diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5415 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.0,
+  "eval_steps": 500,
+  "global_step": 2442,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.012285012285012284,
+      "grad_norm": 19.340452072016063,
+      "learning_rate": 6.530612244897961e-07,
+      "loss": 1.0268,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 1.0363750457763672,
+      "step": 5,
+      "valid_targets_mean": 551.8,
+      "valid_targets_min": 379
+    },
+    {
+      "epoch": 0.02457002457002457,
+      "grad_norm": 17.08893563458667,
+      "learning_rate": 1.469387755102041e-06,
+      "loss": 0.9836,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.9560065865516663,
+      "step": 10,
+      "valid_targets_mean": 630.4,
+      "valid_targets_min": 357
+    },
+    {
+      "epoch": 0.036855036855036855,
+      "grad_norm": 15.515149146739999,
+      "learning_rate": 2.285714285714286e-06,
+      "loss": 0.9821,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.9447790384292603,
+      "step": 15,
+      "valid_targets_mean": 589.2,
+      "valid_targets_min": 395
+    },
+    {
+      "epoch": 0.04914004914004914,
+      "grad_norm": 9.407548224971022,
+      "learning_rate": 3.1020408163265307e-06,
+      "loss": 0.8773,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.8385097980499268,
+      "step": 20,
+      "valid_targets_mean": 603.5,
+      "valid_targets_min": 308
+    },
+    {
+      "epoch": 0.06142506142506143,
+      "grad_norm": 7.055702490270053,
+      "learning_rate": 3.9183673469387755e-06,
+      "loss": 0.82,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.7672586441040039,
+      "step": 25,
+      "valid_targets_mean": 571.4,
+      "valid_targets_min": 341
+    },
+    {
+      "epoch": 0.07371007371007371,
+      "grad_norm": 5.105243917945062,
+      "learning_rate": 4.734693877551021e-06,
+      "loss": 0.6919,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.6746348142623901,
+      "step": 30,
+      "valid_targets_mean": 494.1,
+      "valid_targets_min": 350
+    },
+    {
+      "epoch": 0.085995085995086,
+      "grad_norm": 3.791005812926048,
+      "learning_rate": 5.551020408163266e-06,
+      "loss": 0.6329,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.6144911050796509,
+      "step": 35,
+      "valid_targets_mean": 499.4,
+      "valid_targets_min": 327
+    },
+    {
+      "epoch": 0.09828009828009827,
+      "grad_norm": 3.0694531628102637,
+      "learning_rate": 6.36734693877551e-06,
+      "loss": 0.5631,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.5236115455627441,
+      "step": 40,
+      "valid_targets_mean": 531.1,
+      "valid_targets_min": 303
+    },
+    {
+      "epoch": 0.11056511056511056,
+      "grad_norm": 2.3798079841537203,
+      "learning_rate": 7.183673469387755e-06,
+      "loss": 0.4885,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.4935950040817261,
+      "step": 45,
+      "valid_targets_mean": 519.8,
+      "valid_targets_min": 309
+    },
+    {
+      "epoch": 0.12285012285012285,
+      "grad_norm": 2.326009549639266,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.4598,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.44546374678611755,
+      "step": 50,
+      "valid_targets_mean": 478.7,
+      "valid_targets_min": 325
+    },
+    {
+      "epoch": 0.13513513513513514,
+      "grad_norm": 1.8861705823036319,
+      "learning_rate": 8.816326530612247e-06,
+      "loss": 0.4127,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.4182875454425812,
+      "step": 55,
+      "valid_targets_mean": 551.9,
+      "valid_targets_min": 314
+    },
+    {
+      "epoch": 0.14742014742014742,
+      "grad_norm": 1.8700358687363636,
+      "learning_rate": 9.63265306122449e-06,
+      "loss": 0.4035,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3552461266517639,
+      "step": 60,
+      "valid_targets_mean": 541.9,
+      "valid_targets_min": 317
+    },
+    {
+      "epoch": 0.1597051597051597,
+      "grad_norm": 2.0142370613194434,
+      "learning_rate": 1.0448979591836737e-05,
+      "loss": 0.3729,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3656679391860962,
+      "step": 65,
+      "valid_targets_mean": 562.7,
+      "valid_targets_min": 358
+    },
+    {
+      "epoch": 0.171990171990172,
+      "grad_norm": 1.7529734182420011,
+      "learning_rate": 1.126530612244898e-05,
+      "loss": 0.3651,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3600107431411743,
+      "step": 70,
+      "valid_targets_mean": 590.9,
+      "valid_targets_min": 328
+    },
+    {
+      "epoch": 0.18427518427518427,
+      "grad_norm": 1.5209870168026196,
+      "learning_rate": 1.2081632653061225e-05,
+      "loss": 0.3538,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3366089463233948,
+      "step": 75,
+      "valid_targets_mean": 656.9,
+      "valid_targets_min": 366
+    },
+    {
+      "epoch": 0.19656019656019655,
+      "grad_norm": 1.71123642654244,
+      "learning_rate": 1.2897959183673469e-05,
+      "loss": 0.3303,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.35618680715560913,
+      "step": 80,
+      "valid_targets_mean": 606.8,
+      "valid_targets_min": 370
+    },
+    {
+      "epoch": 0.20884520884520885,
+      "grad_norm": 1.4853648763338492,
+      "learning_rate": 1.3714285714285716e-05,
+      "loss": 0.3245,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3359481692314148,
+      "step": 85,
+      "valid_targets_mean": 622.8,
+      "valid_targets_min": 380
+    },
+    {
+      "epoch": 0.22113022113022113,
+      "grad_norm": 1.361528089135814,
+      "learning_rate": 1.4530612244897961e-05,
+      "loss": 0.2952,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2936898469924927,
+      "step": 90,
+      "valid_targets_mean": 587.8,
+      "valid_targets_min": 388
+    },
+    {
+      "epoch": 0.2334152334152334,
+      "grad_norm": 4.740219098514282,
+      "learning_rate": 1.5346938775510204e-05,
+      "loss": 0.3211,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3229856491088867,
+      "step": 95,
+      "valid_targets_mean": 641.1,
+      "valid_targets_min": 319
+    },
+    {
+      "epoch": 0.2457002457002457,
+      "grad_norm": 1.5245883787059937,
+      "learning_rate": 1.616326530612245e-05,
+      "loss": 0.3009,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.280606210231781,
+      "step": 100,
+      "valid_targets_mean": 569.8,
+      "valid_targets_min": 329
+    },
+    {
+      "epoch": 0.257985257985258,
+      "grad_norm": 1.5406725035413686,
+      "learning_rate": 1.6979591836734695e-05,
+      "loss": 0.2913,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3024246394634247,
+      "step": 105,
+      "valid_targets_mean": 608.4,
+      "valid_targets_min": 326
+    },
+    {
+      "epoch": 0.2702702702702703,
+      "grad_norm": 1.4482316981295145,
+      "learning_rate": 1.779591836734694e-05,
+      "loss": 0.2983,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3170296549797058,
+      "step": 110,
+      "valid_targets_mean": 627.4,
+      "valid_targets_min": 353
+    },
+    {
+      "epoch": 0.28255528255528256,
+      "grad_norm": 1.4660214762255919,
+      "learning_rate": 1.8612244897959185e-05,
+      "loss": 0.2739,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.26000386476516724,
+      "step": 115,
+      "valid_targets_mean": 612.3,
+      "valid_targets_min": 389
+    },
+    {
+      "epoch": 0.29484029484029484,
+      "grad_norm": 1.6409256874657099,
+      "learning_rate": 1.942857142857143e-05,
+      "loss": 0.2961,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.3217686414718628,
+      "step": 120,
+      "valid_targets_mean": 561.9,
+      "valid_targets_min": 327
+    },
+    {
+      "epoch": 0.3071253071253071,
+      "grad_norm": 1.4065164837723425,
+      "learning_rate": 2.0244897959183672e-05,
+      "loss": 0.2921,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2969103753566742,
+      "step": 125,
+      "valid_targets_mean": 589.6,
+      "valid_targets_min": 310
+    },
+    {
+      "epoch": 0.3194103194103194,
+      "grad_norm": 1.5813750575415058,
+      "learning_rate": 2.106122448979592e-05,
+      "loss": 0.2818,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2890018820762634,
+      "step": 130,
+      "valid_targets_mean": 586.1,
+      "valid_targets_min": 300
+    },
+    {
+      "epoch": 0.3316953316953317,
+      "grad_norm": 1.5224653877090757,
+      "learning_rate": 2.1877551020408166e-05,
+      "loss": 0.2917,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2904941439628601,
+      "step": 135,
+      "valid_targets_mean": 576.4,
+      "valid_targets_min": 354
+    },
+    {
+      "epoch": 0.343980343980344,
+      "grad_norm": 1.3452861170507189,
+      "learning_rate": 2.269387755102041e-05,
+      "loss": 0.2826,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.31476208567619324,
+      "step": 140,
+      "valid_targets_mean": 692.4,
+      "valid_targets_min": 352
+    },
+    {
+      "epoch": 0.35626535626535627,
+      "grad_norm": 1.5523700640116074,
+      "learning_rate": 2.3510204081632656e-05,
+      "loss": 0.2606,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23628346621990204,
+      "step": 145,
+      "valid_targets_mean": 509.4,
+      "valid_targets_min": 329
+    },
+    {
+      "epoch": 0.36855036855036855,
+      "grad_norm": 1.4055510862867506,
+      "learning_rate": 2.4326530612244898e-05,
+      "loss": 0.2773,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.28314584493637085,
+      "step": 150,
+      "valid_targets_mean": 613.8,
+      "valid_targets_min": 387
+    },
+    {
+      "epoch": 0.3808353808353808,
+      "grad_norm": 1.5047918312671191,
+      "learning_rate": 2.5142857142857143e-05,
+      "loss": 0.2662,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2703036665916443,
+      "step": 155,
+      "valid_targets_mean": 611.1,
+      "valid_targets_min": 373
+    },
+    {
+      "epoch": 0.3931203931203931,
+      "grad_norm": 1.6190760418995298,
+      "learning_rate": 2.5959183673469392e-05,
+      "loss": 0.2813,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2736574411392212,
+      "step": 160,
+      "valid_targets_mean": 575.7,
+      "valid_targets_min": 380
+    },
+    {
+      "epoch": 0.40540540540540543,
+      "grad_norm": 1.5123475030605578,
+      "learning_rate": 2.6775510204081637e-05,
+      "loss": 0.2607,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2557530105113983,
+      "step": 165,
+      "valid_targets_mean": 565.6,
+      "valid_targets_min": 344
+    },
+    {
+      "epoch": 0.4176904176904177,
+      "grad_norm": 1.619694103526443,
+      "learning_rate": 2.7591836734693882e-05,
+      "loss": 0.2724,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27611520886421204,
+      "step": 170,
+      "valid_targets_mean": 541.1,
+      "valid_targets_min": 342
+    },
+    {
+      "epoch": 0.42997542997543,
+      "grad_norm": 1.4826756269933499,
+      "learning_rate": 2.8408163265306124e-05,
+      "loss": 0.2629,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2644728720188141,
+      "step": 175,
+      "valid_targets_mean": 522.6,
+      "valid_targets_min": 364
+    },
+    {
+      "epoch": 0.44226044226044225,
+      "grad_norm": 1.5670463191851074,
+      "learning_rate": 2.922448979591837e-05,
+      "loss": 0.2757,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.27483248710632324,
+      "step": 180,
+      "valid_targets_mean": 602.3,
+      "valid_targets_min": 327
+    },
+    {
+      "epoch": 0.45454545454545453,
+      "grad_norm": 1.408617532690139,
+      "learning_rate": 3.0040816326530614e-05,
+      "loss": 0.2585,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24957937002182007,
+      "step": 185,
+      "valid_targets_mean": 599.8,
+      "valid_targets_min": 399
+    },
+    {
+      "epoch": 0.4668304668304668,
+      "grad_norm": 1.4355940504839466,
+      "learning_rate": 3.085714285714286e-05,
+      "loss": 0.2652,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2865978181362152,
+      "step": 190,
+      "valid_targets_mean": 613.9,
+      "valid_targets_min": 364
+    },
+    {
+      "epoch": 0.47911547911547914,
+      "grad_norm": 1.3616233501948924,
+      "learning_rate": 3.1673469387755105e-05,
+      "loss": 0.2643,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24286630749702454,
+      "step": 195,
+      "valid_targets_mean": 604.1,
+      "valid_targets_min": 362
+    },
+    {
+      "epoch": 0.4914004914004914,
+      "grad_norm": 1.3541147315022435,
+      "learning_rate": 3.2489795918367346e-05,
+      "loss": 0.2535,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22083410620689392,
+      "step": 200,
+      "valid_targets_mean": 519.1,
+      "valid_targets_min": 318
+    },
+    {
+      "epoch": 0.5036855036855037,
+      "grad_norm": 1.5699583522473184,
+      "learning_rate": 3.3306122448979595e-05,
+      "loss": 0.2663,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2536630928516388,
+      "step": 205,
+      "valid_targets_mean": 503.4,
+      "valid_targets_min": 291
+    },
+    {
+      "epoch": 0.515970515970516,
+      "grad_norm": 1.4524492019396484,
+      "learning_rate": 3.4122448979591843e-05,
+      "loss": 0.2574,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2488933801651001,
+      "step": 210,
+      "valid_targets_mean": 568.8,
+      "valid_targets_min": 358
+    },
+    {
+      "epoch": 0.5282555282555282,
+      "grad_norm": 1.455074638920383,
+      "learning_rate": 3.4938775510204085e-05,
+      "loss": 0.253,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2419266402721405,
+      "step": 215,
+      "valid_targets_mean": 550.0,
+      "valid_targets_min": 332
+    },
+    {
+      "epoch": 0.5405405405405406,
+      "grad_norm": 1.4702983842648538,
+      "learning_rate": 3.575510204081633e-05,
+      "loss": 0.2443,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24988505244255066,
+      "step": 220,
+      "valid_targets_mean": 522.4,
+      "valid_targets_min": 314
+    },
+    {
+      "epoch": 0.5528255528255528,
+      "grad_norm": 1.2678095680289083,
+      "learning_rate": 3.6571428571428576e-05,
+      "loss": 0.2535,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23922167718410492,
+      "step": 225,
+      "valid_targets_mean": 618.5,
+      "valid_targets_min": 391
+    },
+    {
+      "epoch": 0.5651105651105651,
+      "grad_norm": 1.3124904196473028,
+      "learning_rate": 3.738775510204082e-05,
+      "loss": 0.2513,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2686520218849182,
+      "step": 230,
+      "valid_targets_mean": 610.7,
+      "valid_targets_min": 348
+    },
+    {
+      "epoch": 0.5773955773955773,
+      "grad_norm": 1.2321365465764669,
+      "learning_rate": 3.8204081632653066e-05,
+      "loss": 0.2429,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22676542401313782,
+      "step": 235,
+      "valid_targets_mean": 529.8,
+      "valid_targets_min": 340
+    },
+    {
+      "epoch": 0.5896805896805897,
+      "grad_norm": 1.4304495261227945,
+      "learning_rate": 3.902040816326531e-05,
+      "loss": 0.2402,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22461196780204773,
+      "step": 240,
+      "valid_targets_mean": 525.8,
+      "valid_targets_min": 351
+    },
+    {
+      "epoch": 0.601965601965602,
+      "grad_norm": 1.2501593563439926,
+      "learning_rate": 3.983673469387755e-05,
+      "loss": 0.2553,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2536337077617645,
+      "step": 245,
+      "valid_targets_mean": 624.1,
+      "valid_targets_min": 341
+    },
+    {
+      "epoch": 0.6142506142506142,
+      "grad_norm": 1.365723699724294,
+      "learning_rate": 3.9999672841332876e-05,
+      "loss": 0.2569,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24196216464042664,
+      "step": 250,
+      "valid_targets_mean": 572.6,
+      "valid_targets_min": 377
+    },
+    {
+      "epoch": 0.6265356265356266,
+      "grad_norm": 1.4290316134985837,
+      "learning_rate": 3.999834377759164e-05,
+      "loss": 0.2583,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2442513257265091,
+      "step": 255,
+      "valid_targets_mean": 570.1,
+      "valid_targets_min": 392
+    },
+    {
+      "epoch": 0.6388206388206388,
+      "grad_norm": 1.279025523831711,
+      "learning_rate": 3.999599242924703e-05,
+      "loss": 0.2602,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2676926255226135,
+      "step": 260,
+      "valid_targets_mean": 616.0,
+      "valid_targets_min": 359
+    },
+    {
+      "epoch": 0.6511056511056511,
+      "grad_norm": 1.3742183041274103,
+      "learning_rate": 3.999261891649637e-05,
+      "loss": 0.2446,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23325861990451813,
+      "step": 265,
+      "valid_targets_mean": 535.8,
+      "valid_targets_min": 362
+    },
+    {
+      "epoch": 0.6633906633906634,
+      "grad_norm": 1.3319257895644228,
+      "learning_rate": 3.9988223411788436e-05,
+      "loss": 0.2553,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23690742254257202,
+      "step": 270,
+      "valid_targets_mean": 529.6,
+      "valid_targets_min": 336
+    },
+    {
+      "epoch": 0.6756756756756757,
+      "grad_norm": 1.2333251207411267,
+      "learning_rate": 3.998280613981468e-05,
+      "loss": 0.2441,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2443428933620453,
+      "step": 275,
+      "valid_targets_mean": 592.6,
+      "valid_targets_min": 381
+    },
+    {
+      "epoch": 0.687960687960688,
+      "grad_norm": 1.2653219638572915,
+      "learning_rate": 3.9976367377497725e-05,
+      "loss": 0.2533,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2635766565799713,
+      "step": 280,
+      "valid_targets_mean": 661.1,
+      "valid_targets_min": 333
+    },
+    {
+      "epoch": 0.7002457002457002,
+      "grad_norm": 1.2132987899715728,
+      "learning_rate": 3.99689074539772e-05,
+      "loss": 0.2428,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23148810863494873,
+      "step": 285,
+      "valid_targets_mean": 590.9,
+      "valid_targets_min": 342
+    },
+    {
+      "epoch": 0.7125307125307125,
+      "grad_norm": 1.3232052555393115,
+      "learning_rate": 3.9960426750592936e-05,
+      "loss": 0.2366,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2554059624671936,
+      "step": 290,
+      "valid_targets_mean": 586.4,
+      "valid_targets_min": 391
+    },
+    {
+      "epoch": 0.7248157248157249,
+      "grad_norm": 1.3250013840237953,
+      "learning_rate": 3.995092570086546e-05,
+      "loss": 0.2372,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23205137252807617,
+      "step": 295,
+      "valid_targets_mean": 534.9,
+      "valid_targets_min": 379
+    },
+    {
+      "epoch": 0.7371007371007371,
+      "grad_norm": 1.174536495094598,
+      "learning_rate": 3.9940404790473825e-05,
+      "loss": 0.2362,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23327840864658356,
+      "step": 300,
+      "valid_targets_mean": 577.4,
+      "valid_targets_min": 338
+    },
+    {
+      "epoch": 0.7493857493857494,
+      "grad_norm": 1.2736986293286983,
+      "learning_rate": 3.992886455723082e-05,
+      "loss": 0.2423,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23847690224647522,
+      "step": 305,
+      "valid_targets_mean": 604.8,
+      "valid_targets_min": 356
+    },
+    {
+      "epoch": 0.7616707616707616,
+      "grad_norm": 1.202738294769371,
+      "learning_rate": 3.991630559105541e-05,
+      "loss": 0.2429,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.25206199288368225,
+      "step": 310,
+      "valid_targets_mean": 595.6,
+      "valid_targets_min": 336
+    },
+    {
+      "epoch": 0.773955773955774,
+      "grad_norm": 1.1705643307443048,
+      "learning_rate": 3.990272853394268e-05,
+      "loss": 0.2401,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2267996072769165,
+      "step": 315,
+      "valid_targets_mean": 609.8,
+      "valid_targets_min": 350
+    },
+    {
+      "epoch": 0.7862407862407862,
+      "grad_norm": 1.2741625837837056,
+      "learning_rate": 3.988813407993089e-05,
+      "loss": 0.2272,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22659529745578766,
+      "step": 320,
+      "valid_targets_mean": 520.9,
+      "valid_targets_min": 339
+    },
+    {
+      "epoch": 0.7985257985257985,
+      "grad_norm": 1.2095767910140882,
+      "learning_rate": 3.987252297506613e-05,
+      "loss": 0.2387,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23505187034606934,
+      "step": 325,
+      "valid_targets_mean": 557.3,
+      "valid_targets_min": 351
+    },
+    {
+      "epoch": 0.8108108108108109,
+      "grad_norm": 1.2469501504557738,
+      "learning_rate": 3.9855896017364075e-05,
+      "loss": 0.2361,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2419613003730774,
+      "step": 330,
+      "valid_targets_mean": 522.1,
+      "valid_targets_min": 333
+    },
+    {
+      "epoch": 0.8230958230958231,
+      "grad_norm": 1.3525431983270617,
+      "learning_rate": 3.983825405676927e-05,
+      "loss": 0.2425,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23498916625976562,
+      "step": 335,
+      "valid_targets_mean": 526.7,
+      "valid_targets_min": 364
+    },
+    {
+      "epoch": 0.8353808353808354,
+      "grad_norm": 1.266098276438524,
+      "learning_rate": 3.981959799511161e-05,
+      "loss": 0.2361,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22201555967330933,
+      "step": 340,
+      "valid_targets_mean": 516.1,
+      "valid_targets_min": 287
+    },
+    {
+      "epoch": 0.8476658476658476,
+      "grad_norm": 1.1950372008453742,
+      "learning_rate": 3.979992878606032e-05,
+      "loss": 0.2391,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23893003165721893,
+      "step": 345,
+      "valid_targets_mean": 604.9,
+      "valid_targets_min": 358
+    },
+    {
+      "epoch": 0.85995085995086,
+      "grad_norm": 1.1942813421589011,
+      "learning_rate": 3.977924743507513e-05,
+      "loss": 0.2423,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2164364457130432,
+      "step": 350,
+      "valid_targets_mean": 568.4,
+      "valid_targets_min": 386
+    },
+    {
+      "epoch": 0.8722358722358723,
+      "grad_norm": 1.1521319253994895,
+      "learning_rate": 3.975755499935492e-05,
+      "loss": 0.2541,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2973342835903168,
+      "step": 355,
+      "valid_targets_mean": 677.1,
+      "valid_targets_min": 359
+    },
+    {
+      "epoch": 0.8845208845208845,
+      "grad_norm": 1.2819519264235832,
+      "learning_rate": 3.973485258778368e-05,
+      "loss": 0.2315,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2431134581565857,
+      "step": 360,
+      "valid_targets_mean": 520.0,
+      "valid_targets_min": 320
+    },
+    {
+      "epoch": 0.8968058968058968,
+      "grad_norm": 1.121053883863795,
+      "learning_rate": 3.971114136087379e-05,
+      "loss": 0.242,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2276688814163208,
+      "step": 365,
+      "valid_targets_mean": 623.6,
+      "valid_targets_min": 314
+    },
+    {
+      "epoch": 0.9090909090909091,
+      "grad_norm": 1.2335240034008033,
+      "learning_rate": 3.968642253070675e-05,
+      "loss": 0.2506,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24868804216384888,
+      "step": 370,
+      "valid_targets_mean": 486.9,
+      "valid_targets_min": 349
+    },
+    {
+      "epoch": 0.9213759213759214,
+      "grad_norm": 1.084721824161106,
+      "learning_rate": 3.966069736087116e-05,
+      "loss": 0.2419,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21974554657936096,
+      "step": 375,
+      "valid_targets_mean": 572.9,
+      "valid_targets_min": 388
+    },
+    {
+      "epoch": 0.9336609336609336,
+      "grad_norm": 1.5415223968516378,
+      "learning_rate": 3.963396716639818e-05,
+      "loss": 0.23,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22481045126914978,
+      "step": 380,
+      "valid_targets_mean": 513.1,
+      "valid_targets_min": 335
+    },
+    {
+      "epoch": 0.9459459459459459,
+      "grad_norm": 1.184005524489551,
+      "learning_rate": 3.960623331369427e-05,
+      "loss": 0.229,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20717152953147888,
+      "step": 385,
+      "valid_targets_mean": 569.8,
+      "valid_targets_min": 378
+    },
+    {
+      "epoch": 0.9582309582309583,
+      "grad_norm": 1.1736401735916975,
+      "learning_rate": 3.957749722047138e-05,
+      "loss": 0.2246,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24852100014686584,
+      "step": 390,
+      "valid_targets_mean": 562.8,
+      "valid_targets_min": 340
+    },
+    {
+      "epoch": 0.9705159705159705,
+      "grad_norm": 1.2154851241242357,
+      "learning_rate": 3.9547760355674405e-05,
+      "loss": 0.2306,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2460574209690094,
+      "step": 395,
+      "valid_targets_mean": 592.2,
+      "valid_targets_min": 315
+    },
+    {
+      "epoch": 0.9828009828009828,
+      "grad_norm": 1.1414024083345906,
+      "learning_rate": 3.951702423940621e-05,
+      "loss": 0.2432,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.273346483707428,
+      "step": 400,
+      "valid_targets_mean": 704.4,
+      "valid_targets_min": 372
+    },
+    {
+      "epoch": 0.995085995085995,
+      "grad_norm": 1.2888653065517275,
+      "learning_rate": 3.948529044284981e-05,
+      "loss": 0.2285,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2338770031929016,
+      "step": 405,
+      "valid_targets_mean": 528.1,
+      "valid_targets_min": 383
+    },
+    {
+      "epoch": 1.0073710073710074,
+      "grad_norm": 1.273222474953753,
+      "learning_rate": 3.9452560588188135e-05,
+      "loss": 0.2126,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20872265100479126,
+      "step": 410,
+      "valid_targets_mean": 483.4,
+      "valid_targets_min": 345
+    },
+    {
+      "epoch": 1.0196560196560196,
+      "grad_norm": 1.1230278716937114,
+      "learning_rate": 3.9418836348521045e-05,
+      "loss": 0.2122,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1878654807806015,
+      "step": 415,
+      "valid_targets_mean": 560.2,
+      "valid_targets_min": 350
+    },
+    {
+      "epoch": 1.031941031941032,
+      "grad_norm": 1.3385969964323565,
+      "learning_rate": 3.9384119447779854e-05,
+      "loss": 0.2053,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2124098837375641,
+      "step": 420,
+      "valid_targets_mean": 531.4,
+      "valid_targets_min": 346
+    },
+    {
+      "epoch": 1.0442260442260443,
+      "grad_norm": 1.133701044120653,
+      "learning_rate": 3.934841166063919e-05,
+      "loss": 0.2151,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21468473970890045,
+      "step": 425,
+      "valid_targets_mean": 593.8,
+      "valid_targets_min": 337
+    },
+    {
+      "epoch": 1.0565110565110565,
+      "grad_norm": 1.1884236103320405,
+      "learning_rate": 3.931171481242625e-05,
+      "loss": 0.1999,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20533575117588043,
+      "step": 430,
+      "valid_targets_mean": 586.6,
+      "valid_targets_min": 364
+    },
+    {
+      "epoch": 1.0687960687960687,
+      "grad_norm": 1.2358471031407317,
+      "learning_rate": 3.927403077902753e-05,
+      "loss": 0.2029,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20509561896324158,
+      "step": 435,
+      "valid_targets_mean": 554.1,
+      "valid_targets_min": 303
+    },
+    {
+      "epoch": 1.0810810810810811,
+      "grad_norm": 1.1162627713390312,
+      "learning_rate": 3.9235361486792905e-05,
+      "loss": 0.2007,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2194371223449707,
+      "step": 440,
+      "valid_targets_mean": 592.0,
+      "valid_targets_min": 391
+    },
+    {
+      "epoch": 1.0933660933660934,
+      "grad_norm": 1.1664113838151917,
+      "learning_rate": 3.9195708912437176e-05,
+      "loss": 0.2222,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19551971554756165,
+      "step": 445,
+      "valid_targets_mean": 562.9,
+      "valid_targets_min": 299
+    },
+    {
+      "epoch": 1.1056511056511056,
+      "grad_norm": 1.1148563823576316,
+      "learning_rate": 3.915507508293901e-05,
+      "loss": 0.2028,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.224244624376297,
+      "step": 450,
+      "valid_targets_mean": 601.6,
+      "valid_targets_min": 341
+    },
+    {
+      "epoch": 1.117936117936118,
+      "grad_norm": 1.209728679709457,
+      "learning_rate": 3.911346207543734e-05,
+      "loss": 0.1981,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21454249322414398,
+      "step": 455,
+      "valid_targets_mean": 560.9,
+      "valid_targets_min": 333
+    },
+    {
+      "epoch": 1.1302211302211302,
+      "grad_norm": 1.0703696823302942,
+      "learning_rate": 3.907087201712515e-05,
+      "loss": 0.2154,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22700481116771698,
+      "step": 460,
+      "valid_targets_mean": 657.5,
+      "valid_targets_min": 321
+    },
+    {
+      "epoch": 1.1425061425061425,
+      "grad_norm": 1.087498535135239,
+      "learning_rate": 3.902730708514078e-05,
+      "loss": 0.2037,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19121837615966797,
+      "step": 465,
+      "valid_targets_mean": 605.1,
+      "valid_targets_min": 414
+    },
+    {
+      "epoch": 1.154791154791155,
+      "grad_norm": 1.1754498968382059,
+      "learning_rate": 3.8982769506456616e-05,
+      "loss": 0.1997,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2021336704492569,
+      "step": 470,
+      "valid_targets_mean": 610.5,
+      "valid_targets_min": 390
+    },
+    {
+      "epoch": 1.1670761670761671,
+      "grad_norm": 1.0324644891352066,
+      "learning_rate": 3.893726155776524e-05,
+      "loss": 0.2022,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16948124766349792,
+      "step": 475,
+      "valid_targets_mean": 563.5,
+      "valid_targets_min": 355
+    },
+    {
+      "epoch": 1.1793611793611793,
+      "grad_norm": 1.1444059489981688,
+      "learning_rate": 3.8890785565363046e-05,
+      "loss": 0.2131,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22946152091026306,
+      "step": 480,
+      "valid_targets_mean": 585.4,
+      "valid_targets_min": 333
+    },
+    {
+      "epoch": 1.1916461916461916,
+      "grad_norm": 1.0980010911655984,
+      "learning_rate": 3.884334390503136e-05,
+      "loss": 0.2136,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20137803256511688,
+      "step": 485,
+      "valid_targets_mean": 556.1,
+      "valid_targets_min": 352
+    },
+    {
+      "epoch": 1.203931203931204,
+      "grad_norm": 1.2187757226237967,
+      "learning_rate": 3.8794939001914955e-05,
+      "loss": 0.209,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2199264019727707,
+      "step": 490,
+      "valid_targets_mean": 605.1,
+      "valid_targets_min": 350
+    },
+    {
+      "epoch": 1.2162162162162162,
+      "grad_norm": 1.291377657263299,
+      "learning_rate": 3.87455733303981e-05,
+      "loss": 0.2161,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22223210334777832,
+      "step": 495,
+      "valid_targets_mean": 562.2,
+      "valid_targets_min": 316
+    },
+    {
+      "epoch": 1.2285012285012284,
+      "grad_norm": 1.087891213866882,
+      "learning_rate": 3.869524941397805e-05,
+      "loss": 0.2058,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22227486968040466,
+      "step": 500,
+      "valid_targets_mean": 561.1,
+      "valid_targets_min": 374
+    },
+    {
+      "epoch": 1.2407862407862407,
+      "grad_norm": 1.0875951457065944,
+      "learning_rate": 3.8643969825136095e-05,
+      "loss": 0.209,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21871165931224823,
+      "step": 505,
+      "valid_targets_mean": 605.8,
+      "valid_targets_min": 291
+    },
+    {
+      "epoch": 1.253071253071253,
+      "grad_norm": 1.1345287983708072,
+      "learning_rate": 3.8591737185206024e-05,
+      "loss": 0.2134,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19657248258590698,
+      "step": 510,
+      "valid_targets_mean": 517.6,
+      "valid_targets_min": 365
+    },
+    {
+      "epoch": 1.2653562653562653,
+      "grad_norm": 1.1578099843167293,
+      "learning_rate": 3.853855416424011e-05,
+      "loss": 0.2046,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22923174500465393,
+      "step": 515,
+      "valid_targets_mean": 629.9,
+      "valid_targets_min": 351
+    },
+    {
+      "epoch": 1.2776412776412776,
+      "grad_norm": 1.3467752290200405,
+      "learning_rate": 3.848442348087267e-05,
+      "loss": 0.2129,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22567245364189148,
+      "step": 520,
+      "valid_targets_mean": 493.1,
+      "valid_targets_min": 348
+    },
+    {
+      "epoch": 1.28992628992629,
+      "grad_norm": 1.0355038419138378,
+      "learning_rate": 3.842934790218106e-05,
+      "loss": 0.2027,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1945795863866806,
+      "step": 525,
+      "valid_targets_mean": 609.4,
+      "valid_targets_min": 355
+    },
+    {
+      "epoch": 1.3022113022113022,
+      "grad_norm": 1.098408635608927,
+      "learning_rate": 3.837333024354422e-05,
+      "loss": 0.2111,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21321016550064087,
+      "step": 530,
+      "valid_targets_mean": 629.6,
+      "valid_targets_min": 334
+    },
+    {
+      "epoch": 1.3144963144963144,
+      "grad_norm": 1.1965501456460534,
+      "learning_rate": 3.8316373368498794e-05,
+      "loss": 0.197,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20622071623802185,
+      "step": 535,
+      "valid_targets_mean": 498.1,
+      "valid_targets_min": 329
+    },
+    {
+      "epoch": 1.3267813267813269,
+      "grad_norm": 1.2780117337853627,
+      "learning_rate": 3.82584801885927e-05,
+      "loss": 0.2112,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21127185225486755,
+      "step": 540,
+      "valid_targets_mean": 512.1,
+      "valid_targets_min": 272
+    },
+    {
+      "epoch": 1.339066339066339,
+      "grad_norm": 1.0208209030208835,
+      "learning_rate": 3.8199653663236336e-05,
+      "loss": 0.2215,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23393374681472778,
+      "step": 545,
+      "valid_targets_mean": 637.0,
+      "valid_targets_min": 389
+    },
+    {
+      "epoch": 1.3513513513513513,
+      "grad_norm": 1.0756398207865907,
+      "learning_rate": 3.813989679955128e-05,
+      "loss": 0.2102,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21633684635162354,
+      "step": 550,
+      "valid_targets_mean": 600.6,
+      "valid_targets_min": 332
+    },
+    {
+      "epoch": 1.3636363636363638,
+      "grad_norm": 1.2709984419070997,
+      "learning_rate": 3.8079212652216595e-05,
+      "loss": 0.1973,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19809314608573914,
+      "step": 555,
+      "valid_targets_mean": 507.6,
+      "valid_targets_min": 391
+    },
+    {
+      "epoch": 1.375921375921376,
+      "grad_norm": 1.064418446515292,
+      "learning_rate": 3.8017604323312616e-05,
+      "loss": 0.2009,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2035004198551178,
+      "step": 560,
+      "valid_targets_mean": 572.9,
+      "valid_targets_min": 366
+    },
+    {
+      "epoch": 1.3882063882063882,
+      "grad_norm": 1.1078788278566427,
+      "learning_rate": 3.795507496216246e-05,
+      "loss": 0.2136,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.24907588958740234,
+      "step": 565,
+      "valid_targets_mean": 608.2,
+      "valid_targets_min": 386
+    },
+    {
+      "epoch": 1.4004914004914004,
+      "grad_norm": 1.0256660334768586,
+      "learning_rate": 3.789162776517098e-05,
+      "loss": 0.2026,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2145499885082245,
+      "step": 570,
+      "valid_targets_mean": 657.9,
+      "valid_targets_min": 373
+    },
+    {
+      "epoch": 1.4127764127764126,
+      "grad_norm": 1.034031043079854,
+      "learning_rate": 3.78272659756614e-05,
+      "loss": 0.2183,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2272603064775467,
+      "step": 575,
+      "valid_targets_mean": 645.5,
+      "valid_targets_min": 340
+    },
+    {
+      "epoch": 1.425061425061425,
+      "grad_norm": 1.2896668422957693,
+      "learning_rate": 3.776199288370948e-05,
+      "loss": 0.2177,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23518535494804382,
+      "step": 580,
+      "valid_targets_mean": 468.1,
+      "valid_targets_min": 319
+    },
+    {
+      "epoch": 1.4373464373464373,
+      "grad_norm": 0.8865050790970145,
+      "learning_rate": 3.7695811825975386e-05,
+      "loss": 0.2022,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1835131049156189,
+      "step": 585,
+      "valid_targets_mean": 766.4,
+      "valid_targets_min": 556
+    },
+    {
+      "epoch": 1.4496314496314495,
+      "grad_norm": 1.0052694619139135,
+      "learning_rate": 3.76287261855331e-05,
+      "loss": 0.1979,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21307772397994995,
+      "step": 590,
+      "valid_targets_mean": 663.4,
+      "valid_targets_min": 324
+    },
+    {
+      "epoch": 1.461916461916462,
+      "grad_norm": 0.968909885622136,
+      "learning_rate": 3.7560739391697465e-05,
+      "loss": 0.2058,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1846340447664261,
+      "step": 595,
+      "valid_targets_mean": 574.1,
+      "valid_targets_min": 357
+    },
+    {
+      "epoch": 1.4742014742014742,
+      "grad_norm": 1.008739229853549,
+      "learning_rate": 3.749185491984891e-05,
+      "loss": 0.2011,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1903940886259079,
+      "step": 600,
+      "valid_targets_mean": 651.4,
+      "valid_targets_min": 349
+    },
+    {
+      "epoch": 1.4864864864864864,
+      "grad_norm": 1.0894449533360777,
+      "learning_rate": 3.7422076291255785e-05,
+      "loss": 0.1951,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19377467036247253,
+      "step": 605,
+      "valid_targets_mean": 501.9,
+      "valid_targets_min": 360
+    },
+    {
+      "epoch": 1.4987714987714988,
+      "grad_norm": 1.1134116216367664,
+      "learning_rate": 3.7351407072894356e-05,
+      "loss": 0.2005,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19366130232810974,
+      "step": 610,
+      "valid_targets_mean": 540.0,
+      "valid_targets_min": 323
+    },
+    {
+      "epoch": 1.511056511056511,
+      "grad_norm": 1.1085285737625885,
+      "learning_rate": 3.7279850877266486e-05,
+      "loss": 0.2165,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.23652777075767517,
+      "step": 615,
+      "valid_targets_mean": 552.9,
+      "valid_targets_min": 330
+    },
+    {
+      "epoch": 1.5233415233415233,
+      "grad_norm": 1.1659336753490528,
+      "learning_rate": 3.720741136221491e-05,
+      "loss": 0.1904,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20393647253513336,
+      "step": 620,
+      "valid_targets_mean": 562.8,
+      "valid_targets_min": 339
+    },
+    {
+      "epoch": 1.5356265356265357,
+      "grad_norm": 1.1256183805366937,
+      "learning_rate": 3.713409223073636e-05,
+      "loss": 0.2087,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21131408214569092,
+      "step": 625,
+      "valid_targets_mean": 596.2,
+      "valid_targets_min": 361
+    },
+    {
+      "epoch": 1.547911547911548,
+      "grad_norm": 1.1025314376028277,
+      "learning_rate": 3.705989723079214e-05,
+      "loss": 0.1998,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2073504477739334,
+      "step": 630,
+      "valid_targets_mean": 534.4,
+      "valid_targets_min": 347
+    },
+    {
+      "epoch": 1.5601965601965602,
+      "grad_norm": 1.2328995169055594,
+      "learning_rate": 3.698483015511665e-05,
+      "loss": 0.2005,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20556914806365967,
+      "step": 635,
+      "valid_targets_mean": 556.0,
+      "valid_targets_min": 393
+    },
+    {
+      "epoch": 1.5724815724815726,
+      "grad_norm": 1.13051808020514,
+      "learning_rate": 3.690889484102344e-05,
+      "loss": 0.2034,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20527680218219757,
+      "step": 640,
+      "valid_targets_mean": 560.2,
+      "valid_targets_min": 387
+    },
+    {
+      "epoch": 1.5847665847665846,
+      "grad_norm": 1.0356033195955525,
+      "learning_rate": 3.683209517020908e-05,
+      "loss": 0.213,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21655723452568054,
+      "step": 645,
+      "valid_targets_mean": 582.9,
+      "valid_targets_min": 338
+    },
+    {
+      "epoch": 1.597051597051597,
+      "grad_norm": 1.1170971606578726,
+      "learning_rate": 3.675443506855473e-05,
+      "loss": 0.2036,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2006620168685913,
+      "step": 650,
+      "valid_targets_mean": 613.5,
+      "valid_targets_min": 318
+    },
+    {
+      "epoch": 1.6093366093366095,
+      "grad_norm": 1.133535405808262,
+      "learning_rate": 3.6675918505925456e-05,
+      "loss": 0.201,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18652908504009247,
+      "step": 655,
+      "valid_targets_mean": 495.4,
+      "valid_targets_min": 330
+    },
+    {
+      "epoch": 1.6216216216216215,
+      "grad_norm": 0.9925583371315405,
+      "learning_rate": 3.6596549495967276e-05,
+      "loss": 0.2068,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2260739654302597,
+      "step": 660,
+      "valid_targets_mean": 691.1,
+      "valid_targets_min": 372
+    },
+    {
+      "epoch": 1.633906633906634,
+      "grad_norm": 1.145416405295079,
+      "learning_rate": 3.651633209590202e-05,
+      "loss": 0.1991,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1762043833732605,
+      "step": 665,
+      "valid_targets_mean": 534.9,
+      "valid_targets_min": 313
+    },
+    {
+      "epoch": 1.6461916461916462,
+      "grad_norm": 1.1479183598008718,
+      "learning_rate": 3.6435270406319914e-05,
+      "loss": 0.2345,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22825786471366882,
+      "step": 670,
+      "valid_targets_mean": 511.7,
+      "valid_targets_min": 382
+    },
+    {
+      "epoch": 1.6584766584766584,
+      "grad_norm": 1.0177952792861085,
+      "learning_rate": 3.635336857096997e-05,
+      "loss": 0.2002,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1958564817905426,
+      "step": 675,
+      "valid_targets_mean": 596.1,
+      "valid_targets_min": 367
+    },
+    {
+      "epoch": 1.6707616707616708,
+      "grad_norm": 0.9497476755880511,
+      "learning_rate": 3.627063077654815e-05,
+      "loss": 0.1991,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18790635466575623,
+      "step": 680,
+      "valid_targets_mean": 577.7,
+      "valid_targets_min": 364
+    },
+    {
+      "epoch": 1.683046683046683,
+      "grad_norm": 1.0364511769228686,
+      "learning_rate": 3.618706125248337e-05,
+      "loss": 0.2105,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20568892359733582,
+      "step": 685,
+      "valid_targets_mean": 610.6,
+      "valid_targets_min": 395
+    },
+    {
+      "epoch": 1.6953316953316953,
+      "grad_norm": 1.0653626980123834,
+      "learning_rate": 3.6102664270721275e-05,
+      "loss": 0.2097,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2012636959552765,
+      "step": 690,
+      "valid_targets_mean": 584.4,
+      "valid_targets_min": 380
+    },
+    {
+      "epoch": 1.7076167076167077,
+      "grad_norm": 1.0782173350105624,
+      "learning_rate": 3.601744414550589e-05,
+      "loss": 0.2029,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21700087189674377,
+      "step": 695,
+      "valid_targets_mean": 545.4,
+      "valid_targets_min": 363
+    },
+    {
+      "epoch": 1.71990171990172,
+      "grad_norm": 1.0898960424875124,
+      "learning_rate": 3.593140523315906e-05,
+      "loss": 0.2099,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22257234156131744,
+      "step": 700,
+      "valid_targets_mean": 520.8,
+      "valid_targets_min": 347
+    },
+    {
+      "epoch": 1.7321867321867321,
+      "grad_norm": 1.0486466431772246,
+      "learning_rate": 3.584455193185778e-05,
+      "loss": 0.202,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20295602083206177,
+      "step": 705,
+      "valid_targets_mean": 544.9,
+      "valid_targets_min": 340
+    },
+    {
+      "epoch": 1.7444717444717446,
+      "grad_norm": 1.084800118329105,
+      "learning_rate": 3.575688868140933e-05,
+      "loss": 0.2032,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19122770428657532,
+      "step": 710,
+      "valid_targets_mean": 488.4,
+      "valid_targets_min": 318
+    },
+    {
+      "epoch": 1.7567567567567568,
+      "grad_norm": 1.084292899345759,
+      "learning_rate": 3.566841996302438e-05,
+      "loss": 0.2051,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19632896780967712,
+      "step": 715,
+      "valid_targets_mean": 510.2,
+      "valid_targets_min": 335
+    },
+    {
+      "epoch": 1.769041769041769,
+      "grad_norm": 1.0557684475273712,
+      "learning_rate": 3.557915029908787e-05,
+      "loss": 0.2064,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2040451020002365,
+      "step": 720,
+      "valid_targets_mean": 558.2,
+      "valid_targets_min": 312
+    },
+    {
+      "epoch": 1.7813267813267815,
+      "grad_norm": 0.9261448145365117,
+      "learning_rate": 3.548908425292784e-05,
+      "loss": 0.206,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20299643278121948,
+      "step": 725,
+      "valid_targets_mean": 621.5,
+      "valid_targets_min": 315
+    },
+    {
+      "epoch": 1.7936117936117935,
+      "grad_norm": 1.059967470462753,
+      "learning_rate": 3.5398226428582165e-05,
+      "loss": 0.2016,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1749897301197052,
+      "step": 730,
+      "valid_targets_mean": 587.3,
+      "valid_targets_min": 327
+    },
+    {
+      "epoch": 1.805896805896806,
+      "grad_norm": 1.0927225436828998,
+      "learning_rate": 3.530658147056321e-05,
+      "loss": 0.2119,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21093472838401794,
+      "step": 735,
+      "valid_targets_mean": 536.0,
+      "valid_targets_min": 367
+    },
+    {
+      "epoch": 1.8181818181818183,
+      "grad_norm": 1.0767012332996604,
+      "learning_rate": 3.521415406362041e-05,
+      "loss": 0.208,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2154802829027176,
+      "step": 740,
+      "valid_targets_mean": 587.4,
+      "valid_targets_min": 358
+    },
+    {
+      "epoch": 1.8304668304668303,
+      "grad_norm": 1.0258650874116662,
+      "learning_rate": 3.512094893250076e-05,
+      "loss": 0.2038,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18948307633399963,
+      "step": 745,
+      "valid_targets_mean": 556.7,
+      "valid_targets_min": 325
+    },
+    {
+      "epoch": 1.8427518427518428,
+      "grad_norm": 1.0917952930239936,
+      "learning_rate": 3.5026970841707366e-05,
+      "loss": 0.1969,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20295953750610352,
+      "step": 750,
+      "valid_targets_mean": 637.2,
+      "valid_targets_min": 302
+    },
+    {
+      "epoch": 1.855036855036855,
+      "grad_norm": 1.0259051559636694,
+      "learning_rate": 3.493222459525579e-05,
+      "loss": 0.1963,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18988922238349915,
+      "step": 755,
+      "valid_targets_mean": 574.6,
+      "valid_targets_min": 389
+    },
+    {
+      "epoch": 1.8673218673218672,
+      "grad_norm": 1.107801334038011,
+      "learning_rate": 3.483671503642858e-05,
+      "loss": 0.2029,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20162153244018555,
+      "step": 760,
+      "valid_targets_mean": 559.1,
+      "valid_targets_min": 341
+    },
+    {
+      "epoch": 1.8796068796068797,
+      "grad_norm": 0.9513895961250298,
+      "learning_rate": 3.474044704752761e-05,
+      "loss": 0.2019,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20117336511611938,
+      "step": 765,
+      "valid_targets_mean": 614.8,
+      "valid_targets_min": 344
+    },
+    {
+      "epoch": 1.8918918918918919,
+      "grad_norm": 1.087713893272352,
+      "learning_rate": 3.464342554962454e-05,
+      "loss": 0.2028,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.21448519825935364,
+      "step": 770,
+      "valid_targets_mean": 522.4,
+      "valid_targets_min": 330
+    },
+    {
+      "epoch": 1.904176904176904,
+      "grad_norm": 0.9467693162634699,
+      "learning_rate": 3.4545655502309254e-05,
+      "loss": 0.2004,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2018240988254547,
+      "step": 775,
+      "valid_targets_mean": 646.3,
+      "valid_targets_min": 320
+    },
+    {
+      "epoch": 1.9164619164619165,
+      "grad_norm": 0.994395677212433,
+      "learning_rate": 3.444714190343633e-05,
+      "loss": 0.1997,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1869840919971466,
+      "step": 780,
+      "valid_targets_mean": 559.4,
+      "valid_targets_min": 315
+    },
+    {
+      "epoch": 1.9287469287469288,
+      "grad_norm": 1.1350837336602309,
+      "learning_rate": 3.434788978886957e-05,
+      "loss": 0.2049,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19758516550064087,
+      "step": 785,
+      "valid_targets_mean": 514.1,
+      "valid_targets_min": 360
+    },
+    {
+      "epoch": 1.941031941031941,
+      "grad_norm": 1.0742750813094744,
+      "learning_rate": 3.424790423222455e-05,
+      "loss": 0.1911,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19571363925933838,
+      "step": 790,
+      "valid_targets_mean": 575.7,
+      "valid_targets_min": 333
+    },
+    {
+      "epoch": 1.9533169533169534,
+      "grad_norm": 1.1044557075952837,
+      "learning_rate": 3.414719034460928e-05,
+      "loss": 0.1911,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.22668087482452393,
+      "step": 795,
+      "valid_targets_mean": 564.0,
+      "valid_targets_min": 358
+    },
+    {
+      "epoch": 1.9656019656019657,
+      "grad_norm": 1.0816888460619751,
+      "learning_rate": 3.404575327436294e-05,
+      "loss": 0.1979,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.2019120752811432,
+      "step": 800,
+      "valid_targets_mean": 579.1,
+      "valid_targets_min": 368
+    },
+    {
+      "epoch": 1.9778869778869779,
+      "grad_norm": 0.9653532634042872,
+      "learning_rate": 3.3943598206792665e-05,
+      "loss": 0.1951,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1838877946138382,
+      "step": 805,
+      "valid_targets_mean": 610.0,
+      "valid_targets_min": 383
+    },
+    {
+      "epoch": 1.9901719901719903,
+      "grad_norm": 1.1415332941821148,
+      "learning_rate": 3.384073036390857e-05,
+      "loss": 0.1965,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20956549048423767,
+      "step": 810,
+      "valid_targets_mean": 501.6,
+      "valid_targets_min": 304
+    },
+    {
+      "epoch": 2.0024570024570023,
+      "grad_norm": 0.9598781701745787,
+      "learning_rate": 3.373715500415667e-05,
+      "loss": 0.1931,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1695285439491272,
+      "step": 815,
+      "valid_targets_mean": 518.5,
+      "valid_targets_min": 333
+    },
+    {
+      "epoch": 2.0147420147420148,
+      "grad_norm": 1.0104897625706792,
+      "learning_rate": 3.363287742215023e-05,
+      "loss": 0.1674,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1609475016593933,
+      "step": 820,
+      "valid_targets_mean": 606.6,
+      "valid_targets_min": 355
+    },
+    {
+      "epoch": 2.027027027027027,
+      "grad_norm": 1.1012228379960987,
+      "learning_rate": 3.352790294839898e-05,
+      "loss": 0.1721,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16471633315086365,
+      "step": 825,
+      "valid_targets_mean": 617.8,
+      "valid_targets_min": 360
+    },
+    {
+      "epoch": 2.039312039312039,
+      "grad_norm": 1.0191591931441848,
+      "learning_rate": 3.3422236949036726e-05,
+      "loss": 0.1628,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1552569568157196,
+      "step": 830,
+      "valid_targets_mean": 604.6,
+      "valid_targets_min": 405
+    },
+    {
+      "epoch": 2.0515970515970516,
+      "grad_norm": 0.9959292379606938,
+      "learning_rate": 3.331588482554697e-05,
+      "loss": 0.1633,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.165554940700531,
+      "step": 835,
+      "valid_targets_mean": 608.4,
+      "valid_targets_min": 381
+    },
+    {
+      "epoch": 2.063882063882064,
+      "grad_norm": 0.9912133098541214,
+      "learning_rate": 3.320885201448684e-05,
+      "loss": 0.1711,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1614345908164978,
+      "step": 840,
+      "valid_targets_mean": 567.6,
+      "valid_targets_min": 317
+    },
+    {
+      "epoch": 2.076167076167076,
+      "grad_norm": 1.310631256990444,
+      "learning_rate": 3.310114398720917e-05,
+      "loss": 0.1621,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17154039442539215,
+      "step": 845,
+      "valid_targets_mean": 599.4,
+      "valid_targets_min": 411
+    },
+    {
+      "epoch": 2.0884520884520885,
+      "grad_norm": 1.1094181190105217,
+      "learning_rate": 3.299276624958281e-05,
+      "loss": 0.1673,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1730373501777649,
+      "step": 850,
+      "valid_targets_mean": 633.8,
+      "valid_targets_min": 379
+    },
+    {
+      "epoch": 2.100737100737101,
+      "grad_norm": 1.0089696788731093,
+      "learning_rate": 3.288372434171116e-05,
+      "loss": 0.1621,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1542433649301529,
+      "step": 855,
+      "valid_targets_mean": 556.9,
+      "valid_targets_min": 350
+    },
+    {
+      "epoch": 2.113022113022113,
+      "grad_norm": 1.1668954830382665,
+      "learning_rate": 3.2774023837648986e-05,
+      "loss": 0.1638,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18032445013523102,
+      "step": 860,
+      "valid_targets_mean": 595.8,
+      "valid_targets_min": 372
+    },
+    {
+      "epoch": 2.1253071253071254,
+      "grad_norm": 0.9857434649722174,
+      "learning_rate": 3.26636703451175e-05,
+      "loss": 0.1642,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.14477500319480896,
+      "step": 865,
+      "valid_targets_mean": 607.5,
+      "valid_targets_min": 356
+    },
+    {
+      "epoch": 2.1375921375921374,
+      "grad_norm": 0.968780124957847,
+      "learning_rate": 3.2552669505217646e-05,
+      "loss": 0.1623,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15161213278770447,
+      "step": 870,
+      "valid_targets_mean": 597.6,
+      "valid_targets_min": 360
+    },
+    {
+      "epoch": 2.14987714987715,
+      "grad_norm": 0.9765410163261161,
+      "learning_rate": 3.24410269921418e-05,
+      "loss": 0.1623,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.14507704973220825,
+      "step": 875,
+      "valid_targets_mean": 593.9,
+      "valid_targets_min": 417
+    },
+    {
+      "epoch": 2.1621621621621623,
+      "grad_norm": 1.0775768198373612,
+      "learning_rate": 3.232874851288367e-05,
+      "loss": 0.1691,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1555686593055725,
+      "step": 880,
+      "valid_targets_mean": 521.9,
+      "valid_targets_min": 341
+    },
+    {
+      "epoch": 2.1744471744471743,
+      "grad_norm": 0.9514600306163379,
+      "learning_rate": 3.221583980694659e-05,
+      "loss": 0.1697,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1574643850326538,
+      "step": 885,
+      "valid_targets_mean": 628.7,
+      "valid_targets_min": 357
+    },
+    {
+      "epoch": 2.1867321867321867,
+      "grad_norm": 1.2201470319941308,
+      "learning_rate": 3.21023066460501e-05,
+      "loss": 0.1635,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16941586136817932,
+      "step": 890,
+      "valid_targets_mean": 501.3,
+      "valid_targets_min": 357
+    },
+    {
+      "epoch": 2.199017199017199,
+      "grad_norm": 0.9973297712090284,
+      "learning_rate": 3.198815483383492e-05,
+      "loss": 0.1661,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1548662930727005,
+      "step": 895,
+      "valid_targets_mean": 588.8,
+      "valid_targets_min": 447
+    },
+    {
+      "epoch": 2.211302211302211,
+      "grad_norm": 1.109909348492073,
+      "learning_rate": 3.1873390205566295e-05,
+      "loss": 0.1625,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18065625429153442,
+      "step": 900,
+      "valid_targets_mean": 611.1,
+      "valid_targets_min": 381
+    },
+    {
+      "epoch": 2.2235872235872236,
+      "grad_norm": 1.150607769729905,
+      "learning_rate": 3.175801862783565e-05,
+      "loss": 0.1676,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1760374903678894,
+      "step": 905,
+      "valid_targets_mean": 579.7,
+      "valid_targets_min": 330
+    },
+    {
+      "epoch": 2.235872235872236,
+      "grad_norm": 1.0134504400046016,
+      "learning_rate": 3.164204599826077e-05,
+      "loss": 0.1679,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15869474411010742,
+      "step": 910,
+      "valid_targets_mean": 628.8,
+      "valid_targets_min": 387
+    },
+    {
+      "epoch": 2.248157248157248,
+      "grad_norm": 1.03636879858501,
+      "learning_rate": 3.1525478245184245e-05,
+      "loss": 0.161,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17024025321006775,
+      "step": 915,
+      "valid_targets_mean": 638.8,
+      "valid_targets_min": 338
+    },
+    {
+      "epoch": 2.2604422604422605,
+      "grad_norm": 1.027260075646475,
+      "learning_rate": 3.140832132737051e-05,
+      "loss": 0.1598,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16474905610084534,
+      "step": 920,
+      "valid_targets_mean": 634.4,
+      "valid_targets_min": 351
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 1.1516883647794431,
+      "learning_rate": 3.129058123370116e-05,
+      "loss": 0.1655,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16940803825855255,
+      "step": 925,
+      "valid_targets_mean": 556.1,
+      "valid_targets_min": 357
+    },
+    {
+      "epoch": 2.285012285012285,
+      "grad_norm": 0.9664843540102078,
+      "learning_rate": 3.117226398286887e-05,
+      "loss": 0.1602,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1558990180492401,
+      "step": 930,
+      "valid_targets_mean": 644.6,
+      "valid_targets_min": 341
+    },
+    {
+      "epoch": 2.2972972972972974,
+      "grad_norm": 1.2097643368196167,
+      "learning_rate": 3.105337562306968e-05,
+      "loss": 0.1674,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16575542092323303,
+      "step": 935,
+      "valid_targets_mean": 512.9,
+      "valid_targets_min": 375
+    },
+    {
+      "epoch": 2.30958230958231,
+      "grad_norm": 1.0251584186027345,
+      "learning_rate": 3.0933922231693854e-05,
+      "loss": 0.1607,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16751596331596375,
+      "step": 940,
+      "valid_targets_mean": 659.8,
+      "valid_targets_min": 330
+    },
+    {
+      "epoch": 2.321867321867322,
+      "grad_norm": 1.8223152408400631,
+      "learning_rate": 3.08139099150152e-05,
+      "loss": 0.1648,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17123517394065857,
+      "step": 945,
+      "valid_targets_mean": 623.1,
+      "valid_targets_min": 346
+    },
+    {
+      "epoch": 2.3341523341523343,
+      "grad_norm": 1.0197979025893136,
+      "learning_rate": 3.069334480787893e-05,
+      "loss": 0.1638,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1619846671819687,
+      "step": 950,
+      "valid_targets_mean": 516.6,
+      "valid_targets_min": 360
+    },
+    {
+      "epoch": 2.3464373464373462,
+      "grad_norm": 1.0185481805967684,
+      "learning_rate": 3.057223307338806e-05,
+      "loss": 0.1587,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16147619485855103,
+      "step": 955,
+      "valid_targets_mean": 610.8,
+      "valid_targets_min": 321
+    },
+    {
+      "epoch": 2.3587223587223587,
+      "grad_norm": 1.1629489871911958,
+      "learning_rate": 3.0450580902588346e-05,
+      "loss": 0.1675,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18816746771335602,
+      "step": 960,
+      "valid_targets_mean": 530.4,
+      "valid_targets_min": 370
+    },
+    {
+      "epoch": 2.371007371007371,
+      "grad_norm": 0.9701315397422264,
+      "learning_rate": 3.032839451415182e-05,
+      "loss": 0.1623,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15979017317295074,
+      "step": 965,
+      "valid_targets_mean": 634.5,
+      "valid_targets_min": 366
+    },
+    {
+      "epoch": 2.383292383292383,
+      "grad_norm": 0.99312578094489,
+      "learning_rate": 3.0205680154058904e-05,
+      "loss": 0.1573,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15236841142177582,
+      "step": 970,
+      "valid_targets_mean": 553.4,
+      "valid_targets_min": 362
+    },
+    {
+      "epoch": 2.3955773955773956,
+      "grad_norm": 1.0612042490649,
+      "learning_rate": 3.0082444095279117e-05,
+      "loss": 0.1681,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17524561285972595,
+      "step": 975,
+      "valid_targets_mean": 595.1,
+      "valid_targets_min": 377
+    },
+    {
+      "epoch": 2.407862407862408,
+      "grad_norm": 1.038886957377111,
+      "learning_rate": 2.9958692637450406e-05,
+      "loss": 0.1636,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.147619366645813,
+      "step": 980,
+      "valid_targets_mean": 583.9,
+      "valid_targets_min": 369
+    },
+    {
+      "epoch": 2.42014742014742,
+      "grad_norm": 1.037085869783086,
+      "learning_rate": 2.983443210655714e-05,
+      "loss": 0.1584,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15583443641662598,
+      "step": 985,
+      "valid_targets_mean": 515.8,
+      "valid_targets_min": 328
+    },
+    {
+      "epoch": 2.4324324324324325,
+      "grad_norm": 1.0353541133091022,
+      "learning_rate": 2.9709668854606706e-05,
+      "loss": 0.1705,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15763302147388458,
+      "step": 990,
+      "valid_targets_mean": 585.6,
+      "valid_targets_min": 397
+    },
+    {
+      "epoch": 2.444717444717445,
+      "grad_norm": 1.022215789393668,
+      "learning_rate": 2.9584409259304828e-05,
+      "loss": 0.1747,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17135146260261536,
+      "step": 995,
+      "valid_targets_mean": 552.5,
+      "valid_targets_min": 318
+    },
+    {
+      "epoch": 2.457002457002457,
+      "grad_norm": 1.1712513338241275,
+      "learning_rate": 2.945865972372954e-05,
+      "loss": 0.1699,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1690291315317154,
+      "step": 1000,
+      "valid_targets_mean": 564.1,
+      "valid_targets_min": 362
+    },
+    {
+      "epoch": 2.4692874692874693,
+      "grad_norm": 1.0317993574285669,
+      "learning_rate": 2.9332426676003858e-05,
+      "loss": 0.1627,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.14596088230609894,
+      "step": 1005,
+      "valid_targets_mean": 628.4,
+      "valid_targets_min": 333
+    },
+    {
+      "epoch": 2.4815724815724813,
+      "grad_norm": 1.15263750619694,
+      "learning_rate": 2.920571656896722e-05,
+      "loss": 0.163,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1633550375699997,
+      "step": 1010,
+      "valid_targets_mean": 595.0,
+      "valid_targets_min": 338
+    },
+    {
+      "epoch": 2.493857493857494,
+      "grad_norm": 1.196927272455729,
+      "learning_rate": 2.907853587984558e-05,
+      "loss": 0.1665,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18414945900440216,
+      "step": 1015,
+      "valid_targets_mean": 529.7,
+      "valid_targets_min": 324
+    },
+    {
+      "epoch": 2.506142506142506,
+      "grad_norm": 1.008390669336626,
+      "learning_rate": 2.8950891109920333e-05,
+      "loss": 0.171,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17250725626945496,
+      "step": 1020,
+      "valid_targets_mean": 554.2,
+      "valid_targets_min": 341
+    },
+    {
+      "epoch": 2.5184275184275187,
+      "grad_norm": 1.1830974634404077,
+      "learning_rate": 2.882278878419597e-05,
+      "loss": 0.1704,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17889085412025452,
+      "step": 1025,
+      "valid_targets_mean": 532.0,
+      "valid_targets_min": 336
+    },
+    {
+      "epoch": 2.5307125307125307,
+      "grad_norm": 1.225171314277819,
+      "learning_rate": 2.8694235451066538e-05,
+      "loss": 0.1646,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1738959401845932,
+      "step": 1030,
+      "valid_targets_mean": 521.2,
+      "valid_targets_min": 343
+    },
+    {
+      "epoch": 2.542997542997543,
+      "grad_norm": 1.0699171978912032,
+      "learning_rate": 2.8565237681980876e-05,
+      "loss": 0.1597,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1724793016910553,
+      "step": 1035,
+      "valid_targets_mean": 602.1,
+      "valid_targets_min": 365
+    },
+    {
+      "epoch": 2.555282555282555,
+      "grad_norm": 0.9758834051911167,
+      "learning_rate": 2.843580207110672e-05,
+      "loss": 0.164,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1547844111919403,
+      "step": 1040,
+      "valid_targets_mean": 624.6,
+      "valid_targets_min": 348
+    },
+    {
+      "epoch": 2.5675675675675675,
+      "grad_norm": 1.0648767118155311,
+      "learning_rate": 2.830593523499361e-05,
+      "loss": 0.1712,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17022374272346497,
+      "step": 1045,
+      "valid_targets_mean": 528.1,
+      "valid_targets_min": 316
+    },
+    {
+      "epoch": 2.57985257985258,
+      "grad_norm": 1.1786460270152923,
+      "learning_rate": 2.8175643812234627e-05,
+      "loss": 0.1657,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17464269697666168,
+      "step": 1050,
+      "valid_targets_mean": 482.8,
+      "valid_targets_min": 326
+    },
+    {
+      "epoch": 2.592137592137592,
+      "grad_norm": 1.015941200408563,
+      "learning_rate": 2.8044934463127108e-05,
+      "loss": 0.1635,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15763753652572632,
+      "step": 1055,
+      "valid_targets_mean": 595.5,
+      "valid_targets_min": 415
+    },
+    {
+      "epoch": 2.6044226044226044,
+      "grad_norm": 1.5455311111736163,
+      "learning_rate": 2.7913813869332112e-05,
+      "loss": 0.1613,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16405543684959412,
+      "step": 1060,
+      "valid_targets_mean": 553.4,
+      "valid_targets_min": 337
+    },
+    {
+      "epoch": 2.616707616707617,
+      "grad_norm": 1.0206262986883972,
+      "learning_rate": 2.7782288733532915e-05,
+      "loss": 0.1653,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15845853090286255,
+      "step": 1065,
+      "valid_targets_mean": 617.5,
+      "valid_targets_min": 377
+    },
+    {
+      "epoch": 2.628992628992629,
+      "grad_norm": 1.0127994524563275,
+      "learning_rate": 2.7650365779092346e-05,
+      "loss": 0.1574,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15306022763252258,
+      "step": 1070,
+      "valid_targets_mean": 574.9,
+      "valid_targets_min": 343
+    },
+    {
+      "epoch": 2.6412776412776413,
+      "grad_norm": 1.0920854825298392,
+      "learning_rate": 2.751805174970912e-05,
+      "loss": 0.1656,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1654532253742218,
+      "step": 1075,
+      "valid_targets_mean": 563.6,
+      "valid_targets_min": 372
+    },
+    {
+      "epoch": 2.6535626535626538,
+      "grad_norm": 1.3337934562654803,
+      "learning_rate": 2.7385353409073093e-05,
+      "loss": 0.1684,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18576306104660034,
+      "step": 1080,
+      "valid_targets_mean": 563.4,
+      "valid_targets_min": 340
+    },
+    {
+      "epoch": 2.6658476658476657,
+      "grad_norm": 1.1468407631418238,
+      "learning_rate": 2.725227754051953e-05,
+      "loss": 0.1625,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17222510278224945,
+      "step": 1085,
+      "valid_targets_mean": 532.2,
+      "valid_targets_min": 330
+    },
+    {
+      "epoch": 2.678132678132678,
+      "grad_norm": 1.0145689177041828,
+      "learning_rate": 2.711883094668234e-05,
+      "loss": 0.1576,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16334547102451324,
+      "step": 1090,
+      "valid_targets_mean": 712.4,
+      "valid_targets_min": 330
+    },
+    {
+      "epoch": 2.69041769041769,
+      "grad_norm": 1.0436018416176482,
+      "learning_rate": 2.698502044914633e-05,
+      "loss": 0.1675,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18585729598999023,
+      "step": 1095,
+      "valid_targets_mean": 637.7,
+      "valid_targets_min": 358
+    },
+    {
+      "epoch": 2.7027027027027026,
+      "grad_norm": 1.065555873590499,
+      "learning_rate": 2.685085288809853e-05,
+      "loss": 0.1645,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17872926592826843,
+      "step": 1100,
+      "valid_targets_mean": 615.6,
+      "valid_targets_min": 339
+    },
+    {
+      "epoch": 2.714987714987715,
+      "grad_norm": 1.105393129040058,
+      "learning_rate": 2.671633512197848e-05,
+      "loss": 0.165,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1631755530834198,
+      "step": 1105,
+      "valid_targets_mean": 546.8,
+      "valid_targets_min": 318
+    },
+    {
+      "epoch": 2.7272727272727275,
+      "grad_norm": 1.1349764871912573,
+      "learning_rate": 2.658147402712768e-05,
+      "loss": 0.1717,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1780092418193817,
+      "step": 1110,
+      "valid_targets_mean": 549.3,
+      "valid_targets_min": 308
+    },
+    {
+      "epoch": 2.7395577395577395,
+      "grad_norm": 1.0248807207449195,
+      "learning_rate": 2.6446276497438064e-05,
+      "loss": 0.1595,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1788291037082672,
+      "step": 1115,
+      "valid_targets_mean": 619.8,
+      "valid_targets_min": 315
+    },
+    {
+      "epoch": 2.751842751842752,
+      "grad_norm": 1.0236466528288748,
+      "learning_rate": 2.6310749443999593e-05,
+      "loss": 0.167,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.18527953326702118,
+      "step": 1120,
+      "valid_targets_mean": 704.7,
+      "valid_targets_min": 386
+    },
+    {
+      "epoch": 2.764127764127764,
+      "grad_norm": 1.0161666743813242,
+      "learning_rate": 2.617489979474699e-05,
+      "loss": 0.159,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16755788028240204,
+      "step": 1125,
+      "valid_targets_mean": 635.2,
+      "valid_targets_min": 327
+    },
+    {
+      "epoch": 2.7764127764127764,
+      "grad_norm": 1.1229342619607572,
+      "learning_rate": 2.6038734494105562e-05,
+      "loss": 0.1683,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17469602823257446,
+      "step": 1130,
+      "valid_targets_mean": 522.4,
+      "valid_targets_min": 306
+    },
+    {
+      "epoch": 2.788697788697789,
+      "grad_norm": 1.1045159313327992,
+      "learning_rate": 2.590226050263625e-05,
+      "loss": 0.1604,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1623562127351761,
+      "step": 1135,
+      "valid_targets_mean": 559.4,
+      "valid_targets_min": 340
+    },
+    {
+      "epoch": 2.800982800982801,
+      "grad_norm": 1.0512684811170392,
+      "learning_rate": 2.5765484796679768e-05,
+      "loss": 0.1664,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17947736382484436,
+      "step": 1140,
+      "valid_targets_mean": 643.2,
+      "valid_targets_min": 343
+    },
+    {
+      "epoch": 2.8132678132678133,
+      "grad_norm": 1.0378515532484835,
+      "learning_rate": 2.5628414368000035e-05,
+      "loss": 0.1651,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15963560342788696,
+      "step": 1145,
+      "valid_targets_mean": 604.6,
+      "valid_targets_min": 381
+    },
+    {
+      "epoch": 2.8255528255528253,
+      "grad_norm": 1.1752740327029623,
+      "learning_rate": 2.5491056223426746e-05,
+      "loss": 0.1685,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17332467436790466,
+      "step": 1150,
+      "valid_targets_mean": 508.6,
+      "valid_targets_min": 329
+    },
+    {
+      "epoch": 2.8378378378378377,
+      "grad_norm": 1.0171284826487994,
+      "learning_rate": 2.5353417384497166e-05,
+      "loss": 0.1632,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1706382930278778,
+      "step": 1155,
+      "valid_targets_mean": 620.8,
+      "valid_targets_min": 406
+    },
+    {
+      "epoch": 2.85012285012285,
+      "grad_norm": 1.1287447526542342,
+      "learning_rate": 2.5215504887097243e-05,
+      "loss": 0.1673,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.19561892747879028,
+      "step": 1160,
+      "valid_targets_mean": 601.8,
+      "valid_targets_min": 340
+    },
+    {
+      "epoch": 2.8624078624078626,
+      "grad_norm": 0.9610766649005913,
+      "learning_rate": 2.5077325781101918e-05,
+      "loss": 0.1594,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.14945566654205322,
+      "step": 1165,
+      "valid_targets_mean": 570.4,
+      "valid_targets_min": 342
+    },
+    {
+      "epoch": 2.8746928746928746,
+      "grad_norm": 1.0227531905951595,
+      "learning_rate": 2.493888713001476e-05,
+      "loss": 0.1673,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16126927733421326,
+      "step": 1170,
+      "valid_targets_mean": 642.7,
+      "valid_targets_min": 359
+    },
+    {
+      "epoch": 2.886977886977887,
+      "grad_norm": 1.0837878704674906,
+      "learning_rate": 2.480019601060687e-05,
+      "loss": 0.1593,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17691218852996826,
+      "step": 1175,
+      "valid_targets_mean": 650.2,
+      "valid_targets_min": 333
+    },
+    {
+      "epoch": 2.899262899262899,
+      "grad_norm": 1.1939077938787526,
+      "learning_rate": 2.4661259512555176e-05,
+      "loss": 0.1712,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.17172303795814514,
+      "step": 1180,
+      "valid_targets_mean": 516.2,
+      "valid_targets_min": 359
+    },
+    {
+      "epoch": 2.9115479115479115,
+      "grad_norm": 1.0876622788145698,
+      "learning_rate": 2.4522084738079933e-05,
+      "loss": 0.1724,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.20079737901687622,
+      "step": 1185,
+      "valid_targets_mean": 650.1,
+      "valid_targets_min": 347
+    },
+    {
+      "epoch": 2.923832923832924,
+      "grad_norm": 1.139247105803865,
+      "learning_rate": 2.4382678801581762e-05,
+      "loss": 0.1625,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15431401133537292,
+      "step": 1190,
+      "valid_targets_mean": 504.2,
+      "valid_targets_min": 345
+    },
+    {
+      "epoch": 2.9361179361179364,
+      "grad_norm": 1.1904829689631444,
+      "learning_rate": 2.4243048829277916e-05,
+      "loss": 0.1646,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.15002189576625824,
+      "step": 1195,
+      "valid_targets_mean": 496.9,
+      "valid_targets_min": 359
+    },
+    {
+      "epoch": 2.9484029484029484,
+      "grad_norm": 1.0235188956010766,
+      "learning_rate": 2.410320195883802e-05,
+      "loss": 0.1628,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16833597421646118,
+      "step": 1200,
+      "valid_targets_mean": 663.0,
+      "valid_targets_min": 341
+    },
+    {
+      "epoch": 2.960687960687961,
+      "grad_norm": 1.0386284807339494,
+      "learning_rate": 2.396314533901918e-05,
+      "loss": 0.1679,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1625211238861084,
+      "step": 1205,
+      "valid_targets_mean": 541.9,
+      "valid_targets_min": 357
+    },
+    {
+      "epoch": 2.972972972972973,
+      "grad_norm": 1.0553260194827612,
+      "learning_rate": 2.3822886129300603e-05,
+      "loss": 0.1636,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16509494185447693,
+      "step": 1210,
+      "valid_targets_mean": 593.0,
+      "valid_targets_min": 350
+    },
+    {
+      "epoch": 2.9852579852579852,
+      "grad_norm": 1.1291087525354948,
+      "learning_rate": 2.368243149951755e-05,
+      "loss": 0.1639,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.16854259371757507,
+      "step": 1215,
+      "valid_targets_mean": 540.3,
+      "valid_targets_min": 336
+    },
+    {
+      "epoch": 2.9975429975429977,
+      "grad_norm": 1.0280421294913595,
+      "learning_rate": 2.3541788629494865e-05,
+      "loss": 0.1668,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.154227614402771,
+      "step": 1220,
+      "valid_targets_mean": 537.2,
+      "valid_targets_min": 353
+    },
+    {
+      "epoch": 3.0098280098280097,
+      "grad_norm": 0.9438151898624106,
+      "learning_rate": 2.3400964708679944e-05,
+      "loss": 0.1331,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11106042563915253,
+      "step": 1225,
+      "valid_targets_mean": 543.3,
+      "valid_targets_min": 348
+    },
+    {
+      "epoch": 3.022113022113022,
+      "grad_norm": 1.3092340644616152,
+      "learning_rate": 2.325996693577522e-05,
+      "loss": 0.1314,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13556016981601715,
+      "step": 1230,
+      "valid_targets_mean": 569.0,
+      "valid_targets_min": 344
+    },
+    {
+      "epoch": 3.0343980343980346,
+      "grad_norm": 1.1917250259707166,
+      "learning_rate": 2.311880251837019e-05,
+      "loss": 0.1298,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12731008231639862,
+      "step": 1235,
+      "valid_targets_mean": 621.3,
+      "valid_targets_min": 363
+    },
+    {
+      "epoch": 3.0466830466830466,
+      "grad_norm": 1.171746419744826,
+      "learning_rate": 2.2977478672572933e-05,
+      "loss": 0.1348,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.14058241248130798,
+      "step": 1240,
+      "valid_targets_mean": 583.0,
+      "valid_targets_min": 330
+    },
+    {
+      "epoch": 3.058968058968059,
+      "grad_norm": 0.9850462514719055,
+      "learning_rate": 2.2836002622641297e-05,
+      "loss": 0.1246,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12502151727676392,
+      "step": 1245,
+      "valid_targets_mean": 655.2,
+      "valid_targets_min": 426
+    },
+    {
+      "epoch": 3.0712530712530715,
+      "grad_norm": 1.1523919057407686,
+      "learning_rate": 2.269438160061354e-05,
+      "loss": 0.1286,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1243138313293457,
+      "step": 1250,
+      "valid_targets_mean": 538.6,
+      "valid_targets_min": 298
+    },
+    {
+      "epoch": 3.0835380835380835,
+      "grad_norm": 1.1105751349907576,
+      "learning_rate": 2.2552622845938698e-05,
+      "loss": 0.1226,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11567364633083344,
+      "step": 1255,
+      "valid_targets_mean": 508.1,
+      "valid_targets_min": 357
+    },
+    {
+      "epoch": 3.095823095823096,
+      "grad_norm": 1.1173952059466155,
+      "learning_rate": 2.2410733605106462e-05,
+      "loss": 0.1335,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.14052309095859528,
+      "step": 1260,
+      "valid_targets_mean": 600.1,
+      "valid_targets_min": 369
+    },
+    {
+      "epoch": 3.108108108108108,
+      "grad_norm": 1.133256035223599,
+      "learning_rate": 2.2268721131276805e-05,
+      "loss": 0.1306,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1453954577445984,
+      "step": 1265,
+      "valid_targets_mean": 594.8,
+      "valid_targets_min": 364
+    },
+    {
+      "epoch": 3.1203931203931203,
+      "grad_norm": 1.045210794343584,
+      "learning_rate": 2.2126592683909154e-05,
+      "loss": 0.1228,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12265928834676743,
+      "step": 1270,
+      "valid_targets_mean": 683.3,
+      "valid_targets_min": 386
+    },
+    {
+      "epoch": 3.1326781326781328,
+      "grad_norm": 1.2110743775006845,
+      "learning_rate": 2.1984355528391342e-05,
+      "loss": 0.1279,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13180884718894958,
+      "step": 1275,
+      "valid_targets_mean": 560.5,
+      "valid_targets_min": 329
+    },
+    {
+      "epoch": 3.1449631449631448,
+      "grad_norm": 1.1434541255684711,
+      "learning_rate": 2.1842016935668188e-05,
+      "loss": 0.1282,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12741702795028687,
+      "step": 1280,
+      "valid_targets_mean": 549.2,
+      "valid_targets_min": 379
+    },
+    {
+      "epoch": 3.157248157248157,
+      "grad_norm": 1.0439156376434802,
+      "learning_rate": 2.169958418186982e-05,
+      "loss": 0.1341,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12404695153236389,
+      "step": 1285,
+      "valid_targets_mean": 665.5,
+      "valid_targets_min": 318
+    },
+    {
+      "epoch": 3.1695331695331697,
+      "grad_norm": 1.0124516604323626,
+      "learning_rate": 2.1557064547939754e-05,
+      "loss": 0.1315,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11568133533000946,
+      "step": 1290,
+      "valid_targets_mean": 624.2,
+      "valid_targets_min": 350
+    },
+    {
+      "epoch": 3.1818181818181817,
+      "grad_norm": 1.0468568345799445,
+      "learning_rate": 2.1414465319262666e-05,
+      "loss": 0.1371,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12586909532546997,
+      "step": 1295,
+      "valid_targets_mean": 592.6,
+      "valid_targets_min": 377
+    },
+    {
+      "epoch": 3.194103194103194,
+      "grad_norm": 1.2355339893235782,
+      "learning_rate": 2.1271793785291997e-05,
+      "loss": 0.1272,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12561175227165222,
+      "step": 1300,
+      "valid_targets_mean": 503.3,
+      "valid_targets_min": 326
+    },
+    {
+      "epoch": 3.2063882063882065,
+      "grad_norm": 1.2213272421816683,
+      "learning_rate": 2.1129057239177337e-05,
+      "loss": 0.1316,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13759493827819824,
+      "step": 1305,
+      "valid_targets_mean": 592.8,
+      "valid_targets_min": 336
+    },
+    {
+      "epoch": 3.2186732186732185,
+      "grad_norm": 1.1733266254515748,
+      "learning_rate": 2.0986262977391577e-05,
+      "loss": 0.1319,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1188465803861618,
+      "step": 1310,
+      "valid_targets_mean": 577.1,
+      "valid_targets_min": 366
+    },
+    {
+      "epoch": 3.230958230958231,
+      "grad_norm": 1.0329034457428024,
+      "learning_rate": 2.084341829935796e-05,
+      "loss": 0.1312,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13047614693641663,
+      "step": 1315,
+      "valid_targets_mean": 648.7,
+      "valid_targets_min": 429
+    },
+    {
+      "epoch": 3.2432432432432434,
+      "grad_norm": 1.0872058545288361,
+      "learning_rate": 2.0700530507076916e-05,
+      "loss": 0.1219,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11639399826526642,
+      "step": 1320,
+      "valid_targets_mean": 631.0,
+      "valid_targets_min": 355
+    },
+    {
+      "epoch": 3.2555282555282554,
+      "grad_norm": 1.1820198984468047,
+      "learning_rate": 2.0557606904752833e-05,
+      "loss": 0.1278,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13082380592823029,
+      "step": 1325,
+      "valid_targets_mean": 574.2,
+      "valid_targets_min": 365
+    },
+    {
+      "epoch": 3.267813267813268,
+      "grad_norm": 1.1630414180551119,
+      "learning_rate": 2.0414654798420622e-05,
+      "loss": 0.1221,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13212239742279053,
+      "step": 1330,
+      "valid_targets_mean": 576.9,
+      "valid_targets_min": 300
+    },
+    {
+      "epoch": 3.2800982800982803,
+      "grad_norm": 1.0578455352391878,
+      "learning_rate": 2.02716814955723e-05,
+      "loss": 0.1269,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11416147649288177,
+      "step": 1335,
+      "valid_targets_mean": 588.4,
+      "valid_targets_min": 375
+    },
+    {
+      "epoch": 3.2923832923832923,
+      "grad_norm": 1.1805958220382198,
+      "learning_rate": 2.0128694304783406e-05,
+      "loss": 0.1305,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1387619972229004,
+      "step": 1340,
+      "valid_targets_mean": 635.7,
+      "valid_targets_min": 351
+    },
+    {
+      "epoch": 3.3046683046683047,
+      "grad_norm": 1.0482090682153697,
+      "learning_rate": 1.9985700535339406e-05,
+      "loss": 0.1296,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11787254363298416,
+      "step": 1345,
+      "valid_targets_mean": 583.8,
+      "valid_targets_min": 351
+    },
+    {
+      "epoch": 3.3169533169533167,
+      "grad_norm": 1.1169879054755618,
+      "learning_rate": 1.984270749686207e-05,
+      "loss": 0.1332,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12067554891109467,
+      "step": 1350,
+      "valid_targets_mean": 516.5,
+      "valid_targets_min": 333
+    },
+    {
+      "epoch": 3.329238329238329,
+      "grad_norm": 1.1551139226743778,
+      "learning_rate": 1.9699722498935786e-05,
+      "loss": 0.1347,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13348305225372314,
+      "step": 1355,
+      "valid_targets_mean": 618.8,
+      "valid_targets_min": 333
+    },
+    {
+      "epoch": 3.3415233415233416,
+      "grad_norm": 1.0249399702817776,
+      "learning_rate": 1.9556752850733933e-05,
+      "loss": 0.1271,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11070968210697174,
+      "step": 1360,
+      "valid_targets_mean": 561.4,
+      "valid_targets_min": 347
+    },
+    {
+      "epoch": 3.3538083538083536,
+      "grad_norm": 1.1499498208012953,
+      "learning_rate": 1.9413805860645242e-05,
+      "loss": 0.1286,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12810730934143066,
+      "step": 1365,
+      "valid_targets_mean": 703.3,
+      "valid_targets_min": 311
+    },
+    {
+      "epoch": 3.366093366093366,
+      "grad_norm": 1.0876681376203914,
+      "learning_rate": 1.9270888835900165e-05,
+      "loss": 0.1296,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13525883853435516,
+      "step": 1370,
+      "valid_targets_mean": 606.2,
+      "valid_targets_min": 345
+    },
+    {
+      "epoch": 3.3783783783783785,
+      "grad_norm": 1.225696579696641,
+      "learning_rate": 1.9128009082197417e-05,
+      "loss": 0.1287,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13166263699531555,
+      "step": 1375,
+      "valid_targets_mean": 592.9,
+      "valid_targets_min": 378
+    },
+    {
+      "epoch": 3.3906633906633905,
+      "grad_norm": 1.066556068908003,
+      "learning_rate": 1.8985173903330428e-05,
+      "loss": 0.1292,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12290339171886444,
+      "step": 1380,
+      "valid_targets_mean": 688.7,
+      "valid_targets_min": 357
+    },
+    {
+      "epoch": 3.402948402948403,
+      "grad_norm": 1.0572944355622975,
+      "learning_rate": 1.884239060081407e-05,
+      "loss": 0.1264,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12135250866413116,
+      "step": 1385,
+      "valid_targets_mean": 587.2,
+      "valid_targets_min": 318
+    },
+    {
+      "epoch": 3.4152334152334154,
+      "grad_norm": 1.1838129653133687,
+      "learning_rate": 1.869966647351135e-05,
+      "loss": 0.1234,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12669309973716736,
+      "step": 1390,
+      "valid_targets_mean": 583.1,
+      "valid_targets_min": 351
+    },
+    {
+      "epoch": 3.4275184275184274,
+      "grad_norm": 1.0319091113711576,
+      "learning_rate": 1.8557008817260343e-05,
+      "loss": 0.1306,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12052074819803238,
+      "step": 1395,
+      "valid_targets_mean": 611.9,
+      "valid_targets_min": 335
+    },
+    {
+      "epoch": 3.43980343980344,
+      "grad_norm": 1.1399305429197617,
+      "learning_rate": 1.8414424924501222e-05,
+      "loss": 0.1301,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11919272691011429,
+      "step": 1400,
+      "valid_targets_mean": 465.2,
+      "valid_targets_min": 344
+    },
+    {
+      "epoch": 3.4520884520884523,
+      "grad_norm": 1.177328920151525,
+      "learning_rate": 1.827192208390347e-05,
+      "loss": 0.1266,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12182799726724625,
+      "step": 1405,
+      "valid_targets_mean": 594.9,
+      "valid_targets_min": 320
+    },
+    {
+      "epoch": 3.4643734643734643,
+      "grad_norm": 1.1035517940031225,
+      "learning_rate": 1.812950757999334e-05,
+      "loss": 0.1279,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12335000932216644,
+      "step": 1410,
+      "valid_targets_mean": 601.6,
+      "valid_targets_min": 375
+    },
+    {
+      "epoch": 3.4766584766584767,
+      "grad_norm": 1.1686625404125646,
+      "learning_rate": 1.7987188692781417e-05,
+      "loss": 0.1275,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12197911739349365,
+      "step": 1415,
+      "valid_targets_mean": 518.9,
+      "valid_targets_min": 372
+    },
+    {
+      "epoch": 3.488943488943489,
+      "grad_norm": 1.1409244395482991,
+      "learning_rate": 1.784497269739052e-05,
+      "loss": 0.1339,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12735916674137115,
+      "step": 1420,
+      "valid_targets_mean": 555.7,
+      "valid_targets_min": 326
+    },
+    {
+      "epoch": 3.501228501228501,
+      "grad_norm": 1.15898956512608,
+      "learning_rate": 1.770286686368381e-05,
+      "loss": 0.1283,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13501927256584167,
+      "step": 1425,
+      "valid_targets_mean": 619.3,
+      "valid_targets_min": 359
+    },
+    {
+      "epoch": 3.5135135135135136,
+      "grad_norm": 1.098433144324709,
+      "learning_rate": 1.756087845589312e-05,
+      "loss": 0.131,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12755000591278076,
+      "step": 1430,
+      "valid_targets_mean": 583.7,
+      "valid_targets_min": 348
+    },
+    {
+      "epoch": 3.5257985257985256,
+      "grad_norm": 1.2285008083017597,
+      "learning_rate": 1.7419014732247683e-05,
+      "loss": 0.1323,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1387604922056198,
+      "step": 1435,
+      "valid_targets_mean": 548.3,
+      "valid_targets_min": 321
+    },
+    {
+      "epoch": 3.538083538083538,
+      "grad_norm": 1.2016331632631438,
+      "learning_rate": 1.7277282944603047e-05,
+      "loss": 0.1373,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1299329698085785,
+      "step": 1440,
+      "valid_targets_mean": 572.7,
+      "valid_targets_min": 331
+    },
+    {
+      "epoch": 3.5503685503685505,
+      "grad_norm": 1.0798388503657812,
+      "learning_rate": 1.713569033807041e-05,
+      "loss": 0.1313,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12629565596580505,
+      "step": 1445,
+      "valid_targets_mean": 598.0,
+      "valid_targets_min": 312
+    },
+    {
+      "epoch": 3.562653562653563,
+      "grad_norm": 1.1335346583884485,
+      "learning_rate": 1.6994244150646244e-05,
+      "loss": 0.1237,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1171446442604065,
+      "step": 1450,
+      "valid_targets_mean": 486.2,
+      "valid_targets_min": 338
+    },
+    {
+      "epoch": 3.574938574938575,
+      "grad_norm": 1.026365044309373,
+      "learning_rate": 1.6852951612842278e-05,
+      "loss": 0.1238,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11981429159641266,
+      "step": 1455,
+      "valid_targets_mean": 735.9,
+      "valid_targets_min": 392
+    },
+    {
+      "epoch": 3.5872235872235874,
+      "grad_norm": 1.1509255430926,
+      "learning_rate": 1.671181994731595e-05,
+      "loss": 0.1299,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12611302733421326,
+      "step": 1460,
+      "valid_targets_mean": 626.5,
+      "valid_targets_min": 378
+    },
+    {
+      "epoch": 3.5995085995085994,
+      "grad_norm": 1.1055170772072278,
+      "learning_rate": 1.6570856368501108e-05,
+      "loss": 0.1339,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1377750188112259,
+      "step": 1465,
+      "valid_targets_mean": 683.5,
+      "valid_targets_min": 388
+    },
+    {
+      "epoch": 3.611793611793612,
+      "grad_norm": 1.1283525993380816,
+      "learning_rate": 1.643006808223931e-05,
+      "loss": 0.1275,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11461061239242554,
+      "step": 1470,
+      "valid_targets_mean": 523.2,
+      "valid_targets_min": 365
+    },
+    {
+      "epoch": 3.6240786240786242,
+      "grad_norm": 1.1522419980137233,
+      "learning_rate": 1.6289462285411387e-05,
+      "loss": 0.1257,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13068246841430664,
+      "step": 1475,
+      "valid_targets_mean": 668.4,
+      "valid_targets_min": 385
+    },
+    {
+      "epoch": 3.6363636363636362,
+      "grad_norm": 1.1631017873745493,
+      "learning_rate": 1.614904616556962e-05,
+      "loss": 0.1281,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12972646951675415,
+      "step": 1480,
+      "valid_targets_mean": 511.1,
+      "valid_targets_min": 319
+    },
+    {
+      "epoch": 3.6486486486486487,
+      "grad_norm": 1.0535715929600475,
+      "learning_rate": 1.6008826900570294e-05,
+      "loss": 0.126,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12002302706241608,
+      "step": 1485,
+      "valid_targets_mean": 619.9,
+      "valid_targets_min": 305
+    },
+    {
+      "epoch": 3.6609336609336607,
+      "grad_norm": 1.0292067655871422,
+      "learning_rate": 1.586881165820675e-05,
+      "loss": 0.1284,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12251895666122437,
+      "step": 1490,
+      "valid_targets_mean": 545.2,
+      "valid_targets_min": 336
+    },
+    {
+      "epoch": 3.673218673218673,
+      "grad_norm": 1.165710854153103,
+      "learning_rate": 1.5729007595843037e-05,
+      "loss": 0.1276,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1191115528345108,
+      "step": 1495,
+      "valid_targets_mean": 561.9,
+      "valid_targets_min": 370
+    },
+    {
+      "epoch": 3.6855036855036856,
+      "grad_norm": 1.1331503839407906,
+      "learning_rate": 1.5589421860047986e-05,
+      "loss": 0.1325,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13271397352218628,
+      "step": 1500,
+      "valid_targets_mean": 619.1,
+      "valid_targets_min": 343
+    },
+    {
+      "epoch": 3.697788697788698,
+      "grad_norm": 1.1461701235051638,
+      "learning_rate": 1.5450061586229903e-05,
+      "loss": 0.1309,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12879186868667603,
+      "step": 1505,
+      "valid_targets_mean": 600.9,
+      "valid_targets_min": 371
+    },
+    {
+      "epoch": 3.71007371007371,
+      "grad_norm": 1.1847404529689227,
+      "learning_rate": 1.5310933898271864e-05,
+      "loss": 0.1259,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1365867555141449,
+      "step": 1510,
+      "valid_targets_mean": 611.1,
+      "valid_targets_min": 340
+    },
+    {
+      "epoch": 3.7223587223587224,
+      "grad_norm": 1.092144235161736,
+      "learning_rate": 1.5172045908167462e-05,
+      "loss": 0.1321,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13623477518558502,
+      "step": 1515,
+      "valid_targets_mean": 634.0,
+      "valid_targets_min": 367
+    },
+    {
+      "epoch": 3.7346437346437344,
+      "grad_norm": 1.1632897050883428,
+      "learning_rate": 1.5033404715657344e-05,
+      "loss": 0.1275,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12677055597305298,
+      "step": 1520,
+      "valid_targets_mean": 525.4,
+      "valid_targets_min": 353
+    },
+    {
+      "epoch": 3.746928746928747,
+      "grad_norm": 1.1658235409021696,
+      "learning_rate": 1.4895017407866217e-05,
+      "loss": 0.1372,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.14368900656700134,
+      "step": 1525,
+      "valid_targets_mean": 561.8,
+      "valid_targets_min": 358
+    },
+    {
+      "epoch": 3.7592137592137593,
+      "grad_norm": 1.161265061332121,
+      "learning_rate": 1.4756891058940606e-05,
+      "loss": 0.1289,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12873947620391846,
+      "step": 1530,
+      "valid_targets_mean": 572.2,
+      "valid_targets_min": 303
+    },
+    {
+      "epoch": 3.7714987714987718,
+      "grad_norm": 1.194746454201963,
+      "learning_rate": 1.4619032729687223e-05,
+      "loss": 0.1302,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12980486452579498,
+      "step": 1535,
+      "valid_targets_mean": 531.4,
+      "valid_targets_min": 403
+    },
+    {
+      "epoch": 3.7837837837837838,
+      "grad_norm": 1.1787818795890501,
+      "learning_rate": 1.4481449467212004e-05,
+      "loss": 0.1325,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13438472151756287,
+      "step": 1540,
+      "valid_targets_mean": 502.7,
+      "valid_targets_min": 337
+    },
+    {
+      "epoch": 3.796068796068796,
+      "grad_norm": 1.1499950845573694,
+      "learning_rate": 1.4344148304559926e-05,
+      "loss": 0.1333,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1443779170513153,
+      "step": 1545,
+      "valid_targets_mean": 646.7,
+      "valid_targets_min": 396
+    },
+    {
+      "epoch": 3.808353808353808,
+      "grad_norm": 1.1616430554436303,
+      "learning_rate": 1.4207136260355426e-05,
+      "loss": 0.1265,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1254158318042755,
+      "step": 1550,
+      "valid_targets_mean": 638.1,
+      "valid_targets_min": 338
+    },
+    {
+      "epoch": 3.8206388206388207,
+      "grad_norm": 1.132424544684703,
+      "learning_rate": 1.4070420338443667e-05,
+      "loss": 0.1298,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13349393010139465,
+      "step": 1555,
+      "valid_targets_mean": 654.6,
+      "valid_targets_min": 327
+    },
+    {
+      "epoch": 3.832923832923833,
+      "grad_norm": 1.3001413881701709,
+      "learning_rate": 1.3934007527532494e-05,
+      "loss": 0.1296,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13770049810409546,
+      "step": 1560,
+      "valid_targets_mean": 481.9,
+      "valid_targets_min": 336
+    },
+    {
+      "epoch": 3.845208845208845,
+      "grad_norm": 1.1715428722278898,
+      "learning_rate": 1.3797904800835174e-05,
+      "loss": 0.1239,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12449394166469574,
+      "step": 1565,
+      "valid_targets_mean": 480.0,
+      "valid_targets_min": 356
+    },
+    {
+      "epoch": 3.8574938574938575,
+      "grad_norm": 1.0380113852952901,
+      "learning_rate": 1.3662119115713968e-05,
+      "loss": 0.1237,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11352840811014175,
+      "step": 1570,
+      "valid_targets_mean": 607.8,
+      "valid_targets_min": 358
+    },
+    {
+      "epoch": 3.8697788697788695,
+      "grad_norm": 1.0491947528953767,
+      "learning_rate": 1.3526657413324427e-05,
+      "loss": 0.1263,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10872411727905273,
+      "step": 1575,
+      "valid_targets_mean": 613.9,
+      "valid_targets_min": 346
+    },
+    {
+      "epoch": 3.882063882063882,
+      "grad_norm": 1.1342755535198472,
+      "learning_rate": 1.3391526618260636e-05,
+      "loss": 0.1319,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13144788146018982,
+      "step": 1580,
+      "valid_targets_mean": 571.2,
+      "valid_targets_min": 396
+    },
+    {
+      "epoch": 3.8943488943488944,
+      "grad_norm": 1.2509197238722758,
+      "learning_rate": 1.3256733638201172e-05,
+      "loss": 0.1329,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.14529390633106232,
+      "step": 1585,
+      "valid_targets_mean": 546.2,
+      "valid_targets_min": 328
+    },
+    {
+      "epoch": 3.906633906633907,
+      "grad_norm": 0.9973094279197666,
+      "learning_rate": 1.3122285363556053e-05,
+      "loss": 0.1251,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11257536709308624,
+      "step": 1590,
+      "valid_targets_mean": 564.9,
+      "valid_targets_min": 374
+    },
+    {
+      "epoch": 3.918918918918919,
+      "grad_norm": 1.2025598432808287,
+      "learning_rate": 1.2988188667114487e-05,
+      "loss": 0.1255,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13427409529685974,
+      "step": 1595,
+      "valid_targets_mean": 562.8,
+      "valid_targets_min": 315
+    },
+    {
+      "epoch": 3.9312039312039313,
+      "grad_norm": 1.1529440857186466,
+      "learning_rate": 1.2854450403693526e-05,
+      "loss": 0.1296,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1339278221130371,
+      "step": 1600,
+      "valid_targets_mean": 503.7,
+      "valid_targets_min": 339
+    },
+    {
+      "epoch": 3.9434889434889433,
+      "grad_norm": 1.0860172451290426,
+      "learning_rate": 1.272107740978769e-05,
+      "loss": 0.1278,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1276274025440216,
+      "step": 1605,
+      "valid_targets_mean": 653.6,
+      "valid_targets_min": 320
+    },
+    {
+      "epoch": 3.9557739557739557,
+      "grad_norm": 1.1427290885983061,
+      "learning_rate": 1.2588076503219475e-05,
+      "loss": 0.1282,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12042656540870667,
+      "step": 1610,
+      "valid_targets_mean": 583.0,
+      "valid_targets_min": 344
+    },
+    {
+      "epoch": 3.968058968058968,
+      "grad_norm": 1.1290073089466048,
+      "learning_rate": 1.2455454482790859e-05,
+      "loss": 0.1253,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.13513417541980743,
+      "step": 1615,
+      "valid_targets_mean": 533.4,
+      "valid_targets_min": 318
+    },
+    {
+      "epoch": 3.98034398034398,
+      "grad_norm": 1.1928783790551887,
+      "learning_rate": 1.2323218127935714e-05,
+      "loss": 0.1251,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1335742324590683,
+      "step": 1620,
+      "valid_targets_mean": 557.1,
+      "valid_targets_min": 368
+    },
+    {
+      "epoch": 3.9926289926289926,
+      "grad_norm": 1.150530284349401,
+      "learning_rate": 1.2191374198373309e-05,
+      "loss": 0.1271,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12559539079666138,
+      "step": 1625,
+      "valid_targets_mean": 588.1,
+      "valid_targets_min": 317
+    },
+    {
+      "epoch": 4.004914004914005,
+      "grad_norm": 1.0048824233082845,
+      "learning_rate": 1.2059929433762734e-05,
+      "loss": 0.12,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10455832630395889,
+      "step": 1630,
+      "valid_targets_mean": 675.5,
+      "valid_targets_min": 341
+    },
+    {
+      "epoch": 4.017199017199017,
+      "grad_norm": 1.163545632299262,
+      "learning_rate": 1.1928890553358352e-05,
+      "loss": 0.1024,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10858234018087387,
+      "step": 1635,
+      "valid_targets_mean": 502.5,
+      "valid_targets_min": 365
+    },
+    {
+      "epoch": 4.0294840294840295,
+      "grad_norm": 1.2412134097978402,
+      "learning_rate": 1.1798264255666387e-05,
+      "loss": 0.1027,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12209483981132507,
+      "step": 1640,
+      "valid_targets_mean": 621.5,
+      "valid_targets_min": 387
+    },
+    {
+      "epoch": 4.041769041769042,
+      "grad_norm": 1.3264719047529132,
+      "learning_rate": 1.1668057218102436e-05,
+      "loss": 0.1032,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10184046626091003,
+      "step": 1645,
+      "valid_targets_mean": 567.1,
+      "valid_targets_min": 327
+    },
+    {
+      "epoch": 4.054054054054054,
+      "grad_norm": 1.2340618039253701,
+      "learning_rate": 1.1538276096650175e-05,
+      "loss": 0.1013,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09597574919462204,
+      "step": 1650,
+      "valid_targets_mean": 625.8,
+      "valid_targets_min": 406
+    },
+    {
+      "epoch": 4.066339066339066,
+      "grad_norm": 1.1327237171212756,
+      "learning_rate": 1.1408927525521118e-05,
+      "loss": 0.0984,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09209711849689484,
+      "step": 1655,
+      "valid_targets_mean": 571.6,
+      "valid_targets_min": 347
+    },
+    {
+      "epoch": 4.078624078624078,
+      "grad_norm": 1.1342526879934762,
+      "learning_rate": 1.1280018116815438e-05,
+      "loss": 0.0993,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09200088679790497,
+      "step": 1660,
+      "valid_targets_mean": 566.7,
+      "valid_targets_min": 330
+    },
+    {
+      "epoch": 4.090909090909091,
+      "grad_norm": 1.2009691446799167,
+      "learning_rate": 1.115155446018404e-05,
+      "loss": 0.0998,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08615422993898392,
+      "step": 1665,
+      "valid_targets_mean": 539.9,
+      "valid_targets_min": 347
+    },
+    {
+      "epoch": 4.103194103194103,
+      "grad_norm": 1.2104780895909246,
+      "learning_rate": 1.1023543122491626e-05,
+      "loss": 0.0972,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10453509539365768,
+      "step": 1670,
+      "valid_targets_mean": 530.2,
+      "valid_targets_min": 357
+    },
+    {
+      "epoch": 4.115479115479116,
+      "grad_norm": 1.2910856332936864,
+      "learning_rate": 1.089599064748108e-05,
+      "loss": 0.1014,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09106092900037766,
+      "step": 1675,
+      "valid_targets_mean": 607.2,
+      "valid_targets_min": 347
+    },
+    {
+      "epoch": 4.127764127764128,
+      "grad_norm": 1.230623396109194,
+      "learning_rate": 1.0768903555438927e-05,
+      "loss": 0.1002,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11302720755338669,
+      "step": 1680,
+      "valid_targets_mean": 638.8,
+      "valid_targets_min": 314
+    },
+    {
+      "epoch": 4.14004914004914,
+      "grad_norm": 1.071186203200074,
+      "learning_rate": 1.0642288342862007e-05,
+      "loss": 0.097,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09703318774700165,
+      "step": 1685,
+      "valid_targets_mean": 661.2,
+      "valid_targets_min": 434
+    },
+    {
+      "epoch": 4.152334152334152,
+      "grad_norm": 1.0873993018103425,
+      "learning_rate": 1.051615148212544e-05,
+      "loss": 0.0981,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09849526733160019,
+      "step": 1690,
+      "valid_targets_mean": 622.8,
+      "valid_targets_min": 390
+    },
+    {
+      "epoch": 4.164619164619165,
+      "grad_norm": 1.1263995338301178,
+      "learning_rate": 1.0390499421151706e-05,
+      "loss": 0.0987,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09109313786029816,
+      "step": 1695,
+      "valid_targets_mean": 578.1,
+      "valid_targets_min": 355
+    },
+    {
+      "epoch": 4.176904176904177,
+      "grad_norm": 1.0985910939618122,
+      "learning_rate": 1.0265338583081088e-05,
+      "loss": 0.0984,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08648423850536346,
+      "step": 1700,
+      "valid_targets_mean": 595.2,
+      "valid_targets_min": 364
+    },
+    {
+      "epoch": 4.1891891891891895,
+      "grad_norm": 1.1590572484293804,
+      "learning_rate": 1.0140675365943284e-05,
+      "loss": 0.1003,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09820704907178879,
+      "step": 1705,
+      "valid_targets_mean": 609.7,
+      "valid_targets_min": 338
+    },
+    {
+      "epoch": 4.201474201474202,
+      "grad_norm": 1.1427962066433188,
+      "learning_rate": 1.0016516142330404e-05,
+      "loss": 0.1038,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09883490204811096,
+      "step": 1710,
+      "valid_targets_mean": 652.7,
+      "valid_targets_min": 360
+    },
+    {
+      "epoch": 4.2137592137592135,
+      "grad_norm": 1.2229775700196275,
+      "learning_rate": 9.89286725907117e-06,
+      "loss": 0.1029,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09851114451885223,
+      "step": 1715,
+      "valid_targets_mean": 616.8,
+      "valid_targets_min": 335
+    },
+    {
+      "epoch": 4.226044226044226,
+      "grad_norm": 1.2892267743946126,
+      "learning_rate": 9.769735036906475e-06,
+      "loss": 0.1015,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.094369076192379,
+      "step": 1720,
+      "valid_targets_mean": 535.9,
+      "valid_targets_min": 371
+    },
+    {
+      "epoch": 4.238329238329238,
+      "grad_norm": 1.1935112749649825,
+      "learning_rate": 9.647125770166321e-06,
+      "loss": 0.1034,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10198774188756943,
+      "step": 1725,
+      "valid_targets_mean": 545.2,
+      "valid_targets_min": 350
+    },
+    {
+      "epoch": 4.250614250614251,
+      "grad_norm": 1.2238824327091253,
+      "learning_rate": 9.525045726448001e-06,
+      "loss": 0.1022,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10707564651966095,
+      "step": 1730,
+      "valid_targets_mean": 562.8,
+      "valid_targets_min": 352
+    },
+    {
+      "epoch": 4.262899262899263,
+      "grad_norm": 1.2437855878561452,
+      "learning_rate": 9.40350114629577e-06,
+      "loss": 0.0999,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10580414533615112,
+      "step": 1735,
+      "valid_targets_mean": 605.5,
+      "valid_targets_min": 366
+    },
+    {
+      "epoch": 4.275184275184275,
+      "grad_norm": 1.2388603553940354,
+      "learning_rate": 9.282498242881784e-06,
+      "loss": 0.1007,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10578162968158722,
+      "step": 1740,
+      "valid_targets_mean": 588.9,
+      "valid_targets_min": 364
+    },
+    {
+      "epoch": 4.287469287469287,
+      "grad_norm": 1.1790865419807355,
+      "learning_rate": 9.162043201688517e-06,
+      "loss": 0.0992,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.105898916721344,
+      "step": 1745,
+      "valid_targets_mean": 573.7,
+      "valid_targets_min": 304
+    },
+    {
+      "epoch": 4.2997542997543,
+      "grad_norm": 1.2072953523804615,
+      "learning_rate": 9.042142180192596e-06,
+      "loss": 0.0969,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10502354800701141,
+      "step": 1750,
+      "valid_targets_mean": 618.2,
+      "valid_targets_min": 311
+    },
+    {
+      "epoch": 4.312039312039312,
+      "grad_norm": 1.2456975643832313,
+      "learning_rate": 8.92280130754998e-06,
+      "loss": 0.0978,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10343105345964432,
+      "step": 1755,
+      "valid_targets_mean": 583.5,
+      "valid_targets_min": 367
+    },
+    {
+      "epoch": 4.324324324324325,
+      "grad_norm": 1.1047621262208223,
+      "learning_rate": 8.804026684282694e-06,
+      "loss": 0.1017,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09652222692966461,
+      "step": 1760,
+      "valid_targets_mean": 556.1,
+      "valid_targets_min": 353
+    },
+    {
+      "epoch": 4.336609336609337,
+      "grad_norm": 1.1690242316360768,
+      "learning_rate": 8.685824381966975e-06,
+      "loss": 0.0998,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.1021624505519867,
+      "step": 1765,
+      "valid_targets_mean": 600.4,
+      "valid_targets_min": 376
+    },
+    {
+      "epoch": 4.348894348894349,
+      "grad_norm": 1.168482386851969,
+      "learning_rate": 8.568200442922865e-06,
+      "loss": 0.098,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09414025396108627,
+      "step": 1770,
+      "valid_targets_mean": 533.4,
+      "valid_targets_min": 353
+    },
+    {
+      "epoch": 4.361179361179361,
+      "grad_norm": 1.2014191056193684,
+      "learning_rate": 8.451160879905398e-06,
+      "loss": 0.1044,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11013706028461456,
+      "step": 1775,
+      "valid_targets_mean": 574.8,
+      "valid_targets_min": 358
+    },
+    {
+      "epoch": 4.3734643734643734,
+      "grad_norm": 1.3497231211942646,
+      "learning_rate": 8.33471167579717e-06,
+      "loss": 0.0998,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11065986752510071,
+      "step": 1780,
+      "valid_targets_mean": 560.4,
+      "valid_targets_min": 338
+    },
+    {
+      "epoch": 4.385749385749386,
+      "grad_norm": 1.1746303621149712,
+      "learning_rate": 8.218858783302566e-06,
+      "loss": 0.1058,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10344547033309937,
+      "step": 1785,
+      "valid_targets_mean": 543.9,
+      "valid_targets_min": 365
+    },
+    {
+      "epoch": 4.398034398034398,
+      "grad_norm": 1.2044261375281227,
+      "learning_rate": 8.103608124643412e-06,
+      "loss": 0.1008,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09899267554283142,
+      "step": 1790,
+      "valid_targets_mean": 519.8,
+      "valid_targets_min": 343
+    },
+    {
+      "epoch": 4.41031941031941,
+      "grad_norm": 1.2175130594542445,
+      "learning_rate": 7.988965591256284e-06,
+      "loss": 0.1021,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10632935166358948,
+      "step": 1795,
+      "valid_targets_mean": 628.6,
+      "valid_targets_min": 325
+    },
+    {
+      "epoch": 4.422604422604422,
+      "grad_norm": 1.2792552447352683,
+      "learning_rate": 7.874937043491331e-06,
+      "loss": 0.1047,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10912451148033142,
+      "step": 1800,
+      "valid_targets_mean": 513.6,
+      "valid_targets_min": 377
+    },
+    {
+      "epoch": 4.434889434889435,
+      "grad_norm": 1.182229707392635,
+      "learning_rate": 7.761528310312679e-06,
+      "loss": 0.1008,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09684855490922928,
+      "step": 1805,
+      "valid_targets_mean": 550.6,
+      "valid_targets_min": 338
+    },
+    {
+      "epoch": 4.447174447174447,
+      "grad_norm": 1.2263550156104428,
+      "learning_rate": 7.648745189000511e-06,
+      "loss": 0.1001,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10374526679515839,
+      "step": 1810,
+      "valid_targets_mean": 553.9,
+      "valid_targets_min": 310
+    },
+    {
+      "epoch": 4.45945945945946,
+      "grad_norm": 1.1731823115544264,
+      "learning_rate": 7.536593444854663e-06,
+      "loss": 0.1034,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10558365285396576,
+      "step": 1815,
+      "valid_targets_mean": 633.2,
+      "valid_targets_min": 338
+    },
+    {
+      "epoch": 4.471744471744472,
+      "grad_norm": 1.1474916322105462,
+      "learning_rate": 7.4250788108999686e-06,
+      "loss": 0.1001,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09885166585445404,
+      "step": 1820,
+      "valid_targets_mean": 717.4,
+      "valid_targets_min": 377
+    },
+    {
+      "epoch": 4.484029484029484,
+      "grad_norm": 1.094647206055445,
+      "learning_rate": 7.314206987593162e-06,
+      "loss": 0.1036,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10133668780326843,
+      "step": 1825,
+      "valid_targets_mean": 636.3,
+      "valid_targets_min": 329
+    },
+    {
+      "epoch": 4.496314496314496,
+      "grad_norm": 1.1905531326550673,
+      "learning_rate": 7.203983642531462e-06,
+      "loss": 0.1006,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09761211276054382,
+      "step": 1830,
+      "valid_targets_mean": 589.3,
+      "valid_targets_min": 378
+    },
+    {
+      "epoch": 4.5085995085995085,
+      "grad_norm": 1.231956473792296,
+      "learning_rate": 7.094414410162913e-06,
+      "loss": 0.1075,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12320096790790558,
+      "step": 1835,
+      "valid_targets_mean": 671.7,
+      "valid_targets_min": 318
+    },
+    {
+      "epoch": 4.520884520884521,
+      "grad_norm": 1.1916369852456652,
+      "learning_rate": 6.985504891498291e-06,
+      "loss": 0.103,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09794484078884125,
+      "step": 1840,
+      "valid_targets_mean": 506.5,
+      "valid_targets_min": 348
+    },
+    {
+      "epoch": 4.533169533169533,
+      "grad_norm": 1.1392602085497459,
+      "learning_rate": 6.8772606538248285e-06,
+      "loss": 0.1007,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0966009795665741,
+      "step": 1845,
+      "valid_targets_mean": 601.4,
+      "valid_targets_min": 377
+    },
+    {
+      "epoch": 4.545454545454545,
+      "grad_norm": 1.3903615271877925,
+      "learning_rate": 6.769687230421638e-06,
+      "loss": 0.1052,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11042024195194244,
+      "step": 1850,
+      "valid_targets_mean": 554.5,
+      "valid_targets_min": 278
+    },
+    {
+      "epoch": 4.557739557739557,
+      "grad_norm": 1.2300336106127345,
+      "learning_rate": 6.662790120276803e-06,
+      "loss": 0.1027,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09811799973249435,
+      "step": 1855,
+      "valid_targets_mean": 534.4,
+      "valid_targets_min": 333
+    },
+    {
+      "epoch": 4.57002457002457,
+      "grad_norm": 1.2200961784026068,
+      "learning_rate": 6.556574787806344e-06,
+      "loss": 0.1013,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10353550314903259,
+      "step": 1860,
+      "valid_targets_mean": 529.3,
+      "valid_targets_min": 340
+    },
+    {
+      "epoch": 4.582309582309582,
+      "grad_norm": 1.1625800754502562,
+      "learning_rate": 6.451046662574831e-06,
+      "loss": 0.1009,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10336748510599136,
+      "step": 1865,
+      "valid_targets_mean": 633.1,
+      "valid_targets_min": 419
+    },
+    {
+      "epoch": 4.594594594594595,
+      "grad_norm": 1.2713762384531742,
+      "learning_rate": 6.346211139017877e-06,
+      "loss": 0.1012,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10492067039012909,
+      "step": 1870,
+      "valid_targets_mean": 561.4,
+      "valid_targets_min": 327
+    },
+    {
+      "epoch": 4.606879606879607,
+      "grad_norm": 1.3231558835585264,
+      "learning_rate": 6.242073576166337e-06,
+      "loss": 0.1015,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10041987895965576,
+      "step": 1875,
+      "valid_targets_mean": 548.6,
+      "valid_targets_min": 376
+    },
+    {
+      "epoch": 4.61916461916462,
+      "grad_norm": 1.2601382420703833,
+      "learning_rate": 6.138639297372404e-06,
+      "loss": 0.0959,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09813426434993744,
+      "step": 1880,
+      "valid_targets_mean": 577.2,
+      "valid_targets_min": 376
+    },
+    {
+      "epoch": 4.631449631449631,
+      "grad_norm": 1.2257187798744478,
+      "learning_rate": 6.035913590037479e-06,
+      "loss": 0.0998,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.11072864383459091,
+      "step": 1885,
+      "valid_targets_mean": 576.0,
+      "valid_targets_min": 315
+    },
+    {
+      "epoch": 4.643734643734644,
+      "grad_norm": 1.2004615205444573,
+      "learning_rate": 5.933901705341851e-06,
+      "loss": 0.1016,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09538184106349945,
+      "step": 1890,
+      "valid_targets_mean": 589.8,
+      "valid_targets_min": 380
+    },
+    {
+      "epoch": 4.656019656019656,
+      "grad_norm": 1.1868393241698096,
+      "learning_rate": 5.832608857976321e-06,
+      "loss": 0.095,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10105034708976746,
+      "step": 1895,
+      "valid_targets_mean": 611.6,
+      "valid_targets_min": 385
+    },
+    {
+      "epoch": 4.6683046683046685,
+      "grad_norm": 1.3081946253917116,
+      "learning_rate": 5.732040225875584e-06,
+      "loss": 0.1022,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.12059606611728668,
+      "step": 1900,
+      "valid_targets_mean": 560.9,
+      "valid_targets_min": 354
+    },
+    {
+      "epoch": 4.680589680589681,
+      "grad_norm": 1.2987377377767717,
+      "learning_rate": 5.632200949953579e-06,
+      "loss": 0.1008,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10647489130496979,
+      "step": 1905,
+      "valid_targets_mean": 531.2,
+      "valid_targets_min": 355
+    },
+    {
+      "epoch": 4.6928746928746925,
+      "grad_norm": 1.3291852046527648,
+      "learning_rate": 5.533096133840677e-06,
+      "loss": 0.0966,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09767606109380722,
+      "step": 1910,
+      "valid_targets_mean": 544.9,
+      "valid_targets_min": 330
+    },
+    {
+      "epoch": 4.705159705159705,
+      "grad_norm": 1.0561152027218779,
+      "learning_rate": 5.434730843622778e-06,
+      "loss": 0.0929,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0867503434419632,
+      "step": 1915,
+      "valid_targets_mean": 566.1,
+      "valid_targets_min": 385
+    },
+    {
+      "epoch": 4.717444717444717,
+      "grad_norm": 1.1538215958161149,
+      "learning_rate": 5.337110107582377e-06,
+      "loss": 0.0953,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09404034167528152,
+      "step": 1920,
+      "valid_targets_mean": 560.6,
+      "valid_targets_min": 393
+    },
+    {
+      "epoch": 4.72972972972973,
+      "grad_norm": 1.1289210852598999,
+      "learning_rate": 5.2402389159414755e-06,
+      "loss": 0.1005,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10088243335485458,
+      "step": 1925,
+      "valid_targets_mean": 601.7,
+      "valid_targets_min": 391
+    },
+    {
+      "epoch": 4.742014742014742,
+      "grad_norm": 1.1358514090172158,
+      "learning_rate": 5.144122220606542e-06,
+      "loss": 0.0945,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0944918692111969,
+      "step": 1930,
+      "valid_targets_mean": 662.3,
+      "valid_targets_min": 363
+    },
+    {
+      "epoch": 4.754299754299755,
+      "grad_norm": 1.1514337440143445,
+      "learning_rate": 5.048764934915349e-06,
+      "loss": 0.0964,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09842099249362946,
+      "step": 1935,
+      "valid_targets_mean": 574.4,
+      "valid_targets_min": 354
+    },
+    {
+      "epoch": 4.766584766584766,
+      "grad_norm": 1.1230515925428892,
+      "learning_rate": 4.954171933385805e-06,
+      "loss": 0.0987,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09137429296970367,
+      "step": 1940,
+      "valid_targets_mean": 570.8,
+      "valid_targets_min": 337
+    },
+    {
+      "epoch": 4.778869778869779,
+      "grad_norm": 1.2121531782344195,
+      "learning_rate": 4.8603480514667836e-06,
+      "loss": 0.1008,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09575438499450684,
+      "step": 1945,
+      "valid_targets_mean": 558.6,
+      "valid_targets_min": 364
+    },
+    {
+      "epoch": 4.791154791154791,
+      "grad_norm": 1.1949570463904897,
+      "learning_rate": 4.767298085290963e-06,
+      "loss": 0.1025,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10193642973899841,
+      "step": 1950,
+      "valid_targets_mean": 609.2,
+      "valid_targets_min": 330
+    },
+    {
+      "epoch": 4.803439803439804,
+      "grad_norm": 1.2296273425861448,
+      "learning_rate": 4.675026791429624e-06,
+      "loss": 0.1015,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10295425355434418,
+      "step": 1955,
+      "valid_targets_mean": 519.8,
+      "valid_targets_min": 316
+    },
+    {
+      "epoch": 4.815724815724816,
+      "grad_norm": 1.1149494013206527,
+      "learning_rate": 4.583538886649525e-06,
+      "loss": 0.0955,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0870693176984787,
+      "step": 1960,
+      "valid_targets_mean": 525.1,
+      "valid_targets_min": 360
+    },
+    {
+      "epoch": 4.828009828009828,
+      "grad_norm": 1.1527137920360166,
+      "learning_rate": 4.492839047671764e-06,
+      "loss": 0.1003,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10126141458749771,
+      "step": 1965,
+      "valid_targets_mean": 597.5,
+      "valid_targets_min": 345
+    },
+    {
+      "epoch": 4.84029484029484,
+      "grad_norm": 1.285693094621376,
+      "learning_rate": 4.4029319109327465e-06,
+      "loss": 0.098,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10001290589570999,
+      "step": 1970,
+      "valid_targets_mean": 564.9,
+      "valid_targets_min": 410
+    },
+    {
+      "epoch": 4.8525798525798525,
+      "grad_norm": 1.1948838524886878,
+      "learning_rate": 4.313822072347136e-06,
+      "loss": 0.0973,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10110676288604736,
+      "step": 1975,
+      "valid_targets_mean": 609.7,
+      "valid_targets_min": 361
+    },
+    {
+      "epoch": 4.864864864864865,
+      "grad_norm": 1.0867537916477328,
+      "learning_rate": 4.22551408707296e-06,
+      "loss": 0.0965,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09373465180397034,
+      "step": 1980,
+      "valid_targets_mean": 669.0,
+      "valid_targets_min": 400
+    },
+    {
+      "epoch": 4.877149877149877,
+      "grad_norm": 1.1751891930352016,
+      "learning_rate": 4.138012469278714e-06,
+      "loss": 0.1015,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09852367639541626,
+      "step": 1985,
+      "valid_targets_mean": 546.5,
+      "valid_targets_min": 350
+    },
+    {
+      "epoch": 4.88943488943489,
+      "grad_norm": 1.256508408435214,
+      "learning_rate": 4.051321691912649e-06,
+      "loss": 0.1031,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10389189422130585,
+      "step": 1990,
+      "valid_targets_mean": 519.4,
+      "valid_targets_min": 321
+    },
+    {
+      "epoch": 4.901719901719901,
+      "grad_norm": 1.154895355871411,
+      "learning_rate": 3.9654461864740935e-06,
+      "loss": 0.0991,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09166670590639114,
+      "step": 1995,
+      "valid_targets_mean": 577.4,
+      "valid_targets_min": 324
+    },
+    {
+      "epoch": 4.914004914004914,
+      "grad_norm": 1.4017783580922674,
+      "learning_rate": 3.880390342786915e-06,
+      "loss": 0.0968,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09913163632154465,
+      "step": 2000,
+      "valid_targets_mean": 477.3,
+      "valid_targets_min": 350
+    },
+    {
+      "epoch": 4.926289926289926,
+      "grad_norm": 1.1230883383998704,
+      "learning_rate": 3.7961585087751516e-06,
+      "loss": 0.0932,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0872161015868187,
+      "step": 2005,
+      "valid_targets_mean": 639.8,
+      "valid_targets_min": 389
+    },
+    {
+      "epoch": 4.938574938574939,
+      "grad_norm": 1.2215546893236813,
+      "learning_rate": 3.71275499024071e-06,
+      "loss": 0.0997,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10271018743515015,
+      "step": 2010,
+      "valid_targets_mean": 554.2,
+      "valid_targets_min": 349
+    },
+    {
+      "epoch": 4.950859950859951,
+      "grad_norm": 1.1371869103474366,
+      "learning_rate": 3.6301840506433083e-06,
+      "loss": 0.0942,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09029437601566315,
+      "step": 2015,
+      "valid_targets_mean": 575.8,
+      "valid_targets_min": 378
+    },
+    {
+      "epoch": 4.963144963144963,
+      "grad_norm": 1.5723266744911633,
+      "learning_rate": 3.5484499108824853e-06,
+      "loss": 0.0978,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09650382399559021,
+      "step": 2020,
+      "valid_targets_mean": 605.9,
+      "valid_targets_min": 300
+    },
+    {
+      "epoch": 4.975429975429975,
+      "grad_norm": 1.201414516571051,
+      "learning_rate": 3.4675567490818727e-06,
+      "loss": 0.1029,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.10197515785694122,
+      "step": 2025,
+      "valid_targets_mean": 624.2,
+      "valid_targets_min": 368
+    },
+    {
+      "epoch": 4.987714987714988,
+      "grad_norm": 1.1607939133459078,
+      "learning_rate": 3.3875087003756036e-06,
+      "loss": 0.0975,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09291709959506989,
+      "step": 2030,
+      "valid_targets_mean": 587.6,
+      "valid_targets_min": 346
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 1.1476704642538418,
+      "learning_rate": 3.30830985669691e-06,
+      "loss": 0.0966,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08938758075237274,
+      "step": 2035,
+      "valid_targets_mean": 497.5,
+      "valid_targets_min": 374
+    },
+    {
+      "epoch": 5.012285012285012,
+      "grad_norm": 1.0154622918334222,
+      "learning_rate": 3.22996426656899e-06,
+      "loss": 0.0854,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08019639551639557,
+      "step": 2040,
+      "valid_targets_mean": 645.4,
+      "valid_targets_min": 318
+    },
+    {
+      "epoch": 5.024570024570025,
+      "grad_norm": 1.0873366416439045,
+      "learning_rate": 3.1524759348980096e-06,
+      "loss": 0.0836,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08791200816631317,
+      "step": 2045,
+      "valid_targets_mean": 640.9,
+      "valid_targets_min": 353
+    },
+    {
+      "epoch": 5.036855036855036,
+      "grad_norm": 1.2422621105770808,
+      "learning_rate": 3.0758488227684212e-06,
+      "loss": 0.0828,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08541159331798553,
+      "step": 2050,
+      "valid_targets_mean": 473.1,
+      "valid_targets_min": 315
+    },
+    {
+      "epoch": 5.049140049140049,
+      "grad_norm": 1.1245818830882286,
+      "learning_rate": 3.0000868472404423e-06,
+      "loss": 0.0889,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0811869204044342,
+      "step": 2055,
+      "valid_targets_mean": 620.8,
+      "valid_targets_min": 350
+    },
+    {
+      "epoch": 5.061425061425061,
+      "grad_norm": 1.1981733121122524,
+      "learning_rate": 2.9251938811498436e-06,
+      "loss": 0.0868,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08584136515855789,
+      "step": 2060,
+      "valid_targets_mean": 559.6,
+      "valid_targets_min": 366
+    },
+    {
+      "epoch": 5.073710073710074,
+      "grad_norm": 1.143387432773036,
+      "learning_rate": 2.8511737529099704e-06,
+      "loss": 0.0859,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08999934047460556,
+      "step": 2065,
+      "valid_targets_mean": 638.4,
+      "valid_targets_min": 375
+    },
+    {
+      "epoch": 5.085995085995086,
+      "grad_norm": 1.143479733060532,
+      "learning_rate": 2.7780302463160235e-06,
+      "loss": 0.0875,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0899144858121872,
+      "step": 2070,
+      "valid_targets_mean": 600.1,
+      "valid_targets_min": 305
+    },
+    {
+      "epoch": 5.098280098280099,
+      "grad_norm": 1.2493107418250313,
+      "learning_rate": 2.705767100351673e-06,
+      "loss": 0.0875,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08807513862848282,
+      "step": 2075,
+      "valid_targets_mean": 477.2,
+      "valid_targets_min": 328
+    },
+    {
+      "epoch": 5.11056511056511,
+      "grad_norm": 1.277539917750501,
+      "learning_rate": 2.634388008997899e-06,
+      "loss": 0.0872,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09081301093101501,
+      "step": 2080,
+      "valid_targets_mean": 548.4,
+      "valid_targets_min": 326
+    },
+    {
+      "epoch": 5.122850122850123,
+      "grad_norm": 1.041138079187501,
+      "learning_rate": 2.5638966210441597e-06,
+      "loss": 0.0817,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07195958495140076,
+      "step": 2085,
+      "valid_targets_mean": 630.1,
+      "valid_targets_min": 398
+    },
+    {
+      "epoch": 5.135135135135135,
+      "grad_norm": 1.2357154232642422,
+      "learning_rate": 2.4942965399018926e-06,
+      "loss": 0.0865,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09714603424072266,
+      "step": 2090,
+      "valid_targets_mean": 696.2,
+      "valid_targets_min": 365
+    },
+    {
+      "epoch": 5.1474201474201475,
+      "grad_norm": 1.2604606849246982,
+      "learning_rate": 2.425591323420289e-06,
+      "loss": 0.0867,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08725640177726746,
+      "step": 2095,
+      "valid_targets_mean": 555.8,
+      "valid_targets_min": 287
+    },
+    {
+      "epoch": 5.15970515970516,
+      "grad_norm": 1.2025961132788572,
+      "learning_rate": 2.357784483704444e-06,
+      "loss": 0.0851,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0895378589630127,
+      "step": 2100,
+      "valid_targets_mean": 580.9,
+      "valid_targets_min": 343
+    },
+    {
+      "epoch": 5.171990171990172,
+      "grad_norm": 1.253165445892313,
+      "learning_rate": 2.2908794869358044e-06,
+      "loss": 0.0835,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08695147931575775,
+      "step": 2105,
+      "valid_targets_mean": 505.1,
+      "valid_targets_min": 352
+    },
+    {
+      "epoch": 5.184275184275184,
+      "grad_norm": 1.2289552360721567,
+      "learning_rate": 2.2248797531949952e-06,
+      "loss": 0.0891,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0879005640745163,
+      "step": 2110,
+      "valid_targets_mean": 531.9,
+      "valid_targets_min": 321
+    },
+    {
+      "epoch": 5.196560196560196,
+      "grad_norm": 1.1908990615471768,
+      "learning_rate": 2.1597886562869917e-06,
+      "loss": 0.0859,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0877969041466713,
+      "step": 2115,
+      "valid_targets_mean": 577.0,
+      "valid_targets_min": 337
+    },
+    {
+      "epoch": 5.208845208845209,
+      "grad_norm": 1.192626976342137,
+      "learning_rate": 2.095609523568638e-06,
+      "loss": 0.0842,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07734735310077667,
+      "step": 2120,
+      "valid_targets_mean": 562.0,
+      "valid_targets_min": 329
+    },
+    {
+      "epoch": 5.221130221130221,
+      "grad_norm": 1.2082555924579998,
+      "learning_rate": 2.0323456357785855e-06,
+      "loss": 0.087,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0842600166797638,
+      "step": 2125,
+      "valid_targets_mean": 585.1,
+      "valid_targets_min": 352
+    },
+    {
+      "epoch": 5.233415233415234,
+      "grad_norm": 1.2386435260363529,
+      "learning_rate": 1.970000226869553e-06,
+      "loss": 0.0878,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09272652119398117,
+      "step": 2130,
+      "valid_targets_mean": 524.5,
+      "valid_targets_min": 320
+    },
+    {
+      "epoch": 5.245700245700245,
+      "grad_norm": 1.126056243469827,
+      "learning_rate": 1.90857648384305e-06,
+      "loss": 0.0806,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07395152747631073,
+      "step": 2135,
+      "valid_targets_mean": 619.2,
+      "valid_targets_min": 313
+    },
+    {
+      "epoch": 5.257985257985258,
+      "grad_norm": 1.1666537336061145,
+      "learning_rate": 1.848077546586431e-06,
+      "loss": 0.0828,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08457744866609573,
+      "step": 2140,
+      "valid_targets_mean": 546.1,
+      "valid_targets_min": 374
+    },
+    {
+      "epoch": 5.27027027027027,
+      "grad_norm": 1.199622886892468,
+      "learning_rate": 1.7885065077123976e-06,
+      "loss": 0.083,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08328172564506531,
+      "step": 2145,
+      "valid_targets_mean": 554.4,
+      "valid_targets_min": 387
+    },
+    {
+      "epoch": 5.282555282555283,
+      "grad_norm": 1.1728412716455279,
+      "learning_rate": 1.7298664124009245e-06,
+      "loss": 0.0834,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08228134363889694,
+      "step": 2150,
+      "valid_targets_mean": 528.0,
+      "valid_targets_min": 312
+    },
+    {
+      "epoch": 5.294840294840295,
+      "grad_norm": 1.1803113468201745,
+      "learning_rate": 1.672160258243567e-06,
+      "loss": 0.0826,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08139941096305847,
+      "step": 2155,
+      "valid_targets_mean": 572.4,
+      "valid_targets_min": 356
+    },
+    {
+      "epoch": 5.3071253071253075,
+      "grad_norm": 1.2619621031879709,
+      "learning_rate": 1.615390995090258e-06,
+      "loss": 0.0868,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09053842723369598,
+      "step": 2160,
+      "valid_targets_mean": 577.2,
+      "valid_targets_min": 350
+    },
+    {
+      "epoch": 5.319410319410319,
+      "grad_norm": 1.2248890955579523,
+      "learning_rate": 1.559561524898492e-06,
+      "loss": 0.0812,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08257773518562317,
+      "step": 2165,
+      "valid_targets_mean": 617.9,
+      "valid_targets_min": 381
+    },
+    {
+      "epoch": 5.3316953316953315,
+      "grad_norm": 1.2467046188005135,
+      "learning_rate": 1.5046747015849893e-06,
+      "loss": 0.0875,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0834282860159874,
+      "step": 2170,
+      "valid_targets_mean": 543.5,
+      "valid_targets_min": 387
+    },
+    {
+      "epoch": 5.343980343980344,
+      "grad_norm": 1.1148418333356434,
+      "learning_rate": 1.4507333308798255e-06,
+      "loss": 0.087,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07972699403762817,
+      "step": 2175,
+      "valid_targets_mean": 576.8,
+      "valid_targets_min": 317
+    },
+    {
+      "epoch": 5.356265356265356,
+      "grad_norm": 1.0664970769867355,
+      "learning_rate": 1.3977401701829752e-06,
+      "loss": 0.0771,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07614714652299881,
+      "step": 2180,
+      "valid_targets_mean": 600.1,
+      "valid_targets_min": 344
+    },
+    {
+      "epoch": 5.368550368550369,
+      "grad_norm": 1.0693404257105559,
+      "learning_rate": 1.345697928423384e-06,
+      "loss": 0.0871,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07640987634658813,
+      "step": 2185,
+      "valid_targets_mean": 665.3,
+      "valid_targets_min": 353
+    },
+    {
+      "epoch": 5.38083538083538,
+      "grad_norm": 1.189930580211284,
+      "learning_rate": 1.2946092659204767e-06,
+      "loss": 0.0835,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09601159393787384,
+      "step": 2190,
+      "valid_targets_mean": 593.2,
+      "valid_targets_min": 353
+    },
+    {
+      "epoch": 5.393120393120393,
+      "grad_norm": 1.1238320625578555,
+      "learning_rate": 1.244476794248175e-06,
+      "loss": 0.0843,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08733360469341278,
+      "step": 2195,
+      "valid_targets_mean": 635.1,
+      "valid_targets_min": 374
+    },
+    {
+      "epoch": 5.405405405405405,
+      "grad_norm": 1.2770058118624072,
+      "learning_rate": 1.1953030761014017e-06,
+      "loss": 0.0834,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09015761315822601,
+      "step": 2200,
+      "valid_targets_mean": 556.9,
+      "valid_targets_min": 293
+    },
+    {
+      "epoch": 5.417690417690418,
+      "grad_norm": 1.13370178694208,
+      "learning_rate": 1.147090625165055e-06,
+      "loss": 0.0834,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07899008691310883,
+      "step": 2205,
+      "valid_targets_mean": 598.9,
+      "valid_targets_min": 346
+    },
+    {
+      "epoch": 5.42997542997543,
+      "grad_norm": 1.1383568574789888,
+      "learning_rate": 1.0998419059855503e-06,
+      "loss": 0.0868,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08398991823196411,
+      "step": 2210,
+      "valid_targets_mean": 588.1,
+      "valid_targets_min": 355
+    },
+    {
+      "epoch": 5.442260442260443,
+      "grad_norm": 1.2374596577372317,
+      "learning_rate": 1.053559333844798e-06,
+      "loss": 0.0836,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0927569717168808,
+      "step": 2215,
+      "valid_targets_mean": 554.8,
+      "valid_targets_min": 299
+    },
+    {
+      "epoch": 5.454545454545454,
+      "grad_norm": 1.060062165304812,
+      "learning_rate": 1.0082452746367721e-06,
+      "loss": 0.0798,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07243770360946655,
+      "step": 2220,
+      "valid_targets_mean": 688.8,
+      "valid_targets_min": 412
+    },
+    {
+      "epoch": 5.466830466830467,
+      "grad_norm": 1.234886143589627,
+      "learning_rate": 9.639020447465475e-07,
+      "loss": 0.0847,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08597145974636078,
+      "step": 2225,
+      "valid_targets_mean": 531.4,
+      "valid_targets_min": 368
+    },
+    {
+      "epoch": 5.479115479115479,
+      "grad_norm": 1.0957537106814175,
+      "learning_rate": 9.205319109318922e-07,
+      "loss": 0.0825,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07663384079933167,
+      "step": 2230,
+      "valid_targets_mean": 577.2,
+      "valid_targets_min": 349
+    },
+    {
+      "epoch": 5.4914004914004915,
+      "grad_norm": 1.3059978658871396,
+      "learning_rate": 8.781370902074049e-07,
+      "loss": 0.0844,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0905524268746376,
+      "step": 2235,
+      "valid_targets_mean": 514.3,
+      "valid_targets_min": 304
+    },
+    {
+      "epoch": 5.503685503685504,
+      "grad_norm": 1.145886936771296,
+      "learning_rate": 8.367197497311719e-07,
+      "loss": 0.0829,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08202914893627167,
+      "step": 2240,
+      "valid_targets_mean": 638.8,
+      "valid_targets_min": 349
+    },
+    {
+      "epoch": 5.515970515970516,
+      "grad_norm": 1.1560278108403075,
+      "learning_rate": 7.962820066939958e-07,
+      "loss": 0.0832,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08465921878814697,
+      "step": 2245,
+      "valid_targets_mean": 617.2,
+      "valid_targets_min": 330
+    },
+    {
+      "epoch": 5.528255528255528,
+      "grad_norm": 1.146827460792829,
+      "learning_rate": 7.568259282111645e-07,
+      "loss": 0.0835,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08355319499969482,
+      "step": 2250,
+      "valid_targets_mean": 589.8,
+      "valid_targets_min": 408
+    },
+    {
+      "epoch": 5.54054054054054,
+      "grad_norm": 1.2678244005272907,
+      "learning_rate": 7.183535312167755e-07,
+      "loss": 0.0889,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08739691227674484,
+      "step": 2255,
+      "valid_targets_mean": 537.2,
+      "valid_targets_min": 387
+    },
+    {
+      "epoch": 5.552825552825553,
+      "grad_norm": 1.4375448638349997,
+      "learning_rate": 6.808667823606474e-07,
+      "loss": 0.0791,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07598749548196793,
+      "step": 2260,
+      "valid_targets_mean": 557.4,
+      "valid_targets_min": 344
+    },
+    {
+      "epoch": 5.565110565110565,
+      "grad_norm": 1.2784791536932771,
+      "learning_rate": 6.443675979077779e-07,
+      "loss": 0.0836,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09129341691732407,
+      "step": 2265,
+      "valid_targets_mean": 611.4,
+      "valid_targets_min": 355
+    },
+    {
+      "epoch": 5.577395577395578,
+      "grad_norm": 1.146114809603193,
+      "learning_rate": 6.088578436403847e-07,
+      "loss": 0.0777,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07668814063072205,
+      "step": 2270,
+      "valid_targets_mean": 548.8,
+      "valid_targets_min": 333
+    },
+    {
+      "epoch": 5.58968058968059,
+      "grad_norm": 1.1701260037394332,
+      "learning_rate": 5.743393347625436e-07,
+      "loss": 0.0809,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0863061472773552,
+      "step": 2275,
+      "valid_targets_mean": 631.6,
+      "valid_targets_min": 328
+    },
+    {
+      "epoch": 5.601965601965602,
+      "grad_norm": 1.1336198412288852,
+      "learning_rate": 5.408138358073833e-07,
+      "loss": 0.0829,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07773379981517792,
+      "step": 2280,
+      "valid_targets_mean": 707.2,
+      "valid_targets_min": 335
+    },
+    {
+      "epoch": 5.614250614250614,
+      "grad_norm": 1.1732238369009445,
+      "learning_rate": 5.082830605468969e-07,
+      "loss": 0.0816,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0849945917725563,
+      "step": 2285,
+      "valid_targets_mean": 632.2,
+      "valid_targets_min": 278
+    },
+    {
+      "epoch": 5.6265356265356266,
+      "grad_norm": 1.1247949445590306,
+      "learning_rate": 4.767486719043235e-07,
+      "loss": 0.0834,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07646386325359344,
+      "step": 2290,
+      "valid_targets_mean": 657.5,
+      "valid_targets_min": 409
+    },
+    {
+      "epoch": 5.638820638820639,
+      "grad_norm": 1.2775182803680445,
+      "learning_rate": 4.4621228186915833e-07,
+      "loss": 0.0824,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08856301754713058,
+      "step": 2295,
+      "valid_targets_mean": 620.8,
+      "valid_targets_min": 401
+    },
+    {
+      "epoch": 5.651105651105651,
+      "grad_norm": 1.2333249170791625,
+      "learning_rate": 4.166754514147275e-07,
+      "loss": 0.084,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08326977491378784,
+      "step": 2300,
+      "valid_targets_mean": 534.1,
+      "valid_targets_min": 379
+    },
+    {
+      "epoch": 5.663390663390663,
+      "grad_norm": 1.1108265985271855,
+      "learning_rate": 3.881396904184231e-07,
+      "loss": 0.0791,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07909250259399414,
+      "step": 2305,
+      "valid_targets_mean": 623.9,
+      "valid_targets_min": 372
+    },
+    {
+      "epoch": 5.675675675675675,
+      "grad_norm": 1.1071999111918622,
+      "learning_rate": 3.6060645758449584e-07,
+      "loss": 0.0793,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.06771372258663177,
+      "step": 2310,
+      "valid_targets_mean": 535.7,
+      "valid_targets_min": 328
+    },
+    {
+      "epoch": 5.687960687960688,
+      "grad_norm": 1.2811978558463222,
+      "learning_rate": 3.34077160369497e-07,
+      "loss": 0.0851,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08796574175357819,
+      "step": 2315,
+      "valid_targets_mean": 582.8,
+      "valid_targets_min": 272
+    },
+    {
+      "epoch": 5.7002457002457,
+      "grad_norm": 1.1575219121136995,
+      "learning_rate": 3.08553154910336e-07,
+      "loss": 0.0877,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08139719069004059,
+      "step": 2320,
+      "valid_targets_mean": 620.1,
+      "valid_targets_min": 372
+    },
+    {
+      "epoch": 5.712530712530713,
+      "grad_norm": 1.2556732274676712,
+      "learning_rate": 2.840357459549492e-07,
+      "loss": 0.0809,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08443877846002579,
+      "step": 2325,
+      "valid_targets_mean": 542.2,
+      "valid_targets_min": 399
+    },
+    {
+      "epoch": 5.724815724815725,
+      "grad_norm": 1.259874729323289,
+      "learning_rate": 2.6052618679560884e-07,
+      "loss": 0.085,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09191766381263733,
+      "step": 2330,
+      "valid_targets_mean": 461.4,
+      "valid_targets_min": 322
+    },
+    {
+      "epoch": 5.737100737100737,
+      "grad_norm": 1.1917534753808974,
+      "learning_rate": 2.380256792048541e-07,
+      "loss": 0.0789,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07446075975894928,
+      "step": 2335,
+      "valid_targets_mean": 511.6,
+      "valid_targets_min": 342
+    },
+    {
+      "epoch": 5.749385749385749,
+      "grad_norm": 1.2373562089047119,
+      "learning_rate": 2.1653537337405383e-07,
+      "loss": 0.0823,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08613431453704834,
+      "step": 2340,
+      "valid_targets_mean": 577.6,
+      "valid_targets_min": 396
+    },
+    {
+      "epoch": 5.761670761670762,
+      "grad_norm": 1.2034405427568784,
+      "learning_rate": 1.9605636785462234e-07,
+      "loss": 0.0795,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08189545571804047,
+      "step": 2345,
+      "valid_targets_mean": 621.9,
+      "valid_targets_min": 365
+    },
+    {
+      "epoch": 5.773955773955774,
+      "grad_norm": 1.0799325407284226,
+      "learning_rate": 1.7658970950185095e-07,
+      "loss": 0.0809,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07476474344730377,
+      "step": 2350,
+      "valid_targets_mean": 609.9,
+      "valid_targets_min": 321
+    },
+    {
+      "epoch": 5.7862407862407865,
+      "grad_norm": 1.1757906849036914,
+      "learning_rate": 1.5813639342140197e-07,
+      "loss": 0.0815,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08601002395153046,
+      "step": 2355,
+      "valid_targets_mean": 550.2,
+      "valid_targets_min": 379
+    },
+    {
+      "epoch": 5.798525798525798,
+      "grad_norm": 1.188427364444162,
+      "learning_rate": 1.4069736291843605e-07,
+      "loss": 0.0809,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07079163193702698,
+      "step": 2360,
+      "valid_targets_mean": 589.3,
+      "valid_targets_min": 338
+    },
+    {
+      "epoch": 5.8108108108108105,
+      "grad_norm": 1.2501121164250937,
+      "learning_rate": 1.242735094493952e-07,
+      "loss": 0.0839,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08723746240139008,
+      "step": 2365,
+      "valid_targets_mean": 535.8,
+      "valid_targets_min": 373
+    },
+    {
+      "epoch": 5.823095823095823,
+      "grad_norm": 1.2533918670853614,
+      "learning_rate": 1.0886567257643033e-07,
+      "loss": 0.082,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08290581405162811,
+      "step": 2370,
+      "valid_targets_mean": 550.7,
+      "valid_targets_min": 360
+    },
+    {
+      "epoch": 5.835380835380835,
+      "grad_norm": 1.1564818042371932,
+      "learning_rate": 9.447463992448891e-08,
+      "loss": 0.0847,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08502306044101715,
+      "step": 2375,
+      "valid_targets_mean": 659.5,
+      "valid_targets_min": 325
+    },
+    {
+      "epoch": 5.847665847665848,
+      "grad_norm": 1.1755092919019083,
+      "learning_rate": 8.110114714104277e-08,
+      "loss": 0.0819,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07557385414838791,
+      "step": 2380,
+      "valid_targets_mean": 598.8,
+      "valid_targets_min": 342
+    },
+    {
+      "epoch": 5.85995085995086,
+      "grad_norm": 1.1962496012954607,
+      "learning_rate": 6.874587785849152e-08,
+      "loss": 0.0812,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07918815314769745,
+      "step": 2385,
+      "valid_targets_mean": 603.8,
+      "valid_targets_min": 377
+    },
+    {
+      "epoch": 5.872235872235873,
+      "grad_norm": 1.306024990482639,
+      "learning_rate": 5.7409463659219286e-08,
+      "loss": 0.0845,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08919475227594376,
+      "step": 2390,
+      "valid_targets_mean": 493.2,
+      "valid_targets_min": 303
+    },
+    {
+      "epoch": 5.884520884520884,
+      "grad_norm": 1.2110406679406136,
+      "learning_rate": 4.709248404329625e-08,
+      "loss": 0.0805,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08352182805538177,
+      "step": 2395,
+      "valid_targets_mean": 576.7,
+      "valid_targets_min": 314
+    },
+    {
+      "epoch": 5.896805896805897,
+      "grad_norm": 1.1494918965138259,
+      "learning_rate": 3.7795466398868885e-08,
+      "loss": 0.0807,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07362495362758636,
+      "step": 2400,
+      "valid_targets_mean": 632.3,
+      "valid_targets_min": 376
+    },
+    {
+      "epoch": 5.909090909090909,
+      "grad_norm": 1.2125273811964747,
+      "learning_rate": 2.9518885975192702e-08,
+      "loss": 0.0813,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0766986608505249,
+      "step": 2405,
+      "valid_targets_mean": 540.9,
+      "valid_targets_min": 298
+    },
+    {
+      "epoch": 5.921375921375922,
+      "grad_norm": 1.3055089690040842,
+      "learning_rate": 2.226316585833832e-08,
+      "loss": 0.088,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.09940752387046814,
+      "step": 2410,
+      "valid_targets_mean": 630.2,
+      "valid_targets_min": 366
+    },
+    {
+      "epoch": 5.933660933660933,
+      "grad_norm": 1.2734724231764385,
+      "learning_rate": 1.6028676949570997e-08,
+      "loss": 0.0863,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08724012225866318,
+      "step": 2415,
+      "valid_targets_mean": 496.8,
+      "valid_targets_min": 343
+    },
+    {
+      "epoch": 5.945945945945946,
+      "grad_norm": 1.3140331206962972,
+      "learning_rate": 1.0815737946383575e-08,
+      "loss": 0.0797,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.0807531550526619,
+      "step": 2420,
+      "valid_targets_mean": 511.9,
+      "valid_targets_min": 355
+    },
+    {
+      "epoch": 5.958230958230958,
+      "grad_norm": 1.2133753164899292,
+      "learning_rate": 6.624615326207284e-09,
+      "loss": 0.086,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08304368704557419,
+      "step": 2425,
+      "valid_targets_mean": 509.8,
+      "valid_targets_min": 319
+    },
+    {
+      "epoch": 5.9705159705159705,
+      "grad_norm": 1.2628163299904318,
+      "learning_rate": 3.4555233327893124e-09,
+      "loss": 0.0888,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08596052974462509,
+      "step": 2430,
+      "valid_targets_mean": 489.0,
+      "valid_targets_min": 355
+    },
+    {
+      "epoch": 5.982800982800983,
+      "grad_norm": 1.1137438849834025,
+      "learning_rate": 1.3086239652415621e-09,
+      "loss": 0.0785,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.07572906464338303,
+      "step": 2435,
+      "valid_targets_mean": 634.5,
+      "valid_targets_min": 374
+    },
+    {
+      "epoch": 5.995085995085995,
+      "grad_norm": 1.256225517041073,
+      "learning_rate": 1.840269697628294e-10,
+      "loss": 0.0843,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08771345764398575,
+      "step": 2440,
+      "valid_targets_mean": 510.3,
+      "valid_targets_min": 316
+    },
+    {
+      "epoch": 6.0,
+      "loss_nan_ranks": 0,
+      "loss_rank_avg": 0.08076979219913483,
+      "step": 2442,
+      "total_flos": 140364736167936.0,
+      "train_loss": 0.16821050625614803,
+      "train_runtime": 7348.4915,
+      "train_samples_per_second": 5.31,
+      "train_steps_per_second": 0.332,
+      "valid_targets_mean": 624.3,
+      "valid_targets_min": 380
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 2442,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 6,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 140364736167936.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}