
+      "valid_targets_min": 731
+    },
+    {
+      "epoch": 6.283783783783784,
+      "grad_norm": 0.5675843811457961,
+      "learning_rate": 1.2677949896679408e-06,
+      "loss": 0.2117,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20922914147377014,
+      "step": 4185,
+      "valid_targets_mean": 4288.4,
+      "valid_targets_min": 434
+    },
+    {
+      "epoch": 6.291291291291291,
+      "grad_norm": 0.5380343131647501,
+      "learning_rate": 1.241687295939966e-06,
+      "loss": 0.1725,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.14380943775177002,
+      "step": 4190,
+      "valid_targets_mean": 4689.5,
+      "valid_targets_min": 367
+    },
+    {
+      "epoch": 6.298798798798799,
+      "grad_norm": 0.5899919262888533,
+      "learning_rate": 1.2158426102900345e-06,
+      "loss": 0.189,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19863969087600708,
+      "step": 4195,
+      "valid_targets_mean": 4058.4,
+      "valid_targets_min": 727
+    },
+    {
+      "epoch": 6.306306306306306,
+      "grad_norm": 0.5264653142973713,
+      "learning_rate": 1.1902612950834213e-06,
+      "loss": 0.2007,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.197919100522995,
+      "step": 4200,
+      "valid_targets_mean": 4504.0,
+      "valid_targets_min": 727
+    },
+    {
+      "epoch": 6.313813813813814,
+      "grad_norm": 0.5816807471532681,
+      "learning_rate": 1.164943708992714e-06,
+      "loss": 0.1954,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19351375102996826,
+      "step": 4205,
+      "valid_targets_mean": 3772.9,
+      "valid_targets_min": 725
+    },
+    {
+      "epoch": 6.321321321321321,
+      "grad_norm": 0.6196400182011077,
+      "learning_rate": 1.1398902069927842e-06,
+      "loss": 0.232,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3165440261363983,
+      "step": 4210,
+      "valid_targets_mean": 4516.8,
+      "valid_targets_min": 374
+    },
+    {
+      "epoch": 6.328828828828829,
+      "grad_norm": 0.5403053431375547,
+      "learning_rate": 1.115101140355812e-06,
+      "loss": 0.2077,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20008732378482819,
+      "step": 4215,
+      "valid_targets_mean": 4993.6,
+      "valid_targets_min": 667
+    },
+    {
+      "epoch": 6.336336336336337,
+      "grad_norm": 0.5311224743152807,
+      "learning_rate": 1.0905768566463593e-06,
+      "loss": 0.2158,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2345670610666275,
+      "step": 4220,
+      "valid_targets_mean": 5031.6,
+      "valid_targets_min": 559
+    },
+    {
+      "epoch": 6.343843843843844,
+      "grad_norm": 0.7193489501948159,
+      "learning_rate": 1.0663176997165037e-06,
+      "loss": 0.2083,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18373778462409973,
+      "step": 4225,
+      "valid_targets_mean": 2563.9,
+      "valid_targets_min": 614
+    },
+    {
+      "epoch": 6.351351351351352,
+      "grad_norm": 0.6194973179199644,
+      "learning_rate": 1.0423240097010056e-06,
+      "loss": 0.1904,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17724689841270447,
+      "step": 4230,
+      "valid_targets_mean": 2932.5,
+      "valid_targets_min": 550
+    },
+    {
+      "epoch": 6.358858858858859,
+      "grad_norm": 0.5938101431895507,
+      "learning_rate": 1.018596123012545e-06,
+      "loss": 0.2014,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15485897660255432,
+      "step": 4235,
+      "valid_targets_mean": 3859.1,
+      "valid_targets_min": 565
+    },
+    {
+      "epoch": 6.366366366366367,
+      "grad_norm": 0.5832105338347596,
+      "learning_rate": 9.951343723370145e-07,
+      "loss": 0.2217,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27296674251556396,
+      "step": 4240,
+      "valid_targets_mean": 4697.3,
+      "valid_targets_min": 651
+    },
+    {
+      "epoch": 6.373873873873874,
+      "grad_norm": 0.6257938479877299,
+      "learning_rate": 9.719390866288325e-07,
+      "loss": 0.2018,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21452078223228455,
+      "step": 4245,
+      "valid_targets_mean": 3980.8,
+      "valid_targets_min": 644
+    },
+    {
+      "epoch": 6.381381381381382,
+      "grad_norm": 0.4748467821348681,
+      "learning_rate": 9.490105911063519e-07,
+      "loss": 0.1984,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19311422109603882,
+      "step": 4250,
+      "valid_targets_mean": 4970.2,
+      "valid_targets_min": 958
+    },
+    {
+      "epoch": 6.388888888888889,
+      "grad_norm": 0.49848343014176094,
+      "learning_rate": 9.263492072472879e-07,
+      "loss": 0.2041,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20664718747138977,
+      "step": 4255,
+      "valid_targets_mean": 5091.6,
+      "valid_targets_min": 530
+    },
+    {
+      "epoch": 6.396396396396397,
+      "grad_norm": 0.4839998287321489,
+      "learning_rate": 9.039552527842188e-07,
+      "loss": 0.1931,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15590183436870575,
+      "step": 4260,
+      "valid_targets_mean": 4334.7,
+      "valid_targets_min": 755
+    },
+    {
+      "epoch": 6.403903903903904,
+      "grad_norm": 0.47675581127591066,
+      "learning_rate": 8.818290417001218e-07,
+      "loss": 0.1877,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18418073654174805,
+      "step": 4265,
+      "valid_targets_mean": 5461.1,
+      "valid_targets_min": 875
+    },
+    {
+      "epoch": 6.411411411411412,
+      "grad_norm": 0.5148421630943572,
+      "learning_rate": 8.599708842239795e-07,
+      "loss": 0.2056,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1673521250486374,
+      "step": 4270,
+      "valid_targets_mean": 5613.2,
+      "valid_targets_min": 791
+    },
+    {
+      "epoch": 6.418918918918919,
+      "grad_norm": 0.6034041442260967,
+      "learning_rate": 8.383810868264253e-07,
+      "loss": 0.2101,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2042214274406433,
+      "step": 4275,
+      "valid_targets_mean": 4745.4,
+      "valid_targets_min": 521
+    },
+    {
+      "epoch": 6.426426426426427,
+      "grad_norm": 0.44620270158602615,
+      "learning_rate": 8.17059952215451e-07,
+      "loss": 0.1892,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.14886009693145752,
+      "step": 4280,
+      "valid_targets_mean": 5529.8,
+      "valid_targets_min": 1189
+    },
+    {
+      "epoch": 6.433933933933934,
+      "grad_norm": 0.5070235125913082,
+      "learning_rate": 7.960077793321552e-07,
+      "loss": 0.1769,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19193494319915771,
+      "step": 4285,
+      "valid_targets_mean": 5251.6,
+      "valid_targets_min": 425
+    },
+    {
+      "epoch": 6.441441441441442,
+      "grad_norm": 0.5615395471791965,
+      "learning_rate": 7.752248633465576e-07,
+      "loss": 0.215,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22860676050186157,
+      "step": 4290,
+      "valid_targets_mean": 3936.7,
+      "valid_targets_min": 583
+    },
+    {
+      "epoch": 6.448948948948949,
+      "grad_norm": 0.5442260146056317,
+      "learning_rate": 7.547114956534574e-07,
+      "loss": 0.1909,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1491059809923172,
+      "step": 4295,
+      "valid_targets_mean": 3889.1,
+      "valid_targets_min": 485
+    },
+    {
+      "epoch": 6.456456456456457,
+      "grad_norm": 0.5835357955664173,
+      "learning_rate": 7.344679638683527e-07,
+      "loss": 0.2332,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2454603910446167,
+      "step": 4300,
+      "valid_targets_mean": 5267.1,
+      "valid_targets_min": 578
+    },
+    {
+      "epoch": 6.463963963963964,
+      "grad_norm": 0.6370221393360167,
+      "learning_rate": 7.144945518234014e-07,
+      "loss": 0.2188,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20245075225830078,
+      "step": 4305,
+      "valid_targets_mean": 3993.6,
+      "valid_targets_min": 292
+    },
+    {
+      "epoch": 6.471471471471472,
+      "grad_norm": 0.487081794320264,
+      "learning_rate": 6.947915395634463e-07,
+      "loss": 0.2125,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22089102864265442,
+      "step": 4310,
+      "valid_targets_mean": 4927.1,
+      "valid_targets_min": 801
+    },
+    {
+      "epoch": 6.478978978978979,
+      "grad_norm": 0.6970402670081776,
+      "learning_rate": 6.753592033420809e-07,
+      "loss": 0.1902,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19842252135276794,
+      "step": 4315,
+      "valid_targets_mean": 2896.4,
+      "valid_targets_min": 475
+    },
+    {
+      "epoch": 6.486486486486487,
+      "grad_norm": 0.5079593962075888,
+      "learning_rate": 6.561978156177939e-07,
+      "loss": 0.1899,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1689176708459854,
+      "step": 4320,
+      "valid_targets_mean": 4254.1,
+      "valid_targets_min": 861
+    },
+    {
+      "epoch": 6.493993993993994,
+      "grad_norm": 0.6574113141398381,
+      "learning_rate": 6.37307645050127e-07,
+      "loss": 0.2036,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2085978239774704,
+      "step": 4325,
+      "valid_targets_mean": 3372.8,
+      "valid_targets_min": 478
+    },
+    {
+      "epoch": 6.501501501501502,
+      "grad_norm": 0.508520487800978,
+      "learning_rate": 6.186889564959209e-07,
+      "loss": 0.2103,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23924149572849274,
+      "step": 4330,
+      "valid_targets_mean": 5251.2,
+      "valid_targets_min": 565
+    },
+    {
+      "epoch": 6.509009009009009,
+      "grad_norm": 0.6324952400411343,
+      "learning_rate": 6.003420110056035e-07,
+      "loss": 0.204,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2557426691055298,
+      "step": 4335,
+      "valid_targets_mean": 4012.0,
+      "valid_targets_min": 503
+    },
+    {
+      "epoch": 6.516516516516517,
+      "grad_norm": 0.553904609232918,
+      "learning_rate": 5.822670658195173e-07,
+      "loss": 0.1907,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21928340196609497,
+      "step": 4340,
+      "valid_targets_mean": 4482.4,
+      "valid_targets_min": 462
+    },
+    {
+      "epoch": 6.524024024024024,
+      "grad_norm": 0.616465266517478,
+      "learning_rate": 5.644643743643308e-07,
+      "loss": 0.1911,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21219468116760254,
+      "step": 4345,
+      "valid_targets_mean": 3722.6,
+      "valid_targets_min": 504
+    },
+    {
+      "epoch": 6.531531531531532,
+      "grad_norm": 0.7300734830816725,
+      "learning_rate": 5.469341862494637e-07,
+      "loss": 0.2046,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2306327223777771,
+      "step": 4350,
+      "valid_targets_mean": 2665.6,
+      "valid_targets_min": 527
+    },
+    {
+      "epoch": 6.539039039039039,
+      "grad_norm": 0.5642202986656522,
+      "learning_rate": 5.296767472636077e-07,
+      "loss": 0.1916,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1920192539691925,
+      "step": 4355,
+      "valid_targets_mean": 3940.9,
+      "valid_targets_min": 461
+    },
+    {
+      "epoch": 6.546546546546547,
+      "grad_norm": 0.5150972885704285,
+      "learning_rate": 5.12692299371269e-07,
+      "loss": 0.1942,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.14734071493148804,
+      "step": 4360,
+      "valid_targets_mean": 4389.3,
+      "valid_targets_min": 590
+    },
+    {
+      "epoch": 6.554054054054054,
+      "grad_norm": 0.6913329068038451,
+      "learning_rate": 4.959810807093779e-07,
+      "loss": 0.2347,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2543680667877197,
+      "step": 4365,
+      "valid_targets_mean": 4525.6,
+      "valid_targets_min": 554
+    },
+    {
+      "epoch": 6.561561561561562,
+      "grad_norm": 0.5451201542998624,
+      "learning_rate": 4.795433255839488e-07,
+      "loss": 0.2049,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21131403744220734,
+      "step": 4370,
+      "valid_targets_mean": 4360.6,
+      "valid_targets_min": 654
+    },
+    {
+      "epoch": 6.569069069069069,
+      "grad_norm": 0.47040078136948726,
+      "learning_rate": 4.633792644667967e-07,
+      "loss": 0.1887,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19125640392303467,
+      "step": 4375,
+      "valid_targets_mean": 5975.6,
+      "valid_targets_min": 1693
+    },
+    {
+      "epoch": 6.576576576576577,
+      "grad_norm": 0.5920967555082476,
+      "learning_rate": 4.474891239923063e-07,
+      "loss": 0.209,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21285581588745117,
+      "step": 4380,
+      "valid_targets_mean": 4406.5,
+      "valid_targets_min": 643
+    },
+    {
+      "epoch": 6.584084084084084,
+      "grad_norm": 0.7205742446023933,
+      "learning_rate": 4.318731269542542e-07,
+      "loss": 0.1867,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21254310011863708,
+      "step": 4385,
+      "valid_targets_mean": 3167.4,
+      "valid_targets_min": 489
+    },
+    {
+      "epoch": 6.591591591591592,
+      "grad_norm": 0.5312844013389322,
+      "learning_rate": 4.1653149230268087e-07,
+      "loss": 0.2147,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18900719285011292,
+      "step": 4390,
+      "valid_targets_mean": 4654.4,
+      "valid_targets_min": 479
+    },
+    {
+      "epoch": 6.599099099099099,
+      "grad_norm": 0.5723780339209558,
+      "learning_rate": 4.014644351408259e-07,
+      "loss": 0.1911,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22007179260253906,
+      "step": 4395,
+      "valid_targets_mean": 4301.8,
+      "valid_targets_min": 879
+    },
+    {
+      "epoch": 6.606606606606607,
+      "grad_norm": 0.5052598547107464,
+      "learning_rate": 3.8667216672211515e-07,
+      "loss": 0.2064,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17064377665519714,
+      "step": 4400,
+      "valid_targets_mean": 4867.7,
+      "valid_targets_min": 651
+    },
+    {
+      "epoch": 6.614114114114114,
+      "grad_norm": 0.7316891946951816,
+      "learning_rate": 3.7215489444718313e-07,
+      "loss": 0.2098,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22059103846549988,
+      "step": 4405,
+      "valid_targets_mean": 2672.8,
+      "valid_targets_min": 724
+    },
+    {
+      "epoch": 6.621621621621622,
+      "grad_norm": 0.6761848493284879,
+      "learning_rate": 3.5791282186099064e-07,
+      "loss": 0.2279,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23260068893432617,
+      "step": 4410,
+      "valid_targets_mean": 3445.1,
+      "valid_targets_min": 613
+    },
+    {
+      "epoch": 6.629129129129129,
+      "grad_norm": 0.6708523109273121,
+      "learning_rate": 3.439461486499451e-07,
+      "loss": 0.2085,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23754608631134033,
+      "step": 4415,
+      "valid_targets_mean": 3454.1,
+      "valid_targets_min": 495
+    },
+    {
+      "epoch": 6.636636636636637,
+      "grad_norm": 0.5130944622824821,
+      "learning_rate": 3.302550706391161e-07,
+      "loss": 0.2022,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1540905237197876,
+      "step": 4420,
+      "valid_targets_mean": 4449.4,
+      "valid_targets_min": 676
+    },
+    {
+      "epoch": 6.6441441441441444,
+      "grad_norm": 0.5253445851442728,
+      "learning_rate": 3.168397797894818e-07,
+      "loss": 0.2039,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19345726072788239,
+      "step": 4425,
+      "valid_targets_mean": 5712.4,
+      "valid_targets_min": 814
+    },
+    {
+      "epoch": 6.651651651651652,
+      "grad_norm": 0.4999642956754218,
+      "learning_rate": 3.03700464195249e-07,
+      "loss": 0.1771,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1533123254776001,
+      "step": 4430,
+      "valid_targets_mean": 4014.4,
+      "valid_targets_min": 448
+    },
+    {
+      "epoch": 6.6591591591591595,
+      "grad_norm": 0.6600694282993609,
+      "learning_rate": 2.9083730808120214e-07,
+      "loss": 0.252,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26119160652160645,
+      "step": 4435,
+      "valid_targets_mean": 3377.4,
+      "valid_targets_min": 445
+    },
+    {
+      "epoch": 6.666666666666667,
+      "grad_norm": 0.6967687532254948,
+      "learning_rate": 2.78250491800125e-07,
+      "loss": 0.2081,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1988351047039032,
+      "step": 4440,
+      "valid_targets_mean": 2430.9,
+      "valid_targets_min": 396
+    },
+    {
+      "epoch": 6.6741741741741745,
+      "grad_norm": 0.6534875940602458,
+      "learning_rate": 2.6594019183027844e-07,
+      "loss": 0.2099,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2011052370071411,
+      "step": 4445,
+      "valid_targets_mean": 3745.1,
+      "valid_targets_min": 709
+    },
+    {
+      "epoch": 6.681681681681682,
+      "grad_norm": 0.6527183652558144,
+      "learning_rate": 2.5390658077292017e-07,
+      "loss": 0.2069,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19569085538387299,
+      "step": 4450,
+      "valid_targets_mean": 3687.9,
+      "valid_targets_min": 539
+    },
+    {
+      "epoch": 6.6891891891891895,
+      "grad_norm": 0.8645812862940178,
+      "learning_rate": 2.4214982734988013e-07,
+      "loss": 0.2033,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22479039430618286,
+      "step": 4455,
+      "valid_targets_mean": 2105.8,
+      "valid_targets_min": 704
+    },
+    {
+      "epoch": 6.696696696696697,
+      "grad_norm": 0.5255004317144049,
+      "learning_rate": 2.3067009640120652e-07,
+      "loss": 0.2365,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26815474033355713,
+      "step": 4460,
+      "valid_targets_mean": 5218.8,
+      "valid_targets_min": 565
+    },
+    {
+      "epoch": 6.7042042042042045,
+      "grad_norm": 0.5916763601809271,
+      "learning_rate": 2.194675488828457e-07,
+      "loss": 0.1791,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1834341287612915,
+      "step": 4465,
+      "valid_targets_mean": 3968.9,
+      "valid_targets_min": 704
+    },
+    {
+      "epoch": 6.711711711711712,
+      "grad_norm": 0.614175690495789,
+      "learning_rate": 2.0854234186438837e-07,
+      "loss": 0.1942,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24795103073120117,
+      "step": 4470,
+      "valid_targets_mean": 3751.3,
+      "valid_targets_min": 613
+    },
+    {
+      "epoch": 6.7192192192192195,
+      "grad_norm": 0.5171279571165366,
+      "learning_rate": 1.9789462852686237e-07,
+      "loss": 0.2013,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19156940281391144,
+      "step": 4475,
+      "valid_targets_mean": 4793.4,
+      "valid_targets_min": 679
+    },
+    {
+      "epoch": 6.726726726726727,
+      "grad_norm": 0.5178172693023037,
+      "learning_rate": 1.875245581605989e-07,
+      "loss": 0.187,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16094420850276947,
+      "step": 4480,
+      "valid_targets_mean": 4665.6,
+      "valid_targets_min": 471
+    },
+    {
+      "epoch": 6.7342342342342345,
+      "grad_norm": 0.48834241533960204,
+      "learning_rate": 1.7743227616312308e-07,
+      "loss": 0.213,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16197466850280762,
+      "step": 4485,
+      "valid_targets_mean": 4962.2,
+      "valid_targets_min": 1052
+    },
+    {
+      "epoch": 6.741741741741742,
+      "grad_norm": 0.5234925189927083,
+      "learning_rate": 1.676179240371245e-07,
+      "loss": 0.1807,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16373220086097717,
+      "step": 4490,
+      "valid_targets_mean": 3922.8,
+      "valid_targets_min": 464
+    },
+    {
+      "epoch": 6.7492492492492495,
+      "grad_norm": 0.6352684311920057,
+      "learning_rate": 1.5808163938846986e-07,
+      "loss": 0.1976,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19435149431228638,
+      "step": 4495,
+      "valid_targets_mean": 3218.4,
+      "valid_targets_min": 478
+    },
+    {
+      "epoch": 6.756756756756757,
+      "grad_norm": 0.6229594370415693,
+      "learning_rate": 1.4882355592427344e-07,
+      "loss": 0.191,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16698959469795227,
+      "step": 4500,
+      "valid_targets_mean": 3428.0,
+      "valid_targets_min": 665
+    },
+    {
+      "epoch": 6.7642642642642645,
+      "grad_norm": 0.6138075392270004,
+      "learning_rate": 1.3984380345102966e-07,
+      "loss": 0.2124,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21965721249580383,
+      "step": 4505,
+      "valid_targets_mean": 3452.5,
+      "valid_targets_min": 408
+    },
+    {
+      "epoch": 6.771771771771772,
+      "grad_norm": 0.5674625446735158,
+      "learning_rate": 1.3114250787278128e-07,
+      "loss": 0.2276,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17410314083099365,
+      "step": 4510,
+      "valid_targets_mean": 5005.2,
+      "valid_targets_min": 560
+    },
+    {
+      "epoch": 6.7792792792792795,
+      "grad_norm": 0.5558176857114882,
+      "learning_rate": 1.2271979118936074e-07,
+      "loss": 0.2114,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17912578582763672,
+      "step": 4515,
+      "valid_targets_mean": 3863.8,
+      "valid_targets_min": 853
+    },
+    {
+      "epoch": 6.786786786786787,
+      "grad_norm": 0.635474987483276,
+      "learning_rate": 1.1457577149468268e-07,
+      "loss": 0.1884,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17979639768600464,
+      "step": 4520,
+      "valid_targets_mean": 3391.7,
+      "valid_targets_min": 429
+    },
+    {
+      "epoch": 6.7942942942942945,
+      "grad_norm": 0.794907141047451,
+      "learning_rate": 1.0671056297507642e-07,
+      "loss": 0.2055,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19865292310714722,
+      "step": 4525,
+      "valid_targets_mean": 2033.5,
+      "valid_targets_min": 584
+    },
+    {
+      "epoch": 6.801801801801802,
+      "grad_norm": 0.49215782912003203,
+      "learning_rate": 9.91242759077049e-08,
+      "loss": 0.2204,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15488871932029724,
+      "step": 4530,
+      "valid_targets_mean": 4630.8,
+      "valid_targets_min": 639
+    },
+    {
+      "epoch": 6.8093093093093096,
+      "grad_norm": 0.6173142764754961,
+      "learning_rate": 9.181701665899268e-08,
+      "loss": 0.1977,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18007370829582214,
+      "step": 4535,
+      "valid_targets_mean": 4323.4,
+      "valid_targets_min": 696
+    },
+    {
+      "epoch": 6.816816816816817,
+      "grad_norm": 0.7575394360176801,
+      "learning_rate": 8.47888876831604e-08,
+      "loss": 0.2296,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24876675009727478,
+      "step": 4540,
+      "valid_targets_mean": 3193.3,
+      "valid_targets_min": 277
+    },
+    {
+      "epoch": 6.824324324324325,
+      "grad_norm": 0.5819569698944355,
+      "learning_rate": 7.803998752076824e-08,
+      "loss": 0.1881,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17346233129501343,
+      "step": 4545,
+      "valid_targets_mean": 5187.9,
+      "valid_targets_min": 643
+    },
+    {
+      "epoch": 6.831831831831832,
+      "grad_norm": 0.4396136217756035,
+      "learning_rate": 7.157041079734584e-08,
+      "loss": 0.2325,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18130743503570557,
+      "step": 4550,
+      "valid_targets_mean": 5762.4,
+      "valid_targets_min": 1228
+    },
+    {
+      "epoch": 6.83933933933934,
+      "grad_norm": 0.4687226820184863,
+      "learning_rate": 6.538024822206224e-08,
+      "loss": 0.2008,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1852032095193863,
+      "step": 4555,
+      "valid_targets_mean": 5690.4,
+      "valid_targets_min": 702
+    },
+    {
+      "epoch": 6.846846846846847,
+      "grad_norm": 0.6527972240202853,
+      "learning_rate": 5.946958658645141e-08,
+      "loss": 0.2255,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21088294684886932,
+      "step": 4560,
+      "valid_targets_mean": 3305.9,
+      "valid_targets_min": 371
+    },
+    {
+      "epoch": 6.854354354354355,
+      "grad_norm": 0.5747007613275369,
+      "learning_rate": 5.383850876319763e-08,
+      "loss": 0.1834,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16404512524604797,
+      "step": 4565,
+      "valid_targets_mean": 3822.6,
+      "valid_targets_min": 609
+    },
+    {
+      "epoch": 6.861861861861862,
+      "grad_norm": 0.5712891848122477,
+      "learning_rate": 4.848709370498084e-08,
+      "loss": 0.2022,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24901172518730164,
+      "step": 4570,
+      "valid_targets_mean": 5079.1,
+      "valid_targets_min": 574
+    },
+    {
+      "epoch": 6.86936936936937,
+      "grad_norm": 0.4949929365077814,
+      "learning_rate": 4.34154164433509e-08,
+      "loss": 0.1955,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17748858034610748,
+      "step": 4575,
+      "valid_targets_mean": 4491.8,
+      "valid_targets_min": 634
+    },
+    {
+      "epoch": 6.876876876876877,
+      "grad_norm": 0.5203752172969807,
+      "learning_rate": 3.86235480876973e-08,
+      "loss": 0.2374,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2086370289325714,
+      "step": 4580,
+      "valid_targets_mean": 5460.8,
+      "valid_targets_min": 622
+    },
+    {
+      "epoch": 6.884384384384385,
+      "grad_norm": 0.5124535100635832,
+      "learning_rate": 3.411155582423664e-08,
+      "loss": 0.1985,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16645345091819763,
+      "step": 4585,
+      "valid_targets_mean": 4649.4,
+      "valid_targets_min": 549
+    },
+    {
+      "epoch": 6.891891891891892,
+      "grad_norm": 0.6965621530697311,
+      "learning_rate": 2.9879502915075574e-08,
+      "loss": 0.2214,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21695250272750854,
+      "step": 4590,
+      "valid_targets_mean": 3357.6,
+      "valid_targets_min": 704
+    },
+    {
+      "epoch": 6.8993993993994,
+      "grad_norm": 0.5009549067493665,
+      "learning_rate": 2.592744869732933e-08,
+      "loss": 0.1793,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2433704286813736,
+      "step": 4595,
+      "valid_targets_mean": 5021.1,
+      "valid_targets_min": 535
+    },
+    {
+      "epoch": 6.906906906906907,
+      "grad_norm": 0.6325006211055703,
+      "learning_rate": 2.2255448582280127e-08,
+      "loss": 0.1923,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1930675059556961,
+      "step": 4600,
+      "valid_targets_mean": 3954.0,
+      "valid_targets_min": 562
+    },
+    {
+      "epoch": 6.914414414414415,
+      "grad_norm": 0.5272145532637147,
+      "learning_rate": 1.8863554054606715e-08,
+      "loss": 0.1888,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1625824272632599,
+      "step": 4605,
+      "valid_targets_mean": 4323.2,
+      "valid_targets_min": 1190
+    },
+    {
+      "epoch": 6.921921921921922,
+      "grad_norm": 0.5329661463696542,
+      "learning_rate": 1.5751812671658264e-08,
+      "loss": 0.2071,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19332969188690186,
+      "step": 4610,
+      "valid_targets_mean": 4747.9,
+      "valid_targets_min": 906
+    },
+    {
+      "epoch": 6.92942942942943,
+      "grad_norm": 0.5417369482444132,
+      "learning_rate": 1.2920268062794894e-08,
+      "loss": 0.2086,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19969046115875244,
+      "step": 4615,
+      "valid_targets_mean": 4192.8,
+      "valid_targets_min": 728
+    },
+    {
+      "epoch": 6.936936936936937,
+      "grad_norm": 0.6878033594654033,
+      "learning_rate": 1.0368959928763744e-08,
+      "loss": 0.2053,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21517297625541687,
+      "step": 4620,
+      "valid_targets_mean": 3462.9,
+      "valid_targets_min": 709
+    },
+    {
+      "epoch": 6.944444444444445,
+      "grad_norm": 0.5692311923901623,
+      "learning_rate": 8.097924041157168e-09,
+      "loss": 0.1876,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17874173820018768,
+      "step": 4625,
+      "valid_targets_mean": 3834.8,
+      "valid_targets_min": 605
+    },
+    {
+      "epoch": 6.951951951951952,
+      "grad_norm": 0.8120108501447949,
+      "learning_rate": 6.107192241897597e-09,
+      "loss": 0.1783,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19116508960723877,
+      "step": 4630,
+      "valid_targets_mean": 3737.9,
+      "valid_targets_min": 756
+    },
+    {
+      "epoch": 6.95945945945946,
+      "grad_norm": 0.6343758559646698,
+      "learning_rate": 4.396792442800113e-09,
+      "loss": 0.2174,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2372244894504547,
+      "step": 4635,
+      "valid_targets_mean": 5020.9,
+      "valid_targets_min": 821
+    },
+    {
+      "epoch": 6.966966966966967,
+      "grad_norm": 0.4747536212588191,
+      "learning_rate": 2.9667486251794274e-09,
+      "loss": 0.1961,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15623793005943298,
+      "step": 4640,
+      "valid_targets_mean": 5340.8,
+      "valid_targets_min": 611
+    },
+    {
+      "epoch": 6.974474474474475,
+      "grad_norm": 0.4663889436387785,
+      "learning_rate": 1.8170808395079342e-09,
+      "loss": 0.2291,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.204878568649292,
+      "step": 4645,
+      "valid_targets_mean": 5754.1,
+      "valid_targets_min": 2157
+    },
+    {
+      "epoch": 6.981981981981982,
+      "grad_norm": 0.5618969419816342,
+      "learning_rate": 9.47805205140373e-10,
+      "loss": 0.196,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2027861475944519,
+      "step": 4650,
+      "valid_targets_mean": 4324.8,
+      "valid_targets_min": 736
+    },
+    {
+      "epoch": 6.98948948948949,
+      "grad_norm": 0.440752938378733,
+      "learning_rate": 3.589339100917855e-10,
+      "loss": 0.1985,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21129390597343445,
+      "step": 4655,
+      "valid_targets_mean": 6267.5,
+      "valid_targets_min": 686
+    },
+    {
+      "epoch": 6.996996996996997,
+      "grad_norm": 0.5759092607170879,
+      "learning_rate": 5.0475210853218045e-11,
+      "loss": 0.2195,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21579566597938538,
+      "step": 4660,
+      "valid_targets_mean": 4775.2,
+      "valid_targets_min": 824
+    },
+    {
+      "epoch": 7.0,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1870056539773941,
+      "step": 4662,
+      "total_flos": 1420700030337024.0,
+      "train_loss": 0.25631797409579254,
+      "train_runtime": 31285.8836,
+      "train_samples_per_second": 2.384,
+      "train_steps_per_second": 0.149,
+      "valid_targets_mean": 4345.4,
+      "valid_targets_min": 624
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 4662,
+  "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": 1420700030337024.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}