diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5524 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 782,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0025575447570332483,
+      "grad_norm": 21.953563303736964,
+      "learning_rate": 9.999959651660741e-06,
+      "loss": 1.3458,
+      "step": 1
+    },
+    {
+      "epoch": 0.005115089514066497,
+      "grad_norm": 11.652464724143648,
+      "learning_rate": 9.999838607294157e-06,
+      "loss": 1.293,
+      "step": 2
+    },
+    {
+      "epoch": 0.0076726342710997444,
+      "grad_norm": 14.373069724258055,
+      "learning_rate": 9.999636868853824e-06,
+      "loss": 0.9527,
+      "step": 3
+    },
+    {
+      "epoch": 0.010230179028132993,
+      "grad_norm": 9.875205922264723,
+      "learning_rate": 9.999354439595668e-06,
+      "loss": 0.9394,
+      "step": 4
+    },
+    {
+      "epoch": 0.01278772378516624,
+      "grad_norm": 15.116196776322745,
+      "learning_rate": 9.998991324077906e-06,
+      "loss": 0.8439,
+      "step": 5
+    },
+    {
+      "epoch": 0.015345268542199489,
+      "grad_norm": 11.057152894995923,
+      "learning_rate": 9.998547528160987e-06,
+      "loss": 0.8405,
+      "step": 6
+    },
+    {
+      "epoch": 0.017902813299232736,
+      "grad_norm": 9.365688071554938,
+      "learning_rate": 9.998023059007477e-06,
+      "loss": 0.8734,
+      "step": 7
+    },
+    {
+      "epoch": 0.020460358056265986,
+      "grad_norm": 8.392619519229692,
+      "learning_rate": 9.997417925081963e-06,
+      "loss": 0.7449,
+      "step": 8
+    },
+    {
+      "epoch": 0.023017902813299233,
+      "grad_norm": 9.340778471253389,
+      "learning_rate": 9.996732136150902e-06,
+      "loss": 0.7876,
+      "step": 9
+    },
+    {
+      "epoch": 0.02557544757033248,
+      "grad_norm": 8.20102806123342,
+      "learning_rate": 9.995965703282472e-06,
+      "loss": 0.7954,
+      "step": 10
+    },
+    {
+      "epoch": 0.028132992327365727,
+      "grad_norm": 8.76971572309029,
+      "learning_rate": 9.995118638846394e-06,
+      "loss": 0.8333,
+      "step": 11
+    },
+    {
+      "epoch": 0.030690537084398978,
+      "grad_norm": 7.557620850007077,
+      "learning_rate": 9.99419095651372e-06,
+      "loss": 0.8554,
+      "step": 12
+    },
+    {
+      "epoch": 0.03324808184143223,
+      "grad_norm": 7.743714182477662,
+      "learning_rate": 9.993182671256633e-06,
+      "loss": 0.8637,
+      "step": 13
+    },
+    {
+      "epoch": 0.03580562659846547,
+      "grad_norm": 7.32084774886768,
+      "learning_rate": 9.992093799348182e-06,
+      "loss": 0.8621,
+      "step": 14
+    },
+    {
+      "epoch": 0.03836317135549872,
+      "grad_norm": 7.6800585762630655,
+      "learning_rate": 9.990924358362037e-06,
+      "loss": 0.7075,
+      "step": 15
+    },
+    {
+      "epoch": 0.04092071611253197,
+      "grad_norm": 7.053712401961599,
+      "learning_rate": 9.9896743671722e-06,
+      "loss": 0.8078,
+      "step": 16
+    },
+    {
+      "epoch": 0.043478260869565216,
+      "grad_norm": 7.957894711517596,
+      "learning_rate": 9.988343845952697e-06,
+      "loss": 0.8534,
+      "step": 17
+    },
+    {
+      "epoch": 0.04603580562659847,
+      "grad_norm": 7.77708191899406,
+      "learning_rate": 9.986932816177258e-06,
+      "loss": 0.7898,
+      "step": 18
+    },
+    {
+      "epoch": 0.04859335038363171,
+      "grad_norm": 8.86563729131372,
+      "learning_rate": 9.985441300618966e-06,
+      "loss": 0.7867,
+      "step": 19
+    },
+    {
+      "epoch": 0.05115089514066496,
+      "grad_norm": 7.9582904003704655,
+      "learning_rate": 9.98386932334989e-06,
+      "loss": 0.7112,
+      "step": 20
+    },
+    {
+      "epoch": 0.05370843989769821,
+      "grad_norm": 7.46439913319433,
+      "learning_rate": 9.982216909740703e-06,
+      "loss": 0.7207,
+      "step": 21
+    },
+    {
+      "epoch": 0.056265984654731455,
+      "grad_norm": 7.502417267747896,
+      "learning_rate": 9.980484086460258e-06,
+      "loss": 0.7635,
+      "step": 22
+    },
+    {
+      "epoch": 0.058823529411764705,
+      "grad_norm": 8.077410576519545,
+      "learning_rate": 9.978670881475173e-06,
+      "loss": 0.7568,
+      "step": 23
+    },
+    {
+      "epoch": 0.061381074168797956,
+      "grad_norm": 8.530645418517887,
+      "learning_rate": 9.976777324049374e-06,
+      "loss": 0.8006,
+      "step": 24
+    },
+    {
+      "epoch": 0.0639386189258312,
+      "grad_norm": 7.884014258298183,
+      "learning_rate": 9.974803444743617e-06,
+      "loss": 0.741,
+      "step": 25
+    },
+    {
+      "epoch": 0.06649616368286446,
+      "grad_norm": 6.917821196530823,
+      "learning_rate": 9.972749275415005e-06,
+      "loss": 0.6668,
+      "step": 26
+    },
+    {
+      "epoch": 0.06905370843989769,
+      "grad_norm": 6.951997744847425,
+      "learning_rate": 9.970614849216465e-06,
+      "loss": 0.7619,
+      "step": 27
+    },
+    {
+      "epoch": 0.07161125319693094,
+      "grad_norm": 6.541689929654185,
+      "learning_rate": 9.96840020059622e-06,
+      "loss": 0.7667,
+      "step": 28
+    },
+    {
+      "epoch": 0.0741687979539642,
+      "grad_norm": 7.022380585505507,
+      "learning_rate": 9.966105365297226e-06,
+      "loss": 0.7176,
+      "step": 29
+    },
+    {
+      "epoch": 0.07672634271099744,
+      "grad_norm": 6.912790106638039,
+      "learning_rate": 9.963730380356599e-06,
+      "loss": 0.7199,
+      "step": 30
+    },
+    {
+      "epoch": 0.0792838874680307,
+      "grad_norm": 6.0140784283728275,
+      "learning_rate": 9.96127528410502e-06,
+      "loss": 0.7356,
+      "step": 31
+    },
+    {
+      "epoch": 0.08184143222506395,
+      "grad_norm": 6.50099127837641,
+      "learning_rate": 9.958740116166113e-06,
+      "loss": 0.6741,
+      "step": 32
+    },
+    {
+      "epoch": 0.08439897698209718,
+      "grad_norm": 6.196159967289502,
+      "learning_rate": 9.9561249174558e-06,
+      "loss": 0.6453,
+      "step": 33
+    },
+    {
+      "epoch": 0.08695652173913043,
+      "grad_norm": 7.115083432855918,
+      "learning_rate": 9.953429730181653e-06,
+      "loss": 0.6921,
+      "step": 34
+    },
+    {
+      "epoch": 0.08951406649616368,
+      "grad_norm": 8.373221612447116,
+      "learning_rate": 9.950654597842209e-06,
+      "loss": 0.6904,
+      "step": 35
+    },
+    {
+      "epoch": 0.09207161125319693,
+      "grad_norm": 7.708793853244428,
+      "learning_rate": 9.947799565226253e-06,
+      "loss": 0.7295,
+      "step": 36
+    },
+    {
+      "epoch": 0.09462915601023018,
+      "grad_norm": 8.81215615740888,
+      "learning_rate": 9.944864678412118e-06,
+      "loss": 0.6806,
+      "step": 37
+    },
+    {
+      "epoch": 0.09718670076726342,
+      "grad_norm": 7.456995220385941,
+      "learning_rate": 9.94184998476693e-06,
+      "loss": 0.6659,
+      "step": 38
+    },
+    {
+      "epoch": 0.09974424552429667,
+      "grad_norm": 7.636846126721769,
+      "learning_rate": 9.938755532945838e-06,
+      "loss": 0.7073,
+      "step": 39
+    },
+    {
+      "epoch": 0.10230179028132992,
+      "grad_norm": 8.018319876671674,
+      "learning_rate": 9.93558137289124e-06,
+      "loss": 0.8063,
+      "step": 40
+    },
+    {
+      "epoch": 0.10485933503836317,
+      "grad_norm": 6.429546682427428,
+      "learning_rate": 9.932327555831972e-06,
+      "loss": 0.5498,
+      "step": 41
+    },
+    {
+      "epoch": 0.10741687979539642,
+      "grad_norm": 6.106558471191083,
+      "learning_rate": 9.928994134282477e-06,
+      "loss": 0.6125,
+      "step": 42
+    },
+    {
+      "epoch": 0.10997442455242967,
+      "grad_norm": 7.220242425194578,
+      "learning_rate": 9.925581162041967e-06,
+      "loss": 0.6553,
+      "step": 43
+    },
+    {
+      "epoch": 0.11253196930946291,
+      "grad_norm": 7.406518370609504,
+      "learning_rate": 9.922088694193546e-06,
+      "loss": 0.7235,
+      "step": 44
+    },
+    {
+      "epoch": 0.11508951406649616,
+      "grad_norm": 6.232863127816167,
+      "learning_rate": 9.918516787103322e-06,
+      "loss": 0.6302,
+      "step": 45
+    },
+    {
+      "epoch": 0.11764705882352941,
+      "grad_norm": 6.960585716404723,
+      "learning_rate": 9.91486549841951e-06,
+      "loss": 0.6631,
+      "step": 46
+    },
+    {
+      "epoch": 0.12020460358056266,
+      "grad_norm": 7.014565196822424,
+      "learning_rate": 9.911134887071477e-06,
+      "loss": 0.6013,
+      "step": 47
+    },
+    {
+      "epoch": 0.12276214833759591,
+      "grad_norm": 6.747724971317866,
+      "learning_rate": 9.907325013268816e-06,
+      "loss": 0.7311,
+      "step": 48
+    },
+    {
+      "epoch": 0.12531969309462915,
+      "grad_norm": 6.343001355283702,
+      "learning_rate": 9.903435938500356e-06,
+      "loss": 0.5662,
+      "step": 49
+    },
+    {
+      "epoch": 0.1278772378516624,
+      "grad_norm": 6.68715844293194,
+      "learning_rate": 9.899467725533181e-06,
+      "loss": 0.6746,
+      "step": 50
+    },
+    {
+      "epoch": 0.13043478260869565,
+      "grad_norm": 6.670242029824649,
+      "learning_rate": 9.895420438411616e-06,
+      "loss": 0.5636,
+      "step": 51
+    },
+    {
+      "epoch": 0.1329923273657289,
+      "grad_norm": 7.6565238394007,
+      "learning_rate": 9.89129414245618e-06,
+      "loss": 0.6959,
+      "step": 52
+    },
+    {
+      "epoch": 0.13554987212276215,
+      "grad_norm": 6.322260761537018,
+      "learning_rate": 9.887088904262557e-06,
+      "loss": 0.6475,
+      "step": 53
+    },
+    {
+      "epoch": 0.13810741687979539,
+      "grad_norm": 7.424958009645482,
+      "learning_rate": 9.882804791700488e-06,
+      "loss": 0.7381,
+      "step": 54
+    },
+    {
+      "epoch": 0.14066496163682865,
+      "grad_norm": 6.484080185567617,
+      "learning_rate": 9.878441873912712e-06,
+      "loss": 0.5978,
+      "step": 55
+    },
+    {
+      "epoch": 0.1432225063938619,
+      "grad_norm": 7.904610864584768,
+      "learning_rate": 9.87400022131382e-06,
+      "loss": 0.7156,
+      "step": 56
+    },
+    {
+      "epoch": 0.14578005115089515,
+      "grad_norm": 6.091768577876145,
+      "learning_rate": 9.869479905589136e-06,
+      "loss": 0.5674,
+      "step": 57
+    },
+    {
+      "epoch": 0.1483375959079284,
+      "grad_norm": 6.338206394229405,
+      "learning_rate": 9.864880999693551e-06,
+      "loss": 0.5511,
+      "step": 58
+    },
+    {
+      "epoch": 0.15089514066496162,
+      "grad_norm": 7.393436860670507,
+      "learning_rate": 9.860203577850353e-06,
+      "loss": 0.6305,
+      "step": 59
+    },
+    {
+      "epoch": 0.1534526854219949,
+      "grad_norm": 6.520098900315275,
+      "learning_rate": 9.855447715550024e-06,
+      "loss": 0.6191,
+      "step": 60
+    },
+    {
+      "epoch": 0.15601023017902813,
+      "grad_norm": 6.415362809418058,
+      "learning_rate": 9.850613489549018e-06,
+      "loss": 0.6296,
+      "step": 61
+    },
+    {
+      "epoch": 0.1585677749360614,
+      "grad_norm": 7.9580764103093635,
+      "learning_rate": 9.845700977868536e-06,
+      "loss": 0.6384,
+      "step": 62
+    },
+    {
+      "epoch": 0.16112531969309463,
+      "grad_norm": 7.1082899489573315,
+      "learning_rate": 9.840710259793251e-06,
+      "loss": 0.6021,
+      "step": 63
+    },
+    {
+      "epoch": 0.1636828644501279,
+      "grad_norm": 6.386504712691233,
+      "learning_rate": 9.835641415870038e-06,
+      "loss": 0.6873,
+      "step": 64
+    },
+    {
+      "epoch": 0.16624040920716113,
+      "grad_norm": 7.4797007816006555,
+      "learning_rate": 9.830494527906671e-06,
+      "loss": 0.5648,
+      "step": 65
+    },
+    {
+      "epoch": 0.16879795396419436,
+      "grad_norm": 6.748679766187366,
+      "learning_rate": 9.825269678970502e-06,
+      "loss": 0.566,
+      "step": 66
+    },
+    {
+      "epoch": 0.17135549872122763,
+      "grad_norm": 6.56170099270124,
+      "learning_rate": 9.819966953387122e-06,
+      "loss": 0.6398,
+      "step": 67
+    },
+    {
+      "epoch": 0.17391304347826086,
+      "grad_norm": 5.941005918686002,
+      "learning_rate": 9.814586436738998e-06,
+      "loss": 0.6658,
+      "step": 68
+    },
+    {
+      "epoch": 0.17647058823529413,
+      "grad_norm": 5.942719440981736,
+      "learning_rate": 9.809128215864096e-06,
+      "loss": 0.5971,
+      "step": 69
+    },
+    {
+      "epoch": 0.17902813299232737,
+      "grad_norm": 6.4968806279365925,
+      "learning_rate": 9.803592378854476e-06,
+      "loss": 0.6047,
+      "step": 70
+    },
+    {
+      "epoch": 0.1815856777493606,
+      "grad_norm": 5.577555703482892,
+      "learning_rate": 9.797979015054868e-06,
+      "loss": 0.5534,
+      "step": 71
+    },
+    {
+      "epoch": 0.18414322250639387,
+      "grad_norm": 6.262423254220968,
+      "learning_rate": 9.792288215061237e-06,
+      "loss": 0.5755,
+      "step": 72
+    },
+    {
+      "epoch": 0.1867007672634271,
+      "grad_norm": 7.17658874763877,
+      "learning_rate": 9.786520070719313e-06,
+      "loss": 0.5511,
+      "step": 73
+    },
+    {
+      "epoch": 0.18925831202046037,
+      "grad_norm": 5.532267867303149,
+      "learning_rate": 9.780674675123113e-06,
+      "loss": 0.4965,
+      "step": 74
+    },
+    {
+      "epoch": 0.1918158567774936,
+      "grad_norm": 7.040268722404183,
+      "learning_rate": 9.77475212261344e-06,
+      "loss": 0.6644,
+      "step": 75
+    },
+    {
+      "epoch": 0.19437340153452684,
+      "grad_norm": 6.774497918030386,
+      "learning_rate": 9.768752508776358e-06,
+      "loss": 0.491,
+      "step": 76
+    },
+    {
+      "epoch": 0.1969309462915601,
+      "grad_norm": 5.776934238941505,
+      "learning_rate": 9.762675930441647e-06,
+      "loss": 0.4861,
+      "step": 77
+    },
+    {
+      "epoch": 0.19948849104859334,
+      "grad_norm": 6.47046232974907,
+      "learning_rate": 9.756522485681247e-06,
+      "loss": 0.6273,
+      "step": 78
+    },
+    {
+      "epoch": 0.2020460358056266,
+      "grad_norm": 6.712438744552012,
+      "learning_rate": 9.750292273807666e-06,
+      "loss": 0.7368,
+      "step": 79
+    },
+    {
+      "epoch": 0.20460358056265984,
+      "grad_norm": 8.615585919768426,
+      "learning_rate": 9.743985395372387e-06,
+      "loss": 0.6335,
+      "step": 80
+    },
+    {
+      "epoch": 0.2071611253196931,
+      "grad_norm": 8.071799145378323,
+      "learning_rate": 9.737601952164238e-06,
+      "loss": 0.6612,
+      "step": 81
+    },
+    {
+      "epoch": 0.20971867007672634,
+      "grad_norm": 6.823667160150434,
+      "learning_rate": 9.73114204720775e-06,
+      "loss": 0.5456,
+      "step": 82
+    },
+    {
+      "epoch": 0.21227621483375958,
+      "grad_norm": 5.888063194216176,
+      "learning_rate": 9.724605784761501e-06,
+      "loss": 0.5959,
+      "step": 83
+    },
+    {
+      "epoch": 0.21483375959079284,
+      "grad_norm": 7.210449720464657,
+      "learning_rate": 9.717993270316421e-06,
+      "loss": 0.5919,
+      "step": 84
+    },
+    {
+      "epoch": 0.21739130434782608,
+      "grad_norm": 7.06901541191221,
+      "learning_rate": 9.711304610594104e-06,
+      "loss": 0.5882,
+      "step": 85
+    },
+    {
+      "epoch": 0.21994884910485935,
+      "grad_norm": 6.374122555190542,
+      "learning_rate": 9.704539913545073e-06,
+      "loss": 0.6133,
+      "step": 86
+    },
+    {
+      "epoch": 0.22250639386189258,
+      "grad_norm": 6.630925756972144,
+      "learning_rate": 9.697699288347043e-06,
+      "loss": 0.5664,
+      "step": 87
+    },
+    {
+      "epoch": 0.22506393861892582,
+      "grad_norm": 5.966658672241774,
+      "learning_rate": 9.690782845403164e-06,
+      "loss": 0.6244,
+      "step": 88
+    },
+    {
+      "epoch": 0.22762148337595908,
+      "grad_norm": 5.5770246123900264,
+      "learning_rate": 9.683790696340229e-06,
+      "loss": 0.5334,
+      "step": 89
+    },
+    {
+      "epoch": 0.23017902813299232,
+      "grad_norm": 6.522499319244262,
+      "learning_rate": 9.676722954006878e-06,
+      "loss": 0.6739,
+      "step": 90
+    },
+    {
+      "epoch": 0.23273657289002558,
+      "grad_norm": 6.4235805218363025,
+      "learning_rate": 9.669579732471779e-06,
+      "loss": 0.6595,
+      "step": 91
+    },
+    {
+      "epoch": 0.23529411764705882,
+      "grad_norm": 6.3431696131235284,
+      "learning_rate": 9.66236114702178e-06,
+      "loss": 0.6023,
+      "step": 92
+    },
+    {
+      "epoch": 0.23785166240409208,
+      "grad_norm": 6.429883886786425,
+      "learning_rate": 9.655067314160058e-06,
+      "loss": 0.5986,
+      "step": 93
+    },
+    {
+      "epoch": 0.24040920716112532,
+      "grad_norm": 6.701368521252926,
+      "learning_rate": 9.647698351604227e-06,
+      "loss": 0.6569,
+      "step": 94
+    },
+    {
+      "epoch": 0.24296675191815856,
+      "grad_norm": 6.626338500498468,
+      "learning_rate": 9.640254378284447e-06,
+      "loss": 0.5552,
+      "step": 95
+    },
+    {
+      "epoch": 0.24552429667519182,
+      "grad_norm": 6.872036104522696,
+      "learning_rate": 9.632735514341508e-06,
+      "loss": 0.5384,
+      "step": 96
+    },
+    {
+      "epoch": 0.24808184143222506,
+      "grad_norm": 7.976561364452108,
+      "learning_rate": 9.625141881124874e-06,
+      "loss": 0.6225,
+      "step": 97
+    },
+    {
+      "epoch": 0.2506393861892583,
+      "grad_norm": 6.979721775572128,
+      "learning_rate": 9.617473601190743e-06,
+      "loss": 0.5937,
+      "step": 98
+    },
+    {
+      "epoch": 0.2531969309462916,
+      "grad_norm": 7.519173375318072,
+      "learning_rate": 9.609730798300056e-06,
+      "loss": 0.5673,
+      "step": 99
+    },
+    {
+      "epoch": 0.2557544757033248,
+      "grad_norm": 6.501376080036557,
+      "learning_rate": 9.601913597416513e-06,
+      "loss": 0.6167,
+      "step": 100
+    },
+    {
+      "epoch": 0.25831202046035806,
+      "grad_norm": 7.232025532728033,
+      "learning_rate": 9.594022124704541e-06,
+      "loss": 0.6528,
+      "step": 101
+    },
+    {
+      "epoch": 0.2608695652173913,
+      "grad_norm": 5.821804513422366,
+      "learning_rate": 9.586056507527266e-06,
+      "loss": 0.5839,
+      "step": 102
+    },
+    {
+      "epoch": 0.26342710997442453,
+      "grad_norm": 7.447246210872055,
+      "learning_rate": 9.578016874444459e-06,
+      "loss": 0.5425,
+      "step": 103
+    },
+    {
+      "epoch": 0.2659846547314578,
+      "grad_norm": 5.023721202880345,
+      "learning_rate": 9.569903355210457e-06,
+      "loss": 0.4649,
+      "step": 104
+    },
+    {
+      "epoch": 0.26854219948849106,
+      "grad_norm": 5.3855220410063165,
+      "learning_rate": 9.561716080772072e-06,
+      "loss": 0.5362,
+      "step": 105
+    },
+    {
+      "epoch": 0.2710997442455243,
+      "grad_norm": 5.755012965635693,
+      "learning_rate": 9.55345518326647e-06,
+      "loss": 0.637,
+      "step": 106
+    },
+    {
+      "epoch": 0.27365728900255754,
+      "grad_norm": 6.317278618265475,
+      "learning_rate": 9.545120796019056e-06,
+      "loss": 0.6073,
+      "step": 107
+    },
+    {
+      "epoch": 0.27621483375959077,
+      "grad_norm": 6.556750652969424,
+      "learning_rate": 9.5367130535413e-06,
+      "loss": 0.6106,
+      "step": 108
+    },
+    {
+      "epoch": 0.27877237851662406,
+      "grad_norm": 7.234204791297877,
+      "learning_rate": 9.528232091528578e-06,
+      "loss": 0.5537,
+      "step": 109
+    },
+    {
+      "epoch": 0.2813299232736573,
+      "grad_norm": 6.44863975829238,
+      "learning_rate": 9.519678046857987e-06,
+      "loss": 0.6654,
+      "step": 110
+    },
+    {
+      "epoch": 0.28388746803069054,
+      "grad_norm": 5.882964477582141,
+      "learning_rate": 9.511051057586125e-06,
+      "loss": 0.5723,
+      "step": 111
+    },
+    {
+      "epoch": 0.2864450127877238,
+      "grad_norm": 5.682881601819309,
+      "learning_rate": 9.502351262946865e-06,
+      "loss": 0.5325,
+      "step": 112
+    },
+    {
+      "epoch": 0.289002557544757,
+      "grad_norm": 5.907234833224297,
+      "learning_rate": 9.493578803349117e-06,
+      "loss": 0.6238,
+      "step": 113
+    },
+    {
+      "epoch": 0.2915601023017903,
+      "grad_norm": 6.539157912120955,
+      "learning_rate": 9.48473382037455e-06,
+      "loss": 0.6228,
+      "step": 114
+    },
+    {
+      "epoch": 0.29411764705882354,
+      "grad_norm": 6.263387218751874,
+      "learning_rate": 9.475816456775313e-06,
+      "loss": 0.5954,
+      "step": 115
+    },
+    {
+      "epoch": 0.2966751918158568,
+      "grad_norm": 5.93110985413684,
+      "learning_rate": 9.466826856471728e-06,
+      "loss": 0.6008,
+      "step": 116
+    },
+    {
+      "epoch": 0.29923273657289,
+      "grad_norm": 6.326093043591936,
+      "learning_rate": 9.457765164549979e-06,
+      "loss": 0.5834,
+      "step": 117
+    },
+    {
+      "epoch": 0.30179028132992325,
+      "grad_norm": 8.333855109469555,
+      "learning_rate": 9.448631527259749e-06,
+      "loss": 0.7357,
+      "step": 118
+    },
+    {
+      "epoch": 0.30434782608695654,
+      "grad_norm": 7.352912958641833,
+      "learning_rate": 9.439426092011877e-06,
+      "loss": 0.5621,
+      "step": 119
+    },
+    {
+      "epoch": 0.3069053708439898,
+      "grad_norm": 6.812130929920315,
+      "learning_rate": 9.430149007375974e-06,
+      "loss": 0.6281,
+      "step": 120
+    },
+    {
+      "epoch": 0.309462915601023,
+      "grad_norm": 6.244862089177942,
+      "learning_rate": 9.42080042307802e-06,
+      "loss": 0.6083,
+      "step": 121
+    },
+    {
+      "epoch": 0.31202046035805625,
+      "grad_norm": 5.91824427054237,
+      "learning_rate": 9.411380489997962e-06,
+      "loss": 0.5141,
+      "step": 122
+    },
+    {
+      "epoch": 0.3145780051150895,
+      "grad_norm": 5.9138584985365075,
+      "learning_rate": 9.401889360167256e-06,
+      "loss": 0.5525,
+      "step": 123
+    },
+    {
+      "epoch": 0.3171355498721228,
+      "grad_norm": 6.264682424037789,
+      "learning_rate": 9.392327186766434e-06,
+      "loss": 0.5049,
+      "step": 124
+    },
+    {
+      "epoch": 0.319693094629156,
+      "grad_norm": 5.838854049045799,
+      "learning_rate": 9.382694124122624e-06,
+      "loss": 0.5835,
+      "step": 125
+    },
+    {
+      "epoch": 0.32225063938618925,
+      "grad_norm": 6.39551091760669,
+      "learning_rate": 9.372990327707057e-06,
+      "loss": 0.5132,
+      "step": 126
+    },
+    {
+      "epoch": 0.3248081841432225,
+      "grad_norm": 6.716488574379217,
+      "learning_rate": 9.36321595413256e-06,
+      "loss": 0.5372,
+      "step": 127
+    },
+    {
+      "epoch": 0.3273657289002558,
+      "grad_norm": 5.742568960781894,
+      "learning_rate": 9.353371161151032e-06,
+      "loss": 0.6203,
+      "step": 128
+    },
+    {
+      "epoch": 0.329923273657289,
+      "grad_norm": 6.772868273583815,
+      "learning_rate": 9.34345610765089e-06,
+      "loss": 0.4926,
+      "step": 129
+    },
+    {
+      "epoch": 0.33248081841432225,
+      "grad_norm": 7.185477633140159,
+      "learning_rate": 9.333470953654513e-06,
+      "loss": 0.6842,
+      "step": 130
+    },
+    {
+      "epoch": 0.3350383631713555,
+      "grad_norm": 6.296716433927121,
+      "learning_rate": 9.32341586031565e-06,
+      "loss": 0.5163,
+      "step": 131
+    },
+    {
+      "epoch": 0.3375959079283887,
+      "grad_norm": 8.081152679238958,
+      "learning_rate": 9.31329098991683e-06,
+      "loss": 0.7902,
+      "step": 132
+    },
+    {
+      "epoch": 0.340153452685422,
+      "grad_norm": 6.861874747525335,
+      "learning_rate": 9.303096505866734e-06,
+      "loss": 0.5645,
+      "step": 133
+    },
+    {
+      "epoch": 0.34271099744245526,
+      "grad_norm": 6.171268132782227,
+      "learning_rate": 9.292832572697566e-06,
+      "loss": 0.6114,
+      "step": 134
+    },
+    {
+      "epoch": 0.3452685421994885,
+      "grad_norm": 6.552747924757749,
+      "learning_rate": 9.282499356062385e-06,
+      "loss": 0.5954,
+      "step": 135
+    },
+    {
+      "epoch": 0.34782608695652173,
+      "grad_norm": 6.847724657541522,
+      "learning_rate": 9.272097022732444e-06,
+      "loss": 0.6485,
+      "step": 136
+    },
+    {
+      "epoch": 0.35038363171355497,
+      "grad_norm": 6.507209809740027,
+      "learning_rate": 9.261625740594494e-06,
+      "loss": 0.6159,
+      "step": 137
+    },
+    {
+      "epoch": 0.35294117647058826,
+      "grad_norm": 7.252605300424873,
+      "learning_rate": 9.251085678648072e-06,
+      "loss": 0.5576,
+      "step": 138
+    },
+    {
+      "epoch": 0.3554987212276215,
+      "grad_norm": 7.119115054572448,
+      "learning_rate": 9.240477007002777e-06,
+      "loss": 0.7135,
+      "step": 139
+    },
+    {
+      "epoch": 0.35805626598465473,
+      "grad_norm": 7.349847252060697,
+      "learning_rate": 9.22979989687552e-06,
+      "loss": 0.6444,
+      "step": 140
+    },
+    {
+      "epoch": 0.36061381074168797,
+      "grad_norm": 5.765325529890724,
+      "learning_rate": 9.219054520587766e-06,
+      "loss": 0.4233,
+      "step": 141
+    },
+    {
+      "epoch": 0.3631713554987212,
+      "grad_norm": 5.112678313504443,
+      "learning_rate": 9.208241051562753e-06,
+      "loss": 0.5447,
+      "step": 142
+    },
+    {
+      "epoch": 0.3657289002557545,
+      "grad_norm": 7.144208807574378,
+      "learning_rate": 9.197359664322684e-06,
+      "loss": 0.5891,
+      "step": 143
+    },
+    {
+      "epoch": 0.36828644501278773,
+      "grad_norm": 6.278104080681577,
+      "learning_rate": 9.186410534485924e-06,
+      "loss": 0.5701,
+      "step": 144
+    },
+    {
+      "epoch": 0.37084398976982097,
+      "grad_norm": 6.887145963626663,
+      "learning_rate": 9.175393838764153e-06,
+      "loss": 0.5502,
+      "step": 145
+    },
+    {
+      "epoch": 0.3734015345268542,
+      "grad_norm": 6.023815922626531,
+      "learning_rate": 9.164309754959523e-06,
+      "loss": 0.5286,
+      "step": 146
+    },
+    {
+      "epoch": 0.37595907928388744,
+      "grad_norm": 5.517318940161725,
+      "learning_rate": 9.153158461961782e-06,
+      "loss": 0.4433,
+      "step": 147
+    },
+    {
+      "epoch": 0.37851662404092073,
+      "grad_norm": 6.019318968473694,
+      "learning_rate": 9.14194013974539e-06,
+      "loss": 0.5065,
+      "step": 148
+    },
+    {
+      "epoch": 0.38107416879795397,
+      "grad_norm": 5.6444749821227385,
+      "learning_rate": 9.130654969366619e-06,
+      "loss": 0.501,
+      "step": 149
+    },
+    {
+      "epoch": 0.3836317135549872,
+      "grad_norm": 6.559732586020477,
+      "learning_rate": 9.11930313296062e-06,
+      "loss": 0.6101,
+      "step": 150
+    },
+    {
+      "epoch": 0.38618925831202044,
+      "grad_norm": 6.2279396393587705,
+      "learning_rate": 9.107884813738492e-06,
+      "loss": 0.5938,
+      "step": 151
+    },
+    {
+      "epoch": 0.3887468030690537,
+      "grad_norm": 7.319750378815782,
+      "learning_rate": 9.096400195984322e-06,
+      "loss": 0.4252,
+      "step": 152
+    },
+    {
+      "epoch": 0.391304347826087,
+      "grad_norm": 7.088600256578154,
+      "learning_rate": 9.08484946505221e-06,
+      "loss": 0.5793,
+      "step": 153
+    },
+    {
+      "epoch": 0.3938618925831202,
+      "grad_norm": 6.573409174775933,
+      "learning_rate": 9.073232807363283e-06,
+      "loss": 0.5026,
+      "step": 154
+    },
+    {
+      "epoch": 0.39641943734015345,
+      "grad_norm": 7.980317521312881,
+      "learning_rate": 9.061550410402677e-06,
+      "loss": 0.6736,
+      "step": 155
+    },
+    {
+      "epoch": 0.3989769820971867,
+      "grad_norm": 6.87810163307716,
+      "learning_rate": 9.049802462716521e-06,
+      "loss": 0.493,
+      "step": 156
+    },
+    {
+      "epoch": 0.40153452685422,
+      "grad_norm": 7.407535434490462,
+      "learning_rate": 9.037989153908882e-06,
+      "loss": 0.5762,
+      "step": 157
+    },
+    {
+      "epoch": 0.4040920716112532,
+      "grad_norm": 6.310545687817972,
+      "learning_rate": 9.026110674638722e-06,
+      "loss": 0.5802,
+      "step": 158
+    },
+    {
+      "epoch": 0.40664961636828645,
+      "grad_norm": 6.538278944739297,
+      "learning_rate": 9.0141672166168e-06,
+      "loss": 0.4665,
+      "step": 159
+    },
+    {
+      "epoch": 0.4092071611253197,
+      "grad_norm": 6.25525186329276,
+      "learning_rate": 9.002158972602599e-06,
+      "loss": 0.65,
+      "step": 160
+    },
+    {
+      "epoch": 0.4117647058823529,
+      "grad_norm": 7.134439997933502,
+      "learning_rate": 8.990086136401199e-06,
+      "loss": 0.6436,
+      "step": 161
+    },
+    {
+      "epoch": 0.4143222506393862,
+      "grad_norm": 6.907730334313879,
+      "learning_rate": 8.977948902860154e-06,
+      "loss": 0.6688,
+      "step": 162
+    },
+    {
+      "epoch": 0.41687979539641945,
+      "grad_norm": 5.60949076779962,
+      "learning_rate": 8.965747467866355e-06,
+      "loss": 0.4263,
+      "step": 163
+    },
+    {
+      "epoch": 0.4194373401534527,
+      "grad_norm": 7.153866044006984,
+      "learning_rate": 8.953482028342853e-06,
+      "loss": 0.707,
+      "step": 164
+    },
+    {
+      "epoch": 0.4219948849104859,
+      "grad_norm": 5.749873195369449,
+      "learning_rate": 8.9411527822457e-06,
+      "loss": 0.5522,
+      "step": 165
+    },
+    {
+      "epoch": 0.42455242966751916,
+      "grad_norm": 6.450113679037296,
+      "learning_rate": 8.92875992856073e-06,
+      "loss": 0.491,
+      "step": 166
+    },
+    {
+      "epoch": 0.42710997442455245,
+      "grad_norm": 7.10664387732353,
+      "learning_rate": 8.916303667300373e-06,
+      "loss": 0.5526,
+      "step": 167
+    },
+    {
+      "epoch": 0.4296675191815857,
+      "grad_norm": 5.908129226489756,
+      "learning_rate": 8.903784199500412e-06,
+      "loss": 0.4589,
+      "step": 168
+    },
+    {
+      "epoch": 0.4322250639386189,
+      "grad_norm": 6.036965674381112,
+      "learning_rate": 8.89120172721674e-06,
+      "loss": 0.5393,
+      "step": 169
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "grad_norm": 6.557422143271899,
+      "learning_rate": 8.8785564535221e-06,
+      "loss": 0.4947,
+      "step": 170
+    },
+    {
+      "epoch": 0.4373401534526854,
+      "grad_norm": 7.091654604179938,
+      "learning_rate": 8.86584858250281e-06,
+      "loss": 0.5744,
+      "step": 171
+    },
+    {
+      "epoch": 0.4398976982097187,
+      "grad_norm": 6.839942726126082,
+      "learning_rate": 8.853078319255466e-06,
+      "loss": 0.5621,
+      "step": 172
+    },
+    {
+      "epoch": 0.4424552429667519,
+      "grad_norm": 6.705305452145543,
+      "learning_rate": 8.840245869883635e-06,
+      "loss": 0.6277,
+      "step": 173
+    },
+    {
+      "epoch": 0.44501278772378516,
+      "grad_norm": 6.410914353629197,
+      "learning_rate": 8.827351441494525e-06,
+      "loss": 0.5795,
+      "step": 174
+    },
+    {
+      "epoch": 0.4475703324808184,
+      "grad_norm": 6.178508928681474,
+      "learning_rate": 8.814395242195642e-06,
+      "loss": 0.5039,
+      "step": 175
+    },
+    {
+      "epoch": 0.45012787723785164,
+      "grad_norm": 5.283454576226612,
+      "learning_rate": 8.80137748109144e-06,
+      "loss": 0.4565,
+      "step": 176
+    },
+    {
+      "epoch": 0.45268542199488493,
+      "grad_norm": 5.78397157032685,
+      "learning_rate": 8.78829836827993e-06,
+      "loss": 0.5435,
+      "step": 177
+    },
+    {
+      "epoch": 0.45524296675191817,
+      "grad_norm": 5.014711664858047,
+      "learning_rate": 8.77515811484931e-06,
+      "loss": 0.4744,
+      "step": 178
+    },
+    {
+      "epoch": 0.4578005115089514,
+      "grad_norm": 5.673473816390766,
+      "learning_rate": 8.761956932874539e-06,
+      "loss": 0.4794,
+      "step": 179
+    },
+    {
+      "epoch": 0.46035805626598464,
+      "grad_norm": 6.548843702355434,
+      "learning_rate": 8.748695035413925e-06,
+      "loss": 0.5124,
+      "step": 180
+    },
+    {
+      "epoch": 0.4629156010230179,
+      "grad_norm": 5.294743068866496,
+      "learning_rate": 8.735372636505681e-06,
+      "loss": 0.4964,
+      "step": 181
+    },
+    {
+      "epoch": 0.46547314578005117,
+      "grad_norm": 5.307479046129796,
+      "learning_rate": 8.72198995116448e-06,
+      "loss": 0.4848,
+      "step": 182
+    },
+    {
+      "epoch": 0.4680306905370844,
+      "grad_norm": 5.862290813329295,
+      "learning_rate": 8.708547195377968e-06,
+      "loss": 0.6168,
+      "step": 183
+    },
+    {
+      "epoch": 0.47058823529411764,
+      "grad_norm": 6.660472406940894,
+      "learning_rate": 8.695044586103297e-06,
+      "loss": 0.6317,
+      "step": 184
+    },
+    {
+      "epoch": 0.4731457800511509,
+      "grad_norm": 6.239173715990654,
+      "learning_rate": 8.68148234126361e-06,
+      "loss": 0.5712,
+      "step": 185
+    },
+    {
+      "epoch": 0.47570332480818417,
+      "grad_norm": 6.855169313192307,
+      "learning_rate": 8.667860679744529e-06,
+      "loss": 0.4569,
+      "step": 186
+    },
+    {
+      "epoch": 0.4782608695652174,
+      "grad_norm": 6.098286103768463,
+      "learning_rate": 8.65417982139062e-06,
+      "loss": 0.5377,
+      "step": 187
+    },
+    {
+      "epoch": 0.48081841432225064,
+      "grad_norm": 6.687927077404218,
+      "learning_rate": 8.640439987001855e-06,
+      "loss": 0.5466,
+      "step": 188
+    },
+    {
+      "epoch": 0.4833759590792839,
+      "grad_norm": 5.06986215439764,
+      "learning_rate": 8.626641398330027e-06,
+      "loss": 0.4597,
+      "step": 189
+    },
+    {
+      "epoch": 0.4859335038363171,
+      "grad_norm": 6.615207228778788,
+      "learning_rate": 8.612784278075195e-06,
+      "loss": 0.6486,
+      "step": 190
+    },
+    {
+      "epoch": 0.4884910485933504,
+      "grad_norm": 8.206032710195597,
+      "learning_rate": 8.598868849882074e-06,
+      "loss": 0.53,
+      "step": 191
+    },
+    {
+      "epoch": 0.49104859335038364,
+      "grad_norm": 6.512201203410748,
+      "learning_rate": 8.58489533833643e-06,
+      "loss": 0.5075,
+      "step": 192
+    },
+    {
+      "epoch": 0.4936061381074169,
+      "grad_norm": 6.623708653660542,
+      "learning_rate": 8.570863968961456e-06,
+      "loss": 0.4697,
+      "step": 193
+    },
+    {
+      "epoch": 0.4961636828644501,
+      "grad_norm": 4.93093459028815,
+      "learning_rate": 8.556774968214134e-06,
+      "loss": 0.5169,
+      "step": 194
+    },
+    {
+      "epoch": 0.49872122762148335,
+      "grad_norm": 5.787452319450779,
+      "learning_rate": 8.542628563481577e-06,
+      "loss": 0.5482,
+      "step": 195
+    },
+    {
+      "epoch": 0.5012787723785166,
+      "grad_norm": 5.88850708880366,
+      "learning_rate": 8.52842498307736e-06,
+      "loss": 0.6134,
+      "step": 196
+    },
+    {
+      "epoch": 0.5038363171355499,
+      "grad_norm": 6.696557687225988,
+      "learning_rate": 8.514164456237835e-06,
+      "loss": 0.6447,
+      "step": 197
+    },
+    {
+      "epoch": 0.5063938618925832,
+      "grad_norm": 7.195386414110228,
+      "learning_rate": 8.499847213118431e-06,
+      "loss": 0.5117,
+      "step": 198
+    },
+    {
+      "epoch": 0.5089514066496164,
+      "grad_norm": 5.825208905742397,
+      "learning_rate": 8.485473484789944e-06,
+      "loss": 0.5152,
+      "step": 199
+    },
+    {
+      "epoch": 0.5115089514066496,
+      "grad_norm": 6.118462869888847,
+      "learning_rate": 8.471043503234796e-06,
+      "loss": 0.6536,
+      "step": 200
+    },
+    {
+      "epoch": 0.5140664961636828,
+      "grad_norm": 7.551332041886624,
+      "learning_rate": 8.45655750134331e-06,
+      "loss": 0.6084,
+      "step": 201
+    },
+    {
+      "epoch": 0.5166240409207161,
+      "grad_norm": 5.225118027592022,
+      "learning_rate": 8.442015712909926e-06,
+      "loss": 0.4555,
+      "step": 202
+    },
+    {
+      "epoch": 0.5191815856777494,
+      "grad_norm": 5.3592959184851265,
+      "learning_rate": 8.427418372629456e-06,
+      "loss": 0.4821,
+      "step": 203
+    },
+    {
+      "epoch": 0.5217391304347826,
+      "grad_norm": 6.212770160868918,
+      "learning_rate": 8.412765716093273e-06,
+      "loss": 0.5149,
+      "step": 204
+    },
+    {
+      "epoch": 0.5242966751918159,
+      "grad_norm": 5.519259201547804,
+      "learning_rate": 8.398057979785515e-06,
+      "loss": 0.4876,
+      "step": 205
+    },
+    {
+      "epoch": 0.5268542199488491,
+      "grad_norm": 6.847844140759948,
+      "learning_rate": 8.383295401079284e-06,
+      "loss": 0.5245,
+      "step": 206
+    },
+    {
+      "epoch": 0.5294117647058824,
+      "grad_norm": 6.473882379231715,
+      "learning_rate": 8.368478218232787e-06,
+      "loss": 0.5319,
+      "step": 207
+    },
+    {
+      "epoch": 0.5319693094629157,
+      "grad_norm": 4.995240865011453,
+      "learning_rate": 8.353606670385514e-06,
+      "loss": 0.4201,
+      "step": 208
+    },
+    {
+      "epoch": 0.5345268542199488,
+      "grad_norm": 5.272197527549254,
+      "learning_rate": 8.338680997554372e-06,
+      "loss": 0.4832,
+      "step": 209
+    },
+    {
+      "epoch": 0.5370843989769821,
+      "grad_norm": 5.585738561949535,
+      "learning_rate": 8.3237014406298e-06,
+      "loss": 0.4929,
+      "step": 210
+    },
+    {
+      "epoch": 0.5396419437340153,
+      "grad_norm": 6.239332915949274,
+      "learning_rate": 8.308668241371897e-06,
+      "loss": 0.4171,
+      "step": 211
+    },
+    {
+      "epoch": 0.5421994884910486,
+      "grad_norm": 5.322513595323884,
+      "learning_rate": 8.293581642406517e-06,
+      "loss": 0.4073,
+      "step": 212
+    },
+    {
+      "epoch": 0.5447570332480819,
+      "grad_norm": 7.46575629890418,
+      "learning_rate": 8.278441887221338e-06,
+      "loss": 0.6626,
+      "step": 213
+    },
+    {
+      "epoch": 0.5473145780051151,
+      "grad_norm": 6.052661000824651,
+      "learning_rate": 8.263249220161957e-06,
+      "loss": 0.5068,
+      "step": 214
+    },
+    {
+      "epoch": 0.5498721227621484,
+      "grad_norm": 6.708569886061961,
+      "learning_rate": 8.248003886427927e-06,
+      "loss": 0.4966,
+      "step": 215
+    },
+    {
+      "epoch": 0.5524296675191815,
+      "grad_norm": 5.68545952396897,
+      "learning_rate": 8.232706132068806e-06,
+      "loss": 0.3861,
+      "step": 216
+    },
+    {
+      "epoch": 0.5549872122762148,
+      "grad_norm": 5.421452942064916,
+      "learning_rate": 8.217356203980187e-06,
+      "loss": 0.3885,
+      "step": 217
+    },
+    {
+      "epoch": 0.5575447570332481,
+      "grad_norm": 6.028892220556533,
+      "learning_rate": 8.201954349899712e-06,
+      "loss": 0.5848,
+      "step": 218
+    },
+    {
+      "epoch": 0.5601023017902813,
+      "grad_norm": 5.85865954700368,
+      "learning_rate": 8.186500818403076e-06,
+      "loss": 0.4014,
+      "step": 219
+    },
+    {
+      "epoch": 0.5626598465473146,
+      "grad_norm": 7.224336653003557,
+      "learning_rate": 8.17099585890001e-06,
+      "loss": 0.6191,
+      "step": 220
+    },
+    {
+      "epoch": 0.5652173913043478,
+      "grad_norm": 6.30772616724702,
+      "learning_rate": 8.155439721630265e-06,
+      "loss": 0.4756,
+      "step": 221
+    },
+    {
+      "epoch": 0.5677749360613811,
+      "grad_norm": 6.236289350658551,
+      "learning_rate": 8.139832657659557e-06,
+      "loss": 0.4964,
+      "step": 222
+    },
+    {
+      "epoch": 0.5703324808184144,
+      "grad_norm": 6.209791638515158,
+      "learning_rate": 8.124174918875532e-06,
+      "loss": 0.5958,
+      "step": 223
+    },
+    {
+      "epoch": 0.5728900255754475,
+      "grad_norm": 6.377139101398714,
+      "learning_rate": 8.108466757983695e-06,
+      "loss": 0.3906,
+      "step": 224
+    },
+    {
+      "epoch": 0.5754475703324808,
+      "grad_norm": 5.907378705805276,
+      "learning_rate": 8.092708428503324e-06,
+      "loss": 0.5376,
+      "step": 225
+    },
+    {
+      "epoch": 0.578005115089514,
+      "grad_norm": 6.743429798340147,
+      "learning_rate": 8.076900184763394e-06,
+      "loss": 0.4802,
+      "step": 226
+    },
+    {
+      "epoch": 0.5805626598465473,
+      "grad_norm": 5.376964104341389,
+      "learning_rate": 8.061042281898453e-06,
+      "loss": 0.4509,
+      "step": 227
+    },
+    {
+      "epoch": 0.5831202046035806,
+      "grad_norm": 6.511105613676549,
+      "learning_rate": 8.04513497584452e-06,
+      "loss": 0.4214,
+      "step": 228
+    },
+    {
+      "epoch": 0.5856777493606138,
+      "grad_norm": 6.852114152108356,
+      "learning_rate": 8.02917852333495e-06,
+      "loss": 0.6038,
+      "step": 229
+    },
+    {
+      "epoch": 0.5882352941176471,
+      "grad_norm": 5.572255565814968,
+      "learning_rate": 8.013173181896283e-06,
+      "loss": 0.5224,
+      "step": 230
+    },
+    {
+      "epoch": 0.5907928388746803,
+      "grad_norm": 7.0760399259778435,
+      "learning_rate": 7.9971192098441e-06,
+      "loss": 0.4502,
+      "step": 231
+    },
+    {
+      "epoch": 0.5933503836317136,
+      "grad_norm": 5.207560218712082,
+      "learning_rate": 7.981016866278843e-06,
+      "loss": 0.4027,
+      "step": 232
+    },
+    {
+      "epoch": 0.5959079283887468,
+      "grad_norm": 5.796635616276233,
+      "learning_rate": 7.964866411081645e-06,
+      "loss": 0.5675,
+      "step": 233
+    },
+    {
+      "epoch": 0.59846547314578,
+      "grad_norm": 5.708164631421739,
+      "learning_rate": 7.94866810491012e-06,
+      "loss": 0.4437,
+      "step": 234
+    },
+    {
+      "epoch": 0.6010230179028133,
+      "grad_norm": 6.436431203161013,
+      "learning_rate": 7.93242220919417e-06,
+      "loss": 0.5474,
+      "step": 235
+    },
+    {
+      "epoch": 0.6035805626598465,
+      "grad_norm": 5.574144310350591,
+      "learning_rate": 7.916128986131761e-06,
+      "loss": 0.5439,
+      "step": 236
+    },
+    {
+      "epoch": 0.6061381074168798,
+      "grad_norm": 4.8240456797819835,
+      "learning_rate": 7.899788698684687e-06,
+      "loss": 0.4686,
+      "step": 237
+    },
+    {
+      "epoch": 0.6086956521739131,
+      "grad_norm": 6.5914849870729055,
+      "learning_rate": 7.883401610574338e-06,
+      "loss": 0.5512,
+      "step": 238
+    },
+    {
+      "epoch": 0.6112531969309463,
+      "grad_norm": 4.739332602957458,
+      "learning_rate": 7.866967986277423e-06,
+      "loss": 0.4204,
+      "step": 239
+    },
+    {
+      "epoch": 0.6138107416879796,
+      "grad_norm": 6.116120073202256,
+      "learning_rate": 7.850488091021726e-06,
+      "loss": 0.5596,
+      "step": 240
+    },
+    {
+      "epoch": 0.6163682864450127,
+      "grad_norm": 6.299579832647148,
+      "learning_rate": 7.833962190781809e-06,
+      "loss": 0.5729,
+      "step": 241
+    },
+    {
+      "epoch": 0.618925831202046,
+      "grad_norm": 5.77832842987742,
+      "learning_rate": 7.817390552274721e-06,
+      "loss": 0.4062,
+      "step": 242
+    },
+    {
+      "epoch": 0.6214833759590793,
+      "grad_norm": 5.6928424134185365,
+      "learning_rate": 7.800773442955703e-06,
+      "loss": 0.562,
+      "step": 243
+    },
+    {
+      "epoch": 0.6240409207161125,
+      "grad_norm": 5.754032663780959,
+      "learning_rate": 7.784111131013858e-06,
+      "loss": 0.4763,
+      "step": 244
+    },
+    {
+      "epoch": 0.6265984654731458,
+      "grad_norm": 6.1254790208347,
+      "learning_rate": 7.767403885367832e-06,
+      "loss": 0.4931,
+      "step": 245
+    },
+    {
+      "epoch": 0.629156010230179,
+      "grad_norm": 6.088136454995643,
+      "learning_rate": 7.750651975661471e-06,
+      "loss": 0.5366,
+      "step": 246
+    },
+    {
+      "epoch": 0.6317135549872123,
+      "grad_norm": 6.525801581028963,
+      "learning_rate": 7.733855672259472e-06,
+      "loss": 0.5869,
+      "step": 247
+    },
+    {
+      "epoch": 0.6342710997442456,
+      "grad_norm": 5.142226783902718,
+      "learning_rate": 7.717015246243012e-06,
+      "loss": 0.4107,
+      "step": 248
+    },
+    {
+      "epoch": 0.6368286445012787,
+      "grad_norm": 5.884475685733821,
+      "learning_rate": 7.700130969405377e-06,
+      "loss": 0.5575,
+      "step": 249
+    },
+    {
+      "epoch": 0.639386189258312,
+      "grad_norm": 5.430956350007929,
+      "learning_rate": 7.683203114247587e-06,
+      "loss": 0.4316,
+      "step": 250
+    },
+    {
+      "epoch": 0.6419437340153452,
+      "grad_norm": 5.852470007876826,
+      "learning_rate": 7.66623195397397e-06,
+      "loss": 0.5228,
+      "step": 251
+    },
+    {
+      "epoch": 0.6445012787723785,
+      "grad_norm": 6.316931409524609,
+      "learning_rate": 7.649217762487786e-06,
+      "loss": 0.6069,
+      "step": 252
+    },
+    {
+      "epoch": 0.6470588235294118,
+      "grad_norm": 5.981156306158716,
+      "learning_rate": 7.63216081438678e-06,
+      "loss": 0.4525,
+      "step": 253
+    },
+    {
+      "epoch": 0.649616368286445,
+      "grad_norm": 6.193124213697377,
+      "learning_rate": 7.615061384958764e-06,
+      "loss": 0.6367,
+      "step": 254
+    },
+    {
+      "epoch": 0.6521739130434783,
+      "grad_norm": 6.145394633019291,
+      "learning_rate": 7.597919750177168e-06,
+      "loss": 0.5622,
+      "step": 255
+    },
+    {
+      "epoch": 0.6547314578005116,
+      "grad_norm": 6.076895232152138,
+      "learning_rate": 7.580736186696593e-06,
+      "loss": 0.5016,
+      "step": 256
+    },
+    {
+      "epoch": 0.6572890025575447,
+      "grad_norm": 5.562852949209647,
+      "learning_rate": 7.563510971848339e-06,
+      "loss": 0.5739,
+      "step": 257
+    },
+    {
+      "epoch": 0.659846547314578,
+      "grad_norm": 6.358261854476947,
+      "learning_rate": 7.546244383635929e-06,
+      "loss": 0.5783,
+      "step": 258
+    },
+    {
+      "epoch": 0.6624040920716112,
+      "grad_norm": 5.61873313563532,
+      "learning_rate": 7.528936700730627e-06,
+      "loss": 0.5671,
+      "step": 259
+    },
+    {
+      "epoch": 0.6649616368286445,
+      "grad_norm": 5.007471092297137,
+      "learning_rate": 7.5115882024669375e-06,
+      "loss": 0.4238,
+      "step": 260
+    },
+    {
+      "epoch": 0.6675191815856778,
+      "grad_norm": 5.02718344671977,
+      "learning_rate": 7.494199168838099e-06,
+      "loss": 0.431,
+      "step": 261
+    },
+    {
+      "epoch": 0.670076726342711,
+      "grad_norm": 5.822025275525143,
+      "learning_rate": 7.476769880491561e-06,
+      "loss": 0.555,
+      "step": 262
+    },
+    {
+      "epoch": 0.6726342710997443,
+      "grad_norm": 5.852672525450696,
+      "learning_rate": 7.459300618724462e-06,
+      "loss": 0.4537,
+      "step": 263
+    },
+    {
+      "epoch": 0.6751918158567775,
+      "grad_norm": 6.36830749484907,
+      "learning_rate": 7.44179166547908e-06,
+      "loss": 0.5466,
+      "step": 264
+    },
+    {
+      "epoch": 0.6777493606138107,
+      "grad_norm": 5.187641704740303,
+      "learning_rate": 7.42424330333829e-06,
+      "loss": 0.4966,
+      "step": 265
+    },
+    {
+      "epoch": 0.680306905370844,
+      "grad_norm": 5.862839321803861,
+      "learning_rate": 7.406655815520998e-06,
+      "loss": 0.4902,
+      "step": 266
+    },
+    {
+      "epoch": 0.6828644501278772,
+      "grad_norm": 6.529161660718858,
+      "learning_rate": 7.389029485877577e-06,
+      "loss": 0.493,
+      "step": 267
+    },
+    {
+      "epoch": 0.6854219948849105,
+      "grad_norm": 5.732050686572585,
+      "learning_rate": 7.371364598885276e-06,
+      "loss": 0.4744,
+      "step": 268
+    },
+    {
+      "epoch": 0.6879795396419437,
+      "grad_norm": 5.533174363200175,
+      "learning_rate": 7.353661439643638e-06,
+      "loss": 0.3833,
+      "step": 269
+    },
+    {
+      "epoch": 0.690537084398977,
+      "grad_norm": 5.768168615527615,
+      "learning_rate": 7.335920293869891e-06,
+      "loss": 0.423,
+      "step": 270
+    },
+    {
+      "epoch": 0.6930946291560103,
+      "grad_norm": 5.852266644103708,
+      "learning_rate": 7.318141447894344e-06,
+      "loss": 0.3371,
+      "step": 271
+    },
+    {
+      "epoch": 0.6956521739130435,
+      "grad_norm": 6.038563114564619,
+      "learning_rate": 7.300325188655762e-06,
+      "loss": 0.4891,
+      "step": 272
+    },
+    {
+      "epoch": 0.6982097186700768,
+      "grad_norm": 6.280803826327464,
+      "learning_rate": 7.28247180369673e-06,
+      "loss": 0.5385,
+      "step": 273
+    },
+    {
+      "epoch": 0.7007672634271099,
+      "grad_norm": 7.292365659382516,
+      "learning_rate": 7.264581581159024e-06,
+      "loss": 0.6148,
+      "step": 274
+    },
+    {
+      "epoch": 0.7033248081841432,
+      "grad_norm": 6.763240999324924,
+      "learning_rate": 7.246654809778951e-06,
+      "loss": 0.5272,
+      "step": 275
+    },
+    {
+      "epoch": 0.7058823529411765,
+      "grad_norm": 6.444401975777849,
+      "learning_rate": 7.2286917788826926e-06,
+      "loss": 0.4879,
+      "step": 276
+    },
+    {
+      "epoch": 0.7084398976982097,
+      "grad_norm": 5.037923525497081,
+      "learning_rate": 7.210692778381634e-06,
+      "loss": 0.5377,
+      "step": 277
+    },
+    {
+      "epoch": 0.710997442455243,
+      "grad_norm": 6.327806611970394,
+      "learning_rate": 7.192658098767686e-06,
+      "loss": 0.4654,
+      "step": 278
+    },
+    {
+      "epoch": 0.7135549872122762,
+      "grad_norm": 5.832786135763086,
+      "learning_rate": 7.174588031108598e-06,
+      "loss": 0.5921,
+      "step": 279
+    },
+    {
+      "epoch": 0.7161125319693095,
+      "grad_norm": 6.032098832742715,
+      "learning_rate": 7.1564828670432595e-06,
+      "loss": 0.5032,
+      "step": 280
+    },
+    {
+      "epoch": 0.7186700767263428,
+      "grad_norm": 7.528807903355475,
+      "learning_rate": 7.138342898776989e-06,
+      "loss": 0.5143,
+      "step": 281
+    },
+    {
+      "epoch": 0.7212276214833759,
+      "grad_norm": 5.662236290695636,
+      "learning_rate": 7.120168419076825e-06,
+      "loss": 0.5752,
+      "step": 282
+    },
+    {
+      "epoch": 0.7237851662404092,
+      "grad_norm": 5.922836277812778,
+      "learning_rate": 7.101959721266798e-06,
+      "loss": 0.5907,
+      "step": 283
+    },
+    {
+      "epoch": 0.7263427109974424,
+      "grad_norm": 6.258012070363337,
+      "learning_rate": 7.083717099223192e-06,
+      "loss": 0.5447,
+      "step": 284
+    },
+    {
+      "epoch": 0.7289002557544757,
+      "grad_norm": 5.261480296532744,
+      "learning_rate": 7.0654408473698084e-06,
+      "loss": 0.4521,
+      "step": 285
+    },
+    {
+      "epoch": 0.731457800511509,
+      "grad_norm": 5.918110722172615,
+      "learning_rate": 7.047131260673214e-06,
+      "loss": 0.4637,
+      "step": 286
+    },
+    {
+      "epoch": 0.7340153452685422,
+      "grad_norm": 5.741282290810403,
+      "learning_rate": 7.0287886346379755e-06,
+      "loss": 0.4131,
+      "step": 287
+    },
+    {
+      "epoch": 0.7365728900255755,
+      "grad_norm": 5.943919434881143,
+      "learning_rate": 7.010413265301888e-06,
+      "loss": 0.4712,
+      "step": 288
+    },
+    {
+      "epoch": 0.7391304347826086,
+      "grad_norm": 6.1059644383499885,
+      "learning_rate": 6.9920054492312086e-06,
+      "loss": 0.6022,
+      "step": 289
+    },
+    {
+      "epoch": 0.7416879795396419,
+      "grad_norm": 6.884474367848085,
+      "learning_rate": 6.97356548351586e-06,
+      "loss": 0.5212,
+      "step": 290
+    },
+    {
+      "epoch": 0.7442455242966752,
+      "grad_norm": 5.758493578440039,
+      "learning_rate": 6.9550936657646386e-06,
+      "loss": 0.507,
+      "step": 291
+    },
+    {
+      "epoch": 0.7468030690537084,
+      "grad_norm": 5.5122177192122415,
+      "learning_rate": 6.936590294100414e-06,
+      "loss": 0.4096,
+      "step": 292
+    },
+    {
+      "epoch": 0.7493606138107417,
+      "grad_norm": 6.529147733060143,
+      "learning_rate": 6.918055667155311e-06,
+      "loss": 0.4668,
+      "step": 293
+    },
+    {
+      "epoch": 0.7519181585677749,
+      "grad_norm": 5.580434227838566,
+      "learning_rate": 6.899490084065897e-06,
+      "loss": 0.4825,
+      "step": 294
+    },
+    {
+      "epoch": 0.7544757033248082,
+      "grad_norm": 6.141771398723171,
+      "learning_rate": 6.8808938444683505e-06,
+      "loss": 0.5189,
+      "step": 295
+    },
+    {
+      "epoch": 0.7570332480818415,
+      "grad_norm": 6.258445455734687,
+      "learning_rate": 6.862267248493624e-06,
+      "loss": 0.4217,
+      "step": 296
+    },
+    {
+      "epoch": 0.7595907928388747,
+      "grad_norm": 5.577447249480196,
+      "learning_rate": 6.843610596762606e-06,
+      "loss": 0.4574,
+      "step": 297
+    },
+    {
+      "epoch": 0.7621483375959079,
+      "grad_norm": 6.508949986966596,
+      "learning_rate": 6.824924190381257e-06,
+      "loss": 0.4512,
+      "step": 298
+    },
+    {
+      "epoch": 0.7647058823529411,
+      "grad_norm": 6.487835617711066,
+      "learning_rate": 6.806208330935766e-06,
+      "loss": 0.4817,
+      "step": 299
+    },
+    {
+      "epoch": 0.7672634271099744,
+      "grad_norm": 6.733508290939032,
+      "learning_rate": 6.7874633204876705e-06,
+      "loss": 0.4648,
+      "step": 300
+    },
+    {
+      "epoch": 0.7698209718670077,
+      "grad_norm": 6.6391089444926195,
+      "learning_rate": 6.768689461568987e-06,
+      "loss": 0.4959,
+      "step": 301
+    },
+    {
+      "epoch": 0.7723785166240409,
+      "grad_norm": 5.426573235588597,
+      "learning_rate": 6.7498870571773275e-06,
+      "loss": 0.4101,
+      "step": 302
+    },
+    {
+      "epoch": 0.7749360613810742,
+      "grad_norm": 5.27403325499086,
+      "learning_rate": 6.731056410771008e-06,
+      "loss": 0.4183,
+      "step": 303
+    },
+    {
+      "epoch": 0.7774936061381074,
+      "grad_norm": 5.602097858442588,
+      "learning_rate": 6.712197826264154e-06,
+      "loss": 0.4712,
+      "step": 304
+    },
+    {
+      "epoch": 0.7800511508951407,
+      "grad_norm": 5.602023911663575,
+      "learning_rate": 6.69331160802179e-06,
+      "loss": 0.376,
+      "step": 305
+    },
+    {
+      "epoch": 0.782608695652174,
+      "grad_norm": 6.057331404811353,
+      "learning_rate": 6.674398060854931e-06,
+      "loss": 0.3333,
+      "step": 306
+    },
+    {
+      "epoch": 0.7851662404092071,
+      "grad_norm": 5.453910778706793,
+      "learning_rate": 6.655457490015667e-06,
+      "loss": 0.5251,
+      "step": 307
+    },
+    {
+      "epoch": 0.7877237851662404,
+      "grad_norm": 6.209463244054028,
+      "learning_rate": 6.636490201192229e-06,
+      "loss": 0.5256,
+      "step": 308
+    },
+    {
+      "epoch": 0.7902813299232737,
+      "grad_norm": 5.6125006489249145,
+      "learning_rate": 6.617496500504056e-06,
+      "loss": 0.35,
+      "step": 309
+    },
+    {
+      "epoch": 0.7928388746803069,
+      "grad_norm": 6.414498410153366,
+      "learning_rate": 6.5984766944968636e-06,
+      "loss": 0.5181,
+      "step": 310
+    },
+    {
+      "epoch": 0.7953964194373402,
+      "grad_norm": 6.101747662704522,
+      "learning_rate": 6.579431090137681e-06,
+      "loss": 0.4106,
+      "step": 311
+    },
+    {
+      "epoch": 0.7979539641943734,
+      "grad_norm": 6.462777333488606,
+      "learning_rate": 6.560359994809916e-06,
+      "loss": 0.6125,
+      "step": 312
+    },
+    {
+      "epoch": 0.8005115089514067,
+      "grad_norm": 5.9920053305051875,
+      "learning_rate": 6.541263716308375e-06,
+      "loss": 0.4968,
+      "step": 313
+    },
+    {
+      "epoch": 0.80306905370844,
+      "grad_norm": 6.671005371719509,
+      "learning_rate": 6.522142562834307e-06,
+      "loss": 0.5637,
+      "step": 314
+    },
+    {
+      "epoch": 0.8056265984654731,
+      "grad_norm": 5.361336122168199,
+      "learning_rate": 6.502996842990431e-06,
+      "loss": 0.4208,
+      "step": 315
+    },
+    {
+      "epoch": 0.8081841432225064,
+      "grad_norm": 5.670064103939166,
+      "learning_rate": 6.483826865775941e-06,
+      "loss": 0.5278,
+      "step": 316
+    },
+    {
+      "epoch": 0.8107416879795396,
+      "grad_norm": 5.5103998057715105,
+      "learning_rate": 6.46463294058154e-06,
+      "loss": 0.4007,
+      "step": 317
+    },
+    {
+      "epoch": 0.8132992327365729,
+      "grad_norm": 5.659095784663181,
+      "learning_rate": 6.445415377184427e-06,
+      "loss": 0.4742,
+      "step": 318
+    },
+    {
+      "epoch": 0.8158567774936062,
+      "grad_norm": 6.30132561670194,
+      "learning_rate": 6.426174485743309e-06,
+      "loss": 0.4078,
+      "step": 319
+    },
+    {
+      "epoch": 0.8184143222506394,
+      "grad_norm": 5.643268096385628,
+      "learning_rate": 6.4069105767933944e-06,
+      "loss": 0.46,
+      "step": 320
+    },
+    {
+      "epoch": 0.8209718670076727,
+      "grad_norm": 7.788725418859061,
+      "learning_rate": 6.387623961241375e-06,
+      "loss": 0.6119,
+      "step": 321
+    },
+    {
+      "epoch": 0.8235294117647058,
+      "grad_norm": 5.927896020719375,
+      "learning_rate": 6.368314950360416e-06,
+      "loss": 0.5225,
+      "step": 322
+    },
+    {
+      "epoch": 0.8260869565217391,
+      "grad_norm": 5.296955151964955,
+      "learning_rate": 6.348983855785122e-06,
+      "loss": 0.3126,
+      "step": 323
+    },
+    {
+      "epoch": 0.8286445012787724,
+      "grad_norm": 5.095733276738074,
+      "learning_rate": 6.3296309895065215e-06,
+      "loss": 0.3639,
+      "step": 324
+    },
+    {
+      "epoch": 0.8312020460358056,
+      "grad_norm": 6.080988913298908,
+      "learning_rate": 6.310256663867019e-06,
+      "loss": 0.5063,
+      "step": 325
+    },
+    {
+      "epoch": 0.8337595907928389,
+      "grad_norm": 7.036245894709906,
+      "learning_rate": 6.290861191555359e-06,
+      "loss": 0.4578,
+      "step": 326
+    },
+    {
+      "epoch": 0.8363171355498721,
+      "grad_norm": 5.580633409599807,
+      "learning_rate": 6.271444885601583e-06,
+      "loss": 0.4639,
+      "step": 327
+    },
+    {
+      "epoch": 0.8388746803069054,
+      "grad_norm": 6.523213064272758,
+      "learning_rate": 6.252008059371968e-06,
+      "loss": 0.4699,
+      "step": 328
+    },
+    {
+      "epoch": 0.8414322250639387,
+      "grad_norm": 4.85798591447732,
+      "learning_rate": 6.2325510265639785e-06,
+      "loss": 0.2973,
+      "step": 329
+    },
+    {
+      "epoch": 0.8439897698209718,
+      "grad_norm": 6.805564012992218,
+      "learning_rate": 6.213074101201202e-06,
+      "loss": 0.4894,
+      "step": 330
+    },
+    {
+      "epoch": 0.8465473145780051,
+      "grad_norm": 6.052583253032932,
+      "learning_rate": 6.193577597628268e-06,
+      "loss": 0.5193,
+      "step": 331
+    },
+    {
+      "epoch": 0.8491048593350383,
+      "grad_norm": 6.230894069829904,
+      "learning_rate": 6.174061830505801e-06,
+      "loss": 0.5028,
+      "step": 332
+    },
+    {
+      "epoch": 0.8516624040920716,
+      "grad_norm": 7.201079594756455,
+      "learning_rate": 6.154527114805312e-06,
+      "loss": 0.618,
+      "step": 333
+    },
+    {
+      "epoch": 0.8542199488491049,
+      "grad_norm": 6.600993850416883,
+      "learning_rate": 6.1349737658041385e-06,
+      "loss": 0.5133,
+      "step": 334
+    },
+    {
+      "epoch": 0.8567774936061381,
+      "grad_norm": 5.724822321191247,
+      "learning_rate": 6.115402099080345e-06,
+      "loss": 0.4838,
+      "step": 335
+    },
+    {
+      "epoch": 0.8593350383631714,
+      "grad_norm": 5.613108225355487,
+      "learning_rate": 6.095812430507627e-06,
+      "loss": 0.3442,
+      "step": 336
+    },
+    {
+      "epoch": 0.8618925831202046,
+      "grad_norm": 5.965477408637899,
+      "learning_rate": 6.076205076250227e-06,
+      "loss": 0.5109,
+      "step": 337
+    },
+    {
+      "epoch": 0.8644501278772379,
+      "grad_norm": 5.9857901067552,
+      "learning_rate": 6.056580352757813e-06,
+      "loss": 0.4511,
+      "step": 338
+    },
+    {
+      "epoch": 0.8670076726342711,
+      "grad_norm": 6.174131332105638,
+      "learning_rate": 6.036938576760388e-06,
+      "loss": 0.4419,
+      "step": 339
+    },
+    {
+      "epoch": 0.8695652173913043,
+      "grad_norm": 5.075847963553367,
+      "learning_rate": 6.0172800652631706e-06,
+      "loss": 0.3777,
+      "step": 340
+    },
+    {
+      "epoch": 0.8721227621483376,
+      "grad_norm": 7.190594951132575,
+      "learning_rate": 5.997605135541472e-06,
+      "loss": 0.5106,
+      "step": 341
+    },
+    {
+      "epoch": 0.8746803069053708,
+      "grad_norm": 6.747196680979683,
+      "learning_rate": 5.977914105135594e-06,
+      "loss": 0.4762,
+      "step": 342
+    },
+    {
+      "epoch": 0.8772378516624041,
+      "grad_norm": 5.714370912906624,
+      "learning_rate": 5.9582072918456805e-06,
+      "loss": 0.362,
+      "step": 343
+    },
+    {
+      "epoch": 0.8797953964194374,
+      "grad_norm": 5.236876408531075,
+      "learning_rate": 5.938485013726612e-06,
+      "loss": 0.3947,
+      "step": 344
+    },
+    {
+      "epoch": 0.8823529411764706,
+      "grad_norm": 5.116217278468624,
+      "learning_rate": 5.918747589082853e-06,
+      "loss": 0.4747,
+      "step": 345
+    },
+    {
+      "epoch": 0.8849104859335039,
+      "grad_norm": 4.749747087957306,
+      "learning_rate": 5.898995336463326e-06,
+      "loss": 0.4274,
+      "step": 346
+    },
+    {
+      "epoch": 0.887468030690537,
+      "grad_norm": 5.230418676152823,
+      "learning_rate": 5.879228574656269e-06,
+      "loss": 0.3441,
+      "step": 347
+    },
+    {
+      "epoch": 0.8900255754475703,
+      "grad_norm": 6.303242756857959,
+      "learning_rate": 5.859447622684084e-06,
+      "loss": 0.5131,
+      "step": 348
+    },
+    {
+      "epoch": 0.8925831202046036,
+      "grad_norm": 5.4435933577635645,
+      "learning_rate": 5.839652799798197e-06,
+      "loss": 0.4243,
+      "step": 349
+    },
+    {
+      "epoch": 0.8951406649616368,
+      "grad_norm": 6.921408500128556,
+      "learning_rate": 5.819844425473899e-06,
+      "loss": 0.5549,
+      "step": 350
+    },
+    {
+      "epoch": 0.8976982097186701,
+      "grad_norm": 6.635487149449039,
+      "learning_rate": 5.800022819405194e-06,
+      "loss": 0.5061,
+      "step": 351
+    },
+    {
+      "epoch": 0.9002557544757033,
+      "grad_norm": 6.3209563326259515,
+      "learning_rate": 5.780188301499636e-06,
+      "loss": 0.5999,
+      "step": 352
+    },
+    {
+      "epoch": 0.9028132992327366,
+      "grad_norm": 6.077422261762329,
+      "learning_rate": 5.760341191873167e-06,
+      "loss": 0.5111,
+      "step": 353
+    },
+    {
+      "epoch": 0.9053708439897699,
+      "grad_norm": 5.867259188193717,
+      "learning_rate": 5.740481810844952e-06,
+      "loss": 0.4771,
+      "step": 354
+    },
+    {
+      "epoch": 0.907928388746803,
+      "grad_norm": 4.777893746653604,
+      "learning_rate": 5.720610478932211e-06,
+      "loss": 0.3242,
+      "step": 355
+    },
+    {
+      "epoch": 0.9104859335038363,
+      "grad_norm": 5.764419825204445,
+      "learning_rate": 5.700727516845038e-06,
+      "loss": 0.3306,
+      "step": 356
+    },
+    {
+      "epoch": 0.9130434782608695,
+      "grad_norm": 5.510318712985209,
+      "learning_rate": 5.680833245481234e-06,
+      "loss": 0.4642,
+      "step": 357
+    },
+    {
+      "epoch": 0.9156010230179028,
+      "grad_norm": 6.53182819796998,
+      "learning_rate": 5.660927985921122e-06,
+      "loss": 0.51,
+      "step": 358
+    },
+    {
+      "epoch": 0.9181585677749361,
+      "grad_norm": 6.086318246788371,
+      "learning_rate": 5.641012059422369e-06,
+      "loss": 0.5472,
+      "step": 359
+    },
+    {
+      "epoch": 0.9207161125319693,
+      "grad_norm": 5.544550425971534,
+      "learning_rate": 5.621085787414799e-06,
+      "loss": 0.4603,
+      "step": 360
+    },
+    {
+      "epoch": 0.9232736572890026,
+      "grad_norm": 6.1043161575819616,
+      "learning_rate": 5.601149491495206e-06,
+      "loss": 0.485,
+      "step": 361
+    },
+    {
+      "epoch": 0.9258312020460358,
+      "grad_norm": 5.827986489165051,
+      "learning_rate": 5.581203493422161e-06,
+      "loss": 0.5864,
+      "step": 362
+    },
+    {
+      "epoch": 0.928388746803069,
+      "grad_norm": 5.172049016763376,
+      "learning_rate": 5.561248115110822e-06,
+      "loss": 0.4517,
+      "step": 363
+    },
+    {
+      "epoch": 0.9309462915601023,
+      "grad_norm": 6.640832021653832,
+      "learning_rate": 5.541283678627742e-06,
+      "loss": 0.3703,
+      "step": 364
+    },
+    {
+      "epoch": 0.9335038363171355,
+      "grad_norm": 5.361432485515485,
+      "learning_rate": 5.521310506185661e-06,
+      "loss": 0.4262,
+      "step": 365
+    },
+    {
+      "epoch": 0.9360613810741688,
+      "grad_norm": 5.808037599792696,
+      "learning_rate": 5.501328920138314e-06,
+      "loss": 0.544,
+      "step": 366
+    },
+    {
+      "epoch": 0.9386189258312021,
+      "grad_norm": 6.0541662716251095,
+      "learning_rate": 5.481339242975227e-06,
+      "loss": 0.4024,
+      "step": 367
+    },
+    {
+      "epoch": 0.9411764705882353,
+      "grad_norm": 6.232047697753629,
+      "learning_rate": 5.46134179731651e-06,
+      "loss": 0.4862,
+      "step": 368
+    },
+    {
+      "epoch": 0.9437340153452686,
+      "grad_norm": 6.000932213910604,
+      "learning_rate": 5.441336905907653e-06,
+      "loss": 0.4635,
+      "step": 369
+    },
+    {
+      "epoch": 0.9462915601023018,
+      "grad_norm": 7.611546241496742,
+      "learning_rate": 5.421324891614312e-06,
+      "loss": 0.4135,
+      "step": 370
+    },
+    {
+      "epoch": 0.948849104859335,
+      "grad_norm": 5.362330490202002,
+      "learning_rate": 5.4013060774171055e-06,
+      "loss": 0.4506,
+      "step": 371
+    },
+    {
+      "epoch": 0.9514066496163683,
+      "grad_norm": 5.346120483245015,
+      "learning_rate": 5.3812807864063946e-06,
+      "loss": 0.4576,
+      "step": 372
+    },
+    {
+      "epoch": 0.9539641943734015,
+      "grad_norm": 6.100965821026688,
+      "learning_rate": 5.361249341777075e-06,
+      "loss": 0.5165,
+      "step": 373
+    },
+    {
+      "epoch": 0.9565217391304348,
+      "grad_norm": 4.676613084116823,
+      "learning_rate": 5.341212066823356e-06,
+      "loss": 0.4383,
+      "step": 374
+    },
+    {
+      "epoch": 0.959079283887468,
+      "grad_norm": 6.564693244827939,
+      "learning_rate": 5.321169284933543e-06,
+      "loss": 0.5044,
+      "step": 375
+    },
+    {
+      "epoch": 0.9616368286445013,
+      "grad_norm": 4.752286756978166,
+      "learning_rate": 5.3011213195848245e-06,
+      "loss": 0.5422,
+      "step": 376
+    },
+    {
+      "epoch": 0.9641943734015346,
+      "grad_norm": 6.298216680263071,
+      "learning_rate": 5.281068494338039e-06,
+      "loss": 0.3751,
+      "step": 377
+    },
+    {
+      "epoch": 0.9667519181585678,
+      "grad_norm": 5.041453892793781,
+      "learning_rate": 5.26101113283247e-06,
+      "loss": 0.3732,
+      "step": 378
+    },
+    {
+      "epoch": 0.969309462915601,
+      "grad_norm": 6.072922237394072,
+      "learning_rate": 5.240949558780605e-06,
+      "loss": 0.4873,
+      "step": 379
+    },
+    {
+      "epoch": 0.9718670076726342,
+      "grad_norm": 5.712375997105542,
+      "learning_rate": 5.220884095962924e-06,
+      "loss": 0.4877,
+      "step": 380
+    },
+    {
+      "epoch": 0.9744245524296675,
+      "grad_norm": 5.5897510763688585,
+      "learning_rate": 5.200815068222666e-06,
+      "loss": 0.386,
+      "step": 381
+    },
+    {
+      "epoch": 0.9769820971867008,
+      "grad_norm": 6.160592176666778,
+      "learning_rate": 5.1807427994606065e-06,
+      "loss": 0.369,
+      "step": 382
+    },
+    {
+      "epoch": 0.979539641943734,
+      "grad_norm": 5.482848722330923,
+      "learning_rate": 5.1606676136298305e-06,
+      "loss": 0.4618,
+      "step": 383
+    },
+    {
+      "epoch": 0.9820971867007673,
+      "grad_norm": 5.96411837712334,
+      "learning_rate": 5.140589834730503e-06,
+      "loss": 0.4286,
+      "step": 384
+    },
+    {
+      "epoch": 0.9846547314578005,
+      "grad_norm": 6.021924443213883,
+      "learning_rate": 5.120509786804635e-06,
+      "loss": 0.4545,
+      "step": 385
+    },
+    {
+      "epoch": 0.9872122762148338,
+      "grad_norm": 4.860619712046072,
+      "learning_rate": 5.100427793930862e-06,
+      "loss": 0.4847,
+      "step": 386
+    },
+    {
+      "epoch": 0.989769820971867,
+      "grad_norm": 6.16249371180202,
+      "learning_rate": 5.08034418021921e-06,
+      "loss": 0.4119,
+      "step": 387
+    },
+    {
+      "epoch": 0.9923273657289002,
+      "grad_norm": 5.190813395638479,
+      "learning_rate": 5.06025926980586e-06,
+      "loss": 0.3609,
+      "step": 388
+    },
+    {
+      "epoch": 0.9948849104859335,
+      "grad_norm": 5.3092155597782025,
+      "learning_rate": 5.040173386847926e-06,
+      "loss": 0.3655,
+      "step": 389
+    },
+    {
+      "epoch": 0.9974424552429667,
+      "grad_norm": 5.7321767766000935,
+      "learning_rate": 5.0200868555182155e-06,
+      "loss": 0.5465,
+      "step": 390
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.719257228559791,
+      "learning_rate": 5e-06,
+      "loss": 0.4347,
+      "step": 391
+    },
+    {
+      "epoch": 1.0025575447570332,
+      "grad_norm": 4.2391486408050145,
+      "learning_rate": 4.979913144481785e-06,
+      "loss": 0.2179,
+      "step": 392
+    },
+    {
+      "epoch": 1.0051150895140666,
+      "grad_norm": 4.604106412801713,
+      "learning_rate": 4.959826613152074e-06,
+      "loss": 0.2675,
+      "step": 393
+    },
+    {
+      "epoch": 1.0076726342710998,
+      "grad_norm": 4.113700302207306,
+      "learning_rate": 4.939740730194141e-06,
+      "loss": 0.2586,
+      "step": 394
+    },
+    {
+      "epoch": 1.010230179028133,
+      "grad_norm": 4.894518417900452,
+      "learning_rate": 4.919655819780792e-06,
+      "loss": 0.278,
+      "step": 395
+    },
+    {
+      "epoch": 1.0127877237851663,
+      "grad_norm": 4.3903147870940815,
+      "learning_rate": 4.899572206069138e-06,
+      "loss": 0.2175,
+      "step": 396
+    },
+    {
+      "epoch": 1.0153452685421995,
+      "grad_norm": 4.395292367085451,
+      "learning_rate": 4.879490213195366e-06,
+      "loss": 0.2597,
+      "step": 397
+    },
+    {
+      "epoch": 1.0179028132992327,
+      "grad_norm": 4.399892144315872,
+      "learning_rate": 4.8594101652694996e-06,
+      "loss": 0.2806,
+      "step": 398
+    },
+    {
+      "epoch": 1.020460358056266,
+      "grad_norm": 5.345621237626395,
+      "learning_rate": 4.839332386370171e-06,
+      "loss": 0.2571,
+      "step": 399
+    },
+    {
+      "epoch": 1.0230179028132993,
+      "grad_norm": 4.533667138931473,
+      "learning_rate": 4.819257200539394e-06,
+      "loss": 0.2646,
+      "step": 400
+    },
+    {
+      "epoch": 1.0255754475703325,
+      "grad_norm": 5.037019368108468,
+      "learning_rate": 4.799184931777337e-06,
+      "loss": 0.1862,
+      "step": 401
+    },
+    {
+      "epoch": 1.0281329923273657,
+      "grad_norm": 5.628048303078724,
+      "learning_rate": 4.779115904037079e-06,
+      "loss": 0.239,
+      "step": 402
+    },
+    {
+      "epoch": 1.030690537084399,
+      "grad_norm": 6.657437275552853,
+      "learning_rate": 4.759050441219395e-06,
+      "loss": 0.2032,
+      "step": 403
+    },
+    {
+      "epoch": 1.0332480818414322,
+      "grad_norm": 7.066364647253674,
+      "learning_rate": 4.738988867167531e-06,
+      "loss": 0.2686,
+      "step": 404
+    },
+    {
+      "epoch": 1.0358056265984654,
+      "grad_norm": 6.242504066160711,
+      "learning_rate": 4.718931505661961e-06,
+      "loss": 0.2039,
+      "step": 405
+    },
+    {
+      "epoch": 1.0383631713554988,
+      "grad_norm": 7.053347520596888,
+      "learning_rate": 4.698878680415176e-06,
+      "loss": 0.2677,
+      "step": 406
+    },
+    {
+      "epoch": 1.040920716112532,
+      "grad_norm": 6.980982091790442,
+      "learning_rate": 4.678830715066458e-06,
+      "loss": 0.2986,
+      "step": 407
+    },
+    {
+      "epoch": 1.0434782608695652,
+      "grad_norm": 7.1587957747958715,
+      "learning_rate": 4.6587879331766465e-06,
+      "loss": 0.2742,
+      "step": 408
+    },
+    {
+      "epoch": 1.0460358056265984,
+      "grad_norm": 6.506842088570043,
+      "learning_rate": 4.638750658222927e-06,
+      "loss": 0.327,
+      "step": 409
+    },
+    {
+      "epoch": 1.0485933503836318,
+      "grad_norm": 5.343374587345889,
+      "learning_rate": 4.618719213593605e-06,
+      "loss": 0.2399,
+      "step": 410
+    },
+    {
+      "epoch": 1.051150895140665,
+      "grad_norm": 5.952604432226821,
+      "learning_rate": 4.598693922582896e-06,
+      "loss": 0.3362,
+      "step": 411
+    },
+    {
+      "epoch": 1.0537084398976981,
+      "grad_norm": 5.292017124731555,
+      "learning_rate": 4.5786751083856895e-06,
+      "loss": 0.2229,
+      "step": 412
+    },
+    {
+      "epoch": 1.0562659846547315,
+      "grad_norm": 5.7674633387114,
+      "learning_rate": 4.558663094092348e-06,
+      "loss": 0.2716,
+      "step": 413
+    },
+    {
+      "epoch": 1.0588235294117647,
+      "grad_norm": 4.654675739900228,
+      "learning_rate": 4.53865820268349e-06,
+      "loss": 0.2439,
+      "step": 414
+    },
+    {
+      "epoch": 1.061381074168798,
+      "grad_norm": 6.404992169914481,
+      "learning_rate": 4.518660757024774e-06,
+      "loss": 0.2713,
+      "step": 415
+    },
+    {
+      "epoch": 1.0639386189258313,
+      "grad_norm": 4.940754639892674,
+      "learning_rate": 4.498671079861686e-06,
+      "loss": 0.2225,
+      "step": 416
+    },
+    {
+      "epoch": 1.0664961636828645,
+      "grad_norm": 4.76891987704093,
+      "learning_rate": 4.478689493814341e-06,
+      "loss": 0.2676,
+      "step": 417
+    },
+    {
+      "epoch": 1.0690537084398977,
+      "grad_norm": 5.023886841031966,
+      "learning_rate": 4.4587163213722595e-06,
+      "loss": 0.2594,
+      "step": 418
+    },
+    {
+      "epoch": 1.0716112531969308,
+      "grad_norm": 4.184649476108494,
+      "learning_rate": 4.438751884889179e-06,
+      "loss": 0.1791,
+      "step": 419
+    },
+    {
+      "epoch": 1.0741687979539642,
+      "grad_norm": 5.029459043202698,
+      "learning_rate": 4.41879650657784e-06,
+      "loss": 0.2588,
+      "step": 420
+    },
+    {
+      "epoch": 1.0767263427109974,
+      "grad_norm": 4.795405784853634,
+      "learning_rate": 4.398850508504795e-06,
+      "loss": 0.2273,
+      "step": 421
+    },
+    {
+      "epoch": 1.0792838874680306,
+      "grad_norm": 4.137020254719259,
+      "learning_rate": 4.3789142125852015e-06,
+      "loss": 0.2682,
+      "step": 422
+    },
+    {
+      "epoch": 1.081841432225064,
+      "grad_norm": 5.222698810349911,
+      "learning_rate": 4.358987940577631e-06,
+      "loss": 0.2186,
+      "step": 423
+    },
+    {
+      "epoch": 1.0843989769820972,
+      "grad_norm": 5.743293194926402,
+      "learning_rate": 4.339072014078879e-06,
+      "loss": 0.1999,
+      "step": 424
+    },
+    {
+      "epoch": 1.0869565217391304,
+      "grad_norm": 5.052413496789277,
+      "learning_rate": 4.319166754518768e-06,
+      "loss": 0.2047,
+      "step": 425
+    },
+    {
+      "epoch": 1.0895140664961638,
+      "grad_norm": 6.345187477594701,
+      "learning_rate": 4.299272483154963e-06,
+      "loss": 0.314,
+      "step": 426
+    },
+    {
+      "epoch": 1.092071611253197,
+      "grad_norm": 5.440791188116355,
+      "learning_rate": 4.27938952106779e-06,
+      "loss": 0.1874,
+      "step": 427
+    },
+    {
+      "epoch": 1.0946291560102301,
+      "grad_norm": 4.59572793131487,
+      "learning_rate": 4.259518189155049e-06,
+      "loss": 0.2259,
+      "step": 428
+    },
+    {
+      "epoch": 1.0971867007672633,
+      "grad_norm": 7.1276467094843925,
+      "learning_rate": 4.2396588081268355e-06,
+      "loss": 0.2556,
+      "step": 429
+    },
+    {
+      "epoch": 1.0997442455242967,
+      "grad_norm": 6.0356348861027564,
+      "learning_rate": 4.219811698500365e-06,
+      "loss": 0.3379,
+      "step": 430
+    },
+    {
+      "epoch": 1.10230179028133,
+      "grad_norm": 5.364328953584096,
+      "learning_rate": 4.199977180594807e-06,
+      "loss": 0.1789,
+      "step": 431
+    },
+    {
+      "epoch": 1.104859335038363,
+      "grad_norm": 6.395865991308588,
+      "learning_rate": 4.1801555745261025e-06,
+      "loss": 0.3364,
+      "step": 432
+    },
+    {
+      "epoch": 1.1074168797953965,
+      "grad_norm": 5.307623822753343,
+      "learning_rate": 4.160347200201804e-06,
+      "loss": 0.3123,
+      "step": 433
+    },
+    {
+      "epoch": 1.1099744245524297,
+      "grad_norm": 5.8960456968886055,
+      "learning_rate": 4.140552377315918e-06,
+      "loss": 0.3115,
+      "step": 434
+    },
+    {
+      "epoch": 1.1125319693094629,
+      "grad_norm": 5.5625443910431915,
+      "learning_rate": 4.120771425343733e-06,
+      "loss": 0.2276,
+      "step": 435
+    },
+    {
+      "epoch": 1.1150895140664963,
+      "grad_norm": 5.922218273593377,
+      "learning_rate": 4.101004663536675e-06,
+      "loss": 0.2569,
+      "step": 436
+    },
+    {
+      "epoch": 1.1176470588235294,
+      "grad_norm": 5.263570887474491,
+      "learning_rate": 4.081252410917148e-06,
+      "loss": 0.2444,
+      "step": 437
+    },
+    {
+      "epoch": 1.1202046035805626,
+      "grad_norm": 6.181574590354267,
+      "learning_rate": 4.061514986273391e-06,
+      "loss": 0.251,
+      "step": 438
+    },
+    {
+      "epoch": 1.1227621483375958,
+      "grad_norm": 6.1580046071843935,
+      "learning_rate": 4.041792708154321e-06,
+      "loss": 0.2596,
+      "step": 439
+    },
+    {
+      "epoch": 1.1253196930946292,
+      "grad_norm": 5.458696870991704,
+      "learning_rate": 4.022085894864408e-06,
+      "loss": 0.2237,
+      "step": 440
+    },
+    {
+      "epoch": 1.1278772378516624,
+      "grad_norm": 4.70048293838936,
+      "learning_rate": 4.0023948644585294e-06,
+      "loss": 0.2476,
+      "step": 441
+    },
+    {
+      "epoch": 1.1304347826086956,
+      "grad_norm": 4.466773072234314,
+      "learning_rate": 3.982719934736832e-06,
+      "loss": 0.2,
+      "step": 442
+    },
+    {
+      "epoch": 1.132992327365729,
+      "grad_norm": 4.27013775097264,
+      "learning_rate": 3.963061423239612e-06,
+      "loss": 0.1978,
+      "step": 443
+    },
+    {
+      "epoch": 1.1355498721227621,
+      "grad_norm": 5.689677983167152,
+      "learning_rate": 3.943419647242189e-06,
+      "loss": 0.3034,
+      "step": 444
+    },
+    {
+      "epoch": 1.1381074168797953,
+      "grad_norm": 4.25975811296343,
+      "learning_rate": 3.923794923749775e-06,
+      "loss": 0.2187,
+      "step": 445
+    },
+    {
+      "epoch": 1.1406649616368287,
+      "grad_norm": 4.876551747060212,
+      "learning_rate": 3.904187569492373e-06,
+      "loss": 0.257,
+      "step": 446
+    },
+    {
+      "epoch": 1.143222506393862,
+      "grad_norm": 7.4635507464539765,
+      "learning_rate": 3.884597900919656e-06,
+      "loss": 0.2295,
+      "step": 447
+    },
+    {
+      "epoch": 1.145780051150895,
+      "grad_norm": 5.715365590596455,
+      "learning_rate": 3.865026234195863e-06,
+      "loss": 0.2771,
+      "step": 448
+    },
+    {
+      "epoch": 1.1483375959079285,
+      "grad_norm": 5.275576356708176,
+      "learning_rate": 3.8454728851946885e-06,
+      "loss": 0.1754,
+      "step": 449
+    },
+    {
+      "epoch": 1.1508951406649617,
+      "grad_norm": 5.062656319020689,
+      "learning_rate": 3.8259381694942e-06,
+      "loss": 0.2382,
+      "step": 450
+    },
+    {
+      "epoch": 1.1534526854219949,
+      "grad_norm": 5.798164920877668,
+      "learning_rate": 3.806422402371733e-06,
+      "loss": 0.288,
+      "step": 451
+    },
+    {
+      "epoch": 1.156010230179028,
+      "grad_norm": 5.272021205396943,
+      "learning_rate": 3.786925898798801e-06,
+      "loss": 0.2654,
+      "step": 452
+    },
+    {
+      "epoch": 1.1585677749360614,
+      "grad_norm": 4.739102069258541,
+      "learning_rate": 3.767448973436021e-06,
+      "loss": 0.1996,
+      "step": 453
+    },
+    {
+      "epoch": 1.1611253196930946,
+      "grad_norm": 5.933924764643274,
+      "learning_rate": 3.7479919406280334e-06,
+      "loss": 0.3014,
+      "step": 454
+    },
+    {
+      "epoch": 1.1636828644501278,
+      "grad_norm": 5.230108631129741,
+      "learning_rate": 3.728555114398419e-06,
+      "loss": 0.2306,
+      "step": 455
+    },
+    {
+      "epoch": 1.1662404092071612,
+      "grad_norm": 5.1311748950507905,
+      "learning_rate": 3.709138808444641e-06,
+      "loss": 0.2426,
+      "step": 456
+    },
+    {
+      "epoch": 1.1687979539641944,
+      "grad_norm": 4.8518342992156365,
+      "learning_rate": 3.689743336132982e-06,
+      "loss": 0.148,
+      "step": 457
+    },
+    {
+      "epoch": 1.1713554987212276,
+      "grad_norm": 4.141242492696908,
+      "learning_rate": 3.6703690104934806e-06,
+      "loss": 0.1969,
+      "step": 458
+    },
+    {
+      "epoch": 1.1739130434782608,
+      "grad_norm": 5.575504964227216,
+      "learning_rate": 3.6510161442148783e-06,
+      "loss": 0.2847,
+      "step": 459
+    },
+    {
+      "epoch": 1.1764705882352942,
+      "grad_norm": 5.205832366429269,
+      "learning_rate": 3.6316850496395863e-06,
+      "loss": 0.3003,
+      "step": 460
+    },
+    {
+      "epoch": 1.1790281329923273,
+      "grad_norm": 5.640735065432005,
+      "learning_rate": 3.6123760387586265e-06,
+      "loss": 0.2454,
+      "step": 461
+    },
+    {
+      "epoch": 1.1815856777493605,
+      "grad_norm": 5.129282050106555,
+      "learning_rate": 3.5930894232066072e-06,
+      "loss": 0.1335,
+      "step": 462
+    },
+    {
+      "epoch": 1.184143222506394,
+      "grad_norm": 5.640092025988574,
+      "learning_rate": 3.5738255142566912e-06,
+      "loss": 0.2181,
+      "step": 463
+    },
+    {
+      "epoch": 1.186700767263427,
+      "grad_norm": 5.386833673551597,
+      "learning_rate": 3.5545846228155743e-06,
+      "loss": 0.2176,
+      "step": 464
+    },
+    {
+      "epoch": 1.1892583120204603,
+      "grad_norm": 5.704105112383357,
+      "learning_rate": 3.5353670594184623e-06,
+      "loss": 0.2497,
+      "step": 465
+    },
+    {
+      "epoch": 1.1918158567774937,
+      "grad_norm": 5.93251011789455,
+      "learning_rate": 3.516173134224059e-06,
+      "loss": 0.3096,
+      "step": 466
+    },
+    {
+      "epoch": 1.1943734015345269,
+      "grad_norm": 5.135070353719214,
+      "learning_rate": 3.4970031570095707e-06,
+      "loss": 0.2201,
+      "step": 467
+    },
+    {
+      "epoch": 1.19693094629156,
+      "grad_norm": 5.865866790423223,
+      "learning_rate": 3.477857437165694e-06,
+      "loss": 0.2694,
+      "step": 468
+    },
+    {
+      "epoch": 1.1994884910485935,
+      "grad_norm": 6.268703077398429,
+      "learning_rate": 3.458736283691626e-06,
+      "loss": 0.2555,
+      "step": 469
+    },
+    {
+      "epoch": 1.2020460358056266,
+      "grad_norm": 6.588443786412173,
+      "learning_rate": 3.4396400051900846e-06,
+      "loss": 0.2546,
+      "step": 470
+    },
+    {
+      "epoch": 1.2046035805626598,
+      "grad_norm": 5.327149282563148,
+      "learning_rate": 3.4205689098623195e-06,
+      "loss": 0.1783,
+      "step": 471
+    },
+    {
+      "epoch": 1.207161125319693,
+      "grad_norm": 4.958795590477734,
+      "learning_rate": 3.401523305503139e-06,
+      "loss": 0.1755,
+      "step": 472
+    },
+    {
+      "epoch": 1.2097186700767264,
+      "grad_norm": 5.8814512868036735,
+      "learning_rate": 3.3825034994959445e-06,
+      "loss": 0.2751,
+      "step": 473
+    },
+    {
+      "epoch": 1.2122762148337596,
+      "grad_norm": 6.346180998183863,
+      "learning_rate": 3.3635097988077724e-06,
+      "loss": 0.2825,
+      "step": 474
+    },
+    {
+      "epoch": 1.2148337595907928,
+      "grad_norm": 4.824141112150008,
+      "learning_rate": 3.3445425099843343e-06,
+      "loss": 0.193,
+      "step": 475
+    },
+    {
+      "epoch": 1.2173913043478262,
+      "grad_norm": 5.212763506921242,
+      "learning_rate": 3.3256019391450696e-06,
+      "loss": 0.2539,
+      "step": 476
+    },
+    {
+      "epoch": 1.2199488491048593,
+      "grad_norm": 5.731843763443905,
+      "learning_rate": 3.3066883919782116e-06,
+      "loss": 0.2647,
+      "step": 477
+    },
+    {
+      "epoch": 1.2225063938618925,
+      "grad_norm": 5.069916401820976,
+      "learning_rate": 3.287802173735848e-06,
+      "loss": 0.2466,
+      "step": 478
+    },
+    {
+      "epoch": 1.2250639386189257,
+      "grad_norm": 5.351892499599417,
+      "learning_rate": 3.268943589228992e-06,
+      "loss": 0.2369,
+      "step": 479
+    },
+    {
+      "epoch": 1.227621483375959,
+      "grad_norm": 5.226527301587174,
+      "learning_rate": 3.250112942822673e-06,
+      "loss": 0.2063,
+      "step": 480
+    },
+    {
+      "epoch": 1.2301790281329923,
+      "grad_norm": 5.599367983952576,
+      "learning_rate": 3.231310538431015e-06,
+      "loss": 0.3108,
+      "step": 481
+    },
+    {
+      "epoch": 1.2327365728900257,
+      "grad_norm": 4.949232012109662,
+      "learning_rate": 3.212536679512332e-06,
+      "loss": 0.2778,
+      "step": 482
+    },
+    {
+      "epoch": 1.2352941176470589,
+      "grad_norm": 4.692407945943108,
+      "learning_rate": 3.1937916690642356e-06,
+      "loss": 0.2972,
+      "step": 483
+    },
+    {
+      "epoch": 1.237851662404092,
+      "grad_norm": 4.946087662063809,
+      "learning_rate": 3.1750758096187446e-06,
+      "loss": 0.2155,
+      "step": 484
+    },
+    {
+      "epoch": 1.2404092071611252,
+      "grad_norm": 4.8219004232287,
+      "learning_rate": 3.1563894032373977e-06,
+      "loss": 0.22,
+      "step": 485
+    },
+    {
+      "epoch": 1.2429667519181586,
+      "grad_norm": 5.621400027096903,
+      "learning_rate": 3.137732751506376e-06,
+      "loss": 0.2286,
+      "step": 486
+    },
+    {
+      "epoch": 1.2455242966751918,
+      "grad_norm": 5.479705903908969,
+      "learning_rate": 3.1191061555316503e-06,
+      "loss": 0.2534,
+      "step": 487
+    },
+    {
+      "epoch": 1.248081841432225,
+      "grad_norm": 5.794553272942662,
+      "learning_rate": 3.1005099159341044e-06,
+      "loss": 0.2618,
+      "step": 488
+    },
+    {
+      "epoch": 1.2506393861892584,
+      "grad_norm": 6.017425638082386,
+      "learning_rate": 3.08194433284469e-06,
+      "loss": 0.2843,
+      "step": 489
+    },
+    {
+      "epoch": 1.2531969309462916,
+      "grad_norm": 4.5326844807976805,
+      "learning_rate": 3.0634097058995877e-06,
+      "loss": 0.1928,
+      "step": 490
+    },
+    {
+      "epoch": 1.2557544757033248,
+      "grad_norm": 6.4463104104564115,
+      "learning_rate": 3.0449063342353635e-06,
+      "loss": 0.2478,
+      "step": 491
+    },
+    {
+      "epoch": 1.258312020460358,
+      "grad_norm": 4.5887061422293565,
+      "learning_rate": 3.0264345164841426e-06,
+      "loss": 0.204,
+      "step": 492
+    },
+    {
+      "epoch": 1.2608695652173914,
+      "grad_norm": 5.508160093203401,
+      "learning_rate": 3.007994550768793e-06,
+      "loss": 0.2159,
+      "step": 493
+    },
+    {
+      "epoch": 1.2634271099744245,
+      "grad_norm": 4.737111003577314,
+      "learning_rate": 2.989586734698113e-06,
+      "loss": 0.2093,
+      "step": 494
+    },
+    {
+      "epoch": 1.265984654731458,
+      "grad_norm": 4.947584431847608,
+      "learning_rate": 2.971211365362028e-06,
+      "loss": 0.1984,
+      "step": 495
+    },
+    {
+      "epoch": 1.2685421994884911,
+      "grad_norm": 4.800447464091687,
+      "learning_rate": 2.9528687393267865e-06,
+      "loss": 0.2396,
+      "step": 496
+    },
+    {
+      "epoch": 1.2710997442455243,
+      "grad_norm": 5.424287981614177,
+      "learning_rate": 2.934559152630192e-06,
+      "loss": 0.2752,
+      "step": 497
+    },
+    {
+      "epoch": 1.2736572890025575,
+      "grad_norm": 4.885472933623559,
+      "learning_rate": 2.9162829007768103e-06,
+      "loss": 0.2778,
+      "step": 498
+    },
+    {
+      "epoch": 1.2762148337595907,
+      "grad_norm": 5.051334757075423,
+      "learning_rate": 2.898040278733203e-06,
+      "loss": 0.2728,
+      "step": 499
+    },
+    {
+      "epoch": 1.278772378516624,
+      "grad_norm": 5.103121044913984,
+      "learning_rate": 2.879831580923176e-06,
+      "loss": 0.2276,
+      "step": 500
+    },
+    {
+      "epoch": 1.278772378516624,
+      "eval_loss": 0.4383206367492676,
+      "eval_runtime": 0.9493,
+      "eval_samples_per_second": 33.71,
+      "eval_steps_per_second": 8.428,
+      "step": 500
+    },
+    {
+      "epoch": 1.2813299232736572,
+      "grad_norm": 6.302964407184529,
+      "learning_rate": 2.8616571012230134e-06,
+      "loss": 0.324,
+      "step": 501
+    },
+    {
+      "epoch": 1.2838874680306906,
+      "grad_norm": 4.570682847412226,
+      "learning_rate": 2.843517132956742e-06,
+      "loss": 0.2638,
+      "step": 502
+    },
+    {
+      "epoch": 1.2864450127877238,
+      "grad_norm": 5.1557886110649065,
+      "learning_rate": 2.8254119688914017e-06,
+      "loss": 0.2901,
+      "step": 503
+    },
+    {
+      "epoch": 1.289002557544757,
+      "grad_norm": 4.59205694943921,
+      "learning_rate": 2.8073419012323154e-06,
+      "loss": 0.246,
+      "step": 504
+    },
+    {
+      "epoch": 1.2915601023017902,
+      "grad_norm": 5.220862830802117,
+      "learning_rate": 2.789307221618369e-06,
+      "loss": 0.2665,
+      "step": 505
+    },
+    {
+      "epoch": 1.2941176470588236,
+      "grad_norm": 5.6445669568957895,
+      "learning_rate": 2.771308221117309e-06,
+      "loss": 0.2565,
+      "step": 506
+    },
+    {
+      "epoch": 1.2966751918158568,
+      "grad_norm": 5.578642462958587,
+      "learning_rate": 2.7533451902210512e-06,
+      "loss": 0.2805,
+      "step": 507
+    },
+    {
+      "epoch": 1.29923273657289,
+      "grad_norm": 5.9960066898200814,
+      "learning_rate": 2.7354184188409773e-06,
+      "loss": 0.3228,
+      "step": 508
+    },
+    {
+      "epoch": 1.3017902813299234,
+      "grad_norm": 4.257760920803555,
+      "learning_rate": 2.71752819630327e-06,
+      "loss": 0.1833,
+      "step": 509
+    },
+    {
+      "epoch": 1.3043478260869565,
+      "grad_norm": 6.1683068408323605,
+      "learning_rate": 2.6996748113442397e-06,
+      "loss": 0.185,
+      "step": 510
+    },
+    {
+      "epoch": 1.3069053708439897,
+      "grad_norm": 5.042423224754131,
+      "learning_rate": 2.6818585521056573e-06,
+      "loss": 0.241,
+      "step": 511
+    },
+    {
+      "epoch": 1.309462915601023,
+      "grad_norm": 4.727233078424547,
+      "learning_rate": 2.66407970613011e-06,
+      "loss": 0.1559,
+      "step": 512
+    },
+    {
+      "epoch": 1.3120204603580563,
+      "grad_norm": 5.0241427635079505,
+      "learning_rate": 2.646338560356363e-06,
+      "loss": 0.1877,
+      "step": 513
+    },
+    {
+      "epoch": 1.3145780051150895,
+      "grad_norm": 6.126061319891909,
+      "learning_rate": 2.6286354011147252e-06,
+      "loss": 0.2004,
+      "step": 514
+    },
+    {
+      "epoch": 1.317135549872123,
+      "grad_norm": 5.24912663925928,
+      "learning_rate": 2.6109705141224255e-06,
+      "loss": 0.2364,
+      "step": 515
+    },
+    {
+      "epoch": 1.319693094629156,
+      "grad_norm": 6.034124083029447,
+      "learning_rate": 2.593344184479003e-06,
+      "loss": 0.2788,
+      "step": 516
+    },
+    {
+      "epoch": 1.3222506393861893,
+      "grad_norm": 6.608163697606231,
+      "learning_rate": 2.575756696661713e-06,
+      "loss": 0.2416,
+      "step": 517
+    },
+    {
+      "epoch": 1.3248081841432224,
+      "grad_norm": 4.758894846786754,
+      "learning_rate": 2.5582083345209217e-06,
+      "loss": 0.2124,
+      "step": 518
+    },
+    {
+      "epoch": 1.3273657289002558,
+      "grad_norm": 5.719504905312417,
+      "learning_rate": 2.540699381275539e-06,
+      "loss": 0.297,
+      "step": 519
+    },
+    {
+      "epoch": 1.329923273657289,
+      "grad_norm": 5.86724061218003,
+      "learning_rate": 2.5232301195084395e-06,
+      "loss": 0.3234,
+      "step": 520
+    },
+    {
+      "epoch": 1.3324808184143222,
+      "grad_norm": 5.7644650842566945,
+      "learning_rate": 2.5058008311619035e-06,
+      "loss": 0.2615,
+      "step": 521
+    },
+    {
+      "epoch": 1.3350383631713556,
+      "grad_norm": 5.745064713338971,
+      "learning_rate": 2.488411797533064e-06,
+      "loss": 0.2118,
+      "step": 522
+    },
+    {
+      "epoch": 1.3375959079283888,
+      "grad_norm": 4.576869292927031,
+      "learning_rate": 2.4710632992693737e-06,
+      "loss": 0.2278,
+      "step": 523
+    },
+    {
+      "epoch": 1.340153452685422,
+      "grad_norm": 5.037272698300459,
+      "learning_rate": 2.4537556163640726e-06,
+      "loss": 0.236,
+      "step": 524
+    },
+    {
+      "epoch": 1.3427109974424551,
+      "grad_norm": 5.421366472711992,
+      "learning_rate": 2.4364890281516633e-06,
+      "loss": 0.2577,
+      "step": 525
+    },
+    {
+      "epoch": 1.3452685421994885,
+      "grad_norm": 4.866078798724115,
+      "learning_rate": 2.4192638133034074e-06,
+      "loss": 0.2318,
+      "step": 526
+    },
+    {
+      "epoch": 1.3478260869565217,
+      "grad_norm": 4.7964708165794265,
+      "learning_rate": 2.4020802498228333e-06,
+      "loss": 0.2054,
+      "step": 527
+    },
+    {
+      "epoch": 1.350383631713555,
+      "grad_norm": 4.928572310449122,
+      "learning_rate": 2.384938615041238e-06,
+      "loss": 0.2681,
+      "step": 528
+    },
+    {
+      "epoch": 1.3529411764705883,
+      "grad_norm": 4.737745386088544,
+      "learning_rate": 2.3678391856132203e-06,
+      "loss": 0.23,
+      "step": 529
+    },
+    {
+      "epoch": 1.3554987212276215,
+      "grad_norm": 4.546792180996154,
+      "learning_rate": 2.350782237512215e-06,
+      "loss": 0.2158,
+      "step": 530
+    },
+    {
+      "epoch": 1.3580562659846547,
+      "grad_norm": 5.14316611376246,
+      "learning_rate": 2.3337680460260314e-06,
+      "loss": 0.196,
+      "step": 531
+    },
+    {
+      "epoch": 1.3606138107416879,
+      "grad_norm": 5.834646238531157,
+      "learning_rate": 2.316796885752415e-06,
+      "loss": 0.2587,
+      "step": 532
+    },
+    {
+      "epoch": 1.3631713554987213,
+      "grad_norm": 5.780876751105411,
+      "learning_rate": 2.299869030594622e-06,
+      "loss": 0.1942,
+      "step": 533
+    },
+    {
+      "epoch": 1.3657289002557544,
+      "grad_norm": 4.548997507254711,
+      "learning_rate": 2.2829847537569904e-06,
+      "loss": 0.176,
+      "step": 534
+    },
+    {
+      "epoch": 1.3682864450127878,
+      "grad_norm": 5.731302369481431,
+      "learning_rate": 2.266144327740531e-06,
+      "loss": 0.3019,
+      "step": 535
+    },
+    {
+      "epoch": 1.370843989769821,
+      "grad_norm": 5.073679478403773,
+      "learning_rate": 2.2493480243385298e-06,
+      "loss": 0.2065,
+      "step": 536
+    },
+    {
+      "epoch": 1.3734015345268542,
+      "grad_norm": 5.850299158046723,
+      "learning_rate": 2.2325961146321683e-06,
+      "loss": 0.2714,
+      "step": 537
+    },
+    {
+      "epoch": 1.3759590792838874,
+      "grad_norm": 5.7360121373777035,
+      "learning_rate": 2.2158888689861434e-06,
+      "loss": 0.2249,
+      "step": 538
+    },
+    {
+      "epoch": 1.3785166240409208,
+      "grad_norm": 4.828788116522103,
+      "learning_rate": 2.1992265570442974e-06,
+      "loss": 0.2304,
+      "step": 539
+    },
+    {
+      "epoch": 1.381074168797954,
+      "grad_norm": 5.160394779683274,
+      "learning_rate": 2.182609447725279e-06,
+      "loss": 0.2259,
+      "step": 540
+    },
+    {
+      "epoch": 1.3836317135549872,
+      "grad_norm": 5.290912122436337,
+      "learning_rate": 2.1660378092181935e-06,
+      "loss": 0.2104,
+      "step": 541
+    },
+    {
+      "epoch": 1.3861892583120206,
+      "grad_norm": 5.983793408233051,
+      "learning_rate": 2.149511908978275e-06,
+      "loss": 0.2372,
+      "step": 542
+    },
+    {
+      "epoch": 1.3887468030690537,
+      "grad_norm": 5.146327357469625,
+      "learning_rate": 2.1330320137225773e-06,
+      "loss": 0.2594,
+      "step": 543
+    },
+    {
+      "epoch": 1.391304347826087,
+      "grad_norm": 6.1200616170554465,
+      "learning_rate": 2.1165983894256647e-06,
+      "loss": 0.2858,
+      "step": 544
+    },
+    {
+      "epoch": 1.39386189258312,
+      "grad_norm": 5.18913029029246,
+      "learning_rate": 2.100211301315315e-06,
+      "loss": 0.1924,
+      "step": 545
+    },
+    {
+      "epoch": 1.3964194373401535,
+      "grad_norm": 5.568160243169466,
+      "learning_rate": 2.0838710138682412e-06,
+      "loss": 0.193,
+      "step": 546
+    },
+    {
+      "epoch": 1.3989769820971867,
+      "grad_norm": 4.998024067501082,
+      "learning_rate": 2.0675777908058307e-06,
+      "loss": 0.2753,
+      "step": 547
+    },
+    {
+      "epoch": 1.40153452685422,
+      "grad_norm": 5.644184048080268,
+      "learning_rate": 2.051331895089882e-06,
+      "loss": 0.2865,
+      "step": 548
+    },
+    {
+      "epoch": 1.4040920716112533,
+      "grad_norm": 5.40264585527916,
+      "learning_rate": 2.035133588918356e-06,
+      "loss": 0.183,
+      "step": 549
+    },
+    {
+      "epoch": 1.4066496163682864,
+      "grad_norm": 4.477185903868405,
+      "learning_rate": 2.0189831337211573e-06,
+      "loss": 0.1957,
+      "step": 550
+    },
+    {
+      "epoch": 1.4092071611253196,
+      "grad_norm": 5.199329102721851,
+      "learning_rate": 2.0028807901559027e-06,
+      "loss": 0.2067,
+      "step": 551
+    },
+    {
+      "epoch": 1.4117647058823528,
+      "grad_norm": 4.040859440248337,
+      "learning_rate": 1.9868268181037186e-06,
+      "loss": 0.188,
+      "step": 552
+    },
+    {
+      "epoch": 1.4143222506393862,
+      "grad_norm": 4.554963712348274,
+      "learning_rate": 1.970821476665051e-06,
+      "loss": 0.1872,
+      "step": 553
+    },
+    {
+      "epoch": 1.4168797953964194,
+      "grad_norm": 5.430125296226688,
+      "learning_rate": 1.9548650241554812e-06,
+      "loss": 0.2154,
+      "step": 554
+    },
+    {
+      "epoch": 1.4194373401534528,
+      "grad_norm": 5.328931412586536,
+      "learning_rate": 1.9389577181015496e-06,
+      "loss": 0.1755,
+      "step": 555
+    },
+    {
+      "epoch": 1.421994884910486,
+      "grad_norm": 4.249854605239616,
+      "learning_rate": 1.923099815236608e-06,
+      "loss": 0.1593,
+      "step": 556
+    },
+    {
+      "epoch": 1.4245524296675192,
+      "grad_norm": 3.907795738528054,
+      "learning_rate": 1.9072915714966761e-06,
+      "loss": 0.1251,
+      "step": 557
+    },
+    {
+      "epoch": 1.4271099744245523,
+      "grad_norm": 5.529382370073762,
+      "learning_rate": 1.8915332420163074e-06,
+      "loss": 0.1817,
+      "step": 558
+    },
+    {
+      "epoch": 1.4296675191815857,
+      "grad_norm": 5.035709320601147,
+      "learning_rate": 1.8758250811244682e-06,
+      "loss": 0.2079,
+      "step": 559
+    },
+    {
+      "epoch": 1.432225063938619,
+      "grad_norm": 4.581843899498873,
+      "learning_rate": 1.8601673423404449e-06,
+      "loss": 0.2037,
+      "step": 560
+    },
+    {
+      "epoch": 1.434782608695652,
+      "grad_norm": 6.5976247360019045,
+      "learning_rate": 1.8445602783697375e-06,
+      "loss": 0.2877,
+      "step": 561
+    },
+    {
+      "epoch": 1.4373401534526855,
+      "grad_norm": 4.959100730308061,
+      "learning_rate": 1.8290041410999893e-06,
+      "loss": 0.1957,
+      "step": 562
+    },
+    {
+      "epoch": 1.4398976982097187,
+      "grad_norm": 5.169599476739708,
+      "learning_rate": 1.8134991815969238e-06,
+      "loss": 0.2358,
+      "step": 563
+    },
+    {
+      "epoch": 1.4424552429667519,
+      "grad_norm": 5.894324327311631,
+      "learning_rate": 1.798045650100289e-06,
+      "loss": 0.224,
+      "step": 564
+    },
+    {
+      "epoch": 1.445012787723785,
+      "grad_norm": 5.802424880479901,
+      "learning_rate": 1.782643796019814e-06,
+      "loss": 0.2429,
+      "step": 565
+    },
+    {
+      "epoch": 1.4475703324808185,
+      "grad_norm": 4.922016237101965,
+      "learning_rate": 1.7672938679311957e-06,
+      "loss": 0.2266,
+      "step": 566
+    },
+    {
+      "epoch": 1.4501278772378516,
+      "grad_norm": 5.755816715889569,
+      "learning_rate": 1.7519961135720737e-06,
+      "loss": 0.2376,
+      "step": 567
+    },
+    {
+      "epoch": 1.452685421994885,
+      "grad_norm": 5.220912348487513,
+      "learning_rate": 1.736750779838044e-06,
+      "loss": 0.2216,
+      "step": 568
+    },
+    {
+      "epoch": 1.4552429667519182,
+      "grad_norm": 5.210521733834325,
+      "learning_rate": 1.7215581127786624e-06,
+      "loss": 0.2615,
+      "step": 569
+    },
+    {
+      "epoch": 1.4578005115089514,
+      "grad_norm": 5.082754952247903,
+      "learning_rate": 1.7064183575934856e-06,
+      "loss": 0.2395,
+      "step": 570
+    },
+    {
+      "epoch": 1.4603580562659846,
+      "grad_norm": 4.668118713570096,
+      "learning_rate": 1.6913317586281048e-06,
+      "loss": 0.1761,
+      "step": 571
+    },
+    {
+      "epoch": 1.4629156010230178,
+      "grad_norm": 5.582283863667842,
+      "learning_rate": 1.676298559370202e-06,
+      "loss": 0.2342,
+      "step": 572
+    },
+    {
+      "epoch": 1.4654731457800512,
+      "grad_norm": 5.88971284160264,
+      "learning_rate": 1.6613190024456293e-06,
+      "loss": 0.3086,
+      "step": 573
+    },
+    {
+      "epoch": 1.4680306905370843,
+      "grad_norm": 5.053648614417428,
+      "learning_rate": 1.6463933296144863e-06,
+      "loss": 0.2169,
+      "step": 574
+    },
+    {
+      "epoch": 1.4705882352941178,
+      "grad_norm": 5.3893999941494055,
+      "learning_rate": 1.6315217817672142e-06,
+      "loss": 0.2483,
+      "step": 575
+    },
+    {
+      "epoch": 1.473145780051151,
+      "grad_norm": 6.003001211927369,
+      "learning_rate": 1.6167045989207185e-06,
+      "loss": 0.2488,
+      "step": 576
+    },
+    {
+      "epoch": 1.4757033248081841,
+      "grad_norm": 5.8557134317975965,
+      "learning_rate": 1.6019420202144853e-06,
+      "loss": 0.2349,
+      "step": 577
+    },
+    {
+      "epoch": 1.4782608695652173,
+      "grad_norm": 5.263808302802599,
+      "learning_rate": 1.5872342839067305e-06,
+      "loss": 0.1857,
+      "step": 578
+    },
+    {
+      "epoch": 1.4808184143222507,
+      "grad_norm": 4.433242818261617,
+      "learning_rate": 1.5725816273705453e-06,
+      "loss": 0.1888,
+      "step": 579
+    },
+    {
+      "epoch": 1.4833759590792839,
+      "grad_norm": 4.993905414062928,
+      "learning_rate": 1.5579842870900746e-06,
+      "loss": 0.2154,
+      "step": 580
+    },
+    {
+      "epoch": 1.485933503836317,
+      "grad_norm": 4.381712629412114,
+      "learning_rate": 1.5434424986566938e-06,
+      "loss": 0.222,
+      "step": 581
+    },
+    {
+      "epoch": 1.4884910485933505,
+      "grad_norm": 4.174610676856918,
+      "learning_rate": 1.5289564967652033e-06,
+      "loss": 0.1991,
+      "step": 582
+    },
+    {
+      "epoch": 1.4910485933503836,
+      "grad_norm": 4.819672292661032,
+      "learning_rate": 1.5145265152100574e-06,
+      "loss": 0.2425,
+      "step": 583
+    },
+    {
+      "epoch": 1.4936061381074168,
+      "grad_norm": 5.297609781727035,
+      "learning_rate": 1.5001527868815702e-06,
+      "loss": 0.3006,
+      "step": 584
+    },
+    {
+      "epoch": 1.49616368286445,
+      "grad_norm": 5.79687615069299,
+      "learning_rate": 1.4858355437621663e-06,
+      "loss": 0.27,
+      "step": 585
+    },
+    {
+      "epoch": 1.4987212276214834,
+      "grad_norm": 4.762605653912011,
+      "learning_rate": 1.4715750169226417e-06,
+      "loss": 0.2548,
+      "step": 586
+    },
+    {
+      "epoch": 1.5012787723785166,
+      "grad_norm": 5.7198048630752725,
+      "learning_rate": 1.457371436518424e-06,
+      "loss": 0.2594,
+      "step": 587
+    },
+    {
+      "epoch": 1.50383631713555,
+      "grad_norm": 4.802403477803473,
+      "learning_rate": 1.4432250317858675e-06,
+      "loss": 0.19,
+      "step": 588
+    },
+    {
+      "epoch": 1.5063938618925832,
+      "grad_norm": 6.635420914435053,
+      "learning_rate": 1.4291360310385455e-06,
+      "loss": 0.2405,
+      "step": 589
+    },
+    {
+      "epoch": 1.5089514066496164,
+      "grad_norm": 4.539889740139929,
+      "learning_rate": 1.4151046616635727e-06,
+      "loss": 0.1389,
+      "step": 590
+    },
+    {
+      "epoch": 1.5115089514066495,
+      "grad_norm": 5.38811895384659,
+      "learning_rate": 1.4011311501179287e-06,
+      "loss": 0.2662,
+      "step": 591
+    },
+    {
+      "epoch": 1.5140664961636827,
+      "grad_norm": 5.109717895825474,
+      "learning_rate": 1.3872157219248045e-06,
+      "loss": 0.2043,
+      "step": 592
+    },
+    {
+      "epoch": 1.5166240409207161,
+      "grad_norm": 5.0959884186276625,
+      "learning_rate": 1.373358601669973e-06,
+      "loss": 0.1719,
+      "step": 593
+    },
+    {
+      "epoch": 1.5191815856777495,
+      "grad_norm": 5.469942867350689,
+      "learning_rate": 1.3595600129981469e-06,
+      "loss": 0.1644,
+      "step": 594
+    },
+    {
+      "epoch": 1.5217391304347827,
+      "grad_norm": 3.8395720981267885,
+      "learning_rate": 1.3458201786093795e-06,
+      "loss": 0.1706,
+      "step": 595
+    },
+    {
+      "epoch": 1.5242966751918159,
+      "grad_norm": 5.518424328264906,
+      "learning_rate": 1.3321393202554739e-06,
+      "loss": 0.23,
+      "step": 596
+    },
+    {
+      "epoch": 1.526854219948849,
+      "grad_norm": 6.325750802451315,
+      "learning_rate": 1.3185176587363919e-06,
+      "loss": 0.3093,
+      "step": 597
+    },
+    {
+      "epoch": 1.5294117647058822,
+      "grad_norm": 5.306860678214338,
+      "learning_rate": 1.3049554138967052e-06,
+      "loss": 0.1669,
+      "step": 598
+    },
+    {
+      "epoch": 1.5319693094629157,
+      "grad_norm": 5.531828559483773,
+      "learning_rate": 1.2914528046220332e-06,
+      "loss": 0.2521,
+      "step": 599
+    },
+    {
+      "epoch": 1.5345268542199488,
+      "grad_norm": 4.928627869404515,
+      "learning_rate": 1.278010048835523e-06,
+      "loss": 0.1606,
+      "step": 600
+    },
+    {
+      "epoch": 1.5370843989769822,
+      "grad_norm": 4.736176509824429,
+      "learning_rate": 1.2646273634943195e-06,
+      "loss": 0.2382,
+      "step": 601
+    },
+    {
+      "epoch": 1.5396419437340154,
+      "grad_norm": 5.273813096512506,
+      "learning_rate": 1.2513049645860759e-06,
+      "loss": 0.1809,
+      "step": 602
+    },
+    {
+      "epoch": 1.5421994884910486,
+      "grad_norm": 5.824236440729651,
+      "learning_rate": 1.2380430671254618e-06,
+      "loss": 0.2194,
+      "step": 603
+    },
+    {
+      "epoch": 1.5447570332480818,
+      "grad_norm": 5.0892118395960395,
+      "learning_rate": 1.224841885150691e-06,
+      "loss": 0.2326,
+      "step": 604
+    },
+    {
+      "epoch": 1.547314578005115,
+      "grad_norm": 4.776070837603081,
+      "learning_rate": 1.2117016317200702e-06,
+      "loss": 0.202,
+      "step": 605
+    },
+    {
+      "epoch": 1.5498721227621484,
+      "grad_norm": 4.827682184305226,
+      "learning_rate": 1.1986225189085627e-06,
+      "loss": 0.2047,
+      "step": 606
+    },
+    {
+      "epoch": 1.5524296675191815,
+      "grad_norm": 6.354947289868109,
+      "learning_rate": 1.185604757804359e-06,
+      "loss": 0.2489,
+      "step": 607
+    },
+    {
+      "epoch": 1.554987212276215,
+      "grad_norm": 6.389932140184679,
+      "learning_rate": 1.172648558505477e-06,
+      "loss": 0.2039,
+      "step": 608
+    },
+    {
+      "epoch": 1.5575447570332481,
+      "grad_norm": 5.4707664838796815,
+      "learning_rate": 1.1597541301163655e-06,
+      "loss": 0.2242,
+      "step": 609
+    },
+    {
+      "epoch": 1.5601023017902813,
+      "grad_norm": 5.852672190013367,
+      "learning_rate": 1.1469216807445348e-06,
+      "loss": 0.1804,
+      "step": 610
+    },
+    {
+      "epoch": 1.5626598465473145,
+      "grad_norm": 6.351885487364482,
+      "learning_rate": 1.1341514174971907e-06,
+      "loss": 0.2128,
+      "step": 611
+    },
+    {
+      "epoch": 1.5652173913043477,
+      "grad_norm": 5.1770067311650605,
+      "learning_rate": 1.1214435464779006e-06,
+      "loss": 0.2993,
+      "step": 612
+    },
+    {
+      "epoch": 1.567774936061381,
+      "grad_norm": 5.447984746676171,
+      "learning_rate": 1.1087982727832613e-06,
+      "loss": 0.307,
+      "step": 613
+    },
+    {
+      "epoch": 1.5703324808184145,
+      "grad_norm": 5.182727825447981,
+      "learning_rate": 1.0962158004995893e-06,
+      "loss": 0.2687,
+      "step": 614
+    },
+    {
+      "epoch": 1.5728900255754477,
+      "grad_norm": 4.64809564738005,
+      "learning_rate": 1.083696332699628e-06,
+      "loss": 0.1572,
+      "step": 615
+    },
+    {
+      "epoch": 1.5754475703324808,
+      "grad_norm": 4.7325713746238876,
+      "learning_rate": 1.0712400714392723e-06,
+      "loss": 0.18,
+      "step": 616
+    },
+    {
+      "epoch": 1.578005115089514,
+      "grad_norm": 5.269623806320352,
+      "learning_rate": 1.058847217754303e-06,
+      "loss": 0.2037,
+      "step": 617
+    },
+    {
+      "epoch": 1.5805626598465472,
+      "grad_norm": 5.304757901803767,
+      "learning_rate": 1.0465179716571467e-06,
+      "loss": 0.1971,
+      "step": 618
+    },
+    {
+      "epoch": 1.5831202046035806,
+      "grad_norm": 6.728413029180757,
+      "learning_rate": 1.034252532133646e-06,
+      "loss": 0.2494,
+      "step": 619
+    },
+    {
+      "epoch": 1.5856777493606138,
+      "grad_norm": 3.4223539307887454,
+      "learning_rate": 1.0220510971398473e-06,
+      "loss": 0.1613,
+      "step": 620
+    },
+    {
+      "epoch": 1.5882352941176472,
+      "grad_norm": 5.305411837383943,
+      "learning_rate": 1.0099138635988026e-06,
+      "loss": 0.2349,
+      "step": 621
+    },
+    {
+      "epoch": 1.5907928388746804,
+      "grad_norm": 4.856914484373969,
+      "learning_rate": 9.978410273974015e-07,
+      "loss": 0.1723,
+      "step": 622
+    },
+    {
+      "epoch": 1.5933503836317136,
+      "grad_norm": 5.358684756870008,
+      "learning_rate": 9.858327833832004e-07,
+      "loss": 0.2593,
+      "step": 623
+    },
+    {
+      "epoch": 1.5959079283887467,
+      "grad_norm": 4.542731858430077,
+      "learning_rate": 9.738893253612808e-07,
+      "loss": 0.1939,
+      "step": 624
+    },
+    {
+      "epoch": 1.59846547314578,
+      "grad_norm": 4.5148886714925345,
+      "learning_rate": 9.620108460911181e-07,
+      "loss": 0.1901,
+      "step": 625
+    },
+    {
+      "epoch": 1.6010230179028133,
+      "grad_norm": 5.107745783747687,
+      "learning_rate": 9.50197537283481e-07,
+      "loss": 0.2062,
+      "step": 626
+    },
+    {
+      "epoch": 1.6035805626598465,
+      "grad_norm": 6.591147405424452,
+      "learning_rate": 9.384495895973227e-07,
+      "loss": 0.2293,
+      "step": 627
+    },
+    {
+      "epoch": 1.60613810741688,
+      "grad_norm": 5.607830445145212,
+      "learning_rate": 9.267671926367166e-07,
+      "loss": 0.2449,
+      "step": 628
+    },
+    {
+      "epoch": 1.608695652173913,
+      "grad_norm": 5.457362074950791,
+      "learning_rate": 9.151505349477901e-07,
+      "loss": 0.2638,
+      "step": 629
+    },
+    {
+      "epoch": 1.6112531969309463,
+      "grad_norm": 4.821051497667833,
+      "learning_rate": 9.035998040156801e-07,
+      "loss": 0.1902,
+      "step": 630
+    },
+    {
+      "epoch": 1.6138107416879794,
+      "grad_norm": 5.171122943753494,
+      "learning_rate": 8.921151862615091e-07,
+      "loss": 0.1422,
+      "step": 631
+    },
+    {
+      "epoch": 1.6163682864450126,
+      "grad_norm": 6.259146518225325,
+      "learning_rate": 8.806968670393801e-07,
+      "loss": 0.315,
+      "step": 632
+    },
+    {
+      "epoch": 1.618925831202046,
+      "grad_norm": 5.397528592846895,
+      "learning_rate": 8.693450306333818e-07,
+      "loss": 0.21,
+      "step": 633
+    },
+    {
+      "epoch": 1.6214833759590794,
+      "grad_norm": 5.180249332098257,
+      "learning_rate": 8.580598602546109e-07,
+      "loss": 0.2556,
+      "step": 634
+    },
+    {
+      "epoch": 1.6240409207161126,
+      "grad_norm": 4.562634677645485,
+      "learning_rate": 8.4684153803822e-07,
+      "loss": 0.2216,
+      "step": 635
+    },
+    {
+      "epoch": 1.6265984654731458,
+      "grad_norm": 5.390722528674729,
+      "learning_rate": 8.356902450404792e-07,
+      "loss": 0.233,
+      "step": 636
+    },
+    {
+      "epoch": 1.629156010230179,
+      "grad_norm": 4.775191692612564,
+      "learning_rate": 8.246061612358475e-07,
+      "loss": 0.2287,
+      "step": 637
+    },
+    {
+      "epoch": 1.6317135549872122,
+      "grad_norm": 4.852828279434425,
+      "learning_rate": 8.135894655140758e-07,
+      "loss": 0.2191,
+      "step": 638
+    },
+    {
+      "epoch": 1.6342710997442456,
+      "grad_norm": 5.704310246599364,
+      "learning_rate": 8.026403356773161e-07,
+      "loss": 0.2047,
+      "step": 639
+    },
+    {
+      "epoch": 1.6368286445012787,
+      "grad_norm": 4.793807479675851,
+      "learning_rate": 7.91758948437249e-07,
+      "loss": 0.1618,
+      "step": 640
+    },
+    {
+      "epoch": 1.6393861892583121,
+      "grad_norm": 4.754502329271982,
+      "learning_rate": 7.809454794122346e-07,
+      "loss": 0.2781,
+      "step": 641
+    },
+    {
+      "epoch": 1.6419437340153453,
+      "grad_norm": 7.018986504016533,
+      "learning_rate": 7.702001031244816e-07,
+      "loss": 0.2729,
+      "step": 642
+    },
+    {
+      "epoch": 1.6445012787723785,
+      "grad_norm": 4.41916704046766,
+      "learning_rate": 7.595229929972253e-07,
+      "loss": 0.2092,
+      "step": 643
+    },
+    {
+      "epoch": 1.6470588235294117,
+      "grad_norm": 5.336932050022912,
+      "learning_rate": 7.489143213519301e-07,
+      "loss": 0.2267,
+      "step": 644
+    },
+    {
+      "epoch": 1.6496163682864449,
+      "grad_norm": 4.654218849496965,
+      "learning_rate": 7.383742594055077e-07,
+      "loss": 0.2136,
+      "step": 645
+    },
+    {
+      "epoch": 1.6521739130434783,
+      "grad_norm": 5.1623224835718515,
+      "learning_rate": 7.279029772675572e-07,
+      "loss": 0.221,
+      "step": 646
+    },
+    {
+      "epoch": 1.6547314578005117,
+      "grad_norm": 5.686897560893601,
+      "learning_rate": 7.17500643937617e-07,
+      "loss": 0.243,
+      "step": 647
+    },
+    {
+      "epoch": 1.6572890025575449,
+      "grad_norm": 4.738802336926236,
+      "learning_rate": 7.071674273024353e-07,
+      "loss": 0.2059,
+      "step": 648
+    },
+    {
+      "epoch": 1.659846547314578,
+      "grad_norm": 6.305944976355385,
+      "learning_rate": 6.969034941332664e-07,
+      "loss": 0.2147,
+      "step": 649
+    },
+    {
+      "epoch": 1.6624040920716112,
+      "grad_norm": 5.190380929928794,
+      "learning_rate": 6.86709010083172e-07,
+      "loss": 0.1909,
+      "step": 650
+    },
+    {
+      "epoch": 1.6649616368286444,
+      "grad_norm": 5.659990773064172,
+      "learning_rate": 6.765841396843514e-07,
+      "loss": 0.1913,
+      "step": 651
+    },
+    {
+      "epoch": 1.6675191815856778,
+      "grad_norm": 4.841853616494702,
+      "learning_rate": 6.665290463454882e-07,
+      "loss": 0.201,
+      "step": 652
+    },
+    {
+      "epoch": 1.670076726342711,
+      "grad_norm": 6.0589020041416575,
+      "learning_rate": 6.565438923491102e-07,
+      "loss": 0.2908,
+      "step": 653
+    },
+    {
+      "epoch": 1.6726342710997444,
+      "grad_norm": 4.716079545501844,
+      "learning_rate": 6.466288388489689e-07,
+      "loss": 0.2169,
+      "step": 654
+    },
+    {
+      "epoch": 1.6751918158567776,
+      "grad_norm": 4.268196527173102,
+      "learning_rate": 6.367840458674401e-07,
+      "loss": 0.202,
+      "step": 655
+    },
+    {
+      "epoch": 1.6777493606138107,
+      "grad_norm": 5.8944147568821625,
+      "learning_rate": 6.270096722929442e-07,
+      "loss": 0.2586,
+      "step": 656
+    },
+    {
+      "epoch": 1.680306905370844,
+      "grad_norm": 5.564876452450793,
+      "learning_rate": 6.173058758773775e-07,
+      "loss": 0.2391,
+      "step": 657
+    },
+    {
+      "epoch": 1.682864450127877,
+      "grad_norm": 5.003498689191179,
+      "learning_rate": 6.076728132335669e-07,
+      "loss": 0.1756,
+      "step": 658
+    },
+    {
+      "epoch": 1.6854219948849105,
+      "grad_norm": 4.993122576064924,
+      "learning_rate": 5.981106398327463e-07,
+      "loss": 0.1984,
+      "step": 659
+    },
+    {
+      "epoch": 1.6879795396419437,
+      "grad_norm": 5.118160492207756,
+      "learning_rate": 5.886195100020408e-07,
+      "loss": 0.2879,
+      "step": 660
+    },
+    {
+      "epoch": 1.690537084398977,
+      "grad_norm": 5.396969385626712,
+      "learning_rate": 5.7919957692198e-07,
+      "loss": 0.1767,
+      "step": 661
+    },
+    {
+      "epoch": 1.6930946291560103,
+      "grad_norm": 4.709841482974692,
+      "learning_rate": 5.698509926240275e-07,
+      "loss": 0.2265,
+      "step": 662
+    },
+    {
+      "epoch": 1.6956521739130435,
+      "grad_norm": 5.74378925311425,
+      "learning_rate": 5.60573907988124e-07,
+      "loss": 0.2585,
+      "step": 663
+    },
+    {
+      "epoch": 1.6982097186700766,
+      "grad_norm": 5.340843111211026,
+      "learning_rate": 5.513684727402529e-07,
+      "loss": 0.1713,
+      "step": 664
+    },
+    {
+      "epoch": 1.7007672634271098,
+      "grad_norm": 5.095292121289202,
+      "learning_rate": 5.422348354500217e-07,
+      "loss": 0.1969,
+      "step": 665
+    },
+    {
+      "epoch": 1.7033248081841432,
+      "grad_norm": 6.38491199555891,
+      "learning_rate": 5.331731435282705e-07,
+      "loss": 0.1764,
+      "step": 666
+    },
+    {
+      "epoch": 1.7058823529411766,
+      "grad_norm": 6.33773326776292,
+      "learning_rate": 5.241835432246888e-07,
+      "loss": 0.2176,
+      "step": 667
+    },
+    {
+      "epoch": 1.7084398976982098,
+      "grad_norm": 4.747290198021267,
+      "learning_rate": 5.152661796254505e-07,
+      "loss": 0.2194,
+      "step": 668
+    },
+    {
+      "epoch": 1.710997442455243,
+      "grad_norm": 5.180305167036278,
+      "learning_rate": 5.064211966508837e-07,
+      "loss": 0.1838,
+      "step": 669
+    },
+    {
+      "epoch": 1.7135549872122762,
+      "grad_norm": 5.158569886057061,
+      "learning_rate": 4.976487370531352e-07,
+      "loss": 0.1714,
+      "step": 670
+    },
+    {
+      "epoch": 1.7161125319693094,
+      "grad_norm": 6.7699503709640405,
+      "learning_rate": 4.88948942413876e-07,
+      "loss": 0.2794,
+      "step": 671
+    },
+    {
+      "epoch": 1.7186700767263428,
+      "grad_norm": 4.330269645663795,
+      "learning_rate": 4.803219531420128e-07,
+      "loss": 0.1567,
+      "step": 672
+    },
+    {
+      "epoch": 1.721227621483376,
+      "grad_norm": 6.071163454166977,
+      "learning_rate": 4.717679084714222e-07,
+      "loss": 0.2268,
+      "step": 673
+    },
+    {
+      "epoch": 1.7237851662404093,
+      "grad_norm": 4.455492690019572,
+      "learning_rate": 4.6328694645870254e-07,
+      "loss": 0.2092,
+      "step": 674
+    },
+    {
+      "epoch": 1.7263427109974425,
+      "grad_norm": 5.5297942207932165,
+      "learning_rate": 4.5487920398094465e-07,
+      "loss": 0.2037,
+      "step": 675
+    },
+    {
+      "epoch": 1.7289002557544757,
+      "grad_norm": 4.8818412705137115,
+      "learning_rate": 4.46544816733529e-07,
+      "loss": 0.2475,
+      "step": 676
+    },
+    {
+      "epoch": 1.7314578005115089,
+      "grad_norm": 5.627569483357689,
+      "learning_rate": 4.382839192279303e-07,
+      "loss": 0.2585,
+      "step": 677
+    },
+    {
+      "epoch": 1.734015345268542,
+      "grad_norm": 5.1192641037653965,
+      "learning_rate": 4.3009664478954384e-07,
+      "loss": 0.2449,
+      "step": 678
+    },
+    {
+      "epoch": 1.7365728900255755,
+      "grad_norm": 5.286870696275029,
+      "learning_rate": 4.219831255555423e-07,
+      "loss": 0.2004,
+      "step": 679
+    },
+    {
+      "epoch": 1.7391304347826086,
+      "grad_norm": 5.5327277136832675,
+      "learning_rate": 4.139434924727359e-07,
+      "loss": 0.1921,
+      "step": 680
+    },
+    {
+      "epoch": 1.741687979539642,
+      "grad_norm": 4.552608688952828,
+      "learning_rate": 4.059778752954607e-07,
+      "loss": 0.1432,
+      "step": 681
+    },
+    {
+      "epoch": 1.7442455242966752,
+      "grad_norm": 5.216152442361206,
+      "learning_rate": 3.9808640258348686e-07,
+      "loss": 0.1754,
+      "step": 682
+    },
+    {
+      "epoch": 1.7468030690537084,
+      "grad_norm": 5.756818223369158,
+      "learning_rate": 3.9026920169994374e-07,
+      "loss": 0.21,
+      "step": 683
+    },
+    {
+      "epoch": 1.7493606138107416,
+      "grad_norm": 5.29491810051506,
+      "learning_rate": 3.825263988092587e-07,
+      "loss": 0.2228,
+      "step": 684
+    },
+    {
+      "epoch": 1.7519181585677748,
+      "grad_norm": 5.164127220528273,
+      "learning_rate": 3.7485811887512714e-07,
+      "loss": 0.2144,
+      "step": 685
+    },
+    {
+      "epoch": 1.7544757033248082,
+      "grad_norm": 5.314203439099788,
+      "learning_rate": 3.672644856584928e-07,
+      "loss": 0.2911,
+      "step": 686
+    },
+    {
+      "epoch": 1.7570332480818416,
+      "grad_norm": 4.972659541709094,
+      "learning_rate": 3.597456217155526e-07,
+      "loss": 0.1519,
+      "step": 687
+    },
+    {
+      "epoch": 1.7595907928388748,
+      "grad_norm": 5.199722512215739,
+      "learning_rate": 3.523016483957742e-07,
+      "loss": 0.2739,
+      "step": 688
+    },
+    {
+      "epoch": 1.762148337595908,
+      "grad_norm": 5.21999031799005,
+      "learning_rate": 3.4493268583994434e-07,
+      "loss": 0.201,
+      "step": 689
+    },
+    {
+      "epoch": 1.7647058823529411,
+      "grad_norm": 4.364821712930802,
+      "learning_rate": 3.3763885297822153e-07,
+      "loss": 0.1881,
+      "step": 690
+    },
+    {
+      "epoch": 1.7672634271099743,
+      "grad_norm": 5.202401016558692,
+      "learning_rate": 3.3042026752822254e-07,
+      "loss": 0.2125,
+      "step": 691
+    },
+    {
+      "epoch": 1.7698209718670077,
+      "grad_norm": 5.684342025889591,
+      "learning_rate": 3.2327704599312283e-07,
+      "loss": 0.2636,
+      "step": 692
+    },
+    {
+      "epoch": 1.772378516624041,
+      "grad_norm": 5.59338748722469,
+      "learning_rate": 3.16209303659773e-07,
+      "loss": 0.2489,
+      "step": 693
+    },
+    {
+      "epoch": 1.7749360613810743,
+      "grad_norm": 4.23116707170702,
+      "learning_rate": 3.0921715459683753e-07,
+      "loss": 0.153,
+      "step": 694
+    },
+    {
+      "epoch": 1.7774936061381075,
+      "grad_norm": 5.558061912478462,
+      "learning_rate": 3.0230071165295804e-07,
+      "loss": 0.2291,
+      "step": 695
+    },
+    {
+      "epoch": 1.7800511508951407,
+      "grad_norm": 5.197951332859319,
+      "learning_rate": 2.95460086454929e-07,
+      "loss": 0.2267,
+      "step": 696
+    },
+    {
+      "epoch": 1.7826086956521738,
+      "grad_norm": 5.668569655261014,
+      "learning_rate": 2.88695389405898e-07,
+      "loss": 0.2191,
+      "step": 697
+    },
+    {
+      "epoch": 1.785166240409207,
+      "grad_norm": 4.00218617616878,
+      "learning_rate": 2.820067296835799e-07,
+      "loss": 0.1415,
+      "step": 698
+    },
+    {
+      "epoch": 1.7877237851662404,
+      "grad_norm": 5.580094964915473,
+      "learning_rate": 2.753942152385014e-07,
+      "loss": 0.2282,
+      "step": 699
+    },
+    {
+      "epoch": 1.7902813299232738,
+      "grad_norm": 5.157143860228644,
+      "learning_rate": 2.688579527922514e-07,
+      "loss": 0.228,
+      "step": 700
+    },
+    {
+      "epoch": 1.792838874680307,
+      "grad_norm": 6.375062738221611,
+      "learning_rate": 2.6239804783576294e-07,
+      "loss": 0.227,
+      "step": 701
+    },
+    {
+      "epoch": 1.7953964194373402,
+      "grad_norm": 5.7437796904243115,
+      "learning_rate": 2.560146046276135e-07,
+      "loss": 0.2823,
+      "step": 702
+    },
+    {
+      "epoch": 1.7979539641943734,
+      "grad_norm": 5.624690347371044,
+      "learning_rate": 2.4970772619233475e-07,
+      "loss": 0.2042,
+      "step": 703
+    },
+    {
+      "epoch": 1.8005115089514065,
+      "grad_norm": 6.498262506148002,
+      "learning_rate": 2.4347751431875453e-07,
+      "loss": 0.2598,
+      "step": 704
+    },
+    {
+      "epoch": 1.80306905370844,
+      "grad_norm": 6.186981848018963,
+      "learning_rate": 2.373240695583534e-07,
+      "loss": 0.2505,
+      "step": 705
+    },
+    {
+      "epoch": 1.8056265984654731,
+      "grad_norm": 4.425992548953008,
+      "learning_rate": 2.3124749122364286e-07,
+      "loss": 0.2145,
+      "step": 706
+    },
+    {
+      "epoch": 1.8081841432225065,
+      "grad_norm": 4.690718811893992,
+      "learning_rate": 2.2524787738656073e-07,
+      "loss": 0.2106,
+      "step": 707
+    },
+    {
+      "epoch": 1.8107416879795397,
+      "grad_norm": 6.156573679807675,
+      "learning_rate": 2.1932532487688784e-07,
+      "loss": 0.222,
+      "step": 708
+    },
+    {
+      "epoch": 1.813299232736573,
+      "grad_norm": 5.586951769770138,
+      "learning_rate": 2.1347992928068884e-07,
+      "loss": 0.2469,
+      "step": 709
+    },
+    {
+      "epoch": 1.815856777493606,
+      "grad_norm": 5.817151417919622,
+      "learning_rate": 2.0771178493876387e-07,
+      "loss": 0.2514,
+      "step": 710
+    },
+    {
+      "epoch": 1.8184143222506393,
+      "grad_norm": 5.920190459702125,
+      "learning_rate": 2.0202098494513157e-07,
+      "loss": 0.2034,
+      "step": 711
+    },
+    {
+      "epoch": 1.8209718670076727,
+      "grad_norm": 5.8230410453933015,
+      "learning_rate": 1.964076211455246e-07,
+      "loss": 0.2364,
+      "step": 712
+    },
+    {
+      "epoch": 1.8235294117647058,
+      "grad_norm": 5.7392724174161405,
+      "learning_rate": 1.908717841359048e-07,
+      "loss": 0.2135,
+      "step": 713
+    },
+    {
+      "epoch": 1.8260869565217392,
+      "grad_norm": 5.775722433349359,
+      "learning_rate": 1.8541356326100436e-07,
+      "loss": 0.2462,
+      "step": 714
+    },
+    {
+      "epoch": 1.8286445012787724,
+      "grad_norm": 5.591670023546011,
+      "learning_rate": 1.800330466128808e-07,
+      "loss": 0.228,
+      "step": 715
+    },
+    {
+      "epoch": 1.8312020460358056,
+      "grad_norm": 4.711360255327199,
+      "learning_rate": 1.7473032102949983e-07,
+      "loss": 0.2075,
+      "step": 716
+    },
+    {
+      "epoch": 1.8337595907928388,
+      "grad_norm": 5.929377308741077,
+      "learning_rate": 1.695054720933309e-07,
+      "loss": 0.257,
+      "step": 717
+    },
+    {
+      "epoch": 1.836317135549872,
+      "grad_norm": 5.279542039345601,
+      "learning_rate": 1.6435858412996275e-07,
+      "loss": 0.1529,
+      "step": 718
+    },
+    {
+      "epoch": 1.8388746803069054,
+      "grad_norm": 5.1397676431266595,
+      "learning_rate": 1.5928974020674947e-07,
+      "loss": 0.1491,
+      "step": 719
+    },
+    {
+      "epoch": 1.8414322250639388,
+      "grad_norm": 5.512326623165319,
+      "learning_rate": 1.542990221314644e-07,
+      "loss": 0.1774,
+      "step": 720
+    },
+    {
+      "epoch": 1.843989769820972,
+      "grad_norm": 5.66627970312488,
+      "learning_rate": 1.4938651045098174e-07,
+      "loss": 0.2278,
+      "step": 721
+    },
+    {
+      "epoch": 1.8465473145780051,
+      "grad_norm": 4.424031839884274,
+      "learning_rate": 1.445522844499775e-07,
+      "loss": 0.1713,
+      "step": 722
+    },
+    {
+      "epoch": 1.8491048593350383,
+      "grad_norm": 6.625716850892438,
+      "learning_rate": 1.3979642214964728e-07,
+      "loss": 0.254,
+      "step": 723
+    },
+    {
+      "epoch": 1.8516624040920715,
+      "grad_norm": 5.181446281414914,
+      "learning_rate": 1.3511900030644954e-07,
+      "loss": 0.1753,
+      "step": 724
+    },
+    {
+      "epoch": 1.854219948849105,
+      "grad_norm": 6.694348870428919,
+      "learning_rate": 1.3052009441086533e-07,
+      "loss": 0.2535,
+      "step": 725
+    },
+    {
+      "epoch": 1.856777493606138,
+      "grad_norm": 6.09347774518519,
+      "learning_rate": 1.2599977868618052e-07,
+      "loss": 0.2617,
+      "step": 726
+    },
+    {
+      "epoch": 1.8593350383631715,
+      "grad_norm": 6.126918278593983,
+      "learning_rate": 1.215581260872889e-07,
+      "loss": 0.1734,
+      "step": 727
+    },
+    {
+      "epoch": 1.8618925831202047,
+      "grad_norm": 4.9394061667025175,
+      "learning_rate": 1.1719520829951203e-07,
+      "loss": 0.1727,
+      "step": 728
+    },
+    {
+      "epoch": 1.8644501278772379,
+      "grad_norm": 5.679604325861919,
+      "learning_rate": 1.1291109573744574e-07,
+      "loss": 0.2195,
+      "step": 729
+    },
+    {
+      "epoch": 1.867007672634271,
+      "grad_norm": 5.5914554870990525,
+      "learning_rate": 1.087058575438199e-07,
+      "loss": 0.2528,
+      "step": 730
+    },
+    {
+      "epoch": 1.8695652173913042,
+      "grad_norm": 5.877663743949046,
+      "learning_rate": 1.0457956158838545e-07,
+      "loss": 0.2491,
+      "step": 731
+    },
+    {
+      "epoch": 1.8721227621483376,
+      "grad_norm": 3.8851223721617663,
+      "learning_rate": 1.0053227446681912e-07,
+      "loss": 0.1317,
+      "step": 732
+    },
+    {
+      "epoch": 1.8746803069053708,
+      "grad_norm": 5.412598503950858,
+      "learning_rate": 9.656406149964548e-08,
+      "loss": 0.2314,
+      "step": 733
+    },
+    {
+      "epoch": 1.8772378516624042,
+      "grad_norm": 4.937142526302332,
+      "learning_rate": 9.267498673118547e-08,
+      "loss": 0.1964,
+      "step": 734
+    },
+    {
+      "epoch": 1.8797953964194374,
+      "grad_norm": 5.6093925489434735,
+      "learning_rate": 8.886511292852395e-08,
+      "loss": 0.2169,
+      "step": 735
+    },
+    {
+      "epoch": 1.8823529411764706,
+      "grad_norm": 5.544589584646933,
+      "learning_rate": 8.513450158049109e-08,
+      "loss": 0.2262,
+      "step": 736
+    },
+    {
+      "epoch": 1.8849104859335037,
+      "grad_norm": 4.522441136690872,
+      "learning_rate": 8.148321289667749e-08,
+      "loss": 0.1836,
+      "step": 737
+    },
+    {
+      "epoch": 1.887468030690537,
+      "grad_norm": 4.735747110712485,
+      "learning_rate": 7.791130580645623e-08,
+      "loss": 0.2037,
+      "step": 738
+    },
+    {
+      "epoch": 1.8900255754475703,
+      "grad_norm": 5.424105506041224,
+      "learning_rate": 7.441883795803462e-08,
+      "loss": 0.181,
+      "step": 739
+    },
+    {
+      "epoch": 1.8925831202046037,
+      "grad_norm": 5.57291250707421,
+      "learning_rate": 7.100586571752444e-08,
+      "loss": 0.2009,
+      "step": 740
+    },
+    {
+      "epoch": 1.895140664961637,
+      "grad_norm": 4.723852414529119,
+      "learning_rate": 6.767244416802988e-08,
+      "loss": 0.2135,
+      "step": 741
+    },
+    {
+      "epoch": 1.89769820971867,
+      "grad_norm": 4.625168893146648,
+      "learning_rate": 6.441862710876102e-08,
+      "loss": 0.182,
+      "step": 742
+    },
+    {
+      "epoch": 1.9002557544757033,
+      "grad_norm": 5.742248230826967,
+      "learning_rate": 6.124446705416343e-08,
+      "loss": 0.2256,
+      "step": 743
+    },
+    {
+      "epoch": 1.9028132992327365,
+      "grad_norm": 5.446320051746138,
+      "learning_rate": 5.815001523307162e-08,
+      "loss": 0.1844,
+      "step": 744
+    },
+    {
+      "epoch": 1.9053708439897699,
+      "grad_norm": 5.776856594363444,
+      "learning_rate": 5.513532158788193e-08,
+      "loss": 0.2793,
+      "step": 745
+    },
+    {
+      "epoch": 1.907928388746803,
+      "grad_norm": 6.11898381729892,
+      "learning_rate": 5.220043477374759e-08,
+      "loss": 0.228,
+      "step": 746
+    },
+    {
+      "epoch": 1.9104859335038364,
+      "grad_norm": 5.485214100078597,
+      "learning_rate": 4.934540215779271e-08,
+      "loss": 0.2315,
+      "step": 747
+    },
+    {
+      "epoch": 1.9130434782608696,
+      "grad_norm": 5.1671494114407155,
+      "learning_rate": 4.657026981834623e-08,
+      "loss": 0.2191,
+      "step": 748
+    },
+    {
+      "epoch": 1.9156010230179028,
+      "grad_norm": 5.544282249484376,
+      "learning_rate": 4.3875082544201364e-08,
+      "loss": 0.2446,
+      "step": 749
+    },
+    {
+      "epoch": 1.918158567774936,
+      "grad_norm": 4.828585643362734,
+      "learning_rate": 4.125988383388957e-08,
+      "loss": 0.2293,
+      "step": 750
+    },
+    {
+      "epoch": 1.9207161125319692,
+      "grad_norm": 6.32383213055493,
+      "learning_rate": 3.87247158949805e-08,
+      "loss": 0.2332,
+      "step": 751
+    },
+    {
+      "epoch": 1.9232736572890026,
+      "grad_norm": 5.560416321895529,
+      "learning_rate": 3.626961964340203e-08,
+      "loss": 0.2383,
+      "step": 752
+    },
+    {
+      "epoch": 1.9258312020460358,
+      "grad_norm": 5.865352141528642,
+      "learning_rate": 3.389463470277576e-08,
+      "loss": 0.2104,
+      "step": 753
+    },
+    {
+      "epoch": 1.9283887468030692,
+      "grad_norm": 5.341153765939635,
+      "learning_rate": 3.159979940378088e-08,
+      "loss": 0.2128,
+      "step": 754
+    },
+    {
+      "epoch": 1.9309462915601023,
+      "grad_norm": 5.291758252142984,
+      "learning_rate": 2.938515078353521e-08,
+      "loss": 0.1886,
+      "step": 755
+    },
+    {
+      "epoch": 1.9335038363171355,
+      "grad_norm": 5.163447364959534,
+      "learning_rate": 2.725072458499567e-08,
+      "loss": 0.2996,
+      "step": 756
+    },
+    {
+      "epoch": 1.9360613810741687,
+      "grad_norm": 6.676816256652347,
+      "learning_rate": 2.519655525638376e-08,
+      "loss": 0.2902,
+      "step": 757
+    },
+    {
+      "epoch": 1.938618925831202,
+      "grad_norm": 6.29139443707462,
+      "learning_rate": 2.3222675950627106e-08,
+      "loss": 0.1572,
+      "step": 758
+    },
+    {
+      "epoch": 1.9411764705882353,
+      "grad_norm": 4.037989400305309,
+      "learning_rate": 2.1329118524827662e-08,
+      "loss": 0.1417,
+      "step": 759
+    },
+    {
+      "epoch": 1.9437340153452687,
+      "grad_norm": 4.5423783674836695,
+      "learning_rate": 1.9515913539743247e-08,
+      "loss": 0.233,
+      "step": 760
+    },
+    {
+      "epoch": 1.9462915601023019,
+      "grad_norm": 3.859294316006792,
+      "learning_rate": 1.7783090259297918e-08,
+      "loss": 0.1918,
+      "step": 761
+    },
+    {
+      "epoch": 1.948849104859335,
+      "grad_norm": 5.265098117553532,
+      "learning_rate": 1.613067665010959e-08,
+      "loss": 0.1965,
+      "step": 762
+    },
+    {
+      "epoch": 1.9514066496163682,
+      "grad_norm": 4.559779542352908,
+      "learning_rate": 1.4558699381034825e-08,
+      "loss": 0.1888,
+      "step": 763
+    },
+    {
+      "epoch": 1.9539641943734014,
+      "grad_norm": 6.603926656306259,
+      "learning_rate": 1.3067183822742525e-08,
+      "loss": 0.2729,
+      "step": 764
+    },
+    {
+      "epoch": 1.9565217391304348,
+      "grad_norm": 5.966454788112439,
+      "learning_rate": 1.1656154047303691e-08,
+      "loss": 0.1994,
+      "step": 765
+    },
+    {
+      "epoch": 1.959079283887468,
+      "grad_norm": 5.135081772282031,
+      "learning_rate": 1.0325632827801745e-08,
+      "loss": 0.1875,
+      "step": 766
+    },
+    {
+      "epoch": 1.9616368286445014,
+      "grad_norm": 4.6463930984383275,
+      "learning_rate": 9.075641637964483e-09,
+      "loss": 0.213,
+      "step": 767
+    },
+    {
+      "epoch": 1.9641943734015346,
+      "grad_norm": 5.19696586284704,
+      "learning_rate": 7.906200651819907e-09,
+      "loss": 0.188,
+      "step": 768
+    },
+    {
+      "epoch": 1.9667519181585678,
+      "grad_norm": 4.1586996620802585,
+      "learning_rate": 6.817328743368712e-09,
+      "loss": 0.1314,
+      "step": 769
+    },
+    {
+      "epoch": 1.969309462915601,
+      "grad_norm": 6.428796482081383,
+      "learning_rate": 5.809043486279531e-09,
+      "loss": 0.265,
+      "step": 770
+    },
+    {
+      "epoch": 1.9718670076726341,
+      "grad_norm": 6.071118019652648,
+      "learning_rate": 4.881361153606934e-09,
+      "loss": 0.1938,
+      "step": 771
+    },
+    {
+      "epoch": 1.9744245524296675,
+      "grad_norm": 6.004256284137117,
+      "learning_rate": 4.034296717527752e-09,
+      "loss": 0.258,
+      "step": 772
+    },
+    {
+      "epoch": 1.976982097186701,
+      "grad_norm": 4.424170682035935,
+      "learning_rate": 3.2678638490996064e-09,
+      "loss": 0.1944,
+      "step": 773
+    },
+    {
+      "epoch": 1.979539641943734,
+      "grad_norm": 4.5443816733282025,
+      "learning_rate": 2.5820749180388573e-09,
+      "loss": 0.225,
+      "step": 774
+    },
+    {
+      "epoch": 1.9820971867007673,
+      "grad_norm": 6.813889752181503,
+      "learning_rate": 1.976940992523546e-09,
+      "loss": 0.242,
+      "step": 775
+    },
+    {
+      "epoch": 1.9846547314578005,
+      "grad_norm": 6.943362317683093,
+      "learning_rate": 1.4524718390140913e-09,
+      "loss": 0.3044,
+      "step": 776
+    },
+    {
+      "epoch": 1.9872122762148337,
+      "grad_norm": 5.440758095117894,
+      "learning_rate": 1.0086759220934162e-09,
+      "loss": 0.2411,
+      "step": 777
+    },
+    {
+      "epoch": 1.989769820971867,
+      "grad_norm": 3.8628110410030914,
+      "learning_rate": 6.455604043331676e-10,
+      "loss": 0.1659,
+      "step": 778
+    },
+    {
+      "epoch": 1.9923273657289002,
+      "grad_norm": 5.170928482570995,
+      "learning_rate": 3.631311461765874e-10,
+      "loss": 0.1393,
+      "step": 779
+    },
+    {
+      "epoch": 1.9948849104859336,
+      "grad_norm": 5.769015016748028,
+      "learning_rate": 1.6139270584358823e-10,
+      "loss": 0.178,
+      "step": 780
+    },
+    {
+      "epoch": 1.9974424552429668,
+      "grad_norm": 5.070011448256778,
+      "learning_rate": 4.034833925969928e-11,
+      "loss": 0.2705,
+      "step": 781
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 6.222117398419603,
+      "learning_rate": 0.0,
+      "loss": 0.1879,
+      "step": 782
+    },
+    {
+      "epoch": 2.0,
+      "step": 782,
+      "total_flos": 850857689088.0,
+      "train_loss": 0.39088617330012115,
+      "train_runtime": 416.9902,
+      "train_samples_per_second": 15.003,
+      "train_steps_per_second": 1.875
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 782,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 70000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 850857689088.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}