diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,20754 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 2960,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0013513513513513514,
+      "grad_norm": 5.150245666503906,
+      "learning_rate": 0.0,
+      "loss": 4.368040561676025,
+      "step": 1
+    },
+    {
+      "epoch": 0.002702702702702703,
+      "grad_norm": 5.183511734008789,
+      "learning_rate": 6.25e-06,
+      "loss": 4.384284973144531,
+      "step": 2
+    },
+    {
+      "epoch": 0.004054054054054054,
+      "grad_norm": 5.011528491973877,
+      "learning_rate": 1.25e-05,
+      "loss": 4.308831214904785,
+      "step": 3
+    },
+    {
+      "epoch": 0.005405405405405406,
+      "grad_norm": 4.968644142150879,
+      "learning_rate": 1.875e-05,
+      "loss": 4.294878005981445,
+      "step": 4
+    },
+    {
+      "epoch": 0.006756756756756757,
+      "grad_norm": 4.489753723144531,
+      "learning_rate": 2.5e-05,
+      "loss": 4.2850213050842285,
+      "step": 5
+    },
+    {
+      "epoch": 0.008108108108108109,
+      "grad_norm": 3.832670211791992,
+      "learning_rate": 3.125e-05,
+      "loss": 4.17610502243042,
+      "step": 6
+    },
+    {
+      "epoch": 0.00945945945945946,
+      "grad_norm": 3.231327772140503,
+      "learning_rate": 3.75e-05,
+      "loss": 4.0854082107543945,
+      "step": 7
+    },
+    {
+      "epoch": 0.010810810810810811,
+      "grad_norm": 2.834808111190796,
+      "learning_rate": 4.375e-05,
+      "loss": 3.9746999740600586,
+      "step": 8
+    },
+    {
+      "epoch": 0.012162162162162163,
+      "grad_norm": 2.6464433670043945,
+      "learning_rate": 5e-05,
+      "loss": 3.7933273315429688,
+      "step": 9
+    },
+    {
+      "epoch": 0.013513513513513514,
+      "grad_norm": 2.6169960498809814,
+      "learning_rate": 5.6250000000000005e-05,
+      "loss": 3.7386374473571777,
+      "step": 10
+    },
+    {
+      "epoch": 0.014864864864864866,
+      "grad_norm": 2.7303273677825928,
+      "learning_rate": 6.25e-05,
+      "loss": 3.649820327758789,
+      "step": 11
+    },
+    {
+      "epoch": 0.016216216216216217,
+      "grad_norm": 2.342320442199707,
+      "learning_rate": 6.875e-05,
+      "loss": 3.4914913177490234,
+      "step": 12
+    },
+    {
+      "epoch": 0.01756756756756757,
+      "grad_norm": 2.1329972743988037,
+      "learning_rate": 7.5e-05,
+      "loss": 3.3290600776672363,
+      "step": 13
+    },
+    {
+      "epoch": 0.01891891891891892,
+      "grad_norm": 2.039605140686035,
+      "learning_rate": 8.125000000000001e-05,
+      "loss": 3.320183753967285,
+      "step": 14
+    },
+    {
+      "epoch": 0.02027027027027027,
+      "grad_norm": 1.7343803644180298,
+      "learning_rate": 8.75e-05,
+      "loss": 3.17116641998291,
+      "step": 15
+    },
+    {
+      "epoch": 0.021621621621621623,
+      "grad_norm": 1.630784273147583,
+      "learning_rate": 9.375e-05,
+      "loss": 3.1205813884735107,
+      "step": 16
+    },
+    {
+      "epoch": 0.022972972972972974,
+      "grad_norm": 1.6119762659072876,
+      "learning_rate": 0.0001,
+      "loss": 2.8811912536621094,
+      "step": 17
+    },
+    {
+      "epoch": 0.024324324324324326,
+      "grad_norm": 1.475785255432129,
+      "learning_rate": 0.00010625,
+      "loss": 2.9480514526367188,
+      "step": 18
+    },
+    {
+      "epoch": 0.025675675675675677,
+      "grad_norm": 1.4030277729034424,
+      "learning_rate": 0.00011250000000000001,
+      "loss": 2.7605443000793457,
+      "step": 19
+    },
+    {
+      "epoch": 0.02702702702702703,
+      "grad_norm": 1.2840570211410522,
+      "learning_rate": 0.00011875,
+      "loss": 2.697035789489746,
+      "step": 20
+    },
+    {
+      "epoch": 0.02837837837837838,
+      "grad_norm": 1.1836718320846558,
+      "learning_rate": 0.000125,
+      "loss": 2.6500611305236816,
+      "step": 21
+    },
+    {
+      "epoch": 0.02972972972972973,
+      "grad_norm": 1.2846307754516602,
+      "learning_rate": 0.00013125000000000002,
+      "loss": 2.548082113265991,
+      "step": 22
+    },
+    {
+      "epoch": 0.031081081081081083,
+      "grad_norm": 1.4446133375167847,
+      "learning_rate": 0.0001375,
+      "loss": 2.4596900939941406,
+      "step": 23
+    },
+    {
+      "epoch": 0.032432432432432434,
+      "grad_norm": 1.204020380973816,
+      "learning_rate": 0.00014375,
+      "loss": 2.3601913452148438,
+      "step": 24
+    },
+    {
+      "epoch": 0.033783783783783786,
+      "grad_norm": 1.2149330377578735,
+      "learning_rate": 0.00015,
+      "loss": 2.4212570190429688,
+      "step": 25
+    },
+    {
+      "epoch": 0.03513513513513514,
+      "grad_norm": 1.3213460445404053,
+      "learning_rate": 0.00015625,
+      "loss": 2.4033050537109375,
+      "step": 26
+    },
+    {
+      "epoch": 0.03648648648648649,
+      "grad_norm": 1.5141443014144897,
+      "learning_rate": 0.00016250000000000002,
+      "loss": 2.420095443725586,
+      "step": 27
+    },
+    {
+      "epoch": 0.03783783783783784,
+      "grad_norm": 1.0965179204940796,
+      "learning_rate": 0.00016875,
+      "loss": 2.161489963531494,
+      "step": 28
+    },
+    {
+      "epoch": 0.03918918918918919,
+      "grad_norm": 1.2932631969451904,
+      "learning_rate": 0.000175,
+      "loss": 2.1635515689849854,
+      "step": 29
+    },
+    {
+      "epoch": 0.04054054054054054,
+      "grad_norm": 1.1419399976730347,
+      "learning_rate": 0.00018125,
+      "loss": 2.146437406539917,
+      "step": 30
+    },
+    {
+      "epoch": 0.041891891891891894,
+      "grad_norm": 1.6522607803344727,
+      "learning_rate": 0.0001875,
+      "loss": 2.145968198776245,
+      "step": 31
+    },
+    {
+      "epoch": 0.043243243243243246,
+      "grad_norm": 1.112465262413025,
+      "learning_rate": 0.00019375000000000002,
+      "loss": 2.0768933296203613,
+      "step": 32
+    },
+    {
+      "epoch": 0.0445945945945946,
+      "grad_norm": 1.455386757850647,
+      "learning_rate": 0.0002,
+      "loss": 2.112654685974121,
+      "step": 33
+    },
+    {
+      "epoch": 0.04594594594594595,
+      "grad_norm": 1.48453688621521,
+      "learning_rate": 0.00020625,
+      "loss": 2.061765670776367,
+      "step": 34
+    },
+    {
+      "epoch": 0.0472972972972973,
+      "grad_norm": 0.8649069666862488,
+      "learning_rate": 0.0002125,
+      "loss": 2.106442928314209,
+      "step": 35
+    },
+    {
+      "epoch": 0.04864864864864865,
+      "grad_norm": 1.1809029579162598,
+      "learning_rate": 0.00021875,
+      "loss": 1.9744726419448853,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9913383722305298,
+      "learning_rate": 0.00022500000000000002,
+      "loss": 1.9472594261169434,
+      "step": 37
+    },
+    {
+      "epoch": 0.051351351351351354,
+      "grad_norm": 1.2904082536697388,
+      "learning_rate": 0.00023125,
+      "loss": 1.9079926013946533,
+      "step": 38
+    },
+    {
+      "epoch": 0.052702702702702706,
+      "grad_norm": 1.1915795803070068,
+      "learning_rate": 0.0002375,
+      "loss": 1.8431737422943115,
+      "step": 39
+    },
+    {
+      "epoch": 0.05405405405405406,
+      "grad_norm": 1.1800377368927002,
+      "learning_rate": 0.00024375,
+      "loss": 1.8914250135421753,
+      "step": 40
+    },
+    {
+      "epoch": 0.05540540540540541,
+      "grad_norm": 1.3670918941497803,
+      "learning_rate": 0.00025,
+      "loss": 1.9523389339447021,
+      "step": 41
+    },
+    {
+      "epoch": 0.05675675675675676,
+      "grad_norm": 1.6109960079193115,
+      "learning_rate": 0.00025624999999999997,
+      "loss": 1.9186618328094482,
+      "step": 42
+    },
+    {
+      "epoch": 0.05810810810810811,
+      "grad_norm": 1.3506876230239868,
+      "learning_rate": 0.00026250000000000004,
+      "loss": 1.847217082977295,
+      "step": 43
+    },
+    {
+      "epoch": 0.05945945945945946,
+      "grad_norm": 1.093510627746582,
+      "learning_rate": 0.00026875,
+      "loss": 1.8938560485839844,
+      "step": 44
+    },
+    {
+      "epoch": 0.060810810810810814,
+      "grad_norm": 1.3800203800201416,
+      "learning_rate": 0.000275,
+      "loss": 1.795160174369812,
+      "step": 45
+    },
+    {
+      "epoch": 0.062162162162162166,
+      "grad_norm": 1.6892707347869873,
+      "learning_rate": 0.00028125000000000003,
+      "loss": 1.8281978368759155,
+      "step": 46
+    },
+    {
+      "epoch": 0.06351351351351352,
+      "grad_norm": 1.25480055809021,
+      "learning_rate": 0.0002875,
+      "loss": 1.7173875570297241,
+      "step": 47
+    },
+    {
+      "epoch": 0.06486486486486487,
+      "grad_norm": 1.0360338687896729,
+      "learning_rate": 0.00029375,
+      "loss": 1.692659616470337,
+      "step": 48
+    },
+    {
+      "epoch": 0.06621621621621622,
+      "grad_norm": 1.1564885377883911,
+      "learning_rate": 0.0003,
+      "loss": 1.7551082372665405,
+      "step": 49
+    },
+    {
+      "epoch": 0.06756756756756757,
+      "grad_norm": 0.9665911793708801,
+      "learning_rate": 0.00030625000000000004,
+      "loss": 1.705405354499817,
+      "step": 50
+    },
+    {
+      "epoch": 0.06891891891891892,
+      "grad_norm": 0.9467185139656067,
+      "learning_rate": 0.0003125,
+      "loss": 1.6615545749664307,
+      "step": 51
+    },
+    {
+      "epoch": 0.07027027027027027,
+      "grad_norm": 0.9772281646728516,
+      "learning_rate": 0.00031874999999999997,
+      "loss": 1.7018463611602783,
+      "step": 52
+    },
+    {
+      "epoch": 0.07162162162162163,
+      "grad_norm": 1.0119168758392334,
+      "learning_rate": 0.00032500000000000004,
+      "loss": 1.680145502090454,
+      "step": 53
+    },
+    {
+      "epoch": 0.07297297297297298,
+      "grad_norm": 1.0849117040634155,
+      "learning_rate": 0.00033125,
+      "loss": 1.6759579181671143,
+      "step": 54
+    },
+    {
+      "epoch": 0.07432432432432433,
+      "grad_norm": 1.0580090284347534,
+      "learning_rate": 0.0003375,
+      "loss": 1.6751251220703125,
+      "step": 55
+    },
+    {
+      "epoch": 0.07567567567567568,
+      "grad_norm": 2.2117326259613037,
+      "learning_rate": 0.00034375,
+      "loss": 1.8973729610443115,
+      "step": 56
+    },
+    {
+      "epoch": 0.07702702702702703,
+      "grad_norm": 1.1481927633285522,
+      "learning_rate": 0.00035,
+      "loss": 1.6814296245574951,
+      "step": 57
+    },
+    {
+      "epoch": 0.07837837837837838,
+      "grad_norm": 1.5936975479125977,
+      "learning_rate": 0.00035625,
+      "loss": 1.7655748128890991,
+      "step": 58
+    },
+    {
+      "epoch": 0.07972972972972973,
+      "grad_norm": 2.0108320713043213,
+      "learning_rate": 0.0003625,
+      "loss": 1.663893461227417,
+      "step": 59
+    },
+    {
+      "epoch": 0.08108108108108109,
+      "grad_norm": 1.1752243041992188,
+      "learning_rate": 0.00036875000000000005,
+      "loss": 1.6854264736175537,
+      "step": 60
+    },
+    {
+      "epoch": 0.08243243243243244,
+      "grad_norm": 1.1751480102539062,
+      "learning_rate": 0.000375,
+      "loss": 1.590531587600708,
+      "step": 61
+    },
+    {
+      "epoch": 0.08378378378378379,
+      "grad_norm": 1.2346240282058716,
+      "learning_rate": 0.00038124999999999997,
+      "loss": 1.6115528345108032,
+      "step": 62
+    },
+    {
+      "epoch": 0.08513513513513514,
+      "grad_norm": 0.9958378076553345,
+      "learning_rate": 0.00038750000000000004,
+      "loss": 1.589698076248169,
+      "step": 63
+    },
+    {
+      "epoch": 0.08648648648648649,
+      "grad_norm": 1.1387269496917725,
+      "learning_rate": 0.00039375,
+      "loss": 1.6706581115722656,
+      "step": 64
+    },
+    {
+      "epoch": 0.08783783783783784,
+      "grad_norm": 1.3066505193710327,
+      "learning_rate": 0.0004,
+      "loss": 1.6112594604492188,
+      "step": 65
+    },
+    {
+      "epoch": 0.0891891891891892,
+      "grad_norm": 1.0126328468322754,
+      "learning_rate": 0.00040625000000000004,
+      "loss": 1.5779879093170166,
+      "step": 66
+    },
+    {
+      "epoch": 0.09054054054054055,
+      "grad_norm": 1.1921765804290771,
+      "learning_rate": 0.0004125,
+      "loss": 1.6781558990478516,
+      "step": 67
+    },
+    {
+      "epoch": 0.0918918918918919,
+      "grad_norm": 2.6628432273864746,
+      "learning_rate": 0.00041875,
+      "loss": 1.5804617404937744,
+      "step": 68
+    },
+    {
+      "epoch": 0.09324324324324325,
+      "grad_norm": 1.3377033472061157,
+      "learning_rate": 0.000425,
+      "loss": 1.598684310913086,
+      "step": 69
+    },
+    {
+      "epoch": 0.0945945945945946,
+      "grad_norm": 2.759659767150879,
+      "learning_rate": 0.00043125000000000005,
+      "loss": 1.6314220428466797,
+      "step": 70
+    },
+    {
+      "epoch": 0.09594594594594595,
+      "grad_norm": 1.852536916732788,
+      "learning_rate": 0.0004375,
+      "loss": 1.5850790739059448,
+      "step": 71
+    },
+    {
+      "epoch": 0.0972972972972973,
+      "grad_norm": 2.1744425296783447,
+      "learning_rate": 0.00044374999999999997,
+      "loss": 1.615290641784668,
+      "step": 72
+    },
+    {
+      "epoch": 0.09864864864864865,
+      "grad_norm": 1.993006706237793,
+      "learning_rate": 0.00045000000000000004,
+      "loss": 1.6611815690994263,
+      "step": 73
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.2113990783691406,
+      "learning_rate": 0.00045625,
+      "loss": 1.532459020614624,
+      "step": 74
+    },
+    {
+      "epoch": 0.10135135135135136,
+      "grad_norm": 1.4300771951675415,
+      "learning_rate": 0.0004625,
+      "loss": 1.6086134910583496,
+      "step": 75
+    },
+    {
+      "epoch": 0.10270270270270271,
+      "grad_norm": 1.553735375404358,
+      "learning_rate": 0.00046875,
+      "loss": 1.6302762031555176,
+      "step": 76
+    },
+    {
+      "epoch": 0.10405405405405406,
+      "grad_norm": 1.153415322303772,
+      "learning_rate": 0.000475,
+      "loss": 1.5829484462738037,
+      "step": 77
+    },
+    {
+      "epoch": 0.10540540540540541,
+      "grad_norm": 1.509520173072815,
+      "learning_rate": 0.00048125,
+      "loss": 1.4917460680007935,
+      "step": 78
+    },
+    {
+      "epoch": 0.10675675675675676,
+      "grad_norm": 1.0911402702331543,
+      "learning_rate": 0.0004875,
+      "loss": 1.5634562969207764,
+      "step": 79
+    },
+    {
+      "epoch": 0.10810810810810811,
+      "grad_norm": 1.4336016178131104,
+      "learning_rate": 0.00049375,
+      "loss": 1.5104329586029053,
+      "step": 80
+    },
+    {
+      "epoch": 0.10945945945945947,
+      "grad_norm": 1.1370353698730469,
+      "learning_rate": 0.0005,
+      "loss": 1.5228022336959839,
+      "step": 81
+    },
+    {
+      "epoch": 0.11081081081081082,
+      "grad_norm": 1.512107014656067,
+      "learning_rate": 0.0005,
+      "loss": 1.481480598449707,
+      "step": 82
+    },
+    {
+      "epoch": 0.11216216216216217,
+      "grad_norm": 0.9868727326393127,
+      "learning_rate": 0.0005,
+      "loss": 1.5119633674621582,
+      "step": 83
+    },
+    {
+      "epoch": 0.11351351351351352,
+      "grad_norm": 0.853813111782074,
+      "learning_rate": 0.0005,
+      "loss": 1.482405662536621,
+      "step": 84
+    },
+    {
+      "epoch": 0.11486486486486487,
+      "grad_norm": 1.2503330707550049,
+      "learning_rate": 0.0005,
+      "loss": 1.4970108270645142,
+      "step": 85
+    },
+    {
+      "epoch": 0.11621621621621622,
+      "grad_norm": 0.8130242824554443,
+      "learning_rate": 0.0005,
+      "loss": 1.4712978601455688,
+      "step": 86
+    },
+    {
+      "epoch": 0.11756756756756757,
+      "grad_norm": 1.2808117866516113,
+      "learning_rate": 0.0005,
+      "loss": 1.5203461647033691,
+      "step": 87
+    },
+    {
+      "epoch": 0.11891891891891893,
+      "grad_norm": 1.2095636129379272,
+      "learning_rate": 0.0005,
+      "loss": 1.4850702285766602,
+      "step": 88
+    },
+    {
+      "epoch": 0.12027027027027028,
+      "grad_norm": 1.2033348083496094,
+      "learning_rate": 0.0005,
+      "loss": 1.533637523651123,
+      "step": 89
+    },
+    {
+      "epoch": 0.12162162162162163,
+      "grad_norm": 1.0917384624481201,
+      "learning_rate": 0.0005,
+      "loss": 1.5291318893432617,
+      "step": 90
+    },
+    {
+      "epoch": 0.12297297297297298,
+      "grad_norm": 1.1534318923950195,
+      "learning_rate": 0.0005,
+      "loss": 1.4992069005966187,
+      "step": 91
+    },
+    {
+      "epoch": 0.12432432432432433,
+      "grad_norm": 1.1794413328170776,
+      "learning_rate": 0.0005,
+      "loss": 1.4403340816497803,
+      "step": 92
+    },
+    {
+      "epoch": 0.12567567567567567,
+      "grad_norm": 0.9999462366104126,
+      "learning_rate": 0.0005,
+      "loss": 1.4309759140014648,
+      "step": 93
+    },
+    {
+      "epoch": 0.12702702702702703,
+      "grad_norm": 1.244626760482788,
+      "learning_rate": 0.0005,
+      "loss": 1.4745709896087646,
+      "step": 94
+    },
+    {
+      "epoch": 0.12837837837837837,
+      "grad_norm": 0.8035940527915955,
+      "learning_rate": 0.0005,
+      "loss": 1.3985300064086914,
+      "step": 95
+    },
+    {
+      "epoch": 0.12972972972972974,
+      "grad_norm": 1.1606130599975586,
+      "learning_rate": 0.0005,
+      "loss": 1.4202492237091064,
+      "step": 96
+    },
+    {
+      "epoch": 0.13108108108108107,
+      "grad_norm": 0.9500803351402283,
+      "learning_rate": 0.0005,
+      "loss": 1.4417552947998047,
+      "step": 97
+    },
+    {
+      "epoch": 0.13243243243243244,
+      "grad_norm": 0.9110660552978516,
+      "learning_rate": 0.0005,
+      "loss": 1.4029029607772827,
+      "step": 98
+    },
+    {
+      "epoch": 0.13378378378378378,
+      "grad_norm": 1.1800588369369507,
+      "learning_rate": 0.0005,
+      "loss": 1.4258697032928467,
+      "step": 99
+    },
+    {
+      "epoch": 0.13513513513513514,
+      "grad_norm": 0.841268002986908,
+      "learning_rate": 0.0005,
+      "loss": 1.4004430770874023,
+      "step": 100
+    },
+    {
+      "epoch": 0.13648648648648648,
+      "grad_norm": 1.383272647857666,
+      "learning_rate": 0.0005,
+      "loss": 1.3981729745864868,
+      "step": 101
+    },
+    {
+      "epoch": 0.13783783783783785,
+      "grad_norm": 0.9028334021568298,
+      "learning_rate": 0.0005,
+      "loss": 1.4220244884490967,
+      "step": 102
+    },
+    {
+      "epoch": 0.13918918918918918,
+      "grad_norm": 1.230581283569336,
+      "learning_rate": 0.0005,
+      "loss": 1.4683095216751099,
+      "step": 103
+    },
+    {
+      "epoch": 0.14054054054054055,
+      "grad_norm": 1.2321827411651611,
+      "learning_rate": 0.0005,
+      "loss": 1.4666341543197632,
+      "step": 104
+    },
+    {
+      "epoch": 0.14189189189189189,
+      "grad_norm": 0.9921591281890869,
+      "learning_rate": 0.0005,
+      "loss": 1.3607821464538574,
+      "step": 105
+    },
+    {
+      "epoch": 0.14324324324324325,
+      "grad_norm": 1.0495153665542603,
+      "learning_rate": 0.0005,
+      "loss": 1.3390944004058838,
+      "step": 106
+    },
+    {
+      "epoch": 0.1445945945945946,
+      "grad_norm": 0.8765786290168762,
+      "learning_rate": 0.0005,
+      "loss": 1.333587646484375,
+      "step": 107
+    },
+    {
+      "epoch": 0.14594594594594595,
+      "grad_norm": 0.9104813933372498,
+      "learning_rate": 0.0005,
+      "loss": 1.3725016117095947,
+      "step": 108
+    },
+    {
+      "epoch": 0.1472972972972973,
+      "grad_norm": 0.9035435914993286,
+      "learning_rate": 0.0005,
+      "loss": 1.364345669746399,
+      "step": 109
+    },
+    {
+      "epoch": 0.14864864864864866,
+      "grad_norm": 0.9638692736625671,
+      "learning_rate": 0.0005,
+      "loss": 1.4002130031585693,
+      "step": 110
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1487019062042236,
+      "learning_rate": 0.0005,
+      "loss": 1.3804394006729126,
+      "step": 111
+    },
+    {
+      "epoch": 0.15135135135135136,
+      "grad_norm": 1.3018951416015625,
+      "learning_rate": 0.0005,
+      "loss": 1.3483757972717285,
+      "step": 112
+    },
+    {
+      "epoch": 0.1527027027027027,
+      "grad_norm": 0.9842585921287537,
+      "learning_rate": 0.0005,
+      "loss": 1.3281912803649902,
+      "step": 113
+    },
+    {
+      "epoch": 0.15405405405405406,
+      "grad_norm": 1.0535917282104492,
+      "learning_rate": 0.0005,
+      "loss": 1.3392138481140137,
+      "step": 114
+    },
+    {
+      "epoch": 0.1554054054054054,
+      "grad_norm": 1.0435677766799927,
+      "learning_rate": 0.0005,
+      "loss": 1.3455313444137573,
+      "step": 115
+    },
+    {
+      "epoch": 0.15675675675675677,
+      "grad_norm": 0.8129881620407104,
+      "learning_rate": 0.0005,
+      "loss": 1.3637219667434692,
+      "step": 116
+    },
+    {
+      "epoch": 0.1581081081081081,
+      "grad_norm": 1.0077428817749023,
+      "learning_rate": 0.0005,
+      "loss": 1.3303861618041992,
+      "step": 117
+    },
+    {
+      "epoch": 0.15945945945945947,
+      "grad_norm": 0.9998123645782471,
+      "learning_rate": 0.0005,
+      "loss": 1.3071932792663574,
+      "step": 118
+    },
+    {
+      "epoch": 0.1608108108108108,
+      "grad_norm": 0.9819269776344299,
+      "learning_rate": 0.0005,
+      "loss": 1.2924962043762207,
+      "step": 119
+    },
+    {
+      "epoch": 0.16216216216216217,
+      "grad_norm": 1.0808758735656738,
+      "learning_rate": 0.0005,
+      "loss": 1.2974927425384521,
+      "step": 120
+    },
+    {
+      "epoch": 0.1635135135135135,
+      "grad_norm": 0.9639748930931091,
+      "learning_rate": 0.0005,
+      "loss": 1.2871202230453491,
+      "step": 121
+    },
+    {
+      "epoch": 0.16486486486486487,
+      "grad_norm": 1.2070515155792236,
+      "learning_rate": 0.0005,
+      "loss": 1.3268383741378784,
+      "step": 122
+    },
+    {
+      "epoch": 0.1662162162162162,
+      "grad_norm": 0.9738483428955078,
+      "learning_rate": 0.0005,
+      "loss": 1.316206932067871,
+      "step": 123
+    },
+    {
+      "epoch": 0.16756756756756758,
+      "grad_norm": 0.8679337501525879,
+      "learning_rate": 0.0005,
+      "loss": 1.3247699737548828,
+      "step": 124
+    },
+    {
+      "epoch": 0.16891891891891891,
+      "grad_norm": 1.3547780513763428,
+      "learning_rate": 0.0005,
+      "loss": 1.3003835678100586,
+      "step": 125
+    },
+    {
+      "epoch": 0.17027027027027028,
+      "grad_norm": 0.8746247887611389,
+      "learning_rate": 0.0005,
+      "loss": 1.2414307594299316,
+      "step": 126
+    },
+    {
+      "epoch": 0.17162162162162162,
+      "grad_norm": 1.2968964576721191,
+      "learning_rate": 0.0005,
+      "loss": 1.303656816482544,
+      "step": 127
+    },
+    {
+      "epoch": 0.17297297297297298,
+      "grad_norm": 0.839564859867096,
+      "learning_rate": 0.0005,
+      "loss": 1.270803451538086,
+      "step": 128
+    },
+    {
+      "epoch": 0.17432432432432432,
+      "grad_norm": 1.0784152746200562,
+      "learning_rate": 0.0005,
+      "loss": 1.2504408359527588,
+      "step": 129
+    },
+    {
+      "epoch": 0.17567567567567569,
+      "grad_norm": 0.8033078908920288,
+      "learning_rate": 0.0005,
+      "loss": 1.286864995956421,
+      "step": 130
+    },
+    {
+      "epoch": 0.17702702702702702,
+      "grad_norm": 1.0974067449569702,
+      "learning_rate": 0.0005,
+      "loss": 1.2653756141662598,
+      "step": 131
+    },
+    {
+      "epoch": 0.1783783783783784,
+      "grad_norm": 0.7733312845230103,
+      "learning_rate": 0.0005,
+      "loss": 1.2879849672317505,
+      "step": 132
+    },
+    {
+      "epoch": 0.17972972972972973,
+      "grad_norm": 0.9355564713478088,
+      "learning_rate": 0.0005,
+      "loss": 1.2926647663116455,
+      "step": 133
+    },
+    {
+      "epoch": 0.1810810810810811,
+      "grad_norm": 0.7724138498306274,
+      "learning_rate": 0.0005,
+      "loss": 1.2400695085525513,
+      "step": 134
+    },
+    {
+      "epoch": 0.18243243243243243,
+      "grad_norm": 0.7194676995277405,
+      "learning_rate": 0.0005,
+      "loss": 1.257643461227417,
+      "step": 135
+    },
+    {
+      "epoch": 0.1837837837837838,
+      "grad_norm": 0.8569849133491516,
+      "learning_rate": 0.0005,
+      "loss": 1.2732133865356445,
+      "step": 136
+    },
+    {
+      "epoch": 0.18513513513513513,
+      "grad_norm": 0.8977859616279602,
+      "learning_rate": 0.0005,
+      "loss": 1.2396153211593628,
+      "step": 137
+    },
+    {
+      "epoch": 0.1864864864864865,
+      "grad_norm": 0.9070987105369568,
+      "learning_rate": 0.0005,
+      "loss": 1.2443931102752686,
+      "step": 138
+    },
+    {
+      "epoch": 0.18783783783783783,
+      "grad_norm": 0.8386971354484558,
+      "learning_rate": 0.0005,
+      "loss": 1.2399685382843018,
+      "step": 139
+    },
+    {
+      "epoch": 0.1891891891891892,
+      "grad_norm": 0.8963634967803955,
+      "learning_rate": 0.0005,
+      "loss": 1.2585588693618774,
+      "step": 140
+    },
+    {
+      "epoch": 0.19054054054054054,
+      "grad_norm": 1.0193976163864136,
+      "learning_rate": 0.0005,
+      "loss": 1.2888814210891724,
+      "step": 141
+    },
+    {
+      "epoch": 0.1918918918918919,
+      "grad_norm": 0.8432244658470154,
+      "learning_rate": 0.0005,
+      "loss": 1.3043186664581299,
+      "step": 142
+    },
+    {
+      "epoch": 0.19324324324324324,
+      "grad_norm": 1.2887746095657349,
+      "learning_rate": 0.0005,
+      "loss": 1.2295067310333252,
+      "step": 143
+    },
+    {
+      "epoch": 0.1945945945945946,
+      "grad_norm": 0.90013587474823,
+      "learning_rate": 0.0005,
+      "loss": 1.211667537689209,
+      "step": 144
+    },
+    {
+      "epoch": 0.19594594594594594,
+      "grad_norm": 1.0162228345870972,
+      "learning_rate": 0.0005,
+      "loss": 1.2645812034606934,
+      "step": 145
+    },
+    {
+      "epoch": 0.1972972972972973,
+      "grad_norm": 0.9358816146850586,
+      "learning_rate": 0.0005,
+      "loss": 1.2465994358062744,
+      "step": 146
+    },
+    {
+      "epoch": 0.19864864864864865,
+      "grad_norm": 0.8003714680671692,
+      "learning_rate": 0.0005,
+      "loss": 1.294866919517517,
+      "step": 147
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9675484895706177,
+      "learning_rate": 0.0005,
+      "loss": 1.2389206886291504,
+      "step": 148
+    },
+    {
+      "epoch": 0.20135135135135135,
+      "grad_norm": 0.7660173773765564,
+      "learning_rate": 0.0005,
+      "loss": 1.2126613855361938,
+      "step": 149
+    },
+    {
+      "epoch": 0.20270270270270271,
+      "grad_norm": 0.9532003402709961,
+      "learning_rate": 0.0005,
+      "loss": 1.2369132041931152,
+      "step": 150
+    },
+    {
+      "epoch": 0.20405405405405405,
+      "grad_norm": 0.7144771218299866,
+      "learning_rate": 0.0005,
+      "loss": 1.215743899345398,
+      "step": 151
+    },
+    {
+      "epoch": 0.20540540540540542,
+      "grad_norm": 0.9232819080352783,
+      "learning_rate": 0.0005,
+      "loss": 1.2163770198822021,
+      "step": 152
+    },
+    {
+      "epoch": 0.20675675675675675,
+      "grad_norm": 0.8011885285377502,
+      "learning_rate": 0.0005,
+      "loss": 1.2160571813583374,
+      "step": 153
+    },
+    {
+      "epoch": 0.20810810810810812,
+      "grad_norm": 0.8463148474693298,
+      "learning_rate": 0.0005,
+      "loss": 1.2045239210128784,
+      "step": 154
+    },
+    {
+      "epoch": 0.20945945945945946,
+      "grad_norm": 0.9044064283370972,
+      "learning_rate": 0.0005,
+      "loss": 1.2509243488311768,
+      "step": 155
+    },
+    {
+      "epoch": 0.21081081081081082,
+      "grad_norm": 0.8431558012962341,
+      "learning_rate": 0.0005,
+      "loss": 1.2250175476074219,
+      "step": 156
+    },
+    {
+      "epoch": 0.21216216216216216,
+      "grad_norm": 0.8079390525817871,
+      "learning_rate": 0.0005,
+      "loss": 1.1982223987579346,
+      "step": 157
+    },
+    {
+      "epoch": 0.21351351351351353,
+      "grad_norm": 0.7152562141418457,
+      "learning_rate": 0.0005,
+      "loss": 1.2292553186416626,
+      "step": 158
+    },
+    {
+      "epoch": 0.21486486486486486,
+      "grad_norm": 0.6953814029693604,
+      "learning_rate": 0.0005,
+      "loss": 1.1695722341537476,
+      "step": 159
+    },
+    {
+      "epoch": 0.21621621621621623,
+      "grad_norm": 0.7286930680274963,
+      "learning_rate": 0.0005,
+      "loss": 1.230505347251892,
+      "step": 160
+    },
+    {
+      "epoch": 0.21756756756756757,
+      "grad_norm": 0.7171347141265869,
+      "learning_rate": 0.0005,
+      "loss": 1.1747171878814697,
+      "step": 161
+    },
+    {
+      "epoch": 0.21891891891891893,
+      "grad_norm": 0.8353352546691895,
+      "learning_rate": 0.0005,
+      "loss": 1.2184884548187256,
+      "step": 162
+    },
+    {
+      "epoch": 0.22027027027027027,
+      "grad_norm": 0.715570330619812,
+      "learning_rate": 0.0005,
+      "loss": 1.2186400890350342,
+      "step": 163
+    },
+    {
+      "epoch": 0.22162162162162163,
+      "grad_norm": 0.8067787885665894,
+      "learning_rate": 0.0005,
+      "loss": 1.2250593900680542,
+      "step": 164
+    },
+    {
+      "epoch": 0.22297297297297297,
+      "grad_norm": 0.650882363319397,
+      "learning_rate": 0.0005,
+      "loss": 1.1829571723937988,
+      "step": 165
+    },
+    {
+      "epoch": 0.22432432432432434,
+      "grad_norm": 0.8616522550582886,
+      "learning_rate": 0.0005,
+      "loss": 1.2750496864318848,
+      "step": 166
+    },
+    {
+      "epoch": 0.22567567567567567,
+      "grad_norm": 0.6728682518005371,
+      "learning_rate": 0.0005,
+      "loss": 1.2199058532714844,
+      "step": 167
+    },
+    {
+      "epoch": 0.22702702702702704,
+      "grad_norm": 0.8413500189781189,
+      "learning_rate": 0.0005,
+      "loss": 1.1925678253173828,
+      "step": 168
+    },
+    {
+      "epoch": 0.22837837837837838,
+      "grad_norm": 0.6642195582389832,
+      "learning_rate": 0.0005,
+      "loss": 1.1659798622131348,
+      "step": 169
+    },
+    {
+      "epoch": 0.22972972972972974,
+      "grad_norm": 0.916282057762146,
+      "learning_rate": 0.0005,
+      "loss": 1.172922134399414,
+      "step": 170
+    },
+    {
+      "epoch": 0.23108108108108108,
+      "grad_norm": 0.6789361238479614,
+      "learning_rate": 0.0005,
+      "loss": 1.2101984024047852,
+      "step": 171
+    },
+    {
+      "epoch": 0.23243243243243245,
+      "grad_norm": 0.7241368293762207,
+      "learning_rate": 0.0005,
+      "loss": 1.143203854560852,
+      "step": 172
+    },
+    {
+      "epoch": 0.23378378378378378,
+      "grad_norm": 0.6976072788238525,
+      "learning_rate": 0.0005,
+      "loss": 1.1931599378585815,
+      "step": 173
+    },
+    {
+      "epoch": 0.23513513513513515,
+      "grad_norm": 0.8945550322532654,
+      "learning_rate": 0.0005,
+      "loss": 1.2164692878723145,
+      "step": 174
+    },
+    {
+      "epoch": 0.23648648648648649,
+      "grad_norm": 0.6579152345657349,
+      "learning_rate": 0.0005,
+      "loss": 1.1654832363128662,
+      "step": 175
+    },
+    {
+      "epoch": 0.23783783783783785,
+      "grad_norm": 0.744730532169342,
+      "learning_rate": 0.0005,
+      "loss": 1.1667935848236084,
+      "step": 176
+    },
+    {
+      "epoch": 0.2391891891891892,
+      "grad_norm": 0.8223527669906616,
+      "learning_rate": 0.0005,
+      "loss": 1.2500083446502686,
+      "step": 177
+    },
+    {
+      "epoch": 0.24054054054054055,
+      "grad_norm": 0.787659764289856,
+      "learning_rate": 0.0005,
+      "loss": 1.2049771547317505,
+      "step": 178
+    },
+    {
+      "epoch": 0.2418918918918919,
+      "grad_norm": 1.0487968921661377,
+      "learning_rate": 0.0005,
+      "loss": 1.204469919204712,
+      "step": 179
+    },
+    {
+      "epoch": 0.24324324324324326,
+      "grad_norm": 0.9354089498519897,
+      "learning_rate": 0.0005,
+      "loss": 1.1825487613677979,
+      "step": 180
+    },
+    {
+      "epoch": 0.2445945945945946,
+      "grad_norm": 0.8767392635345459,
+      "learning_rate": 0.0005,
+      "loss": 1.1680569648742676,
+      "step": 181
+    },
+    {
+      "epoch": 0.24594594594594596,
+      "grad_norm": 0.7324953675270081,
+      "learning_rate": 0.0005,
+      "loss": 1.1974588632583618,
+      "step": 182
+    },
+    {
+      "epoch": 0.2472972972972973,
+      "grad_norm": 0.8366905450820923,
+      "learning_rate": 0.0005,
+      "loss": 1.150967001914978,
+      "step": 183
+    },
+    {
+      "epoch": 0.24864864864864866,
+      "grad_norm": 0.7201752066612244,
+      "learning_rate": 0.0005,
+      "loss": 1.1760510206222534,
+      "step": 184
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8918378353118896,
+      "learning_rate": 0.0005,
+      "loss": 1.1723861694335938,
+      "step": 185
+    },
+    {
+      "epoch": 0.25135135135135134,
+      "grad_norm": 0.7727148532867432,
+      "learning_rate": 0.0005,
+      "loss": 1.1728715896606445,
+      "step": 186
+    },
+    {
+      "epoch": 0.2527027027027027,
+      "grad_norm": 0.8823726773262024,
+      "learning_rate": 0.0005,
+      "loss": 1.1365349292755127,
+      "step": 187
+    },
+    {
+      "epoch": 0.25405405405405407,
+      "grad_norm": 0.7688813805580139,
+      "learning_rate": 0.0005,
+      "loss": 1.1566078662872314,
+      "step": 188
+    },
+    {
+      "epoch": 0.2554054054054054,
+      "grad_norm": 0.6604582667350769,
+      "learning_rate": 0.0005,
+      "loss": 1.1697120666503906,
+      "step": 189
+    },
+    {
+      "epoch": 0.25675675675675674,
+      "grad_norm": 0.8455849289894104,
+      "learning_rate": 0.0005,
+      "loss": 1.1952283382415771,
+      "step": 190
+    },
+    {
+      "epoch": 0.2581081081081081,
+      "grad_norm": 0.7293181419372559,
+      "learning_rate": 0.0005,
+      "loss": 1.218419075012207,
+      "step": 191
+    },
+    {
+      "epoch": 0.2594594594594595,
+      "grad_norm": 0.6691655516624451,
+      "learning_rate": 0.0005,
+      "loss": 1.1880594491958618,
+      "step": 192
+    },
+    {
+      "epoch": 0.2608108108108108,
+      "grad_norm": 0.6989620923995972,
+      "learning_rate": 0.0005,
+      "loss": 1.1166973114013672,
+      "step": 193
+    },
+    {
+      "epoch": 0.26216216216216215,
+      "grad_norm": 0.7248004078865051,
+      "learning_rate": 0.0005,
+      "loss": 1.1571764945983887,
+      "step": 194
+    },
+    {
+      "epoch": 0.2635135135135135,
+      "grad_norm": 0.734848141670227,
+      "learning_rate": 0.0005,
+      "loss": 1.1902449131011963,
+      "step": 195
+    },
+    {
+      "epoch": 0.2648648648648649,
+      "grad_norm": 0.8059806823730469,
+      "learning_rate": 0.0005,
+      "loss": 1.151974081993103,
+      "step": 196
+    },
+    {
+      "epoch": 0.2662162162162162,
+      "grad_norm": 0.8632883429527283,
+      "learning_rate": 0.0005,
+      "loss": 1.197104811668396,
+      "step": 197
+    },
+    {
+      "epoch": 0.26756756756756755,
+      "grad_norm": 0.7579038143157959,
+      "learning_rate": 0.0005,
+      "loss": 1.167602777481079,
+      "step": 198
+    },
+    {
+      "epoch": 0.2689189189189189,
+      "grad_norm": 0.7441277503967285,
+      "learning_rate": 0.0005,
+      "loss": 1.1435925960540771,
+      "step": 199
+    },
+    {
+      "epoch": 0.2702702702702703,
+      "grad_norm": 0.7189327478408813,
+      "learning_rate": 0.0005,
+      "loss": 1.1217926740646362,
+      "step": 200
+    },
+    {
+      "epoch": 0.2716216216216216,
+      "grad_norm": 0.7653085589408875,
+      "learning_rate": 0.0005,
+      "loss": 1.1414499282836914,
+      "step": 201
+    },
+    {
+      "epoch": 0.27297297297297296,
+      "grad_norm": 0.6481093168258667,
+      "learning_rate": 0.0005,
+      "loss": 1.1885757446289062,
+      "step": 202
+    },
+    {
+      "epoch": 0.2743243243243243,
+      "grad_norm": 0.8565916419029236,
+      "learning_rate": 0.0005,
+      "loss": 1.176697015762329,
+      "step": 203
+    },
+    {
+      "epoch": 0.2756756756756757,
+      "grad_norm": 0.7128346562385559,
+      "learning_rate": 0.0005,
+      "loss": 1.1714215278625488,
+      "step": 204
+    },
+    {
+      "epoch": 0.27702702702702703,
+      "grad_norm": 0.8249174952507019,
+      "learning_rate": 0.0005,
+      "loss": 1.1214120388031006,
+      "step": 205
+    },
+    {
+      "epoch": 0.27837837837837837,
+      "grad_norm": 0.9149993658065796,
+      "learning_rate": 0.0005,
+      "loss": 1.2059478759765625,
+      "step": 206
+    },
+    {
+      "epoch": 0.2797297297297297,
+      "grad_norm": 0.6661089658737183,
+      "learning_rate": 0.0005,
+      "loss": 1.1452312469482422,
+      "step": 207
+    },
+    {
+      "epoch": 0.2810810810810811,
+      "grad_norm": 0.7923280596733093,
+      "learning_rate": 0.0005,
+      "loss": 1.1633195877075195,
+      "step": 208
+    },
+    {
+      "epoch": 0.28243243243243243,
+      "grad_norm": 0.7541139721870422,
+      "learning_rate": 0.0005,
+      "loss": 1.1540124416351318,
+      "step": 209
+    },
+    {
+      "epoch": 0.28378378378378377,
+      "grad_norm": 0.8460253477096558,
+      "learning_rate": 0.0005,
+      "loss": 1.1418213844299316,
+      "step": 210
+    },
+    {
+      "epoch": 0.2851351351351351,
+      "grad_norm": 0.7472702264785767,
+      "learning_rate": 0.0005,
+      "loss": 1.1263737678527832,
+      "step": 211
+    },
+    {
+      "epoch": 0.2864864864864865,
+      "grad_norm": 0.7162389159202576,
+      "learning_rate": 0.0005,
+      "loss": 1.1696337461471558,
+      "step": 212
+    },
+    {
+      "epoch": 0.28783783783783784,
+      "grad_norm": 0.5730604529380798,
+      "learning_rate": 0.0005,
+      "loss": 1.15077805519104,
+      "step": 213
+    },
+    {
+      "epoch": 0.2891891891891892,
+      "grad_norm": 0.6078928709030151,
+      "learning_rate": 0.0005,
+      "loss": 1.1649022102355957,
+      "step": 214
+    },
+    {
+      "epoch": 0.2905405405405405,
+      "grad_norm": 0.5806666016578674,
+      "learning_rate": 0.0005,
+      "loss": 1.1314780712127686,
+      "step": 215
+    },
+    {
+      "epoch": 0.2918918918918919,
+      "grad_norm": 0.5917550921440125,
+      "learning_rate": 0.0005,
+      "loss": 1.1359272003173828,
+      "step": 216
+    },
+    {
+      "epoch": 0.29324324324324325,
+      "grad_norm": 0.7103092074394226,
+      "learning_rate": 0.0005,
+      "loss": 1.1766481399536133,
+      "step": 217
+    },
+    {
+      "epoch": 0.2945945945945946,
+      "grad_norm": 0.592679500579834,
+      "learning_rate": 0.0005,
+      "loss": 1.1225545406341553,
+      "step": 218
+    },
+    {
+      "epoch": 0.2959459459459459,
+      "grad_norm": 0.6089760661125183,
+      "learning_rate": 0.0005,
+      "loss": 1.1148836612701416,
+      "step": 219
+    },
+    {
+      "epoch": 0.2972972972972973,
+      "grad_norm": 0.7156877517700195,
+      "learning_rate": 0.0005,
+      "loss": 1.1501141786575317,
+      "step": 220
+    },
+    {
+      "epoch": 0.29864864864864865,
+      "grad_norm": 0.5425162315368652,
+      "learning_rate": 0.0005,
+      "loss": 1.1127963066101074,
+      "step": 221
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6782895922660828,
+      "learning_rate": 0.0005,
+      "loss": 1.128038763999939,
+      "step": 222
+    },
+    {
+      "epoch": 0.3013513513513513,
+      "grad_norm": 0.8271231651306152,
+      "learning_rate": 0.0005,
+      "loss": 1.1497278213500977,
+      "step": 223
+    },
+    {
+      "epoch": 0.3027027027027027,
+      "grad_norm": 0.627651035785675,
+      "learning_rate": 0.0005,
+      "loss": 1.1400760412216187,
+      "step": 224
+    },
+    {
+      "epoch": 0.30405405405405406,
+      "grad_norm": 0.8547243475914001,
+      "learning_rate": 0.0005,
+      "loss": 1.135500431060791,
+      "step": 225
+    },
+    {
+      "epoch": 0.3054054054054054,
+      "grad_norm": 0.584225594997406,
+      "learning_rate": 0.0005,
+      "loss": 1.1399239301681519,
+      "step": 226
+    },
+    {
+      "epoch": 0.30675675675675673,
+      "grad_norm": 0.6446762681007385,
+      "learning_rate": 0.0005,
+      "loss": 1.144736647605896,
+      "step": 227
+    },
+    {
+      "epoch": 0.3081081081081081,
+      "grad_norm": 0.808337926864624,
+      "learning_rate": 0.0005,
+      "loss": 1.1491795778274536,
+      "step": 228
+    },
+    {
+      "epoch": 0.30945945945945946,
+      "grad_norm": 0.5605192184448242,
+      "learning_rate": 0.0005,
+      "loss": 1.1070001125335693,
+      "step": 229
+    },
+    {
+      "epoch": 0.3108108108108108,
+      "grad_norm": 0.6058110594749451,
+      "learning_rate": 0.0005,
+      "loss": 1.0867992639541626,
+      "step": 230
+    },
+    {
+      "epoch": 0.31216216216216214,
+      "grad_norm": 0.7036818861961365,
+      "learning_rate": 0.0005,
+      "loss": 1.1165096759796143,
+      "step": 231
+    },
+    {
+      "epoch": 0.31351351351351353,
+      "grad_norm": 0.6433873176574707,
+      "learning_rate": 0.0005,
+      "loss": 1.1843355894088745,
+      "step": 232
+    },
+    {
+      "epoch": 0.31486486486486487,
+      "grad_norm": 0.6821860671043396,
+      "learning_rate": 0.0005,
+      "loss": 1.1194504499435425,
+      "step": 233
+    },
+    {
+      "epoch": 0.3162162162162162,
+      "grad_norm": 0.6141608953475952,
+      "learning_rate": 0.0005,
+      "loss": 1.1432108879089355,
+      "step": 234
+    },
+    {
+      "epoch": 0.31756756756756754,
+      "grad_norm": 0.7562606334686279,
+      "learning_rate": 0.0005,
+      "loss": 1.1441978216171265,
+      "step": 235
+    },
+    {
+      "epoch": 0.31891891891891894,
+      "grad_norm": 0.5656039118766785,
+      "learning_rate": 0.0005,
+      "loss": 1.111415147781372,
+      "step": 236
+    },
+    {
+      "epoch": 0.3202702702702703,
+      "grad_norm": 0.624939501285553,
+      "learning_rate": 0.0005,
+      "loss": 1.1203961372375488,
+      "step": 237
+    },
+    {
+      "epoch": 0.3216216216216216,
+      "grad_norm": 0.6211335062980652,
+      "learning_rate": 0.0005,
+      "loss": 1.1381049156188965,
+      "step": 238
+    },
+    {
+      "epoch": 0.32297297297297295,
+      "grad_norm": 0.5427518486976624,
+      "learning_rate": 0.0005,
+      "loss": 1.1267451047897339,
+      "step": 239
+    },
+    {
+      "epoch": 0.32432432432432434,
+      "grad_norm": 0.67227703332901,
+      "learning_rate": 0.0005,
+      "loss": 1.116066336631775,
+      "step": 240
+    },
+    {
+      "epoch": 0.3256756756756757,
+      "grad_norm": 0.4935643970966339,
+      "learning_rate": 0.0005,
+      "loss": 1.1238842010498047,
+      "step": 241
+    },
+    {
+      "epoch": 0.327027027027027,
+      "grad_norm": 0.43769606947898865,
+      "learning_rate": 0.0005,
+      "loss": 1.072867512702942,
+      "step": 242
+    },
+    {
+      "epoch": 0.32837837837837835,
+      "grad_norm": 0.6349326372146606,
+      "learning_rate": 0.0005,
+      "loss": 1.1099973917007446,
+      "step": 243
+    },
+    {
+      "epoch": 0.32972972972972975,
+      "grad_norm": 0.5866597294807434,
+      "learning_rate": 0.0005,
+      "loss": 1.084036111831665,
+      "step": 244
+    },
+    {
+      "epoch": 0.3310810810810811,
+      "grad_norm": 0.5462098717689514,
+      "learning_rate": 0.0005,
+      "loss": 1.1326122283935547,
+      "step": 245
+    },
+    {
+      "epoch": 0.3324324324324324,
+      "grad_norm": 0.619070827960968,
+      "learning_rate": 0.0005,
+      "loss": 1.0964711904525757,
+      "step": 246
+    },
+    {
+      "epoch": 0.33378378378378376,
+      "grad_norm": 0.581043004989624,
+      "learning_rate": 0.0005,
+      "loss": 1.0770080089569092,
+      "step": 247
+    },
+    {
+      "epoch": 0.33513513513513515,
+      "grad_norm": 0.7369904518127441,
+      "learning_rate": 0.0005,
+      "loss": 1.174965739250183,
+      "step": 248
+    },
+    {
+      "epoch": 0.3364864864864865,
+      "grad_norm": 0.5907689332962036,
+      "learning_rate": 0.0005,
+      "loss": 1.110600233078003,
+      "step": 249
+    },
+    {
+      "epoch": 0.33783783783783783,
+      "grad_norm": 0.6739678382873535,
+      "learning_rate": 0.0005,
+      "loss": 1.150277018547058,
+      "step": 250
+    },
+    {
+      "epoch": 0.33918918918918917,
+      "grad_norm": 0.653849184513092,
+      "learning_rate": 0.0005,
+      "loss": 1.1306829452514648,
+      "step": 251
+    },
+    {
+      "epoch": 0.34054054054054056,
+      "grad_norm": 0.5729151368141174,
+      "learning_rate": 0.0005,
+      "loss": 1.115358591079712,
+      "step": 252
+    },
+    {
+      "epoch": 0.3418918918918919,
+      "grad_norm": 0.5855023860931396,
+      "learning_rate": 0.0005,
+      "loss": 1.133919596672058,
+      "step": 253
+    },
+    {
+      "epoch": 0.34324324324324323,
+      "grad_norm": 0.6284984946250916,
+      "learning_rate": 0.0005,
+      "loss": 1.1123967170715332,
+      "step": 254
+    },
+    {
+      "epoch": 0.34459459459459457,
+      "grad_norm": 0.6432620882987976,
+      "learning_rate": 0.0005,
+      "loss": 1.0876009464263916,
+      "step": 255
+    },
+    {
+      "epoch": 0.34594594594594597,
+      "grad_norm": 0.5167033076286316,
+      "learning_rate": 0.0005,
+      "loss": 1.096536636352539,
+      "step": 256
+    },
+    {
+      "epoch": 0.3472972972972973,
+      "grad_norm": 0.539940357208252,
+      "learning_rate": 0.0005,
+      "loss": 1.0787551403045654,
+      "step": 257
+    },
+    {
+      "epoch": 0.34864864864864864,
+      "grad_norm": 0.623758852481842,
+      "learning_rate": 0.0005,
+      "loss": 1.1167761087417603,
+      "step": 258
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5210520625114441,
+      "learning_rate": 0.0005,
+      "loss": 1.1265431642532349,
+      "step": 259
+    },
+    {
+      "epoch": 0.35135135135135137,
+      "grad_norm": 0.6331918239593506,
+      "learning_rate": 0.0005,
+      "loss": 1.1037598848342896,
+      "step": 260
+    },
+    {
+      "epoch": 0.3527027027027027,
+      "grad_norm": 0.6131813526153564,
+      "learning_rate": 0.0005,
+      "loss": 1.1285226345062256,
+      "step": 261
+    },
+    {
+      "epoch": 0.35405405405405405,
+      "grad_norm": 0.5961823463439941,
+      "learning_rate": 0.0005,
+      "loss": 1.121457815170288,
+      "step": 262
+    },
+    {
+      "epoch": 0.3554054054054054,
+      "grad_norm": 0.4831971526145935,
+      "learning_rate": 0.0005,
+      "loss": 1.0331010818481445,
+      "step": 263
+    },
+    {
+      "epoch": 0.3567567567567568,
+      "grad_norm": 0.6117404103279114,
+      "learning_rate": 0.0005,
+      "loss": 1.1207304000854492,
+      "step": 264
+    },
+    {
+      "epoch": 0.3581081081081081,
+      "grad_norm": 0.7258090972900391,
+      "learning_rate": 0.0005,
+      "loss": 1.1407687664031982,
+      "step": 265
+    },
+    {
+      "epoch": 0.35945945945945945,
+      "grad_norm": 0.6041075587272644,
+      "learning_rate": 0.0005,
+      "loss": 1.1474555730819702,
+      "step": 266
+    },
+    {
+      "epoch": 0.3608108108108108,
+      "grad_norm": 0.5688941478729248,
+      "learning_rate": 0.0005,
+      "loss": 1.0870087146759033,
+      "step": 267
+    },
+    {
+      "epoch": 0.3621621621621622,
+      "grad_norm": 0.7341166734695435,
+      "learning_rate": 0.0005,
+      "loss": 1.1139581203460693,
+      "step": 268
+    },
+    {
+      "epoch": 0.3635135135135135,
+      "grad_norm": 0.5610862374305725,
+      "learning_rate": 0.0005,
+      "loss": 1.0804221630096436,
+      "step": 269
+    },
+    {
+      "epoch": 0.36486486486486486,
+      "grad_norm": 0.5837327837944031,
+      "learning_rate": 0.0005,
+      "loss": 1.1093471050262451,
+      "step": 270
+    },
+    {
+      "epoch": 0.3662162162162162,
+      "grad_norm": 0.5809403657913208,
+      "learning_rate": 0.0005,
+      "loss": 1.0860390663146973,
+      "step": 271
+    },
+    {
+      "epoch": 0.3675675675675676,
+      "grad_norm": 0.6011520624160767,
+      "learning_rate": 0.0005,
+      "loss": 1.1017963886260986,
+      "step": 272
+    },
+    {
+      "epoch": 0.3689189189189189,
+      "grad_norm": 0.57268226146698,
+      "learning_rate": 0.0005,
+      "loss": 1.0914065837860107,
+      "step": 273
+    },
+    {
+      "epoch": 0.37027027027027026,
+      "grad_norm": 0.591681182384491,
+      "learning_rate": 0.0005,
+      "loss": 1.1633641719818115,
+      "step": 274
+    },
+    {
+      "epoch": 0.3716216216216216,
+      "grad_norm": 0.6246844530105591,
+      "learning_rate": 0.0005,
+      "loss": 1.0821588039398193,
+      "step": 275
+    },
+    {
+      "epoch": 0.372972972972973,
+      "grad_norm": 0.6925287246704102,
+      "learning_rate": 0.0005,
+      "loss": 1.1347568035125732,
+      "step": 276
+    },
+    {
+      "epoch": 0.37432432432432433,
+      "grad_norm": 0.7299679517745972,
+      "learning_rate": 0.0005,
+      "loss": 1.1008014678955078,
+      "step": 277
+    },
+    {
+      "epoch": 0.37567567567567567,
+      "grad_norm": 0.6282519102096558,
+      "learning_rate": 0.0005,
+      "loss": 1.1192901134490967,
+      "step": 278
+    },
+    {
+      "epoch": 0.377027027027027,
+      "grad_norm": 0.6840739250183105,
+      "learning_rate": 0.0005,
+      "loss": 1.1370794773101807,
+      "step": 279
+    },
+    {
+      "epoch": 0.3783783783783784,
+      "grad_norm": 0.6782492995262146,
+      "learning_rate": 0.0005,
+      "loss": 1.1894714832305908,
+      "step": 280
+    },
+    {
+      "epoch": 0.37972972972972974,
+      "grad_norm": 0.6601359248161316,
+      "learning_rate": 0.0005,
+      "loss": 1.1417372226715088,
+      "step": 281
+    },
+    {
+      "epoch": 0.3810810810810811,
+      "grad_norm": 0.6476040482521057,
+      "learning_rate": 0.0005,
+      "loss": 1.1338998079299927,
+      "step": 282
+    },
+    {
+      "epoch": 0.3824324324324324,
+      "grad_norm": 0.630574643611908,
+      "learning_rate": 0.0005,
+      "loss": 1.0719553232192993,
+      "step": 283
+    },
+    {
+      "epoch": 0.3837837837837838,
+      "grad_norm": 0.7719358205795288,
+      "learning_rate": 0.0005,
+      "loss": 1.1107310056686401,
+      "step": 284
+    },
+    {
+      "epoch": 0.38513513513513514,
+      "grad_norm": 0.5485761165618896,
+      "learning_rate": 0.0005,
+      "loss": 1.0505266189575195,
+      "step": 285
+    },
+    {
+      "epoch": 0.3864864864864865,
+      "grad_norm": 0.6068816781044006,
+      "learning_rate": 0.0005,
+      "loss": 1.094908356666565,
+      "step": 286
+    },
+    {
+      "epoch": 0.3878378378378378,
+      "grad_norm": 0.567980170249939,
+      "learning_rate": 0.0005,
+      "loss": 1.1013671159744263,
+      "step": 287
+    },
+    {
+      "epoch": 0.3891891891891892,
+      "grad_norm": 0.6476532220840454,
+      "learning_rate": 0.0005,
+      "loss": 1.0967718362808228,
+      "step": 288
+    },
+    {
+      "epoch": 0.39054054054054055,
+      "grad_norm": 0.5016531944274902,
+      "learning_rate": 0.0005,
+      "loss": 1.0534034967422485,
+      "step": 289
+    },
+    {
+      "epoch": 0.3918918918918919,
+      "grad_norm": 0.531010091304779,
+      "learning_rate": 0.0005,
+      "loss": 1.0800139904022217,
+      "step": 290
+    },
+    {
+      "epoch": 0.3932432432432432,
+      "grad_norm": 0.5679235458374023,
+      "learning_rate": 0.0005,
+      "loss": 1.100247859954834,
+      "step": 291
+    },
+    {
+      "epoch": 0.3945945945945946,
+      "grad_norm": 0.613233208656311,
+      "learning_rate": 0.0005,
+      "loss": 1.13656747341156,
+      "step": 292
+    },
+    {
+      "epoch": 0.39594594594594595,
+      "grad_norm": 0.5559043884277344,
+      "learning_rate": 0.0005,
+      "loss": 1.0944575071334839,
+      "step": 293
+    },
+    {
+      "epoch": 0.3972972972972973,
+      "grad_norm": 0.6198007464408875,
+      "learning_rate": 0.0005,
+      "loss": 1.1411030292510986,
+      "step": 294
+    },
+    {
+      "epoch": 0.39864864864864863,
+      "grad_norm": 0.5842750668525696,
+      "learning_rate": 0.0005,
+      "loss": 1.108482837677002,
+      "step": 295
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5915307402610779,
+      "learning_rate": 0.0005,
+      "loss": 1.112201452255249,
+      "step": 296
+    },
+    {
+      "epoch": 0.40135135135135136,
+      "grad_norm": 0.6103691458702087,
+      "learning_rate": 0.0005,
+      "loss": 1.0683923959732056,
+      "step": 297
+    },
+    {
+      "epoch": 0.4027027027027027,
+      "grad_norm": 0.6326104998588562,
+      "learning_rate": 0.0005,
+      "loss": 1.08837890625,
+      "step": 298
+    },
+    {
+      "epoch": 0.40405405405405403,
+      "grad_norm": 0.5359740257263184,
+      "learning_rate": 0.0005,
+      "loss": 1.0767585039138794,
+      "step": 299
+    },
+    {
+      "epoch": 0.40540540540540543,
+      "grad_norm": 0.6952090263366699,
+      "learning_rate": 0.0005,
+      "loss": 1.1441268920898438,
+      "step": 300
+    },
+    {
+      "epoch": 0.40675675675675677,
+      "grad_norm": 0.6194038391113281,
+      "learning_rate": 0.0005,
+      "loss": 1.1154354810714722,
+      "step": 301
+    },
+    {
+      "epoch": 0.4081081081081081,
+      "grad_norm": 0.5703895092010498,
+      "learning_rate": 0.0005,
+      "loss": 1.1524344682693481,
+      "step": 302
+    },
+    {
+      "epoch": 0.40945945945945944,
+      "grad_norm": 0.6218256950378418,
+      "learning_rate": 0.0005,
+      "loss": 1.1221665143966675,
+      "step": 303
+    },
+    {
+      "epoch": 0.41081081081081083,
+      "grad_norm": 0.506065845489502,
+      "learning_rate": 0.0005,
+      "loss": 1.090808629989624,
+      "step": 304
+    },
+    {
+      "epoch": 0.41216216216216217,
+      "grad_norm": 0.5985586643218994,
+      "learning_rate": 0.0005,
+      "loss": 1.0938026905059814,
+      "step": 305
+    },
+    {
+      "epoch": 0.4135135135135135,
+      "grad_norm": 0.5803443789482117,
+      "learning_rate": 0.0005,
+      "loss": 1.1075477600097656,
+      "step": 306
+    },
+    {
+      "epoch": 0.41486486486486485,
+      "grad_norm": 0.7027707099914551,
+      "learning_rate": 0.0005,
+      "loss": 1.1155507564544678,
+      "step": 307
+    },
+    {
+      "epoch": 0.41621621621621624,
+      "grad_norm": 0.5953061580657959,
+      "learning_rate": 0.0005,
+      "loss": 1.079146146774292,
+      "step": 308
+    },
+    {
+      "epoch": 0.4175675675675676,
+      "grad_norm": 0.6720386147499084,
+      "learning_rate": 0.0005,
+      "loss": 1.1202231645584106,
+      "step": 309
+    },
+    {
+      "epoch": 0.4189189189189189,
+      "grad_norm": 0.5391613841056824,
+      "learning_rate": 0.0005,
+      "loss": 1.096177339553833,
+      "step": 310
+    },
+    {
+      "epoch": 0.42027027027027025,
+      "grad_norm": 0.6461462378501892,
+      "learning_rate": 0.0005,
+      "loss": 1.072554349899292,
+      "step": 311
+    },
+    {
+      "epoch": 0.42162162162162165,
+      "grad_norm": 0.639743447303772,
+      "learning_rate": 0.0005,
+      "loss": 1.0918264389038086,
+      "step": 312
+    },
+    {
+      "epoch": 0.422972972972973,
+      "grad_norm": 0.642130434513092,
+      "learning_rate": 0.0005,
+      "loss": 1.1245808601379395,
+      "step": 313
+    },
+    {
+      "epoch": 0.4243243243243243,
+      "grad_norm": 0.5202526450157166,
+      "learning_rate": 0.0005,
+      "loss": 1.0721776485443115,
+      "step": 314
+    },
+    {
+      "epoch": 0.42567567567567566,
+      "grad_norm": 0.5655060410499573,
+      "learning_rate": 0.0005,
+      "loss": 1.0408949851989746,
+      "step": 315
+    },
+    {
+      "epoch": 0.42702702702702705,
+      "grad_norm": 0.5834784507751465,
+      "learning_rate": 0.0005,
+      "loss": 1.109189510345459,
+      "step": 316
+    },
+    {
+      "epoch": 0.4283783783783784,
+      "grad_norm": 0.6271800398826599,
+      "learning_rate": 0.0005,
+      "loss": 1.1495022773742676,
+      "step": 317
+    },
+    {
+      "epoch": 0.4297297297297297,
+      "grad_norm": 0.6059660911560059,
+      "learning_rate": 0.0005,
+      "loss": 1.100429892539978,
+      "step": 318
+    },
+    {
+      "epoch": 0.43108108108108106,
+      "grad_norm": 0.644235372543335,
+      "learning_rate": 0.0005,
+      "loss": 1.102222204208374,
+      "step": 319
+    },
+    {
+      "epoch": 0.43243243243243246,
+      "grad_norm": 0.591401994228363,
+      "learning_rate": 0.0005,
+      "loss": 1.0779266357421875,
+      "step": 320
+    },
+    {
+      "epoch": 0.4337837837837838,
+      "grad_norm": 0.5998734831809998,
+      "learning_rate": 0.0005,
+      "loss": 1.0868037939071655,
+      "step": 321
+    },
+    {
+      "epoch": 0.43513513513513513,
+      "grad_norm": 0.6761185526847839,
+      "learning_rate": 0.0005,
+      "loss": 1.1631739139556885,
+      "step": 322
+    },
+    {
+      "epoch": 0.43648648648648647,
+      "grad_norm": 0.5894303917884827,
+      "learning_rate": 0.0005,
+      "loss": 1.1114063262939453,
+      "step": 323
+    },
+    {
+      "epoch": 0.43783783783783786,
+      "grad_norm": 0.5219560861587524,
+      "learning_rate": 0.0005,
+      "loss": 1.1048885583877563,
+      "step": 324
+    },
+    {
+      "epoch": 0.4391891891891892,
+      "grad_norm": 0.5143725872039795,
+      "learning_rate": 0.0005,
+      "loss": 1.0909724235534668,
+      "step": 325
+    },
+    {
+      "epoch": 0.44054054054054054,
+      "grad_norm": 0.49115315079689026,
+      "learning_rate": 0.0005,
+      "loss": 1.0430961847305298,
+      "step": 326
+    },
+    {
+      "epoch": 0.4418918918918919,
+      "grad_norm": 0.7028875350952148,
+      "learning_rate": 0.0005,
+      "loss": 1.1015006303787231,
+      "step": 327
+    },
+    {
+      "epoch": 0.44324324324324327,
+      "grad_norm": 0.5508009195327759,
+      "learning_rate": 0.0005,
+      "loss": 1.0856225490570068,
+      "step": 328
+    },
+    {
+      "epoch": 0.4445945945945946,
+      "grad_norm": 0.67892986536026,
+      "learning_rate": 0.0005,
+      "loss": 1.106320858001709,
+      "step": 329
+    },
+    {
+      "epoch": 0.44594594594594594,
+      "grad_norm": 0.6417338848114014,
+      "learning_rate": 0.0005,
+      "loss": 1.1312164068222046,
+      "step": 330
+    },
+    {
+      "epoch": 0.4472972972972973,
+      "grad_norm": 0.6246355175971985,
+      "learning_rate": 0.0005,
+      "loss": 1.1064890623092651,
+      "step": 331
+    },
+    {
+      "epoch": 0.4486486486486487,
+      "grad_norm": 0.6260076761245728,
+      "learning_rate": 0.0005,
+      "loss": 1.0983638763427734,
+      "step": 332
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5905974507331848,
+      "learning_rate": 0.0005,
+      "loss": 1.1219292879104614,
+      "step": 333
+    },
+    {
+      "epoch": 0.45135135135135135,
+      "grad_norm": 0.5573140978813171,
+      "learning_rate": 0.0005,
+      "loss": 1.0991413593292236,
+      "step": 334
+    },
+    {
+      "epoch": 0.4527027027027027,
+      "grad_norm": 0.6058799028396606,
+      "learning_rate": 0.0005,
+      "loss": 1.0862925052642822,
+      "step": 335
+    },
+    {
+      "epoch": 0.4540540540540541,
+      "grad_norm": 0.5283327102661133,
+      "learning_rate": 0.0005,
+      "loss": 1.068895697593689,
+      "step": 336
+    },
+    {
+      "epoch": 0.4554054054054054,
+      "grad_norm": 0.5668305158615112,
+      "learning_rate": 0.0005,
+      "loss": 1.0841717720031738,
+      "step": 337
+    },
+    {
+      "epoch": 0.45675675675675675,
+      "grad_norm": 0.6177982687950134,
+      "learning_rate": 0.0005,
+      "loss": 1.0897855758666992,
+      "step": 338
+    },
+    {
+      "epoch": 0.4581081081081081,
+      "grad_norm": 0.5476327538490295,
+      "learning_rate": 0.0005,
+      "loss": 1.065648078918457,
+      "step": 339
+    },
+    {
+      "epoch": 0.4594594594594595,
+      "grad_norm": 0.5164141058921814,
+      "learning_rate": 0.0005,
+      "loss": 1.050333857536316,
+      "step": 340
+    },
+    {
+      "epoch": 0.4608108108108108,
+      "grad_norm": 0.5637829303741455,
+      "learning_rate": 0.0005,
+      "loss": 1.0772809982299805,
+      "step": 341
+    },
+    {
+      "epoch": 0.46216216216216216,
+      "grad_norm": 0.5503175258636475,
+      "learning_rate": 0.0005,
+      "loss": 1.125168800354004,
+      "step": 342
+    },
+    {
+      "epoch": 0.4635135135135135,
+      "grad_norm": 0.5408004522323608,
+      "learning_rate": 0.0005,
+      "loss": 1.0474896430969238,
+      "step": 343
+    },
+    {
+      "epoch": 0.4648648648648649,
+      "grad_norm": 0.5610090494155884,
+      "learning_rate": 0.0005,
+      "loss": 1.0560894012451172,
+      "step": 344
+    },
+    {
+      "epoch": 0.46621621621621623,
+      "grad_norm": 0.5408453941345215,
+      "learning_rate": 0.0005,
+      "loss": 1.0779005289077759,
+      "step": 345
+    },
+    {
+      "epoch": 0.46756756756756757,
+      "grad_norm": 0.5340505242347717,
+      "learning_rate": 0.0005,
+      "loss": 1.096764326095581,
+      "step": 346
+    },
+    {
+      "epoch": 0.4689189189189189,
+      "grad_norm": 0.6074935793876648,
+      "learning_rate": 0.0005,
+      "loss": 1.0522434711456299,
+      "step": 347
+    },
+    {
+      "epoch": 0.4702702702702703,
+      "grad_norm": 0.562341570854187,
+      "learning_rate": 0.0005,
+      "loss": 1.0841513872146606,
+      "step": 348
+    },
+    {
+      "epoch": 0.47162162162162163,
+      "grad_norm": 0.5711448788642883,
+      "learning_rate": 0.0005,
+      "loss": 1.0946356058120728,
+      "step": 349
+    },
+    {
+      "epoch": 0.47297297297297297,
+      "grad_norm": 0.48436805605888367,
+      "learning_rate": 0.0005,
+      "loss": 1.068122148513794,
+      "step": 350
+    },
+    {
+      "epoch": 0.4743243243243243,
+      "grad_norm": 0.4889150559902191,
+      "learning_rate": 0.0005,
+      "loss": 1.0663830041885376,
+      "step": 351
+    },
+    {
+      "epoch": 0.4756756756756757,
+      "grad_norm": 0.5094220638275146,
+      "learning_rate": 0.0005,
+      "loss": 1.0416780710220337,
+      "step": 352
+    },
+    {
+      "epoch": 0.47702702702702704,
+      "grad_norm": 0.5173611640930176,
+      "learning_rate": 0.0005,
+      "loss": 1.0819835662841797,
+      "step": 353
+    },
+    {
+      "epoch": 0.4783783783783784,
+      "grad_norm": 0.5650801062583923,
+      "learning_rate": 0.0005,
+      "loss": 1.066215991973877,
+      "step": 354
+    },
+    {
+      "epoch": 0.4797297297297297,
+      "grad_norm": 0.5343220233917236,
+      "learning_rate": 0.0005,
+      "loss": 1.136378288269043,
+      "step": 355
+    },
+    {
+      "epoch": 0.4810810810810811,
+      "grad_norm": 0.4983275532722473,
+      "learning_rate": 0.0005,
+      "loss": 1.0694124698638916,
+      "step": 356
+    },
+    {
+      "epoch": 0.48243243243243245,
+      "grad_norm": 0.5148589611053467,
+      "learning_rate": 0.0005,
+      "loss": 1.0821418762207031,
+      "step": 357
+    },
+    {
+      "epoch": 0.4837837837837838,
+      "grad_norm": 0.4957127571105957,
+      "learning_rate": 0.0005,
+      "loss": 1.0951273441314697,
+      "step": 358
+    },
+    {
+      "epoch": 0.4851351351351351,
+      "grad_norm": 0.5073482990264893,
+      "learning_rate": 0.0005,
+      "loss": 1.0820565223693848,
+      "step": 359
+    },
+    {
+      "epoch": 0.4864864864864865,
+      "grad_norm": 0.4286222457885742,
+      "learning_rate": 0.0005,
+      "loss": 1.070575475692749,
+      "step": 360
+    },
+    {
+      "epoch": 0.48783783783783785,
+      "grad_norm": 0.5445823073387146,
+      "learning_rate": 0.0005,
+      "loss": 1.0851056575775146,
+      "step": 361
+    },
+    {
+      "epoch": 0.4891891891891892,
+      "grad_norm": 0.5872825980186462,
+      "learning_rate": 0.0005,
+      "loss": 1.0698063373565674,
+      "step": 362
+    },
+    {
+      "epoch": 0.4905405405405405,
+      "grad_norm": 0.6388886570930481,
+      "learning_rate": 0.0005,
+      "loss": 1.1297681331634521,
+      "step": 363
+    },
+    {
+      "epoch": 0.4918918918918919,
+      "grad_norm": 0.4993698000907898,
+      "learning_rate": 0.0005,
+      "loss": 1.0771889686584473,
+      "step": 364
+    },
+    {
+      "epoch": 0.49324324324324326,
+      "grad_norm": 0.6535164713859558,
+      "learning_rate": 0.0005,
+      "loss": 1.1116045713424683,
+      "step": 365
+    },
+    {
+      "epoch": 0.4945945945945946,
+      "grad_norm": 0.49536725878715515,
+      "learning_rate": 0.0005,
+      "loss": 1.0725308656692505,
+      "step": 366
+    },
+    {
+      "epoch": 0.49594594594594593,
+      "grad_norm": 0.5841764211654663,
+      "learning_rate": 0.0005,
+      "loss": 1.0601493120193481,
+      "step": 367
+    },
+    {
+      "epoch": 0.4972972972972973,
+      "grad_norm": 0.657380998134613,
+      "learning_rate": 0.0005,
+      "loss": 1.0814579725265503,
+      "step": 368
+    },
+    {
+      "epoch": 0.49864864864864866,
+      "grad_norm": 0.5290428400039673,
+      "learning_rate": 0.0005,
+      "loss": 1.1033718585968018,
+      "step": 369
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4980141818523407,
+      "learning_rate": 0.0005,
+      "loss": 1.0430307388305664,
+      "step": 370
+    },
+    {
+      "epoch": 0.5013513513513513,
+      "grad_norm": 0.5604400634765625,
+      "learning_rate": 0.0005,
+      "loss": 1.0714476108551025,
+      "step": 371
+    },
+    {
+      "epoch": 0.5027027027027027,
+      "grad_norm": 0.5509002804756165,
+      "learning_rate": 0.0005,
+      "loss": 1.0782151222229004,
+      "step": 372
+    },
+    {
+      "epoch": 0.504054054054054,
+      "grad_norm": 0.503757119178772,
+      "learning_rate": 0.0005,
+      "loss": 1.0868076086044312,
+      "step": 373
+    },
+    {
+      "epoch": 0.5054054054054054,
+      "grad_norm": 0.5594017505645752,
+      "learning_rate": 0.0005,
+      "loss": 1.0860458612442017,
+      "step": 374
+    },
+    {
+      "epoch": 0.5067567567567568,
+      "grad_norm": 0.5547530651092529,
+      "learning_rate": 0.0005,
+      "loss": 1.0654972791671753,
+      "step": 375
+    },
+    {
+      "epoch": 0.5081081081081081,
+      "grad_norm": 0.5168285369873047,
+      "learning_rate": 0.0005,
+      "loss": 1.0886621475219727,
+      "step": 376
+    },
+    {
+      "epoch": 0.5094594594594595,
+      "grad_norm": 0.5278452634811401,
+      "learning_rate": 0.0005,
+      "loss": 1.1363879442214966,
+      "step": 377
+    },
+    {
+      "epoch": 0.5108108108108108,
+      "grad_norm": 0.5178099870681763,
+      "learning_rate": 0.0005,
+      "loss": 1.069596767425537,
+      "step": 378
+    },
+    {
+      "epoch": 0.5121621621621621,
+      "grad_norm": 0.45746925473213196,
+      "learning_rate": 0.0005,
+      "loss": 1.0277632474899292,
+      "step": 379
+    },
+    {
+      "epoch": 0.5135135135135135,
+      "grad_norm": 0.43722110986709595,
+      "learning_rate": 0.0005,
+      "loss": 1.048799991607666,
+      "step": 380
+    },
+    {
+      "epoch": 0.5148648648648648,
+      "grad_norm": 0.563886284828186,
+      "learning_rate": 0.0005,
+      "loss": 1.0681401491165161,
+      "step": 381
+    },
+    {
+      "epoch": 0.5162162162162162,
+      "grad_norm": 0.6592566967010498,
+      "learning_rate": 0.0005,
+      "loss": 1.0685908794403076,
+      "step": 382
+    },
+    {
+      "epoch": 0.5175675675675676,
+      "grad_norm": 0.5334255695343018,
+      "learning_rate": 0.0005,
+      "loss": 1.1096529960632324,
+      "step": 383
+    },
+    {
+      "epoch": 0.518918918918919,
+      "grad_norm": 0.48358264565467834,
+      "learning_rate": 0.0005,
+      "loss": 1.098144769668579,
+      "step": 384
+    },
+    {
+      "epoch": 0.5202702702702703,
+      "grad_norm": 0.4784682095050812,
+      "learning_rate": 0.0005,
+      "loss": 1.085552453994751,
+      "step": 385
+    },
+    {
+      "epoch": 0.5216216216216216,
+      "grad_norm": 0.47708263993263245,
+      "learning_rate": 0.0005,
+      "loss": 1.0315327644348145,
+      "step": 386
+    },
+    {
+      "epoch": 0.522972972972973,
+      "grad_norm": 0.5705336332321167,
+      "learning_rate": 0.0005,
+      "loss": 1.1196415424346924,
+      "step": 387
+    },
+    {
+      "epoch": 0.5243243243243243,
+      "grad_norm": 0.4651273190975189,
+      "learning_rate": 0.0005,
+      "loss": 1.0968196392059326,
+      "step": 388
+    },
+    {
+      "epoch": 0.5256756756756756,
+      "grad_norm": 0.41358479857444763,
+      "learning_rate": 0.0005,
+      "loss": 1.0315778255462646,
+      "step": 389
+    },
+    {
+      "epoch": 0.527027027027027,
+      "grad_norm": 0.5183364748954773,
+      "learning_rate": 0.0005,
+      "loss": 1.070630431175232,
+      "step": 390
+    },
+    {
+      "epoch": 0.5283783783783784,
+      "grad_norm": 0.5014268755912781,
+      "learning_rate": 0.0005,
+      "loss": 1.0802465677261353,
+      "step": 391
+    },
+    {
+      "epoch": 0.5297297297297298,
+      "grad_norm": 0.4863749146461487,
+      "learning_rate": 0.0005,
+      "loss": 1.0789668560028076,
+      "step": 392
+    },
+    {
+      "epoch": 0.5310810810810811,
+      "grad_norm": 0.4650745391845703,
+      "learning_rate": 0.0005,
+      "loss": 1.0448637008666992,
+      "step": 393
+    },
+    {
+      "epoch": 0.5324324324324324,
+      "grad_norm": 0.4772622287273407,
+      "learning_rate": 0.0005,
+      "loss": 1.0519218444824219,
+      "step": 394
+    },
+    {
+      "epoch": 0.5337837837837838,
+      "grad_norm": 0.4680585563182831,
+      "learning_rate": 0.0005,
+      "loss": 1.068974494934082,
+      "step": 395
+    },
+    {
+      "epoch": 0.5351351351351351,
+      "grad_norm": 0.5340046882629395,
+      "learning_rate": 0.0005,
+      "loss": 1.0795836448669434,
+      "step": 396
+    },
+    {
+      "epoch": 0.5364864864864864,
+      "grad_norm": 0.5459898710250854,
+      "learning_rate": 0.0005,
+      "loss": 1.1046656370162964,
+      "step": 397
+    },
+    {
+      "epoch": 0.5378378378378378,
+      "grad_norm": 0.43207472562789917,
+      "learning_rate": 0.0005,
+      "loss": 1.028637409210205,
+      "step": 398
+    },
+    {
+      "epoch": 0.5391891891891892,
+      "grad_norm": 0.5658368468284607,
+      "learning_rate": 0.0005,
+      "loss": 1.0938193798065186,
+      "step": 399
+    },
+    {
+      "epoch": 0.5405405405405406,
+      "grad_norm": 0.5108270645141602,
+      "learning_rate": 0.0005,
+      "loss": 1.0814203023910522,
+      "step": 400
+    },
+    {
+      "epoch": 0.5418918918918919,
+      "grad_norm": 0.5494470000267029,
+      "learning_rate": 0.0005,
+      "loss": 1.0514384508132935,
+      "step": 401
+    },
+    {
+      "epoch": 0.5432432432432432,
+      "grad_norm": 0.5327450633049011,
+      "learning_rate": 0.0005,
+      "loss": 1.084667682647705,
+      "step": 402
+    },
+    {
+      "epoch": 0.5445945945945946,
+      "grad_norm": 0.5266766548156738,
+      "learning_rate": 0.0005,
+      "loss": 1.0734909772872925,
+      "step": 403
+    },
+    {
+      "epoch": 0.5459459459459459,
+      "grad_norm": 0.4444648325443268,
+      "learning_rate": 0.0005,
+      "loss": 1.0483028888702393,
+      "step": 404
+    },
+    {
+      "epoch": 0.5472972972972973,
+      "grad_norm": 0.4890826940536499,
+      "learning_rate": 0.0005,
+      "loss": 1.048398494720459,
+      "step": 405
+    },
+    {
+      "epoch": 0.5486486486486486,
+      "grad_norm": 0.4687519073486328,
+      "learning_rate": 0.0005,
+      "loss": 1.0529582500457764,
+      "step": 406
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.47229763865470886,
+      "learning_rate": 0.0005,
+      "loss": 1.0743314027786255,
+      "step": 407
+    },
+    {
+      "epoch": 0.5513513513513514,
+      "grad_norm": 0.4878067374229431,
+      "learning_rate": 0.0005,
+      "loss": 1.0679268836975098,
+      "step": 408
+    },
+    {
+      "epoch": 0.5527027027027027,
+      "grad_norm": 0.5673519372940063,
+      "learning_rate": 0.0005,
+      "loss": 1.0871992111206055,
+      "step": 409
+    },
+    {
+      "epoch": 0.5540540540540541,
+      "grad_norm": 0.47779786586761475,
+      "learning_rate": 0.0005,
+      "loss": 1.0572857856750488,
+      "step": 410
+    },
+    {
+      "epoch": 0.5554054054054054,
+      "grad_norm": 0.5081274509429932,
+      "learning_rate": 0.0005,
+      "loss": 1.0563907623291016,
+      "step": 411
+    },
+    {
+      "epoch": 0.5567567567567567,
+      "grad_norm": 0.47870486974716187,
+      "learning_rate": 0.0005,
+      "loss": 1.0480505228042603,
+      "step": 412
+    },
+    {
+      "epoch": 0.5581081081081081,
+      "grad_norm": 0.5674945116043091,
+      "learning_rate": 0.0005,
+      "loss": 1.0267696380615234,
+      "step": 413
+    },
+    {
+      "epoch": 0.5594594594594594,
+      "grad_norm": 0.45697033405303955,
+      "learning_rate": 0.0005,
+      "loss": 1.0346108675003052,
+      "step": 414
+    },
+    {
+      "epoch": 0.5608108108108109,
+      "grad_norm": 0.5555755496025085,
+      "learning_rate": 0.0005,
+      "loss": 1.0735392570495605,
+      "step": 415
+    },
+    {
+      "epoch": 0.5621621621621622,
+      "grad_norm": 0.7655848264694214,
+      "learning_rate": 0.0005,
+      "loss": 1.102205753326416,
+      "step": 416
+    },
+    {
+      "epoch": 0.5635135135135135,
+      "grad_norm": 0.49174973368644714,
+      "learning_rate": 0.0005,
+      "loss": 1.0470848083496094,
+      "step": 417
+    },
+    {
+      "epoch": 0.5648648648648649,
+      "grad_norm": 0.6119154095649719,
+      "learning_rate": 0.0005,
+      "loss": 1.0822882652282715,
+      "step": 418
+    },
+    {
+      "epoch": 0.5662162162162162,
+      "grad_norm": 0.5382258892059326,
+      "learning_rate": 0.0005,
+      "loss": 1.1167161464691162,
+      "step": 419
+    },
+    {
+      "epoch": 0.5675675675675675,
+      "grad_norm": 0.5162979960441589,
+      "learning_rate": 0.0005,
+      "loss": 1.0591048002243042,
+      "step": 420
+    },
+    {
+      "epoch": 0.5689189189189189,
+      "grad_norm": 0.5960574746131897,
+      "learning_rate": 0.0005,
+      "loss": 1.0965948104858398,
+      "step": 421
+    },
+    {
+      "epoch": 0.5702702702702702,
+      "grad_norm": 0.5273575186729431,
+      "learning_rate": 0.0005,
+      "loss": 1.0581772327423096,
+      "step": 422
+    },
+    {
+      "epoch": 0.5716216216216217,
+      "grad_norm": 0.43563807010650635,
+      "learning_rate": 0.0005,
+      "loss": 1.0142557621002197,
+      "step": 423
+    },
+    {
+      "epoch": 0.572972972972973,
+      "grad_norm": 0.6871557831764221,
+      "learning_rate": 0.0005,
+      "loss": 1.0636563301086426,
+      "step": 424
+    },
+    {
+      "epoch": 0.5743243243243243,
+      "grad_norm": 0.47624340653419495,
+      "learning_rate": 0.0005,
+      "loss": 1.0522403717041016,
+      "step": 425
+    },
+    {
+      "epoch": 0.5756756756756757,
+      "grad_norm": 0.48264244198799133,
+      "learning_rate": 0.0005,
+      "loss": 1.0912257432937622,
+      "step": 426
+    },
+    {
+      "epoch": 0.577027027027027,
+      "grad_norm": 0.6931416392326355,
+      "learning_rate": 0.0005,
+      "loss": 1.072869896888733,
+      "step": 427
+    },
+    {
+      "epoch": 0.5783783783783784,
+      "grad_norm": 0.5382377505302429,
+      "learning_rate": 0.0005,
+      "loss": 1.057403802871704,
+      "step": 428
+    },
+    {
+      "epoch": 0.5797297297297297,
+      "grad_norm": 0.5226300954818726,
+      "learning_rate": 0.0005,
+      "loss": 1.0372663736343384,
+      "step": 429
+    },
+    {
+      "epoch": 0.581081081081081,
+      "grad_norm": 0.6336800456047058,
+      "learning_rate": 0.0005,
+      "loss": 1.0802345275878906,
+      "step": 430
+    },
+    {
+      "epoch": 0.5824324324324325,
+      "grad_norm": 0.5072260499000549,
+      "learning_rate": 0.0005,
+      "loss": 1.0680969953536987,
+      "step": 431
+    },
+    {
+      "epoch": 0.5837837837837838,
+      "grad_norm": 0.5386255383491516,
+      "learning_rate": 0.0005,
+      "loss": 1.0726966857910156,
+      "step": 432
+    },
+    {
+      "epoch": 0.5851351351351352,
+      "grad_norm": 0.577531635761261,
+      "learning_rate": 0.0005,
+      "loss": 1.0962961912155151,
+      "step": 433
+    },
+    {
+      "epoch": 0.5864864864864865,
+      "grad_norm": 0.4521787762641907,
+      "learning_rate": 0.0005,
+      "loss": 1.0740089416503906,
+      "step": 434
+    },
+    {
+      "epoch": 0.5878378378378378,
+      "grad_norm": 0.5496105551719666,
+      "learning_rate": 0.0005,
+      "loss": 1.071483850479126,
+      "step": 435
+    },
+    {
+      "epoch": 0.5891891891891892,
+      "grad_norm": 0.4901188611984253,
+      "learning_rate": 0.0005,
+      "loss": 1.0724855661392212,
+      "step": 436
+    },
+    {
+      "epoch": 0.5905405405405405,
+      "grad_norm": 0.4615694284439087,
+      "learning_rate": 0.0005,
+      "loss": 1.04933762550354,
+      "step": 437
+    },
+    {
+      "epoch": 0.5918918918918918,
+      "grad_norm": 0.5415195822715759,
+      "learning_rate": 0.0005,
+      "loss": 1.0820186138153076,
+      "step": 438
+    },
+    {
+      "epoch": 0.5932432432432433,
+      "grad_norm": 0.5417017340660095,
+      "learning_rate": 0.0005,
+      "loss": 1.0480682849884033,
+      "step": 439
+    },
+    {
+      "epoch": 0.5945945945945946,
+      "grad_norm": 0.6528788208961487,
+      "learning_rate": 0.0005,
+      "loss": 1.110504388809204,
+      "step": 440
+    },
+    {
+      "epoch": 0.595945945945946,
+      "grad_norm": 0.48452794551849365,
+      "learning_rate": 0.0005,
+      "loss": 1.0602511167526245,
+      "step": 441
+    },
+    {
+      "epoch": 0.5972972972972973,
+      "grad_norm": 0.5134998559951782,
+      "learning_rate": 0.0005,
+      "loss": 1.0417735576629639,
+      "step": 442
+    },
+    {
+      "epoch": 0.5986486486486486,
+      "grad_norm": 0.5207071900367737,
+      "learning_rate": 0.0005,
+      "loss": 1.0412781238555908,
+      "step": 443
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5858142971992493,
+      "learning_rate": 0.0005,
+      "loss": 1.0803744792938232,
+      "step": 444
+    },
+    {
+      "epoch": 0.6013513513513513,
+      "grad_norm": 0.47337478399276733,
+      "learning_rate": 0.0005,
+      "loss": 1.0673136711120605,
+      "step": 445
+    },
+    {
+      "epoch": 0.6027027027027027,
+      "grad_norm": 0.5226991176605225,
+      "learning_rate": 0.0005,
+      "loss": 1.043373942375183,
+      "step": 446
+    },
+    {
+      "epoch": 0.6040540540540541,
+      "grad_norm": 0.47647494077682495,
+      "learning_rate": 0.0005,
+      "loss": 1.0676965713500977,
+      "step": 447
+    },
+    {
+      "epoch": 0.6054054054054054,
+      "grad_norm": 0.4820915460586548,
+      "learning_rate": 0.0005,
+      "loss": 1.036659598350525,
+      "step": 448
+    },
+    {
+      "epoch": 0.6067567567567568,
+      "grad_norm": 0.4376277029514313,
+      "learning_rate": 0.0005,
+      "loss": 1.0378129482269287,
+      "step": 449
+    },
+    {
+      "epoch": 0.6081081081081081,
+      "grad_norm": 0.604735255241394,
+      "learning_rate": 0.0005,
+      "loss": 1.0333046913146973,
+      "step": 450
+    },
+    {
+      "epoch": 0.6094594594594595,
+      "grad_norm": 0.6223862171173096,
+      "learning_rate": 0.0005,
+      "loss": 1.078611135482788,
+      "step": 451
+    },
+    {
+      "epoch": 0.6108108108108108,
+      "grad_norm": 0.49422475695610046,
+      "learning_rate": 0.0005,
+      "loss": 1.0513859987258911,
+      "step": 452
+    },
+    {
+      "epoch": 0.6121621621621621,
+      "grad_norm": 0.6217668056488037,
+      "learning_rate": 0.0005,
+      "loss": 1.0462939739227295,
+      "step": 453
+    },
+    {
+      "epoch": 0.6135135135135135,
+      "grad_norm": 0.47971969842910767,
+      "learning_rate": 0.0005,
+      "loss": 1.0318353176116943,
+      "step": 454
+    },
+    {
+      "epoch": 0.6148648648648649,
+      "grad_norm": 0.47853562235832214,
+      "learning_rate": 0.0005,
+      "loss": 1.0274955034255981,
+      "step": 455
+    },
+    {
+      "epoch": 0.6162162162162163,
+      "grad_norm": 0.5500760674476624,
+      "learning_rate": 0.0005,
+      "loss": 1.055375099182129,
+      "step": 456
+    },
+    {
+      "epoch": 0.6175675675675676,
+      "grad_norm": 0.5682643055915833,
+      "learning_rate": 0.0005,
+      "loss": 1.0601093769073486,
+      "step": 457
+    },
+    {
+      "epoch": 0.6189189189189189,
+      "grad_norm": 0.4839378595352173,
+      "learning_rate": 0.0005,
+      "loss": 1.0696974992752075,
+      "step": 458
+    },
+    {
+      "epoch": 0.6202702702702703,
+      "grad_norm": 0.5248072743415833,
+      "learning_rate": 0.0005,
+      "loss": 1.0693981647491455,
+      "step": 459
+    },
+    {
+      "epoch": 0.6216216216216216,
+      "grad_norm": 0.41835638880729675,
+      "learning_rate": 0.0005,
+      "loss": 1.033949613571167,
+      "step": 460
+    },
+    {
+      "epoch": 0.6229729729729729,
+      "grad_norm": 0.5551074743270874,
+      "learning_rate": 0.0005,
+      "loss": 1.0392053127288818,
+      "step": 461
+    },
+    {
+      "epoch": 0.6243243243243243,
+      "grad_norm": 0.5182341933250427,
+      "learning_rate": 0.0005,
+      "loss": 1.0577712059020996,
+      "step": 462
+    },
+    {
+      "epoch": 0.6256756756756757,
+      "grad_norm": 0.5023088455200195,
+      "learning_rate": 0.0005,
+      "loss": 1.017266035079956,
+      "step": 463
+    },
+    {
+      "epoch": 0.6270270270270271,
+      "grad_norm": 0.4730609953403473,
+      "learning_rate": 0.0005,
+      "loss": 1.0322251319885254,
+      "step": 464
+    },
+    {
+      "epoch": 0.6283783783783784,
+      "grad_norm": 0.42791587114334106,
+      "learning_rate": 0.0005,
+      "loss": 1.0366151332855225,
+      "step": 465
+    },
+    {
+      "epoch": 0.6297297297297297,
+      "grad_norm": 0.4762161076068878,
+      "learning_rate": 0.0005,
+      "loss": 1.0304794311523438,
+      "step": 466
+    },
+    {
+      "epoch": 0.6310810810810811,
+      "grad_norm": 0.5109895467758179,
+      "learning_rate": 0.0005,
+      "loss": 1.0078113079071045,
+      "step": 467
+    },
+    {
+      "epoch": 0.6324324324324324,
+      "grad_norm": 0.508070170879364,
+      "learning_rate": 0.0005,
+      "loss": 1.0428462028503418,
+      "step": 468
+    },
+    {
+      "epoch": 0.6337837837837837,
+      "grad_norm": 0.47582876682281494,
+      "learning_rate": 0.0005,
+      "loss": 1.0290013551712036,
+      "step": 469
+    },
+    {
+      "epoch": 0.6351351351351351,
+      "grad_norm": 0.42087921500205994,
+      "learning_rate": 0.0005,
+      "loss": 1.0233328342437744,
+      "step": 470
+    },
+    {
+      "epoch": 0.6364864864864865,
+      "grad_norm": 0.5179510116577148,
+      "learning_rate": 0.0005,
+      "loss": 1.0148857831954956,
+      "step": 471
+    },
+    {
+      "epoch": 0.6378378378378379,
+      "grad_norm": 0.48539331555366516,
+      "learning_rate": 0.0005,
+      "loss": 1.06147038936615,
+      "step": 472
+    },
+    {
+      "epoch": 0.6391891891891892,
+      "grad_norm": 0.4295227527618408,
+      "learning_rate": 0.0005,
+      "loss": 1.0240867137908936,
+      "step": 473
+    },
+    {
+      "epoch": 0.6405405405405405,
+      "grad_norm": 0.47125038504600525,
+      "learning_rate": 0.0005,
+      "loss": 1.0361430644989014,
+      "step": 474
+    },
+    {
+      "epoch": 0.6418918918918919,
+      "grad_norm": 0.4819973409175873,
+      "learning_rate": 0.0005,
+      "loss": 1.0537837743759155,
+      "step": 475
+    },
+    {
+      "epoch": 0.6432432432432432,
+      "grad_norm": 0.5186148285865784,
+      "learning_rate": 0.0005,
+      "loss": 1.0720645189285278,
+      "step": 476
+    },
+    {
+      "epoch": 0.6445945945945946,
+      "grad_norm": 0.5041191577911377,
+      "learning_rate": 0.0005,
+      "loss": 1.0358654260635376,
+      "step": 477
+    },
+    {
+      "epoch": 0.6459459459459459,
+      "grad_norm": 0.6293438673019409,
+      "learning_rate": 0.0005,
+      "loss": 1.056816816329956,
+      "step": 478
+    },
+    {
+      "epoch": 0.6472972972972973,
+      "grad_norm": 0.5979795455932617,
+      "learning_rate": 0.0005,
+      "loss": 1.0717554092407227,
+      "step": 479
+    },
+    {
+      "epoch": 0.6486486486486487,
+      "grad_norm": 0.4743674397468567,
+      "learning_rate": 0.0005,
+      "loss": 1.0250202417373657,
+      "step": 480
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.49539586901664734,
+      "learning_rate": 0.0005,
+      "loss": 1.031696081161499,
+      "step": 481
+    },
+    {
+      "epoch": 0.6513513513513514,
+      "grad_norm": 0.47858211398124695,
+      "learning_rate": 0.0005,
+      "loss": 1.0164557695388794,
+      "step": 482
+    },
+    {
+      "epoch": 0.6527027027027027,
+      "grad_norm": 0.4592629671096802,
+      "learning_rate": 0.0005,
+      "loss": 1.0283336639404297,
+      "step": 483
+    },
+    {
+      "epoch": 0.654054054054054,
+      "grad_norm": 0.47926226258277893,
+      "learning_rate": 0.0005,
+      "loss": 1.0222868919372559,
+      "step": 484
+    },
+    {
+      "epoch": 0.6554054054054054,
+      "grad_norm": 0.4727352261543274,
+      "learning_rate": 0.0005,
+      "loss": 1.0491029024124146,
+      "step": 485
+    },
+    {
+      "epoch": 0.6567567567567567,
+      "grad_norm": 0.4233059287071228,
+      "learning_rate": 0.0005,
+      "loss": 1.0376783609390259,
+      "step": 486
+    },
+    {
+      "epoch": 0.6581081081081082,
+      "grad_norm": 0.5205044746398926,
+      "learning_rate": 0.0005,
+      "loss": 1.0493543148040771,
+      "step": 487
+    },
+    {
+      "epoch": 0.6594594594594595,
+      "grad_norm": 0.4925619959831238,
+      "learning_rate": 0.0005,
+      "loss": 1.0298229455947876,
+      "step": 488
+    },
+    {
+      "epoch": 0.6608108108108108,
+      "grad_norm": 0.6024385690689087,
+      "learning_rate": 0.0005,
+      "loss": 1.0500094890594482,
+      "step": 489
+    },
+    {
+      "epoch": 0.6621621621621622,
+      "grad_norm": 0.39845606684684753,
+      "learning_rate": 0.0005,
+      "loss": 1.0017619132995605,
+      "step": 490
+    },
+    {
+      "epoch": 0.6635135135135135,
+      "grad_norm": 0.51519775390625,
+      "learning_rate": 0.0005,
+      "loss": 1.037652611732483,
+      "step": 491
+    },
+    {
+      "epoch": 0.6648648648648648,
+      "grad_norm": 0.5427865982055664,
+      "learning_rate": 0.0005,
+      "loss": 1.0273033380508423,
+      "step": 492
+    },
+    {
+      "epoch": 0.6662162162162162,
+      "grad_norm": 0.41599738597869873,
+      "learning_rate": 0.0005,
+      "loss": 0.9836013317108154,
+      "step": 493
+    },
+    {
+      "epoch": 0.6675675675675675,
+      "grad_norm": 0.573667049407959,
+      "learning_rate": 0.0005,
+      "loss": 1.0532680749893188,
+      "step": 494
+    },
+    {
+      "epoch": 0.668918918918919,
+      "grad_norm": 0.4746394455432892,
+      "learning_rate": 0.0005,
+      "loss": 1.0513215065002441,
+      "step": 495
+    },
+    {
+      "epoch": 0.6702702702702703,
+      "grad_norm": 0.5827715992927551,
+      "learning_rate": 0.0005,
+      "loss": 1.0666645765304565,
+      "step": 496
+    },
+    {
+      "epoch": 0.6716216216216216,
+      "grad_norm": 0.5131393671035767,
+      "learning_rate": 0.0005,
+      "loss": 1.028235912322998,
+      "step": 497
+    },
+    {
+      "epoch": 0.672972972972973,
+      "grad_norm": 0.5110966563224792,
+      "learning_rate": 0.0005,
+      "loss": 1.023898720741272,
+      "step": 498
+    },
+    {
+      "epoch": 0.6743243243243243,
+      "grad_norm": 0.6198959350585938,
+      "learning_rate": 0.0005,
+      "loss": 1.052811861038208,
+      "step": 499
+    },
+    {
+      "epoch": 0.6756756756756757,
+      "grad_norm": 0.5189945697784424,
+      "learning_rate": 0.0005,
+      "loss": 1.058215856552124,
+      "step": 500
+    },
+    {
+      "epoch": 0.677027027027027,
+      "grad_norm": 0.5540105700492859,
+      "learning_rate": 0.0005,
+      "loss": 1.0416383743286133,
+      "step": 501
+    },
+    {
+      "epoch": 0.6783783783783783,
+      "grad_norm": 0.5179439783096313,
+      "learning_rate": 0.0005,
+      "loss": 1.0480440855026245,
+      "step": 502
+    },
+    {
+      "epoch": 0.6797297297297298,
+      "grad_norm": 0.45372235774993896,
+      "learning_rate": 0.0005,
+      "loss": 1.0257372856140137,
+      "step": 503
+    },
+    {
+      "epoch": 0.6810810810810811,
+      "grad_norm": 0.5303036570549011,
+      "learning_rate": 0.0005,
+      "loss": 1.0832569599151611,
+      "step": 504
+    },
+    {
+      "epoch": 0.6824324324324325,
+      "grad_norm": 0.5142026543617249,
+      "learning_rate": 0.0005,
+      "loss": 1.0558518171310425,
+      "step": 505
+    },
+    {
+      "epoch": 0.6837837837837838,
+      "grad_norm": 0.593209445476532,
+      "learning_rate": 0.0005,
+      "loss": 1.0961201190948486,
+      "step": 506
+    },
+    {
+      "epoch": 0.6851351351351351,
+      "grad_norm": 0.485584557056427,
+      "learning_rate": 0.0005,
+      "loss": 1.0598936080932617,
+      "step": 507
+    },
+    {
+      "epoch": 0.6864864864864865,
+      "grad_norm": 0.42617595195770264,
+      "learning_rate": 0.0005,
+      "loss": 1.0283641815185547,
+      "step": 508
+    },
+    {
+      "epoch": 0.6878378378378378,
+      "grad_norm": 0.5115044116973877,
+      "learning_rate": 0.0005,
+      "loss": 1.0470268726348877,
+      "step": 509
+    },
+    {
+      "epoch": 0.6891891891891891,
+      "grad_norm": 0.49908941984176636,
+      "learning_rate": 0.0005,
+      "loss": 1.0529003143310547,
+      "step": 510
+    },
+    {
+      "epoch": 0.6905405405405406,
+      "grad_norm": 0.4704062342643738,
+      "learning_rate": 0.0005,
+      "loss": 1.0366861820220947,
+      "step": 511
+    },
+    {
+      "epoch": 0.6918918918918919,
+      "grad_norm": 0.5176315307617188,
+      "learning_rate": 0.0005,
+      "loss": 1.050870656967163,
+      "step": 512
+    },
+    {
+      "epoch": 0.6932432432432433,
+      "grad_norm": 0.47236931324005127,
+      "learning_rate": 0.0005,
+      "loss": 1.0175198316574097,
+      "step": 513
+    },
+    {
+      "epoch": 0.6945945945945946,
+      "grad_norm": 0.5407692790031433,
+      "learning_rate": 0.0005,
+      "loss": 1.0138609409332275,
+      "step": 514
+    },
+    {
+      "epoch": 0.6959459459459459,
+      "grad_norm": 0.4923318326473236,
+      "learning_rate": 0.0005,
+      "loss": 1.0328869819641113,
+      "step": 515
+    },
+    {
+      "epoch": 0.6972972972972973,
+      "grad_norm": 0.4349938631057739,
+      "learning_rate": 0.0005,
+      "loss": 1.0451037883758545,
+      "step": 516
+    },
+    {
+      "epoch": 0.6986486486486486,
+      "grad_norm": 0.5476527810096741,
+      "learning_rate": 0.0005,
+      "loss": 1.0718934535980225,
+      "step": 517
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.48214271664619446,
+      "learning_rate": 0.0005,
+      "loss": 1.0486044883728027,
+      "step": 518
+    },
+    {
+      "epoch": 0.7013513513513514,
+      "grad_norm": 0.4273481070995331,
+      "learning_rate": 0.0005,
+      "loss": 1.0336437225341797,
+      "step": 519
+    },
+    {
+      "epoch": 0.7027027027027027,
+      "grad_norm": 0.4973129630088806,
+      "learning_rate": 0.0005,
+      "loss": 1.055091381072998,
+      "step": 520
+    },
+    {
+      "epoch": 0.7040540540540541,
+      "grad_norm": 0.49030575156211853,
+      "learning_rate": 0.0005,
+      "loss": 1.045243263244629,
+      "step": 521
+    },
+    {
+      "epoch": 0.7054054054054054,
+      "grad_norm": 0.5066006183624268,
+      "learning_rate": 0.0005,
+      "loss": 1.0333068370819092,
+      "step": 522
+    },
+    {
+      "epoch": 0.7067567567567568,
+      "grad_norm": 0.45373833179473877,
+      "learning_rate": 0.0005,
+      "loss": 0.9968117475509644,
+      "step": 523
+    },
+    {
+      "epoch": 0.7081081081081081,
+      "grad_norm": 0.5595064759254456,
+      "learning_rate": 0.0005,
+      "loss": 1.093470573425293,
+      "step": 524
+    },
+    {
+      "epoch": 0.7094594594594594,
+      "grad_norm": 0.4288385510444641,
+      "learning_rate": 0.0005,
+      "loss": 1.0205581188201904,
+      "step": 525
+    },
+    {
+      "epoch": 0.7108108108108108,
+      "grad_norm": 0.4625224471092224,
+      "learning_rate": 0.0005,
+      "loss": 1.022278070449829,
+      "step": 526
+    },
+    {
+      "epoch": 0.7121621621621622,
+      "grad_norm": 0.45569881796836853,
+      "learning_rate": 0.0005,
+      "loss": 1.0382270812988281,
+      "step": 527
+    },
+    {
+      "epoch": 0.7135135135135136,
+      "grad_norm": 0.4225581884384155,
+      "learning_rate": 0.0005,
+      "loss": 1.036947250366211,
+      "step": 528
+    },
+    {
+      "epoch": 0.7148648648648649,
+      "grad_norm": 0.45782795548439026,
+      "learning_rate": 0.0005,
+      "loss": 1.0075304508209229,
+      "step": 529
+    },
+    {
+      "epoch": 0.7162162162162162,
+      "grad_norm": 0.4990319311618805,
+      "learning_rate": 0.0005,
+      "loss": 1.0402591228485107,
+      "step": 530
+    },
+    {
+      "epoch": 0.7175675675675676,
+      "grad_norm": 0.48327210545539856,
+      "learning_rate": 0.0005,
+      "loss": 1.0420546531677246,
+      "step": 531
+    },
+    {
+      "epoch": 0.7189189189189189,
+      "grad_norm": 0.5273853540420532,
+      "learning_rate": 0.0005,
+      "loss": 1.0421206951141357,
+      "step": 532
+    },
+    {
+      "epoch": 0.7202702702702702,
+      "grad_norm": 0.5067024827003479,
+      "learning_rate": 0.0005,
+      "loss": 1.0412708520889282,
+      "step": 533
+    },
+    {
+      "epoch": 0.7216216216216216,
+      "grad_norm": 0.5357084274291992,
+      "learning_rate": 0.0005,
+      "loss": 1.0633103847503662,
+      "step": 534
+    },
+    {
+      "epoch": 0.722972972972973,
+      "grad_norm": 0.4665685296058655,
+      "learning_rate": 0.0005,
+      "loss": 1.0338220596313477,
+      "step": 535
+    },
+    {
+      "epoch": 0.7243243243243244,
+      "grad_norm": 0.4655638635158539,
+      "learning_rate": 0.0005,
+      "loss": 1.0408687591552734,
+      "step": 536
+    },
+    {
+      "epoch": 0.7256756756756757,
+      "grad_norm": 0.4898051619529724,
+      "learning_rate": 0.0005,
+      "loss": 1.0313316583633423,
+      "step": 537
+    },
+    {
+      "epoch": 0.727027027027027,
+      "grad_norm": 0.4842261075973511,
+      "learning_rate": 0.0005,
+      "loss": 1.03394615650177,
+      "step": 538
+    },
+    {
+      "epoch": 0.7283783783783784,
+      "grad_norm": 0.44329172372817993,
+      "learning_rate": 0.0005,
+      "loss": 1.0366597175598145,
+      "step": 539
+    },
+    {
+      "epoch": 0.7297297297297297,
+      "grad_norm": 0.43810006976127625,
+      "learning_rate": 0.0005,
+      "loss": 1.0213016271591187,
+      "step": 540
+    },
+    {
+      "epoch": 0.731081081081081,
+      "grad_norm": 0.49642184376716614,
+      "learning_rate": 0.0005,
+      "loss": 1.0453033447265625,
+      "step": 541
+    },
+    {
+      "epoch": 0.7324324324324324,
+      "grad_norm": 0.41174235939979553,
+      "learning_rate": 0.0005,
+      "loss": 1.0031802654266357,
+      "step": 542
+    },
+    {
+      "epoch": 0.7337837837837838,
+      "grad_norm": 0.4784287214279175,
+      "learning_rate": 0.0005,
+      "loss": 1.0258064270019531,
+      "step": 543
+    },
+    {
+      "epoch": 0.7351351351351352,
+      "grad_norm": 0.4930378794670105,
+      "learning_rate": 0.0005,
+      "loss": 1.027051329612732,
+      "step": 544
+    },
+    {
+      "epoch": 0.7364864864864865,
+      "grad_norm": 0.5551691651344299,
+      "learning_rate": 0.0005,
+      "loss": 1.0454012155532837,
+      "step": 545
+    },
+    {
+      "epoch": 0.7378378378378379,
+      "grad_norm": 0.49620580673217773,
+      "learning_rate": 0.0005,
+      "loss": 1.065859079360962,
+      "step": 546
+    },
+    {
+      "epoch": 0.7391891891891892,
+      "grad_norm": 0.449829638004303,
+      "learning_rate": 0.0005,
+      "loss": 1.03792405128479,
+      "step": 547
+    },
+    {
+      "epoch": 0.7405405405405405,
+      "grad_norm": 0.5030459761619568,
+      "learning_rate": 0.0005,
+      "loss": 1.065321922302246,
+      "step": 548
+    },
+    {
+      "epoch": 0.7418918918918919,
+      "grad_norm": 0.5905163288116455,
+      "learning_rate": 0.0005,
+      "loss": 1.0636804103851318,
+      "step": 549
+    },
+    {
+      "epoch": 0.7432432432432432,
+      "grad_norm": 0.5087503790855408,
+      "learning_rate": 0.0005,
+      "loss": 1.0318293571472168,
+      "step": 550
+    },
+    {
+      "epoch": 0.7445945945945946,
+      "grad_norm": 0.42909225821495056,
+      "learning_rate": 0.0005,
+      "loss": 1.0346299409866333,
+      "step": 551
+    },
+    {
+      "epoch": 0.745945945945946,
+      "grad_norm": 0.4954073131084442,
+      "learning_rate": 0.0005,
+      "loss": 1.0068690776824951,
+      "step": 552
+    },
+    {
+      "epoch": 0.7472972972972973,
+      "grad_norm": 0.40755385160446167,
+      "learning_rate": 0.0005,
+      "loss": 1.024791955947876,
+      "step": 553
+    },
+    {
+      "epoch": 0.7486486486486487,
+      "grad_norm": 0.5402243733406067,
+      "learning_rate": 0.0005,
+      "loss": 1.0808833837509155,
+      "step": 554
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4142736494541168,
+      "learning_rate": 0.0005,
+      "loss": 1.0196151733398438,
+      "step": 555
+    },
+    {
+      "epoch": 0.7513513513513513,
+      "grad_norm": 0.48605963587760925,
+      "learning_rate": 0.0005,
+      "loss": 1.0608810186386108,
+      "step": 556
+    },
+    {
+      "epoch": 0.7527027027027027,
+      "grad_norm": 0.5101632475852966,
+      "learning_rate": 0.0005,
+      "loss": 1.0146245956420898,
+      "step": 557
+    },
+    {
+      "epoch": 0.754054054054054,
+      "grad_norm": 0.46320557594299316,
+      "learning_rate": 0.0005,
+      "loss": 1.0326318740844727,
+      "step": 558
+    },
+    {
+      "epoch": 0.7554054054054054,
+      "grad_norm": 0.45969563722610474,
+      "learning_rate": 0.0005,
+      "loss": 1.0196032524108887,
+      "step": 559
+    },
+    {
+      "epoch": 0.7567567567567568,
+      "grad_norm": 0.5032131671905518,
+      "learning_rate": 0.0005,
+      "loss": 1.041754961013794,
+      "step": 560
+    },
+    {
+      "epoch": 0.7581081081081081,
+      "grad_norm": 0.5435339212417603,
+      "learning_rate": 0.0005,
+      "loss": 1.0451676845550537,
+      "step": 561
+    },
+    {
+      "epoch": 0.7594594594594595,
+      "grad_norm": 0.4705445170402527,
+      "learning_rate": 0.0005,
+      "loss": 1.0531575679779053,
+      "step": 562
+    },
+    {
+      "epoch": 0.7608108108108108,
+      "grad_norm": 0.5654048323631287,
+      "learning_rate": 0.0005,
+      "loss": 1.0476967096328735,
+      "step": 563
+    },
+    {
+      "epoch": 0.7621621621621621,
+      "grad_norm": 0.38625577092170715,
+      "learning_rate": 0.0005,
+      "loss": 0.9966887831687927,
+      "step": 564
+    },
+    {
+      "epoch": 0.7635135135135135,
+      "grad_norm": 0.5037944316864014,
+      "learning_rate": 0.0005,
+      "loss": 1.066859245300293,
+      "step": 565
+    },
+    {
+      "epoch": 0.7648648648648648,
+      "grad_norm": 0.45881128311157227,
+      "learning_rate": 0.0005,
+      "loss": 1.0178160667419434,
+      "step": 566
+    },
+    {
+      "epoch": 0.7662162162162162,
+      "grad_norm": 0.42770805954933167,
+      "learning_rate": 0.0005,
+      "loss": 1.003728985786438,
+      "step": 567
+    },
+    {
+      "epoch": 0.7675675675675676,
+      "grad_norm": 0.5010950565338135,
+      "learning_rate": 0.0005,
+      "loss": 1.0440328121185303,
+      "step": 568
+    },
+    {
+      "epoch": 0.768918918918919,
+      "grad_norm": 0.45169878005981445,
+      "learning_rate": 0.0005,
+      "loss": 1.0093224048614502,
+      "step": 569
+    },
+    {
+      "epoch": 0.7702702702702703,
+      "grad_norm": 0.4730907082557678,
+      "learning_rate": 0.0005,
+      "loss": 1.038079023361206,
+      "step": 570
+    },
+    {
+      "epoch": 0.7716216216216216,
+      "grad_norm": 0.487634539604187,
+      "learning_rate": 0.0005,
+      "loss": 1.0257151126861572,
+      "step": 571
+    },
+    {
+      "epoch": 0.772972972972973,
+      "grad_norm": 0.4810939133167267,
+      "learning_rate": 0.0005,
+      "loss": 0.993195116519928,
+      "step": 572
+    },
+    {
+      "epoch": 0.7743243243243243,
+      "grad_norm": 0.5274896621704102,
+      "learning_rate": 0.0005,
+      "loss": 1.0808234214782715,
+      "step": 573
+    },
+    {
+      "epoch": 0.7756756756756756,
+      "grad_norm": 0.4638441801071167,
+      "learning_rate": 0.0005,
+      "loss": 1.03310227394104,
+      "step": 574
+    },
+    {
+      "epoch": 0.777027027027027,
+      "grad_norm": 0.5634359121322632,
+      "learning_rate": 0.0005,
+      "loss": 1.0224418640136719,
+      "step": 575
+    },
+    {
+      "epoch": 0.7783783783783784,
+      "grad_norm": 0.48556622862815857,
+      "learning_rate": 0.0005,
+      "loss": 1.0380364656448364,
+      "step": 576
+    },
+    {
+      "epoch": 0.7797297297297298,
+      "grad_norm": 0.4427739977836609,
+      "learning_rate": 0.0005,
+      "loss": 1.018902063369751,
+      "step": 577
+    },
+    {
+      "epoch": 0.7810810810810811,
+      "grad_norm": 0.4209824502468109,
+      "learning_rate": 0.0005,
+      "loss": 1.0146095752716064,
+      "step": 578
+    },
+    {
+      "epoch": 0.7824324324324324,
+      "grad_norm": 0.4424142837524414,
+      "learning_rate": 0.0005,
+      "loss": 1.0146994590759277,
+      "step": 579
+    },
+    {
+      "epoch": 0.7837837837837838,
+      "grad_norm": 0.468938946723938,
+      "learning_rate": 0.0005,
+      "loss": 1.0057779550552368,
+      "step": 580
+    },
+    {
+      "epoch": 0.7851351351351351,
+      "grad_norm": 0.4303213655948639,
+      "learning_rate": 0.0005,
+      "loss": 1.0192193984985352,
+      "step": 581
+    },
+    {
+      "epoch": 0.7864864864864864,
+      "grad_norm": 0.40197330713272095,
+      "learning_rate": 0.0005,
+      "loss": 1.016475796699524,
+      "step": 582
+    },
+    {
+      "epoch": 0.7878378378378378,
+      "grad_norm": 0.5073442459106445,
+      "learning_rate": 0.0005,
+      "loss": 1.0290741920471191,
+      "step": 583
+    },
+    {
+      "epoch": 0.7891891891891892,
+      "grad_norm": 0.5253981351852417,
+      "learning_rate": 0.0005,
+      "loss": 1.0564197301864624,
+      "step": 584
+    },
+    {
+      "epoch": 0.7905405405405406,
+      "grad_norm": 0.37736448645591736,
+      "learning_rate": 0.0005,
+      "loss": 0.9992639422416687,
+      "step": 585
+    },
+    {
+      "epoch": 0.7918918918918919,
+      "grad_norm": 0.5484856963157654,
+      "learning_rate": 0.0005,
+      "loss": 1.0441937446594238,
+      "step": 586
+    },
+    {
+      "epoch": 0.7932432432432432,
+      "grad_norm": 0.41673773527145386,
+      "learning_rate": 0.0005,
+      "loss": 1.0231746435165405,
+      "step": 587
+    },
+    {
+      "epoch": 0.7945945945945946,
+      "grad_norm": 0.4072820544242859,
+      "learning_rate": 0.0005,
+      "loss": 1.0057131052017212,
+      "step": 588
+    },
+    {
+      "epoch": 0.7959459459459459,
+      "grad_norm": 0.5318555235862732,
+      "learning_rate": 0.0005,
+      "loss": 1.006967306137085,
+      "step": 589
+    },
+    {
+      "epoch": 0.7972972972972973,
+      "grad_norm": 0.43006038665771484,
+      "learning_rate": 0.0005,
+      "loss": 1.00039803981781,
+      "step": 590
+    },
+    {
+      "epoch": 0.7986486486486486,
+      "grad_norm": 0.41542840003967285,
+      "learning_rate": 0.0005,
+      "loss": 1.0094499588012695,
+      "step": 591
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4588295519351959,
+      "learning_rate": 0.0005,
+      "loss": 1.034789800643921,
+      "step": 592
+    },
+    {
+      "epoch": 0.8013513513513514,
+      "grad_norm": 0.48823821544647217,
+      "learning_rate": 0.0005,
+      "loss": 1.0368657112121582,
+      "step": 593
+    },
+    {
+      "epoch": 0.8027027027027027,
+      "grad_norm": 0.4459134638309479,
+      "learning_rate": 0.0005,
+      "loss": 0.9959095120429993,
+      "step": 594
+    },
+    {
+      "epoch": 0.8040540540540541,
+      "grad_norm": 0.43293318152427673,
+      "learning_rate": 0.0005,
+      "loss": 0.9924641847610474,
+      "step": 595
+    },
+    {
+      "epoch": 0.8054054054054054,
+      "grad_norm": 0.4316229522228241,
+      "learning_rate": 0.0005,
+      "loss": 1.0188825130462646,
+      "step": 596
+    },
+    {
+      "epoch": 0.8067567567567567,
+      "grad_norm": 0.4091908037662506,
+      "learning_rate": 0.0005,
+      "loss": 0.9827393889427185,
+      "step": 597
+    },
+    {
+      "epoch": 0.8081081081081081,
+      "grad_norm": 0.4476555585861206,
+      "learning_rate": 0.0005,
+      "loss": 1.0209702253341675,
+      "step": 598
+    },
+    {
+      "epoch": 0.8094594594594594,
+      "grad_norm": 0.46138909459114075,
+      "learning_rate": 0.0005,
+      "loss": 1.045775294303894,
+      "step": 599
+    },
+    {
+      "epoch": 0.8108108108108109,
+      "grad_norm": 0.49752670526504517,
+      "learning_rate": 0.0005,
+      "loss": 1.0137711763381958,
+      "step": 600
+    },
+    {
+      "epoch": 0.8121621621621622,
+      "grad_norm": 0.42703142762184143,
+      "learning_rate": 0.0005,
+      "loss": 1.0052390098571777,
+      "step": 601
+    },
+    {
+      "epoch": 0.8135135135135135,
+      "grad_norm": 0.45716193318367004,
+      "learning_rate": 0.0005,
+      "loss": 1.0265538692474365,
+      "step": 602
+    },
+    {
+      "epoch": 0.8148648648648649,
+      "grad_norm": 0.4737812280654907,
+      "learning_rate": 0.0005,
+      "loss": 1.0255980491638184,
+      "step": 603
+    },
+    {
+      "epoch": 0.8162162162162162,
+      "grad_norm": 0.545356035232544,
+      "learning_rate": 0.0005,
+      "loss": 1.031076192855835,
+      "step": 604
+    },
+    {
+      "epoch": 0.8175675675675675,
+      "grad_norm": 0.4238259494304657,
+      "learning_rate": 0.0005,
+      "loss": 1.013535976409912,
+      "step": 605
+    },
+    {
+      "epoch": 0.8189189189189189,
+      "grad_norm": 0.6073107123374939,
+      "learning_rate": 0.0005,
+      "loss": 1.0498228073120117,
+      "step": 606
+    },
+    {
+      "epoch": 0.8202702702702702,
+      "grad_norm": 0.5308000445365906,
+      "learning_rate": 0.0005,
+      "loss": 1.0487077236175537,
+      "step": 607
+    },
+    {
+      "epoch": 0.8216216216216217,
+      "grad_norm": 0.6445031762123108,
+      "learning_rate": 0.0005,
+      "loss": 1.0511088371276855,
+      "step": 608
+    },
+    {
+      "epoch": 0.822972972972973,
+      "grad_norm": 0.47989726066589355,
+      "learning_rate": 0.0005,
+      "loss": 1.0241856575012207,
+      "step": 609
+    },
+    {
+      "epoch": 0.8243243243243243,
+      "grad_norm": 0.46888938546180725,
+      "learning_rate": 0.0005,
+      "loss": 1.0699931383132935,
+      "step": 610
+    },
+    {
+      "epoch": 0.8256756756756757,
+      "grad_norm": 0.5808868408203125,
+      "learning_rate": 0.0005,
+      "loss": 1.0739437341690063,
+      "step": 611
+    },
+    {
+      "epoch": 0.827027027027027,
+      "grad_norm": 0.4044581651687622,
+      "learning_rate": 0.0005,
+      "loss": 1.02779221534729,
+      "step": 612
+    },
+    {
+      "epoch": 0.8283783783783784,
+      "grad_norm": 0.5485680103302002,
+      "learning_rate": 0.0005,
+      "loss": 1.0110399723052979,
+      "step": 613
+    },
+    {
+      "epoch": 0.8297297297297297,
+      "grad_norm": 0.49968525767326355,
+      "learning_rate": 0.0005,
+      "loss": 0.99173903465271,
+      "step": 614
+    },
+    {
+      "epoch": 0.831081081081081,
+      "grad_norm": 0.49087750911712646,
+      "learning_rate": 0.0005,
+      "loss": 1.035444736480713,
+      "step": 615
+    },
+    {
+      "epoch": 0.8324324324324325,
+      "grad_norm": 0.5220504999160767,
+      "learning_rate": 0.0005,
+      "loss": 1.0296630859375,
+      "step": 616
+    },
+    {
+      "epoch": 0.8337837837837838,
+      "grad_norm": 0.4195800721645355,
+      "learning_rate": 0.0005,
+      "loss": 1.0090069770812988,
+      "step": 617
+    },
+    {
+      "epoch": 0.8351351351351352,
+      "grad_norm": 0.5679245591163635,
+      "learning_rate": 0.0005,
+      "loss": 1.0799074172973633,
+      "step": 618
+    },
+    {
+      "epoch": 0.8364864864864865,
+      "grad_norm": 0.5598008632659912,
+      "learning_rate": 0.0005,
+      "loss": 1.01593816280365,
+      "step": 619
+    },
+    {
+      "epoch": 0.8378378378378378,
+      "grad_norm": 0.4850425124168396,
+      "learning_rate": 0.0005,
+      "loss": 1.0145297050476074,
+      "step": 620
+    },
+    {
+      "epoch": 0.8391891891891892,
+      "grad_norm": 0.49778953194618225,
+      "learning_rate": 0.0005,
+      "loss": 1.0130635499954224,
+      "step": 621
+    },
+    {
+      "epoch": 0.8405405405405405,
+      "grad_norm": 0.5111128091812134,
+      "learning_rate": 0.0005,
+      "loss": 1.054506778717041,
+      "step": 622
+    },
+    {
+      "epoch": 0.8418918918918918,
+      "grad_norm": 0.42872992157936096,
+      "learning_rate": 0.0005,
+      "loss": 1.0361979007720947,
+      "step": 623
+    },
+    {
+      "epoch": 0.8432432432432433,
+      "grad_norm": 0.3681606948375702,
+      "learning_rate": 0.0005,
+      "loss": 0.984827995300293,
+      "step": 624
+    },
+    {
+      "epoch": 0.8445945945945946,
+      "grad_norm": 0.48832836747169495,
+      "learning_rate": 0.0005,
+      "loss": 1.0181818008422852,
+      "step": 625
+    },
+    {
+      "epoch": 0.845945945945946,
+      "grad_norm": 0.4458760619163513,
+      "learning_rate": 0.0005,
+      "loss": 1.0090442895889282,
+      "step": 626
+    },
+    {
+      "epoch": 0.8472972972972973,
+      "grad_norm": 0.4188026189804077,
+      "learning_rate": 0.0005,
+      "loss": 1.0031174421310425,
+      "step": 627
+    },
+    {
+      "epoch": 0.8486486486486486,
+      "grad_norm": 0.5052714943885803,
+      "learning_rate": 0.0005,
+      "loss": 1.006934642791748,
+      "step": 628
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4715379774570465,
+      "learning_rate": 0.0005,
+      "loss": 0.9957121014595032,
+      "step": 629
+    },
+    {
+      "epoch": 0.8513513513513513,
+      "grad_norm": 0.49712294340133667,
+      "learning_rate": 0.0005,
+      "loss": 1.0217595100402832,
+      "step": 630
+    },
+    {
+      "epoch": 0.8527027027027027,
+      "grad_norm": 0.5355307459831238,
+      "learning_rate": 0.0005,
+      "loss": 1.036346673965454,
+      "step": 631
+    },
+    {
+      "epoch": 0.8540540540540541,
+      "grad_norm": 0.4915159344673157,
+      "learning_rate": 0.0005,
+      "loss": 1.035682201385498,
+      "step": 632
+    },
+    {
+      "epoch": 0.8554054054054054,
+      "grad_norm": 0.40670403838157654,
+      "learning_rate": 0.0005,
+      "loss": 0.9890744686126709,
+      "step": 633
+    },
+    {
+      "epoch": 0.8567567567567568,
+      "grad_norm": 0.4956667423248291,
+      "learning_rate": 0.0005,
+      "loss": 0.9944498538970947,
+      "step": 634
+    },
+    {
+      "epoch": 0.8581081081081081,
+      "grad_norm": 0.5081890225410461,
+      "learning_rate": 0.0005,
+      "loss": 1.0287578105926514,
+      "step": 635
+    },
+    {
+      "epoch": 0.8594594594594595,
+      "grad_norm": 0.3762257397174835,
+      "learning_rate": 0.0005,
+      "loss": 1.0098940134048462,
+      "step": 636
+    },
+    {
+      "epoch": 0.8608108108108108,
+      "grad_norm": 0.4281370937824249,
+      "learning_rate": 0.0005,
+      "loss": 1.0526270866394043,
+      "step": 637
+    },
+    {
+      "epoch": 0.8621621621621621,
+      "grad_norm": 0.4765768349170685,
+      "learning_rate": 0.0005,
+      "loss": 0.9987397193908691,
+      "step": 638
+    },
+    {
+      "epoch": 0.8635135135135135,
+      "grad_norm": 0.4511881470680237,
+      "learning_rate": 0.0005,
+      "loss": 1.0394165515899658,
+      "step": 639
+    },
+    {
+      "epoch": 0.8648648648648649,
+      "grad_norm": 0.5037724375724792,
+      "learning_rate": 0.0005,
+      "loss": 1.0249526500701904,
+      "step": 640
+    },
+    {
+      "epoch": 0.8662162162162163,
+      "grad_norm": 0.4886779189109802,
+      "learning_rate": 0.0005,
+      "loss": 1.0303761959075928,
+      "step": 641
+    },
+    {
+      "epoch": 0.8675675675675676,
+      "grad_norm": 0.4741458594799042,
+      "learning_rate": 0.0005,
+      "loss": 1.0326656103134155,
+      "step": 642
+    },
+    {
+      "epoch": 0.8689189189189189,
+      "grad_norm": 0.4561908543109894,
+      "learning_rate": 0.0005,
+      "loss": 0.9663345813751221,
+      "step": 643
+    },
+    {
+      "epoch": 0.8702702702702703,
+      "grad_norm": 0.47762688994407654,
+      "learning_rate": 0.0005,
+      "loss": 0.9962032437324524,
+      "step": 644
+    },
+    {
+      "epoch": 0.8716216216216216,
+      "grad_norm": 0.5105867981910706,
+      "learning_rate": 0.0005,
+      "loss": 1.0066442489624023,
+      "step": 645
+    },
+    {
+      "epoch": 0.8729729729729729,
+      "grad_norm": 0.3840254843235016,
+      "learning_rate": 0.0005,
+      "loss": 0.9877827167510986,
+      "step": 646
+    },
+    {
+      "epoch": 0.8743243243243243,
+      "grad_norm": 0.5215223431587219,
+      "learning_rate": 0.0005,
+      "loss": 1.0350215435028076,
+      "step": 647
+    },
+    {
+      "epoch": 0.8756756756756757,
+      "grad_norm": 0.46102967858314514,
+      "learning_rate": 0.0005,
+      "loss": 1.0055605173110962,
+      "step": 648
+    },
+    {
+      "epoch": 0.8770270270270271,
+      "grad_norm": 0.4897022545337677,
+      "learning_rate": 0.0005,
+      "loss": 0.979157567024231,
+      "step": 649
+    },
+    {
+      "epoch": 0.8783783783783784,
+      "grad_norm": 0.4580881893634796,
+      "learning_rate": 0.0005,
+      "loss": 1.02813720703125,
+      "step": 650
+    },
+    {
+      "epoch": 0.8797297297297297,
+      "grad_norm": 0.48723655939102173,
+      "learning_rate": 0.0005,
+      "loss": 1.0245866775512695,
+      "step": 651
+    },
+    {
+      "epoch": 0.8810810810810811,
+      "grad_norm": 0.5133780241012573,
+      "learning_rate": 0.0005,
+      "loss": 1.0023231506347656,
+      "step": 652
+    },
+    {
+      "epoch": 0.8824324324324324,
+      "grad_norm": 0.48277023434638977,
+      "learning_rate": 0.0005,
+      "loss": 1.0280406475067139,
+      "step": 653
+    },
+    {
+      "epoch": 0.8837837837837837,
+      "grad_norm": 0.5185429453849792,
+      "learning_rate": 0.0005,
+      "loss": 1.0330833196640015,
+      "step": 654
+    },
+    {
+      "epoch": 0.8851351351351351,
+      "grad_norm": 0.4708307087421417,
+      "learning_rate": 0.0005,
+      "loss": 0.98390793800354,
+      "step": 655
+    },
+    {
+      "epoch": 0.8864864864864865,
+      "grad_norm": 0.5550637245178223,
+      "learning_rate": 0.0005,
+      "loss": 1.0149857997894287,
+      "step": 656
+    },
+    {
+      "epoch": 0.8878378378378379,
+      "grad_norm": 0.47591426968574524,
+      "learning_rate": 0.0005,
+      "loss": 1.0093480348587036,
+      "step": 657
+    },
+    {
+      "epoch": 0.8891891891891892,
+      "grad_norm": 0.3929627537727356,
+      "learning_rate": 0.0005,
+      "loss": 0.9797130227088928,
+      "step": 658
+    },
+    {
+      "epoch": 0.8905405405405405,
+      "grad_norm": 0.557467520236969,
+      "learning_rate": 0.0005,
+      "loss": 1.032228946685791,
+      "step": 659
+    },
+    {
+      "epoch": 0.8918918918918919,
+      "grad_norm": 0.4516514837741852,
+      "learning_rate": 0.0005,
+      "loss": 1.0333354473114014,
+      "step": 660
+    },
+    {
+      "epoch": 0.8932432432432432,
+      "grad_norm": 0.3958131968975067,
+      "learning_rate": 0.0005,
+      "loss": 0.9846503734588623,
+      "step": 661
+    },
+    {
+      "epoch": 0.8945945945945946,
+      "grad_norm": 0.4606556296348572,
+      "learning_rate": 0.0005,
+      "loss": 0.9966476559638977,
+      "step": 662
+    },
+    {
+      "epoch": 0.8959459459459459,
+      "grad_norm": 0.6221334338188171,
+      "learning_rate": 0.0005,
+      "loss": 1.0364421606063843,
+      "step": 663
+    },
+    {
+      "epoch": 0.8972972972972973,
+      "grad_norm": 0.520056426525116,
+      "learning_rate": 0.0005,
+      "loss": 1.0300077199935913,
+      "step": 664
+    },
+    {
+      "epoch": 0.8986486486486487,
+      "grad_norm": 0.3949083983898163,
+      "learning_rate": 0.0005,
+      "loss": 0.9922760725021362,
+      "step": 665
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4319571256637573,
+      "learning_rate": 0.0005,
+      "loss": 1.0052639245986938,
+      "step": 666
+    },
+    {
+      "epoch": 0.9013513513513514,
+      "grad_norm": 0.4583132565021515,
+      "learning_rate": 0.0005,
+      "loss": 1.0130114555358887,
+      "step": 667
+    },
+    {
+      "epoch": 0.9027027027027027,
+      "grad_norm": 0.5304968953132629,
+      "learning_rate": 0.0005,
+      "loss": 0.9993221163749695,
+      "step": 668
+    },
+    {
+      "epoch": 0.904054054054054,
+      "grad_norm": 0.4572198987007141,
+      "learning_rate": 0.0005,
+      "loss": 1.0175483226776123,
+      "step": 669
+    },
+    {
+      "epoch": 0.9054054054054054,
+      "grad_norm": 0.432370662689209,
+      "learning_rate": 0.0005,
+      "loss": 0.9973339438438416,
+      "step": 670
+    },
+    {
+      "epoch": 0.9067567567567567,
+      "grad_norm": 0.49859416484832764,
+      "learning_rate": 0.0005,
+      "loss": 1.0221576690673828,
+      "step": 671
+    },
+    {
+      "epoch": 0.9081081081081082,
+      "grad_norm": 0.4469922184944153,
+      "learning_rate": 0.0005,
+      "loss": 1.037582516670227,
+      "step": 672
+    },
+    {
+      "epoch": 0.9094594594594595,
+      "grad_norm": 0.45900753140449524,
+      "learning_rate": 0.0005,
+      "loss": 1.0123159885406494,
+      "step": 673
+    },
+    {
+      "epoch": 0.9108108108108108,
+      "grad_norm": 0.39479169249534607,
+      "learning_rate": 0.0005,
+      "loss": 0.9888971447944641,
+      "step": 674
+    },
+    {
+      "epoch": 0.9121621621621622,
+      "grad_norm": 0.3813369870185852,
+      "learning_rate": 0.0005,
+      "loss": 0.9957236647605896,
+      "step": 675
+    },
+    {
+      "epoch": 0.9135135135135135,
+      "grad_norm": 0.4439331591129303,
+      "learning_rate": 0.0005,
+      "loss": 0.9989795684814453,
+      "step": 676
+    },
+    {
+      "epoch": 0.9148648648648648,
+      "grad_norm": 0.4254644513130188,
+      "learning_rate": 0.0005,
+      "loss": 0.9893885850906372,
+      "step": 677
+    },
+    {
+      "epoch": 0.9162162162162162,
+      "grad_norm": 0.4894943833351135,
+      "learning_rate": 0.0005,
+      "loss": 1.0079773664474487,
+      "step": 678
+    },
+    {
+      "epoch": 0.9175675675675675,
+      "grad_norm": 0.4840932786464691,
+      "learning_rate": 0.0005,
+      "loss": 1.0247912406921387,
+      "step": 679
+    },
+    {
+      "epoch": 0.918918918918919,
+      "grad_norm": 0.35252636671066284,
+      "learning_rate": 0.0005,
+      "loss": 0.9916898608207703,
+      "step": 680
+    },
+    {
+      "epoch": 0.9202702702702703,
+      "grad_norm": 0.3770149350166321,
+      "learning_rate": 0.0005,
+      "loss": 1.005816102027893,
+      "step": 681
+    },
+    {
+      "epoch": 0.9216216216216216,
+      "grad_norm": 0.4493076503276825,
+      "learning_rate": 0.0005,
+      "loss": 1.0226973295211792,
+      "step": 682
+    },
+    {
+      "epoch": 0.922972972972973,
+      "grad_norm": 0.5252628326416016,
+      "learning_rate": 0.0005,
+      "loss": 1.0220677852630615,
+      "step": 683
+    },
+    {
+      "epoch": 0.9243243243243243,
+      "grad_norm": 0.42215991020202637,
+      "learning_rate": 0.0005,
+      "loss": 0.9978030920028687,
+      "step": 684
+    },
+    {
+      "epoch": 0.9256756756756757,
+      "grad_norm": 0.4360280930995941,
+      "learning_rate": 0.0005,
+      "loss": 1.0372809171676636,
+      "step": 685
+    },
+    {
+      "epoch": 0.927027027027027,
+      "grad_norm": 0.4627527892589569,
+      "learning_rate": 0.0005,
+      "loss": 1.0065193176269531,
+      "step": 686
+    },
+    {
+      "epoch": 0.9283783783783783,
+      "grad_norm": 0.47111305594444275,
+      "learning_rate": 0.0005,
+      "loss": 1.0256332159042358,
+      "step": 687
+    },
+    {
+      "epoch": 0.9297297297297298,
+      "grad_norm": 0.4233240783214569,
+      "learning_rate": 0.0005,
+      "loss": 1.0063891410827637,
+      "step": 688
+    },
+    {
+      "epoch": 0.9310810810810811,
+      "grad_norm": 0.49735820293426514,
+      "learning_rate": 0.0005,
+      "loss": 1.03660249710083,
+      "step": 689
+    },
+    {
+      "epoch": 0.9324324324324325,
+      "grad_norm": 0.45061346888542175,
+      "learning_rate": 0.0005,
+      "loss": 1.0228159427642822,
+      "step": 690
+    },
+    {
+      "epoch": 0.9337837837837838,
+      "grad_norm": 0.4417983293533325,
+      "learning_rate": 0.0005,
+      "loss": 1.0052520036697388,
+      "step": 691
+    },
+    {
+      "epoch": 0.9351351351351351,
+      "grad_norm": 0.4691475033760071,
+      "learning_rate": 0.0005,
+      "loss": 1.0091984272003174,
+      "step": 692
+    },
+    {
+      "epoch": 0.9364864864864865,
+      "grad_norm": 0.4500192701816559,
+      "learning_rate": 0.0005,
+      "loss": 1.0428805351257324,
+      "step": 693
+    },
+    {
+      "epoch": 0.9378378378378378,
+      "grad_norm": 0.4931966960430145,
+      "learning_rate": 0.0005,
+      "loss": 1.0415328741073608,
+      "step": 694
+    },
+    {
+      "epoch": 0.9391891891891891,
+      "grad_norm": 0.4465312063694,
+      "learning_rate": 0.0005,
+      "loss": 0.9974257946014404,
+      "step": 695
+    },
+    {
+      "epoch": 0.9405405405405406,
+      "grad_norm": 0.4245172142982483,
+      "learning_rate": 0.0005,
+      "loss": 0.9970884323120117,
+      "step": 696
+    },
+    {
+      "epoch": 0.9418918918918919,
+      "grad_norm": 0.3770783841609955,
+      "learning_rate": 0.0005,
+      "loss": 0.9911584258079529,
+      "step": 697
+    },
+    {
+      "epoch": 0.9432432432432433,
+      "grad_norm": 0.4888884723186493,
+      "learning_rate": 0.0005,
+      "loss": 1.0356944799423218,
+      "step": 698
+    },
+    {
+      "epoch": 0.9445945945945946,
+      "grad_norm": 0.5687872171401978,
+      "learning_rate": 0.0005,
+      "loss": 1.0551847219467163,
+      "step": 699
+    },
+    {
+      "epoch": 0.9459459459459459,
+      "grad_norm": 0.4348186254501343,
+      "learning_rate": 0.0005,
+      "loss": 1.0065891742706299,
+      "step": 700
+    },
+    {
+      "epoch": 0.9472972972972973,
+      "grad_norm": 0.4512298107147217,
+      "learning_rate": 0.0005,
+      "loss": 0.994554340839386,
+      "step": 701
+    },
+    {
+      "epoch": 0.9486486486486486,
+      "grad_norm": 0.37331444025039673,
+      "learning_rate": 0.0005,
+      "loss": 0.9718385934829712,
+      "step": 702
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4383757412433624,
+      "learning_rate": 0.0005,
+      "loss": 1.0255049467086792,
+      "step": 703
+    },
+    {
+      "epoch": 0.9513513513513514,
+      "grad_norm": 0.4831158220767975,
+      "learning_rate": 0.0005,
+      "loss": 1.0184235572814941,
+      "step": 704
+    },
+    {
+      "epoch": 0.9527027027027027,
+      "grad_norm": 0.4493253529071808,
+      "learning_rate": 0.0005,
+      "loss": 1.0078189373016357,
+      "step": 705
+    },
+    {
+      "epoch": 0.9540540540540541,
+      "grad_norm": 0.4486171305179596,
+      "learning_rate": 0.0005,
+      "loss": 1.0064935684204102,
+      "step": 706
+    },
+    {
+      "epoch": 0.9554054054054054,
+      "grad_norm": 0.42528945207595825,
+      "learning_rate": 0.0005,
+      "loss": 0.9922582507133484,
+      "step": 707
+    },
+    {
+      "epoch": 0.9567567567567568,
+      "grad_norm": 0.4825116693973541,
+      "learning_rate": 0.0005,
+      "loss": 1.0048103332519531,
+      "step": 708
+    },
+    {
+      "epoch": 0.9581081081081081,
+      "grad_norm": 0.4591309130191803,
+      "learning_rate": 0.0005,
+      "loss": 0.9644536972045898,
+      "step": 709
+    },
+    {
+      "epoch": 0.9594594594594594,
+      "grad_norm": 0.4692993760108948,
+      "learning_rate": 0.0005,
+      "loss": 1.0626860857009888,
+      "step": 710
+    },
+    {
+      "epoch": 0.9608108108108108,
+      "grad_norm": 0.4549885094165802,
+      "learning_rate": 0.0005,
+      "loss": 0.9852452278137207,
+      "step": 711
+    },
+    {
+      "epoch": 0.9621621621621622,
+      "grad_norm": 0.596221923828125,
+      "learning_rate": 0.0005,
+      "loss": 1.057373046875,
+      "step": 712
+    },
+    {
+      "epoch": 0.9635135135135136,
+      "grad_norm": 0.48202916979789734,
+      "learning_rate": 0.0005,
+      "loss": 1.0333809852600098,
+      "step": 713
+    },
+    {
+      "epoch": 0.9648648648648649,
+      "grad_norm": 0.5182270407676697,
+      "learning_rate": 0.0005,
+      "loss": 0.9707871079444885,
+      "step": 714
+    },
+    {
+      "epoch": 0.9662162162162162,
+      "grad_norm": 0.5023300647735596,
+      "learning_rate": 0.0005,
+      "loss": 1.013512372970581,
+      "step": 715
+    },
+    {
+      "epoch": 0.9675675675675676,
+      "grad_norm": 0.4642000198364258,
+      "learning_rate": 0.0005,
+      "loss": 0.9938773512840271,
+      "step": 716
+    },
+    {
+      "epoch": 0.9689189189189189,
+      "grad_norm": 0.4536377787590027,
+      "learning_rate": 0.0005,
+      "loss": 1.0033628940582275,
+      "step": 717
+    },
+    {
+      "epoch": 0.9702702702702702,
+      "grad_norm": 0.47987979650497437,
+      "learning_rate": 0.0005,
+      "loss": 0.9904276132583618,
+      "step": 718
+    },
+    {
+      "epoch": 0.9716216216216216,
+      "grad_norm": 0.43377485871315,
+      "learning_rate": 0.0005,
+      "loss": 1.013549566268921,
+      "step": 719
+    },
+    {
+      "epoch": 0.972972972972973,
+      "grad_norm": 0.5736255645751953,
+      "learning_rate": 0.0005,
+      "loss": 1.0272995233535767,
+      "step": 720
+    },
+    {
+      "epoch": 0.9743243243243244,
+      "grad_norm": 0.44735679030418396,
+      "learning_rate": 0.0005,
+      "loss": 1.0366603136062622,
+      "step": 721
+    },
+    {
+      "epoch": 0.9756756756756757,
+      "grad_norm": 0.4559691846370697,
+      "learning_rate": 0.0005,
+      "loss": 1.0132989883422852,
+      "step": 722
+    },
+    {
+      "epoch": 0.977027027027027,
+      "grad_norm": 0.4669279158115387,
+      "learning_rate": 0.0005,
+      "loss": 1.014146327972412,
+      "step": 723
+    },
+    {
+      "epoch": 0.9783783783783784,
+      "grad_norm": 0.4849174916744232,
+      "learning_rate": 0.0005,
+      "loss": 0.9876691102981567,
+      "step": 724
+    },
+    {
+      "epoch": 0.9797297297297297,
+      "grad_norm": 0.45228511095046997,
+      "learning_rate": 0.0005,
+      "loss": 0.9816712141036987,
+      "step": 725
+    },
+    {
+      "epoch": 0.981081081081081,
+      "grad_norm": 0.4673261344432831,
+      "learning_rate": 0.0005,
+      "loss": 1.003387451171875,
+      "step": 726
+    },
+    {
+      "epoch": 0.9824324324324324,
+      "grad_norm": 0.4340033531188965,
+      "learning_rate": 0.0005,
+      "loss": 0.9941866397857666,
+      "step": 727
+    },
+    {
+      "epoch": 0.9837837837837838,
+      "grad_norm": 0.44876429438591003,
+      "learning_rate": 0.0005,
+      "loss": 0.9699825644493103,
+      "step": 728
+    },
+    {
+      "epoch": 0.9851351351351352,
+      "grad_norm": 0.418673574924469,
+      "learning_rate": 0.0005,
+      "loss": 0.9902230501174927,
+      "step": 729
+    },
+    {
+      "epoch": 0.9864864864864865,
+      "grad_norm": 0.4880320131778717,
+      "learning_rate": 0.0005,
+      "loss": 1.01579749584198,
+      "step": 730
+    },
+    {
+      "epoch": 0.9878378378378379,
+      "grad_norm": 0.41727185249328613,
+      "learning_rate": 0.0005,
+      "loss": 1.030842661857605,
+      "step": 731
+    },
+    {
+      "epoch": 0.9891891891891892,
+      "grad_norm": 0.48642903566360474,
+      "learning_rate": 0.0005,
+      "loss": 0.9593423008918762,
+      "step": 732
+    },
+    {
+      "epoch": 0.9905405405405405,
+      "grad_norm": 0.45963045954704285,
+      "learning_rate": 0.0005,
+      "loss": 1.0037460327148438,
+      "step": 733
+    },
+    {
+      "epoch": 0.9918918918918919,
+      "grad_norm": 0.4807617664337158,
+      "learning_rate": 0.0005,
+      "loss": 1.003387451171875,
+      "step": 734
+    },
+    {
+      "epoch": 0.9932432432432432,
+      "grad_norm": 0.44078657031059265,
+      "learning_rate": 0.0005,
+      "loss": 1.0330944061279297,
+      "step": 735
+    },
+    {
+      "epoch": 0.9945945945945946,
+      "grad_norm": 0.4119824469089508,
+      "learning_rate": 0.0005,
+      "loss": 0.9948251843452454,
+      "step": 736
+    },
+    {
+      "epoch": 0.995945945945946,
+      "grad_norm": 0.3926996886730194,
+      "learning_rate": 0.0005,
+      "loss": 0.9804766178131104,
+      "step": 737
+    },
+    {
+      "epoch": 0.9972972972972973,
+      "grad_norm": 0.3825426399707794,
+      "learning_rate": 0.0005,
+      "loss": 1.02166748046875,
+      "step": 738
+    },
+    {
+      "epoch": 0.9986486486486487,
+      "grad_norm": 0.44135546684265137,
+      "learning_rate": 0.0005,
+      "loss": 0.9986615777015686,
+      "step": 739
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.44254422187805176,
+      "learning_rate": 0.0005,
+      "loss": 1.0358532667160034,
+      "step": 740
+    },
+    {
+      "epoch": 1.0013513513513514,
+      "grad_norm": 0.37885668873786926,
+      "learning_rate": 0.0005,
+      "loss": 0.9826116561889648,
+      "step": 741
+    },
+    {
+      "epoch": 1.0027027027027027,
+      "grad_norm": 0.3668298125267029,
+      "learning_rate": 0.0005,
+      "loss": 0.986068844795227,
+      "step": 742
+    },
+    {
+      "epoch": 1.0040540540540541,
+      "grad_norm": 0.415840208530426,
+      "learning_rate": 0.0005,
+      "loss": 0.9781303405761719,
+      "step": 743
+    },
+    {
+      "epoch": 1.0054054054054054,
+      "grad_norm": 0.37847238779067993,
+      "learning_rate": 0.0005,
+      "loss": 0.9830222725868225,
+      "step": 744
+    },
+    {
+      "epoch": 1.0067567567567568,
+      "grad_norm": 0.48941272497177124,
+      "learning_rate": 0.0005,
+      "loss": 0.9910082221031189,
+      "step": 745
+    },
+    {
+      "epoch": 1.008108108108108,
+      "grad_norm": 0.43017080426216125,
+      "learning_rate": 0.0005,
+      "loss": 1.0052462816238403,
+      "step": 746
+    },
+    {
+      "epoch": 1.0094594594594595,
+      "grad_norm": 0.4088289439678192,
+      "learning_rate": 0.0005,
+      "loss": 0.9904218316078186,
+      "step": 747
+    },
+    {
+      "epoch": 1.0108108108108107,
+      "grad_norm": 0.383180171251297,
+      "learning_rate": 0.0005,
+      "loss": 0.9789303541183472,
+      "step": 748
+    },
+    {
+      "epoch": 1.0121621621621621,
+      "grad_norm": 0.440793514251709,
+      "learning_rate": 0.0005,
+      "loss": 1.021952748298645,
+      "step": 749
+    },
+    {
+      "epoch": 1.0135135135135136,
+      "grad_norm": 0.39574360847473145,
+      "learning_rate": 0.0005,
+      "loss": 0.967638373374939,
+      "step": 750
+    },
+    {
+      "epoch": 1.0148648648648648,
+      "grad_norm": 0.3977818787097931,
+      "learning_rate": 0.0005,
+      "loss": 0.9699274301528931,
+      "step": 751
+    },
+    {
+      "epoch": 1.0162162162162163,
+      "grad_norm": 0.4296707212924957,
+      "learning_rate": 0.0005,
+      "loss": 0.9848112463951111,
+      "step": 752
+    },
+    {
+      "epoch": 1.0175675675675675,
+      "grad_norm": 0.4194265305995941,
+      "learning_rate": 0.0005,
+      "loss": 1.004648208618164,
+      "step": 753
+    },
+    {
+      "epoch": 1.018918918918919,
+      "grad_norm": 0.3951985538005829,
+      "learning_rate": 0.0005,
+      "loss": 0.981384813785553,
+      "step": 754
+    },
+    {
+      "epoch": 1.0202702702702702,
+      "grad_norm": 0.44012582302093506,
+      "learning_rate": 0.0005,
+      "loss": 1.0142056941986084,
+      "step": 755
+    },
+    {
+      "epoch": 1.0216216216216216,
+      "grad_norm": 0.43094557523727417,
+      "learning_rate": 0.0005,
+      "loss": 0.9952385425567627,
+      "step": 756
+    },
+    {
+      "epoch": 1.022972972972973,
+      "grad_norm": 0.4329661428928375,
+      "learning_rate": 0.0005,
+      "loss": 1.0096514225006104,
+      "step": 757
+    },
+    {
+      "epoch": 1.0243243243243243,
+      "grad_norm": 0.4235393702983856,
+      "learning_rate": 0.0005,
+      "loss": 0.9986224174499512,
+      "step": 758
+    },
+    {
+      "epoch": 1.0256756756756757,
+      "grad_norm": 0.3728649318218231,
+      "learning_rate": 0.0005,
+      "loss": 0.9821822643280029,
+      "step": 759
+    },
+    {
+      "epoch": 1.027027027027027,
+      "grad_norm": 0.46902191638946533,
+      "learning_rate": 0.0005,
+      "loss": 1.0150288343429565,
+      "step": 760
+    },
+    {
+      "epoch": 1.0283783783783784,
+      "grad_norm": 0.43301528692245483,
+      "learning_rate": 0.0005,
+      "loss": 0.9662719964981079,
+      "step": 761
+    },
+    {
+      "epoch": 1.0297297297297296,
+      "grad_norm": 0.5160084366798401,
+      "learning_rate": 0.0005,
+      "loss": 1.0578382015228271,
+      "step": 762
+    },
+    {
+      "epoch": 1.031081081081081,
+      "grad_norm": 0.4197075366973877,
+      "learning_rate": 0.0005,
+      "loss": 0.9986023902893066,
+      "step": 763
+    },
+    {
+      "epoch": 1.0324324324324325,
+      "grad_norm": 0.4478892385959625,
+      "learning_rate": 0.0005,
+      "loss": 0.9914935827255249,
+      "step": 764
+    },
+    {
+      "epoch": 1.0337837837837838,
+      "grad_norm": 0.41060134768486023,
+      "learning_rate": 0.0005,
+      "loss": 0.9743323922157288,
+      "step": 765
+    },
+    {
+      "epoch": 1.0351351351351352,
+      "grad_norm": 0.31929466128349304,
+      "learning_rate": 0.0005,
+      "loss": 0.9460826516151428,
+      "step": 766
+    },
+    {
+      "epoch": 1.0364864864864864,
+      "grad_norm": 0.4298437833786011,
+      "learning_rate": 0.0005,
+      "loss": 0.9774943590164185,
+      "step": 767
+    },
+    {
+      "epoch": 1.037837837837838,
+      "grad_norm": 0.4531768262386322,
+      "learning_rate": 0.0005,
+      "loss": 1.0028963088989258,
+      "step": 768
+    },
+    {
+      "epoch": 1.0391891891891891,
+      "grad_norm": 0.40110403299331665,
+      "learning_rate": 0.0005,
+      "loss": 0.9730525016784668,
+      "step": 769
+    },
+    {
+      "epoch": 1.0405405405405406,
+      "grad_norm": 0.4374914765357971,
+      "learning_rate": 0.0005,
+      "loss": 0.9628942012786865,
+      "step": 770
+    },
+    {
+      "epoch": 1.0418918918918918,
+      "grad_norm": 0.4203137159347534,
+      "learning_rate": 0.0005,
+      "loss": 0.9660401344299316,
+      "step": 771
+    },
+    {
+      "epoch": 1.0432432432432432,
+      "grad_norm": 0.49443280696868896,
+      "learning_rate": 0.0005,
+      "loss": 1.0047507286071777,
+      "step": 772
+    },
+    {
+      "epoch": 1.0445945945945947,
+      "grad_norm": 0.37317997217178345,
+      "learning_rate": 0.0005,
+      "loss": 0.9557087421417236,
+      "step": 773
+    },
+    {
+      "epoch": 1.045945945945946,
+      "grad_norm": 0.39154237508773804,
+      "learning_rate": 0.0005,
+      "loss": 0.9681394100189209,
+      "step": 774
+    },
+    {
+      "epoch": 1.0472972972972974,
+      "grad_norm": 0.4881087839603424,
+      "learning_rate": 0.0005,
+      "loss": 1.0006524324417114,
+      "step": 775
+    },
+    {
+      "epoch": 1.0486486486486486,
+      "grad_norm": 0.4131862223148346,
+      "learning_rate": 0.0005,
+      "loss": 1.0090842247009277,
+      "step": 776
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.4063643515110016,
+      "learning_rate": 0.0005,
+      "loss": 0.9942796230316162,
+      "step": 777
+    },
+    {
+      "epoch": 1.0513513513513513,
+      "grad_norm": 0.511608898639679,
+      "learning_rate": 0.0005,
+      "loss": 0.9985835552215576,
+      "step": 778
+    },
+    {
+      "epoch": 1.0527027027027027,
+      "grad_norm": 0.37490230798721313,
+      "learning_rate": 0.0005,
+      "loss": 0.9590276479721069,
+      "step": 779
+    },
+    {
+      "epoch": 1.054054054054054,
+      "grad_norm": 0.37600603699684143,
+      "learning_rate": 0.0005,
+      "loss": 0.970512330532074,
+      "step": 780
+    },
+    {
+      "epoch": 1.0554054054054054,
+      "grad_norm": 0.4623628258705139,
+      "learning_rate": 0.0005,
+      "loss": 0.9996064901351929,
+      "step": 781
+    },
+    {
+      "epoch": 1.0567567567567568,
+      "grad_norm": 0.44364771246910095,
+      "learning_rate": 0.0005,
+      "loss": 0.9878872036933899,
+      "step": 782
+    },
+    {
+      "epoch": 1.058108108108108,
+      "grad_norm": 0.41689711809158325,
+      "learning_rate": 0.0005,
+      "loss": 0.9908361434936523,
+      "step": 783
+    },
+    {
+      "epoch": 1.0594594594594595,
+      "grad_norm": 0.3942039906978607,
+      "learning_rate": 0.0005,
+      "loss": 1.0063025951385498,
+      "step": 784
+    },
+    {
+      "epoch": 1.0608108108108107,
+      "grad_norm": 0.46740373969078064,
+      "learning_rate": 0.0005,
+      "loss": 1.0207891464233398,
+      "step": 785
+    },
+    {
+      "epoch": 1.0621621621621622,
+      "grad_norm": 0.4881350100040436,
+      "learning_rate": 0.0005,
+      "loss": 0.9953904151916504,
+      "step": 786
+    },
+    {
+      "epoch": 1.0635135135135134,
+      "grad_norm": 0.3690902292728424,
+      "learning_rate": 0.0005,
+      "loss": 0.9922307729721069,
+      "step": 787
+    },
+    {
+      "epoch": 1.0648648648648649,
+      "grad_norm": 0.3918338119983673,
+      "learning_rate": 0.0005,
+      "loss": 1.0009398460388184,
+      "step": 788
+    },
+    {
+      "epoch": 1.0662162162162163,
+      "grad_norm": 0.36540499329566956,
+      "learning_rate": 0.0005,
+      "loss": 0.9735972881317139,
+      "step": 789
+    },
+    {
+      "epoch": 1.0675675675675675,
+      "grad_norm": 0.4220549166202545,
+      "learning_rate": 0.0005,
+      "loss": 0.9815663695335388,
+      "step": 790
+    },
+    {
+      "epoch": 1.068918918918919,
+      "grad_norm": 0.42926225066185,
+      "learning_rate": 0.0005,
+      "loss": 0.996827244758606,
+      "step": 791
+    },
+    {
+      "epoch": 1.0702702702702702,
+      "grad_norm": 0.39149850606918335,
+      "learning_rate": 0.0005,
+      "loss": 0.9812740087509155,
+      "step": 792
+    },
+    {
+      "epoch": 1.0716216216216217,
+      "grad_norm": 0.4387616217136383,
+      "learning_rate": 0.0005,
+      "loss": 0.9877827763557434,
+      "step": 793
+    },
+    {
+      "epoch": 1.072972972972973,
+      "grad_norm": 0.37845513224601746,
+      "learning_rate": 0.0005,
+      "loss": 0.9870060086250305,
+      "step": 794
+    },
+    {
+      "epoch": 1.0743243243243243,
+      "grad_norm": 0.3980821669101715,
+      "learning_rate": 0.0005,
+      "loss": 0.9776878952980042,
+      "step": 795
+    },
+    {
+      "epoch": 1.0756756756756758,
+      "grad_norm": 0.42030075192451477,
+      "learning_rate": 0.0005,
+      "loss": 0.9920201301574707,
+      "step": 796
+    },
+    {
+      "epoch": 1.077027027027027,
+      "grad_norm": 0.4027842879295349,
+      "learning_rate": 0.0005,
+      "loss": 0.9821338653564453,
+      "step": 797
+    },
+    {
+      "epoch": 1.0783783783783785,
+      "grad_norm": 0.4237661361694336,
+      "learning_rate": 0.0005,
+      "loss": 1.0183025598526,
+      "step": 798
+    },
+    {
+      "epoch": 1.0797297297297297,
+      "grad_norm": 0.4317850172519684,
+      "learning_rate": 0.0005,
+      "loss": 0.9741783142089844,
+      "step": 799
+    },
+    {
+      "epoch": 1.0810810810810811,
+      "grad_norm": 0.4845432937145233,
+      "learning_rate": 0.0005,
+      "loss": 1.025728464126587,
+      "step": 800
+    },
+    {
+      "epoch": 1.0824324324324324,
+      "grad_norm": 0.5068687796592712,
+      "learning_rate": 0.0005,
+      "loss": 0.9984809756278992,
+      "step": 801
+    },
+    {
+      "epoch": 1.0837837837837838,
+      "grad_norm": 0.459915429353714,
+      "learning_rate": 0.0005,
+      "loss": 1.012566328048706,
+      "step": 802
+    },
+    {
+      "epoch": 1.085135135135135,
+      "grad_norm": 0.4352189302444458,
+      "learning_rate": 0.0005,
+      "loss": 0.9765508770942688,
+      "step": 803
+    },
+    {
+      "epoch": 1.0864864864864865,
+      "grad_norm": 0.4997720420360565,
+      "learning_rate": 0.0005,
+      "loss": 1.0253517627716064,
+      "step": 804
+    },
+    {
+      "epoch": 1.087837837837838,
+      "grad_norm": 0.40069344639778137,
+      "learning_rate": 0.0005,
+      "loss": 0.9930379986763,
+      "step": 805
+    },
+    {
+      "epoch": 1.0891891891891892,
+      "grad_norm": 0.49873989820480347,
+      "learning_rate": 0.0005,
+      "loss": 0.9931718707084656,
+      "step": 806
+    },
+    {
+      "epoch": 1.0905405405405406,
+      "grad_norm": 0.5117582678794861,
+      "learning_rate": 0.0005,
+      "loss": 1.0159685611724854,
+      "step": 807
+    },
+    {
+      "epoch": 1.0918918918918918,
+      "grad_norm": 0.4322623014450073,
+      "learning_rate": 0.0005,
+      "loss": 0.9705570936203003,
+      "step": 808
+    },
+    {
+      "epoch": 1.0932432432432433,
+      "grad_norm": 0.4055192172527313,
+      "learning_rate": 0.0005,
+      "loss": 1.0086268186569214,
+      "step": 809
+    },
+    {
+      "epoch": 1.0945945945945945,
+      "grad_norm": 0.4433877468109131,
+      "learning_rate": 0.0005,
+      "loss": 0.9911096096038818,
+      "step": 810
+    },
+    {
+      "epoch": 1.095945945945946,
+      "grad_norm": 0.4258456826210022,
+      "learning_rate": 0.0005,
+      "loss": 0.967620849609375,
+      "step": 811
+    },
+    {
+      "epoch": 1.0972972972972972,
+      "grad_norm": 0.3873337209224701,
+      "learning_rate": 0.0005,
+      "loss": 0.9909137487411499,
+      "step": 812
+    },
+    {
+      "epoch": 1.0986486486486486,
+      "grad_norm": 0.48253577947616577,
+      "learning_rate": 0.0005,
+      "loss": 1.0168708562850952,
+      "step": 813
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.46015387773513794,
+      "learning_rate": 0.0005,
+      "loss": 1.006185531616211,
+      "step": 814
+    },
+    {
+      "epoch": 1.1013513513513513,
+      "grad_norm": 0.37485411763191223,
+      "learning_rate": 0.0005,
+      "loss": 0.9788612127304077,
+      "step": 815
+    },
+    {
+      "epoch": 1.1027027027027028,
+      "grad_norm": 0.3654661178588867,
+      "learning_rate": 0.0005,
+      "loss": 0.983730673789978,
+      "step": 816
+    },
+    {
+      "epoch": 1.104054054054054,
+      "grad_norm": 0.41117939352989197,
+      "learning_rate": 0.0005,
+      "loss": 0.9914546012878418,
+      "step": 817
+    },
+    {
+      "epoch": 1.1054054054054054,
+      "grad_norm": 0.4432920515537262,
+      "learning_rate": 0.0005,
+      "loss": 0.9879539012908936,
+      "step": 818
+    },
+    {
+      "epoch": 1.1067567567567567,
+      "grad_norm": 0.42025670409202576,
+      "learning_rate": 0.0005,
+      "loss": 0.9653362035751343,
+      "step": 819
+    },
+    {
+      "epoch": 1.1081081081081081,
+      "grad_norm": 0.4595154821872711,
+      "learning_rate": 0.0005,
+      "loss": 0.9819380044937134,
+      "step": 820
+    },
+    {
+      "epoch": 1.1094594594594596,
+      "grad_norm": 0.38901224732398987,
+      "learning_rate": 0.0005,
+      "loss": 0.9525375366210938,
+      "step": 821
+    },
+    {
+      "epoch": 1.1108108108108108,
+      "grad_norm": 0.5015517473220825,
+      "learning_rate": 0.0005,
+      "loss": 0.9997296333312988,
+      "step": 822
+    },
+    {
+      "epoch": 1.1121621621621622,
+      "grad_norm": 0.46231982111930847,
+      "learning_rate": 0.0005,
+      "loss": 0.992017924785614,
+      "step": 823
+    },
+    {
+      "epoch": 1.1135135135135135,
+      "grad_norm": 0.40264010429382324,
+      "learning_rate": 0.0005,
+      "loss": 0.9629871845245361,
+      "step": 824
+    },
+    {
+      "epoch": 1.114864864864865,
+      "grad_norm": 0.3680342733860016,
+      "learning_rate": 0.0005,
+      "loss": 0.9535119533538818,
+      "step": 825
+    },
+    {
+      "epoch": 1.1162162162162161,
+      "grad_norm": 0.4165852963924408,
+      "learning_rate": 0.0005,
+      "loss": 0.9866558313369751,
+      "step": 826
+    },
+    {
+      "epoch": 1.1175675675675676,
+      "grad_norm": 0.4650227725505829,
+      "learning_rate": 0.0005,
+      "loss": 1.0092779397964478,
+      "step": 827
+    },
+    {
+      "epoch": 1.118918918918919,
+      "grad_norm": 0.4051968455314636,
+      "learning_rate": 0.0005,
+      "loss": 0.9942563772201538,
+      "step": 828
+    },
+    {
+      "epoch": 1.1202702702702703,
+      "grad_norm": 0.4908919334411621,
+      "learning_rate": 0.0005,
+      "loss": 0.9954160451889038,
+      "step": 829
+    },
+    {
+      "epoch": 1.1216216216216217,
+      "grad_norm": 0.43544965982437134,
+      "learning_rate": 0.0005,
+      "loss": 0.9777190685272217,
+      "step": 830
+    },
+    {
+      "epoch": 1.122972972972973,
+      "grad_norm": 0.395830899477005,
+      "learning_rate": 0.0005,
+      "loss": 0.9893279671669006,
+      "step": 831
+    },
+    {
+      "epoch": 1.1243243243243244,
+      "grad_norm": 0.37831902503967285,
+      "learning_rate": 0.0005,
+      "loss": 0.9790507555007935,
+      "step": 832
+    },
+    {
+      "epoch": 1.1256756756756756,
+      "grad_norm": 0.4537504315376282,
+      "learning_rate": 0.0005,
+      "loss": 0.9905118346214294,
+      "step": 833
+    },
+    {
+      "epoch": 1.127027027027027,
+      "grad_norm": 0.3877454400062561,
+      "learning_rate": 0.0005,
+      "loss": 0.9418052434921265,
+      "step": 834
+    },
+    {
+      "epoch": 1.1283783783783783,
+      "grad_norm": 0.4418070316314697,
+      "learning_rate": 0.0005,
+      "loss": 0.9883332848548889,
+      "step": 835
+    },
+    {
+      "epoch": 1.1297297297297297,
+      "grad_norm": 0.551307737827301,
+      "learning_rate": 0.0005,
+      "loss": 0.9886205196380615,
+      "step": 836
+    },
+    {
+      "epoch": 1.131081081081081,
+      "grad_norm": 0.39950889348983765,
+      "learning_rate": 0.0005,
+      "loss": 0.9822620153427124,
+      "step": 837
+    },
+    {
+      "epoch": 1.1324324324324324,
+      "grad_norm": 0.4230376183986664,
+      "learning_rate": 0.0005,
+      "loss": 0.9725573062896729,
+      "step": 838
+    },
+    {
+      "epoch": 1.1337837837837839,
+      "grad_norm": 0.48267409205436707,
+      "learning_rate": 0.0005,
+      "loss": 1.0101895332336426,
+      "step": 839
+    },
+    {
+      "epoch": 1.135135135135135,
+      "grad_norm": 0.41957321763038635,
+      "learning_rate": 0.0005,
+      "loss": 0.996134877204895,
+      "step": 840
+    },
+    {
+      "epoch": 1.1364864864864865,
+      "grad_norm": 0.4457937180995941,
+      "learning_rate": 0.0005,
+      "loss": 0.9777142405509949,
+      "step": 841
+    },
+    {
+      "epoch": 1.1378378378378378,
+      "grad_norm": 0.39050379395484924,
+      "learning_rate": 0.0005,
+      "loss": 0.997581422328949,
+      "step": 842
+    },
+    {
+      "epoch": 1.1391891891891892,
+      "grad_norm": 0.37530946731567383,
+      "learning_rate": 0.0005,
+      "loss": 0.9691987633705139,
+      "step": 843
+    },
+    {
+      "epoch": 1.1405405405405404,
+      "grad_norm": 0.5874009728431702,
+      "learning_rate": 0.0005,
+      "loss": 1.0070269107818604,
+      "step": 844
+    },
+    {
+      "epoch": 1.1418918918918919,
+      "grad_norm": 0.41032925248146057,
+      "learning_rate": 0.0005,
+      "loss": 0.9640153646469116,
+      "step": 845
+    },
+    {
+      "epoch": 1.1432432432432433,
+      "grad_norm": 0.4330558776855469,
+      "learning_rate": 0.0005,
+      "loss": 0.968708336353302,
+      "step": 846
+    },
+    {
+      "epoch": 1.1445945945945946,
+      "grad_norm": 0.5227062106132507,
+      "learning_rate": 0.0005,
+      "loss": 0.9941346645355225,
+      "step": 847
+    },
+    {
+      "epoch": 1.145945945945946,
+      "grad_norm": 0.35492250323295593,
+      "learning_rate": 0.0005,
+      "loss": 0.9570294618606567,
+      "step": 848
+    },
+    {
+      "epoch": 1.1472972972972972,
+      "grad_norm": 0.3751964867115021,
+      "learning_rate": 0.0005,
+      "loss": 0.9554640054702759,
+      "step": 849
+    },
+    {
+      "epoch": 1.1486486486486487,
+      "grad_norm": 0.5041049718856812,
+      "learning_rate": 0.0005,
+      "loss": 1.0386898517608643,
+      "step": 850
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.44854143261909485,
+      "learning_rate": 0.0005,
+      "loss": 0.9746886491775513,
+      "step": 851
+    },
+    {
+      "epoch": 1.1513513513513514,
+      "grad_norm": 0.3769172430038452,
+      "learning_rate": 0.0005,
+      "loss": 0.9356929659843445,
+      "step": 852
+    },
+    {
+      "epoch": 1.1527027027027028,
+      "grad_norm": 0.4356090724468231,
+      "learning_rate": 0.0005,
+      "loss": 0.9625841975212097,
+      "step": 853
+    },
+    {
+      "epoch": 1.154054054054054,
+      "grad_norm": 0.4069555103778839,
+      "learning_rate": 0.0005,
+      "loss": 0.993695855140686,
+      "step": 854
+    },
+    {
+      "epoch": 1.1554054054054055,
+      "grad_norm": 0.5022456645965576,
+      "learning_rate": 0.0005,
+      "loss": 0.9907647371292114,
+      "step": 855
+    },
+    {
+      "epoch": 1.1567567567567567,
+      "grad_norm": 0.45313477516174316,
+      "learning_rate": 0.0005,
+      "loss": 1.0015625953674316,
+      "step": 856
+    },
+    {
+      "epoch": 1.1581081081081082,
+      "grad_norm": 0.3406035900115967,
+      "learning_rate": 0.0005,
+      "loss": 0.9502585530281067,
+      "step": 857
+    },
+    {
+      "epoch": 1.1594594594594594,
+      "grad_norm": 0.48522844910621643,
+      "learning_rate": 0.0005,
+      "loss": 1.0128092765808105,
+      "step": 858
+    },
+    {
+      "epoch": 1.1608108108108108,
+      "grad_norm": 0.3666231334209442,
+      "learning_rate": 0.0005,
+      "loss": 0.9618005752563477,
+      "step": 859
+    },
+    {
+      "epoch": 1.1621621621621623,
+      "grad_norm": 0.42206043004989624,
+      "learning_rate": 0.0005,
+      "loss": 0.9620918035507202,
+      "step": 860
+    },
+    {
+      "epoch": 1.1635135135135135,
+      "grad_norm": 0.43938133120536804,
+      "learning_rate": 0.0005,
+      "loss": 0.9881910681724548,
+      "step": 861
+    },
+    {
+      "epoch": 1.164864864864865,
+      "grad_norm": 0.4852261245250702,
+      "learning_rate": 0.0005,
+      "loss": 0.9718068838119507,
+      "step": 862
+    },
+    {
+      "epoch": 1.1662162162162162,
+      "grad_norm": 0.5131728649139404,
+      "learning_rate": 0.0005,
+      "loss": 0.9992783665657043,
+      "step": 863
+    },
+    {
+      "epoch": 1.1675675675675676,
+      "grad_norm": 0.3744819164276123,
+      "learning_rate": 0.0005,
+      "loss": 0.967980146408081,
+      "step": 864
+    },
+    {
+      "epoch": 1.1689189189189189,
+      "grad_norm": 0.5058520436286926,
+      "learning_rate": 0.0005,
+      "loss": 1.0149688720703125,
+      "step": 865
+    },
+    {
+      "epoch": 1.1702702702702703,
+      "grad_norm": 0.3992828130722046,
+      "learning_rate": 0.0005,
+      "loss": 0.9878426790237427,
+      "step": 866
+    },
+    {
+      "epoch": 1.1716216216216215,
+      "grad_norm": 0.40857842564582825,
+      "learning_rate": 0.0005,
+      "loss": 0.9946902990341187,
+      "step": 867
+    },
+    {
+      "epoch": 1.172972972972973,
+      "grad_norm": 0.4472295045852661,
+      "learning_rate": 0.0005,
+      "loss": 0.9823582768440247,
+      "step": 868
+    },
+    {
+      "epoch": 1.1743243243243242,
+      "grad_norm": 0.4157494306564331,
+      "learning_rate": 0.0005,
+      "loss": 0.9786185026168823,
+      "step": 869
+    },
+    {
+      "epoch": 1.1756756756756757,
+      "grad_norm": 0.43595990538597107,
+      "learning_rate": 0.0005,
+      "loss": 0.9506538510322571,
+      "step": 870
+    },
+    {
+      "epoch": 1.177027027027027,
+      "grad_norm": 0.44766631722450256,
+      "learning_rate": 0.0005,
+      "loss": 0.9727016687393188,
+      "step": 871
+    },
+    {
+      "epoch": 1.1783783783783783,
+      "grad_norm": 0.42148756980895996,
+      "learning_rate": 0.0005,
+      "loss": 0.9690545201301575,
+      "step": 872
+    },
+    {
+      "epoch": 1.1797297297297298,
+      "grad_norm": 0.47412413358688354,
+      "learning_rate": 0.0005,
+      "loss": 0.9919174313545227,
+      "step": 873
+    },
+    {
+      "epoch": 1.181081081081081,
+      "grad_norm": 0.5118183493614197,
+      "learning_rate": 0.0005,
+      "loss": 0.9823079109191895,
+      "step": 874
+    },
+    {
+      "epoch": 1.1824324324324325,
+      "grad_norm": 0.46459686756134033,
+      "learning_rate": 0.0005,
+      "loss": 1.0050690174102783,
+      "step": 875
+    },
+    {
+      "epoch": 1.1837837837837837,
+      "grad_norm": 0.44819289445877075,
+      "learning_rate": 0.0005,
+      "loss": 0.981884241104126,
+      "step": 876
+    },
+    {
+      "epoch": 1.1851351351351351,
+      "grad_norm": 0.4234590232372284,
+      "learning_rate": 0.0005,
+      "loss": 0.9750386476516724,
+      "step": 877
+    },
+    {
+      "epoch": 1.1864864864864866,
+      "grad_norm": 0.4923408329486847,
+      "learning_rate": 0.0005,
+      "loss": 0.9821457862854004,
+      "step": 878
+    },
+    {
+      "epoch": 1.1878378378378378,
+      "grad_norm": 0.36872532963752747,
+      "learning_rate": 0.0005,
+      "loss": 0.9502338171005249,
+      "step": 879
+    },
+    {
+      "epoch": 1.1891891891891893,
+      "grad_norm": 0.44929102063179016,
+      "learning_rate": 0.0005,
+      "loss": 1.0309237241744995,
+      "step": 880
+    },
+    {
+      "epoch": 1.1905405405405405,
+      "grad_norm": 0.4514937102794647,
+      "learning_rate": 0.0005,
+      "loss": 0.9855178594589233,
+      "step": 881
+    },
+    {
+      "epoch": 1.191891891891892,
+      "grad_norm": 0.36711838841438293,
+      "learning_rate": 0.0005,
+      "loss": 0.9521955251693726,
+      "step": 882
+    },
+    {
+      "epoch": 1.1932432432432432,
+      "grad_norm": 0.3700568675994873,
+      "learning_rate": 0.0005,
+      "loss": 0.9675923585891724,
+      "step": 883
+    },
+    {
+      "epoch": 1.1945945945945946,
+      "grad_norm": 0.3703731596469879,
+      "learning_rate": 0.0005,
+      "loss": 0.9626119136810303,
+      "step": 884
+    },
+    {
+      "epoch": 1.195945945945946,
+      "grad_norm": 0.45183035731315613,
+      "learning_rate": 0.0005,
+      "loss": 0.9842014312744141,
+      "step": 885
+    },
+    {
+      "epoch": 1.1972972972972973,
+      "grad_norm": 0.40317144989967346,
+      "learning_rate": 0.0005,
+      "loss": 0.9743790030479431,
+      "step": 886
+    },
+    {
+      "epoch": 1.1986486486486487,
+      "grad_norm": 0.4029059410095215,
+      "learning_rate": 0.0005,
+      "loss": 0.9507811665534973,
+      "step": 887
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.3990880250930786,
+      "learning_rate": 0.0005,
+      "loss": 0.940119743347168,
+      "step": 888
+    },
+    {
+      "epoch": 1.2013513513513514,
+      "grad_norm": 0.4123392105102539,
+      "learning_rate": 0.0005,
+      "loss": 0.9745818376541138,
+      "step": 889
+    },
+    {
+      "epoch": 1.2027027027027026,
+      "grad_norm": 0.49898815155029297,
+      "learning_rate": 0.0005,
+      "loss": 0.987059473991394,
+      "step": 890
+    },
+    {
+      "epoch": 1.204054054054054,
+      "grad_norm": 0.45183414220809937,
+      "learning_rate": 0.0005,
+      "loss": 0.9818235039710999,
+      "step": 891
+    },
+    {
+      "epoch": 1.2054054054054055,
+      "grad_norm": 0.46222662925720215,
+      "learning_rate": 0.0005,
+      "loss": 1.0357893705368042,
+      "step": 892
+    },
+    {
+      "epoch": 1.2067567567567568,
+      "grad_norm": 0.38398125767707825,
+      "learning_rate": 0.0005,
+      "loss": 0.9495432376861572,
+      "step": 893
+    },
+    {
+      "epoch": 1.2081081081081082,
+      "grad_norm": 0.3631584346294403,
+      "learning_rate": 0.0005,
+      "loss": 0.9883661866188049,
+      "step": 894
+    },
+    {
+      "epoch": 1.2094594594594594,
+      "grad_norm": 0.4845236539840698,
+      "learning_rate": 0.0005,
+      "loss": 0.9936016798019409,
+      "step": 895
+    },
+    {
+      "epoch": 1.2108108108108109,
+      "grad_norm": 0.46293485164642334,
+      "learning_rate": 0.0005,
+      "loss": 0.9642701745033264,
+      "step": 896
+    },
+    {
+      "epoch": 1.212162162162162,
+      "grad_norm": 0.36568236351013184,
+      "learning_rate": 0.0005,
+      "loss": 0.9666458964347839,
+      "step": 897
+    },
+    {
+      "epoch": 1.2135135135135136,
+      "grad_norm": 0.45591163635253906,
+      "learning_rate": 0.0005,
+      "loss": 0.9928126931190491,
+      "step": 898
+    },
+    {
+      "epoch": 1.2148648648648648,
+      "grad_norm": 0.4901658594608307,
+      "learning_rate": 0.0005,
+      "loss": 1.028685450553894,
+      "step": 899
+    },
+    {
+      "epoch": 1.2162162162162162,
+      "grad_norm": 0.37241387367248535,
+      "learning_rate": 0.0005,
+      "loss": 0.9426456689834595,
+      "step": 900
+    },
+    {
+      "epoch": 1.2175675675675675,
+      "grad_norm": 0.4686684310436249,
+      "learning_rate": 0.0005,
+      "loss": 1.0105820894241333,
+      "step": 901
+    },
+    {
+      "epoch": 1.218918918918919,
+      "grad_norm": 0.4077851176261902,
+      "learning_rate": 0.0005,
+      "loss": 0.9757517576217651,
+      "step": 902
+    },
+    {
+      "epoch": 1.2202702702702704,
+      "grad_norm": 0.3939678966999054,
+      "learning_rate": 0.0005,
+      "loss": 0.9840407967567444,
+      "step": 903
+    },
+    {
+      "epoch": 1.2216216216216216,
+      "grad_norm": 0.4174797534942627,
+      "learning_rate": 0.0005,
+      "loss": 0.9804435968399048,
+      "step": 904
+    },
+    {
+      "epoch": 1.222972972972973,
+      "grad_norm": 0.438797265291214,
+      "learning_rate": 0.0005,
+      "loss": 0.9985328912734985,
+      "step": 905
+    },
+    {
+      "epoch": 1.2243243243243243,
+      "grad_norm": 0.37842798233032227,
+      "learning_rate": 0.0005,
+      "loss": 0.9844509363174438,
+      "step": 906
+    },
+    {
+      "epoch": 1.2256756756756757,
+      "grad_norm": 0.41690170764923096,
+      "learning_rate": 0.0005,
+      "loss": 0.9616929292678833,
+      "step": 907
+    },
+    {
+      "epoch": 1.227027027027027,
+      "grad_norm": 0.33231908082962036,
+      "learning_rate": 0.0005,
+      "loss": 0.952788233757019,
+      "step": 908
+    },
+    {
+      "epoch": 1.2283783783783784,
+      "grad_norm": 0.41679006814956665,
+      "learning_rate": 0.0005,
+      "loss": 0.9759970903396606,
+      "step": 909
+    },
+    {
+      "epoch": 1.2297297297297298,
+      "grad_norm": 0.3732956051826477,
+      "learning_rate": 0.0005,
+      "loss": 0.9596526622772217,
+      "step": 910
+    },
+    {
+      "epoch": 1.231081081081081,
+      "grad_norm": 0.42420679330825806,
+      "learning_rate": 0.0005,
+      "loss": 1.0124093294143677,
+      "step": 911
+    },
+    {
+      "epoch": 1.2324324324324325,
+      "grad_norm": 0.465690016746521,
+      "learning_rate": 0.0005,
+      "loss": 0.9834306240081787,
+      "step": 912
+    },
+    {
+      "epoch": 1.2337837837837837,
+      "grad_norm": 0.4970438778400421,
+      "learning_rate": 0.0005,
+      "loss": 0.9862213134765625,
+      "step": 913
+    },
+    {
+      "epoch": 1.2351351351351352,
+      "grad_norm": 0.4276185929775238,
+      "learning_rate": 0.0005,
+      "loss": 0.9632750153541565,
+      "step": 914
+    },
+    {
+      "epoch": 1.2364864864864864,
+      "grad_norm": 0.40072494745254517,
+      "learning_rate": 0.0005,
+      "loss": 0.9751795530319214,
+      "step": 915
+    },
+    {
+      "epoch": 1.2378378378378379,
+      "grad_norm": 0.5106648802757263,
+      "learning_rate": 0.0005,
+      "loss": 0.9795105457305908,
+      "step": 916
+    },
+    {
+      "epoch": 1.2391891891891893,
+      "grad_norm": 0.4717729091644287,
+      "learning_rate": 0.0005,
+      "loss": 0.9964693784713745,
+      "step": 917
+    },
+    {
+      "epoch": 1.2405405405405405,
+      "grad_norm": 0.36892959475517273,
+      "learning_rate": 0.0005,
+      "loss": 0.9960974454879761,
+      "step": 918
+    },
+    {
+      "epoch": 1.241891891891892,
+      "grad_norm": 0.4919690191745758,
+      "learning_rate": 0.0005,
+      "loss": 0.9645310044288635,
+      "step": 919
+    },
+    {
+      "epoch": 1.2432432432432432,
+      "grad_norm": 0.40984392166137695,
+      "learning_rate": 0.0005,
+      "loss": 0.9893777370452881,
+      "step": 920
+    },
+    {
+      "epoch": 1.2445945945945946,
+      "grad_norm": 0.39653337001800537,
+      "learning_rate": 0.0005,
+      "loss": 1.007763147354126,
+      "step": 921
+    },
+    {
+      "epoch": 1.2459459459459459,
+      "grad_norm": 0.42782747745513916,
+      "learning_rate": 0.0005,
+      "loss": 0.9583773612976074,
+      "step": 922
+    },
+    {
+      "epoch": 1.2472972972972973,
+      "grad_norm": 0.4824948012828827,
+      "learning_rate": 0.0005,
+      "loss": 0.9618140459060669,
+      "step": 923
+    },
+    {
+      "epoch": 1.2486486486486488,
+      "grad_norm": 0.44346165657043457,
+      "learning_rate": 0.0005,
+      "loss": 0.9623422026634216,
+      "step": 924
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5287876129150391,
+      "learning_rate": 0.0005,
+      "loss": 0.9742758274078369,
+      "step": 925
+    },
+    {
+      "epoch": 1.2513513513513512,
+      "grad_norm": 0.37247171998023987,
+      "learning_rate": 0.0005,
+      "loss": 0.962054967880249,
+      "step": 926
+    },
+    {
+      "epoch": 1.2527027027027027,
+      "grad_norm": 0.46343907713890076,
+      "learning_rate": 0.0005,
+      "loss": 0.9845504760742188,
+      "step": 927
+    },
+    {
+      "epoch": 1.2540540540540541,
+      "grad_norm": 0.4039747416973114,
+      "learning_rate": 0.0005,
+      "loss": 0.9674913287162781,
+      "step": 928
+    },
+    {
+      "epoch": 1.2554054054054054,
+      "grad_norm": 0.5462550520896912,
+      "learning_rate": 0.0005,
+      "loss": 1.0072369575500488,
+      "step": 929
+    },
+    {
+      "epoch": 1.2567567567567568,
+      "grad_norm": 0.46702712774276733,
+      "learning_rate": 0.0005,
+      "loss": 0.9652385115623474,
+      "step": 930
+    },
+    {
+      "epoch": 1.258108108108108,
+      "grad_norm": 0.47799572348594666,
+      "learning_rate": 0.0005,
+      "loss": 0.9889237284660339,
+      "step": 931
+    },
+    {
+      "epoch": 1.2594594594594595,
+      "grad_norm": 0.4744010865688324,
+      "learning_rate": 0.0005,
+      "loss": 0.9937708377838135,
+      "step": 932
+    },
+    {
+      "epoch": 1.2608108108108107,
+      "grad_norm": 0.48625755310058594,
+      "learning_rate": 0.0005,
+      "loss": 0.9937934279441833,
+      "step": 933
+    },
+    {
+      "epoch": 1.2621621621621621,
+      "grad_norm": 0.481217622756958,
+      "learning_rate": 0.0005,
+      "loss": 0.9892547726631165,
+      "step": 934
+    },
+    {
+      "epoch": 1.2635135135135136,
+      "grad_norm": 0.41946983337402344,
+      "learning_rate": 0.0005,
+      "loss": 0.9595297574996948,
+      "step": 935
+    },
+    {
+      "epoch": 1.2648648648648648,
+      "grad_norm": 0.49870550632476807,
+      "learning_rate": 0.0005,
+      "loss": 0.9604277014732361,
+      "step": 936
+    },
+    {
+      "epoch": 1.2662162162162163,
+      "grad_norm": 0.4098280668258667,
+      "learning_rate": 0.0005,
+      "loss": 0.95635586977005,
+      "step": 937
+    },
+    {
+      "epoch": 1.2675675675675675,
+      "grad_norm": 0.4272097945213318,
+      "learning_rate": 0.0005,
+      "loss": 0.9779930114746094,
+      "step": 938
+    },
+    {
+      "epoch": 1.268918918918919,
+      "grad_norm": 0.49592000246047974,
+      "learning_rate": 0.0005,
+      "loss": 0.9965775609016418,
+      "step": 939
+    },
+    {
+      "epoch": 1.2702702702702702,
+      "grad_norm": 0.4345223009586334,
+      "learning_rate": 0.0005,
+      "loss": 1.0044713020324707,
+      "step": 940
+    },
+    {
+      "epoch": 1.2716216216216216,
+      "grad_norm": 0.4642544686794281,
+      "learning_rate": 0.0005,
+      "loss": 1.0084874629974365,
+      "step": 941
+    },
+    {
+      "epoch": 1.272972972972973,
+      "grad_norm": 0.3916667699813843,
+      "learning_rate": 0.0005,
+      "loss": 0.9626407027244568,
+      "step": 942
+    },
+    {
+      "epoch": 1.2743243243243243,
+      "grad_norm": 0.5512668490409851,
+      "learning_rate": 0.0005,
+      "loss": 0.9727204442024231,
+      "step": 943
+    },
+    {
+      "epoch": 1.2756756756756757,
+      "grad_norm": 0.46183592081069946,
+      "learning_rate": 0.0005,
+      "loss": 0.9692797064781189,
+      "step": 944
+    },
+    {
+      "epoch": 1.277027027027027,
+      "grad_norm": 0.4638822376728058,
+      "learning_rate": 0.0005,
+      "loss": 0.9989911317825317,
+      "step": 945
+    },
+    {
+      "epoch": 1.2783783783783784,
+      "grad_norm": 0.44250357151031494,
+      "learning_rate": 0.0005,
+      "loss": 0.9643974304199219,
+      "step": 946
+    },
+    {
+      "epoch": 1.2797297297297296,
+      "grad_norm": 0.36233463883399963,
+      "learning_rate": 0.0005,
+      "loss": 0.9523816108703613,
+      "step": 947
+    },
+    {
+      "epoch": 1.281081081081081,
+      "grad_norm": 0.41845154762268066,
+      "learning_rate": 0.0005,
+      "loss": 0.9806405305862427,
+      "step": 948
+    },
+    {
+      "epoch": 1.2824324324324325,
+      "grad_norm": 0.47220781445503235,
+      "learning_rate": 0.0005,
+      "loss": 0.9782319068908691,
+      "step": 949
+    },
+    {
+      "epoch": 1.2837837837837838,
+      "grad_norm": 0.3892207145690918,
+      "learning_rate": 0.0005,
+      "loss": 0.9769282937049866,
+      "step": 950
+    },
+    {
+      "epoch": 1.285135135135135,
+      "grad_norm": 0.3648059666156769,
+      "learning_rate": 0.0005,
+      "loss": 0.9834216833114624,
+      "step": 951
+    },
+    {
+      "epoch": 1.2864864864864864,
+      "grad_norm": 0.4654182195663452,
+      "learning_rate": 0.0005,
+      "loss": 0.9817870259284973,
+      "step": 952
+    },
+    {
+      "epoch": 1.287837837837838,
+      "grad_norm": 0.472591757774353,
+      "learning_rate": 0.0005,
+      "loss": 0.99371737241745,
+      "step": 953
+    },
+    {
+      "epoch": 1.2891891891891891,
+      "grad_norm": 0.3880445957183838,
+      "learning_rate": 0.0005,
+      "loss": 0.957332968711853,
+      "step": 954
+    },
+    {
+      "epoch": 1.2905405405405406,
+      "grad_norm": 0.3981819748878479,
+      "learning_rate": 0.0005,
+      "loss": 0.9519734382629395,
+      "step": 955
+    },
+    {
+      "epoch": 1.291891891891892,
+      "grad_norm": 0.4885483682155609,
+      "learning_rate": 0.0005,
+      "loss": 0.9721481800079346,
+      "step": 956
+    },
+    {
+      "epoch": 1.2932432432432432,
+      "grad_norm": 0.3798266649246216,
+      "learning_rate": 0.0005,
+      "loss": 0.9571473002433777,
+      "step": 957
+    },
+    {
+      "epoch": 1.2945945945945945,
+      "grad_norm": 0.4167742133140564,
+      "learning_rate": 0.0005,
+      "loss": 0.9730678796768188,
+      "step": 958
+    },
+    {
+      "epoch": 1.295945945945946,
+      "grad_norm": 0.4541209936141968,
+      "learning_rate": 0.0005,
+      "loss": 0.996231198310852,
+      "step": 959
+    },
+    {
+      "epoch": 1.2972972972972974,
+      "grad_norm": 0.37232959270477295,
+      "learning_rate": 0.0005,
+      "loss": 0.9755107760429382,
+      "step": 960
+    },
+    {
+      "epoch": 1.2986486486486486,
+      "grad_norm": 0.38288724422454834,
+      "learning_rate": 0.0005,
+      "loss": 0.9631942510604858,
+      "step": 961
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.46887311339378357,
+      "learning_rate": 0.0005,
+      "loss": 1.011840581893921,
+      "step": 962
+    },
+    {
+      "epoch": 1.3013513513513513,
+      "grad_norm": 0.42255890369415283,
+      "learning_rate": 0.0005,
+      "loss": 0.9645510911941528,
+      "step": 963
+    },
+    {
+      "epoch": 1.3027027027027027,
+      "grad_norm": 0.4499732255935669,
+      "learning_rate": 0.0005,
+      "loss": 0.9763890504837036,
+      "step": 964
+    },
+    {
+      "epoch": 1.304054054054054,
+      "grad_norm": 0.4645133316516876,
+      "learning_rate": 0.0005,
+      "loss": 0.9889826774597168,
+      "step": 965
+    },
+    {
+      "epoch": 1.3054054054054054,
+      "grad_norm": 0.5662233829498291,
+      "learning_rate": 0.0005,
+      "loss": 0.9706462621688843,
+      "step": 966
+    },
+    {
+      "epoch": 1.3067567567567568,
+      "grad_norm": 0.4030493199825287,
+      "learning_rate": 0.0005,
+      "loss": 0.9838024377822876,
+      "step": 967
+    },
+    {
+      "epoch": 1.308108108108108,
+      "grad_norm": 0.4581279158592224,
+      "learning_rate": 0.0005,
+      "loss": 1.004866123199463,
+      "step": 968
+    },
+    {
+      "epoch": 1.3094594594594595,
+      "grad_norm": 0.5351725816726685,
+      "learning_rate": 0.0005,
+      "loss": 1.0074418783187866,
+      "step": 969
+    },
+    {
+      "epoch": 1.3108108108108107,
+      "grad_norm": 0.43033337593078613,
+      "learning_rate": 0.0005,
+      "loss": 0.9701049327850342,
+      "step": 970
+    },
+    {
+      "epoch": 1.3121621621621622,
+      "grad_norm": 0.43474316596984863,
+      "learning_rate": 0.0005,
+      "loss": 0.9867752194404602,
+      "step": 971
+    },
+    {
+      "epoch": 1.3135135135135134,
+      "grad_norm": 0.4598759412765503,
+      "learning_rate": 0.0005,
+      "loss": 0.9737513065338135,
+      "step": 972
+    },
+    {
+      "epoch": 1.3148648648648649,
+      "grad_norm": 0.3885399401187897,
+      "learning_rate": 0.0005,
+      "loss": 0.9631804823875427,
+      "step": 973
+    },
+    {
+      "epoch": 1.3162162162162163,
+      "grad_norm": 0.3821704387664795,
+      "learning_rate": 0.0005,
+      "loss": 0.9453619122505188,
+      "step": 974
+    },
+    {
+      "epoch": 1.3175675675675675,
+      "grad_norm": 0.4695705771446228,
+      "learning_rate": 0.0005,
+      "loss": 0.9943991899490356,
+      "step": 975
+    },
+    {
+      "epoch": 1.318918918918919,
+      "grad_norm": 0.3535776138305664,
+      "learning_rate": 0.0005,
+      "loss": 0.9506361484527588,
+      "step": 976
+    },
+    {
+      "epoch": 1.3202702702702702,
+      "grad_norm": 0.35998857021331787,
+      "learning_rate": 0.0005,
+      "loss": 0.9464927911758423,
+      "step": 977
+    },
+    {
+      "epoch": 1.3216216216216217,
+      "grad_norm": 0.48262956738471985,
+      "learning_rate": 0.0005,
+      "loss": 1.0024775266647339,
+      "step": 978
+    },
+    {
+      "epoch": 1.322972972972973,
+      "grad_norm": 0.46606209874153137,
+      "learning_rate": 0.0005,
+      "loss": 0.9593669772148132,
+      "step": 979
+    },
+    {
+      "epoch": 1.3243243243243243,
+      "grad_norm": 0.4292880892753601,
+      "learning_rate": 0.0005,
+      "loss": 1.0080875158309937,
+      "step": 980
+    },
+    {
+      "epoch": 1.3256756756756758,
+      "grad_norm": 0.39110124111175537,
+      "learning_rate": 0.0005,
+      "loss": 0.9667671918869019,
+      "step": 981
+    },
+    {
+      "epoch": 1.327027027027027,
+      "grad_norm": 0.4362320601940155,
+      "learning_rate": 0.0005,
+      "loss": 0.9895745515823364,
+      "step": 982
+    },
+    {
+      "epoch": 1.3283783783783782,
+      "grad_norm": 0.34816741943359375,
+      "learning_rate": 0.0005,
+      "loss": 0.9362595081329346,
+      "step": 983
+    },
+    {
+      "epoch": 1.3297297297297297,
+      "grad_norm": 0.4204411506652832,
+      "learning_rate": 0.0005,
+      "loss": 0.9596747159957886,
+      "step": 984
+    },
+    {
+      "epoch": 1.3310810810810811,
+      "grad_norm": 0.4056480824947357,
+      "learning_rate": 0.0005,
+      "loss": 0.9546859264373779,
+      "step": 985
+    },
+    {
+      "epoch": 1.3324324324324324,
+      "grad_norm": 0.39050814509391785,
+      "learning_rate": 0.0005,
+      "loss": 0.95786452293396,
+      "step": 986
+    },
+    {
+      "epoch": 1.3337837837837838,
+      "grad_norm": 0.3578621447086334,
+      "learning_rate": 0.0005,
+      "loss": 0.9716924428939819,
+      "step": 987
+    },
+    {
+      "epoch": 1.3351351351351353,
+      "grad_norm": 0.4584849178791046,
+      "learning_rate": 0.0005,
+      "loss": 1.0097193717956543,
+      "step": 988
+    },
+    {
+      "epoch": 1.3364864864864865,
+      "grad_norm": 0.37440621852874756,
+      "learning_rate": 0.0005,
+      "loss": 0.9407498240470886,
+      "step": 989
+    },
+    {
+      "epoch": 1.3378378378378377,
+      "grad_norm": 0.3984168469905853,
+      "learning_rate": 0.0005,
+      "loss": 0.9809123873710632,
+      "step": 990
+    },
+    {
+      "epoch": 1.3391891891891892,
+      "grad_norm": 0.4240108132362366,
+      "learning_rate": 0.0005,
+      "loss": 1.0012762546539307,
+      "step": 991
+    },
+    {
+      "epoch": 1.3405405405405406,
+      "grad_norm": 0.42318645119667053,
+      "learning_rate": 0.0005,
+      "loss": 0.973394513130188,
+      "step": 992
+    },
+    {
+      "epoch": 1.3418918918918918,
+      "grad_norm": 0.38429686427116394,
+      "learning_rate": 0.0005,
+      "loss": 0.9899868965148926,
+      "step": 993
+    },
+    {
+      "epoch": 1.3432432432432433,
+      "grad_norm": 0.39727410674095154,
+      "learning_rate": 0.0005,
+      "loss": 0.9914262294769287,
+      "step": 994
+    },
+    {
+      "epoch": 1.3445945945945945,
+      "grad_norm": 0.35498425364494324,
+      "learning_rate": 0.0005,
+      "loss": 0.9689297080039978,
+      "step": 995
+    },
+    {
+      "epoch": 1.345945945945946,
+      "grad_norm": 0.3783055543899536,
+      "learning_rate": 0.0005,
+      "loss": 0.9784989356994629,
+      "step": 996
+    },
+    {
+      "epoch": 1.3472972972972972,
+      "grad_norm": 0.44674813747406006,
+      "learning_rate": 0.0005,
+      "loss": 0.997064471244812,
+      "step": 997
+    },
+    {
+      "epoch": 1.3486486486486486,
+      "grad_norm": 0.4055602550506592,
+      "learning_rate": 0.0005,
+      "loss": 0.9845970869064331,
+      "step": 998
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.46533408761024475,
+      "learning_rate": 0.0005,
+      "loss": 0.9597142934799194,
+      "step": 999
+    },
+    {
+      "epoch": 1.3513513513513513,
+      "grad_norm": 0.4324585497379303,
+      "learning_rate": 0.0005,
+      "loss": 0.9859974384307861,
+      "step": 1000
+    },
+    {
+      "epoch": 1.3527027027027028,
+      "grad_norm": 0.3993183970451355,
+      "learning_rate": 0.0005,
+      "loss": 0.9444106221199036,
+      "step": 1001
+    },
+    {
+      "epoch": 1.354054054054054,
+      "grad_norm": 0.3790976107120514,
+      "learning_rate": 0.0005,
+      "loss": 0.960803747177124,
+      "step": 1002
+    },
+    {
+      "epoch": 1.3554054054054054,
+      "grad_norm": 0.4591469466686249,
+      "learning_rate": 0.0005,
+      "loss": 1.0034210681915283,
+      "step": 1003
+    },
+    {
+      "epoch": 1.3567567567567567,
+      "grad_norm": 0.35321488976478577,
+      "learning_rate": 0.0005,
+      "loss": 0.9552630186080933,
+      "step": 1004
+    },
+    {
+      "epoch": 1.3581081081081081,
+      "grad_norm": 0.46684369444847107,
+      "learning_rate": 0.0005,
+      "loss": 1.0116569995880127,
+      "step": 1005
+    },
+    {
+      "epoch": 1.3594594594594596,
+      "grad_norm": 0.42274075746536255,
+      "learning_rate": 0.0005,
+      "loss": 0.9782974720001221,
+      "step": 1006
+    },
+    {
+      "epoch": 1.3608108108108108,
+      "grad_norm": 0.4363124668598175,
+      "learning_rate": 0.0005,
+      "loss": 0.9614741802215576,
+      "step": 1007
+    },
+    {
+      "epoch": 1.3621621621621622,
+      "grad_norm": 0.3986418843269348,
+      "learning_rate": 0.0005,
+      "loss": 0.9615585207939148,
+      "step": 1008
+    },
+    {
+      "epoch": 1.3635135135135135,
+      "grad_norm": 0.42405083775520325,
+      "learning_rate": 0.0005,
+      "loss": 0.9859756827354431,
+      "step": 1009
+    },
+    {
+      "epoch": 1.364864864864865,
+      "grad_norm": 0.432868093252182,
+      "learning_rate": 0.0005,
+      "loss": 0.9904283881187439,
+      "step": 1010
+    },
+    {
+      "epoch": 1.3662162162162161,
+      "grad_norm": 0.35173189640045166,
+      "learning_rate": 0.0005,
+      "loss": 0.9630738496780396,
+      "step": 1011
+    },
+    {
+      "epoch": 1.3675675675675676,
+      "grad_norm": 0.383989155292511,
+      "learning_rate": 0.0005,
+      "loss": 0.9615758061408997,
+      "step": 1012
+    },
+    {
+      "epoch": 1.368918918918919,
+      "grad_norm": 0.3957752287387848,
+      "learning_rate": 0.0005,
+      "loss": 0.9648306965827942,
+      "step": 1013
+    },
+    {
+      "epoch": 1.3702702702702703,
+      "grad_norm": 0.41450852155685425,
+      "learning_rate": 0.0005,
+      "loss": 0.9763829112052917,
+      "step": 1014
+    },
+    {
+      "epoch": 1.3716216216216215,
+      "grad_norm": 0.3912656903266907,
+      "learning_rate": 0.0005,
+      "loss": 0.9662620425224304,
+      "step": 1015
+    },
+    {
+      "epoch": 1.372972972972973,
+      "grad_norm": 0.3875739574432373,
+      "learning_rate": 0.0005,
+      "loss": 0.9596538543701172,
+      "step": 1016
+    },
+    {
+      "epoch": 1.3743243243243244,
+      "grad_norm": 0.37504637241363525,
+      "learning_rate": 0.0005,
+      "loss": 0.9556881189346313,
+      "step": 1017
+    },
+    {
+      "epoch": 1.3756756756756756,
+      "grad_norm": 0.408834308385849,
+      "learning_rate": 0.0005,
+      "loss": 0.988312840461731,
+      "step": 1018
+    },
+    {
+      "epoch": 1.377027027027027,
+      "grad_norm": 0.376113623380661,
+      "learning_rate": 0.0005,
+      "loss": 0.9467796683311462,
+      "step": 1019
+    },
+    {
+      "epoch": 1.3783783783783785,
+      "grad_norm": 0.4163655936717987,
+      "learning_rate": 0.0005,
+      "loss": 0.9784432649612427,
+      "step": 1020
+    },
+    {
+      "epoch": 1.3797297297297297,
+      "grad_norm": 0.35616591572761536,
+      "learning_rate": 0.0005,
+      "loss": 0.9544124603271484,
+      "step": 1021
+    },
+    {
+      "epoch": 1.381081081081081,
+      "grad_norm": 0.3302839696407318,
+      "learning_rate": 0.0005,
+      "loss": 0.9766925573348999,
+      "step": 1022
+    },
+    {
+      "epoch": 1.3824324324324324,
+      "grad_norm": 0.4313965439796448,
+      "learning_rate": 0.0005,
+      "loss": 0.980260968208313,
+      "step": 1023
+    },
+    {
+      "epoch": 1.3837837837837839,
+      "grad_norm": 0.44246983528137207,
+      "learning_rate": 0.0005,
+      "loss": 1.0087214708328247,
+      "step": 1024
+    },
+    {
+      "epoch": 1.385135135135135,
+      "grad_norm": 0.43066537380218506,
+      "learning_rate": 0.0005,
+      "loss": 0.9940868616104126,
+      "step": 1025
+    },
+    {
+      "epoch": 1.3864864864864865,
+      "grad_norm": 0.45920509099960327,
+      "learning_rate": 0.0005,
+      "loss": 0.9820013046264648,
+      "step": 1026
+    },
+    {
+      "epoch": 1.3878378378378378,
+      "grad_norm": 0.3817983567714691,
+      "learning_rate": 0.0005,
+      "loss": 0.9762750267982483,
+      "step": 1027
+    },
+    {
+      "epoch": 1.3891891891891892,
+      "grad_norm": 0.39356833696365356,
+      "learning_rate": 0.0005,
+      "loss": 0.9445961713790894,
+      "step": 1028
+    },
+    {
+      "epoch": 1.3905405405405404,
+      "grad_norm": 0.4399636387825012,
+      "learning_rate": 0.0005,
+      "loss": 0.9797091484069824,
+      "step": 1029
+    },
+    {
+      "epoch": 1.3918918918918919,
+      "grad_norm": 0.40890172123908997,
+      "learning_rate": 0.0005,
+      "loss": 0.9671077132225037,
+      "step": 1030
+    },
+    {
+      "epoch": 1.3932432432432433,
+      "grad_norm": 0.4137714207172394,
+      "learning_rate": 0.0005,
+      "loss": 0.983716607093811,
+      "step": 1031
+    },
+    {
+      "epoch": 1.3945945945945946,
+      "grad_norm": 0.4261699914932251,
+      "learning_rate": 0.0005,
+      "loss": 0.977470874786377,
+      "step": 1032
+    },
+    {
+      "epoch": 1.395945945945946,
+      "grad_norm": 0.4437422454357147,
+      "learning_rate": 0.0005,
+      "loss": 0.9873587489128113,
+      "step": 1033
+    },
+    {
+      "epoch": 1.3972972972972972,
+      "grad_norm": 0.42654410004615784,
+      "learning_rate": 0.0005,
+      "loss": 0.9767049551010132,
+      "step": 1034
+    },
+    {
+      "epoch": 1.3986486486486487,
+      "grad_norm": 0.38874828815460205,
+      "learning_rate": 0.0005,
+      "loss": 0.9596085548400879,
+      "step": 1035
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.38053402304649353,
+      "learning_rate": 0.0005,
+      "loss": 0.9573982357978821,
+      "step": 1036
+    },
+    {
+      "epoch": 1.4013513513513514,
+      "grad_norm": 0.4289546608924866,
+      "learning_rate": 0.0005,
+      "loss": 0.9838269352912903,
+      "step": 1037
+    },
+    {
+      "epoch": 1.4027027027027028,
+      "grad_norm": 0.35731303691864014,
+      "learning_rate": 0.0005,
+      "loss": 0.9533226490020752,
+      "step": 1038
+    },
+    {
+      "epoch": 1.404054054054054,
+      "grad_norm": 0.3575209975242615,
+      "learning_rate": 0.0005,
+      "loss": 0.9698132872581482,
+      "step": 1039
+    },
+    {
+      "epoch": 1.4054054054054055,
+      "grad_norm": 0.3679664433002472,
+      "learning_rate": 0.0005,
+      "loss": 0.9616092443466187,
+      "step": 1040
+    },
+    {
+      "epoch": 1.4067567567567567,
+      "grad_norm": 0.39746198058128357,
+      "learning_rate": 0.0005,
+      "loss": 0.958643913269043,
+      "step": 1041
+    },
+    {
+      "epoch": 1.4081081081081082,
+      "grad_norm": 0.4270752966403961,
+      "learning_rate": 0.0005,
+      "loss": 0.9839266538619995,
+      "step": 1042
+    },
+    {
+      "epoch": 1.4094594594594594,
+      "grad_norm": 0.4682374894618988,
+      "learning_rate": 0.0005,
+      "loss": 0.9918210506439209,
+      "step": 1043
+    },
+    {
+      "epoch": 1.4108108108108108,
+      "grad_norm": 0.3882635533809662,
+      "learning_rate": 0.0005,
+      "loss": 0.9528246521949768,
+      "step": 1044
+    },
+    {
+      "epoch": 1.4121621621621623,
+      "grad_norm": 0.4094189405441284,
+      "learning_rate": 0.0005,
+      "loss": 0.9620130062103271,
+      "step": 1045
+    },
+    {
+      "epoch": 1.4135135135135135,
+      "grad_norm": 0.4247358441352844,
+      "learning_rate": 0.0005,
+      "loss": 0.965696394443512,
+      "step": 1046
+    },
+    {
+      "epoch": 1.4148648648648647,
+      "grad_norm": 0.5095513463020325,
+      "learning_rate": 0.0005,
+      "loss": 1.0103189945220947,
+      "step": 1047
+    },
+    {
+      "epoch": 1.4162162162162162,
+      "grad_norm": 0.3551409840583801,
+      "learning_rate": 0.0005,
+      "loss": 0.9478543996810913,
+      "step": 1048
+    },
+    {
+      "epoch": 1.4175675675675676,
+      "grad_norm": 0.3779126703739166,
+      "learning_rate": 0.0005,
+      "loss": 0.9760139584541321,
+      "step": 1049
+    },
+    {
+      "epoch": 1.4189189189189189,
+      "grad_norm": 0.3009454607963562,
+      "learning_rate": 0.0005,
+      "loss": 0.9341579675674438,
+      "step": 1050
+    },
+    {
+      "epoch": 1.4202702702702703,
+      "grad_norm": 0.3730173110961914,
+      "learning_rate": 0.0005,
+      "loss": 0.9517393708229065,
+      "step": 1051
+    },
+    {
+      "epoch": 1.4216216216216218,
+      "grad_norm": 0.3778931498527527,
+      "learning_rate": 0.0005,
+      "loss": 0.9633306264877319,
+      "step": 1052
+    },
+    {
+      "epoch": 1.422972972972973,
+      "grad_norm": 0.4728538990020752,
+      "learning_rate": 0.0005,
+      "loss": 0.9935925602912903,
+      "step": 1053
+    },
+    {
+      "epoch": 1.4243243243243242,
+      "grad_norm": 0.3563747704029083,
+      "learning_rate": 0.0005,
+      "loss": 0.9514280557632446,
+      "step": 1054
+    },
+    {
+      "epoch": 1.4256756756756757,
+      "grad_norm": 0.34488409757614136,
+      "learning_rate": 0.0005,
+      "loss": 0.9695899486541748,
+      "step": 1055
+    },
+    {
+      "epoch": 1.427027027027027,
+      "grad_norm": 0.36349353194236755,
+      "learning_rate": 0.0005,
+      "loss": 0.9630669951438904,
+      "step": 1056
+    },
+    {
+      "epoch": 1.4283783783783783,
+      "grad_norm": 0.4190460741519928,
+      "learning_rate": 0.0005,
+      "loss": 0.959124743938446,
+      "step": 1057
+    },
+    {
+      "epoch": 1.4297297297297298,
+      "grad_norm": 0.3972044587135315,
+      "learning_rate": 0.0005,
+      "loss": 0.9444021582603455,
+      "step": 1058
+    },
+    {
+      "epoch": 1.431081081081081,
+      "grad_norm": 0.35883891582489014,
+      "learning_rate": 0.0005,
+      "loss": 0.9487593770027161,
+      "step": 1059
+    },
+    {
+      "epoch": 1.4324324324324325,
+      "grad_norm": 0.3603910803794861,
+      "learning_rate": 0.0005,
+      "loss": 0.949885368347168,
+      "step": 1060
+    },
+    {
+      "epoch": 1.4337837837837837,
+      "grad_norm": 0.3623533248901367,
+      "learning_rate": 0.0005,
+      "loss": 0.9584531784057617,
+      "step": 1061
+    },
+    {
+      "epoch": 1.4351351351351351,
+      "grad_norm": 0.4085288941860199,
+      "learning_rate": 0.0005,
+      "loss": 1.0058828592300415,
+      "step": 1062
+    },
+    {
+      "epoch": 1.4364864864864866,
+      "grad_norm": 0.3539946377277374,
+      "learning_rate": 0.0005,
+      "loss": 0.9507929086685181,
+      "step": 1063
+    },
+    {
+      "epoch": 1.4378378378378378,
+      "grad_norm": 0.3604002594947815,
+      "learning_rate": 0.0005,
+      "loss": 0.9445990324020386,
+      "step": 1064
+    },
+    {
+      "epoch": 1.4391891891891893,
+      "grad_norm": 0.41712507605552673,
+      "learning_rate": 0.0005,
+      "loss": 0.9690744876861572,
+      "step": 1065
+    },
+    {
+      "epoch": 1.4405405405405405,
+      "grad_norm": 0.44544804096221924,
+      "learning_rate": 0.0005,
+      "loss": 0.9992185831069946,
+      "step": 1066
+    },
+    {
+      "epoch": 1.441891891891892,
+      "grad_norm": 0.3901347815990448,
+      "learning_rate": 0.0005,
+      "loss": 0.9638530611991882,
+      "step": 1067
+    },
+    {
+      "epoch": 1.4432432432432432,
+      "grad_norm": 0.37077733874320984,
+      "learning_rate": 0.0005,
+      "loss": 0.96112060546875,
+      "step": 1068
+    },
+    {
+      "epoch": 1.4445945945945946,
+      "grad_norm": 0.3466204106807709,
+      "learning_rate": 0.0005,
+      "loss": 0.9500885009765625,
+      "step": 1069
+    },
+    {
+      "epoch": 1.445945945945946,
+      "grad_norm": 0.39560431241989136,
+      "learning_rate": 0.0005,
+      "loss": 0.9483655691146851,
+      "step": 1070
+    },
+    {
+      "epoch": 1.4472972972972973,
+      "grad_norm": 0.48067599534988403,
+      "learning_rate": 0.0005,
+      "loss": 0.9826964139938354,
+      "step": 1071
+    },
+    {
+      "epoch": 1.4486486486486487,
+      "grad_norm": 0.4487379491329193,
+      "learning_rate": 0.0005,
+      "loss": 0.9551558494567871,
+      "step": 1072
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.3846507966518402,
+      "learning_rate": 0.0005,
+      "loss": 0.9642122983932495,
+      "step": 1073
+    },
+    {
+      "epoch": 1.4513513513513514,
+      "grad_norm": 0.41805896162986755,
+      "learning_rate": 0.0005,
+      "loss": 0.9856113195419312,
+      "step": 1074
+    },
+    {
+      "epoch": 1.4527027027027026,
+      "grad_norm": 0.39456889033317566,
+      "learning_rate": 0.0005,
+      "loss": 0.9842128753662109,
+      "step": 1075
+    },
+    {
+      "epoch": 1.454054054054054,
+      "grad_norm": 0.37122422456741333,
+      "learning_rate": 0.0005,
+      "loss": 0.9658066034317017,
+      "step": 1076
+    },
+    {
+      "epoch": 1.4554054054054055,
+      "grad_norm": 0.3735131323337555,
+      "learning_rate": 0.0005,
+      "loss": 0.9580011963844299,
+      "step": 1077
+    },
+    {
+      "epoch": 1.4567567567567568,
+      "grad_norm": 0.40505287051200867,
+      "learning_rate": 0.0005,
+      "loss": 0.9720847606658936,
+      "step": 1078
+    },
+    {
+      "epoch": 1.458108108108108,
+      "grad_norm": 0.37636807560920715,
+      "learning_rate": 0.0005,
+      "loss": 0.98288893699646,
+      "step": 1079
+    },
+    {
+      "epoch": 1.4594594594594594,
+      "grad_norm": 0.37467196583747864,
+      "learning_rate": 0.0005,
+      "loss": 0.9786258935928345,
+      "step": 1080
+    },
+    {
+      "epoch": 1.4608108108108109,
+      "grad_norm": 0.45389866828918457,
+      "learning_rate": 0.0005,
+      "loss": 0.9824802875518799,
+      "step": 1081
+    },
+    {
+      "epoch": 1.462162162162162,
+      "grad_norm": 0.39140114188194275,
+      "learning_rate": 0.0005,
+      "loss": 0.9311952590942383,
+      "step": 1082
+    },
+    {
+      "epoch": 1.4635135135135136,
+      "grad_norm": 0.33884966373443604,
+      "learning_rate": 0.0005,
+      "loss": 0.9388359785079956,
+      "step": 1083
+    },
+    {
+      "epoch": 1.464864864864865,
+      "grad_norm": 0.3367622196674347,
+      "learning_rate": 0.0005,
+      "loss": 0.9692588448524475,
+      "step": 1084
+    },
+    {
+      "epoch": 1.4662162162162162,
+      "grad_norm": 0.37058714032173157,
+      "learning_rate": 0.0005,
+      "loss": 0.9605975151062012,
+      "step": 1085
+    },
+    {
+      "epoch": 1.4675675675675675,
+      "grad_norm": 0.38006410002708435,
+      "learning_rate": 0.0005,
+      "loss": 0.9490060806274414,
+      "step": 1086
+    },
+    {
+      "epoch": 1.468918918918919,
+      "grad_norm": 0.3355022072792053,
+      "learning_rate": 0.0005,
+      "loss": 0.9409561157226562,
+      "step": 1087
+    },
+    {
+      "epoch": 1.4702702702702704,
+      "grad_norm": 0.30517950654029846,
+      "learning_rate": 0.0005,
+      "loss": 0.9305635690689087,
+      "step": 1088
+    },
+    {
+      "epoch": 1.4716216216216216,
+      "grad_norm": 0.37387213110923767,
+      "learning_rate": 0.0005,
+      "loss": 0.9584960341453552,
+      "step": 1089
+    },
+    {
+      "epoch": 1.472972972972973,
+      "grad_norm": 0.4188067615032196,
+      "learning_rate": 0.0005,
+      "loss": 0.978485643863678,
+      "step": 1090
+    },
+    {
+      "epoch": 1.4743243243243243,
+      "grad_norm": 0.461946040391922,
+      "learning_rate": 0.0005,
+      "loss": 0.9867110252380371,
+      "step": 1091
+    },
+    {
+      "epoch": 1.4756756756756757,
+      "grad_norm": 0.4881296455860138,
+      "learning_rate": 0.0005,
+      "loss": 0.9920364618301392,
+      "step": 1092
+    },
+    {
+      "epoch": 1.477027027027027,
+      "grad_norm": 0.3795720934867859,
+      "learning_rate": 0.0005,
+      "loss": 0.9500658512115479,
+      "step": 1093
+    },
+    {
+      "epoch": 1.4783783783783784,
+      "grad_norm": 0.4172173738479614,
+      "learning_rate": 0.0005,
+      "loss": 0.9833842515945435,
+      "step": 1094
+    },
+    {
+      "epoch": 1.4797297297297298,
+      "grad_norm": 0.42282652854919434,
+      "learning_rate": 0.0005,
+      "loss": 0.9697315692901611,
+      "step": 1095
+    },
+    {
+      "epoch": 1.481081081081081,
+      "grad_norm": 0.4294726252555847,
+      "learning_rate": 0.0005,
+      "loss": 0.9585219621658325,
+      "step": 1096
+    },
+    {
+      "epoch": 1.4824324324324325,
+      "grad_norm": 0.55002760887146,
+      "learning_rate": 0.0005,
+      "loss": 0.9929457902908325,
+      "step": 1097
+    },
+    {
+      "epoch": 1.4837837837837837,
+      "grad_norm": 0.44223859906196594,
+      "learning_rate": 0.0005,
+      "loss": 0.9795092344284058,
+      "step": 1098
+    },
+    {
+      "epoch": 1.4851351351351352,
+      "grad_norm": 0.4319678843021393,
+      "learning_rate": 0.0005,
+      "loss": 0.9487788677215576,
+      "step": 1099
+    },
+    {
+      "epoch": 1.4864864864864864,
+      "grad_norm": 0.43504250049591064,
+      "learning_rate": 0.0005,
+      "loss": 0.9461289644241333,
+      "step": 1100
+    },
+    {
+      "epoch": 1.4878378378378379,
+      "grad_norm": 0.36632558703422546,
+      "learning_rate": 0.0005,
+      "loss": 0.9295979142189026,
+      "step": 1101
+    },
+    {
+      "epoch": 1.4891891891891893,
+      "grad_norm": 0.4383622705936432,
+      "learning_rate": 0.0005,
+      "loss": 0.9894219636917114,
+      "step": 1102
+    },
+    {
+      "epoch": 1.4905405405405405,
+      "grad_norm": 0.36873993277549744,
+      "learning_rate": 0.0005,
+      "loss": 0.9403143525123596,
+      "step": 1103
+    },
+    {
+      "epoch": 1.491891891891892,
+      "grad_norm": 0.4351787567138672,
+      "learning_rate": 0.0005,
+      "loss": 0.983810305595398,
+      "step": 1104
+    },
+    {
+      "epoch": 1.4932432432432432,
+      "grad_norm": 0.3586796820163727,
+      "learning_rate": 0.0005,
+      "loss": 0.9363659620285034,
+      "step": 1105
+    },
+    {
+      "epoch": 1.4945945945945946,
+      "grad_norm": 0.5187349319458008,
+      "learning_rate": 0.0005,
+      "loss": 0.9657735824584961,
+      "step": 1106
+    },
+    {
+      "epoch": 1.4959459459459459,
+      "grad_norm": 0.35093122720718384,
+      "learning_rate": 0.0005,
+      "loss": 0.9541574716567993,
+      "step": 1107
+    },
+    {
+      "epoch": 1.4972972972972973,
+      "grad_norm": 0.4604719877243042,
+      "learning_rate": 0.0005,
+      "loss": 1.0138943195343018,
+      "step": 1108
+    },
+    {
+      "epoch": 1.4986486486486488,
+      "grad_norm": 0.4752703905105591,
+      "learning_rate": 0.0005,
+      "loss": 0.9676473140716553,
+      "step": 1109
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3966812193393707,
+      "learning_rate": 0.0005,
+      "loss": 0.941551923751831,
+      "step": 1110
+    },
+    {
+      "epoch": 1.5013513513513512,
+      "grad_norm": 0.38008174300193787,
+      "learning_rate": 0.0005,
+      "loss": 0.977961540222168,
+      "step": 1111
+    },
+    {
+      "epoch": 1.5027027027027027,
+      "grad_norm": 0.4349740445613861,
+      "learning_rate": 0.0005,
+      "loss": 0.950982928276062,
+      "step": 1112
+    },
+    {
+      "epoch": 1.5040540540540541,
+      "grad_norm": 0.38316860795021057,
+      "learning_rate": 0.0005,
+      "loss": 0.9531576633453369,
+      "step": 1113
+    },
+    {
+      "epoch": 1.5054054054054054,
+      "grad_norm": 0.45138925313949585,
+      "learning_rate": 0.0005,
+      "loss": 0.9819865226745605,
+      "step": 1114
+    },
+    {
+      "epoch": 1.5067567567567568,
+      "grad_norm": 0.37200289964675903,
+      "learning_rate": 0.0005,
+      "loss": 0.9359767436981201,
+      "step": 1115
+    },
+    {
+      "epoch": 1.5081081081081082,
+      "grad_norm": 0.4309891164302826,
+      "learning_rate": 0.0005,
+      "loss": 0.9660316705703735,
+      "step": 1116
+    },
+    {
+      "epoch": 1.5094594594594595,
+      "grad_norm": 0.39992231130599976,
+      "learning_rate": 0.0005,
+      "loss": 0.9640520215034485,
+      "step": 1117
+    },
+    {
+      "epoch": 1.5108108108108107,
+      "grad_norm": 0.3775927722454071,
+      "learning_rate": 0.0005,
+      "loss": 0.946336030960083,
+      "step": 1118
+    },
+    {
+      "epoch": 1.5121621621621621,
+      "grad_norm": 0.3919059634208679,
+      "learning_rate": 0.0005,
+      "loss": 0.9613589644432068,
+      "step": 1119
+    },
+    {
+      "epoch": 1.5135135135135136,
+      "grad_norm": 0.34711575508117676,
+      "learning_rate": 0.0005,
+      "loss": 0.9538503885269165,
+      "step": 1120
+    },
+    {
+      "epoch": 1.5148648648648648,
+      "grad_norm": 0.4510093927383423,
+      "learning_rate": 0.0005,
+      "loss": 0.982893705368042,
+      "step": 1121
+    },
+    {
+      "epoch": 1.516216216216216,
+      "grad_norm": 0.4375079274177551,
+      "learning_rate": 0.0005,
+      "loss": 0.9656348824501038,
+      "step": 1122
+    },
+    {
+      "epoch": 1.5175675675675677,
+      "grad_norm": 0.39934200048446655,
+      "learning_rate": 0.0005,
+      "loss": 0.9534391164779663,
+      "step": 1123
+    },
+    {
+      "epoch": 1.518918918918919,
+      "grad_norm": 0.39143383502960205,
+      "learning_rate": 0.0005,
+      "loss": 0.9301098585128784,
+      "step": 1124
+    },
+    {
+      "epoch": 1.5202702702702702,
+      "grad_norm": 0.41351252794265747,
+      "learning_rate": 0.0005,
+      "loss": 0.965934157371521,
+      "step": 1125
+    },
+    {
+      "epoch": 1.5216216216216216,
+      "grad_norm": 0.4403027892112732,
+      "learning_rate": 0.0005,
+      "loss": 0.9657508134841919,
+      "step": 1126
+    },
+    {
+      "epoch": 1.522972972972973,
+      "grad_norm": 0.3678840696811676,
+      "learning_rate": 0.0005,
+      "loss": 0.9431184530258179,
+      "step": 1127
+    },
+    {
+      "epoch": 1.5243243243243243,
+      "grad_norm": 0.4261213541030884,
+      "learning_rate": 0.0005,
+      "loss": 0.9288582801818848,
+      "step": 1128
+    },
+    {
+      "epoch": 1.5256756756756755,
+      "grad_norm": 0.4752713441848755,
+      "learning_rate": 0.0005,
+      "loss": 0.9587739706039429,
+      "step": 1129
+    },
+    {
+      "epoch": 1.527027027027027,
+      "grad_norm": 0.36653268337249756,
+      "learning_rate": 0.0005,
+      "loss": 0.9640433192253113,
+      "step": 1130
+    },
+    {
+      "epoch": 1.5283783783783784,
+      "grad_norm": 0.3999096751213074,
+      "learning_rate": 0.0005,
+      "loss": 0.9762459397315979,
+      "step": 1131
+    },
+    {
+      "epoch": 1.5297297297297296,
+      "grad_norm": 0.49856898188591003,
+      "learning_rate": 0.0005,
+      "loss": 0.9680496454238892,
+      "step": 1132
+    },
+    {
+      "epoch": 1.531081081081081,
+      "grad_norm": 0.38013532757759094,
+      "learning_rate": 0.0005,
+      "loss": 0.9568884372711182,
+      "step": 1133
+    },
+    {
+      "epoch": 1.5324324324324325,
+      "grad_norm": 0.41031593084335327,
+      "learning_rate": 0.0005,
+      "loss": 0.9492183327674866,
+      "step": 1134
+    },
+    {
+      "epoch": 1.5337837837837838,
+      "grad_norm": 0.41555631160736084,
+      "learning_rate": 0.0005,
+      "loss": 0.980380654335022,
+      "step": 1135
+    },
+    {
+      "epoch": 1.535135135135135,
+      "grad_norm": 0.34103474020957947,
+      "learning_rate": 0.0005,
+      "loss": 0.9659240245819092,
+      "step": 1136
+    },
+    {
+      "epoch": 1.5364864864864864,
+      "grad_norm": 0.4161185622215271,
+      "learning_rate": 0.0005,
+      "loss": 0.9560171961784363,
+      "step": 1137
+    },
+    {
+      "epoch": 1.537837837837838,
+      "grad_norm": 0.41111308336257935,
+      "learning_rate": 0.0005,
+      "loss": 0.9513988494873047,
+      "step": 1138
+    },
+    {
+      "epoch": 1.5391891891891891,
+      "grad_norm": 0.488657683134079,
+      "learning_rate": 0.0005,
+      "loss": 0.9720638394355774,
+      "step": 1139
+    },
+    {
+      "epoch": 1.5405405405405406,
+      "grad_norm": 0.42847687005996704,
+      "learning_rate": 0.0005,
+      "loss": 0.9507311582565308,
+      "step": 1140
+    },
+    {
+      "epoch": 1.541891891891892,
+      "grad_norm": 0.38240742683410645,
+      "learning_rate": 0.0005,
+      "loss": 0.9944350123405457,
+      "step": 1141
+    },
+    {
+      "epoch": 1.5432432432432432,
+      "grad_norm": 0.4380272924900055,
+      "learning_rate": 0.0005,
+      "loss": 0.933574378490448,
+      "step": 1142
+    },
+    {
+      "epoch": 1.5445945945945945,
+      "grad_norm": 0.4834255278110504,
+      "learning_rate": 0.0005,
+      "loss": 0.9555559158325195,
+      "step": 1143
+    },
+    {
+      "epoch": 1.545945945945946,
+      "grad_norm": 0.4544822871685028,
+      "learning_rate": 0.0005,
+      "loss": 0.9745993614196777,
+      "step": 1144
+    },
+    {
+      "epoch": 1.5472972972972974,
+      "grad_norm": 0.4300711452960968,
+      "learning_rate": 0.0005,
+      "loss": 0.9709230661392212,
+      "step": 1145
+    },
+    {
+      "epoch": 1.5486486486486486,
+      "grad_norm": 0.3389417827129364,
+      "learning_rate": 0.0005,
+      "loss": 0.9415884017944336,
+      "step": 1146
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.4029770791530609,
+      "learning_rate": 0.0005,
+      "loss": 0.965229868888855,
+      "step": 1147
+    },
+    {
+      "epoch": 1.5513513513513515,
+      "grad_norm": 0.41696083545684814,
+      "learning_rate": 0.0005,
+      "loss": 0.9584909081459045,
+      "step": 1148
+    },
+    {
+      "epoch": 1.5527027027027027,
+      "grad_norm": 0.37948331236839294,
+      "learning_rate": 0.0005,
+      "loss": 0.939961850643158,
+      "step": 1149
+    },
+    {
+      "epoch": 1.554054054054054,
+      "grad_norm": 0.47973009943962097,
+      "learning_rate": 0.0005,
+      "loss": 0.975281834602356,
+      "step": 1150
+    },
+    {
+      "epoch": 1.5554054054054054,
+      "grad_norm": 0.3136143982410431,
+      "learning_rate": 0.0005,
+      "loss": 0.929936408996582,
+      "step": 1151
+    },
+    {
+      "epoch": 1.5567567567567568,
+      "grad_norm": 0.3751409947872162,
+      "learning_rate": 0.0005,
+      "loss": 0.9261614084243774,
+      "step": 1152
+    },
+    {
+      "epoch": 1.558108108108108,
+      "grad_norm": 0.41457903385162354,
+      "learning_rate": 0.0005,
+      "loss": 0.9502162933349609,
+      "step": 1153
+    },
+    {
+      "epoch": 1.5594594594594593,
+      "grad_norm": 0.39995279908180237,
+      "learning_rate": 0.0005,
+      "loss": 0.9548132419586182,
+      "step": 1154
+    },
+    {
+      "epoch": 1.560810810810811,
+      "grad_norm": 0.37232062220573425,
+      "learning_rate": 0.0005,
+      "loss": 0.9596434831619263,
+      "step": 1155
+    },
+    {
+      "epoch": 1.5621621621621622,
+      "grad_norm": 0.37979596853256226,
+      "learning_rate": 0.0005,
+      "loss": 0.9468961358070374,
+      "step": 1156
+    },
+    {
+      "epoch": 1.5635135135135134,
+      "grad_norm": 0.3939518332481384,
+      "learning_rate": 0.0005,
+      "loss": 0.9677606225013733,
+      "step": 1157
+    },
+    {
+      "epoch": 1.5648648648648649,
+      "grad_norm": 0.4171069860458374,
+      "learning_rate": 0.0005,
+      "loss": 0.964557945728302,
+      "step": 1158
+    },
+    {
+      "epoch": 1.5662162162162163,
+      "grad_norm": 0.41495436429977417,
+      "learning_rate": 0.0005,
+      "loss": 0.9766783714294434,
+      "step": 1159
+    },
+    {
+      "epoch": 1.5675675675675675,
+      "grad_norm": 0.4327319264411926,
+      "learning_rate": 0.0005,
+      "loss": 0.9697217345237732,
+      "step": 1160
+    },
+    {
+      "epoch": 1.5689189189189188,
+      "grad_norm": 0.5252425670623779,
+      "learning_rate": 0.0005,
+      "loss": 0.9752391576766968,
+      "step": 1161
+    },
+    {
+      "epoch": 1.5702702702702702,
+      "grad_norm": 0.4131445288658142,
+      "learning_rate": 0.0005,
+      "loss": 0.9639368057250977,
+      "step": 1162
+    },
+    {
+      "epoch": 1.5716216216216217,
+      "grad_norm": 0.3978724181652069,
+      "learning_rate": 0.0005,
+      "loss": 0.9491093158721924,
+      "step": 1163
+    },
+    {
+      "epoch": 1.572972972972973,
+      "grad_norm": 0.40715426206588745,
+      "learning_rate": 0.0005,
+      "loss": 0.9473484754562378,
+      "step": 1164
+    },
+    {
+      "epoch": 1.5743243243243243,
+      "grad_norm": 0.4214545786380768,
+      "learning_rate": 0.0005,
+      "loss": 0.9348351955413818,
+      "step": 1165
+    },
+    {
+      "epoch": 1.5756756756756758,
+      "grad_norm": 0.3865233063697815,
+      "learning_rate": 0.0005,
+      "loss": 0.964876651763916,
+      "step": 1166
+    },
+    {
+      "epoch": 1.577027027027027,
+      "grad_norm": 0.39758339524269104,
+      "learning_rate": 0.0005,
+      "loss": 0.9673240184783936,
+      "step": 1167
+    },
+    {
+      "epoch": 1.5783783783783782,
+      "grad_norm": 0.5669912099838257,
+      "learning_rate": 0.0005,
+      "loss": 0.9576663374900818,
+      "step": 1168
+    },
+    {
+      "epoch": 1.5797297297297297,
+      "grad_norm": 0.47453853487968445,
+      "learning_rate": 0.0005,
+      "loss": 0.9692167043685913,
+      "step": 1169
+    },
+    {
+      "epoch": 1.5810810810810811,
+      "grad_norm": 0.4097541570663452,
+      "learning_rate": 0.0005,
+      "loss": 0.9721696376800537,
+      "step": 1170
+    },
+    {
+      "epoch": 1.5824324324324324,
+      "grad_norm": 0.4696161150932312,
+      "learning_rate": 0.0005,
+      "loss": 0.9757513403892517,
+      "step": 1171
+    },
+    {
+      "epoch": 1.5837837837837838,
+      "grad_norm": 0.4041876196861267,
+      "learning_rate": 0.0005,
+      "loss": 0.9461511373519897,
+      "step": 1172
+    },
+    {
+      "epoch": 1.5851351351351353,
+      "grad_norm": 0.3806861639022827,
+      "learning_rate": 0.0005,
+      "loss": 0.9471021890640259,
+      "step": 1173
+    },
+    {
+      "epoch": 1.5864864864864865,
+      "grad_norm": 0.31367865204811096,
+      "learning_rate": 0.0005,
+      "loss": 0.9388632774353027,
+      "step": 1174
+    },
+    {
+      "epoch": 1.5878378378378377,
+      "grad_norm": 0.3768991529941559,
+      "learning_rate": 0.0005,
+      "loss": 0.9513887763023376,
+      "step": 1175
+    },
+    {
+      "epoch": 1.5891891891891892,
+      "grad_norm": 0.3847515285015106,
+      "learning_rate": 0.0005,
+      "loss": 0.9322637915611267,
+      "step": 1176
+    },
+    {
+      "epoch": 1.5905405405405406,
+      "grad_norm": 0.38120102882385254,
+      "learning_rate": 0.0005,
+      "loss": 0.9445083141326904,
+      "step": 1177
+    },
+    {
+      "epoch": 1.5918918918918918,
+      "grad_norm": 0.3941683769226074,
+      "learning_rate": 0.0005,
+      "loss": 0.9767742156982422,
+      "step": 1178
+    },
+    {
+      "epoch": 1.5932432432432433,
+      "grad_norm": 0.36340633034706116,
+      "learning_rate": 0.0005,
+      "loss": 0.958519458770752,
+      "step": 1179
+    },
+    {
+      "epoch": 1.5945945945945947,
+      "grad_norm": 0.40474262833595276,
+      "learning_rate": 0.0005,
+      "loss": 0.9651819467544556,
+      "step": 1180
+    },
+    {
+      "epoch": 1.595945945945946,
+      "grad_norm": 0.4016048014163971,
+      "learning_rate": 0.0005,
+      "loss": 0.9725466370582581,
+      "step": 1181
+    },
+    {
+      "epoch": 1.5972972972972972,
+      "grad_norm": 0.38242003321647644,
+      "learning_rate": 0.0005,
+      "loss": 0.9725283980369568,
+      "step": 1182
+    },
+    {
+      "epoch": 1.5986486486486486,
+      "grad_norm": 0.3724381923675537,
+      "learning_rate": 0.0005,
+      "loss": 0.9591721892356873,
+      "step": 1183
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.4221299886703491,
+      "learning_rate": 0.0005,
+      "loss": 0.9785435795783997,
+      "step": 1184
+    },
+    {
+      "epoch": 1.6013513513513513,
+      "grad_norm": 0.4546847939491272,
+      "learning_rate": 0.0005,
+      "loss": 0.948093831539154,
+      "step": 1185
+    },
+    {
+      "epoch": 1.6027027027027025,
+      "grad_norm": 0.45676088333129883,
+      "learning_rate": 0.0005,
+      "loss": 0.9727705717086792,
+      "step": 1186
+    },
+    {
+      "epoch": 1.6040540540540542,
+      "grad_norm": 0.4651713967323303,
+      "learning_rate": 0.0005,
+      "loss": 0.9824694395065308,
+      "step": 1187
+    },
+    {
+      "epoch": 1.6054054054054054,
+      "grad_norm": 0.39438700675964355,
+      "learning_rate": 0.0005,
+      "loss": 0.9537489414215088,
+      "step": 1188
+    },
+    {
+      "epoch": 1.6067567567567567,
+      "grad_norm": 0.4643537402153015,
+      "learning_rate": 0.0005,
+      "loss": 0.9734922647476196,
+      "step": 1189
+    },
+    {
+      "epoch": 1.6081081081081081,
+      "grad_norm": 0.40369081497192383,
+      "learning_rate": 0.0005,
+      "loss": 0.971176028251648,
+      "step": 1190
+    },
+    {
+      "epoch": 1.6094594594594596,
+      "grad_norm": 0.34440624713897705,
+      "learning_rate": 0.0005,
+      "loss": 0.9252766370773315,
+      "step": 1191
+    },
+    {
+      "epoch": 1.6108108108108108,
+      "grad_norm": 0.3914441466331482,
+      "learning_rate": 0.0005,
+      "loss": 0.9372780323028564,
+      "step": 1192
+    },
+    {
+      "epoch": 1.612162162162162,
+      "grad_norm": 0.3409269452095032,
+      "learning_rate": 0.0005,
+      "loss": 0.9285401105880737,
+      "step": 1193
+    },
+    {
+      "epoch": 1.6135135135135135,
+      "grad_norm": 0.40754765272140503,
+      "learning_rate": 0.0005,
+      "loss": 0.9640967845916748,
+      "step": 1194
+    },
+    {
+      "epoch": 1.614864864864865,
+      "grad_norm": 0.3607959747314453,
+      "learning_rate": 0.0005,
+      "loss": 0.9201816320419312,
+      "step": 1195
+    },
+    {
+      "epoch": 1.6162162162162161,
+      "grad_norm": 0.37694624066352844,
+      "learning_rate": 0.0005,
+      "loss": 0.9603697061538696,
+      "step": 1196
+    },
+    {
+      "epoch": 1.6175675675675676,
+      "grad_norm": 0.3907509744167328,
+      "learning_rate": 0.0005,
+      "loss": 0.9513974785804749,
+      "step": 1197
+    },
+    {
+      "epoch": 1.618918918918919,
+      "grad_norm": 0.3641899526119232,
+      "learning_rate": 0.0005,
+      "loss": 0.9499997496604919,
+      "step": 1198
+    },
+    {
+      "epoch": 1.6202702702702703,
+      "grad_norm": 0.45657679438591003,
+      "learning_rate": 0.0005,
+      "loss": 1.007421612739563,
+      "step": 1199
+    },
+    {
+      "epoch": 1.6216216216216215,
+      "grad_norm": 0.4201827943325043,
+      "learning_rate": 0.0005,
+      "loss": 0.9701443910598755,
+      "step": 1200
+    },
+    {
+      "epoch": 1.622972972972973,
+      "grad_norm": 0.4502195119857788,
+      "learning_rate": 0.0005,
+      "loss": 0.9742072224617004,
+      "step": 1201
+    },
+    {
+      "epoch": 1.6243243243243244,
+      "grad_norm": 0.34931516647338867,
+      "learning_rate": 0.0005,
+      "loss": 0.9445006847381592,
+      "step": 1202
+    },
+    {
+      "epoch": 1.6256756756756756,
+      "grad_norm": 0.3743537962436676,
+      "learning_rate": 0.0005,
+      "loss": 0.9711391925811768,
+      "step": 1203
+    },
+    {
+      "epoch": 1.627027027027027,
+      "grad_norm": 0.3788037896156311,
+      "learning_rate": 0.0005,
+      "loss": 0.9668350219726562,
+      "step": 1204
+    },
+    {
+      "epoch": 1.6283783783783785,
+      "grad_norm": 0.3578028082847595,
+      "learning_rate": 0.0005,
+      "loss": 0.9445785880088806,
+      "step": 1205
+    },
+    {
+      "epoch": 1.6297297297297297,
+      "grad_norm": 0.42827922105789185,
+      "learning_rate": 0.0005,
+      "loss": 0.9668441414833069,
+      "step": 1206
+    },
+    {
+      "epoch": 1.631081081081081,
+      "grad_norm": 0.3780052959918976,
+      "learning_rate": 0.0005,
+      "loss": 0.953309178352356,
+      "step": 1207
+    },
+    {
+      "epoch": 1.6324324324324324,
+      "grad_norm": 0.3911561667919159,
+      "learning_rate": 0.0005,
+      "loss": 0.9693876504898071,
+      "step": 1208
+    },
+    {
+      "epoch": 1.6337837837837839,
+      "grad_norm": 0.4281550347805023,
+      "learning_rate": 0.0005,
+      "loss": 0.9392358660697937,
+      "step": 1209
+    },
+    {
+      "epoch": 1.635135135135135,
+      "grad_norm": 0.36240312457084656,
+      "learning_rate": 0.0005,
+      "loss": 0.9469308853149414,
+      "step": 1210
+    },
+    {
+      "epoch": 1.6364864864864865,
+      "grad_norm": 0.38603439927101135,
+      "learning_rate": 0.0005,
+      "loss": 0.9637563228607178,
+      "step": 1211
+    },
+    {
+      "epoch": 1.637837837837838,
+      "grad_norm": 0.48646900057792664,
+      "learning_rate": 0.0005,
+      "loss": 0.9786785840988159,
+      "step": 1212
+    },
+    {
+      "epoch": 1.6391891891891892,
+      "grad_norm": 0.36049801111221313,
+      "learning_rate": 0.0005,
+      "loss": 0.980427086353302,
+      "step": 1213
+    },
+    {
+      "epoch": 1.6405405405405404,
+      "grad_norm": 0.45904287695884705,
+      "learning_rate": 0.0005,
+      "loss": 0.9667958617210388,
+      "step": 1214
+    },
+    {
+      "epoch": 1.6418918918918919,
+      "grad_norm": 0.4146110415458679,
+      "learning_rate": 0.0005,
+      "loss": 0.9552158117294312,
+      "step": 1215
+    },
+    {
+      "epoch": 1.6432432432432433,
+      "grad_norm": 0.35458043217658997,
+      "learning_rate": 0.0005,
+      "loss": 0.9569350481033325,
+      "step": 1216
+    },
+    {
+      "epoch": 1.6445945945945946,
+      "grad_norm": 0.36103397607803345,
+      "learning_rate": 0.0005,
+      "loss": 0.9625058174133301,
+      "step": 1217
+    },
+    {
+      "epoch": 1.6459459459459458,
+      "grad_norm": 0.3746268153190613,
+      "learning_rate": 0.0005,
+      "loss": 0.958650529384613,
+      "step": 1218
+    },
+    {
+      "epoch": 1.6472972972972975,
+      "grad_norm": 0.46736523509025574,
+      "learning_rate": 0.0005,
+      "loss": 0.9778891801834106,
+      "step": 1219
+    },
+    {
+      "epoch": 1.6486486486486487,
+      "grad_norm": 0.41505011916160583,
+      "learning_rate": 0.0005,
+      "loss": 0.9795930981636047,
+      "step": 1220
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.385784387588501,
+      "learning_rate": 0.0005,
+      "loss": 0.9613853693008423,
+      "step": 1221
+    },
+    {
+      "epoch": 1.6513513513513514,
+      "grad_norm": 0.36682605743408203,
+      "learning_rate": 0.0005,
+      "loss": 0.9281511306762695,
+      "step": 1222
+    },
+    {
+      "epoch": 1.6527027027027028,
+      "grad_norm": 0.404877632856369,
+      "learning_rate": 0.0005,
+      "loss": 0.9210977554321289,
+      "step": 1223
+    },
+    {
+      "epoch": 1.654054054054054,
+      "grad_norm": 0.4140811264514923,
+      "learning_rate": 0.0005,
+      "loss": 0.9336795806884766,
+      "step": 1224
+    },
+    {
+      "epoch": 1.6554054054054053,
+      "grad_norm": 0.39744633436203003,
+      "learning_rate": 0.0005,
+      "loss": 0.9723071455955505,
+      "step": 1225
+    },
+    {
+      "epoch": 1.6567567567567567,
+      "grad_norm": 0.48660629987716675,
+      "learning_rate": 0.0005,
+      "loss": 0.9576288461685181,
+      "step": 1226
+    },
+    {
+      "epoch": 1.6581081081081082,
+      "grad_norm": 0.3451976776123047,
+      "learning_rate": 0.0005,
+      "loss": 0.9174661636352539,
+      "step": 1227
+    },
+    {
+      "epoch": 1.6594594594594594,
+      "grad_norm": 0.4076247811317444,
+      "learning_rate": 0.0005,
+      "loss": 0.9387319684028625,
+      "step": 1228
+    },
+    {
+      "epoch": 1.6608108108108108,
+      "grad_norm": 0.38520604372024536,
+      "learning_rate": 0.0005,
+      "loss": 0.9399251937866211,
+      "step": 1229
+    },
+    {
+      "epoch": 1.6621621621621623,
+      "grad_norm": 0.3771904706954956,
+      "learning_rate": 0.0005,
+      "loss": 0.9440655708312988,
+      "step": 1230
+    },
+    {
+      "epoch": 1.6635135135135135,
+      "grad_norm": 0.5821990966796875,
+      "learning_rate": 0.0005,
+      "loss": 0.9983311295509338,
+      "step": 1231
+    },
+    {
+      "epoch": 1.6648648648648647,
+      "grad_norm": 0.4073484539985657,
+      "learning_rate": 0.0005,
+      "loss": 0.9344413876533508,
+      "step": 1232
+    },
+    {
+      "epoch": 1.6662162162162162,
+      "grad_norm": 0.35189175605773926,
+      "learning_rate": 0.0005,
+      "loss": 0.9279830455780029,
+      "step": 1233
+    },
+    {
+      "epoch": 1.6675675675675676,
+      "grad_norm": 0.43726661801338196,
+      "learning_rate": 0.0005,
+      "loss": 0.9855589270591736,
+      "step": 1234
+    },
+    {
+      "epoch": 1.6689189189189189,
+      "grad_norm": 0.37590906023979187,
+      "learning_rate": 0.0005,
+      "loss": 0.9481592774391174,
+      "step": 1235
+    },
+    {
+      "epoch": 1.6702702702702703,
+      "grad_norm": 0.3665043115615845,
+      "learning_rate": 0.0005,
+      "loss": 0.9500851631164551,
+      "step": 1236
+    },
+    {
+      "epoch": 1.6716216216216218,
+      "grad_norm": 0.46543580293655396,
+      "learning_rate": 0.0005,
+      "loss": 0.9742764234542847,
+      "step": 1237
+    },
+    {
+      "epoch": 1.672972972972973,
+      "grad_norm": 0.3802144229412079,
+      "learning_rate": 0.0005,
+      "loss": 0.951168417930603,
+      "step": 1238
+    },
+    {
+      "epoch": 1.6743243243243242,
+      "grad_norm": 0.43063002824783325,
+      "learning_rate": 0.0005,
+      "loss": 0.9634685516357422,
+      "step": 1239
+    },
+    {
+      "epoch": 1.6756756756756757,
+      "grad_norm": 0.36052894592285156,
+      "learning_rate": 0.0005,
+      "loss": 0.9410466551780701,
+      "step": 1240
+    },
+    {
+      "epoch": 1.677027027027027,
+      "grad_norm": 0.3777155578136444,
+      "learning_rate": 0.0005,
+      "loss": 0.9269376397132874,
+      "step": 1241
+    },
+    {
+      "epoch": 1.6783783783783783,
+      "grad_norm": 0.40749186277389526,
+      "learning_rate": 0.0005,
+      "loss": 0.9703361988067627,
+      "step": 1242
+    },
+    {
+      "epoch": 1.6797297297297298,
+      "grad_norm": 0.4770202040672302,
+      "learning_rate": 0.0005,
+      "loss": 0.9860799312591553,
+      "step": 1243
+    },
+    {
+      "epoch": 1.6810810810810812,
+      "grad_norm": 0.4007508456707001,
+      "learning_rate": 0.0005,
+      "loss": 0.9422894716262817,
+      "step": 1244
+    },
+    {
+      "epoch": 1.6824324324324325,
+      "grad_norm": 0.41895344853401184,
+      "learning_rate": 0.0005,
+      "loss": 0.9578995704650879,
+      "step": 1245
+    },
+    {
+      "epoch": 1.6837837837837837,
+      "grad_norm": 0.35357925295829773,
+      "learning_rate": 0.0005,
+      "loss": 0.9448098540306091,
+      "step": 1246
+    },
+    {
+      "epoch": 1.6851351351351351,
+      "grad_norm": 0.33891761302948,
+      "learning_rate": 0.0005,
+      "loss": 0.948429524898529,
+      "step": 1247
+    },
+    {
+      "epoch": 1.6864864864864866,
+      "grad_norm": 0.45763805508613586,
+      "learning_rate": 0.0005,
+      "loss": 0.9669528603553772,
+      "step": 1248
+    },
+    {
+      "epoch": 1.6878378378378378,
+      "grad_norm": 0.3814607262611389,
+      "learning_rate": 0.0005,
+      "loss": 0.9681079387664795,
+      "step": 1249
+    },
+    {
+      "epoch": 1.689189189189189,
+      "grad_norm": 0.31308087706565857,
+      "learning_rate": 0.0005,
+      "loss": 0.9187357425689697,
+      "step": 1250
+    },
+    {
+      "epoch": 1.6905405405405407,
+      "grad_norm": 0.37015965580940247,
+      "learning_rate": 0.0005,
+      "loss": 0.9551773071289062,
+      "step": 1251
+    },
+    {
+      "epoch": 1.691891891891892,
+      "grad_norm": 0.40885522961616516,
+      "learning_rate": 0.0005,
+      "loss": 0.9582571983337402,
+      "step": 1252
+    },
+    {
+      "epoch": 1.6932432432432432,
+      "grad_norm": 0.34575751423835754,
+      "learning_rate": 0.0005,
+      "loss": 0.9393103718757629,
+      "step": 1253
+    },
+    {
+      "epoch": 1.6945945945945946,
+      "grad_norm": 0.4375525414943695,
+      "learning_rate": 0.0005,
+      "loss": 0.9790081977844238,
+      "step": 1254
+    },
+    {
+      "epoch": 1.695945945945946,
+      "grad_norm": 0.38791894912719727,
+      "learning_rate": 0.0005,
+      "loss": 0.9412522912025452,
+      "step": 1255
+    },
+    {
+      "epoch": 1.6972972972972973,
+      "grad_norm": 0.4078836143016815,
+      "learning_rate": 0.0005,
+      "loss": 0.9804825782775879,
+      "step": 1256
+    },
+    {
+      "epoch": 1.6986486486486485,
+      "grad_norm": 0.36182916164398193,
+      "learning_rate": 0.0005,
+      "loss": 0.9552974700927734,
+      "step": 1257
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.43866604566574097,
+      "learning_rate": 0.0005,
+      "loss": 0.9400486946105957,
+      "step": 1258
+    },
+    {
+      "epoch": 1.7013513513513514,
+      "grad_norm": 0.48193955421447754,
+      "learning_rate": 0.0005,
+      "loss": 0.9671605825424194,
+      "step": 1259
+    },
+    {
+      "epoch": 1.7027027027027026,
+      "grad_norm": 0.3479158878326416,
+      "learning_rate": 0.0005,
+      "loss": 0.9568958282470703,
+      "step": 1260
+    },
+    {
+      "epoch": 1.704054054054054,
+      "grad_norm": 0.34355294704437256,
+      "learning_rate": 0.0005,
+      "loss": 0.925106406211853,
+      "step": 1261
+    },
+    {
+      "epoch": 1.7054054054054055,
+      "grad_norm": 0.3802018463611603,
+      "learning_rate": 0.0005,
+      "loss": 0.9471943378448486,
+      "step": 1262
+    },
+    {
+      "epoch": 1.7067567567567568,
+      "grad_norm": 0.3532378375530243,
+      "learning_rate": 0.0005,
+      "loss": 0.9386012554168701,
+      "step": 1263
+    },
+    {
+      "epoch": 1.708108108108108,
+      "grad_norm": 0.35207465291023254,
+      "learning_rate": 0.0005,
+      "loss": 0.9406929016113281,
+      "step": 1264
+    },
+    {
+      "epoch": 1.7094594594594594,
+      "grad_norm": 0.3474140465259552,
+      "learning_rate": 0.0005,
+      "loss": 0.9419785141944885,
+      "step": 1265
+    },
+    {
+      "epoch": 1.7108108108108109,
+      "grad_norm": 0.3086281716823578,
+      "learning_rate": 0.0005,
+      "loss": 0.9216618537902832,
+      "step": 1266
+    },
+    {
+      "epoch": 1.712162162162162,
+      "grad_norm": 0.3527332842350006,
+      "learning_rate": 0.0005,
+      "loss": 0.9282899498939514,
+      "step": 1267
+    },
+    {
+      "epoch": 1.7135135135135136,
+      "grad_norm": 0.361875981092453,
+      "learning_rate": 0.0005,
+      "loss": 0.9454764127731323,
+      "step": 1268
+    },
+    {
+      "epoch": 1.714864864864865,
+      "grad_norm": 0.37262043356895447,
+      "learning_rate": 0.0005,
+      "loss": 0.9359268546104431,
+      "step": 1269
+    },
+    {
+      "epoch": 1.7162162162162162,
+      "grad_norm": 0.35754793882369995,
+      "learning_rate": 0.0005,
+      "loss": 0.9423198699951172,
+      "step": 1270
+    },
+    {
+      "epoch": 1.7175675675675675,
+      "grad_norm": 0.3368144929409027,
+      "learning_rate": 0.0005,
+      "loss": 0.9250342845916748,
+      "step": 1271
+    },
+    {
+      "epoch": 1.718918918918919,
+      "grad_norm": 0.33008527755737305,
+      "learning_rate": 0.0005,
+      "loss": 0.9536451101303101,
+      "step": 1272
+    },
+    {
+      "epoch": 1.7202702702702704,
+      "grad_norm": 0.37251296639442444,
+      "learning_rate": 0.0005,
+      "loss": 0.9420700073242188,
+      "step": 1273
+    },
+    {
+      "epoch": 1.7216216216216216,
+      "grad_norm": 0.43446239829063416,
+      "learning_rate": 0.0005,
+      "loss": 0.9801032543182373,
+      "step": 1274
+    },
+    {
+      "epoch": 1.722972972972973,
+      "grad_norm": 0.29864367842674255,
+      "learning_rate": 0.0005,
+      "loss": 0.913515567779541,
+      "step": 1275
+    },
+    {
+      "epoch": 1.7243243243243245,
+      "grad_norm": 0.330944687128067,
+      "learning_rate": 0.0005,
+      "loss": 0.9295227527618408,
+      "step": 1276
+    },
+    {
+      "epoch": 1.7256756756756757,
+      "grad_norm": 0.34798383712768555,
+      "learning_rate": 0.0005,
+      "loss": 0.9426395297050476,
+      "step": 1277
+    },
+    {
+      "epoch": 1.727027027027027,
+      "grad_norm": 0.3548194169998169,
+      "learning_rate": 0.0005,
+      "loss": 0.9407570958137512,
+      "step": 1278
+    },
+    {
+      "epoch": 1.7283783783783784,
+      "grad_norm": 0.3421894609928131,
+      "learning_rate": 0.0005,
+      "loss": 0.9514966607093811,
+      "step": 1279
+    },
+    {
+      "epoch": 1.7297297297297298,
+      "grad_norm": 0.4020922780036926,
+      "learning_rate": 0.0005,
+      "loss": 0.9394033551216125,
+      "step": 1280
+    },
+    {
+      "epoch": 1.731081081081081,
+      "grad_norm": 0.39720290899276733,
+      "learning_rate": 0.0005,
+      "loss": 0.9514843225479126,
+      "step": 1281
+    },
+    {
+      "epoch": 1.7324324324324323,
+      "grad_norm": 0.3352365493774414,
+      "learning_rate": 0.0005,
+      "loss": 0.949224054813385,
+      "step": 1282
+    },
+    {
+      "epoch": 1.733783783783784,
+      "grad_norm": 0.3273203372955322,
+      "learning_rate": 0.0005,
+      "loss": 0.9294737577438354,
+      "step": 1283
+    },
+    {
+      "epoch": 1.7351351351351352,
+      "grad_norm": 0.37867429852485657,
+      "learning_rate": 0.0005,
+      "loss": 0.9657316207885742,
+      "step": 1284
+    },
+    {
+      "epoch": 1.7364864864864864,
+      "grad_norm": 0.36902305483818054,
+      "learning_rate": 0.0005,
+      "loss": 0.9425586462020874,
+      "step": 1285
+    },
+    {
+      "epoch": 1.7378378378378379,
+      "grad_norm": 0.33285003900527954,
+      "learning_rate": 0.0005,
+      "loss": 0.9313395023345947,
+      "step": 1286
+    },
+    {
+      "epoch": 1.7391891891891893,
+      "grad_norm": 0.30613237619400024,
+      "learning_rate": 0.0005,
+      "loss": 0.9284007549285889,
+      "step": 1287
+    },
+    {
+      "epoch": 1.7405405405405405,
+      "grad_norm": 0.3469081521034241,
+      "learning_rate": 0.0005,
+      "loss": 0.9538443684577942,
+      "step": 1288
+    },
+    {
+      "epoch": 1.7418918918918918,
+      "grad_norm": 0.33172714710235596,
+      "learning_rate": 0.0005,
+      "loss": 0.9168393611907959,
+      "step": 1289
+    },
+    {
+      "epoch": 1.7432432432432432,
+      "grad_norm": 0.41409584879875183,
+      "learning_rate": 0.0005,
+      "loss": 0.9589287042617798,
+      "step": 1290
+    },
+    {
+      "epoch": 1.7445945945945946,
+      "grad_norm": 0.37759485840797424,
+      "learning_rate": 0.0005,
+      "loss": 0.9516587257385254,
+      "step": 1291
+    },
+    {
+      "epoch": 1.7459459459459459,
+      "grad_norm": 0.34625741839408875,
+      "learning_rate": 0.0005,
+      "loss": 0.9683810472488403,
+      "step": 1292
+    },
+    {
+      "epoch": 1.7472972972972973,
+      "grad_norm": 0.4217616021633148,
+      "learning_rate": 0.0005,
+      "loss": 0.9660884141921997,
+      "step": 1293
+    },
+    {
+      "epoch": 1.7486486486486488,
+      "grad_norm": 0.3414537012577057,
+      "learning_rate": 0.0005,
+      "loss": 0.9375584721565247,
+      "step": 1294
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33917030692100525,
+      "learning_rate": 0.0005,
+      "loss": 0.9571197032928467,
+      "step": 1295
+    },
+    {
+      "epoch": 1.7513513513513512,
+      "grad_norm": 0.3505309522151947,
+      "learning_rate": 0.0005,
+      "loss": 0.9542548060417175,
+      "step": 1296
+    },
+    {
+      "epoch": 1.7527027027027027,
+      "grad_norm": 0.39167532324790955,
+      "learning_rate": 0.0005,
+      "loss": 0.9758152365684509,
+      "step": 1297
+    },
+    {
+      "epoch": 1.7540540540540541,
+      "grad_norm": 0.42795953154563904,
+      "learning_rate": 0.0005,
+      "loss": 0.9473451375961304,
+      "step": 1298
+    },
+    {
+      "epoch": 1.7554054054054054,
+      "grad_norm": 0.37586045265197754,
+      "learning_rate": 0.0005,
+      "loss": 0.9735001921653748,
+      "step": 1299
+    },
+    {
+      "epoch": 1.7567567567567568,
+      "grad_norm": 0.3182435929775238,
+      "learning_rate": 0.0005,
+      "loss": 0.9236515760421753,
+      "step": 1300
+    },
+    {
+      "epoch": 1.7581081081081082,
+      "grad_norm": 0.29968342185020447,
+      "learning_rate": 0.0005,
+      "loss": 0.9464074373245239,
+      "step": 1301
+    },
+    {
+      "epoch": 1.7594594594594595,
+      "grad_norm": 0.3941636085510254,
+      "learning_rate": 0.0005,
+      "loss": 0.9422377347946167,
+      "step": 1302
+    },
+    {
+      "epoch": 1.7608108108108107,
+      "grad_norm": 0.3471173048019409,
+      "learning_rate": 0.0005,
+      "loss": 0.9548400640487671,
+      "step": 1303
+    },
+    {
+      "epoch": 1.7621621621621621,
+      "grad_norm": 0.38001760840415955,
+      "learning_rate": 0.0005,
+      "loss": 0.9351468682289124,
+      "step": 1304
+    },
+    {
+      "epoch": 1.7635135135135136,
+      "grad_norm": 0.3752715587615967,
+      "learning_rate": 0.0005,
+      "loss": 0.9495323300361633,
+      "step": 1305
+    },
+    {
+      "epoch": 1.7648648648648648,
+      "grad_norm": 0.3597239553928375,
+      "learning_rate": 0.0005,
+      "loss": 0.934257447719574,
+      "step": 1306
+    },
+    {
+      "epoch": 1.766216216216216,
+      "grad_norm": 0.31978368759155273,
+      "learning_rate": 0.0005,
+      "loss": 0.9354345798492432,
+      "step": 1307
+    },
+    {
+      "epoch": 1.7675675675675677,
+      "grad_norm": 0.4084911644458771,
+      "learning_rate": 0.0005,
+      "loss": 0.952942967414856,
+      "step": 1308
+    },
+    {
+      "epoch": 1.768918918918919,
+      "grad_norm": 0.40581151843070984,
+      "learning_rate": 0.0005,
+      "loss": 0.9481770992279053,
+      "step": 1309
+    },
+    {
+      "epoch": 1.7702702702702702,
+      "grad_norm": 0.40713587403297424,
+      "learning_rate": 0.0005,
+      "loss": 0.9898827075958252,
+      "step": 1310
+    },
+    {
+      "epoch": 1.7716216216216216,
+      "grad_norm": 0.4113260805606842,
+      "learning_rate": 0.0005,
+      "loss": 0.9642786979675293,
+      "step": 1311
+    },
+    {
+      "epoch": 1.772972972972973,
+      "grad_norm": 0.3911372721195221,
+      "learning_rate": 0.0005,
+      "loss": 0.9563602209091187,
+      "step": 1312
+    },
+    {
+      "epoch": 1.7743243243243243,
+      "grad_norm": 0.37108027935028076,
+      "learning_rate": 0.0005,
+      "loss": 0.9664313793182373,
+      "step": 1313
+    },
+    {
+      "epoch": 1.7756756756756755,
+      "grad_norm": 0.32468128204345703,
+      "learning_rate": 0.0005,
+      "loss": 0.9200162887573242,
+      "step": 1314
+    },
+    {
+      "epoch": 1.777027027027027,
+      "grad_norm": 0.27554982900619507,
+      "learning_rate": 0.0005,
+      "loss": 0.9165645837783813,
+      "step": 1315
+    },
+    {
+      "epoch": 1.7783783783783784,
+      "grad_norm": 0.3835579454898834,
+      "learning_rate": 0.0005,
+      "loss": 0.9459726214408875,
+      "step": 1316
+    },
+    {
+      "epoch": 1.7797297297297296,
+      "grad_norm": 0.4653474688529968,
+      "learning_rate": 0.0005,
+      "loss": 0.9705330729484558,
+      "step": 1317
+    },
+    {
+      "epoch": 1.781081081081081,
+      "grad_norm": 0.36051836609840393,
+      "learning_rate": 0.0005,
+      "loss": 0.9477840065956116,
+      "step": 1318
+    },
+    {
+      "epoch": 1.7824324324324325,
+      "grad_norm": 0.3824165463447571,
+      "learning_rate": 0.0005,
+      "loss": 0.9394538402557373,
+      "step": 1319
+    },
+    {
+      "epoch": 1.7837837837837838,
+      "grad_norm": 0.36151307821273804,
+      "learning_rate": 0.0005,
+      "loss": 0.9293473958969116,
+      "step": 1320
+    },
+    {
+      "epoch": 1.785135135135135,
+      "grad_norm": 0.36345183849334717,
+      "learning_rate": 0.0005,
+      "loss": 0.9451315402984619,
+      "step": 1321
+    },
+    {
+      "epoch": 1.7864864864864864,
+      "grad_norm": 0.4185807704925537,
+      "learning_rate": 0.0005,
+      "loss": 0.9430806040763855,
+      "step": 1322
+    },
+    {
+      "epoch": 1.787837837837838,
+      "grad_norm": 0.3188069462776184,
+      "learning_rate": 0.0005,
+      "loss": 0.9344263672828674,
+      "step": 1323
+    },
+    {
+      "epoch": 1.7891891891891891,
+      "grad_norm": 0.36382561922073364,
+      "learning_rate": 0.0005,
+      "loss": 0.9628078937530518,
+      "step": 1324
+    },
+    {
+      "epoch": 1.7905405405405406,
+      "grad_norm": 0.3531954884529114,
+      "learning_rate": 0.0005,
+      "loss": 0.9440940022468567,
+      "step": 1325
+    },
+    {
+      "epoch": 1.791891891891892,
+      "grad_norm": 0.44106167554855347,
+      "learning_rate": 0.0005,
+      "loss": 0.9672216176986694,
+      "step": 1326
+    },
+    {
+      "epoch": 1.7932432432432432,
+      "grad_norm": 0.35423558950424194,
+      "learning_rate": 0.0005,
+      "loss": 0.9940869808197021,
+      "step": 1327
+    },
+    {
+      "epoch": 1.7945945945945945,
+      "grad_norm": 0.34942126274108887,
+      "learning_rate": 0.0005,
+      "loss": 0.9418994188308716,
+      "step": 1328
+    },
+    {
+      "epoch": 1.795945945945946,
+      "grad_norm": 0.37636587023735046,
+      "learning_rate": 0.0005,
+      "loss": 0.9247602820396423,
+      "step": 1329
+    },
+    {
+      "epoch": 1.7972972972972974,
+      "grad_norm": 0.43075239658355713,
+      "learning_rate": 0.0005,
+      "loss": 0.9883710145950317,
+      "step": 1330
+    },
+    {
+      "epoch": 1.7986486486486486,
+      "grad_norm": 0.379227876663208,
+      "learning_rate": 0.0005,
+      "loss": 0.95688796043396,
+      "step": 1331
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.36325812339782715,
+      "learning_rate": 0.0005,
+      "loss": 0.9530050754547119,
+      "step": 1332
+    },
+    {
+      "epoch": 1.8013513513513515,
+      "grad_norm": 0.363756000995636,
+      "learning_rate": 0.0005,
+      "loss": 0.9650669097900391,
+      "step": 1333
+    },
+    {
+      "epoch": 1.8027027027027027,
+      "grad_norm": 0.3621388077735901,
+      "learning_rate": 0.0005,
+      "loss": 0.9331346750259399,
+      "step": 1334
+    },
+    {
+      "epoch": 1.804054054054054,
+      "grad_norm": 0.36614158749580383,
+      "learning_rate": 0.0005,
+      "loss": 0.9272724390029907,
+      "step": 1335
+    },
+    {
+      "epoch": 1.8054054054054054,
+      "grad_norm": 0.3582662343978882,
+      "learning_rate": 0.0005,
+      "loss": 0.9390696883201599,
+      "step": 1336
+    },
+    {
+      "epoch": 1.8067567567567568,
+      "grad_norm": 0.38801127672195435,
+      "learning_rate": 0.0005,
+      "loss": 0.9571510553359985,
+      "step": 1337
+    },
+    {
+      "epoch": 1.808108108108108,
+      "grad_norm": 0.39441925287246704,
+      "learning_rate": 0.0005,
+      "loss": 0.9477448463439941,
+      "step": 1338
+    },
+    {
+      "epoch": 1.8094594594594593,
+      "grad_norm": 0.3712886571884155,
+      "learning_rate": 0.0005,
+      "loss": 0.958232045173645,
+      "step": 1339
+    },
+    {
+      "epoch": 1.810810810810811,
+      "grad_norm": 0.44147565960884094,
+      "learning_rate": 0.0005,
+      "loss": 0.9515318870544434,
+      "step": 1340
+    },
+    {
+      "epoch": 1.8121621621621622,
+      "grad_norm": 0.4741097688674927,
+      "learning_rate": 0.0005,
+      "loss": 0.9860102534294128,
+      "step": 1341
+    },
+    {
+      "epoch": 1.8135135135135134,
+      "grad_norm": 0.341325044631958,
+      "learning_rate": 0.0005,
+      "loss": 0.9162441492080688,
+      "step": 1342
+    },
+    {
+      "epoch": 1.8148648648648649,
+      "grad_norm": 0.3461504280567169,
+      "learning_rate": 0.0005,
+      "loss": 0.9373440742492676,
+      "step": 1343
+    },
+    {
+      "epoch": 1.8162162162162163,
+      "grad_norm": 0.423786461353302,
+      "learning_rate": 0.0005,
+      "loss": 0.9608849883079529,
+      "step": 1344
+    },
+    {
+      "epoch": 1.8175675675675675,
+      "grad_norm": 0.41154277324676514,
+      "learning_rate": 0.0005,
+      "loss": 0.9329888820648193,
+      "step": 1345
+    },
+    {
+      "epoch": 1.8189189189189188,
+      "grad_norm": 0.4746367335319519,
+      "learning_rate": 0.0005,
+      "loss": 0.9342315196990967,
+      "step": 1346
+    },
+    {
+      "epoch": 1.8202702702702702,
+      "grad_norm": 0.3427903652191162,
+      "learning_rate": 0.0005,
+      "loss": 0.9200377464294434,
+      "step": 1347
+    },
+    {
+      "epoch": 1.8216216216216217,
+      "grad_norm": 0.4284798502922058,
+      "learning_rate": 0.0005,
+      "loss": 0.9653714895248413,
+      "step": 1348
+    },
+    {
+      "epoch": 1.822972972972973,
+      "grad_norm": 0.30078208446502686,
+      "learning_rate": 0.0005,
+      "loss": 0.9194636344909668,
+      "step": 1349
+    },
+    {
+      "epoch": 1.8243243243243243,
+      "grad_norm": 0.47070783376693726,
+      "learning_rate": 0.0005,
+      "loss": 0.9507732391357422,
+      "step": 1350
+    },
+    {
+      "epoch": 1.8256756756756758,
+      "grad_norm": 0.38122686743736267,
+      "learning_rate": 0.0005,
+      "loss": 0.9322240352630615,
+      "step": 1351
+    },
+    {
+      "epoch": 1.827027027027027,
+      "grad_norm": 0.32157304883003235,
+      "learning_rate": 0.0005,
+      "loss": 0.9259010553359985,
+      "step": 1352
+    },
+    {
+      "epoch": 1.8283783783783782,
+      "grad_norm": 0.37224552035331726,
+      "learning_rate": 0.0005,
+      "loss": 0.9574956893920898,
+      "step": 1353
+    },
+    {
+      "epoch": 1.8297297297297297,
+      "grad_norm": 0.38265272974967957,
+      "learning_rate": 0.0005,
+      "loss": 0.9440553188323975,
+      "step": 1354
+    },
+    {
+      "epoch": 1.8310810810810811,
+      "grad_norm": 0.44474589824676514,
+      "learning_rate": 0.0005,
+      "loss": 0.9608544111251831,
+      "step": 1355
+    },
+    {
+      "epoch": 1.8324324324324324,
+      "grad_norm": 0.3621498942375183,
+      "learning_rate": 0.0005,
+      "loss": 0.9605438709259033,
+      "step": 1356
+    },
+    {
+      "epoch": 1.8337837837837838,
+      "grad_norm": 0.38977953791618347,
+      "learning_rate": 0.0005,
+      "loss": 0.9342190027236938,
+      "step": 1357
+    },
+    {
+      "epoch": 1.8351351351351353,
+      "grad_norm": 0.3946993947029114,
+      "learning_rate": 0.0005,
+      "loss": 0.9407868981361389,
+      "step": 1358
+    },
+    {
+      "epoch": 1.8364864864864865,
+      "grad_norm": 0.3613739609718323,
+      "learning_rate": 0.0005,
+      "loss": 0.9307279586791992,
+      "step": 1359
+    },
+    {
+      "epoch": 1.8378378378378377,
+      "grad_norm": 0.3524796664714813,
+      "learning_rate": 0.0005,
+      "loss": 0.9496296644210815,
+      "step": 1360
+    },
+    {
+      "epoch": 1.8391891891891892,
+      "grad_norm": 0.4176379442214966,
+      "learning_rate": 0.0005,
+      "loss": 0.9596730470657349,
+      "step": 1361
+    },
+    {
+      "epoch": 1.8405405405405406,
+      "grad_norm": 0.4638357162475586,
+      "learning_rate": 0.0005,
+      "loss": 0.9571579098701477,
+      "step": 1362
+    },
+    {
+      "epoch": 1.8418918918918918,
+      "grad_norm": 0.3618636727333069,
+      "learning_rate": 0.0005,
+      "loss": 0.9582734704017639,
+      "step": 1363
+    },
+    {
+      "epoch": 1.8432432432432433,
+      "grad_norm": 0.37664595246315,
+      "learning_rate": 0.0005,
+      "loss": 0.9518369436264038,
+      "step": 1364
+    },
+    {
+      "epoch": 1.8445945945945947,
+      "grad_norm": 0.3229392170906067,
+      "learning_rate": 0.0005,
+      "loss": 0.9269097447395325,
+      "step": 1365
+    },
+    {
+      "epoch": 1.845945945945946,
+      "grad_norm": 0.44305238127708435,
+      "learning_rate": 0.0005,
+      "loss": 0.9620966911315918,
+      "step": 1366
+    },
+    {
+      "epoch": 1.8472972972972972,
+      "grad_norm": 0.4314405620098114,
+      "learning_rate": 0.0005,
+      "loss": 0.9577270746231079,
+      "step": 1367
+    },
+    {
+      "epoch": 1.8486486486486486,
+      "grad_norm": 0.3493940532207489,
+      "learning_rate": 0.0005,
+      "loss": 0.9176316261291504,
+      "step": 1368
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3546578586101532,
+      "learning_rate": 0.0005,
+      "loss": 0.9193241596221924,
+      "step": 1369
+    },
+    {
+      "epoch": 1.8513513513513513,
+      "grad_norm": 0.2961341142654419,
+      "learning_rate": 0.0005,
+      "loss": 0.9329721927642822,
+      "step": 1370
+    },
+    {
+      "epoch": 1.8527027027027025,
+      "grad_norm": 0.38422322273254395,
+      "learning_rate": 0.0005,
+      "loss": 0.9528566598892212,
+      "step": 1371
+    },
+    {
+      "epoch": 1.8540540540540542,
+      "grad_norm": 0.4025252163410187,
+      "learning_rate": 0.0005,
+      "loss": 0.972783088684082,
+      "step": 1372
+    },
+    {
+      "epoch": 1.8554054054054054,
+      "grad_norm": 0.34658491611480713,
+      "learning_rate": 0.0005,
+      "loss": 0.9474144577980042,
+      "step": 1373
+    },
+    {
+      "epoch": 1.8567567567567567,
+      "grad_norm": 0.36818045377731323,
+      "learning_rate": 0.0005,
+      "loss": 0.9204481840133667,
+      "step": 1374
+    },
+    {
+      "epoch": 1.8581081081081081,
+      "grad_norm": 0.39111366868019104,
+      "learning_rate": 0.0005,
+      "loss": 0.9724769592285156,
+      "step": 1375
+    },
+    {
+      "epoch": 1.8594594594594596,
+      "grad_norm": 0.3950488567352295,
+      "learning_rate": 0.0005,
+      "loss": 0.9361264705657959,
+      "step": 1376
+    },
+    {
+      "epoch": 1.8608108108108108,
+      "grad_norm": 0.3541398048400879,
+      "learning_rate": 0.0005,
+      "loss": 0.938693642616272,
+      "step": 1377
+    },
+    {
+      "epoch": 1.862162162162162,
+      "grad_norm": 0.3569153845310211,
+      "learning_rate": 0.0005,
+      "loss": 0.957511305809021,
+      "step": 1378
+    },
+    {
+      "epoch": 1.8635135135135135,
+      "grad_norm": 0.3842785954475403,
+      "learning_rate": 0.0005,
+      "loss": 0.9395632743835449,
+      "step": 1379
+    },
+    {
+      "epoch": 1.864864864864865,
+      "grad_norm": 0.3487560451030731,
+      "learning_rate": 0.0005,
+      "loss": 0.9583709836006165,
+      "step": 1380
+    },
+    {
+      "epoch": 1.8662162162162161,
+      "grad_norm": 0.3955117166042328,
+      "learning_rate": 0.0005,
+      "loss": 0.9496450424194336,
+      "step": 1381
+    },
+    {
+      "epoch": 1.8675675675675676,
+      "grad_norm": 0.4035010039806366,
+      "learning_rate": 0.0005,
+      "loss": 0.9666591882705688,
+      "step": 1382
+    },
+    {
+      "epoch": 1.868918918918919,
+      "grad_norm": 0.4327566921710968,
+      "learning_rate": 0.0005,
+      "loss": 0.9512765407562256,
+      "step": 1383
+    },
+    {
+      "epoch": 1.8702702702702703,
+      "grad_norm": 0.35988229513168335,
+      "learning_rate": 0.0005,
+      "loss": 0.9413249492645264,
+      "step": 1384
+    },
+    {
+      "epoch": 1.8716216216216215,
+      "grad_norm": 0.4086672365665436,
+      "learning_rate": 0.0005,
+      "loss": 0.9603999853134155,
+      "step": 1385
+    },
+    {
+      "epoch": 1.872972972972973,
+      "grad_norm": 0.3710937798023224,
+      "learning_rate": 0.0005,
+      "loss": 0.9230751991271973,
+      "step": 1386
+    },
+    {
+      "epoch": 1.8743243243243244,
+      "grad_norm": 0.3920607268810272,
+      "learning_rate": 0.0005,
+      "loss": 0.9450048804283142,
+      "step": 1387
+    },
+    {
+      "epoch": 1.8756756756756756,
+      "grad_norm": 0.3824506402015686,
+      "learning_rate": 0.0005,
+      "loss": 0.9396034479141235,
+      "step": 1388
+    },
+    {
+      "epoch": 1.877027027027027,
+      "grad_norm": 0.3349030911922455,
+      "learning_rate": 0.0005,
+      "loss": 0.9392379522323608,
+      "step": 1389
+    },
+    {
+      "epoch": 1.8783783783783785,
+      "grad_norm": 0.42908236384391785,
+      "learning_rate": 0.0005,
+      "loss": 0.9738844633102417,
+      "step": 1390
+    },
+    {
+      "epoch": 1.8797297297297297,
+      "grad_norm": 0.35405510663986206,
+      "learning_rate": 0.0005,
+      "loss": 0.9493197202682495,
+      "step": 1391
+    },
+    {
+      "epoch": 1.881081081081081,
+      "grad_norm": 0.3650103509426117,
+      "learning_rate": 0.0005,
+      "loss": 0.9414617419242859,
+      "step": 1392
+    },
+    {
+      "epoch": 1.8824324324324324,
+      "grad_norm": 0.3536660373210907,
+      "learning_rate": 0.0005,
+      "loss": 0.9284870028495789,
+      "step": 1393
+    },
+    {
+      "epoch": 1.8837837837837839,
+      "grad_norm": 0.2977929413318634,
+      "learning_rate": 0.0005,
+      "loss": 0.9188510179519653,
+      "step": 1394
+    },
+    {
+      "epoch": 1.885135135135135,
+      "grad_norm": 0.26982608437538147,
+      "learning_rate": 0.0005,
+      "loss": 0.9086926579475403,
+      "step": 1395
+    },
+    {
+      "epoch": 1.8864864864864865,
+      "grad_norm": 0.35776862502098083,
+      "learning_rate": 0.0005,
+      "loss": 0.9337420463562012,
+      "step": 1396
+    },
+    {
+      "epoch": 1.887837837837838,
+      "grad_norm": 0.2743357717990875,
+      "learning_rate": 0.0005,
+      "loss": 0.9350926280021667,
+      "step": 1397
+    },
+    {
+      "epoch": 1.8891891891891892,
+      "grad_norm": 0.3329916298389435,
+      "learning_rate": 0.0005,
+      "loss": 0.9291895627975464,
+      "step": 1398
+    },
+    {
+      "epoch": 1.8905405405405404,
+      "grad_norm": 0.39804255962371826,
+      "learning_rate": 0.0005,
+      "loss": 0.9650388956069946,
+      "step": 1399
+    },
+    {
+      "epoch": 1.8918918918918919,
+      "grad_norm": 0.3707389235496521,
+      "learning_rate": 0.0005,
+      "loss": 0.9499279260635376,
+      "step": 1400
+    },
+    {
+      "epoch": 1.8932432432432433,
+      "grad_norm": 0.35786834359169006,
+      "learning_rate": 0.0005,
+      "loss": 0.9388126730918884,
+      "step": 1401
+    },
+    {
+      "epoch": 1.8945945945945946,
+      "grad_norm": 0.4045484662055969,
+      "learning_rate": 0.0005,
+      "loss": 0.9439940452575684,
+      "step": 1402
+    },
+    {
+      "epoch": 1.8959459459459458,
+      "grad_norm": 0.4282008409500122,
+      "learning_rate": 0.0005,
+      "loss": 0.9736782312393188,
+      "step": 1403
+    },
+    {
+      "epoch": 1.8972972972972975,
+      "grad_norm": 0.38950687646865845,
+      "learning_rate": 0.0005,
+      "loss": 0.9427982568740845,
+      "step": 1404
+    },
+    {
+      "epoch": 1.8986486486486487,
+      "grad_norm": 0.3895266652107239,
+      "learning_rate": 0.0005,
+      "loss": 0.9468898177146912,
+      "step": 1405
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.380068302154541,
+      "learning_rate": 0.0005,
+      "loss": 0.9712737202644348,
+      "step": 1406
+    },
+    {
+      "epoch": 1.9013513513513514,
+      "grad_norm": 0.43507757782936096,
+      "learning_rate": 0.0005,
+      "loss": 0.9601116180419922,
+      "step": 1407
+    },
+    {
+      "epoch": 1.9027027027027028,
+      "grad_norm": 0.3973771929740906,
+      "learning_rate": 0.0005,
+      "loss": 0.9428017139434814,
+      "step": 1408
+    },
+    {
+      "epoch": 1.904054054054054,
+      "grad_norm": 0.44450002908706665,
+      "learning_rate": 0.0005,
+      "loss": 0.9515812397003174,
+      "step": 1409
+    },
+    {
+      "epoch": 1.9054054054054053,
+      "grad_norm": 0.3636583387851715,
+      "learning_rate": 0.0005,
+      "loss": 0.951367974281311,
+      "step": 1410
+    },
+    {
+      "epoch": 1.9067567567567567,
+      "grad_norm": 0.354386568069458,
+      "learning_rate": 0.0005,
+      "loss": 0.9449846148490906,
+      "step": 1411
+    },
+    {
+      "epoch": 1.9081081081081082,
+      "grad_norm": 0.38668006658554077,
+      "learning_rate": 0.0005,
+      "loss": 0.943756103515625,
+      "step": 1412
+    },
+    {
+      "epoch": 1.9094594594594594,
+      "grad_norm": 0.39904263615608215,
+      "learning_rate": 0.0005,
+      "loss": 0.9459818005561829,
+      "step": 1413
+    },
+    {
+      "epoch": 1.9108108108108108,
+      "grad_norm": 0.4588395357131958,
+      "learning_rate": 0.0005,
+      "loss": 0.9618288278579712,
+      "step": 1414
+    },
+    {
+      "epoch": 1.9121621621621623,
+      "grad_norm": 0.39821887016296387,
+      "learning_rate": 0.0005,
+      "loss": 0.937157154083252,
+      "step": 1415
+    },
+    {
+      "epoch": 1.9135135135135135,
+      "grad_norm": 0.3764078617095947,
+      "learning_rate": 0.0005,
+      "loss": 0.9353232383728027,
+      "step": 1416
+    },
+    {
+      "epoch": 1.9148648648648647,
+      "grad_norm": 0.43572765588760376,
+      "learning_rate": 0.0005,
+      "loss": 0.9670611619949341,
+      "step": 1417
+    },
+    {
+      "epoch": 1.9162162162162162,
+      "grad_norm": 0.366218626499176,
+      "learning_rate": 0.0005,
+      "loss": 0.9467562437057495,
+      "step": 1418
+    },
+    {
+      "epoch": 1.9175675675675676,
+      "grad_norm": 0.38250303268432617,
+      "learning_rate": 0.0005,
+      "loss": 0.9249759316444397,
+      "step": 1419
+    },
+    {
+      "epoch": 1.9189189189189189,
+      "grad_norm": 0.30989328026771545,
+      "learning_rate": 0.0005,
+      "loss": 0.913956880569458,
+      "step": 1420
+    },
+    {
+      "epoch": 1.9202702702702703,
+      "grad_norm": 0.39757466316223145,
+      "learning_rate": 0.0005,
+      "loss": 0.9604172110557556,
+      "step": 1421
+    },
+    {
+      "epoch": 1.9216216216216218,
+      "grad_norm": 0.42020246386528015,
+      "learning_rate": 0.0005,
+      "loss": 0.9496593475341797,
+      "step": 1422
+    },
+    {
+      "epoch": 1.922972972972973,
+      "grad_norm": 0.37751004099845886,
+      "learning_rate": 0.0005,
+      "loss": 0.9375120401382446,
+      "step": 1423
+    },
+    {
+      "epoch": 1.9243243243243242,
+      "grad_norm": 0.338034987449646,
+      "learning_rate": 0.0005,
+      "loss": 0.9221707582473755,
+      "step": 1424
+    },
+    {
+      "epoch": 1.9256756756756757,
+      "grad_norm": 0.3542296588420868,
+      "learning_rate": 0.0005,
+      "loss": 0.9329007863998413,
+      "step": 1425
+    },
+    {
+      "epoch": 1.927027027027027,
+      "grad_norm": 0.35013195872306824,
+      "learning_rate": 0.0005,
+      "loss": 0.9499654769897461,
+      "step": 1426
+    },
+    {
+      "epoch": 1.9283783783783783,
+      "grad_norm": 0.3987303674221039,
+      "learning_rate": 0.0005,
+      "loss": 0.9582847952842712,
+      "step": 1427
+    },
+    {
+      "epoch": 1.9297297297297298,
+      "grad_norm": 0.36358097195625305,
+      "learning_rate": 0.0005,
+      "loss": 0.9612246751785278,
+      "step": 1428
+    },
+    {
+      "epoch": 1.9310810810810812,
+      "grad_norm": 0.3315325379371643,
+      "learning_rate": 0.0005,
+      "loss": 0.9445121884346008,
+      "step": 1429
+    },
+    {
+      "epoch": 1.9324324324324325,
+      "grad_norm": 0.3791036605834961,
+      "learning_rate": 0.0005,
+      "loss": 0.9347542524337769,
+      "step": 1430
+    },
+    {
+      "epoch": 1.9337837837837837,
+      "grad_norm": 0.34299755096435547,
+      "learning_rate": 0.0005,
+      "loss": 0.9274984002113342,
+      "step": 1431
+    },
+    {
+      "epoch": 1.9351351351351351,
+      "grad_norm": 0.3397952616214752,
+      "learning_rate": 0.0005,
+      "loss": 0.937746524810791,
+      "step": 1432
+    },
+    {
+      "epoch": 1.9364864864864866,
+      "grad_norm": 0.31259703636169434,
+      "learning_rate": 0.0005,
+      "loss": 0.9210049510002136,
+      "step": 1433
+    },
+    {
+      "epoch": 1.9378378378378378,
+      "grad_norm": 0.3268245756626129,
+      "learning_rate": 0.0005,
+      "loss": 0.9278097748756409,
+      "step": 1434
+    },
+    {
+      "epoch": 1.939189189189189,
+      "grad_norm": 0.40244683623313904,
+      "learning_rate": 0.0005,
+      "loss": 0.9183006286621094,
+      "step": 1435
+    },
+    {
+      "epoch": 1.9405405405405407,
+      "grad_norm": 0.3833875060081482,
+      "learning_rate": 0.0005,
+      "loss": 0.9402996897697449,
+      "step": 1436
+    },
+    {
+      "epoch": 1.941891891891892,
+      "grad_norm": 0.3544410169124603,
+      "learning_rate": 0.0005,
+      "loss": 0.9294213056564331,
+      "step": 1437
+    },
+    {
+      "epoch": 1.9432432432432432,
+      "grad_norm": 0.2884667217731476,
+      "learning_rate": 0.0005,
+      "loss": 0.913970947265625,
+      "step": 1438
+    },
+    {
+      "epoch": 1.9445945945945946,
+      "grad_norm": 0.37266939878463745,
+      "learning_rate": 0.0005,
+      "loss": 0.9208884239196777,
+      "step": 1439
+    },
+    {
+      "epoch": 1.945945945945946,
+      "grad_norm": 0.3247826397418976,
+      "learning_rate": 0.0005,
+      "loss": 0.9036575555801392,
+      "step": 1440
+    },
+    {
+      "epoch": 1.9472972972972973,
+      "grad_norm": 0.3434395492076874,
+      "learning_rate": 0.0005,
+      "loss": 0.9535324573516846,
+      "step": 1441
+    },
+    {
+      "epoch": 1.9486486486486485,
+      "grad_norm": 0.40887948870658875,
+      "learning_rate": 0.0005,
+      "loss": 0.9511103630065918,
+      "step": 1442
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.3621883988380432,
+      "learning_rate": 0.0005,
+      "loss": 0.9601436257362366,
+      "step": 1443
+    },
+    {
+      "epoch": 1.9513513513513514,
+      "grad_norm": 0.3686995804309845,
+      "learning_rate": 0.0005,
+      "loss": 0.9410301446914673,
+      "step": 1444
+    },
+    {
+      "epoch": 1.9527027027027026,
+      "grad_norm": 0.397163063287735,
+      "learning_rate": 0.0005,
+      "loss": 0.9533935189247131,
+      "step": 1445
+    },
+    {
+      "epoch": 1.954054054054054,
+      "grad_norm": 0.393196702003479,
+      "learning_rate": 0.0005,
+      "loss": 0.9490337371826172,
+      "step": 1446
+    },
+    {
+      "epoch": 1.9554054054054055,
+      "grad_norm": 0.339736670255661,
+      "learning_rate": 0.0005,
+      "loss": 0.9453850388526917,
+      "step": 1447
+    },
+    {
+      "epoch": 1.9567567567567568,
+      "grad_norm": 0.40998855233192444,
+      "learning_rate": 0.0005,
+      "loss": 0.9490687847137451,
+      "step": 1448
+    },
+    {
+      "epoch": 1.958108108108108,
+      "grad_norm": 0.3665686249732971,
+      "learning_rate": 0.0005,
+      "loss": 0.9546401500701904,
+      "step": 1449
+    },
+    {
+      "epoch": 1.9594594594594594,
+      "grad_norm": 0.34711989760398865,
+      "learning_rate": 0.0005,
+      "loss": 0.9304929971694946,
+      "step": 1450
+    },
+    {
+      "epoch": 1.9608108108108109,
+      "grad_norm": 0.36591508984565735,
+      "learning_rate": 0.0005,
+      "loss": 0.970090389251709,
+      "step": 1451
+    },
+    {
+      "epoch": 1.962162162162162,
+      "grad_norm": 0.43281546235084534,
+      "learning_rate": 0.0005,
+      "loss": 0.9781184196472168,
+      "step": 1452
+    },
+    {
+      "epoch": 1.9635135135135136,
+      "grad_norm": 0.36678797006607056,
+      "learning_rate": 0.0005,
+      "loss": 0.9482865333557129,
+      "step": 1453
+    },
+    {
+      "epoch": 1.964864864864865,
+      "grad_norm": 0.277685284614563,
+      "learning_rate": 0.0005,
+      "loss": 0.9160224199295044,
+      "step": 1454
+    },
+    {
+      "epoch": 1.9662162162162162,
+      "grad_norm": 0.4385475516319275,
+      "learning_rate": 0.0005,
+      "loss": 0.9369759559631348,
+      "step": 1455
+    },
+    {
+      "epoch": 1.9675675675675675,
+      "grad_norm": 0.32812947034835815,
+      "learning_rate": 0.0005,
+      "loss": 0.9296273589134216,
+      "step": 1456
+    },
+    {
+      "epoch": 1.968918918918919,
+      "grad_norm": 0.35258278250694275,
+      "learning_rate": 0.0005,
+      "loss": 0.9678933620452881,
+      "step": 1457
+    },
+    {
+      "epoch": 1.9702702702702704,
+      "grad_norm": 0.3357744514942169,
+      "learning_rate": 0.0005,
+      "loss": 0.9448915719985962,
+      "step": 1458
+    },
+    {
+      "epoch": 1.9716216216216216,
+      "grad_norm": 0.357829213142395,
+      "learning_rate": 0.0005,
+      "loss": 0.9337435364723206,
+      "step": 1459
+    },
+    {
+      "epoch": 1.972972972972973,
+      "grad_norm": 0.39326247572898865,
+      "learning_rate": 0.0005,
+      "loss": 0.962338924407959,
+      "step": 1460
+    },
+    {
+      "epoch": 1.9743243243243245,
+      "grad_norm": 0.3903563916683197,
+      "learning_rate": 0.0005,
+      "loss": 0.9464069604873657,
+      "step": 1461
+    },
+    {
+      "epoch": 1.9756756756756757,
+      "grad_norm": 0.38473695516586304,
+      "learning_rate": 0.0005,
+      "loss": 0.9433084726333618,
+      "step": 1462
+    },
+    {
+      "epoch": 1.977027027027027,
+      "grad_norm": 0.3408110737800598,
+      "learning_rate": 0.0005,
+      "loss": 0.9166686534881592,
+      "step": 1463
+    },
+    {
+      "epoch": 1.9783783783783784,
+      "grad_norm": 0.4129309356212616,
+      "learning_rate": 0.0005,
+      "loss": 0.9457577466964722,
+      "step": 1464
+    },
+    {
+      "epoch": 1.9797297297297298,
+      "grad_norm": 0.37798449397087097,
+      "learning_rate": 0.0005,
+      "loss": 0.9601933360099792,
+      "step": 1465
+    },
+    {
+      "epoch": 1.981081081081081,
+      "grad_norm": 0.3375338912010193,
+      "learning_rate": 0.0005,
+      "loss": 0.9426185488700867,
+      "step": 1466
+    },
+    {
+      "epoch": 1.9824324324324323,
+      "grad_norm": 0.3807815611362457,
+      "learning_rate": 0.0005,
+      "loss": 0.9425791501998901,
+      "step": 1467
+    },
+    {
+      "epoch": 1.983783783783784,
+      "grad_norm": 0.34266138076782227,
+      "learning_rate": 0.0005,
+      "loss": 0.955440878868103,
+      "step": 1468
+    },
+    {
+      "epoch": 1.9851351351351352,
+      "grad_norm": 0.3668674826622009,
+      "learning_rate": 0.0005,
+      "loss": 0.9458931684494019,
+      "step": 1469
+    },
+    {
+      "epoch": 1.9864864864864864,
+      "grad_norm": 0.366774320602417,
+      "learning_rate": 0.0005,
+      "loss": 0.9480865001678467,
+      "step": 1470
+    },
+    {
+      "epoch": 1.9878378378378379,
+      "grad_norm": 0.3683418333530426,
+      "learning_rate": 0.0005,
+      "loss": 0.9433391690254211,
+      "step": 1471
+    },
+    {
+      "epoch": 1.9891891891891893,
+      "grad_norm": 0.41716790199279785,
+      "learning_rate": 0.0005,
+      "loss": 0.9557572603225708,
+      "step": 1472
+    },
+    {
+      "epoch": 1.9905405405405405,
+      "grad_norm": 0.37185683846473694,
+      "learning_rate": 0.0005,
+      "loss": 0.9274119138717651,
+      "step": 1473
+    },
+    {
+      "epoch": 1.9918918918918918,
+      "grad_norm": 0.32273948192596436,
+      "learning_rate": 0.0005,
+      "loss": 0.9289864301681519,
+      "step": 1474
+    },
+    {
+      "epoch": 1.9932432432432432,
+      "grad_norm": 0.3530927002429962,
+      "learning_rate": 0.0005,
+      "loss": 0.9308085441589355,
+      "step": 1475
+    },
+    {
+      "epoch": 1.9945945945945946,
+      "grad_norm": 0.33832502365112305,
+      "learning_rate": 0.0005,
+      "loss": 0.9202552437782288,
+      "step": 1476
+    },
+    {
+      "epoch": 1.9959459459459459,
+      "grad_norm": 0.4105825126171112,
+      "learning_rate": 0.0005,
+      "loss": 0.9394246935844421,
+      "step": 1477
+    },
+    {
+      "epoch": 1.9972972972972973,
+      "grad_norm": 0.33774396777153015,
+      "learning_rate": 0.0005,
+      "loss": 0.9227287769317627,
+      "step": 1478
+    },
+    {
+      "epoch": 1.9986486486486488,
+      "grad_norm": 0.3160061240196228,
+      "learning_rate": 0.0005,
+      "loss": 0.9324238300323486,
+      "step": 1479
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.30825042724609375,
+      "learning_rate": 0.0005,
+      "loss": 0.9257220029830933,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0013513513513512,
+      "grad_norm": 0.40351760387420654,
+      "learning_rate": 0.0005,
+      "loss": 0.9435147047042847,
+      "step": 1481
+    },
+    {
+      "epoch": 2.002702702702703,
+      "grad_norm": 0.29680851101875305,
+      "learning_rate": 0.0004999994367695001,
+      "loss": 0.9199533462524414,
+      "step": 1482
+    },
+    {
+      "epoch": 2.004054054054054,
+      "grad_norm": 0.36118030548095703,
+      "learning_rate": 0.0004999977470805383,
+      "loss": 0.9379303455352783,
+      "step": 1483
+    },
+    {
+      "epoch": 2.0054054054054054,
+      "grad_norm": 0.3699066638946533,
+      "learning_rate": 0.0004999949309407283,
+      "loss": 0.9634392261505127,
+      "step": 1484
+    },
+    {
+      "epoch": 2.0067567567567566,
+      "grad_norm": 0.28031378984451294,
+      "learning_rate": 0.0004999909883627587,
+      "loss": 0.9041444063186646,
+      "step": 1485
+    },
+    {
+      "epoch": 2.0081081081081082,
+      "grad_norm": 0.34567996859550476,
+      "learning_rate": 0.0004999859193643945,
+      "loss": 0.9505733251571655,
+      "step": 1486
+    },
+    {
+      "epoch": 2.0094594594594595,
+      "grad_norm": 0.3601696789264679,
+      "learning_rate": 0.0004999797239684757,
+      "loss": 0.9408971667289734,
+      "step": 1487
+    },
+    {
+      "epoch": 2.0108108108108107,
+      "grad_norm": 0.34323617815971375,
+      "learning_rate": 0.0004999724022029179,
+      "loss": 0.9247360229492188,
+      "step": 1488
+    },
+    {
+      "epoch": 2.0121621621621624,
+      "grad_norm": 0.28141850233078003,
+      "learning_rate": 0.0004999639541007116,
+      "loss": 0.8996888399124146,
+      "step": 1489
+    },
+    {
+      "epoch": 2.0135135135135136,
+      "grad_norm": 0.3310699760913849,
+      "learning_rate": 0.0004999543796999228,
+      "loss": 0.9173691868782043,
+      "step": 1490
+    },
+    {
+      "epoch": 2.014864864864865,
+      "grad_norm": 0.3400116562843323,
+      "learning_rate": 0.0004999436790436923,
+      "loss": 0.9312518239021301,
+      "step": 1491
+    },
+    {
+      "epoch": 2.016216216216216,
+      "grad_norm": 0.3608151078224182,
+      "learning_rate": 0.0004999318521802356,
+      "loss": 0.9591683149337769,
+      "step": 1492
+    },
+    {
+      "epoch": 2.0175675675675677,
+      "grad_norm": 0.3640655279159546,
+      "learning_rate": 0.0004999188991628425,
+      "loss": 0.9238662719726562,
+      "step": 1493
+    },
+    {
+      "epoch": 2.018918918918919,
+      "grad_norm": 0.31536492705345154,
+      "learning_rate": 0.0004999048200498774,
+      "loss": 0.9129854440689087,
+      "step": 1494
+    },
+    {
+      "epoch": 2.02027027027027,
+      "grad_norm": 0.37519124150276184,
+      "learning_rate": 0.0004998896149047786,
+      "loss": 0.9292105436325073,
+      "step": 1495
+    },
+    {
+      "epoch": 2.0216216216216214,
+      "grad_norm": 0.3185376226902008,
+      "learning_rate": 0.0004998732837960581,
+      "loss": 0.9206859469413757,
+      "step": 1496
+    },
+    {
+      "epoch": 2.022972972972973,
+      "grad_norm": 0.32841283082962036,
+      "learning_rate": 0.0004998558267973013,
+      "loss": 0.9358459711074829,
+      "step": 1497
+    },
+    {
+      "epoch": 2.0243243243243243,
+      "grad_norm": 0.34486502408981323,
+      "learning_rate": 0.0004998372439871668,
+      "loss": 0.8971208333969116,
+      "step": 1498
+    },
+    {
+      "epoch": 2.0256756756756755,
+      "grad_norm": 0.4212709069252014,
+      "learning_rate": 0.0004998175354493857,
+      "loss": 0.9388383030891418,
+      "step": 1499
+    },
+    {
+      "epoch": 2.027027027027027,
+      "grad_norm": 0.34814438223838806,
+      "learning_rate": 0.0004997967012727618,
+      "loss": 0.9290074110031128,
+      "step": 1500
+    },
+    {
+      "epoch": 2.0283783783783784,
+      "grad_norm": 0.3519778549671173,
+      "learning_rate": 0.0004997747415511704,
+      "loss": 0.9226608276367188,
+      "step": 1501
+    },
+    {
+      "epoch": 2.0297297297297296,
+      "grad_norm": 0.3285072147846222,
+      "learning_rate": 0.0004997516563835587,
+      "loss": 0.9417418241500854,
+      "step": 1502
+    },
+    {
+      "epoch": 2.031081081081081,
+      "grad_norm": 0.3209885358810425,
+      "learning_rate": 0.0004997274458739449,
+      "loss": 0.9127825498580933,
+      "step": 1503
+    },
+    {
+      "epoch": 2.0324324324324325,
+      "grad_norm": 0.37093931436538696,
+      "learning_rate": 0.0004997021101314179,
+      "loss": 0.9473943710327148,
+      "step": 1504
+    },
+    {
+      "epoch": 2.0337837837837838,
+      "grad_norm": 0.3965378999710083,
+      "learning_rate": 0.0004996756492701362,
+      "loss": 0.9447396993637085,
+      "step": 1505
+    },
+    {
+      "epoch": 2.035135135135135,
+      "grad_norm": 0.34661123156547546,
+      "learning_rate": 0.0004996480634093287,
+      "loss": 0.9288147687911987,
+      "step": 1506
+    },
+    {
+      "epoch": 2.0364864864864867,
+      "grad_norm": 0.3193095326423645,
+      "learning_rate": 0.0004996193526732929,
+      "loss": 0.9089558720588684,
+      "step": 1507
+    },
+    {
+      "epoch": 2.037837837837838,
+      "grad_norm": 0.41070908308029175,
+      "learning_rate": 0.0004995895171913947,
+      "loss": 0.9320091009140015,
+      "step": 1508
+    },
+    {
+      "epoch": 2.039189189189189,
+      "grad_norm": 0.376377135515213,
+      "learning_rate": 0.0004995585570980684,
+      "loss": 0.944176435470581,
+      "step": 1509
+    },
+    {
+      "epoch": 2.0405405405405403,
+      "grad_norm": 0.322681188583374,
+      "learning_rate": 0.0004995264725328151,
+      "loss": 0.9170739650726318,
+      "step": 1510
+    },
+    {
+      "epoch": 2.041891891891892,
+      "grad_norm": 0.3755601942539215,
+      "learning_rate": 0.0004994932636402031,
+      "loss": 0.9281123280525208,
+      "step": 1511
+    },
+    {
+      "epoch": 2.0432432432432432,
+      "grad_norm": 0.35562393069267273,
+      "learning_rate": 0.0004994589305698663,
+      "loss": 0.9291849136352539,
+      "step": 1512
+    },
+    {
+      "epoch": 2.0445945945945945,
+      "grad_norm": 0.39720407128334045,
+      "learning_rate": 0.0004994234734765043,
+      "loss": 0.9016556739807129,
+      "step": 1513
+    },
+    {
+      "epoch": 2.045945945945946,
+      "grad_norm": 0.3343566358089447,
+      "learning_rate": 0.0004993868925198811,
+      "loss": 0.9050730466842651,
+      "step": 1514
+    },
+    {
+      "epoch": 2.0472972972972974,
+      "grad_norm": 0.3343924283981323,
+      "learning_rate": 0.0004993491878648249,
+      "loss": 0.9193119406700134,
+      "step": 1515
+    },
+    {
+      "epoch": 2.0486486486486486,
+      "grad_norm": 0.37646976113319397,
+      "learning_rate": 0.0004993103596812267,
+      "loss": 0.9137660264968872,
+      "step": 1516
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.4122311770915985,
+      "learning_rate": 0.0004992704081440407,
+      "loss": 0.9322822093963623,
+      "step": 1517
+    },
+    {
+      "epoch": 2.0513513513513515,
+      "grad_norm": 0.42987003922462463,
+      "learning_rate": 0.000499229333433282,
+      "loss": 0.9457452893257141,
+      "step": 1518
+    },
+    {
+      "epoch": 2.0527027027027027,
+      "grad_norm": 0.353111207485199,
+      "learning_rate": 0.0004991871357340269,
+      "loss": 0.8966946601867676,
+      "step": 1519
+    },
+    {
+      "epoch": 2.054054054054054,
+      "grad_norm": 0.34224259853363037,
+      "learning_rate": 0.0004991438152364117,
+      "loss": 0.9349232912063599,
+      "step": 1520
+    },
+    {
+      "epoch": 2.0554054054054056,
+      "grad_norm": 0.3404001295566559,
+      "learning_rate": 0.0004990993721356316,
+      "loss": 0.9146379232406616,
+      "step": 1521
+    },
+    {
+      "epoch": 2.056756756756757,
+      "grad_norm": 0.3475813865661621,
+      "learning_rate": 0.0004990538066319406,
+      "loss": 0.9078060388565063,
+      "step": 1522
+    },
+    {
+      "epoch": 2.058108108108108,
+      "grad_norm": 0.3372005224227905,
+      "learning_rate": 0.0004990071189306495,
+      "loss": 0.9421411752700806,
+      "step": 1523
+    },
+    {
+      "epoch": 2.0594594594594593,
+      "grad_norm": 0.3451443016529083,
+      "learning_rate": 0.0004989593092421258,
+      "loss": 0.9180985689163208,
+      "step": 1524
+    },
+    {
+      "epoch": 2.060810810810811,
+      "grad_norm": 0.42501553893089294,
+      "learning_rate": 0.0004989103777817928,
+      "loss": 0.9358645677566528,
+      "step": 1525
+    },
+    {
+      "epoch": 2.062162162162162,
+      "grad_norm": 0.30565014481544495,
+      "learning_rate": 0.0004988603247701276,
+      "loss": 0.9131169319152832,
+      "step": 1526
+    },
+    {
+      "epoch": 2.0635135135135134,
+      "grad_norm": 0.3487398028373718,
+      "learning_rate": 0.0004988091504326616,
+      "loss": 0.9349451065063477,
+      "step": 1527
+    },
+    {
+      "epoch": 2.064864864864865,
+      "grad_norm": 0.3314070701599121,
+      "learning_rate": 0.0004987568549999782,
+      "loss": 0.9288805723190308,
+      "step": 1528
+    },
+    {
+      "epoch": 2.0662162162162163,
+      "grad_norm": 0.3746781051158905,
+      "learning_rate": 0.0004987034387077126,
+      "loss": 0.9593837857246399,
+      "step": 1529
+    },
+    {
+      "epoch": 2.0675675675675675,
+      "grad_norm": 0.2903171181678772,
+      "learning_rate": 0.0004986489017965501,
+      "loss": 0.9163342714309692,
+      "step": 1530
+    },
+    {
+      "epoch": 2.0689189189189188,
+      "grad_norm": 0.36510220170021057,
+      "learning_rate": 0.0004985932445122257,
+      "loss": 0.9368723034858704,
+      "step": 1531
+    },
+    {
+      "epoch": 2.0702702702702704,
+      "grad_norm": 0.3276277780532837,
+      "learning_rate": 0.0004985364671055223,
+      "loss": 0.911605715751648,
+      "step": 1532
+    },
+    {
+      "epoch": 2.0716216216216217,
+      "grad_norm": 0.33477073907852173,
+      "learning_rate": 0.0004984785698322699,
+      "loss": 0.9151954650878906,
+      "step": 1533
+    },
+    {
+      "epoch": 2.072972972972973,
+      "grad_norm": 0.32165735960006714,
+      "learning_rate": 0.0004984195529533451,
+      "loss": 0.9306996464729309,
+      "step": 1534
+    },
+    {
+      "epoch": 2.074324324324324,
+      "grad_norm": 0.3432990610599518,
+      "learning_rate": 0.0004983594167346681,
+      "loss": 0.9004199504852295,
+      "step": 1535
+    },
+    {
+      "epoch": 2.075675675675676,
+      "grad_norm": 0.33682093024253845,
+      "learning_rate": 0.0004982981614472039,
+      "loss": 0.9015887975692749,
+      "step": 1536
+    },
+    {
+      "epoch": 2.077027027027027,
+      "grad_norm": 0.3301418125629425,
+      "learning_rate": 0.0004982357873669588,
+      "loss": 0.912014365196228,
+      "step": 1537
+    },
+    {
+      "epoch": 2.0783783783783782,
+      "grad_norm": 0.4029909372329712,
+      "learning_rate": 0.0004981722947749811,
+      "loss": 0.9225188493728638,
+      "step": 1538
+    },
+    {
+      "epoch": 2.07972972972973,
+      "grad_norm": 0.37054628133773804,
+      "learning_rate": 0.0004981076839573581,
+      "loss": 0.9330065250396729,
+      "step": 1539
+    },
+    {
+      "epoch": 2.081081081081081,
+      "grad_norm": 0.37817442417144775,
+      "learning_rate": 0.0004980419552052163,
+      "loss": 0.92266845703125,
+      "step": 1540
+    },
+    {
+      "epoch": 2.0824324324324324,
+      "grad_norm": 0.3882576525211334,
+      "learning_rate": 0.0004979751088147192,
+      "loss": 0.9321296215057373,
+      "step": 1541
+    },
+    {
+      "epoch": 2.0837837837837836,
+      "grad_norm": 0.40052473545074463,
+      "learning_rate": 0.0004979071450870662,
+      "loss": 0.9439878463745117,
+      "step": 1542
+    },
+    {
+      "epoch": 2.0851351351351353,
+      "grad_norm": 0.3160717487335205,
+      "learning_rate": 0.0004978380643284912,
+      "loss": 0.9203035831451416,
+      "step": 1543
+    },
+    {
+      "epoch": 2.0864864864864865,
+      "grad_norm": 0.34737688302993774,
+      "learning_rate": 0.0004977678668502614,
+      "loss": 0.9179403185844421,
+      "step": 1544
+    },
+    {
+      "epoch": 2.0878378378378377,
+      "grad_norm": 0.3675885498523712,
+      "learning_rate": 0.0004976965529686756,
+      "loss": 0.9211385250091553,
+      "step": 1545
+    },
+    {
+      "epoch": 2.0891891891891894,
+      "grad_norm": 0.383205771446228,
+      "learning_rate": 0.000497624123005063,
+      "loss": 0.9535226225852966,
+      "step": 1546
+    },
+    {
+      "epoch": 2.0905405405405406,
+      "grad_norm": 0.37953585386276245,
+      "learning_rate": 0.0004975505772857818,
+      "loss": 0.9473111629486084,
+      "step": 1547
+    },
+    {
+      "epoch": 2.091891891891892,
+      "grad_norm": 0.3481861650943756,
+      "learning_rate": 0.0004974759161422175,
+      "loss": 0.9391646385192871,
+      "step": 1548
+    },
+    {
+      "epoch": 2.093243243243243,
+      "grad_norm": 0.3527507781982422,
+      "learning_rate": 0.0004974001399107816,
+      "loss": 0.9470508098602295,
+      "step": 1549
+    },
+    {
+      "epoch": 2.0945945945945947,
+      "grad_norm": 0.35015490651130676,
+      "learning_rate": 0.0004973232489329099,
+      "loss": 0.9472455382347107,
+      "step": 1550
+    },
+    {
+      "epoch": 2.095945945945946,
+      "grad_norm": 0.2705489695072174,
+      "learning_rate": 0.0004972452435550613,
+      "loss": 0.9120519161224365,
+      "step": 1551
+    },
+    {
+      "epoch": 2.097297297297297,
+      "grad_norm": 0.3463180661201477,
+      "learning_rate": 0.0004971661241287157,
+      "loss": 0.9323123693466187,
+      "step": 1552
+    },
+    {
+      "epoch": 2.098648648648649,
+      "grad_norm": 0.39689308404922485,
+      "learning_rate": 0.0004970858910103731,
+      "loss": 0.9348223209381104,
+      "step": 1553
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.3028721511363983,
+      "learning_rate": 0.0004970045445615512,
+      "loss": 0.916419267654419,
+      "step": 1554
+    },
+    {
+      "epoch": 2.1013513513513513,
+      "grad_norm": 0.3976389169692993,
+      "learning_rate": 0.0004969220851487844,
+      "loss": 0.9525309801101685,
+      "step": 1555
+    },
+    {
+      "epoch": 2.1027027027027025,
+      "grad_norm": 0.44023239612579346,
+      "learning_rate": 0.0004968385131436222,
+      "loss": 0.951092004776001,
+      "step": 1556
+    },
+    {
+      "epoch": 2.104054054054054,
+      "grad_norm": 0.3938271701335907,
+      "learning_rate": 0.0004967538289226267,
+      "loss": 0.9260006546974182,
+      "step": 1557
+    },
+    {
+      "epoch": 2.1054054054054054,
+      "grad_norm": 0.33738309144973755,
+      "learning_rate": 0.000496668032867372,
+      "loss": 0.9276026487350464,
+      "step": 1558
+    },
+    {
+      "epoch": 2.1067567567567567,
+      "grad_norm": 0.3689638078212738,
+      "learning_rate": 0.0004965811253644418,
+      "loss": 0.920623779296875,
+      "step": 1559
+    },
+    {
+      "epoch": 2.108108108108108,
+      "grad_norm": 0.35270920395851135,
+      "learning_rate": 0.0004964931068054274,
+      "loss": 0.9153560400009155,
+      "step": 1560
+    },
+    {
+      "epoch": 2.1094594594594596,
+      "grad_norm": 0.3581753075122833,
+      "learning_rate": 0.0004964039775869272,
+      "loss": 0.9311813712120056,
+      "step": 1561
+    },
+    {
+      "epoch": 2.110810810810811,
+      "grad_norm": 0.3324107229709625,
+      "learning_rate": 0.0004963137381105431,
+      "loss": 0.9193117022514343,
+      "step": 1562
+    },
+    {
+      "epoch": 2.112162162162162,
+      "grad_norm": 0.31005147099494934,
+      "learning_rate": 0.0004962223887828803,
+      "loss": 0.9191339612007141,
+      "step": 1563
+    },
+    {
+      "epoch": 2.1135135135135137,
+      "grad_norm": 0.3310469090938568,
+      "learning_rate": 0.0004961299300155446,
+      "loss": 0.9108705520629883,
+      "step": 1564
+    },
+    {
+      "epoch": 2.114864864864865,
+      "grad_norm": 0.36150574684143066,
+      "learning_rate": 0.0004960363622251409,
+      "loss": 0.9215501546859741,
+      "step": 1565
+    },
+    {
+      "epoch": 2.116216216216216,
+      "grad_norm": 0.3973751962184906,
+      "learning_rate": 0.0004959416858332709,
+      "loss": 0.9502665400505066,
+      "step": 1566
+    },
+    {
+      "epoch": 2.1175675675675674,
+      "grad_norm": 0.3337228000164032,
+      "learning_rate": 0.0004958459012665317,
+      "loss": 0.9413602948188782,
+      "step": 1567
+    },
+    {
+      "epoch": 2.118918918918919,
+      "grad_norm": 0.3548646867275238,
+      "learning_rate": 0.0004957490089565137,
+      "loss": 0.9254937171936035,
+      "step": 1568
+    },
+    {
+      "epoch": 2.1202702702702703,
+      "grad_norm": 0.32498660683631897,
+      "learning_rate": 0.0004956510093397983,
+      "loss": 0.9231414794921875,
+      "step": 1569
+    },
+    {
+      "epoch": 2.1216216216216215,
+      "grad_norm": 0.43852025270462036,
+      "learning_rate": 0.0004955519028579568,
+      "loss": 0.9268288612365723,
+      "step": 1570
+    },
+    {
+      "epoch": 2.122972972972973,
+      "grad_norm": 0.4087459146976471,
+      "learning_rate": 0.0004954516899575473,
+      "loss": 0.931428074836731,
+      "step": 1571
+    },
+    {
+      "epoch": 2.1243243243243244,
+      "grad_norm": 0.38924354314804077,
+      "learning_rate": 0.0004953503710901136,
+      "loss": 0.9467356204986572,
+      "step": 1572
+    },
+    {
+      "epoch": 2.1256756756756756,
+      "grad_norm": 0.413943350315094,
+      "learning_rate": 0.0004952479467121827,
+      "loss": 0.948610246181488,
+      "step": 1573
+    },
+    {
+      "epoch": 2.127027027027027,
+      "grad_norm": 0.3289583623409271,
+      "learning_rate": 0.0004951444172852629,
+      "loss": 0.9140801429748535,
+      "step": 1574
+    },
+    {
+      "epoch": 2.1283783783783785,
+      "grad_norm": 0.3322771191596985,
+      "learning_rate": 0.0004950397832758415,
+      "loss": 0.9105451107025146,
+      "step": 1575
+    },
+    {
+      "epoch": 2.1297297297297297,
+      "grad_norm": 0.29896238446235657,
+      "learning_rate": 0.0004949340451553833,
+      "loss": 0.9153224229812622,
+      "step": 1576
+    },
+    {
+      "epoch": 2.131081081081081,
+      "grad_norm": 0.33175188302993774,
+      "learning_rate": 0.0004948272034003275,
+      "loss": 0.9163780212402344,
+      "step": 1577
+    },
+    {
+      "epoch": 2.1324324324324326,
+      "grad_norm": 0.31124094128608704,
+      "learning_rate": 0.0004947192584920866,
+      "loss": 0.8927855491638184,
+      "step": 1578
+    },
+    {
+      "epoch": 2.133783783783784,
+      "grad_norm": 0.3376350700855255,
+      "learning_rate": 0.0004946102109170433,
+      "loss": 0.9334204196929932,
+      "step": 1579
+    },
+    {
+      "epoch": 2.135135135135135,
+      "grad_norm": 0.40761080384254456,
+      "learning_rate": 0.0004945000611665491,
+      "loss": 0.9349830150604248,
+      "step": 1580
+    },
+    {
+      "epoch": 2.1364864864864863,
+      "grad_norm": 0.2953258454799652,
+      "learning_rate": 0.0004943888097369216,
+      "loss": 0.9071077704429626,
+      "step": 1581
+    },
+    {
+      "epoch": 2.137837837837838,
+      "grad_norm": 0.32923176884651184,
+      "learning_rate": 0.0004942764571294422,
+      "loss": 0.9198494553565979,
+      "step": 1582
+    },
+    {
+      "epoch": 2.139189189189189,
+      "grad_norm": 0.3684205710887909,
+      "learning_rate": 0.0004941630038503545,
+      "loss": 0.9333382248878479,
+      "step": 1583
+    },
+    {
+      "epoch": 2.1405405405405404,
+      "grad_norm": 0.3205466568470001,
+      "learning_rate": 0.0004940484504108612,
+      "loss": 0.9127440452575684,
+      "step": 1584
+    },
+    {
+      "epoch": 2.141891891891892,
+      "grad_norm": 0.3288394808769226,
+      "learning_rate": 0.0004939327973271222,
+      "loss": 0.9108311533927917,
+      "step": 1585
+    },
+    {
+      "epoch": 2.1432432432432433,
+      "grad_norm": 0.41623908281326294,
+      "learning_rate": 0.000493816045120252,
+      "loss": 0.9381232857704163,
+      "step": 1586
+    },
+    {
+      "epoch": 2.1445945945945946,
+      "grad_norm": 0.3330880403518677,
+      "learning_rate": 0.0004936981943163182,
+      "loss": 0.924103856086731,
+      "step": 1587
+    },
+    {
+      "epoch": 2.145945945945946,
+      "grad_norm": 0.3432510793209076,
+      "learning_rate": 0.0004935792454463381,
+      "loss": 0.9211220741271973,
+      "step": 1588
+    },
+    {
+      "epoch": 2.1472972972972975,
+      "grad_norm": 0.38387295603752136,
+      "learning_rate": 0.0004934591990462766,
+      "loss": 0.9327381253242493,
+      "step": 1589
+    },
+    {
+      "epoch": 2.1486486486486487,
+      "grad_norm": 0.396767795085907,
+      "learning_rate": 0.000493338055657044,
+      "loss": 0.9473586678504944,
+      "step": 1590
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.3918689787387848,
+      "learning_rate": 0.0004932158158244937,
+      "loss": 0.9229852557182312,
+      "step": 1591
+    },
+    {
+      "epoch": 2.1513513513513516,
+      "grad_norm": 0.33155199885368347,
+      "learning_rate": 0.0004930924800994192,
+      "loss": 0.9128546714782715,
+      "step": 1592
+    },
+    {
+      "epoch": 2.152702702702703,
+      "grad_norm": 0.32360684871673584,
+      "learning_rate": 0.000492968049037552,
+      "loss": 0.9356454610824585,
+      "step": 1593
+    },
+    {
+      "epoch": 2.154054054054054,
+      "grad_norm": 0.32011860609054565,
+      "learning_rate": 0.0004928425231995593,
+      "loss": 0.9254053831100464,
+      "step": 1594
+    },
+    {
+      "epoch": 2.1554054054054053,
+      "grad_norm": 0.34817421436309814,
+      "learning_rate": 0.0004927159031510405,
+      "loss": 0.9244092106819153,
+      "step": 1595
+    },
+    {
+      "epoch": 2.156756756756757,
+      "grad_norm": 0.33490800857543945,
+      "learning_rate": 0.0004925881894625263,
+      "loss": 0.9162651300430298,
+      "step": 1596
+    },
+    {
+      "epoch": 2.158108108108108,
+      "grad_norm": 0.419757217168808,
+      "learning_rate": 0.0004924593827094744,
+      "loss": 0.9346417784690857,
+      "step": 1597
+    },
+    {
+      "epoch": 2.1594594594594594,
+      "grad_norm": 0.3456192910671234,
+      "learning_rate": 0.0004923294834722678,
+      "loss": 0.9282587766647339,
+      "step": 1598
+    },
+    {
+      "epoch": 2.1608108108108106,
+      "grad_norm": 0.30417534708976746,
+      "learning_rate": 0.0004921984923362124,
+      "loss": 0.9149502515792847,
+      "step": 1599
+    },
+    {
+      "epoch": 2.1621621621621623,
+      "grad_norm": 0.3309459090232849,
+      "learning_rate": 0.0004920664098915337,
+      "loss": 0.9259644746780396,
+      "step": 1600
+    },
+    {
+      "epoch": 2.1635135135135135,
+      "grad_norm": 0.3152301609516144,
+      "learning_rate": 0.0004919332367333747,
+      "loss": 0.9188176393508911,
+      "step": 1601
+    },
+    {
+      "epoch": 2.1648648648648647,
+      "grad_norm": 0.3448285162448883,
+      "learning_rate": 0.0004917989734617928,
+      "loss": 0.9362011551856995,
+      "step": 1602
+    },
+    {
+      "epoch": 2.1662162162162164,
+      "grad_norm": 0.406120240688324,
+      "learning_rate": 0.0004916636206817575,
+      "loss": 0.8999876976013184,
+      "step": 1603
+    },
+    {
+      "epoch": 2.1675675675675676,
+      "grad_norm": 0.3502003848552704,
+      "learning_rate": 0.0004915271790031471,
+      "loss": 0.9552854299545288,
+      "step": 1604
+    },
+    {
+      "epoch": 2.168918918918919,
+      "grad_norm": 0.3419058322906494,
+      "learning_rate": 0.0004913896490407467,
+      "loss": 0.9019888043403625,
+      "step": 1605
+    },
+    {
+      "epoch": 2.17027027027027,
+      "grad_norm": 0.31682509183883667,
+      "learning_rate": 0.0004912510314142447,
+      "loss": 0.8971027731895447,
+      "step": 1606
+    },
+    {
+      "epoch": 2.1716216216216218,
+      "grad_norm": 0.36299100518226624,
+      "learning_rate": 0.0004911113267482307,
+      "loss": 0.9274662733078003,
+      "step": 1607
+    },
+    {
+      "epoch": 2.172972972972973,
+      "grad_norm": 0.33910346031188965,
+      "learning_rate": 0.0004909705356721919,
+      "loss": 0.9241058826446533,
+      "step": 1608
+    },
+    {
+      "epoch": 2.174324324324324,
+      "grad_norm": 0.38026130199432373,
+      "learning_rate": 0.000490828658820511,
+      "loss": 0.9293161630630493,
+      "step": 1609
+    },
+    {
+      "epoch": 2.175675675675676,
+      "grad_norm": 0.29337865114212036,
+      "learning_rate": 0.000490685696832463,
+      "loss": 0.9119933247566223,
+      "step": 1610
+    },
+    {
+      "epoch": 2.177027027027027,
+      "grad_norm": 0.3218766748905182,
+      "learning_rate": 0.0004905416503522123,
+      "loss": 0.9101349711418152,
+      "step": 1611
+    },
+    {
+      "epoch": 2.1783783783783783,
+      "grad_norm": 0.3490082323551178,
+      "learning_rate": 0.00049039652002881,
+      "loss": 0.9297454357147217,
+      "step": 1612
+    },
+    {
+      "epoch": 2.1797297297297296,
+      "grad_norm": 0.31801801919937134,
+      "learning_rate": 0.0004902503065161905,
+      "loss": 0.9125286340713501,
+      "step": 1613
+    },
+    {
+      "epoch": 2.1810810810810812,
+      "grad_norm": 0.3560100197792053,
+      "learning_rate": 0.0004901030104731691,
+      "loss": 0.9368000030517578,
+      "step": 1614
+    },
+    {
+      "epoch": 2.1824324324324325,
+      "grad_norm": 0.33292481303215027,
+      "learning_rate": 0.0004899546325634388,
+      "loss": 0.9155285954475403,
+      "step": 1615
+    },
+    {
+      "epoch": 2.1837837837837837,
+      "grad_norm": 0.31558898091316223,
+      "learning_rate": 0.0004898051734555676,
+      "loss": 0.9250460863113403,
+      "step": 1616
+    },
+    {
+      "epoch": 2.185135135135135,
+      "grad_norm": 0.3762737810611725,
+      "learning_rate": 0.0004896546338229945,
+      "loss": 0.8969119787216187,
+      "step": 1617
+    },
+    {
+      "epoch": 2.1864864864864866,
+      "grad_norm": 0.3718250095844269,
+      "learning_rate": 0.0004895030143440278,
+      "loss": 0.9253700971603394,
+      "step": 1618
+    },
+    {
+      "epoch": 2.187837837837838,
+      "grad_norm": 0.28198501467704773,
+      "learning_rate": 0.0004893503157018412,
+      "loss": 0.9053729176521301,
+      "step": 1619
+    },
+    {
+      "epoch": 2.189189189189189,
+      "grad_norm": 0.3092516362667084,
+      "learning_rate": 0.0004891965385844709,
+      "loss": 0.9107194542884827,
+      "step": 1620
+    },
+    {
+      "epoch": 2.1905405405405407,
+      "grad_norm": 0.3224208652973175,
+      "learning_rate": 0.0004890416836848127,
+      "loss": 0.9171070456504822,
+      "step": 1621
+    },
+    {
+      "epoch": 2.191891891891892,
+      "grad_norm": 0.32098761200904846,
+      "learning_rate": 0.0004888857517006186,
+      "loss": 0.9040168523788452,
+      "step": 1622
+    },
+    {
+      "epoch": 2.193243243243243,
+      "grad_norm": 0.38942432403564453,
+      "learning_rate": 0.0004887287433344939,
+      "loss": 0.9369014501571655,
+      "step": 1623
+    },
+    {
+      "epoch": 2.1945945945945944,
+      "grad_norm": 0.31655365228652954,
+      "learning_rate": 0.0004885706592938936,
+      "loss": 0.8992807865142822,
+      "step": 1624
+    },
+    {
+      "epoch": 2.195945945945946,
+      "grad_norm": 0.38557785749435425,
+      "learning_rate": 0.0004884115002911197,
+      "loss": 0.9230440855026245,
+      "step": 1625
+    },
+    {
+      "epoch": 2.1972972972972973,
+      "grad_norm": 0.3073549270629883,
+      "learning_rate": 0.000488251267043318,
+      "loss": 0.9122327566146851,
+      "step": 1626
+    },
+    {
+      "epoch": 2.1986486486486485,
+      "grad_norm": 0.38757145404815674,
+      "learning_rate": 0.00048808996027247453,
+      "loss": 0.9282602667808533,
+      "step": 1627
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.38752660155296326,
+      "learning_rate": 0.00048792758070541234,
+      "loss": 0.9287099838256836,
+      "step": 1628
+    },
+    {
+      "epoch": 2.2013513513513514,
+      "grad_norm": 0.30439382791519165,
+      "learning_rate": 0.0004877641290737884,
+      "loss": 0.9095377922058105,
+      "step": 1629
+    },
+    {
+      "epoch": 2.2027027027027026,
+      "grad_norm": 0.35014665126800537,
+      "learning_rate": 0.00048759960611409036,
+      "loss": 0.9222122430801392,
+      "step": 1630
+    },
+    {
+      "epoch": 2.204054054054054,
+      "grad_norm": 0.29120129346847534,
+      "learning_rate": 0.000487434012567633,
+      "loss": 0.9134736061096191,
+      "step": 1631
+    },
+    {
+      "epoch": 2.2054054054054055,
+      "grad_norm": 0.3225969970226288,
+      "learning_rate": 0.0004872673491805549,
+      "loss": 0.9027793407440186,
+      "step": 1632
+    },
+    {
+      "epoch": 2.2067567567567568,
+      "grad_norm": 0.3528636693954468,
+      "learning_rate": 0.0004870996167038154,
+      "loss": 0.9237062335014343,
+      "step": 1633
+    },
+    {
+      "epoch": 2.208108108108108,
+      "grad_norm": 0.3670106530189514,
+      "learning_rate": 0.0004869308158931909,
+      "loss": 0.9343057870864868,
+      "step": 1634
+    },
+    {
+      "epoch": 2.2094594594594597,
+      "grad_norm": 0.31795746088027954,
+      "learning_rate": 0.00048676094750927144,
+      "loss": 0.9144319295883179,
+      "step": 1635
+    },
+    {
+      "epoch": 2.210810810810811,
+      "grad_norm": 0.4237275719642639,
+      "learning_rate": 0.00048659001231745734,
+      "loss": 0.9481407403945923,
+      "step": 1636
+    },
+    {
+      "epoch": 2.212162162162162,
+      "grad_norm": 0.3293318748474121,
+      "learning_rate": 0.0004864180110879562,
+      "loss": 0.8926504850387573,
+      "step": 1637
+    },
+    {
+      "epoch": 2.2135135135135133,
+      "grad_norm": 0.3425148129463196,
+      "learning_rate": 0.0004862449445957785,
+      "loss": 0.920407772064209,
+      "step": 1638
+    },
+    {
+      "epoch": 2.214864864864865,
+      "grad_norm": 0.38169047236442566,
+      "learning_rate": 0.0004860708136207349,
+      "loss": 0.9240754842758179,
+      "step": 1639
+    },
+    {
+      "epoch": 2.2162162162162162,
+      "grad_norm": 0.39991870522499084,
+      "learning_rate": 0.0004858956189474325,
+      "loss": 0.9396734833717346,
+      "step": 1640
+    },
+    {
+      "epoch": 2.2175675675675675,
+      "grad_norm": 0.4484407603740692,
+      "learning_rate": 0.00048571936136527106,
+      "loss": 0.9516944289207458,
+      "step": 1641
+    },
+    {
+      "epoch": 2.218918918918919,
+      "grad_norm": 0.303413063287735,
+      "learning_rate": 0.0004855420416684398,
+      "loss": 0.9266916513442993,
+      "step": 1642
+    },
+    {
+      "epoch": 2.2202702702702704,
+      "grad_norm": 0.3376297354698181,
+      "learning_rate": 0.00048536366065591354,
+      "loss": 0.9024789333343506,
+      "step": 1643
+    },
+    {
+      "epoch": 2.2216216216216216,
+      "grad_norm": 0.3308083415031433,
+      "learning_rate": 0.0004851842191314494,
+      "loss": 0.9221160411834717,
+      "step": 1644
+    },
+    {
+      "epoch": 2.222972972972973,
+      "grad_norm": 0.30892255902290344,
+      "learning_rate": 0.0004850037179035829,
+      "loss": 0.9091778993606567,
+      "step": 1645
+    },
+    {
+      "epoch": 2.2243243243243245,
+      "grad_norm": 0.32630303502082825,
+      "learning_rate": 0.00048482215778562434,
+      "loss": 0.9243308305740356,
+      "step": 1646
+    },
+    {
+      "epoch": 2.2256756756756757,
+      "grad_norm": 0.35702943801879883,
+      "learning_rate": 0.0004846395395956553,
+      "loss": 0.9348118305206299,
+      "step": 1647
+    },
+    {
+      "epoch": 2.227027027027027,
+      "grad_norm": 0.38458120822906494,
+      "learning_rate": 0.0004844558641565249,
+      "loss": 0.9462944269180298,
+      "step": 1648
+    },
+    {
+      "epoch": 2.2283783783783786,
+      "grad_norm": 0.33020085096359253,
+      "learning_rate": 0.0004842711322958459,
+      "loss": 0.9410030245780945,
+      "step": 1649
+    },
+    {
+      "epoch": 2.22972972972973,
+      "grad_norm": 0.31453147530555725,
+      "learning_rate": 0.00048408534484599143,
+      "loss": 0.9121178388595581,
+      "step": 1650
+    },
+    {
+      "epoch": 2.231081081081081,
+      "grad_norm": 0.35898423194885254,
+      "learning_rate": 0.00048389850264409054,
+      "loss": 0.928550124168396,
+      "step": 1651
+    },
+    {
+      "epoch": 2.2324324324324323,
+      "grad_norm": 0.3258773684501648,
+      "learning_rate": 0.0004837106065320253,
+      "loss": 0.9031070470809937,
+      "step": 1652
+    },
+    {
+      "epoch": 2.233783783783784,
+      "grad_norm": 0.3658798933029175,
+      "learning_rate": 0.00048352165735642607,
+      "loss": 0.9266507625579834,
+      "step": 1653
+    },
+    {
+      "epoch": 2.235135135135135,
+      "grad_norm": 0.31125420331954956,
+      "learning_rate": 0.00048333165596866837,
+      "loss": 0.9096517562866211,
+      "step": 1654
+    },
+    {
+      "epoch": 2.2364864864864864,
+      "grad_norm": 0.3286791443824768,
+      "learning_rate": 0.000483140603224869,
+      "loss": 0.9149118065834045,
+      "step": 1655
+    },
+    {
+      "epoch": 2.237837837837838,
+      "grad_norm": 0.3260885775089264,
+      "learning_rate": 0.00048294849998588155,
+      "loss": 0.9194457530975342,
+      "step": 1656
+    },
+    {
+      "epoch": 2.2391891891891893,
+      "grad_norm": 0.2757807672023773,
+      "learning_rate": 0.0004827553471172935,
+      "loss": 0.8896370530128479,
+      "step": 1657
+    },
+    {
+      "epoch": 2.2405405405405405,
+      "grad_norm": 0.41918718814849854,
+      "learning_rate": 0.0004825611454894213,
+      "loss": 0.920689582824707,
+      "step": 1658
+    },
+    {
+      "epoch": 2.2418918918918918,
+      "grad_norm": 0.3168398439884186,
+      "learning_rate": 0.00048236589597730744,
+      "loss": 0.9059911370277405,
+      "step": 1659
+    },
+    {
+      "epoch": 2.2432432432432434,
+      "grad_norm": 0.3626181483268738,
+      "learning_rate": 0.0004821695994607156,
+      "loss": 0.9240192174911499,
+      "step": 1660
+    },
+    {
+      "epoch": 2.2445945945945946,
+      "grad_norm": 0.3881918489933014,
+      "learning_rate": 0.0004819722568241274,
+      "loss": 0.9143047332763672,
+      "step": 1661
+    },
+    {
+      "epoch": 2.245945945945946,
+      "grad_norm": 0.33343860507011414,
+      "learning_rate": 0.00048177386895673774,
+      "loss": 0.916669487953186,
+      "step": 1662
+    },
+    {
+      "epoch": 2.247297297297297,
+      "grad_norm": 0.4243029057979584,
+      "learning_rate": 0.0004815744367524516,
+      "loss": 0.955696702003479,
+      "step": 1663
+    },
+    {
+      "epoch": 2.2486486486486488,
+      "grad_norm": 0.4440765976905823,
+      "learning_rate": 0.0004813739611098793,
+      "loss": 0.9308452606201172,
+      "step": 1664
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.2955050766468048,
+      "learning_rate": 0.0004811724429323329,
+      "loss": 0.9163042306900024,
+      "step": 1665
+    },
+    {
+      "epoch": 2.2513513513513512,
+      "grad_norm": 0.3259715437889099,
+      "learning_rate": 0.0004809698831278217,
+      "loss": 0.931037187576294,
+      "step": 1666
+    },
+    {
+      "epoch": 2.2527027027027025,
+      "grad_norm": 0.3755234181880951,
+      "learning_rate": 0.0004807662826090488,
+      "loss": 0.9294158220291138,
+      "step": 1667
+    },
+    {
+      "epoch": 2.254054054054054,
+      "grad_norm": 0.41122525930404663,
+      "learning_rate": 0.00048056164229340613,
+      "loss": 0.9260744452476501,
+      "step": 1668
+    },
+    {
+      "epoch": 2.2554054054054054,
+      "grad_norm": 0.3098600506782532,
+      "learning_rate": 0.00048035596310297125,
+      "loss": 0.9060146808624268,
+      "step": 1669
+    },
+    {
+      "epoch": 2.2567567567567566,
+      "grad_norm": 0.38050058484077454,
+      "learning_rate": 0.0004801492459645024,
+      "loss": 0.9354044795036316,
+      "step": 1670
+    },
+    {
+      "epoch": 2.2581081081081082,
+      "grad_norm": 0.3906327784061432,
+      "learning_rate": 0.0004799414918094347,
+      "loss": 0.9285128712654114,
+      "step": 1671
+    },
+    {
+      "epoch": 2.2594594594594595,
+      "grad_norm": 0.3413814902305603,
+      "learning_rate": 0.00047973270157387605,
+      "loss": 0.9065117835998535,
+      "step": 1672
+    },
+    {
+      "epoch": 2.2608108108108107,
+      "grad_norm": 0.32039451599121094,
+      "learning_rate": 0.00047952287619860273,
+      "loss": 0.8982491493225098,
+      "step": 1673
+    },
+    {
+      "epoch": 2.262162162162162,
+      "grad_norm": 0.34785887598991394,
+      "learning_rate": 0.00047931201662905503,
+      "loss": 0.9304678440093994,
+      "step": 1674
+    },
+    {
+      "epoch": 2.2635135135135136,
+      "grad_norm": 0.2917078733444214,
+      "learning_rate": 0.0004791001238153334,
+      "loss": 0.9002468585968018,
+      "step": 1675
+    },
+    {
+      "epoch": 2.264864864864865,
+      "grad_norm": 0.3557056486606598,
+      "learning_rate": 0.00047888719871219367,
+      "loss": 0.9311120510101318,
+      "step": 1676
+    },
+    {
+      "epoch": 2.266216216216216,
+      "grad_norm": 0.3395766317844391,
+      "learning_rate": 0.00047867324227904317,
+      "loss": 0.9243441820144653,
+      "step": 1677
+    },
+    {
+      "epoch": 2.2675675675675677,
+      "grad_norm": 0.3567759692668915,
+      "learning_rate": 0.00047845825547993627,
+      "loss": 0.9488730430603027,
+      "step": 1678
+    },
+    {
+      "epoch": 2.268918918918919,
+      "grad_norm": 0.38428065180778503,
+      "learning_rate": 0.00047824223928356993,
+      "loss": 0.9495503306388855,
+      "step": 1679
+    },
+    {
+      "epoch": 2.27027027027027,
+      "grad_norm": 0.3573961555957794,
+      "learning_rate": 0.00047802519466327945,
+      "loss": 0.9175553321838379,
+      "step": 1680
+    },
+    {
+      "epoch": 2.2716216216216214,
+      "grad_norm": 0.34645822644233704,
+      "learning_rate": 0.00047780712259703394,
+      "loss": 0.9254859685897827,
+      "step": 1681
+    },
+    {
+      "epoch": 2.272972972972973,
+      "grad_norm": 0.38542744517326355,
+      "learning_rate": 0.00047758802406743217,
+      "loss": 0.9182718992233276,
+      "step": 1682
+    },
+    {
+      "epoch": 2.2743243243243243,
+      "grad_norm": 0.33082377910614014,
+      "learning_rate": 0.00047736790006169794,
+      "loss": 0.9165303111076355,
+      "step": 1683
+    },
+    {
+      "epoch": 2.2756756756756755,
+      "grad_norm": 0.33867764472961426,
+      "learning_rate": 0.00047714675157167573,
+      "loss": 0.9407244920730591,
+      "step": 1684
+    },
+    {
+      "epoch": 2.277027027027027,
+      "grad_norm": 0.31942471861839294,
+      "learning_rate": 0.00047692457959382605,
+      "loss": 0.9103283286094666,
+      "step": 1685
+    },
+    {
+      "epoch": 2.2783783783783784,
+      "grad_norm": 0.26950445771217346,
+      "learning_rate": 0.0004767013851292212,
+      "loss": 0.9082165360450745,
+      "step": 1686
+    },
+    {
+      "epoch": 2.2797297297297296,
+      "grad_norm": 0.34477338194847107,
+      "learning_rate": 0.00047647716918354066,
+      "loss": 0.9097682237625122,
+      "step": 1687
+    },
+    {
+      "epoch": 2.281081081081081,
+      "grad_norm": 0.37116265296936035,
+      "learning_rate": 0.0004762519327670664,
+      "loss": 0.9388307332992554,
+      "step": 1688
+    },
+    {
+      "epoch": 2.2824324324324325,
+      "grad_norm": 0.30505824089050293,
+      "learning_rate": 0.0004760256768946787,
+      "loss": 0.8967719674110413,
+      "step": 1689
+    },
+    {
+      "epoch": 2.2837837837837838,
+      "grad_norm": 0.35884320735931396,
+      "learning_rate": 0.0004757984025858511,
+      "loss": 0.9290690422058105,
+      "step": 1690
+    },
+    {
+      "epoch": 2.285135135135135,
+      "grad_norm": 0.2918107509613037,
+      "learning_rate": 0.0004755701108646463,
+      "loss": 0.9260126352310181,
+      "step": 1691
+    },
+    {
+      "epoch": 2.2864864864864867,
+      "grad_norm": 0.30274856090545654,
+      "learning_rate": 0.0004753408027597111,
+      "loss": 0.8979494571685791,
+      "step": 1692
+    },
+    {
+      "epoch": 2.287837837837838,
+      "grad_norm": 0.3197582960128784,
+      "learning_rate": 0.00047511047930427216,
+      "loss": 0.9019080400466919,
+      "step": 1693
+    },
+    {
+      "epoch": 2.289189189189189,
+      "grad_norm": 0.3476748466491699,
+      "learning_rate": 0.0004748791415361309,
+      "loss": 0.9286764860153198,
+      "step": 1694
+    },
+    {
+      "epoch": 2.2905405405405403,
+      "grad_norm": 0.33521780371665955,
+      "learning_rate": 0.00047464679049765926,
+      "loss": 0.9146069884300232,
+      "step": 1695
+    },
+    {
+      "epoch": 2.291891891891892,
+      "grad_norm": 0.37693366408348083,
+      "learning_rate": 0.0004744134272357948,
+      "loss": 0.9236754179000854,
+      "step": 1696
+    },
+    {
+      "epoch": 2.2932432432432432,
+      "grad_norm": 0.29464125633239746,
+      "learning_rate": 0.00047417905280203594,
+      "loss": 0.897982656955719,
+      "step": 1697
+    },
+    {
+      "epoch": 2.2945945945945945,
+      "grad_norm": 0.3195595443248749,
+      "learning_rate": 0.0004739436682524373,
+      "loss": 0.9152045845985413,
+      "step": 1698
+    },
+    {
+      "epoch": 2.295945945945946,
+      "grad_norm": 0.3429083228111267,
+      "learning_rate": 0.0004737072746476051,
+      "loss": 0.9065593481063843,
+      "step": 1699
+    },
+    {
+      "epoch": 2.2972972972972974,
+      "grad_norm": 0.3302845358848572,
+      "learning_rate": 0.00047346987305269184,
+      "loss": 0.9153554439544678,
+      "step": 1700
+    },
+    {
+      "epoch": 2.2986486486486486,
+      "grad_norm": 0.31589290499687195,
+      "learning_rate": 0.0004732314645373921,
+      "loss": 0.9035638570785522,
+      "step": 1701
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.3098660111427307,
+      "learning_rate": 0.0004729920501759376,
+      "loss": 0.9114104509353638,
+      "step": 1702
+    },
+    {
+      "epoch": 2.3013513513513515,
+      "grad_norm": 0.3265870213508606,
+      "learning_rate": 0.00047275163104709196,
+      "loss": 0.9137980937957764,
+      "step": 1703
+    },
+    {
+      "epoch": 2.3027027027027027,
+      "grad_norm": 0.3700502812862396,
+      "learning_rate": 0.0004725102082341464,
+      "loss": 0.9279295802116394,
+      "step": 1704
+    },
+    {
+      "epoch": 2.304054054054054,
+      "grad_norm": 0.28586456179618835,
+      "learning_rate": 0.0004722677828249142,
+      "loss": 0.8954101800918579,
+      "step": 1705
+    },
+    {
+      "epoch": 2.3054054054054056,
+      "grad_norm": 0.3701188266277313,
+      "learning_rate": 0.00047202435591172677,
+      "loss": 0.938246488571167,
+      "step": 1706
+    },
+    {
+      "epoch": 2.306756756756757,
+      "grad_norm": 0.3272833228111267,
+      "learning_rate": 0.0004717799285914276,
+      "loss": 0.9278768301010132,
+      "step": 1707
+    },
+    {
+      "epoch": 2.308108108108108,
+      "grad_norm": 0.3114069402217865,
+      "learning_rate": 0.00047153450196536816,
+      "loss": 0.9017521739006042,
+      "step": 1708
+    },
+    {
+      "epoch": 2.3094594594594593,
+      "grad_norm": 0.31694746017456055,
+      "learning_rate": 0.00047128807713940244,
+      "loss": 0.9085736870765686,
+      "step": 1709
+    },
+    {
+      "epoch": 2.310810810810811,
+      "grad_norm": 0.39185434579849243,
+      "learning_rate": 0.0004710406552238823,
+      "loss": 0.9382093548774719,
+      "step": 1710
+    },
+    {
+      "epoch": 2.312162162162162,
+      "grad_norm": 0.38037556409835815,
+      "learning_rate": 0.00047079223733365234,
+      "loss": 0.9231253862380981,
+      "step": 1711
+    },
+    {
+      "epoch": 2.3135135135135134,
+      "grad_norm": 0.30652523040771484,
+      "learning_rate": 0.00047054282458804477,
+      "loss": 0.9053410291671753,
+      "step": 1712
+    },
+    {
+      "epoch": 2.314864864864865,
+      "grad_norm": 0.29258033633232117,
+      "learning_rate": 0.0004702924181108745,
+      "loss": 0.9089968204498291,
+      "step": 1713
+    },
+    {
+      "epoch": 2.3162162162162163,
+      "grad_norm": 0.3233333230018616,
+      "learning_rate": 0.00047004101903043416,
+      "loss": 0.92583167552948,
+      "step": 1714
+    },
+    {
+      "epoch": 2.3175675675675675,
+      "grad_norm": 0.3245082497596741,
+      "learning_rate": 0.0004697886284794887,
+      "loss": 0.9102175235748291,
+      "step": 1715
+    },
+    {
+      "epoch": 2.3189189189189188,
+      "grad_norm": 0.32355520129203796,
+      "learning_rate": 0.00046953524759527055,
+      "loss": 0.9176565408706665,
+      "step": 1716
+    },
+    {
+      "epoch": 2.3202702702702704,
+      "grad_norm": 0.3228060007095337,
+      "learning_rate": 0.00046928087751947444,
+      "loss": 0.9269182682037354,
+      "step": 1717
+    },
+    {
+      "epoch": 2.3216216216216217,
+      "grad_norm": 0.37442123889923096,
+      "learning_rate": 0.00046902551939825236,
+      "loss": 0.9330718517303467,
+      "step": 1718
+    },
+    {
+      "epoch": 2.322972972972973,
+      "grad_norm": 0.322543740272522,
+      "learning_rate": 0.000468769174382208,
+      "loss": 0.9118773937225342,
+      "step": 1719
+    },
+    {
+      "epoch": 2.3243243243243246,
+      "grad_norm": 0.3613939881324768,
+      "learning_rate": 0.00046851184362639223,
+      "loss": 0.953698456287384,
+      "step": 1720
+    },
+    {
+      "epoch": 2.325675675675676,
+      "grad_norm": 0.30839696526527405,
+      "learning_rate": 0.00046825352829029705,
+      "loss": 0.9090805053710938,
+      "step": 1721
+    },
+    {
+      "epoch": 2.327027027027027,
+      "grad_norm": 0.30680790543556213,
+      "learning_rate": 0.00046799422953785124,
+      "loss": 0.8966723680496216,
+      "step": 1722
+    },
+    {
+      "epoch": 2.3283783783783782,
+      "grad_norm": 0.3211236298084259,
+      "learning_rate": 0.0004677339485374145,
+      "loss": 0.915775716304779,
+      "step": 1723
+    },
+    {
+      "epoch": 2.32972972972973,
+      "grad_norm": 0.3260893225669861,
+      "learning_rate": 0.0004674726864617723,
+      "loss": 0.9303094148635864,
+      "step": 1724
+    },
+    {
+      "epoch": 2.331081081081081,
+      "grad_norm": 0.4222151041030884,
+      "learning_rate": 0.000467210444488131,
+      "loss": 0.9572118520736694,
+      "step": 1725
+    },
+    {
+      "epoch": 2.3324324324324324,
+      "grad_norm": 0.26325348019599915,
+      "learning_rate": 0.0004669472237981118,
+      "loss": 0.8875695466995239,
+      "step": 1726
+    },
+    {
+      "epoch": 2.3337837837837836,
+      "grad_norm": 0.3274972438812256,
+      "learning_rate": 0.0004666830255777462,
+      "loss": 0.9032777547836304,
+      "step": 1727
+    },
+    {
+      "epoch": 2.3351351351351353,
+      "grad_norm": 0.3099897801876068,
+      "learning_rate": 0.0004664178510174702,
+      "loss": 0.926103413105011,
+      "step": 1728
+    },
+    {
+      "epoch": 2.3364864864864865,
+      "grad_norm": 0.3148963749408722,
+      "learning_rate": 0.0004661517013121189,
+      "loss": 0.9164823293685913,
+      "step": 1729
+    },
+    {
+      "epoch": 2.3378378378378377,
+      "grad_norm": 0.3705877363681793,
+      "learning_rate": 0.00046588457766092134,
+      "loss": 0.9058269262313843,
+      "step": 1730
+    },
+    {
+      "epoch": 2.339189189189189,
+      "grad_norm": 0.34459686279296875,
+      "learning_rate": 0.0004656164812674951,
+      "loss": 0.91461181640625,
+      "step": 1731
+    },
+    {
+      "epoch": 2.3405405405405406,
+      "grad_norm": 0.30181315541267395,
+      "learning_rate": 0.0004653474133398408,
+      "loss": 0.9060423970222473,
+      "step": 1732
+    },
+    {
+      "epoch": 2.341891891891892,
+      "grad_norm": 0.38559773564338684,
+      "learning_rate": 0.0004650773750903363,
+      "loss": 0.9496493339538574,
+      "step": 1733
+    },
+    {
+      "epoch": 2.343243243243243,
+      "grad_norm": 0.32058143615722656,
+      "learning_rate": 0.0004648063677357319,
+      "loss": 0.906748354434967,
+      "step": 1734
+    },
+    {
+      "epoch": 2.3445945945945947,
+      "grad_norm": 0.3274504840373993,
+      "learning_rate": 0.00046453439249714466,
+      "loss": 0.9222854971885681,
+      "step": 1735
+    },
+    {
+      "epoch": 2.345945945945946,
+      "grad_norm": 0.29297369718551636,
+      "learning_rate": 0.0004642614506000523,
+      "loss": 0.8992709517478943,
+      "step": 1736
+    },
+    {
+      "epoch": 2.347297297297297,
+      "grad_norm": 0.30893847346305847,
+      "learning_rate": 0.0004639875432742886,
+      "loss": 0.9049811959266663,
+      "step": 1737
+    },
+    {
+      "epoch": 2.3486486486486484,
+      "grad_norm": 0.3543180823326111,
+      "learning_rate": 0.00046371267175403724,
+      "loss": 0.9291818141937256,
+      "step": 1738
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.3835239112377167,
+      "learning_rate": 0.00046343683727782635,
+      "loss": 0.9106743931770325,
+      "step": 1739
+    },
+    {
+      "epoch": 2.3513513513513513,
+      "grad_norm": 0.3083159923553467,
+      "learning_rate": 0.00046316004108852305,
+      "loss": 0.9100226163864136,
+      "step": 1740
+    },
+    {
+      "epoch": 2.3527027027027025,
+      "grad_norm": 0.3212449550628662,
+      "learning_rate": 0.0004628822844333278,
+      "loss": 0.9132744073867798,
+      "step": 1741
+    },
+    {
+      "epoch": 2.354054054054054,
+      "grad_norm": 0.4184679388999939,
+      "learning_rate": 0.00046260356856376884,
+      "loss": 0.933731198310852,
+      "step": 1742
+    },
+    {
+      "epoch": 2.3554054054054054,
+      "grad_norm": 0.3128317594528198,
+      "learning_rate": 0.00046232389473569623,
+      "loss": 0.9154446125030518,
+      "step": 1743
+    },
+    {
+      "epoch": 2.3567567567567567,
+      "grad_norm": 0.39707282185554504,
+      "learning_rate": 0.0004620432642092768,
+      "loss": 0.9393559694290161,
+      "step": 1744
+    },
+    {
+      "epoch": 2.358108108108108,
+      "grad_norm": 0.3394894301891327,
+      "learning_rate": 0.0004617616782489877,
+      "loss": 0.9157607555389404,
+      "step": 1745
+    },
+    {
+      "epoch": 2.3594594594594596,
+      "grad_norm": 0.30203190445899963,
+      "learning_rate": 0.00046147913812361155,
+      "loss": 0.8983393907546997,
+      "step": 1746
+    },
+    {
+      "epoch": 2.360810810810811,
+      "grad_norm": 0.37543177604675293,
+      "learning_rate": 0.00046119564510623,
+      "loss": 0.9291898012161255,
+      "step": 1747
+    },
+    {
+      "epoch": 2.362162162162162,
+      "grad_norm": 0.3701061010360718,
+      "learning_rate": 0.0004609112004742183,
+      "loss": 0.9418013691902161,
+      "step": 1748
+    },
+    {
+      "epoch": 2.3635135135135137,
+      "grad_norm": 0.3433281183242798,
+      "learning_rate": 0.0004606258055092397,
+      "loss": 0.9315159916877747,
+      "step": 1749
+    },
+    {
+      "epoch": 2.364864864864865,
+      "grad_norm": 0.3163180351257324,
+      "learning_rate": 0.0004603394614972393,
+      "loss": 0.9107265472412109,
+      "step": 1750
+    },
+    {
+      "epoch": 2.366216216216216,
+      "grad_norm": 0.3158497214317322,
+      "learning_rate": 0.0004600521697284386,
+      "loss": 0.8959883451461792,
+      "step": 1751
+    },
+    {
+      "epoch": 2.3675675675675674,
+      "grad_norm": 0.3892405331134796,
+      "learning_rate": 0.00045976393149732943,
+      "loss": 0.930243968963623,
+      "step": 1752
+    },
+    {
+      "epoch": 2.368918918918919,
+      "grad_norm": 0.3878495693206787,
+      "learning_rate": 0.00045947474810266844,
+      "loss": 0.9386195540428162,
+      "step": 1753
+    },
+    {
+      "epoch": 2.3702702702702703,
+      "grad_norm": 0.3212476968765259,
+      "learning_rate": 0.0004591846208474707,
+      "loss": 0.9179015159606934,
+      "step": 1754
+    },
+    {
+      "epoch": 2.3716216216216215,
+      "grad_norm": 0.33385151624679565,
+      "learning_rate": 0.0004588935510390045,
+      "loss": 0.9288769960403442,
+      "step": 1755
+    },
+    {
+      "epoch": 2.372972972972973,
+      "grad_norm": 0.33685922622680664,
+      "learning_rate": 0.0004586015399887849,
+      "loss": 0.9184322357177734,
+      "step": 1756
+    },
+    {
+      "epoch": 2.3743243243243244,
+      "grad_norm": 0.42046046257019043,
+      "learning_rate": 0.0004583085890125682,
+      "loss": 0.9540711045265198,
+      "step": 1757
+    },
+    {
+      "epoch": 2.3756756756756756,
+      "grad_norm": 0.36294475197792053,
+      "learning_rate": 0.0004580146994303458,
+      "loss": 0.9246475100517273,
+      "step": 1758
+    },
+    {
+      "epoch": 2.377027027027027,
+      "grad_norm": 0.42263883352279663,
+      "learning_rate": 0.0004577198725663383,
+      "loss": 0.9464613795280457,
+      "step": 1759
+    },
+    {
+      "epoch": 2.3783783783783785,
+      "grad_norm": 0.34650424122810364,
+      "learning_rate": 0.00045742410974898947,
+      "loss": 0.9315037131309509,
+      "step": 1760
+    },
+    {
+      "epoch": 2.3797297297297297,
+      "grad_norm": 0.36014875769615173,
+      "learning_rate": 0.00045712741231096054,
+      "loss": 0.9226959347724915,
+      "step": 1761
+    },
+    {
+      "epoch": 2.381081081081081,
+      "grad_norm": 0.27163341641426086,
+      "learning_rate": 0.00045682978158912384,
+      "loss": 0.892431914806366,
+      "step": 1762
+    },
+    {
+      "epoch": 2.3824324324324326,
+      "grad_norm": 0.3678762912750244,
+      "learning_rate": 0.000456531218924557,
+      "loss": 0.9165188074111938,
+      "step": 1763
+    },
+    {
+      "epoch": 2.383783783783784,
+      "grad_norm": 0.3195040225982666,
+      "learning_rate": 0.00045623172566253676,
+      "loss": 0.9061254262924194,
+      "step": 1764
+    },
+    {
+      "epoch": 2.385135135135135,
+      "grad_norm": 0.2903846502304077,
+      "learning_rate": 0.00045593130315253305,
+      "loss": 0.8938634395599365,
+      "step": 1765
+    },
+    {
+      "epoch": 2.3864864864864863,
+      "grad_norm": 0.3916817307472229,
+      "learning_rate": 0.00045562995274820285,
+      "loss": 0.9287132024765015,
+      "step": 1766
+    },
+    {
+      "epoch": 2.387837837837838,
+      "grad_norm": 0.3391890823841095,
+      "learning_rate": 0.0004553276758073841,
+      "loss": 0.914532482624054,
+      "step": 1767
+    },
+    {
+      "epoch": 2.389189189189189,
+      "grad_norm": 0.38853707909584045,
+      "learning_rate": 0.00045502447369208957,
+      "loss": 0.9409056901931763,
+      "step": 1768
+    },
+    {
+      "epoch": 2.3905405405405404,
+      "grad_norm": 0.3280102610588074,
+      "learning_rate": 0.00045472034776850045,
+      "loss": 0.9200882315635681,
+      "step": 1769
+    },
+    {
+      "epoch": 2.391891891891892,
+      "grad_norm": 0.3334932327270508,
+      "learning_rate": 0.00045441529940696104,
+      "loss": 0.915769100189209,
+      "step": 1770
+    },
+    {
+      "epoch": 2.3932432432432433,
+      "grad_norm": 0.3352791666984558,
+      "learning_rate": 0.0004541093299819714,
+      "loss": 0.9235093593597412,
+      "step": 1771
+    },
+    {
+      "epoch": 2.3945945945945946,
+      "grad_norm": 0.34381210803985596,
+      "learning_rate": 0.00045380244087218224,
+      "loss": 0.9357631206512451,
+      "step": 1772
+    },
+    {
+      "epoch": 2.395945945945946,
+      "grad_norm": 0.3809168338775635,
+      "learning_rate": 0.0004534946334603879,
+      "loss": 0.9138437509536743,
+      "step": 1773
+    },
+    {
+      "epoch": 2.3972972972972975,
+      "grad_norm": 0.3562634587287903,
+      "learning_rate": 0.0004531859091335205,
+      "loss": 0.9072569608688354,
+      "step": 1774
+    },
+    {
+      "epoch": 2.3986486486486487,
+      "grad_norm": 0.3716280162334442,
+      "learning_rate": 0.0004528762692826439,
+      "loss": 0.932267427444458,
+      "step": 1775
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.3389752209186554,
+      "learning_rate": 0.00045256571530294664,
+      "loss": 0.9019883871078491,
+      "step": 1776
+    },
+    {
+      "epoch": 2.4013513513513516,
+      "grad_norm": 0.37173810601234436,
+      "learning_rate": 0.0004522542485937369,
+      "loss": 0.9130491018295288,
+      "step": 1777
+    },
+    {
+      "epoch": 2.402702702702703,
+      "grad_norm": 0.33092325925827026,
+      "learning_rate": 0.0004519418705584348,
+      "loss": 0.922726571559906,
+      "step": 1778
+    },
+    {
+      "epoch": 2.404054054054054,
+      "grad_norm": 0.35117998719215393,
+      "learning_rate": 0.00045162858260456705,
+      "loss": 0.9368631839752197,
+      "step": 1779
+    },
+    {
+      "epoch": 2.4054054054054053,
+      "grad_norm": 0.3560093641281128,
+      "learning_rate": 0.0004513143861437605,
+      "loss": 0.9077574014663696,
+      "step": 1780
+    },
+    {
+      "epoch": 2.406756756756757,
+      "grad_norm": 0.37582510709762573,
+      "learning_rate": 0.00045099928259173516,
+      "loss": 0.9100562334060669,
+      "step": 1781
+    },
+    {
+      "epoch": 2.408108108108108,
+      "grad_norm": 0.3646012842655182,
+      "learning_rate": 0.0004506832733682986,
+      "loss": 0.9070653915405273,
+      "step": 1782
+    },
+    {
+      "epoch": 2.4094594594594594,
+      "grad_norm": 0.4374639093875885,
+      "learning_rate": 0.00045036635989733904,
+      "loss": 0.939071536064148,
+      "step": 1783
+    },
+    {
+      "epoch": 2.410810810810811,
+      "grad_norm": 0.3041325509548187,
+      "learning_rate": 0.0004500485436068191,
+      "loss": 0.9253084063529968,
+      "step": 1784
+    },
+    {
+      "epoch": 2.4121621621621623,
+      "grad_norm": 0.3316110074520111,
+      "learning_rate": 0.0004497298259287696,
+      "loss": 0.9122630953788757,
+      "step": 1785
+    },
+    {
+      "epoch": 2.4135135135135135,
+      "grad_norm": 0.33688560128211975,
+      "learning_rate": 0.00044941020829928247,
+      "loss": 0.9078733325004578,
+      "step": 1786
+    },
+    {
+      "epoch": 2.4148648648648647,
+      "grad_norm": 0.3086908757686615,
+      "learning_rate": 0.00044908969215850495,
+      "loss": 0.9110085964202881,
+      "step": 1787
+    },
+    {
+      "epoch": 2.4162162162162164,
+      "grad_norm": 0.316185861825943,
+      "learning_rate": 0.00044876827895063277,
+      "loss": 0.9066693186759949,
+      "step": 1788
+    },
+    {
+      "epoch": 2.4175675675675676,
+      "grad_norm": 0.40117064118385315,
+      "learning_rate": 0.00044844597012390374,
+      "loss": 0.9343667030334473,
+      "step": 1789
+    },
+    {
+      "epoch": 2.418918918918919,
+      "grad_norm": 0.32079702615737915,
+      "learning_rate": 0.00044812276713059106,
+      "loss": 0.9178594350814819,
+      "step": 1790
+    },
+    {
+      "epoch": 2.42027027027027,
+      "grad_norm": 0.3315507769584656,
+      "learning_rate": 0.00044779867142699713,
+      "loss": 0.9141867160797119,
+      "step": 1791
+    },
+    {
+      "epoch": 2.4216216216216218,
+      "grad_norm": 0.3214103579521179,
+      "learning_rate": 0.0004474736844734465,
+      "loss": 0.9035558700561523,
+      "step": 1792
+    },
+    {
+      "epoch": 2.422972972972973,
+      "grad_norm": 0.3800666332244873,
+      "learning_rate": 0.00044714780773427975,
+      "loss": 0.9251661896705627,
+      "step": 1793
+    },
+    {
+      "epoch": 2.424324324324324,
+      "grad_norm": 0.3060801029205322,
+      "learning_rate": 0.00044682104267784674,
+      "loss": 0.9024914503097534,
+      "step": 1794
+    },
+    {
+      "epoch": 2.4256756756756754,
+      "grad_norm": 0.3356783092021942,
+      "learning_rate": 0.0004464933907764996,
+      "loss": 0.9190641045570374,
+      "step": 1795
+    },
+    {
+      "epoch": 2.427027027027027,
+      "grad_norm": 0.3306639492511749,
+      "learning_rate": 0.0004461648535065869,
+      "loss": 0.9133388996124268,
+      "step": 1796
+    },
+    {
+      "epoch": 2.4283783783783783,
+      "grad_norm": 0.3445473611354828,
+      "learning_rate": 0.00044583543234844616,
+      "loss": 0.9062624573707581,
+      "step": 1797
+    },
+    {
+      "epoch": 2.4297297297297296,
+      "grad_norm": 0.41438028216362,
+      "learning_rate": 0.00044550512878639784,
+      "loss": 0.9398696422576904,
+      "step": 1798
+    },
+    {
+      "epoch": 2.4310810810810812,
+      "grad_norm": 0.33941322565078735,
+      "learning_rate": 0.0004451739443087381,
+      "loss": 0.9248002767562866,
+      "step": 1799
+    },
+    {
+      "epoch": 2.4324324324324325,
+      "grad_norm": 0.31220605969429016,
+      "learning_rate": 0.0004448418804077328,
+      "loss": 0.8930379748344421,
+      "step": 1800
+    },
+    {
+      "epoch": 2.4337837837837837,
+      "grad_norm": 0.3186095654964447,
+      "learning_rate": 0.0004445089385796099,
+      "loss": 0.9105190634727478,
+      "step": 1801
+    },
+    {
+      "epoch": 2.435135135135135,
+      "grad_norm": 0.39297720789909363,
+      "learning_rate": 0.0004441751203245533,
+      "loss": 0.9215264320373535,
+      "step": 1802
+    },
+    {
+      "epoch": 2.4364864864864866,
+      "grad_norm": 0.29654544591903687,
+      "learning_rate": 0.00044384042714669614,
+      "loss": 0.8983317613601685,
+      "step": 1803
+    },
+    {
+      "epoch": 2.437837837837838,
+      "grad_norm": 0.35407447814941406,
+      "learning_rate": 0.00044350486055411354,
+      "loss": 0.9090149402618408,
+      "step": 1804
+    },
+    {
+      "epoch": 2.439189189189189,
+      "grad_norm": 0.3191983699798584,
+      "learning_rate": 0.00044316842205881625,
+      "loss": 0.9002283811569214,
+      "step": 1805
+    },
+    {
+      "epoch": 2.4405405405405407,
+      "grad_norm": 0.35441893339157104,
+      "learning_rate": 0.0004428311131767437,
+      "loss": 0.9118874073028564,
+      "step": 1806
+    },
+    {
+      "epoch": 2.441891891891892,
+      "grad_norm": 0.3604761064052582,
+      "learning_rate": 0.000442492935427757,
+      "loss": 0.9305335283279419,
+      "step": 1807
+    },
+    {
+      "epoch": 2.443243243243243,
+      "grad_norm": 0.3187948167324066,
+      "learning_rate": 0.00044215389033563235,
+      "loss": 0.9124356508255005,
+      "step": 1808
+    },
+    {
+      "epoch": 2.4445945945945944,
+      "grad_norm": 0.31979942321777344,
+      "learning_rate": 0.0004418139794280541,
+      "loss": 0.8940816521644592,
+      "step": 1809
+    },
+    {
+      "epoch": 2.445945945945946,
+      "grad_norm": 0.30196139216423035,
+      "learning_rate": 0.0004414732042366078,
+      "loss": 0.9048377275466919,
+      "step": 1810
+    },
+    {
+      "epoch": 2.4472972972972973,
+      "grad_norm": 0.3922233283519745,
+      "learning_rate": 0.00044113156629677313,
+      "loss": 0.9223603010177612,
+      "step": 1811
+    },
+    {
+      "epoch": 2.4486486486486485,
+      "grad_norm": 0.3320827782154083,
+      "learning_rate": 0.00044078906714791757,
+      "loss": 0.9030029773712158,
+      "step": 1812
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.31748247146606445,
+      "learning_rate": 0.00044044570833328865,
+      "loss": 0.9164584279060364,
+      "step": 1813
+    },
+    {
+      "epoch": 2.4513513513513514,
+      "grad_norm": 0.31214892864227295,
+      "learning_rate": 0.0004401014914000078,
+      "loss": 0.9096285104751587,
+      "step": 1814
+    },
+    {
+      "epoch": 2.4527027027027026,
+      "grad_norm": 0.3224717974662781,
+      "learning_rate": 0.0004397564178990626,
+      "loss": 0.9044045209884644,
+      "step": 1815
+    },
+    {
+      "epoch": 2.454054054054054,
+      "grad_norm": 0.34854814410209656,
+      "learning_rate": 0.0004394104893853007,
+      "loss": 0.9265289902687073,
+      "step": 1816
+    },
+    {
+      "epoch": 2.4554054054054055,
+      "grad_norm": 0.282867968082428,
+      "learning_rate": 0.00043906370741742185,
+      "loss": 0.8990044593811035,
+      "step": 1817
+    },
+    {
+      "epoch": 2.4567567567567568,
+      "grad_norm": 0.36195534467697144,
+      "learning_rate": 0.0004387160735579715,
+      "loss": 0.9095673561096191,
+      "step": 1818
+    },
+    {
+      "epoch": 2.458108108108108,
+      "grad_norm": 0.2779538035392761,
+      "learning_rate": 0.00043836758937333366,
+      "loss": 0.8876782655715942,
+      "step": 1819
+    },
+    {
+      "epoch": 2.4594594594594597,
+      "grad_norm": 0.31564223766326904,
+      "learning_rate": 0.00043801825643372363,
+      "loss": 0.9097212553024292,
+      "step": 1820
+    },
+    {
+      "epoch": 2.460810810810811,
+      "grad_norm": 0.27222245931625366,
+      "learning_rate": 0.0004376680763131811,
+      "loss": 0.8901150226593018,
+      "step": 1821
+    },
+    {
+      "epoch": 2.462162162162162,
+      "grad_norm": 0.34121614694595337,
+      "learning_rate": 0.0004373170505895632,
+      "loss": 0.9018301963806152,
+      "step": 1822
+    },
+    {
+      "epoch": 2.4635135135135133,
+      "grad_norm": 0.3282203674316406,
+      "learning_rate": 0.000436965180844537,
+      "loss": 0.9116355180740356,
+      "step": 1823
+    },
+    {
+      "epoch": 2.464864864864865,
+      "grad_norm": 0.32740673422813416,
+      "learning_rate": 0.0004366124686635727,
+      "loss": 0.9055653214454651,
+      "step": 1824
+    },
+    {
+      "epoch": 2.4662162162162162,
+      "grad_norm": 0.33954450488090515,
+      "learning_rate": 0.0004362589156359363,
+      "loss": 0.9306285381317139,
+      "step": 1825
+    },
+    {
+      "epoch": 2.4675675675675675,
+      "grad_norm": 0.3767278790473938,
+      "learning_rate": 0.00043590452335468265,
+      "loss": 0.9141724705696106,
+      "step": 1826
+    },
+    {
+      "epoch": 2.468918918918919,
+      "grad_norm": 0.33788931369781494,
+      "learning_rate": 0.0004355492934166481,
+      "loss": 0.9194344282150269,
+      "step": 1827
+    },
+    {
+      "epoch": 2.4702702702702704,
+      "grad_norm": 0.46907883882522583,
+      "learning_rate": 0.0004351932274224434,
+      "loss": 0.9251453876495361,
+      "step": 1828
+    },
+    {
+      "epoch": 2.4716216216216216,
+      "grad_norm": 0.3207901418209076,
+      "learning_rate": 0.00043483632697644616,
+      "loss": 0.907422661781311,
+      "step": 1829
+    },
+    {
+      "epoch": 2.472972972972973,
+      "grad_norm": 0.30374690890312195,
+      "learning_rate": 0.0004344785936867942,
+      "loss": 0.9288627505302429,
+      "step": 1830
+    },
+    {
+      "epoch": 2.4743243243243245,
+      "grad_norm": 0.28772252798080444,
+      "learning_rate": 0.0004341200291653781,
+      "loss": 0.9080641865730286,
+      "step": 1831
+    },
+    {
+      "epoch": 2.4756756756756757,
+      "grad_norm": 0.4409027397632599,
+      "learning_rate": 0.00043376063502783337,
+      "loss": 0.9183487892150879,
+      "step": 1832
+    },
+    {
+      "epoch": 2.477027027027027,
+      "grad_norm": 0.3563745319843292,
+      "learning_rate": 0.00043340041289353416,
+      "loss": 0.8854151964187622,
+      "step": 1833
+    },
+    {
+      "epoch": 2.4783783783783786,
+      "grad_norm": 0.29331278800964355,
+      "learning_rate": 0.0004330393643855851,
+      "loss": 0.901701807975769,
+      "step": 1834
+    },
+    {
+      "epoch": 2.47972972972973,
+      "grad_norm": 0.3330759108066559,
+      "learning_rate": 0.0004326774911308145,
+      "loss": 0.911975622177124,
+      "step": 1835
+    },
+    {
+      "epoch": 2.481081081081081,
+      "grad_norm": 0.3731631636619568,
+      "learning_rate": 0.0004323147947597667,
+      "loss": 0.922588586807251,
+      "step": 1836
+    },
+    {
+      "epoch": 2.4824324324324323,
+      "grad_norm": 0.3133493661880493,
+      "learning_rate": 0.00043195127690669486,
+      "loss": 0.9000201225280762,
+      "step": 1837
+    },
+    {
+      "epoch": 2.483783783783784,
+      "grad_norm": 0.311592698097229,
+      "learning_rate": 0.0004315869392095542,
+      "loss": 0.9020200967788696,
+      "step": 1838
+    },
+    {
+      "epoch": 2.485135135135135,
+      "grad_norm": 0.3450436592102051,
+      "learning_rate": 0.00043122178330999296,
+      "loss": 0.9040917158126831,
+      "step": 1839
+    },
+    {
+      "epoch": 2.4864864864864864,
+      "grad_norm": 0.348956823348999,
+      "learning_rate": 0.0004308558108533471,
+      "loss": 0.9117119312286377,
+      "step": 1840
+    },
+    {
+      "epoch": 2.487837837837838,
+      "grad_norm": 0.35695570707321167,
+      "learning_rate": 0.0004304890234886311,
+      "loss": 0.9165434241294861,
+      "step": 1841
+    },
+    {
+      "epoch": 2.4891891891891893,
+      "grad_norm": 0.33306819200515747,
+      "learning_rate": 0.00043012142286853185,
+      "loss": 0.913783848285675,
+      "step": 1842
+    },
+    {
+      "epoch": 2.4905405405405405,
+      "grad_norm": 0.37133845686912537,
+      "learning_rate": 0.00042975301064940026,
+      "loss": 0.9088679552078247,
+      "step": 1843
+    },
+    {
+      "epoch": 2.4918918918918918,
+      "grad_norm": 0.3155823349952698,
+      "learning_rate": 0.0004293837884912444,
+      "loss": 0.8989685773849487,
+      "step": 1844
+    },
+    {
+      "epoch": 2.4932432432432434,
+      "grad_norm": 0.2957735061645508,
+      "learning_rate": 0.0004290137580577216,
+      "loss": 0.8976813554763794,
+      "step": 1845
+    },
+    {
+      "epoch": 2.4945945945945946,
+      "grad_norm": 0.344051331281662,
+      "learning_rate": 0.00042864292101613133,
+      "loss": 0.8990542888641357,
+      "step": 1846
+    },
+    {
+      "epoch": 2.495945945945946,
+      "grad_norm": 0.2935975193977356,
+      "learning_rate": 0.00042827127903740747,
+      "loss": 0.9082565307617188,
+      "step": 1847
+    },
+    {
+      "epoch": 2.4972972972972975,
+      "grad_norm": 0.3455594778060913,
+      "learning_rate": 0.00042789883379611084,
+      "loss": 0.9107598066329956,
+      "step": 1848
+    },
+    {
+      "epoch": 2.4986486486486488,
+      "grad_norm": 0.33428218960762024,
+      "learning_rate": 0.00042752558697042143,
+      "loss": 0.9114682674407959,
+      "step": 1849
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.3454268276691437,
+      "learning_rate": 0.00042715154024213143,
+      "loss": 0.9275224208831787,
+      "step": 1850
+    },
+    {
+      "epoch": 2.5013513513513512,
+      "grad_norm": 0.3415561020374298,
+      "learning_rate": 0.00042677669529663686,
+      "loss": 0.9130823016166687,
+      "step": 1851
+    },
+    {
+      "epoch": 2.5027027027027025,
+      "grad_norm": 0.371301531791687,
+      "learning_rate": 0.00042640105382293073,
+      "loss": 0.9181870222091675,
+      "step": 1852
+    },
+    {
+      "epoch": 2.504054054054054,
+      "grad_norm": 0.35587483644485474,
+      "learning_rate": 0.0004260246175135948,
+      "loss": 0.8941422700881958,
+      "step": 1853
+    },
+    {
+      "epoch": 2.5054054054054054,
+      "grad_norm": 0.3208431005477905,
+      "learning_rate": 0.0004256473880647923,
+      "loss": 0.9251483678817749,
+      "step": 1854
+    },
+    {
+      "epoch": 2.506756756756757,
+      "grad_norm": 0.31760627031326294,
+      "learning_rate": 0.00042526936717626046,
+      "loss": 0.9042402505874634,
+      "step": 1855
+    },
+    {
+      "epoch": 2.5081081081081082,
+      "grad_norm": 0.39646992087364197,
+      "learning_rate": 0.00042489055655130226,
+      "loss": 0.9175405502319336,
+      "step": 1856
+    },
+    {
+      "epoch": 2.5094594594594595,
+      "grad_norm": 0.25203636288642883,
+      "learning_rate": 0.00042451095789677943,
+      "loss": 0.8818904757499695,
+      "step": 1857
+    },
+    {
+      "epoch": 2.5108108108108107,
+      "grad_norm": 0.31227004528045654,
+      "learning_rate": 0.0004241305729231042,
+      "loss": 0.9077086448669434,
+      "step": 1858
+    },
+    {
+      "epoch": 2.512162162162162,
+      "grad_norm": 0.38696590065956116,
+      "learning_rate": 0.00042374940334423194,
+      "loss": 0.9238603115081787,
+      "step": 1859
+    },
+    {
+      "epoch": 2.5135135135135136,
+      "grad_norm": 0.3161331117153168,
+      "learning_rate": 0.0004233674508776533,
+      "loss": 0.8994921445846558,
+      "step": 1860
+    },
+    {
+      "epoch": 2.514864864864865,
+      "grad_norm": 0.29693105816841125,
+      "learning_rate": 0.0004229847172443866,
+      "loss": 0.9123968482017517,
+      "step": 1861
+    },
+    {
+      "epoch": 2.516216216216216,
+      "grad_norm": 0.2906714975833893,
+      "learning_rate": 0.00042260120416896975,
+      "loss": 0.9204038381576538,
+      "step": 1862
+    },
+    {
+      "epoch": 2.5175675675675677,
+      "grad_norm": 0.3538146913051605,
+      "learning_rate": 0.00042221691337945285,
+      "loss": 0.9084588289260864,
+      "step": 1863
+    },
+    {
+      "epoch": 2.518918918918919,
+      "grad_norm": 0.33873820304870605,
+      "learning_rate": 0.00042183184660739027,
+      "loss": 0.9049211144447327,
+      "step": 1864
+    },
+    {
+      "epoch": 2.52027027027027,
+      "grad_norm": 0.30460262298583984,
+      "learning_rate": 0.00042144600558783284,
+      "loss": 0.8979172706604004,
+      "step": 1865
+    },
+    {
+      "epoch": 2.5216216216216214,
+      "grad_norm": 0.3352408707141876,
+      "learning_rate": 0.0004210593920593201,
+      "loss": 0.9137951731681824,
+      "step": 1866
+    },
+    {
+      "epoch": 2.522972972972973,
+      "grad_norm": 0.315025269985199,
+      "learning_rate": 0.00042067200776387215,
+      "loss": 0.8913535475730896,
+      "step": 1867
+    },
+    {
+      "epoch": 2.5243243243243243,
+      "grad_norm": 0.33961451053619385,
+      "learning_rate": 0.0004202838544469822,
+      "loss": 0.9053914546966553,
+      "step": 1868
+    },
+    {
+      "epoch": 2.5256756756756755,
+      "grad_norm": 0.32805007696151733,
+      "learning_rate": 0.0004198949338576086,
+      "loss": 0.9341400861740112,
+      "step": 1869
+    },
+    {
+      "epoch": 2.527027027027027,
+      "grad_norm": 0.3084182143211365,
+      "learning_rate": 0.0004195052477481669,
+      "loss": 0.9133095741271973,
+      "step": 1870
+    },
+    {
+      "epoch": 2.5283783783783784,
+      "grad_norm": 0.2541082799434662,
+      "learning_rate": 0.00041911479787452177,
+      "loss": 0.8836497068405151,
+      "step": 1871
+    },
+    {
+      "epoch": 2.5297297297297296,
+      "grad_norm": 0.3700069189071655,
+      "learning_rate": 0.00041872358599597947,
+      "loss": 0.9051740169525146,
+      "step": 1872
+    },
+    {
+      "epoch": 2.531081081081081,
+      "grad_norm": 0.3845379650592804,
+      "learning_rate": 0.00041833161387527985,
+      "loss": 0.8970038294792175,
+      "step": 1873
+    },
+    {
+      "epoch": 2.5324324324324325,
+      "grad_norm": 0.32960745692253113,
+      "learning_rate": 0.000417938883278588,
+      "loss": 0.9126001000404358,
+      "step": 1874
+    },
+    {
+      "epoch": 2.5337837837837838,
+      "grad_norm": 0.33994802832603455,
+      "learning_rate": 0.0004175453959754867,
+      "loss": 0.9104715585708618,
+      "step": 1875
+    },
+    {
+      "epoch": 2.535135135135135,
+      "grad_norm": 0.39296218752861023,
+      "learning_rate": 0.0004171511537389684,
+      "loss": 0.9066174626350403,
+      "step": 1876
+    },
+    {
+      "epoch": 2.5364864864864867,
+      "grad_norm": 0.3189007639884949,
+      "learning_rate": 0.00041675615834542716,
+      "loss": 0.9140348434448242,
+      "step": 1877
+    },
+    {
+      "epoch": 2.537837837837838,
+      "grad_norm": 0.3002886474132538,
+      "learning_rate": 0.00041636041157465056,
+      "loss": 0.8959788680076599,
+      "step": 1878
+    },
+    {
+      "epoch": 2.539189189189189,
+      "grad_norm": 0.336841881275177,
+      "learning_rate": 0.0004159639152098118,
+      "loss": 0.9172726273536682,
+      "step": 1879
+    },
+    {
+      "epoch": 2.5405405405405403,
+      "grad_norm": 0.329268217086792,
+      "learning_rate": 0.00041556667103746157,
+      "loss": 0.8932222127914429,
+      "step": 1880
+    },
+    {
+      "epoch": 2.541891891891892,
+      "grad_norm": 0.339205801486969,
+      "learning_rate": 0.00041516868084752034,
+      "loss": 0.9075334072113037,
+      "step": 1881
+    },
+    {
+      "epoch": 2.5432432432432432,
+      "grad_norm": 0.3498866856098175,
+      "learning_rate": 0.0004147699464332697,
+      "loss": 0.9209542274475098,
+      "step": 1882
+    },
+    {
+      "epoch": 2.5445945945945945,
+      "grad_norm": 0.3231043815612793,
+      "learning_rate": 0.0004143704695913447,
+      "loss": 0.9138814210891724,
+      "step": 1883
+    },
+    {
+      "epoch": 2.545945945945946,
+      "grad_norm": 0.3205627202987671,
+      "learning_rate": 0.00041397025212172573,
+      "loss": 0.9093181490898132,
+      "step": 1884
+    },
+    {
+      "epoch": 2.5472972972972974,
+      "grad_norm": 0.3205457329750061,
+      "learning_rate": 0.00041356929582773023,
+      "loss": 0.9168918132781982,
+      "step": 1885
+    },
+    {
+      "epoch": 2.5486486486486486,
+      "grad_norm": 0.27052485942840576,
+      "learning_rate": 0.00041316760251600474,
+      "loss": 0.8908618092536926,
+      "step": 1886
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3007594645023346,
+      "learning_rate": 0.00041276517399651657,
+      "loss": 0.9030815362930298,
+      "step": 1887
+    },
+    {
+      "epoch": 2.5513513513513515,
+      "grad_norm": 0.3323644995689392,
+      "learning_rate": 0.0004123620120825459,
+      "loss": 0.9093780517578125,
+      "step": 1888
+    },
+    {
+      "epoch": 2.5527027027027027,
+      "grad_norm": 0.3109865188598633,
+      "learning_rate": 0.00041195811859067756,
+      "loss": 0.922133207321167,
+      "step": 1889
+    },
+    {
+      "epoch": 2.554054054054054,
+      "grad_norm": 0.32004645466804504,
+      "learning_rate": 0.00041155349534079236,
+      "loss": 0.9058867692947388,
+      "step": 1890
+    },
+    {
+      "epoch": 2.5554054054054056,
+      "grad_norm": 0.3767712712287903,
+      "learning_rate": 0.00041114814415605977,
+      "loss": 0.9266032576560974,
+      "step": 1891
+    },
+    {
+      "epoch": 2.556756756756757,
+      "grad_norm": 0.30888187885284424,
+      "learning_rate": 0.0004107420668629289,
+      "loss": 0.9149340987205505,
+      "step": 1892
+    },
+    {
+      "epoch": 2.558108108108108,
+      "grad_norm": 0.2616349160671234,
+      "learning_rate": 0.0004103352652911206,
+      "loss": 0.889664888381958,
+      "step": 1893
+    },
+    {
+      "epoch": 2.5594594594594593,
+      "grad_norm": 0.41275152564048767,
+      "learning_rate": 0.0004099277412736195,
+      "loss": 0.9393388032913208,
+      "step": 1894
+    },
+    {
+      "epoch": 2.560810810810811,
+      "grad_norm": 0.36410650610923767,
+      "learning_rate": 0.00040951949664666504,
+      "loss": 0.9038483500480652,
+      "step": 1895
+    },
+    {
+      "epoch": 2.562162162162162,
+      "grad_norm": 0.278879851102829,
+      "learning_rate": 0.0004091105332497439,
+      "loss": 0.8783587217330933,
+      "step": 1896
+    },
+    {
+      "epoch": 2.5635135135135134,
+      "grad_norm": 0.32042476534843445,
+      "learning_rate": 0.00040870085292558147,
+      "loss": 0.8977522253990173,
+      "step": 1897
+    },
+    {
+      "epoch": 2.564864864864865,
+      "grad_norm": 0.3134475648403168,
+      "learning_rate": 0.00040829045752013317,
+      "loss": 0.9061712622642517,
+      "step": 1898
+    },
+    {
+      "epoch": 2.5662162162162163,
+      "grad_norm": 0.3538665771484375,
+      "learning_rate": 0.00040787934888257673,
+      "loss": 0.9273746013641357,
+      "step": 1899
+    },
+    {
+      "epoch": 2.5675675675675675,
+      "grad_norm": 0.3068298101425171,
+      "learning_rate": 0.0004074675288653037,
+      "loss": 0.8993312120437622,
+      "step": 1900
+    },
+    {
+      "epoch": 2.5689189189189188,
+      "grad_norm": 0.24515053629875183,
+      "learning_rate": 0.0004070549993239106,
+      "loss": 0.8859050273895264,
+      "step": 1901
+    },
+    {
+      "epoch": 2.57027027027027,
+      "grad_norm": 0.3863736093044281,
+      "learning_rate": 0.00040664176211719136,
+      "loss": 0.9223416447639465,
+      "step": 1902
+    },
+    {
+      "epoch": 2.5716216216216217,
+      "grad_norm": 0.3091454803943634,
+      "learning_rate": 0.00040622781910712826,
+      "loss": 0.8968448638916016,
+      "step": 1903
+    },
+    {
+      "epoch": 2.572972972972973,
+      "grad_norm": 0.3157728314399719,
+      "learning_rate": 0.00040581317215888403,
+      "loss": 0.9163667559623718,
+      "step": 1904
+    },
+    {
+      "epoch": 2.5743243243243246,
+      "grad_norm": 0.34956344962120056,
+      "learning_rate": 0.00040539782314079304,
+      "loss": 0.9104989767074585,
+      "step": 1905
+    },
+    {
+      "epoch": 2.575675675675676,
+      "grad_norm": 0.44148772954940796,
+      "learning_rate": 0.0004049817739243532,
+      "loss": 0.9378393888473511,
+      "step": 1906
+    },
+    {
+      "epoch": 2.577027027027027,
+      "grad_norm": 0.2844523787498474,
+      "learning_rate": 0.0004045650263842174,
+      "loss": 0.9052444100379944,
+      "step": 1907
+    },
+    {
+      "epoch": 2.5783783783783782,
+      "grad_norm": 0.35212212800979614,
+      "learning_rate": 0.00040414758239818506,
+      "loss": 0.9261338710784912,
+      "step": 1908
+    },
+    {
+      "epoch": 2.5797297297297295,
+      "grad_norm": 0.3568432033061981,
+      "learning_rate": 0.0004037294438471936,
+      "loss": 0.9275521039962769,
+      "step": 1909
+    },
+    {
+      "epoch": 2.581081081081081,
+      "grad_norm": 0.3275481164455414,
+      "learning_rate": 0.00040331061261531014,
+      "loss": 0.8953363299369812,
+      "step": 1910
+    },
+    {
+      "epoch": 2.5824324324324324,
+      "grad_norm": 0.28254714608192444,
+      "learning_rate": 0.00040289109058972285,
+      "loss": 0.9058210849761963,
+      "step": 1911
+    },
+    {
+      "epoch": 2.583783783783784,
+      "grad_norm": 0.35519900918006897,
+      "learning_rate": 0.00040247087966073253,
+      "loss": 0.9371414184570312,
+      "step": 1912
+    },
+    {
+      "epoch": 2.5851351351351353,
+      "grad_norm": 0.2617085576057434,
+      "learning_rate": 0.0004020499817217441,
+      "loss": 0.8850512504577637,
+      "step": 1913
+    },
+    {
+      "epoch": 2.5864864864864865,
+      "grad_norm": 0.3483463227748871,
+      "learning_rate": 0.00040162839866925804,
+      "loss": 0.9146143198013306,
+      "step": 1914
+    },
+    {
+      "epoch": 2.5878378378378377,
+      "grad_norm": 0.3090210556983948,
+      "learning_rate": 0.0004012061324028617,
+      "loss": 0.906050443649292,
+      "step": 1915
+    },
+    {
+      "epoch": 2.589189189189189,
+      "grad_norm": 0.2953965961933136,
+      "learning_rate": 0.00040078318482522114,
+      "loss": 0.8817096948623657,
+      "step": 1916
+    },
+    {
+      "epoch": 2.5905405405405406,
+      "grad_norm": 0.3506219685077667,
+      "learning_rate": 0.000400359557842072,
+      "loss": 0.8899403214454651,
+      "step": 1917
+    },
+    {
+      "epoch": 2.591891891891892,
+      "grad_norm": 0.3649691641330719,
+      "learning_rate": 0.0003999352533622116,
+      "loss": 0.9380967617034912,
+      "step": 1918
+    },
+    {
+      "epoch": 2.5932432432432435,
+      "grad_norm": 0.2937057316303253,
+      "learning_rate": 0.00039951027329748957,
+      "loss": 0.8942040801048279,
+      "step": 1919
+    },
+    {
+      "epoch": 2.5945945945945947,
+      "grad_norm": 0.4693893492221832,
+      "learning_rate": 0.0003990846195627998,
+      "loss": 0.9250284433364868,
+      "step": 1920
+    },
+    {
+      "epoch": 2.595945945945946,
+      "grad_norm": 0.44072529673576355,
+      "learning_rate": 0.0003986582940760717,
+      "loss": 0.9275575876235962,
+      "step": 1921
+    },
+    {
+      "epoch": 2.597297297297297,
+      "grad_norm": 0.30094221234321594,
+      "learning_rate": 0.00039823129875826127,
+      "loss": 0.9112452268600464,
+      "step": 1922
+    },
+    {
+      "epoch": 2.5986486486486484,
+      "grad_norm": 0.3642895519733429,
+      "learning_rate": 0.000397803635533343,
+      "loss": 0.9000362157821655,
+      "step": 1923
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.35251396894454956,
+      "learning_rate": 0.00039737530632830045,
+      "loss": 0.8967292308807373,
+      "step": 1924
+    },
+    {
+      "epoch": 2.6013513513513513,
+      "grad_norm": 0.3252716362476349,
+      "learning_rate": 0.0003969463130731183,
+      "loss": 0.9046024680137634,
+      "step": 1925
+    },
+    {
+      "epoch": 2.6027027027027025,
+      "grad_norm": 0.3025946617126465,
+      "learning_rate": 0.00039651665770077326,
+      "loss": 0.8906595706939697,
+      "step": 1926
+    },
+    {
+      "epoch": 2.604054054054054,
+      "grad_norm": 0.3056081235408783,
+      "learning_rate": 0.0003960863421472254,
+      "loss": 0.876272439956665,
+      "step": 1927
+    },
+    {
+      "epoch": 2.6054054054054054,
+      "grad_norm": 0.3112778961658478,
+      "learning_rate": 0.00039565536835140934,
+      "loss": 0.8961069583892822,
+      "step": 1928
+    },
+    {
+      "epoch": 2.6067567567567567,
+      "grad_norm": 0.32194626331329346,
+      "learning_rate": 0.000395223738255226,
+      "loss": 0.9010473489761353,
+      "step": 1929
+    },
+    {
+      "epoch": 2.608108108108108,
+      "grad_norm": 0.309995174407959,
+      "learning_rate": 0.00039479145380353313,
+      "loss": 0.9140994548797607,
+      "step": 1930
+    },
+    {
+      "epoch": 2.6094594594594596,
+      "grad_norm": 0.2675793170928955,
+      "learning_rate": 0.00039435851694413705,
+      "loss": 0.8948551416397095,
+      "step": 1931
+    },
+    {
+      "epoch": 2.610810810810811,
+      "grad_norm": 0.3240041732788086,
+      "learning_rate": 0.0003939249296277837,
+      "loss": 0.9059362411499023,
+      "step": 1932
+    },
+    {
+      "epoch": 2.612162162162162,
+      "grad_norm": 0.2921745181083679,
+      "learning_rate": 0.0003934906938081499,
+      "loss": 0.9050780534744263,
+      "step": 1933
+    },
+    {
+      "epoch": 2.6135135135135137,
+      "grad_norm": 0.359138548374176,
+      "learning_rate": 0.0003930558114418345,
+      "loss": 0.8930717706680298,
+      "step": 1934
+    },
+    {
+      "epoch": 2.614864864864865,
+      "grad_norm": 0.24719971418380737,
+      "learning_rate": 0.00039262028448834964,
+      "loss": 0.8832869529724121,
+      "step": 1935
+    },
+    {
+      "epoch": 2.616216216216216,
+      "grad_norm": 0.30986717343330383,
+      "learning_rate": 0.00039218411491011176,
+      "loss": 0.9088811874389648,
+      "step": 1936
+    },
+    {
+      "epoch": 2.6175675675675674,
+      "grad_norm": 0.30103668570518494,
+      "learning_rate": 0.0003917473046724329,
+      "loss": 0.9012068510055542,
+      "step": 1937
+    },
+    {
+      "epoch": 2.618918918918919,
+      "grad_norm": 0.2950650155544281,
+      "learning_rate": 0.0003913098557435121,
+      "loss": 0.9081825017929077,
+      "step": 1938
+    },
+    {
+      "epoch": 2.6202702702702703,
+      "grad_norm": 0.33387863636016846,
+      "learning_rate": 0.00039087177009442567,
+      "loss": 0.8982729911804199,
+      "step": 1939
+    },
+    {
+      "epoch": 2.6216216216216215,
+      "grad_norm": 0.30417221784591675,
+      "learning_rate": 0.0003904330496991194,
+      "loss": 0.8878965377807617,
+      "step": 1940
+    },
+    {
+      "epoch": 2.622972972972973,
+      "grad_norm": 0.31744638085365295,
+      "learning_rate": 0.0003899936965343989,
+      "loss": 0.9052223563194275,
+      "step": 1941
+    },
+    {
+      "epoch": 2.6243243243243244,
+      "grad_norm": 0.2724461257457733,
+      "learning_rate": 0.00038955371257992096,
+      "loss": 0.8975890874862671,
+      "step": 1942
+    },
+    {
+      "epoch": 2.6256756756756756,
+      "grad_norm": 0.3003993034362793,
+      "learning_rate": 0.00038911309981818466,
+      "loss": 0.9078341722488403,
+      "step": 1943
+    },
+    {
+      "epoch": 2.627027027027027,
+      "grad_norm": 0.30752235651016235,
+      "learning_rate": 0.0003886718602345224,
+      "loss": 0.8958864212036133,
+      "step": 1944
+    },
+    {
+      "epoch": 2.6283783783783785,
+      "grad_norm": 0.33105114102363586,
+      "learning_rate": 0.00038822999581709087,
+      "loss": 0.9142142534255981,
+      "step": 1945
+    },
+    {
+      "epoch": 2.6297297297297297,
+      "grad_norm": 0.30755773186683655,
+      "learning_rate": 0.0003877875085568622,
+      "loss": 0.916291356086731,
+      "step": 1946
+    },
+    {
+      "epoch": 2.631081081081081,
+      "grad_norm": 0.3556656837463379,
+      "learning_rate": 0.00038734440044761503,
+      "loss": 0.9339132308959961,
+      "step": 1947
+    },
+    {
+      "epoch": 2.6324324324324326,
+      "grad_norm": 0.27147001028060913,
+      "learning_rate": 0.0003869006734859253,
+      "loss": 0.8926278948783875,
+      "step": 1948
+    },
+    {
+      "epoch": 2.633783783783784,
+      "grad_norm": 0.312071830034256,
+      "learning_rate": 0.00038645632967115753,
+      "loss": 0.9075770974159241,
+      "step": 1949
+    },
+    {
+      "epoch": 2.635135135135135,
+      "grad_norm": 0.3144897520542145,
+      "learning_rate": 0.0003860113710054556,
+      "loss": 0.9099981784820557,
+      "step": 1950
+    },
+    {
+      "epoch": 2.6364864864864863,
+      "grad_norm": 0.33791449666023254,
+      "learning_rate": 0.00038556579949373384,
+      "loss": 0.9206700325012207,
+      "step": 1951
+    },
+    {
+      "epoch": 2.637837837837838,
+      "grad_norm": 0.2925969362258911,
+      "learning_rate": 0.0003851196171436679,
+      "loss": 0.8914316296577454,
+      "step": 1952
+    },
+    {
+      "epoch": 2.639189189189189,
+      "grad_norm": 0.30055493116378784,
+      "learning_rate": 0.000384672825965686,
+      "loss": 0.8882830142974854,
+      "step": 1953
+    },
+    {
+      "epoch": 2.6405405405405404,
+      "grad_norm": 0.30671098828315735,
+      "learning_rate": 0.00038422542797295935,
+      "loss": 0.92866051197052,
+      "step": 1954
+    },
+    {
+      "epoch": 2.641891891891892,
+      "grad_norm": 0.29150334000587463,
+      "learning_rate": 0.0003837774251813936,
+      "loss": 0.8973989486694336,
+      "step": 1955
+    },
+    {
+      "epoch": 2.6432432432432433,
+      "grad_norm": 0.3392993211746216,
+      "learning_rate": 0.00038332881960961943,
+      "loss": 0.908284068107605,
+      "step": 1956
+    },
+    {
+      "epoch": 2.6445945945945946,
+      "grad_norm": 0.27096229791641235,
+      "learning_rate": 0.00038287961327898346,
+      "loss": 0.887292206287384,
+      "step": 1957
+    },
+    {
+      "epoch": 2.645945945945946,
+      "grad_norm": 0.2600380480289459,
+      "learning_rate": 0.00038242980821353954,
+      "loss": 0.8813022971153259,
+      "step": 1958
+    },
+    {
+      "epoch": 2.6472972972972975,
+      "grad_norm": 0.27098140120506287,
+      "learning_rate": 0.0003819794064400389,
+      "loss": 0.8846792578697205,
+      "step": 1959
+    },
+    {
+      "epoch": 2.6486486486486487,
+      "grad_norm": 0.4023469090461731,
+      "learning_rate": 0.0003815284099879218,
+      "loss": 0.9124071598052979,
+      "step": 1960
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.3262810707092285,
+      "learning_rate": 0.00038107682088930797,
+      "loss": 0.9032478332519531,
+      "step": 1961
+    },
+    {
+      "epoch": 2.6513513513513516,
+      "grad_norm": 0.3078111708164215,
+      "learning_rate": 0.0003806246411789872,
+      "loss": 0.8998095989227295,
+      "step": 1962
+    },
+    {
+      "epoch": 2.652702702702703,
+      "grad_norm": 0.3305057883262634,
+      "learning_rate": 0.000380171872894411,
+      "loss": 0.9137475490570068,
+      "step": 1963
+    },
+    {
+      "epoch": 2.654054054054054,
+      "grad_norm": 0.2994779646396637,
+      "learning_rate": 0.00037971851807568237,
+      "loss": 0.9115272164344788,
+      "step": 1964
+    },
+    {
+      "epoch": 2.6554054054054053,
+      "grad_norm": 0.3253689706325531,
+      "learning_rate": 0.0003792645787655476,
+      "loss": 0.8932115435600281,
+      "step": 1965
+    },
+    {
+      "epoch": 2.6567567567567565,
+      "grad_norm": 0.3404305875301361,
+      "learning_rate": 0.0003788100570093863,
+      "loss": 0.8966022729873657,
+      "step": 1966
+    },
+    {
+      "epoch": 2.658108108108108,
+      "grad_norm": 0.29030871391296387,
+      "learning_rate": 0.0003783549548552027,
+      "loss": 0.8939262628555298,
+      "step": 1967
+    },
+    {
+      "epoch": 2.6594594594594594,
+      "grad_norm": 0.286202073097229,
+      "learning_rate": 0.0003778992743536159,
+      "loss": 0.8968703150749207,
+      "step": 1968
+    },
+    {
+      "epoch": 2.660810810810811,
+      "grad_norm": 0.3451937139034271,
+      "learning_rate": 0.00037744301755785137,
+      "loss": 0.9221668243408203,
+      "step": 1969
+    },
+    {
+      "epoch": 2.6621621621621623,
+      "grad_norm": 0.26072317361831665,
+      "learning_rate": 0.000376986186523731,
+      "loss": 0.8977985382080078,
+      "step": 1970
+    },
+    {
+      "epoch": 2.6635135135135135,
+      "grad_norm": 0.3268652558326721,
+      "learning_rate": 0.0003765287833096641,
+      "loss": 0.9147913455963135,
+      "step": 1971
+    },
+    {
+      "epoch": 2.6648648648648647,
+      "grad_norm": 0.2946639955043793,
+      "learning_rate": 0.0003760708099766382,
+      "loss": 0.8838262557983398,
+      "step": 1972
+    },
+    {
+      "epoch": 2.666216216216216,
+      "grad_norm": 0.2908424437046051,
+      "learning_rate": 0.0003756122685882098,
+      "loss": 0.9067938923835754,
+      "step": 1973
+    },
+    {
+      "epoch": 2.6675675675675676,
+      "grad_norm": 0.3317592144012451,
+      "learning_rate": 0.0003751531612104948,
+      "loss": 0.9181103706359863,
+      "step": 1974
+    },
+    {
+      "epoch": 2.668918918918919,
+      "grad_norm": 0.27463945746421814,
+      "learning_rate": 0.00037469348991215934,
+      "loss": 0.8860207796096802,
+      "step": 1975
+    },
+    {
+      "epoch": 2.6702702702702705,
+      "grad_norm": 0.3083774447441101,
+      "learning_rate": 0.00037423325676441064,
+      "loss": 0.9060210585594177,
+      "step": 1976
+    },
+    {
+      "epoch": 2.6716216216216218,
+      "grad_norm": 0.30701640248298645,
+      "learning_rate": 0.0003737724638409876,
+      "loss": 0.8963175415992737,
+      "step": 1977
+    },
+    {
+      "epoch": 2.672972972972973,
+      "grad_norm": 0.30179136991500854,
+      "learning_rate": 0.0003733111132181511,
+      "loss": 0.8974799513816833,
+      "step": 1978
+    },
+    {
+      "epoch": 2.674324324324324,
+      "grad_norm": 0.3115499019622803,
+      "learning_rate": 0.00037284920697467505,
+      "loss": 0.906340479850769,
+      "step": 1979
+    },
+    {
+      "epoch": 2.6756756756756754,
+      "grad_norm": 0.2782456874847412,
+      "learning_rate": 0.0003723867471918371,
+      "loss": 0.8873960971832275,
+      "step": 1980
+    },
+    {
+      "epoch": 2.677027027027027,
+      "grad_norm": 0.3114399015903473,
+      "learning_rate": 0.00037192373595340864,
+      "loss": 0.8955035209655762,
+      "step": 1981
+    },
+    {
+      "epoch": 2.6783783783783783,
+      "grad_norm": 0.3271673619747162,
+      "learning_rate": 0.0003714601753456463,
+      "loss": 0.9069145917892456,
+      "step": 1982
+    },
+    {
+      "epoch": 2.67972972972973,
+      "grad_norm": 0.32321402430534363,
+      "learning_rate": 0.0003709960674572817,
+      "loss": 0.8920638561248779,
+      "step": 1983
+    },
+    {
+      "epoch": 2.6810810810810812,
+      "grad_norm": 0.31119731068611145,
+      "learning_rate": 0.00037053141437951264,
+      "loss": 0.9027121067047119,
+      "step": 1984
+    },
+    {
+      "epoch": 2.6824324324324325,
+      "grad_norm": 0.31413304805755615,
+      "learning_rate": 0.0003700662182059936,
+      "loss": 0.8912751078605652,
+      "step": 1985
+    },
+    {
+      "epoch": 2.6837837837837837,
+      "grad_norm": 0.3564685881137848,
+      "learning_rate": 0.00036960048103282564,
+      "loss": 0.8968368768692017,
+      "step": 1986
+    },
+    {
+      "epoch": 2.685135135135135,
+      "grad_norm": 0.3244015872478485,
+      "learning_rate": 0.00036913420495854793,
+      "loss": 0.9042390584945679,
+      "step": 1987
+    },
+    {
+      "epoch": 2.6864864864864866,
+      "grad_norm": 0.325503945350647,
+      "learning_rate": 0.0003686673920841278,
+      "loss": 0.9101980328559875,
+      "step": 1988
+    },
+    {
+      "epoch": 2.687837837837838,
+      "grad_norm": 0.3087595999240875,
+      "learning_rate": 0.0003682000445129512,
+      "loss": 0.8956350088119507,
+      "step": 1989
+    },
+    {
+      "epoch": 2.689189189189189,
+      "grad_norm": 0.2557029128074646,
+      "learning_rate": 0.00036773216435081335,
+      "loss": 0.8816306591033936,
+      "step": 1990
+    },
+    {
+      "epoch": 2.6905405405405407,
+      "grad_norm": 0.3719809055328369,
+      "learning_rate": 0.00036726375370590926,
+      "loss": 0.8845054507255554,
+      "step": 1991
+    },
+    {
+      "epoch": 2.691891891891892,
+      "grad_norm": 0.3232547640800476,
+      "learning_rate": 0.00036679481468882425,
+      "loss": 0.9000667333602905,
+      "step": 1992
+    },
+    {
+      "epoch": 2.693243243243243,
+      "grad_norm": 0.34377288818359375,
+      "learning_rate": 0.0003663253494125244,
+      "loss": 0.902205228805542,
+      "step": 1993
+    },
+    {
+      "epoch": 2.6945945945945944,
+      "grad_norm": 0.26241326332092285,
+      "learning_rate": 0.00036585535999234697,
+      "loss": 0.8861907124519348,
+      "step": 1994
+    },
+    {
+      "epoch": 2.695945945945946,
+      "grad_norm": 0.30730295181274414,
+      "learning_rate": 0.0003653848485459909,
+      "loss": 0.8876694440841675,
+      "step": 1995
+    },
+    {
+      "epoch": 2.6972972972972973,
+      "grad_norm": 0.2952781915664673,
+      "learning_rate": 0.0003649138171935076,
+      "loss": 0.8840914964675903,
+      "step": 1996
+    },
+    {
+      "epoch": 2.6986486486486485,
+      "grad_norm": 0.29550811648368835,
+      "learning_rate": 0.0003644422680572906,
+      "loss": 0.8962233066558838,
+      "step": 1997
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.3017968237400055,
+      "learning_rate": 0.0003639702032620669,
+      "loss": 0.9069817662239075,
+      "step": 1998
+    },
+    {
+      "epoch": 2.7013513513513514,
+      "grad_norm": 0.29955658316612244,
+      "learning_rate": 0.00036349762493488667,
+      "loss": 0.8895984888076782,
+      "step": 1999
+    },
+    {
+      "epoch": 2.7027027027027026,
+      "grad_norm": 0.29851067066192627,
+      "learning_rate": 0.00036302453520511437,
+      "loss": 0.9061528444290161,
+      "step": 2000
+    },
+    {
+      "epoch": 2.704054054054054,
+      "grad_norm": 0.3488820791244507,
+      "learning_rate": 0.0003625509362044183,
+      "loss": 0.9123108386993408,
+      "step": 2001
+    },
+    {
+      "epoch": 2.7054054054054055,
+      "grad_norm": 0.2549116313457489,
+      "learning_rate": 0.0003620768300667618,
+      "loss": 0.8955531120300293,
+      "step": 2002
+    },
+    {
+      "epoch": 2.7067567567567568,
+      "grad_norm": 0.3269039988517761,
+      "learning_rate": 0.0003616022189283932,
+      "loss": 0.9065597653388977,
+      "step": 2003
+    },
+    {
+      "epoch": 2.708108108108108,
+      "grad_norm": 0.27487996220588684,
+      "learning_rate": 0.00036112710492783605,
+      "loss": 0.885182797908783,
+      "step": 2004
+    },
+    {
+      "epoch": 2.7094594594594597,
+      "grad_norm": 0.314854234457016,
+      "learning_rate": 0.00036065149020588015,
+      "loss": 0.8931908011436462,
+      "step": 2005
+    },
+    {
+      "epoch": 2.710810810810811,
+      "grad_norm": 1.0255366563796997,
+      "learning_rate": 0.0003601753769055711,
+      "loss": 0.9508232474327087,
+      "step": 2006
+    },
+    {
+      "epoch": 2.712162162162162,
+      "grad_norm": 0.269942045211792,
+      "learning_rate": 0.0003596987671722012,
+      "loss": 0.9052224159240723,
+      "step": 2007
+    },
+    {
+      "epoch": 2.7135135135135133,
+      "grad_norm": 0.3443039357662201,
+      "learning_rate": 0.00035922166315329954,
+      "loss": 0.9030207395553589,
+      "step": 2008
+    },
+    {
+      "epoch": 2.714864864864865,
+      "grad_norm": 0.314732164144516,
+      "learning_rate": 0.0003587440669986224,
+      "loss": 0.912821888923645,
+      "step": 2009
+    },
+    {
+      "epoch": 2.7162162162162162,
+      "grad_norm": 0.32170870900154114,
+      "learning_rate": 0.00035826598086014357,
+      "loss": 0.9088935852050781,
+      "step": 2010
+    },
+    {
+      "epoch": 2.7175675675675675,
+      "grad_norm": 0.3373260200023651,
+      "learning_rate": 0.00035778740689204456,
+      "loss": 0.9256114959716797,
+      "step": 2011
+    },
+    {
+      "epoch": 2.718918918918919,
+      "grad_norm": 0.3099922239780426,
+      "learning_rate": 0.0003573083472507051,
+      "loss": 0.8794175386428833,
+      "step": 2012
+    },
+    {
+      "epoch": 2.7202702702702704,
+      "grad_norm": 0.27590784430503845,
+      "learning_rate": 0.0003568288040946931,
+      "loss": 0.8942548036575317,
+      "step": 2013
+    },
+    {
+      "epoch": 2.7216216216216216,
+      "grad_norm": 0.3030194938182831,
+      "learning_rate": 0.00035634877958475535,
+      "loss": 0.8977792263031006,
+      "step": 2014
+    },
+    {
+      "epoch": 2.722972972972973,
+      "grad_norm": 0.2662948668003082,
+      "learning_rate": 0.00035586827588380724,
+      "loss": 0.8871087431907654,
+      "step": 2015
+    },
+    {
+      "epoch": 2.7243243243243245,
+      "grad_norm": 0.27483314275741577,
+      "learning_rate": 0.00035538729515692356,
+      "loss": 0.879793107509613,
+      "step": 2016
+    },
+    {
+      "epoch": 2.7256756756756757,
+      "grad_norm": 0.295896053314209,
+      "learning_rate": 0.0003549058395713285,
+      "loss": 0.8915729522705078,
+      "step": 2017
+    },
+    {
+      "epoch": 2.727027027027027,
+      "grad_norm": 0.2625579535961151,
+      "learning_rate": 0.0003544239112963857,
+      "loss": 0.894089937210083,
+      "step": 2018
+    },
+    {
+      "epoch": 2.7283783783783786,
+      "grad_norm": 0.308627724647522,
+      "learning_rate": 0.00035394151250358886,
+      "loss": 0.8950883150100708,
+      "step": 2019
+    },
+    {
+      "epoch": 2.72972972972973,
+      "grad_norm": 0.2781120836734772,
+      "learning_rate": 0.0003534586453665517,
+      "loss": 0.8923969864845276,
+      "step": 2020
+    },
+    {
+      "epoch": 2.731081081081081,
+      "grad_norm": 0.3309084177017212,
+      "learning_rate": 0.0003529753120609982,
+      "loss": 0.8967127799987793,
+      "step": 2021
+    },
+    {
+      "epoch": 2.7324324324324323,
+      "grad_norm": 0.2957651913166046,
+      "learning_rate": 0.0003524915147647528,
+      "loss": 0.9085147380828857,
+      "step": 2022
+    },
+    {
+      "epoch": 2.733783783783784,
+      "grad_norm": 0.32469162344932556,
+      "learning_rate": 0.0003520072556577306,
+      "loss": 0.9145159721374512,
+      "step": 2023
+    },
+    {
+      "epoch": 2.735135135135135,
+      "grad_norm": 0.30375373363494873,
+      "learning_rate": 0.00035152253692192765,
+      "loss": 0.9057455062866211,
+      "step": 2024
+    },
+    {
+      "epoch": 2.7364864864864864,
+      "grad_norm": 0.3221012353897095,
+      "learning_rate": 0.00035103736074141103,
+      "loss": 0.8848783373832703,
+      "step": 2025
+    },
+    {
+      "epoch": 2.737837837837838,
+      "grad_norm": 0.306880921125412,
+      "learning_rate": 0.0003505517293023088,
+      "loss": 0.8892220854759216,
+      "step": 2026
+    },
+    {
+      "epoch": 2.7391891891891893,
+      "grad_norm": 0.3080769181251526,
+      "learning_rate": 0.0003500656447928005,
+      "loss": 0.9058672189712524,
+      "step": 2027
+    },
+    {
+      "epoch": 2.7405405405405405,
+      "grad_norm": 0.35756972432136536,
+      "learning_rate": 0.00034957910940310716,
+      "loss": 0.9070689678192139,
+      "step": 2028
+    },
+    {
+      "epoch": 2.7418918918918918,
+      "grad_norm": 0.3463623523712158,
+      "learning_rate": 0.0003490921253254813,
+      "loss": 0.9123653769493103,
+      "step": 2029
+    },
+    {
+      "epoch": 2.743243243243243,
+      "grad_norm": 0.29535385966300964,
+      "learning_rate": 0.00034860469475419723,
+      "loss": 0.8981360197067261,
+      "step": 2030
+    },
+    {
+      "epoch": 2.7445945945945946,
+      "grad_norm": 0.3238242268562317,
+      "learning_rate": 0.00034811681988554095,
+      "loss": 0.9063290357589722,
+      "step": 2031
+    },
+    {
+      "epoch": 2.745945945945946,
+      "grad_norm": 0.31268244981765747,
+      "learning_rate": 0.0003476285029178006,
+      "loss": 0.8891018629074097,
+      "step": 2032
+    },
+    {
+      "epoch": 2.7472972972972975,
+      "grad_norm": 0.33000093698501587,
+      "learning_rate": 0.0003471397460512563,
+      "loss": 0.9056144952774048,
+      "step": 2033
+    },
+    {
+      "epoch": 2.7486486486486488,
+      "grad_norm": 0.28237393498420715,
+      "learning_rate": 0.0003466505514881703,
+      "loss": 0.8844913244247437,
+      "step": 2034
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.2983894348144531,
+      "learning_rate": 0.00034616092143277674,
+      "loss": 0.8965655565261841,
+      "step": 2035
+    },
+    {
+      "epoch": 2.7513513513513512,
+      "grad_norm": 0.27541136741638184,
+      "learning_rate": 0.0003456708580912725,
+      "loss": 0.8942146301269531,
+      "step": 2036
+    },
+    {
+      "epoch": 2.7527027027027025,
+      "grad_norm": 0.35277560353279114,
+      "learning_rate": 0.00034518036367180637,
+      "loss": 0.9083061218261719,
+      "step": 2037
+    },
+    {
+      "epoch": 2.754054054054054,
+      "grad_norm": 0.2524873614311218,
+      "learning_rate": 0.0003446894403844698,
+      "loss": 0.890041708946228,
+      "step": 2038
+    },
+    {
+      "epoch": 2.7554054054054054,
+      "grad_norm": 0.29020097851753235,
+      "learning_rate": 0.0003441980904412866,
+      "loss": 0.8794090151786804,
+      "step": 2039
+    },
+    {
+      "epoch": 2.756756756756757,
+      "grad_norm": 0.3138035833835602,
+      "learning_rate": 0.00034370631605620285,
+      "loss": 0.9159804582595825,
+      "step": 2040
+    },
+    {
+      "epoch": 2.7581081081081082,
+      "grad_norm": 0.25865164399147034,
+      "learning_rate": 0.0003432141194450772,
+      "loss": 0.8850780725479126,
+      "step": 2041
+    },
+    {
+      "epoch": 2.7594594594594595,
+      "grad_norm": 0.3289860188961029,
+      "learning_rate": 0.00034272150282567084,
+      "loss": 0.9117475152015686,
+      "step": 2042
+    },
+    {
+      "epoch": 2.7608108108108107,
+      "grad_norm": 0.2783081829547882,
+      "learning_rate": 0.0003422284684176374,
+      "loss": 0.9131172895431519,
+      "step": 2043
+    },
+    {
+      "epoch": 2.762162162162162,
+      "grad_norm": 0.27068454027175903,
+      "learning_rate": 0.00034173501844251305,
+      "loss": 0.888108491897583,
+      "step": 2044
+    },
+    {
+      "epoch": 2.7635135135135136,
+      "grad_norm": 0.2569864094257355,
+      "learning_rate": 0.00034124115512370636,
+      "loss": 0.8832180500030518,
+      "step": 2045
+    },
+    {
+      "epoch": 2.764864864864865,
+      "grad_norm": 0.2993770241737366,
+      "learning_rate": 0.0003407468806864883,
+      "loss": 0.8961409330368042,
+      "step": 2046
+    },
+    {
+      "epoch": 2.766216216216216,
+      "grad_norm": 0.2902892231941223,
+      "learning_rate": 0.0003402521973579826,
+      "loss": 0.8899857997894287,
+      "step": 2047
+    },
+    {
+      "epoch": 2.7675675675675677,
+      "grad_norm": 0.2955883741378784,
+      "learning_rate": 0.00033975710736715504,
+      "loss": 0.9060413837432861,
+      "step": 2048
+    },
+    {
+      "epoch": 2.768918918918919,
+      "grad_norm": 0.24106764793395996,
+      "learning_rate": 0.00033926161294480384,
+      "loss": 0.890876829624176,
+      "step": 2049
+    },
+    {
+      "epoch": 2.77027027027027,
+      "grad_norm": 0.274497389793396,
+      "learning_rate": 0.00033876571632354956,
+      "loss": 0.8796004056930542,
+      "step": 2050
+    },
+    {
+      "epoch": 2.7716216216216214,
+      "grad_norm": 0.2878156304359436,
+      "learning_rate": 0.0003382694197378252,
+      "loss": 0.8955954313278198,
+      "step": 2051
+    },
+    {
+      "epoch": 2.772972972972973,
+      "grad_norm": 0.2751466929912567,
+      "learning_rate": 0.00033777272542386564,
+      "loss": 0.9032543897628784,
+      "step": 2052
+    },
+    {
+      "epoch": 2.7743243243243243,
+      "grad_norm": 0.30835291743278503,
+      "learning_rate": 0.0003372756356196979,
+      "loss": 0.9055225849151611,
+      "step": 2053
+    },
+    {
+      "epoch": 2.7756756756756755,
+      "grad_norm": 0.3433806896209717,
+      "learning_rate": 0.00033677815256513114,
+      "loss": 0.8846802711486816,
+      "step": 2054
+    },
+    {
+      "epoch": 2.777027027027027,
+      "grad_norm": 0.27734121680259705,
+      "learning_rate": 0.0003362802785017464,
+      "loss": 0.8882848024368286,
+      "step": 2055
+    },
+    {
+      "epoch": 2.7783783783783784,
+      "grad_norm": 0.3224279284477234,
+      "learning_rate": 0.0003357820156728866,
+      "loss": 0.8846608996391296,
+      "step": 2056
+    },
+    {
+      "epoch": 2.7797297297297296,
+      "grad_norm": 0.2860287129878998,
+      "learning_rate": 0.00033528336632364624,
+      "loss": 0.8879190683364868,
+      "step": 2057
+    },
+    {
+      "epoch": 2.781081081081081,
+      "grad_norm": 0.34710079431533813,
+      "learning_rate": 0.0003347843327008615,
+      "loss": 0.9077175855636597,
+      "step": 2058
+    },
+    {
+      "epoch": 2.7824324324324325,
+      "grad_norm": 0.2554878294467926,
+      "learning_rate": 0.0003342849170531001,
+      "loss": 0.876301646232605,
+      "step": 2059
+    },
+    {
+      "epoch": 2.7837837837837838,
+      "grad_norm": 0.2859269082546234,
+      "learning_rate": 0.000333785121630651,
+      "loss": 0.8888336420059204,
+      "step": 2060
+    },
+    {
+      "epoch": 2.785135135135135,
+      "grad_norm": 0.2945288121700287,
+      "learning_rate": 0.0003332849486855144,
+      "loss": 0.8917970657348633,
+      "step": 2061
+    },
+    {
+      "epoch": 2.7864864864864867,
+      "grad_norm": 0.2589549422264099,
+      "learning_rate": 0.0003327844004713916,
+      "loss": 0.8733126521110535,
+      "step": 2062
+    },
+    {
+      "epoch": 2.787837837837838,
+      "grad_norm": 0.3589082956314087,
+      "learning_rate": 0.0003322834792436747,
+      "loss": 0.9161995649337769,
+      "step": 2063
+    },
+    {
+      "epoch": 2.789189189189189,
+      "grad_norm": 0.2663860619068146,
+      "learning_rate": 0.00033178218725943666,
+      "loss": 0.8910544514656067,
+      "step": 2064
+    },
+    {
+      "epoch": 2.7905405405405403,
+      "grad_norm": 0.28474897146224976,
+      "learning_rate": 0.0003312805267774209,
+      "loss": 0.8899795413017273,
+      "step": 2065
+    },
+    {
+      "epoch": 2.791891891891892,
+      "grad_norm": 0.3416900634765625,
+      "learning_rate": 0.0003307785000580313,
+      "loss": 0.8873816728591919,
+      "step": 2066
+    },
+    {
+      "epoch": 2.7932432432432432,
+      "grad_norm": 0.2706182599067688,
+      "learning_rate": 0.000330276109363322,
+      "loss": 0.8884077668190002,
+      "step": 2067
+    },
+    {
+      "epoch": 2.7945945945945945,
+      "grad_norm": 0.29542404413223267,
+      "learning_rate": 0.00032977335695698714,
+      "loss": 0.8753232359886169,
+      "step": 2068
+    },
+    {
+      "epoch": 2.795945945945946,
+      "grad_norm": 0.3017449378967285,
+      "learning_rate": 0.00032927024510435055,
+      "loss": 0.8947060108184814,
+      "step": 2069
+    },
+    {
+      "epoch": 2.7972972972972974,
+      "grad_norm": 0.3118552565574646,
+      "learning_rate": 0.00032876677607235566,
+      "loss": 0.910396158695221,
+      "step": 2070
+    },
+    {
+      "epoch": 2.7986486486486486,
+      "grad_norm": 0.3287335932254791,
+      "learning_rate": 0.0003282629521295556,
+      "loss": 0.9103954434394836,
+      "step": 2071
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.2887420058250427,
+      "learning_rate": 0.0003277587755461023,
+      "loss": 0.8984743356704712,
+      "step": 2072
+    },
+    {
+      "epoch": 2.8013513513513515,
+      "grad_norm": 0.3186838924884796,
+      "learning_rate": 0.00032725424859373687,
+      "loss": 0.8951923251152039,
+      "step": 2073
+    },
+    {
+      "epoch": 2.8027027027027027,
+      "grad_norm": 0.2825588881969452,
+      "learning_rate": 0.000326749373545779,
+      "loss": 0.8966405987739563,
+      "step": 2074
+    },
+    {
+      "epoch": 2.804054054054054,
+      "grad_norm": 0.28620630502700806,
+      "learning_rate": 0.00032624415267711694,
+      "loss": 0.8913454413414001,
+      "step": 2075
+    },
+    {
+      "epoch": 2.8054054054054056,
+      "grad_norm": 0.31673333048820496,
+      "learning_rate": 0.0003257385882641971,
+      "loss": 0.9210106134414673,
+      "step": 2076
+    },
+    {
+      "epoch": 2.806756756756757,
+      "grad_norm": 0.29511144757270813,
+      "learning_rate": 0.00032523268258501385,
+      "loss": 0.8919566869735718,
+      "step": 2077
+    },
+    {
+      "epoch": 2.808108108108108,
+      "grad_norm": 0.3063429594039917,
+      "learning_rate": 0.0003247264379190992,
+      "loss": 0.8792030811309814,
+      "step": 2078
+    },
+    {
+      "epoch": 2.8094594594594593,
+      "grad_norm": 0.2539964020252228,
+      "learning_rate": 0.00032421985654751276,
+      "loss": 0.8739627599716187,
+      "step": 2079
+    },
+    {
+      "epoch": 2.810810810810811,
+      "grad_norm": 0.2883901000022888,
+      "learning_rate": 0.0003237129407528311,
+      "loss": 0.9050120115280151,
+      "step": 2080
+    },
+    {
+      "epoch": 2.812162162162162,
+      "grad_norm": 0.2567887306213379,
+      "learning_rate": 0.0003232056928191376,
+      "loss": 0.8891692161560059,
+      "step": 2081
+    },
+    {
+      "epoch": 2.8135135135135134,
+      "grad_norm": 0.295039564371109,
+      "learning_rate": 0.00032269811503201246,
+      "loss": 0.8853375911712646,
+      "step": 2082
+    },
+    {
+      "epoch": 2.814864864864865,
+      "grad_norm": 0.3791702389717102,
+      "learning_rate": 0.000322190209678522,
+      "loss": 0.9231860637664795,
+      "step": 2083
+    },
+    {
+      "epoch": 2.8162162162162163,
+      "grad_norm": 0.29984813928604126,
+      "learning_rate": 0.0003216819790472085,
+      "loss": 0.9060615301132202,
+      "step": 2084
+    },
+    {
+      "epoch": 2.8175675675675675,
+      "grad_norm": 0.2552929222583771,
+      "learning_rate": 0.0003211734254280799,
+      "loss": 0.889758825302124,
+      "step": 2085
+    },
+    {
+      "epoch": 2.8189189189189188,
+      "grad_norm": 0.26178988814353943,
+      "learning_rate": 0.0003206645511125995,
+      "loss": 0.8988333940505981,
+      "step": 2086
+    },
+    {
+      "epoch": 2.82027027027027,
+      "grad_norm": 0.2992238700389862,
+      "learning_rate": 0.00032015535839367544,
+      "loss": 0.8917158842086792,
+      "step": 2087
+    },
+    {
+      "epoch": 2.8216216216216217,
+      "grad_norm": 0.243870809674263,
+      "learning_rate": 0.0003196458495656508,
+      "loss": 0.888929009437561,
+      "step": 2088
+    },
+    {
+      "epoch": 2.822972972972973,
+      "grad_norm": 0.26827332377433777,
+      "learning_rate": 0.0003191360269242928,
+      "loss": 0.8943725228309631,
+      "step": 2089
+    },
+    {
+      "epoch": 2.8243243243243246,
+      "grad_norm": 0.3107629716396332,
+      "learning_rate": 0.00031862589276678276,
+      "loss": 0.9081905484199524,
+      "step": 2090
+    },
+    {
+      "epoch": 2.825675675675676,
+      "grad_norm": 0.3132500946521759,
+      "learning_rate": 0.0003181154493917057,
+      "loss": 0.8958175778388977,
+      "step": 2091
+    },
+    {
+      "epoch": 2.827027027027027,
+      "grad_norm": 0.2778893709182739,
+      "learning_rate": 0.00031760469909903976,
+      "loss": 0.8825944662094116,
+      "step": 2092
+    },
+    {
+      "epoch": 2.8283783783783782,
+      "grad_norm": 0.29173043370246887,
+      "learning_rate": 0.0003170936441901461,
+      "loss": 0.899152398109436,
+      "step": 2093
+    },
+    {
+      "epoch": 2.8297297297297295,
+      "grad_norm": 0.3075809180736542,
+      "learning_rate": 0.00031658228696775835,
+      "loss": 0.9030112028121948,
+      "step": 2094
+    },
+    {
+      "epoch": 2.831081081081081,
+      "grad_norm": 0.3077649176120758,
+      "learning_rate": 0.0003160706297359725,
+      "loss": 0.8934945464134216,
+      "step": 2095
+    },
+    {
+      "epoch": 2.8324324324324324,
+      "grad_norm": 0.2998087406158447,
+      "learning_rate": 0.00031555867480023616,
+      "loss": 0.8962283730506897,
+      "step": 2096
+    },
+    {
+      "epoch": 2.833783783783784,
+      "grad_norm": 0.26584315299987793,
+      "learning_rate": 0.00031504642446733826,
+      "loss": 0.8929603695869446,
+      "step": 2097
+    },
+    {
+      "epoch": 2.8351351351351353,
+      "grad_norm": 0.2816173732280731,
+      "learning_rate": 0.0003145338810453991,
+      "loss": 0.9039076566696167,
+      "step": 2098
+    },
+    {
+      "epoch": 2.8364864864864865,
+      "grad_norm": 0.27676743268966675,
+      "learning_rate": 0.00031402104684385935,
+      "loss": 0.8874030709266663,
+      "step": 2099
+    },
+    {
+      "epoch": 2.8378378378378377,
+      "grad_norm": 0.23759272694587708,
+      "learning_rate": 0.0003135079241734698,
+      "loss": 0.8637881278991699,
+      "step": 2100
+    },
+    {
+      "epoch": 2.839189189189189,
+      "grad_norm": 0.2757917642593384,
+      "learning_rate": 0.0003129945153462813,
+      "loss": 0.9047263860702515,
+      "step": 2101
+    },
+    {
+      "epoch": 2.8405405405405406,
+      "grad_norm": 0.26882481575012207,
+      "learning_rate": 0.00031248082267563385,
+      "loss": 0.8689297437667847,
+      "step": 2102
+    },
+    {
+      "epoch": 2.841891891891892,
+      "grad_norm": 0.2704657316207886,
+      "learning_rate": 0.00031196684847614655,
+      "loss": 0.8811259865760803,
+      "step": 2103
+    },
+    {
+      "epoch": 2.8432432432432435,
+      "grad_norm": 0.2586350739002228,
+      "learning_rate": 0.00031145259506370685,
+      "loss": 0.8780332803726196,
+      "step": 2104
+    },
+    {
+      "epoch": 2.8445945945945947,
+      "grad_norm": 0.25881004333496094,
+      "learning_rate": 0.00031093806475546046,
+      "loss": 0.873753011226654,
+      "step": 2105
+    },
+    {
+      "epoch": 2.845945945945946,
+      "grad_norm": 0.30162373185157776,
+      "learning_rate": 0.00031042325986980064,
+      "loss": 0.8870466947555542,
+      "step": 2106
+    },
+    {
+      "epoch": 2.847297297297297,
+      "grad_norm": 0.2904811501502991,
+      "learning_rate": 0.000309908182726358,
+      "loss": 0.8950209617614746,
+      "step": 2107
+    },
+    {
+      "epoch": 2.8486486486486484,
+      "grad_norm": 0.3141934871673584,
+      "learning_rate": 0.00030939283564598976,
+      "loss": 0.8951435089111328,
+      "step": 2108
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.2910269796848297,
+      "learning_rate": 0.0003088772209507694,
+      "loss": 0.901068925857544,
+      "step": 2109
+    },
+    {
+      "epoch": 2.8513513513513513,
+      "grad_norm": 0.3094770014286041,
+      "learning_rate": 0.0003083613409639764,
+      "loss": 0.916305661201477,
+      "step": 2110
+    },
+    {
+      "epoch": 2.8527027027027025,
+      "grad_norm": 0.26824212074279785,
+      "learning_rate": 0.00030784519801008544,
+      "loss": 0.8764640092849731,
+      "step": 2111
+    },
+    {
+      "epoch": 2.854054054054054,
+      "grad_norm": 0.29381248354911804,
+      "learning_rate": 0.00030732879441475614,
+      "loss": 0.9057637453079224,
+      "step": 2112
+    },
+    {
+      "epoch": 2.8554054054054054,
+      "grad_norm": 0.2525600790977478,
+      "learning_rate": 0.00030681213250482255,
+      "loss": 0.876758873462677,
+      "step": 2113
+    },
+    {
+      "epoch": 2.8567567567567567,
+      "grad_norm": 0.25054264068603516,
+      "learning_rate": 0.0003062952146082826,
+      "loss": 0.8935237526893616,
+      "step": 2114
+    },
+    {
+      "epoch": 2.858108108108108,
+      "grad_norm": 0.2560473382472992,
+      "learning_rate": 0.0003057780430542878,
+      "loss": 0.8924289345741272,
+      "step": 2115
+    },
+    {
+      "epoch": 2.8594594594594596,
+      "grad_norm": 0.283338725566864,
+      "learning_rate": 0.0003052606201731325,
+      "loss": 0.8800677061080933,
+      "step": 2116
+    },
+    {
+      "epoch": 2.860810810810811,
+      "grad_norm": 0.31030112504959106,
+      "learning_rate": 0.0003047429482962433,
+      "loss": 0.9097687005996704,
+      "step": 2117
+    },
+    {
+      "epoch": 2.862162162162162,
+      "grad_norm": 0.29983267188072205,
+      "learning_rate": 0.00030422502975616914,
+      "loss": 0.887913703918457,
+      "step": 2118
+    },
+    {
+      "epoch": 2.8635135135135137,
+      "grad_norm": 0.3344593048095703,
+      "learning_rate": 0.00030370686688657,
+      "loss": 0.9014067053794861,
+      "step": 2119
+    },
+    {
+      "epoch": 2.864864864864865,
+      "grad_norm": 0.30769994854927063,
+      "learning_rate": 0.0003031884620222071,
+      "loss": 0.8888781666755676,
+      "step": 2120
+    },
+    {
+      "epoch": 2.866216216216216,
+      "grad_norm": 0.2764323651790619,
+      "learning_rate": 0.0003026698174989316,
+      "loss": 0.8867925405502319,
+      "step": 2121
+    },
+    {
+      "epoch": 2.8675675675675674,
+      "grad_norm": 0.242692768573761,
+      "learning_rate": 0.0003021509356536751,
+      "loss": 0.8761938214302063,
+      "step": 2122
+    },
+    {
+      "epoch": 2.868918918918919,
+      "grad_norm": 0.2698027789592743,
+      "learning_rate": 0.0003016318188244381,
+      "loss": 0.9020191431045532,
+      "step": 2123
+    },
+    {
+      "epoch": 2.8702702702702703,
+      "grad_norm": 0.2614970803260803,
+      "learning_rate": 0.0003011124693502802,
+      "loss": 0.8902342319488525,
+      "step": 2124
+    },
+    {
+      "epoch": 2.8716216216216215,
+      "grad_norm": 0.24611696600914001,
+      "learning_rate": 0.00030059288957130895,
+      "loss": 0.8826942443847656,
+      "step": 2125
+    },
+    {
+      "epoch": 2.872972972972973,
+      "grad_norm": 0.28727415204048157,
+      "learning_rate": 0.0003000730818286698,
+      "loss": 0.8768051862716675,
+      "step": 2126
+    },
+    {
+      "epoch": 2.8743243243243244,
+      "grad_norm": 0.2899792492389679,
+      "learning_rate": 0.00029955304846453554,
+      "loss": 0.8891726732254028,
+      "step": 2127
+    },
+    {
+      "epoch": 2.8756756756756756,
+      "grad_norm": 0.32515400648117065,
+      "learning_rate": 0.0002990327918220951,
+      "loss": 0.8923490047454834,
+      "step": 2128
+    },
+    {
+      "epoch": 2.877027027027027,
+      "grad_norm": 0.393909752368927,
+      "learning_rate": 0.00029851231424554383,
+      "loss": 0.8921933770179749,
+      "step": 2129
+    },
+    {
+      "epoch": 2.8783783783783785,
+      "grad_norm": 0.2623286843299866,
+      "learning_rate": 0.00029799161808007264,
+      "loss": 0.8742083311080933,
+      "step": 2130
+    },
+    {
+      "epoch": 2.8797297297297297,
+      "grad_norm": 0.2602209448814392,
+      "learning_rate": 0.0002974707056718571,
+      "loss": 0.8871530294418335,
+      "step": 2131
+    },
+    {
+      "epoch": 2.881081081081081,
+      "grad_norm": 0.2739357352256775,
+      "learning_rate": 0.00029694957936804726,
+      "loss": 0.8903975486755371,
+      "step": 2132
+    },
+    {
+      "epoch": 2.8824324324324326,
+      "grad_norm": 0.28526070713996887,
+      "learning_rate": 0.000296428241516757,
+      "loss": 0.8810684084892273,
+      "step": 2133
+    },
+    {
+      "epoch": 2.883783783783784,
+      "grad_norm": 0.36394599080085754,
+      "learning_rate": 0.00029590669446705333,
+      "loss": 0.9005124568939209,
+      "step": 2134
+    },
+    {
+      "epoch": 2.885135135135135,
+      "grad_norm": 0.27970749139785767,
+      "learning_rate": 0.00029538494056894596,
+      "loss": 0.8669013977050781,
+      "step": 2135
+    },
+    {
+      "epoch": 2.8864864864864863,
+      "grad_norm": 0.2756008505821228,
+      "learning_rate": 0.0002948629821733764,
+      "loss": 0.8766254186630249,
+      "step": 2136
+    },
+    {
+      "epoch": 2.887837837837838,
+      "grad_norm": 0.2740667462348938,
+      "learning_rate": 0.0002943408216322077,
+      "loss": 0.8801587820053101,
+      "step": 2137
+    },
+    {
+      "epoch": 2.889189189189189,
+      "grad_norm": 0.2720119059085846,
+      "learning_rate": 0.00029381846129821414,
+      "loss": 0.8682651519775391,
+      "step": 2138
+    },
+    {
+      "epoch": 2.8905405405405404,
+      "grad_norm": 0.3509911596775055,
+      "learning_rate": 0.0002932959035250697,
+      "loss": 0.9004473686218262,
+      "step": 2139
+    },
+    {
+      "epoch": 2.891891891891892,
+      "grad_norm": 0.3801850378513336,
+      "learning_rate": 0.0002927731506673381,
+      "loss": 0.9153184294700623,
+      "step": 2140
+    },
+    {
+      "epoch": 2.8932432432432433,
+      "grad_norm": 0.30604714155197144,
+      "learning_rate": 0.0002922502050804623,
+      "loss": 0.9059813022613525,
+      "step": 2141
+    },
+    {
+      "epoch": 2.8945945945945946,
+      "grad_norm": 0.2990123927593231,
+      "learning_rate": 0.0002917270691207535,
+      "loss": 0.873349666595459,
+      "step": 2142
+    },
+    {
+      "epoch": 2.895945945945946,
+      "grad_norm": 0.3212205469608307,
+      "learning_rate": 0.0002912037451453807,
+      "loss": 0.8815251588821411,
+      "step": 2143
+    },
+    {
+      "epoch": 2.8972972972972975,
+      "grad_norm": 0.2781469523906708,
+      "learning_rate": 0.00029068023551236,
+      "loss": 0.8925422430038452,
+      "step": 2144
+    },
+    {
+      "epoch": 2.8986486486486487,
+      "grad_norm": 0.24458660185337067,
+      "learning_rate": 0.0002901565425805443,
+      "loss": 0.8719623684883118,
+      "step": 2145
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.29449424147605896,
+      "learning_rate": 0.00028963266870961227,
+      "loss": 0.889708399772644,
+      "step": 2146
+    },
+    {
+      "epoch": 2.9013513513513516,
+      "grad_norm": 0.25177979469299316,
+      "learning_rate": 0.00028910861626005774,
+      "loss": 0.8779858946800232,
+      "step": 2147
+    },
+    {
+      "epoch": 2.902702702702703,
+      "grad_norm": 0.3145712912082672,
+      "learning_rate": 0.0002885843875931793,
+      "loss": 0.8936469554901123,
+      "step": 2148
+    },
+    {
+      "epoch": 2.904054054054054,
+      "grad_norm": 0.31224343180656433,
+      "learning_rate": 0.00028805998507106954,
+      "loss": 0.8932575583457947,
+      "step": 2149
+    },
+    {
+      "epoch": 2.9054054054054053,
+      "grad_norm": 0.2613159120082855,
+      "learning_rate": 0.00028753541105660456,
+      "loss": 0.881163477897644,
+      "step": 2150
+    },
+    {
+      "epoch": 2.9067567567567565,
+      "grad_norm": 0.2549487352371216,
+      "learning_rate": 0.00028701066791343287,
+      "loss": 0.8717324137687683,
+      "step": 2151
+    },
+    {
+      "epoch": 2.908108108108108,
+      "grad_norm": 0.28453314304351807,
+      "learning_rate": 0.0002864857580059653,
+      "loss": 0.8948052525520325,
+      "step": 2152
+    },
+    {
+      "epoch": 2.9094594594594594,
+      "grad_norm": 0.25734683871269226,
+      "learning_rate": 0.00028596068369936387,
+      "loss": 0.8819032907485962,
+      "step": 2153
+    },
+    {
+      "epoch": 2.910810810810811,
+      "grad_norm": 0.2809009253978729,
+      "learning_rate": 0.0002854354473595317,
+      "loss": 0.883627712726593,
+      "step": 2154
+    },
+    {
+      "epoch": 2.9121621621621623,
+      "grad_norm": 0.27049344778060913,
+      "learning_rate": 0.00028491005135310166,
+      "loss": 0.8782268762588501,
+      "step": 2155
+    },
+    {
+      "epoch": 2.9135135135135135,
+      "grad_norm": 0.32544058561325073,
+      "learning_rate": 0.00028438449804742626,
+      "loss": 0.9075027704238892,
+      "step": 2156
+    },
+    {
+      "epoch": 2.9148648648648647,
+      "grad_norm": 0.2784898281097412,
+      "learning_rate": 0.0002838587898105666,
+      "loss": 0.8873356580734253,
+      "step": 2157
+    },
+    {
+      "epoch": 2.916216216216216,
+      "grad_norm": 0.243287593126297,
+      "learning_rate": 0.00028333292901128215,
+      "loss": 0.8814469575881958,
+      "step": 2158
+    },
+    {
+      "epoch": 2.9175675675675676,
+      "grad_norm": 0.3021121323108673,
+      "learning_rate": 0.00028280691801901956,
+      "loss": 0.892179548740387,
+      "step": 2159
+    },
+    {
+      "epoch": 2.918918918918919,
+      "grad_norm": 0.26992103457450867,
+      "learning_rate": 0.00028228075920390215,
+      "loss": 0.8827571272850037,
+      "step": 2160
+    },
+    {
+      "epoch": 2.9202702702702705,
+      "grad_norm": 0.2762344181537628,
+      "learning_rate": 0.0002817544549367197,
+      "loss": 0.8880358934402466,
+      "step": 2161
+    },
+    {
+      "epoch": 2.9216216216216218,
+      "grad_norm": 0.3163613975048065,
+      "learning_rate": 0.00028122800758891703,
+      "loss": 0.9106180667877197,
+      "step": 2162
+    },
+    {
+      "epoch": 2.922972972972973,
+      "grad_norm": 0.2807612419128418,
+      "learning_rate": 0.00028070141953258376,
+      "loss": 0.8777960538864136,
+      "step": 2163
+    },
+    {
+      "epoch": 2.924324324324324,
+      "grad_norm": 0.24002785980701447,
+      "learning_rate": 0.00028017469314044354,
+      "loss": 0.8816415071487427,
+      "step": 2164
+    },
+    {
+      "epoch": 2.9256756756756754,
+      "grad_norm": 0.24199317395687103,
+      "learning_rate": 0.00027964783078584333,
+      "loss": 0.8722274303436279,
+      "step": 2165
+    },
+    {
+      "epoch": 2.927027027027027,
+      "grad_norm": 0.2877821922302246,
+      "learning_rate": 0.0002791208348427426,
+      "loss": 0.8967471122741699,
+      "step": 2166
+    },
+    {
+      "epoch": 2.9283783783783783,
+      "grad_norm": 0.30755072832107544,
+      "learning_rate": 0.0002785937076857031,
+      "loss": 0.8987531661987305,
+      "step": 2167
+    },
+    {
+      "epoch": 2.92972972972973,
+      "grad_norm": 0.28164929151535034,
+      "learning_rate": 0.00027806645168987733,
+      "loss": 0.8908619284629822,
+      "step": 2168
+    },
+    {
+      "epoch": 2.9310810810810812,
+      "grad_norm": 0.2570584714412689,
+      "learning_rate": 0.00027753906923099863,
+      "loss": 0.8852226138114929,
+      "step": 2169
+    },
+    {
+      "epoch": 2.9324324324324325,
+      "grad_norm": 0.268938273191452,
+      "learning_rate": 0.00027701156268537016,
+      "loss": 0.8854488730430603,
+      "step": 2170
+    },
+    {
+      "epoch": 2.9337837837837837,
+      "grad_norm": 0.2512552738189697,
+      "learning_rate": 0.000276483934429854,
+      "loss": 0.8835347890853882,
+      "step": 2171
+    },
+    {
+      "epoch": 2.935135135135135,
+      "grad_norm": 0.2786051034927368,
+      "learning_rate": 0.0002759561868418609,
+      "loss": 0.8966101408004761,
+      "step": 2172
+    },
+    {
+      "epoch": 2.9364864864864866,
+      "grad_norm": 0.2516449987888336,
+      "learning_rate": 0.000275428322299339,
+      "loss": 0.8822965025901794,
+      "step": 2173
+    },
+    {
+      "epoch": 2.937837837837838,
+      "grad_norm": 0.2873236835002899,
+      "learning_rate": 0.0002749003431807637,
+      "loss": 0.886327862739563,
+      "step": 2174
+    },
+    {
+      "epoch": 2.939189189189189,
+      "grad_norm": 0.25574326515197754,
+      "learning_rate": 0.00027437225186512657,
+      "loss": 0.8854092359542847,
+      "step": 2175
+    },
+    {
+      "epoch": 2.9405405405405407,
+      "grad_norm": 0.2327030748128891,
+      "learning_rate": 0.00027384405073192455,
+      "loss": 0.86208176612854,
+      "step": 2176
+    },
+    {
+      "epoch": 2.941891891891892,
+      "grad_norm": 0.26710471510887146,
+      "learning_rate": 0.00027331574216114964,
+      "loss": 0.8762836456298828,
+      "step": 2177
+    },
+    {
+      "epoch": 2.943243243243243,
+      "grad_norm": 0.26993924379348755,
+      "learning_rate": 0.0002727873285332778,
+      "loss": 0.8812151551246643,
+      "step": 2178
+    },
+    {
+      "epoch": 2.9445945945945944,
+      "grad_norm": 0.24764086306095123,
+      "learning_rate": 0.0002722588122292585,
+      "loss": 0.8781042098999023,
+      "step": 2179
+    },
+    {
+      "epoch": 2.945945945945946,
+      "grad_norm": 0.32667216658592224,
+      "learning_rate": 0.0002717301956305037,
+      "loss": 0.8851205110549927,
+      "step": 2180
+    },
+    {
+      "epoch": 2.9472972972972973,
+      "grad_norm": 0.29036495089530945,
+      "learning_rate": 0.0002712014811188773,
+      "loss": 0.8669440746307373,
+      "step": 2181
+    },
+    {
+      "epoch": 2.9486486486486485,
+      "grad_norm": 0.26007071137428284,
+      "learning_rate": 0.00027067267107668447,
+      "loss": 0.8753382563591003,
+      "step": 2182
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.2665465474128723,
+      "learning_rate": 0.0002701437678866607,
+      "loss": 0.8848088383674622,
+      "step": 2183
+    },
+    {
+      "epoch": 2.9513513513513514,
+      "grad_norm": 0.35605981945991516,
+      "learning_rate": 0.00026961477393196127,
+      "loss": 0.8874263763427734,
+      "step": 2184
+    },
+    {
+      "epoch": 2.9527027027027026,
+      "grad_norm": 0.2718075215816498,
+      "learning_rate": 0.0002690856915961504,
+      "loss": 0.8709138035774231,
+      "step": 2185
+    },
+    {
+      "epoch": 2.954054054054054,
+      "grad_norm": 0.2586400508880615,
+      "learning_rate": 0.0002685565232631906,
+      "loss": 0.8820140361785889,
+      "step": 2186
+    },
+    {
+      "epoch": 2.9554054054054055,
+      "grad_norm": 0.25579217076301575,
+      "learning_rate": 0.0002680272713174319,
+      "loss": 0.8758094310760498,
+      "step": 2187
+    },
+    {
+      "epoch": 2.9567567567567568,
+      "grad_norm": 0.28764835000038147,
+      "learning_rate": 0.0002674979381436008,
+      "loss": 0.8875294923782349,
+      "step": 2188
+    },
+    {
+      "epoch": 2.958108108108108,
+      "grad_norm": 0.30585774779319763,
+      "learning_rate": 0.0002669685261267902,
+      "loss": 0.8964293003082275,
+      "step": 2189
+    },
+    {
+      "epoch": 2.9594594594594597,
+      "grad_norm": 0.30878904461860657,
+      "learning_rate": 0.000266439037652448,
+      "loss": 0.8929715156555176,
+      "step": 2190
+    },
+    {
+      "epoch": 2.960810810810811,
+      "grad_norm": 0.3084065616130829,
+      "learning_rate": 0.0002659094751063666,
+      "loss": 0.8991612195968628,
+      "step": 2191
+    },
+    {
+      "epoch": 2.962162162162162,
+      "grad_norm": 0.304593950510025,
+      "learning_rate": 0.00026537984087467224,
+      "loss": 0.8949679136276245,
+      "step": 2192
+    },
+    {
+      "epoch": 2.9635135135135133,
+      "grad_norm": 0.2749885320663452,
+      "learning_rate": 0.0002648501373438142,
+      "loss": 0.8705776929855347,
+      "step": 2193
+    },
+    {
+      "epoch": 2.964864864864865,
+      "grad_norm": 0.2865627110004425,
+      "learning_rate": 0.00026432036690055396,
+      "loss": 0.8946327567100525,
+      "step": 2194
+    },
+    {
+      "epoch": 2.9662162162162162,
+      "grad_norm": 0.22356504201889038,
+      "learning_rate": 0.0002637905319319544,
+      "loss": 0.8734426498413086,
+      "step": 2195
+    },
+    {
+      "epoch": 2.9675675675675675,
+      "grad_norm": 0.2326369732618332,
+      "learning_rate": 0.0002632606348253693,
+      "loss": 0.8727171421051025,
+      "step": 2196
+    },
+    {
+      "epoch": 2.968918918918919,
+      "grad_norm": 0.2199176847934723,
+      "learning_rate": 0.0002627306779684324,
+      "loss": 0.8735067844390869,
+      "step": 2197
+    },
+    {
+      "epoch": 2.9702702702702704,
+      "grad_norm": 0.2714870572090149,
+      "learning_rate": 0.00026220066374904653,
+      "loss": 0.8867268562316895,
+      "step": 2198
+    },
+    {
+      "epoch": 2.9716216216216216,
+      "grad_norm": 0.25292378664016724,
+      "learning_rate": 0.0002616705945553732,
+      "loss": 0.8740048408508301,
+      "step": 2199
+    },
+    {
+      "epoch": 2.972972972972973,
+      "grad_norm": 0.34357327222824097,
+      "learning_rate": 0.0002611404727758213,
+      "loss": 0.8997354507446289,
+      "step": 2200
+    },
+    {
+      "epoch": 2.9743243243243245,
+      "grad_norm": 0.356010764837265,
+      "learning_rate": 0.0002606103007990371,
+      "loss": 0.9129589796066284,
+      "step": 2201
+    },
+    {
+      "epoch": 2.9756756756756757,
+      "grad_norm": 0.3109695613384247,
+      "learning_rate": 0.0002600800810138927,
+      "loss": 0.9063622355461121,
+      "step": 2202
+    },
+    {
+      "epoch": 2.977027027027027,
+      "grad_norm": 0.25533419847488403,
+      "learning_rate": 0.0002595498158094757,
+      "loss": 0.8814213275909424,
+      "step": 2203
+    },
+    {
+      "epoch": 2.9783783783783786,
+      "grad_norm": 0.30396729707717896,
+      "learning_rate": 0.00025901950757507847,
+      "loss": 0.8940255641937256,
+      "step": 2204
+    },
+    {
+      "epoch": 2.97972972972973,
+      "grad_norm": 0.3128320574760437,
+      "learning_rate": 0.0002584891587001872,
+      "loss": 0.9051542282104492,
+      "step": 2205
+    },
+    {
+      "epoch": 2.981081081081081,
+      "grad_norm": 0.3098144233226776,
+      "learning_rate": 0.00025795877157447117,
+      "loss": 0.8843621015548706,
+      "step": 2206
+    },
+    {
+      "epoch": 2.9824324324324323,
+      "grad_norm": 0.19813959300518036,
+      "learning_rate": 0.00025742834858777196,
+      "loss": 0.8607967495918274,
+      "step": 2207
+    },
+    {
+      "epoch": 2.983783783783784,
+      "grad_norm": 0.2975059449672699,
+      "learning_rate": 0.0002568978921300928,
+      "loss": 0.8998608589172363,
+      "step": 2208
+    },
+    {
+      "epoch": 2.985135135135135,
+      "grad_norm": 0.2702968716621399,
+      "learning_rate": 0.00025636740459158774,
+      "loss": 0.8886826038360596,
+      "step": 2209
+    },
+    {
+      "epoch": 2.9864864864864864,
+      "grad_norm": 0.26962703466415405,
+      "learning_rate": 0.000255836888362551,
+      "loss": 0.8824656009674072,
+      "step": 2210
+    },
+    {
+      "epoch": 2.987837837837838,
+      "grad_norm": 0.27136754989624023,
+      "learning_rate": 0.00025530634583340587,
+      "loss": 0.8762345314025879,
+      "step": 2211
+    },
+    {
+      "epoch": 2.9891891891891893,
+      "grad_norm": 0.24935993552207947,
+      "learning_rate": 0.0002547757793946942,
+      "loss": 0.8864922523498535,
+      "step": 2212
+    },
+    {
+      "epoch": 2.9905405405405405,
+      "grad_norm": 0.2738676965236664,
+      "learning_rate": 0.0002542451914370656,
+      "loss": 0.8810652494430542,
+      "step": 2213
+    },
+    {
+      "epoch": 2.9918918918918918,
+      "grad_norm": 0.28049764037132263,
+      "learning_rate": 0.00025371458435126664,
+      "loss": 0.89309161901474,
+      "step": 2214
+    },
+    {
+      "epoch": 2.993243243243243,
+      "grad_norm": 0.27761125564575195,
+      "learning_rate": 0.0002531839605281301,
+      "loss": 0.9095168709754944,
+      "step": 2215
+    },
+    {
+      "epoch": 2.9945945945945946,
+      "grad_norm": 0.2239389717578888,
+      "learning_rate": 0.0002526533223585641,
+      "loss": 0.8794859051704407,
+      "step": 2216
+    },
+    {
+      "epoch": 2.995945945945946,
+      "grad_norm": 0.2764374315738678,
+      "learning_rate": 0.0002521226722335414,
+      "loss": 0.8847477436065674,
+      "step": 2217
+    },
+    {
+      "epoch": 2.9972972972972975,
+      "grad_norm": 0.2474570870399475,
+      "learning_rate": 0.0002515920125440888,
+      "loss": 0.8630303740501404,
+      "step": 2218
+    },
+    {
+      "epoch": 2.9986486486486488,
+      "grad_norm": 0.283674418926239,
+      "learning_rate": 0.0002510613456812761,
+      "loss": 0.8801782131195068,
+      "step": 2219
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.26915332674980164,
+      "learning_rate": 0.0002505306740362052,
+      "loss": 0.8812825083732605,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0013513513513512,
+      "grad_norm": 0.28245681524276733,
+      "learning_rate": 0.00025,
+      "loss": 0.8786481022834778,
+      "step": 2221
+    },
+    {
+      "epoch": 3.002702702702703,
+      "grad_norm": 0.24743293225765228,
+      "learning_rate": 0.00024946932596379474,
+      "loss": 0.8796043395996094,
+      "step": 2222
+    },
+    {
+      "epoch": 3.004054054054054,
+      "grad_norm": 0.23386956751346588,
+      "learning_rate": 0.00024893865431872397,
+      "loss": 0.8579753041267395,
+      "step": 2223
+    },
+    {
+      "epoch": 3.0054054054054054,
+      "grad_norm": 0.2509547472000122,
+      "learning_rate": 0.00024840798745591117,
+      "loss": 0.8680855631828308,
+      "step": 2224
+    },
+    {
+      "epoch": 3.0067567567567566,
+      "grad_norm": 0.29318398237228394,
+      "learning_rate": 0.00024787732776645864,
+      "loss": 0.8720840215682983,
+      "step": 2225
+    },
+    {
+      "epoch": 3.0081081081081082,
+      "grad_norm": 0.21161292493343353,
+      "learning_rate": 0.000247346677641436,
+      "loss": 0.8594983816146851,
+      "step": 2226
+    },
+    {
+      "epoch": 3.0094594594594595,
+      "grad_norm": 0.2591874599456787,
+      "learning_rate": 0.00024681603947186996,
+      "loss": 0.8638874888420105,
+      "step": 2227
+    },
+    {
+      "epoch": 3.0108108108108107,
+      "grad_norm": 0.23951996862888336,
+      "learning_rate": 0.00024628541564873337,
+      "loss": 0.8564775586128235,
+      "step": 2228
+    },
+    {
+      "epoch": 3.0121621621621624,
+      "grad_norm": 0.30211642384529114,
+      "learning_rate": 0.0002457548085629345,
+      "loss": 0.8788812160491943,
+      "step": 2229
+    },
+    {
+      "epoch": 3.0135135135135136,
+      "grad_norm": 0.26216399669647217,
+      "learning_rate": 0.00024522422060530583,
+      "loss": 0.8710160255432129,
+      "step": 2230
+    },
+    {
+      "epoch": 3.014864864864865,
+      "grad_norm": 0.22901803255081177,
+      "learning_rate": 0.00024469365416659414,
+      "loss": 0.8678864240646362,
+      "step": 2231
+    },
+    {
+      "epoch": 3.016216216216216,
+      "grad_norm": 0.2699982821941376,
+      "learning_rate": 0.000244163111637449,
+      "loss": 0.8825910091400146,
+      "step": 2232
+    },
+    {
+      "epoch": 3.0175675675675677,
+      "grad_norm": 0.23996087908744812,
+      "learning_rate": 0.00024363259540841222,
+      "loss": 0.8789719343185425,
+      "step": 2233
+    },
+    {
+      "epoch": 3.018918918918919,
+      "grad_norm": 0.2577003240585327,
+      "learning_rate": 0.0002431021078699073,
+      "loss": 0.8715245723724365,
+      "step": 2234
+    },
+    {
+      "epoch": 3.02027027027027,
+      "grad_norm": 0.22879211604595184,
+      "learning_rate": 0.00024257165141222808,
+      "loss": 0.8675684928894043,
+      "step": 2235
+    },
+    {
+      "epoch": 3.0216216216216214,
+      "grad_norm": 0.3341151773929596,
+      "learning_rate": 0.00024204122842552895,
+      "loss": 0.8851441144943237,
+      "step": 2236
+    },
+    {
+      "epoch": 3.022972972972973,
+      "grad_norm": 0.3037666082382202,
+      "learning_rate": 0.0002415108412998128,
+      "loss": 0.8719102740287781,
+      "step": 2237
+    },
+    {
+      "epoch": 3.0243243243243243,
+      "grad_norm": 0.2717050015926361,
+      "learning_rate": 0.00024098049242492152,
+      "loss": 0.8740236759185791,
+      "step": 2238
+    },
+    {
+      "epoch": 3.0256756756756755,
+      "grad_norm": 0.2747006416320801,
+      "learning_rate": 0.0002404501841905243,
+      "loss": 0.8751013278961182,
+      "step": 2239
+    },
+    {
+      "epoch": 3.027027027027027,
+      "grad_norm": 0.2948567271232605,
+      "learning_rate": 0.00023991991898610732,
+      "loss": 0.8710159063339233,
+      "step": 2240
+    },
+    {
+      "epoch": 3.0283783783783784,
+      "grad_norm": 0.2702396512031555,
+      "learning_rate": 0.00023938969920096298,
+      "loss": 0.8775919079780579,
+      "step": 2241
+    },
+    {
+      "epoch": 3.0297297297297296,
+      "grad_norm": 0.2617984712123871,
+      "learning_rate": 0.0002388595272241787,
+      "loss": 0.8687149286270142,
+      "step": 2242
+    },
+    {
+      "epoch": 3.031081081081081,
+      "grad_norm": 0.2886856496334076,
+      "learning_rate": 0.0002383294054446269,
+      "loss": 0.8889174461364746,
+      "step": 2243
+    },
+    {
+      "epoch": 3.0324324324324325,
+      "grad_norm": 0.24505223333835602,
+      "learning_rate": 0.00023779933625095348,
+      "loss": 0.8834069967269897,
+      "step": 2244
+    },
+    {
+      "epoch": 3.0337837837837838,
+      "grad_norm": 0.30133968591690063,
+      "learning_rate": 0.0002372693220315677,
+      "loss": 0.8827613592147827,
+      "step": 2245
+    },
+    {
+      "epoch": 3.035135135135135,
+      "grad_norm": 0.24015666544437408,
+      "learning_rate": 0.00023673936517463074,
+      "loss": 0.8796435594558716,
+      "step": 2246
+    },
+    {
+      "epoch": 3.0364864864864867,
+      "grad_norm": 0.24440433084964752,
+      "learning_rate": 0.00023620946806804561,
+      "loss": 0.8582775592803955,
+      "step": 2247
+    },
+    {
+      "epoch": 3.037837837837838,
+      "grad_norm": 0.28303262591362,
+      "learning_rate": 0.0002356796330994461,
+      "loss": 0.8816821575164795,
+      "step": 2248
+    },
+    {
+      "epoch": 3.039189189189189,
+      "grad_norm": 0.30702245235443115,
+      "learning_rate": 0.0002351498626561858,
+      "loss": 0.8765535354614258,
+      "step": 2249
+    },
+    {
+      "epoch": 3.0405405405405403,
+      "grad_norm": 0.28149691224098206,
+      "learning_rate": 0.00023462015912532782,
+      "loss": 0.8925038576126099,
+      "step": 2250
+    },
+    {
+      "epoch": 3.041891891891892,
+      "grad_norm": 0.2700313925743103,
+      "learning_rate": 0.00023409052489363342,
+      "loss": 0.8919500708580017,
+      "step": 2251
+    },
+    {
+      "epoch": 3.0432432432432432,
+      "grad_norm": 0.26731356978416443,
+      "learning_rate": 0.0002335609623475521,
+      "loss": 0.8774498105049133,
+      "step": 2252
+    },
+    {
+      "epoch": 3.0445945945945945,
+      "grad_norm": 0.25806763768196106,
+      "learning_rate": 0.00023303147387320982,
+      "loss": 0.8749052286148071,
+      "step": 2253
+    },
+    {
+      "epoch": 3.045945945945946,
+      "grad_norm": 0.24659419059753418,
+      "learning_rate": 0.00023250206185639917,
+      "loss": 0.8736898303031921,
+      "step": 2254
+    },
+    {
+      "epoch": 3.0472972972972974,
+      "grad_norm": 0.22123385965824127,
+      "learning_rate": 0.00023197272868256816,
+      "loss": 0.8640708923339844,
+      "step": 2255
+    },
+    {
+      "epoch": 3.0486486486486486,
+      "grad_norm": 0.25848260521888733,
+      "learning_rate": 0.00023144347673680936,
+      "loss": 0.8705750107765198,
+      "step": 2256
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.27677369117736816,
+      "learning_rate": 0.00023091430840384964,
+      "loss": 0.877328634262085,
+      "step": 2257
+    },
+    {
+      "epoch": 3.0513513513513515,
+      "grad_norm": 0.29854586720466614,
+      "learning_rate": 0.0002303852260680388,
+      "loss": 0.8760364055633545,
+      "step": 2258
+    },
+    {
+      "epoch": 3.0527027027027027,
+      "grad_norm": 0.222232386469841,
+      "learning_rate": 0.0002298562321133394,
+      "loss": 0.8541272878646851,
+      "step": 2259
+    },
+    {
+      "epoch": 3.054054054054054,
+      "grad_norm": 0.2587592601776123,
+      "learning_rate": 0.00022932732892331557,
+      "loss": 0.8619615435600281,
+      "step": 2260
+    },
+    {
+      "epoch": 3.0554054054054056,
+      "grad_norm": 0.3104436695575714,
+      "learning_rate": 0.0002287985188811228,
+      "loss": 0.874142050743103,
+      "step": 2261
+    },
+    {
+      "epoch": 3.056756756756757,
+      "grad_norm": 0.2663615047931671,
+      "learning_rate": 0.00022826980436949635,
+      "loss": 0.8642309904098511,
+      "step": 2262
+    },
+    {
+      "epoch": 3.058108108108108,
+      "grad_norm": 0.22837845981121063,
+      "learning_rate": 0.00022774118777074142,
+      "loss": 0.8638479709625244,
+      "step": 2263
+    },
+    {
+      "epoch": 3.0594594594594593,
+      "grad_norm": 0.2972027063369751,
+      "learning_rate": 0.0002272126714667222,
+      "loss": 0.8773185014724731,
+      "step": 2264
+    },
+    {
+      "epoch": 3.060810810810811,
+      "grad_norm": 0.33016878366470337,
+      "learning_rate": 0.0002266842578388504,
+      "loss": 0.8702836632728577,
+      "step": 2265
+    },
+    {
+      "epoch": 3.062162162162162,
+      "grad_norm": 0.2703743577003479,
+      "learning_rate": 0.00022615594926807551,
+      "loss": 0.8774219751358032,
+      "step": 2266
+    },
+    {
+      "epoch": 3.0635135135135134,
+      "grad_norm": 0.30399829149246216,
+      "learning_rate": 0.00022562774813487347,
+      "loss": 0.8712678551673889,
+      "step": 2267
+    },
+    {
+      "epoch": 3.064864864864865,
+      "grad_norm": 0.24527810513973236,
+      "learning_rate": 0.00022509965681923635,
+      "loss": 0.8641320466995239,
+      "step": 2268
+    },
+    {
+      "epoch": 3.0662162162162163,
+      "grad_norm": 0.3068026602268219,
+      "learning_rate": 0.00022457167770066104,
+      "loss": 0.8712729811668396,
+      "step": 2269
+    },
+    {
+      "epoch": 3.0675675675675675,
+      "grad_norm": 0.28483858704566956,
+      "learning_rate": 0.00022404381315813913,
+      "loss": 0.8824198842048645,
+      "step": 2270
+    },
+    {
+      "epoch": 3.0689189189189188,
+      "grad_norm": 0.27471569180488586,
+      "learning_rate": 0.000223516065570146,
+      "loss": 0.8799753189086914,
+      "step": 2271
+    },
+    {
+      "epoch": 3.0702702702702704,
+      "grad_norm": 0.3326389193534851,
+      "learning_rate": 0.00022298843731462985,
+      "loss": 0.8834379315376282,
+      "step": 2272
+    },
+    {
+      "epoch": 3.0716216216216217,
+      "grad_norm": 0.2228376567363739,
+      "learning_rate": 0.00022246093076900143,
+      "loss": 0.864307165145874,
+      "step": 2273
+    },
+    {
+      "epoch": 3.072972972972973,
+      "grad_norm": 0.24693676829338074,
+      "learning_rate": 0.0002219335483101227,
+      "loss": 0.8673359155654907,
+      "step": 2274
+    },
+    {
+      "epoch": 3.074324324324324,
+      "grad_norm": 0.2767293155193329,
+      "learning_rate": 0.00022140629231429698,
+      "loss": 0.8704513311386108,
+      "step": 2275
+    },
+    {
+      "epoch": 3.075675675675676,
+      "grad_norm": 0.2872615456581116,
+      "learning_rate": 0.00022087916515725736,
+      "loss": 0.8693704605102539,
+      "step": 2276
+    },
+    {
+      "epoch": 3.077027027027027,
+      "grad_norm": 0.24123431742191315,
+      "learning_rate": 0.00022035216921415679,
+      "loss": 0.865366518497467,
+      "step": 2277
+    },
+    {
+      "epoch": 3.0783783783783782,
+      "grad_norm": 0.2841516137123108,
+      "learning_rate": 0.00021982530685955653,
+      "loss": 0.8801776170730591,
+      "step": 2278
+    },
+    {
+      "epoch": 3.07972972972973,
+      "grad_norm": 0.2713651657104492,
+      "learning_rate": 0.00021929858046741623,
+      "loss": 0.889110803604126,
+      "step": 2279
+    },
+    {
+      "epoch": 3.081081081081081,
+      "grad_norm": 0.28758594393730164,
+      "learning_rate": 0.00021877199241108304,
+      "loss": 0.8835817575454712,
+      "step": 2280
+    },
+    {
+      "epoch": 3.0824324324324324,
+      "grad_norm": 0.2872033715248108,
+      "learning_rate": 0.0002182455450632803,
+      "loss": 0.8742058277130127,
+      "step": 2281
+    },
+    {
+      "epoch": 3.0837837837837836,
+      "grad_norm": 0.26874151825904846,
+      "learning_rate": 0.00021771924079609788,
+      "loss": 0.868511438369751,
+      "step": 2282
+    },
+    {
+      "epoch": 3.0851351351351353,
+      "grad_norm": 0.27364200353622437,
+      "learning_rate": 0.00021719308198098054,
+      "loss": 0.8673016428947449,
+      "step": 2283
+    },
+    {
+      "epoch": 3.0864864864864865,
+      "grad_norm": 0.2946891784667969,
+      "learning_rate": 0.00021666707098871797,
+      "loss": 0.8946921825408936,
+      "step": 2284
+    },
+    {
+      "epoch": 3.0878378378378377,
+      "grad_norm": 0.28682705760002136,
+      "learning_rate": 0.00021614121018943345,
+      "loss": 0.8577929139137268,
+      "step": 2285
+    },
+    {
+      "epoch": 3.0891891891891894,
+      "grad_norm": 0.27679964900016785,
+      "learning_rate": 0.0002156155019525738,
+      "loss": 0.8696796894073486,
+      "step": 2286
+    },
+    {
+      "epoch": 3.0905405405405406,
+      "grad_norm": 0.2622329294681549,
+      "learning_rate": 0.00021508994864689838,
+      "loss": 0.8667885065078735,
+      "step": 2287
+    },
+    {
+      "epoch": 3.091891891891892,
+      "grad_norm": 0.23634536564350128,
+      "learning_rate": 0.00021456455264046831,
+      "loss": 0.8585928082466125,
+      "step": 2288
+    },
+    {
+      "epoch": 3.093243243243243,
+      "grad_norm": 0.2810300886631012,
+      "learning_rate": 0.00021403931630063617,
+      "loss": 0.8818082809448242,
+      "step": 2289
+    },
+    {
+      "epoch": 3.0945945945945947,
+      "grad_norm": 0.3253014087677002,
+      "learning_rate": 0.00021351424199403477,
+      "loss": 0.8697779178619385,
+      "step": 2290
+    },
+    {
+      "epoch": 3.095945945945946,
+      "grad_norm": 0.2647923529148102,
+      "learning_rate": 0.00021298933208656717,
+      "loss": 0.8702370524406433,
+      "step": 2291
+    },
+    {
+      "epoch": 3.097297297297297,
+      "grad_norm": 0.2581598162651062,
+      "learning_rate": 0.00021246458894339545,
+      "loss": 0.8646286129951477,
+      "step": 2292
+    },
+    {
+      "epoch": 3.098648648648649,
+      "grad_norm": 0.2626897096633911,
+      "learning_rate": 0.0002119400149289305,
+      "loss": 0.8605802059173584,
+      "step": 2293
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.2627605199813843,
+      "learning_rate": 0.00021141561240682068,
+      "loss": 0.8646766543388367,
+      "step": 2294
+    },
+    {
+      "epoch": 3.1013513513513513,
+      "grad_norm": 0.27998417615890503,
+      "learning_rate": 0.00021089138373994224,
+      "loss": 0.8684331774711609,
+      "step": 2295
+    },
+    {
+      "epoch": 3.1027027027027025,
+      "grad_norm": 0.26114803552627563,
+      "learning_rate": 0.0002103673312903878,
+      "loss": 0.8643474578857422,
+      "step": 2296
+    },
+    {
+      "epoch": 3.104054054054054,
+      "grad_norm": 0.2836219072341919,
+      "learning_rate": 0.00020984345741945567,
+      "loss": 0.8714635372161865,
+      "step": 2297
+    },
+    {
+      "epoch": 3.1054054054054054,
+      "grad_norm": 0.30278947949409485,
+      "learning_rate": 0.00020931976448764001,
+      "loss": 0.8761234283447266,
+      "step": 2298
+    },
+    {
+      "epoch": 3.1067567567567567,
+      "grad_norm": 0.2762591540813446,
+      "learning_rate": 0.00020879625485461937,
+      "loss": 0.8700023889541626,
+      "step": 2299
+    },
+    {
+      "epoch": 3.108108108108108,
+      "grad_norm": 0.21707290410995483,
+      "learning_rate": 0.00020827293087924664,
+      "loss": 0.8448672294616699,
+      "step": 2300
+    },
+    {
+      "epoch": 3.1094594594594596,
+      "grad_norm": 0.25777554512023926,
+      "learning_rate": 0.00020774979491953777,
+      "loss": 0.86440110206604,
+      "step": 2301
+    },
+    {
+      "epoch": 3.110810810810811,
+      "grad_norm": 0.2866238057613373,
+      "learning_rate": 0.00020722684933266192,
+      "loss": 0.8698799014091492,
+      "step": 2302
+    },
+    {
+      "epoch": 3.112162162162162,
+      "grad_norm": 0.3080751299858093,
+      "learning_rate": 0.00020670409647493039,
+      "loss": 0.8679055571556091,
+      "step": 2303
+    },
+    {
+      "epoch": 3.1135135135135137,
+      "grad_norm": 0.3039640784263611,
+      "learning_rate": 0.00020618153870178587,
+      "loss": 0.8921195864677429,
+      "step": 2304
+    },
+    {
+      "epoch": 3.114864864864865,
+      "grad_norm": 0.2674875557422638,
+      "learning_rate": 0.0002056591783677923,
+      "loss": 0.8689866065979004,
+      "step": 2305
+    },
+    {
+      "epoch": 3.116216216216216,
+      "grad_norm": 0.2295275628566742,
+      "learning_rate": 0.00020513701782662368,
+      "loss": 0.8703364133834839,
+      "step": 2306
+    },
+    {
+      "epoch": 3.1175675675675674,
+      "grad_norm": 0.23000556230545044,
+      "learning_rate": 0.00020461505943105419,
+      "loss": 0.8620922565460205,
+      "step": 2307
+    },
+    {
+      "epoch": 3.118918918918919,
+      "grad_norm": 0.2588352560997009,
+      "learning_rate": 0.0002040933055329467,
+      "loss": 0.8678509593009949,
+      "step": 2308
+    },
+    {
+      "epoch": 3.1202702702702703,
+      "grad_norm": 0.32819291949272156,
+      "learning_rate": 0.00020357175848324306,
+      "loss": 0.8905925154685974,
+      "step": 2309
+    },
+    {
+      "epoch": 3.1216216216216215,
+      "grad_norm": 0.276021271944046,
+      "learning_rate": 0.00020305042063195275,
+      "loss": 0.8682444095611572,
+      "step": 2310
+    },
+    {
+      "epoch": 3.122972972972973,
+      "grad_norm": 0.2594064474105835,
+      "learning_rate": 0.00020252929432814287,
+      "loss": 0.8691092729568481,
+      "step": 2311
+    },
+    {
+      "epoch": 3.1243243243243244,
+      "grad_norm": 0.2903115749359131,
+      "learning_rate": 0.00020200838191992743,
+      "loss": 0.8799821138381958,
+      "step": 2312
+    },
+    {
+      "epoch": 3.1256756756756756,
+      "grad_norm": 0.2573185861110687,
+      "learning_rate": 0.00020148768575445618,
+      "loss": 0.8725935816764832,
+      "step": 2313
+    },
+    {
+      "epoch": 3.127027027027027,
+      "grad_norm": 0.3153442144393921,
+      "learning_rate": 0.00020096720817790498,
+      "loss": 0.876930832862854,
+      "step": 2314
+    },
+    {
+      "epoch": 3.1283783783783785,
+      "grad_norm": 0.2238197773694992,
+      "learning_rate": 0.00020044695153546456,
+      "loss": 0.8649824261665344,
+      "step": 2315
+    },
+    {
+      "epoch": 3.1297297297297297,
+      "grad_norm": 0.22599418461322784,
+      "learning_rate": 0.00019992691817133024,
+      "loss": 0.866058349609375,
+      "step": 2316
+    },
+    {
+      "epoch": 3.131081081081081,
+      "grad_norm": 0.2963702082633972,
+      "learning_rate": 0.00019940711042869112,
+      "loss": 0.8604861497879028,
+      "step": 2317
+    },
+    {
+      "epoch": 3.1324324324324326,
+      "grad_norm": 0.22070883214473724,
+      "learning_rate": 0.00019888753064971983,
+      "loss": 0.8619557023048401,
+      "step": 2318
+    },
+    {
+      "epoch": 3.133783783783784,
+      "grad_norm": 0.2817903459072113,
+      "learning_rate": 0.00019836818117556187,
+      "loss": 0.8704183101654053,
+      "step": 2319
+    },
+    {
+      "epoch": 3.135135135135135,
+      "grad_norm": 0.2935076057910919,
+      "learning_rate": 0.0001978490643463249,
+      "loss": 0.8642808198928833,
+      "step": 2320
+    },
+    {
+      "epoch": 3.1364864864864863,
+      "grad_norm": 0.23237185180187225,
+      "learning_rate": 0.0001973301825010685,
+      "loss": 0.8581674695014954,
+      "step": 2321
+    },
+    {
+      "epoch": 3.137837837837838,
+      "grad_norm": 0.3275868892669678,
+      "learning_rate": 0.000196811537977793,
+      "loss": 0.8556933403015137,
+      "step": 2322
+    },
+    {
+      "epoch": 3.139189189189189,
+      "grad_norm": 0.2582365870475769,
+      "learning_rate": 0.00019629313311343008,
+      "loss": 0.8749467730522156,
+      "step": 2323
+    },
+    {
+      "epoch": 3.1405405405405404,
+      "grad_norm": 0.28120991587638855,
+      "learning_rate": 0.00019577497024383093,
+      "loss": 0.8877913951873779,
+      "step": 2324
+    },
+    {
+      "epoch": 3.141891891891892,
+      "grad_norm": 0.2379816323518753,
+      "learning_rate": 0.00019525705170375673,
+      "loss": 0.8675013780593872,
+      "step": 2325
+    },
+    {
+      "epoch": 3.1432432432432433,
+      "grad_norm": 0.23793521523475647,
+      "learning_rate": 0.00019473937982686756,
+      "loss": 0.8622690439224243,
+      "step": 2326
+    },
+    {
+      "epoch": 3.1445945945945946,
+      "grad_norm": 0.25070738792419434,
+      "learning_rate": 0.00019422195694571216,
+      "loss": 0.8728247880935669,
+      "step": 2327
+    },
+    {
+      "epoch": 3.145945945945946,
+      "grad_norm": 0.320070743560791,
+      "learning_rate": 0.00019370478539171743,
+      "loss": 0.8793462514877319,
+      "step": 2328
+    },
+    {
+      "epoch": 3.1472972972972975,
+      "grad_norm": 0.2878582179546356,
+      "learning_rate": 0.00019318786749517752,
+      "loss": 0.8732593059539795,
+      "step": 2329
+    },
+    {
+      "epoch": 3.1486486486486487,
+      "grad_norm": 0.2694909870624542,
+      "learning_rate": 0.00019267120558524395,
+      "loss": 0.8776203989982605,
+      "step": 2330
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.23522864282131195,
+      "learning_rate": 0.00019215480198991463,
+      "loss": 0.8610544204711914,
+      "step": 2331
+    },
+    {
+      "epoch": 3.1513513513513516,
+      "grad_norm": 0.23567554354667664,
+      "learning_rate": 0.00019163865903602372,
+      "loss": 0.8638031482696533,
+      "step": 2332
+    },
+    {
+      "epoch": 3.152702702702703,
+      "grad_norm": 0.24203899502754211,
+      "learning_rate": 0.00019112277904923065,
+      "loss": 0.8708548545837402,
+      "step": 2333
+    },
+    {
+      "epoch": 3.154054054054054,
+      "grad_norm": 0.25642532110214233,
+      "learning_rate": 0.00019060716435401025,
+      "loss": 0.8711357712745667,
+      "step": 2334
+    },
+    {
+      "epoch": 3.1554054054054053,
+      "grad_norm": 0.27868375182151794,
+      "learning_rate": 0.00019009181727364205,
+      "loss": 0.8705508708953857,
+      "step": 2335
+    },
+    {
+      "epoch": 3.156756756756757,
+      "grad_norm": 0.22623246908187866,
+      "learning_rate": 0.00018957674013019937,
+      "loss": 0.8557822108268738,
+      "step": 2336
+    },
+    {
+      "epoch": 3.158108108108108,
+      "grad_norm": 0.3316793739795685,
+      "learning_rate": 0.00018906193524453963,
+      "loss": 0.8745378851890564,
+      "step": 2337
+    },
+    {
+      "epoch": 3.1594594594594594,
+      "grad_norm": 0.2182602733373642,
+      "learning_rate": 0.0001885474049362932,
+      "loss": 0.8564372062683105,
+      "step": 2338
+    },
+    {
+      "epoch": 3.1608108108108106,
+      "grad_norm": 0.21598142385482788,
+      "learning_rate": 0.0001880331515238536,
+      "loss": 0.8565239906311035,
+      "step": 2339
+    },
+    {
+      "epoch": 3.1621621621621623,
+      "grad_norm": 0.2625301480293274,
+      "learning_rate": 0.0001875191773243662,
+      "loss": 0.8712631464004517,
+      "step": 2340
+    },
+    {
+      "epoch": 3.1635135135135135,
+      "grad_norm": 0.20701533555984497,
+      "learning_rate": 0.00018700548465371876,
+      "loss": 0.8636097311973572,
+      "step": 2341
+    },
+    {
+      "epoch": 3.1648648648648647,
+      "grad_norm": 0.2954597771167755,
+      "learning_rate": 0.00018649207582653018,
+      "loss": 0.871258556842804,
+      "step": 2342
+    },
+    {
+      "epoch": 3.1662162162162164,
+      "grad_norm": 0.22664862871170044,
+      "learning_rate": 0.00018597895315614066,
+      "loss": 0.8721297979354858,
+      "step": 2343
+    },
+    {
+      "epoch": 3.1675675675675676,
+      "grad_norm": 0.2125791758298874,
+      "learning_rate": 0.00018546611895460093,
+      "loss": 0.8552570343017578,
+      "step": 2344
+    },
+    {
+      "epoch": 3.168918918918919,
+      "grad_norm": 0.2691446542739868,
+      "learning_rate": 0.00018495357553266177,
+      "loss": 0.8832975625991821,
+      "step": 2345
+    },
+    {
+      "epoch": 3.17027027027027,
+      "grad_norm": 0.2728688418865204,
+      "learning_rate": 0.000184441325199764,
+      "loss": 0.867805540561676,
+      "step": 2346
+    },
+    {
+      "epoch": 3.1716216216216218,
+      "grad_norm": 0.22524219751358032,
+      "learning_rate": 0.00018392937026402758,
+      "loss": 0.8463287353515625,
+      "step": 2347
+    },
+    {
+      "epoch": 3.172972972972973,
+      "grad_norm": 0.24862299859523773,
+      "learning_rate": 0.0001834177130322417,
+      "loss": 0.8685007095336914,
+      "step": 2348
+    },
+    {
+      "epoch": 3.174324324324324,
+      "grad_norm": 0.2562634348869324,
+      "learning_rate": 0.00018290635580985392,
+      "loss": 0.8742742538452148,
+      "step": 2349
+    },
+    {
+      "epoch": 3.175675675675676,
+      "grad_norm": 0.2213626205921173,
+      "learning_rate": 0.00018239530090096017,
+      "loss": 0.8630648851394653,
+      "step": 2350
+    },
+    {
+      "epoch": 3.177027027027027,
+      "grad_norm": 0.24784724414348602,
+      "learning_rate": 0.0001818845506082943,
+      "loss": 0.8717747926712036,
+      "step": 2351
+    },
+    {
+      "epoch": 3.1783783783783783,
+      "grad_norm": 0.2979477643966675,
+      "learning_rate": 0.0001813741072332172,
+      "loss": 0.8888581991195679,
+      "step": 2352
+    },
+    {
+      "epoch": 3.1797297297297296,
+      "grad_norm": 0.29202064871788025,
+      "learning_rate": 0.00018086397307570724,
+      "loss": 0.8732571601867676,
+      "step": 2353
+    },
+    {
+      "epoch": 3.1810810810810812,
+      "grad_norm": 0.23958724737167358,
+      "learning_rate": 0.00018035415043434925,
+      "loss": 0.8513171076774597,
+      "step": 2354
+    },
+    {
+      "epoch": 3.1824324324324325,
+      "grad_norm": 0.27801957726478577,
+      "learning_rate": 0.00017984464160632468,
+      "loss": 0.8761996626853943,
+      "step": 2355
+    },
+    {
+      "epoch": 3.1837837837837837,
+      "grad_norm": 0.27619555592536926,
+      "learning_rate": 0.00017933544888740062,
+      "loss": 0.8550466895103455,
+      "step": 2356
+    },
+    {
+      "epoch": 3.185135135135135,
+      "grad_norm": 0.23280490934848785,
+      "learning_rate": 0.00017882657457192014,
+      "loss": 0.8541948199272156,
+      "step": 2357
+    },
+    {
+      "epoch": 3.1864864864864866,
+      "grad_norm": 0.2798883616924286,
+      "learning_rate": 0.00017831802095279149,
+      "loss": 0.8871515989303589,
+      "step": 2358
+    },
+    {
+      "epoch": 3.187837837837838,
+      "grad_norm": 0.24571920931339264,
+      "learning_rate": 0.00017780979032147793,
+      "loss": 0.8640373349189758,
+      "step": 2359
+    },
+    {
+      "epoch": 3.189189189189189,
+      "grad_norm": 0.2816995084285736,
+      "learning_rate": 0.00017730188496798755,
+      "loss": 0.8782883882522583,
+      "step": 2360
+    },
+    {
+      "epoch": 3.1905405405405407,
+      "grad_norm": 0.24954397976398468,
+      "learning_rate": 0.00017679430718086243,
+      "loss": 0.8701125383377075,
+      "step": 2361
+    },
+    {
+      "epoch": 3.191891891891892,
+      "grad_norm": 0.2929428815841675,
+      "learning_rate": 0.00017628705924716903,
+      "loss": 0.8632237315177917,
+      "step": 2362
+    },
+    {
+      "epoch": 3.193243243243243,
+      "grad_norm": 0.27022144198417664,
+      "learning_rate": 0.00017578014345248728,
+      "loss": 0.8659729361534119,
+      "step": 2363
+    },
+    {
+      "epoch": 3.1945945945945944,
+      "grad_norm": 0.3026275634765625,
+      "learning_rate": 0.0001752735620809009,
+      "loss": 0.8769410848617554,
+      "step": 2364
+    },
+    {
+      "epoch": 3.195945945945946,
+      "grad_norm": 0.22482118010520935,
+      "learning_rate": 0.00017476731741498618,
+      "loss": 0.8658411502838135,
+      "step": 2365
+    },
+    {
+      "epoch": 3.1972972972972973,
+      "grad_norm": 0.2588195204734802,
+      "learning_rate": 0.00017426141173580289,
+      "loss": 0.8668730854988098,
+      "step": 2366
+    },
+    {
+      "epoch": 3.1986486486486485,
+      "grad_norm": 0.31844162940979004,
+      "learning_rate": 0.00017375584732288307,
+      "loss": 0.8564999103546143,
+      "step": 2367
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.2446707785129547,
+      "learning_rate": 0.00017325062645422103,
+      "loss": 0.8731265068054199,
+      "step": 2368
+    },
+    {
+      "epoch": 3.2013513513513514,
+      "grad_norm": 0.25380104780197144,
+      "learning_rate": 0.00017274575140626317,
+      "loss": 0.864409327507019,
+      "step": 2369
+    },
+    {
+      "epoch": 3.2027027027027026,
+      "grad_norm": 0.24292118847370148,
+      "learning_rate": 0.0001722412244538977,
+      "loss": 0.8592667579650879,
+      "step": 2370
+    },
+    {
+      "epoch": 3.204054054054054,
+      "grad_norm": 0.2699313163757324,
+      "learning_rate": 0.00017173704787044446,
+      "loss": 0.8659987449645996,
+      "step": 2371
+    },
+    {
+      "epoch": 3.2054054054054055,
+      "grad_norm": 0.19465264678001404,
+      "learning_rate": 0.00017123322392764435,
+      "loss": 0.8652678728103638,
+      "step": 2372
+    },
+    {
+      "epoch": 3.2067567567567568,
+      "grad_norm": 0.24517999589443207,
+      "learning_rate": 0.00017072975489564957,
+      "loss": 0.8616831302642822,
+      "step": 2373
+    },
+    {
+      "epoch": 3.208108108108108,
+      "grad_norm": 0.27129504084587097,
+      "learning_rate": 0.00017022664304301287,
+      "loss": 0.8617523908615112,
+      "step": 2374
+    },
+    {
+      "epoch": 3.2094594594594597,
+      "grad_norm": 0.24193865060806274,
+      "learning_rate": 0.00016972389063667798,
+      "loss": 0.8671574592590332,
+      "step": 2375
+    },
+    {
+      "epoch": 3.210810810810811,
+      "grad_norm": 0.23926737904548645,
+      "learning_rate": 0.0001692214999419688,
+      "loss": 0.8596177101135254,
+      "step": 2376
+    },
+    {
+      "epoch": 3.212162162162162,
+      "grad_norm": 0.2672814130783081,
+      "learning_rate": 0.00016871947322257913,
+      "loss": 0.8882131576538086,
+      "step": 2377
+    },
+    {
+      "epoch": 3.2135135135135133,
+      "grad_norm": 0.31009286642074585,
+      "learning_rate": 0.00016821781274056348,
+      "loss": 0.874397873878479,
+      "step": 2378
+    },
+    {
+      "epoch": 3.214864864864865,
+      "grad_norm": 0.26734107732772827,
+      "learning_rate": 0.00016771652075632537,
+      "loss": 0.86503666639328,
+      "step": 2379
+    },
+    {
+      "epoch": 3.2162162162162162,
+      "grad_norm": 0.27024152874946594,
+      "learning_rate": 0.0001672155995286085,
+      "loss": 0.8673187494277954,
+      "step": 2380
+    },
+    {
+      "epoch": 3.2175675675675675,
+      "grad_norm": 0.28771108388900757,
+      "learning_rate": 0.0001667150513144856,
+      "loss": 0.874567449092865,
+      "step": 2381
+    },
+    {
+      "epoch": 3.218918918918919,
+      "grad_norm": 0.21516448259353638,
+      "learning_rate": 0.00016621487836934897,
+      "loss": 0.8510260581970215,
+      "step": 2382
+    },
+    {
+      "epoch": 3.2202702702702704,
+      "grad_norm": 0.28721946477890015,
+      "learning_rate": 0.0001657150829468999,
+      "loss": 0.8708043098449707,
+      "step": 2383
+    },
+    {
+      "epoch": 3.2216216216216216,
+      "grad_norm": 0.30524682998657227,
+      "learning_rate": 0.0001652156672991385,
+      "loss": 0.8783912062644958,
+      "step": 2384
+    },
+    {
+      "epoch": 3.222972972972973,
+      "grad_norm": 0.2695196568965912,
+      "learning_rate": 0.00016471663367635382,
+      "loss": 0.8528311848640442,
+      "step": 2385
+    },
+    {
+      "epoch": 3.2243243243243245,
+      "grad_norm": 0.26596498489379883,
+      "learning_rate": 0.00016421798432711345,
+      "loss": 0.8712698221206665,
+      "step": 2386
+    },
+    {
+      "epoch": 3.2256756756756757,
+      "grad_norm": 0.3067319095134735,
+      "learning_rate": 0.00016371972149825366,
+      "loss": 0.8509082794189453,
+      "step": 2387
+    },
+    {
+      "epoch": 3.227027027027027,
+      "grad_norm": 0.2347768396139145,
+      "learning_rate": 0.00016322184743486893,
+      "loss": 0.8632569909095764,
+      "step": 2388
+    },
+    {
+      "epoch": 3.2283783783783786,
+      "grad_norm": 0.23592756688594818,
+      "learning_rate": 0.00016272436438030219,
+      "loss": 0.8744109869003296,
+      "step": 2389
+    },
+    {
+      "epoch": 3.22972972972973,
+      "grad_norm": 0.2661348283290863,
+      "learning_rate": 0.00016222727457613446,
+      "loss": 0.8711393475532532,
+      "step": 2390
+    },
+    {
+      "epoch": 3.231081081081081,
+      "grad_norm": 0.2310265600681305,
+      "learning_rate": 0.0001617305802621748,
+      "loss": 0.8603184819221497,
+      "step": 2391
+    },
+    {
+      "epoch": 3.2324324324324323,
+      "grad_norm": 0.2529321312904358,
+      "learning_rate": 0.00016123428367645045,
+      "loss": 0.8778001070022583,
+      "step": 2392
+    },
+    {
+      "epoch": 3.233783783783784,
+      "grad_norm": 0.2600333094596863,
+      "learning_rate": 0.00016073838705519617,
+      "loss": 0.8654484748840332,
+      "step": 2393
+    },
+    {
+      "epoch": 3.235135135135135,
+      "grad_norm": 0.24495479464530945,
+      "learning_rate": 0.00016024289263284508,
+      "loss": 0.8694018721580505,
+      "step": 2394
+    },
+    {
+      "epoch": 3.2364864864864864,
+      "grad_norm": 0.2626786231994629,
+      "learning_rate": 0.00015974780264201743,
+      "loss": 0.872611403465271,
+      "step": 2395
+    },
+    {
+      "epoch": 3.237837837837838,
+      "grad_norm": 0.30014336109161377,
+      "learning_rate": 0.00015925311931351172,
+      "loss": 0.8791029453277588,
+      "step": 2396
+    },
+    {
+      "epoch": 3.2391891891891893,
+      "grad_norm": 0.26000314950942993,
+      "learning_rate": 0.0001587588448762937,
+      "loss": 0.8746799230575562,
+      "step": 2397
+    },
+    {
+      "epoch": 3.2405405405405405,
+      "grad_norm": 0.2414914220571518,
+      "learning_rate": 0.00015826498155748698,
+      "loss": 0.8601603507995605,
+      "step": 2398
+    },
+    {
+      "epoch": 3.2418918918918918,
+      "grad_norm": 0.2715638279914856,
+      "learning_rate": 0.00015777153158236267,
+      "loss": 0.8739718198776245,
+      "step": 2399
+    },
+    {
+      "epoch": 3.2432432432432434,
+      "grad_norm": 0.22535686194896698,
+      "learning_rate": 0.00015727849717432922,
+      "loss": 0.8659794330596924,
+      "step": 2400
+    },
+    {
+      "epoch": 3.2445945945945946,
+      "grad_norm": 0.21759988367557526,
+      "learning_rate": 0.00015678588055492287,
+      "loss": 0.8599169850349426,
+      "step": 2401
+    },
+    {
+      "epoch": 3.245945945945946,
+      "grad_norm": 0.27230122685432434,
+      "learning_rate": 0.0001562936839437972,
+      "loss": 0.8851460218429565,
+      "step": 2402
+    },
+    {
+      "epoch": 3.247297297297297,
+      "grad_norm": 0.2649877071380615,
+      "learning_rate": 0.00015580190955871348,
+      "loss": 0.8784259557723999,
+      "step": 2403
+    },
+    {
+      "epoch": 3.2486486486486488,
+      "grad_norm": 0.29717445373535156,
+      "learning_rate": 0.0001553105596155302,
+      "loss": 0.8771581649780273,
+      "step": 2404
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.2563573718070984,
+      "learning_rate": 0.0001548196363281937,
+      "loss": 0.8568679690361023,
+      "step": 2405
+    },
+    {
+      "epoch": 3.2513513513513512,
+      "grad_norm": 0.2529267966747284,
+      "learning_rate": 0.00015432914190872756,
+      "loss": 0.8681106567382812,
+      "step": 2406
+    },
+    {
+      "epoch": 3.2527027027027025,
+      "grad_norm": 0.22288262844085693,
+      "learning_rate": 0.00015383907856722327,
+      "loss": 0.8601291179656982,
+      "step": 2407
+    },
+    {
+      "epoch": 3.254054054054054,
+      "grad_norm": 0.2490970343351364,
+      "learning_rate": 0.00015334944851182978,
+      "loss": 0.8634223937988281,
+      "step": 2408
+    },
+    {
+      "epoch": 3.2554054054054054,
+      "grad_norm": 0.2792920470237732,
+      "learning_rate": 0.00015286025394874365,
+      "loss": 0.8707645535469055,
+      "step": 2409
+    },
+    {
+      "epoch": 3.2567567567567566,
+      "grad_norm": 0.2623020112514496,
+      "learning_rate": 0.0001523714970821994,
+      "loss": 0.8680872917175293,
+      "step": 2410
+    },
+    {
+      "epoch": 3.2581081081081082,
+      "grad_norm": 0.2253837138414383,
+      "learning_rate": 0.00015188318011445906,
+      "loss": 0.8515494465827942,
+      "step": 2411
+    },
+    {
+      "epoch": 3.2594594594594595,
+      "grad_norm": 0.24258385598659515,
+      "learning_rate": 0.00015139530524580286,
+      "loss": 0.8588720560073853,
+      "step": 2412
+    },
+    {
+      "epoch": 3.2608108108108107,
+      "grad_norm": 0.26553773880004883,
+      "learning_rate": 0.00015090787467451872,
+      "loss": 0.8686118721961975,
+      "step": 2413
+    },
+    {
+      "epoch": 3.262162162162162,
+      "grad_norm": 0.2688366770744324,
+      "learning_rate": 0.0001504208905968929,
+      "loss": 0.8806657195091248,
+      "step": 2414
+    },
+    {
+      "epoch": 3.2635135135135136,
+      "grad_norm": 0.22856402397155762,
+      "learning_rate": 0.00014993435520719954,
+      "loss": 0.8589751720428467,
+      "step": 2415
+    },
+    {
+      "epoch": 3.264864864864865,
+      "grad_norm": 0.27925917506217957,
+      "learning_rate": 0.00014944827069769123,
+      "loss": 0.8699506521224976,
+      "step": 2416
+    },
+    {
+      "epoch": 3.266216216216216,
+      "grad_norm": 0.27406033873558044,
+      "learning_rate": 0.00014896263925858903,
+      "loss": 0.8660067915916443,
+      "step": 2417
+    },
+    {
+      "epoch": 3.2675675675675677,
+      "grad_norm": 0.2580908536911011,
+      "learning_rate": 0.00014847746307807233,
+      "loss": 0.8708088994026184,
+      "step": 2418
+    },
+    {
+      "epoch": 3.268918918918919,
+      "grad_norm": 0.28307676315307617,
+      "learning_rate": 0.0001479927443422695,
+      "loss": 0.8820024728775024,
+      "step": 2419
+    },
+    {
+      "epoch": 3.27027027027027,
+      "grad_norm": 0.24597322940826416,
+      "learning_rate": 0.00014750848523524724,
+      "loss": 0.864234209060669,
+      "step": 2420
+    },
+    {
+      "epoch": 3.2716216216216214,
+      "grad_norm": 0.30561894178390503,
+      "learning_rate": 0.00014702468793900187,
+      "loss": 0.8826373815536499,
+      "step": 2421
+    },
+    {
+      "epoch": 3.272972972972973,
+      "grad_norm": 0.23794260621070862,
+      "learning_rate": 0.00014654135463344832,
+      "loss": 0.8677352666854858,
+      "step": 2422
+    },
+    {
+      "epoch": 3.2743243243243243,
+      "grad_norm": 0.29802313446998596,
+      "learning_rate": 0.0001460584874964111,
+      "loss": 0.8776352405548096,
+      "step": 2423
+    },
+    {
+      "epoch": 3.2756756756756755,
+      "grad_norm": 0.2956795394420624,
+      "learning_rate": 0.00014557608870361432,
+      "loss": 0.8550947904586792,
+      "step": 2424
+    },
+    {
+      "epoch": 3.277027027027027,
+      "grad_norm": 0.28673437237739563,
+      "learning_rate": 0.00014509416042867148,
+      "loss": 0.8721638917922974,
+      "step": 2425
+    },
+    {
+      "epoch": 3.2783783783783784,
+      "grad_norm": 0.2447083443403244,
+      "learning_rate": 0.00014461270484307642,
+      "loss": 0.8576008081436157,
+      "step": 2426
+    },
+    {
+      "epoch": 3.2797297297297296,
+      "grad_norm": 0.29338350892066956,
+      "learning_rate": 0.0001441317241161928,
+      "loss": 0.8732901811599731,
+      "step": 2427
+    },
+    {
+      "epoch": 3.281081081081081,
+      "grad_norm": 0.29690712690353394,
+      "learning_rate": 0.0001436512204152447,
+      "loss": 0.867360532283783,
+      "step": 2428
+    },
+    {
+      "epoch": 3.2824324324324325,
+      "grad_norm": 0.23596090078353882,
+      "learning_rate": 0.0001431711959053069,
+      "loss": 0.8622161746025085,
+      "step": 2429
+    },
+    {
+      "epoch": 3.2837837837837838,
+      "grad_norm": 0.25225022435188293,
+      "learning_rate": 0.00014269165274929496,
+      "loss": 0.8600240349769592,
+      "step": 2430
+    },
+    {
+      "epoch": 3.285135135135135,
+      "grad_norm": 0.2658272981643677,
+      "learning_rate": 0.00014221259310795542,
+      "loss": 0.8702290654182434,
+      "step": 2431
+    },
+    {
+      "epoch": 3.2864864864864867,
+      "grad_norm": 0.23173300921916962,
+      "learning_rate": 0.00014173401913985644,
+      "loss": 0.8619678020477295,
+      "step": 2432
+    },
+    {
+      "epoch": 3.287837837837838,
+      "grad_norm": 0.2377159744501114,
+      "learning_rate": 0.00014125593300137764,
+      "loss": 0.867108941078186,
+      "step": 2433
+    },
+    {
+      "epoch": 3.289189189189189,
+      "grad_norm": 0.24731957912445068,
+      "learning_rate": 0.00014077833684670045,
+      "loss": 0.8574313521385193,
+      "step": 2434
+    },
+    {
+      "epoch": 3.2905405405405403,
+      "grad_norm": 0.2647355794906616,
+      "learning_rate": 0.00014030123282779888,
+      "loss": 0.861417293548584,
+      "step": 2435
+    },
+    {
+      "epoch": 3.291891891891892,
+      "grad_norm": 0.30270543694496155,
+      "learning_rate": 0.0001398246230944289,
+      "loss": 0.8867008686065674,
+      "step": 2436
+    },
+    {
+      "epoch": 3.2932432432432432,
+      "grad_norm": 0.2259078472852707,
+      "learning_rate": 0.0001393485097941199,
+      "loss": 0.8630704879760742,
+      "step": 2437
+    },
+    {
+      "epoch": 3.2945945945945945,
+      "grad_norm": 0.2002897411584854,
+      "learning_rate": 0.00013887289507216394,
+      "loss": 0.8530676364898682,
+      "step": 2438
+    },
+    {
+      "epoch": 3.295945945945946,
+      "grad_norm": 0.23283791542053223,
+      "learning_rate": 0.0001383977810716068,
+      "loss": 0.8627660870552063,
+      "step": 2439
+    },
+    {
+      "epoch": 3.2972972972972974,
+      "grad_norm": 0.2308565080165863,
+      "learning_rate": 0.00013792316993323822,
+      "loss": 0.868086576461792,
+      "step": 2440
+    },
+    {
+      "epoch": 3.2986486486486486,
+      "grad_norm": 0.24681322276592255,
+      "learning_rate": 0.00013744906379558163,
+      "loss": 0.8623136281967163,
+      "step": 2441
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.3092241585254669,
+      "learning_rate": 0.00013697546479488564,
+      "loss": 0.8727505207061768,
+      "step": 2442
+    },
+    {
+      "epoch": 3.3013513513513515,
+      "grad_norm": 0.26917749643325806,
+      "learning_rate": 0.00013650237506511331,
+      "loss": 0.8669689297676086,
+      "step": 2443
+    },
+    {
+      "epoch": 3.3027027027027027,
+      "grad_norm": 0.23903736472129822,
+      "learning_rate": 0.0001360297967379332,
+      "loss": 0.8659352660179138,
+      "step": 2444
+    },
+    {
+      "epoch": 3.304054054054054,
+      "grad_norm": 0.281272828578949,
+      "learning_rate": 0.00013555773194270948,
+      "loss": 0.8804596066474915,
+      "step": 2445
+    },
+    {
+      "epoch": 3.3054054054054056,
+      "grad_norm": 0.2871777415275574,
+      "learning_rate": 0.00013508618280649255,
+      "loss": 0.8721723556518555,
+      "step": 2446
+    },
+    {
+      "epoch": 3.306756756756757,
+      "grad_norm": 0.2571290135383606,
+      "learning_rate": 0.00013461515145400907,
+      "loss": 0.8606833219528198,
+      "step": 2447
+    },
+    {
+      "epoch": 3.308108108108108,
+      "grad_norm": 0.22704248130321503,
+      "learning_rate": 0.0001341446400076531,
+      "loss": 0.8745690584182739,
+      "step": 2448
+    },
+    {
+      "epoch": 3.3094594594594593,
+      "grad_norm": 0.23102006316184998,
+      "learning_rate": 0.00013367465058747565,
+      "loss": 0.8646962642669678,
+      "step": 2449
+    },
+    {
+      "epoch": 3.310810810810811,
+      "grad_norm": 0.24596478044986725,
+      "learning_rate": 0.0001332051853111757,
+      "loss": 0.8688086271286011,
+      "step": 2450
+    },
+    {
+      "epoch": 3.312162162162162,
+      "grad_norm": 0.2292584776878357,
+      "learning_rate": 0.0001327362462940908,
+      "loss": 0.8639126420021057,
+      "step": 2451
+    },
+    {
+      "epoch": 3.3135135135135134,
+      "grad_norm": 0.21922484040260315,
+      "learning_rate": 0.00013226783564918666,
+      "loss": 0.854387104511261,
+      "step": 2452
+    },
+    {
+      "epoch": 3.314864864864865,
+      "grad_norm": 0.25121772289276123,
+      "learning_rate": 0.00013179995548704882,
+      "loss": 0.8659825325012207,
+      "step": 2453
+    },
+    {
+      "epoch": 3.3162162162162163,
+      "grad_norm": 0.26441049575805664,
+      "learning_rate": 0.0001313326079158722,
+      "loss": 0.874781608581543,
+      "step": 2454
+    },
+    {
+      "epoch": 3.3175675675675675,
+      "grad_norm": 0.2536076307296753,
+      "learning_rate": 0.00013086579504145203,
+      "loss": 0.8749950528144836,
+      "step": 2455
+    },
+    {
+      "epoch": 3.3189189189189188,
+      "grad_norm": 0.2678215801715851,
+      "learning_rate": 0.00013039951896717445,
+      "loss": 0.8662208914756775,
+      "step": 2456
+    },
+    {
+      "epoch": 3.3202702702702704,
+      "grad_norm": 0.2735769748687744,
+      "learning_rate": 0.00012993378179400645,
+      "loss": 0.8659250736236572,
+      "step": 2457
+    },
+    {
+      "epoch": 3.3216216216216217,
+      "grad_norm": 0.25811728835105896,
+      "learning_rate": 0.0001294685856204873,
+      "loss": 0.8672993183135986,
+      "step": 2458
+    },
+    {
+      "epoch": 3.322972972972973,
+      "grad_norm": 0.27233678102493286,
+      "learning_rate": 0.0001290039325427183,
+      "loss": 0.8673480749130249,
+      "step": 2459
+    },
+    {
+      "epoch": 3.3243243243243246,
+      "grad_norm": 0.2538042664527893,
+      "learning_rate": 0.00012853982465435375,
+      "loss": 0.8708063364028931,
+      "step": 2460
+    },
+    {
+      "epoch": 3.325675675675676,
+      "grad_norm": 0.21073804795742035,
+      "learning_rate": 0.00012807626404659142,
+      "loss": 0.8564064502716064,
+      "step": 2461
+    },
+    {
+      "epoch": 3.327027027027027,
+      "grad_norm": 0.23770317435264587,
+      "learning_rate": 0.00012761325280816305,
+      "loss": 0.8583852052688599,
+      "step": 2462
+    },
+    {
+      "epoch": 3.3283783783783782,
+      "grad_norm": 0.22055767476558685,
+      "learning_rate": 0.00012715079302532496,
+      "loss": 0.8568918704986572,
+      "step": 2463
+    },
+    {
+      "epoch": 3.32972972972973,
+      "grad_norm": 0.2692430913448334,
+      "learning_rate": 0.00012668888678184892,
+      "loss": 0.8676198124885559,
+      "step": 2464
+    },
+    {
+      "epoch": 3.331081081081081,
+      "grad_norm": 0.23986968398094177,
+      "learning_rate": 0.00012622753615901245,
+      "loss": 0.8660619258880615,
+      "step": 2465
+    },
+    {
+      "epoch": 3.3324324324324324,
+      "grad_norm": 0.2331165373325348,
+      "learning_rate": 0.00012576674323558929,
+      "loss": 0.8573787212371826,
+      "step": 2466
+    },
+    {
+      "epoch": 3.3337837837837836,
+      "grad_norm": 0.2595269978046417,
+      "learning_rate": 0.00012530651008784075,
+      "loss": 0.8644223213195801,
+      "step": 2467
+    },
+    {
+      "epoch": 3.3351351351351353,
+      "grad_norm": 0.24447385966777802,
+      "learning_rate": 0.00012484683878950526,
+      "loss": 0.8570539951324463,
+      "step": 2468
+    },
+    {
+      "epoch": 3.3364864864864865,
+      "grad_norm": 0.2625342905521393,
+      "learning_rate": 0.00012438773141179024,
+      "loss": 0.8697800636291504,
+      "step": 2469
+    },
+    {
+      "epoch": 3.3378378378378377,
+      "grad_norm": 0.22890232503414154,
+      "learning_rate": 0.00012392919002336184,
+      "loss": 0.8676809072494507,
+      "step": 2470
+    },
+    {
+      "epoch": 3.339189189189189,
+      "grad_norm": 0.22221308946609497,
+      "learning_rate": 0.0001234712166903359,
+      "loss": 0.8530213832855225,
+      "step": 2471
+    },
+    {
+      "epoch": 3.3405405405405406,
+      "grad_norm": 0.24853508174419403,
+      "learning_rate": 0.00012301381347626912,
+      "loss": 0.8575125932693481,
+      "step": 2472
+    },
+    {
+      "epoch": 3.341891891891892,
+      "grad_norm": 0.2228991538286209,
+      "learning_rate": 0.00012255698244214864,
+      "loss": 0.8653073310852051,
+      "step": 2473
+    },
+    {
+      "epoch": 3.343243243243243,
+      "grad_norm": 0.24971702694892883,
+      "learning_rate": 0.0001221007256463841,
+      "loss": 0.8712753653526306,
+      "step": 2474
+    },
+    {
+      "epoch": 3.3445945945945947,
+      "grad_norm": 0.2402600646018982,
+      "learning_rate": 0.00012164504514479741,
+      "loss": 0.8602169752120972,
+      "step": 2475
+    },
+    {
+      "epoch": 3.345945945945946,
+      "grad_norm": 0.2728312313556671,
+      "learning_rate": 0.00012118994299061376,
+      "loss": 0.8681799173355103,
+      "step": 2476
+    },
+    {
+      "epoch": 3.347297297297297,
+      "grad_norm": 0.22529709339141846,
+      "learning_rate": 0.00012073542123445239,
+      "loss": 0.8521712422370911,
+      "step": 2477
+    },
+    {
+      "epoch": 3.3486486486486484,
+      "grad_norm": 0.2826855778694153,
+      "learning_rate": 0.00012028148192431771,
+      "loss": 0.8780070543289185,
+      "step": 2478
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.2616288959980011,
+      "learning_rate": 0.00011982812710558905,
+      "loss": 0.8730890154838562,
+      "step": 2479
+    },
+    {
+      "epoch": 3.3513513513513513,
+      "grad_norm": 0.2420555204153061,
+      "learning_rate": 0.00011937535882101281,
+      "loss": 0.8634241819381714,
+      "step": 2480
+    },
+    {
+      "epoch": 3.3527027027027025,
+      "grad_norm": 0.21616621315479279,
+      "learning_rate": 0.00011892317911069211,
+      "loss": 0.8533633351325989,
+      "step": 2481
+    },
+    {
+      "epoch": 3.354054054054054,
+      "grad_norm": 0.2672151029109955,
+      "learning_rate": 0.00011847159001207813,
+      "loss": 0.8716537952423096,
+      "step": 2482
+    },
+    {
+      "epoch": 3.3554054054054054,
+      "grad_norm": 0.23468241095542908,
+      "learning_rate": 0.00011802059355996118,
+      "loss": 0.862542450428009,
+      "step": 2483
+    },
+    {
+      "epoch": 3.3567567567567567,
+      "grad_norm": 0.256318598985672,
+      "learning_rate": 0.0001175701917864605,
+      "loss": 0.8685390949249268,
+      "step": 2484
+    },
+    {
+      "epoch": 3.358108108108108,
+      "grad_norm": 0.268515408039093,
+      "learning_rate": 0.00011712038672101654,
+      "loss": 0.8710564970970154,
+      "step": 2485
+    },
+    {
+      "epoch": 3.3594594594594596,
+      "grad_norm": 0.2363465428352356,
+      "learning_rate": 0.00011667118039038063,
+      "loss": 0.8616529703140259,
+      "step": 2486
+    },
+    {
+      "epoch": 3.360810810810811,
+      "grad_norm": 0.24947911500930786,
+      "learning_rate": 0.00011622257481860637,
+      "loss": 0.8654581308364868,
+      "step": 2487
+    },
+    {
+      "epoch": 3.362162162162162,
+      "grad_norm": 0.2623845338821411,
+      "learning_rate": 0.00011577457202704073,
+      "loss": 0.8674687147140503,
+      "step": 2488
+    },
+    {
+      "epoch": 3.3635135135135137,
+      "grad_norm": 0.26649558544158936,
+      "learning_rate": 0.00011532717403431403,
+      "loss": 0.8842494487762451,
+      "step": 2489
+    },
+    {
+      "epoch": 3.364864864864865,
+      "grad_norm": 0.26572132110595703,
+      "learning_rate": 0.00011488038285633213,
+      "loss": 0.8580397367477417,
+      "step": 2490
+    },
+    {
+      "epoch": 3.366216216216216,
+      "grad_norm": 0.2515789568424225,
+      "learning_rate": 0.00011443420050626624,
+      "loss": 0.8673837184906006,
+      "step": 2491
+    },
+    {
+      "epoch": 3.3675675675675674,
+      "grad_norm": 0.24646705389022827,
+      "learning_rate": 0.00011398862899454449,
+      "loss": 0.8649641871452332,
+      "step": 2492
+    },
+    {
+      "epoch": 3.368918918918919,
+      "grad_norm": 0.27583208680152893,
+      "learning_rate": 0.00011354367032884244,
+      "loss": 0.8623395562171936,
+      "step": 2493
+    },
+    {
+      "epoch": 3.3702702702702703,
+      "grad_norm": 0.290035605430603,
+      "learning_rate": 0.00011309932651407475,
+      "loss": 0.877336859703064,
+      "step": 2494
+    },
+    {
+      "epoch": 3.3716216216216215,
+      "grad_norm": 0.28128883242607117,
+      "learning_rate": 0.00011265559955238496,
+      "loss": 0.8553637266159058,
+      "step": 2495
+    },
+    {
+      "epoch": 3.372972972972973,
+      "grad_norm": 0.2674961984157562,
+      "learning_rate": 0.00011221249144313777,
+      "loss": 0.8784652948379517,
+      "step": 2496
+    },
+    {
+      "epoch": 3.3743243243243244,
+      "grad_norm": 0.279366672039032,
+      "learning_rate": 0.00011177000418290917,
+      "loss": 0.8642014265060425,
+      "step": 2497
+    },
+    {
+      "epoch": 3.3756756756756756,
+      "grad_norm": 0.24750062823295593,
+      "learning_rate": 0.0001113281397654776,
+      "loss": 0.8701891899108887,
+      "step": 2498
+    },
+    {
+      "epoch": 3.377027027027027,
+      "grad_norm": 0.24023734033107758,
+      "learning_rate": 0.00011088690018181544,
+      "loss": 0.8644943237304688,
+      "step": 2499
+    },
+    {
+      "epoch": 3.3783783783783785,
+      "grad_norm": 0.23671427369117737,
+      "learning_rate": 0.00011044628742007909,
+      "loss": 0.8656824827194214,
+      "step": 2500
+    },
+    {
+      "epoch": 3.3797297297297297,
+      "grad_norm": 0.29775694012641907,
+      "learning_rate": 0.00011000630346560118,
+      "loss": 0.8756941556930542,
+      "step": 2501
+    },
+    {
+      "epoch": 3.381081081081081,
+      "grad_norm": 0.25357308983802795,
+      "learning_rate": 0.00010956695030088069,
+      "loss": 0.8523969650268555,
+      "step": 2502
+    },
+    {
+      "epoch": 3.3824324324324326,
+      "grad_norm": 0.25807905197143555,
+      "learning_rate": 0.0001091282299055743,
+      "loss": 0.8691077828407288,
+      "step": 2503
+    },
+    {
+      "epoch": 3.383783783783784,
+      "grad_norm": 0.21109214425086975,
+      "learning_rate": 0.00010869014425648804,
+      "loss": 0.8500497341156006,
+      "step": 2504
+    },
+    {
+      "epoch": 3.385135135135135,
+      "grad_norm": 0.24143348634243011,
+      "learning_rate": 0.00010825269532756707,
+      "loss": 0.8465277552604675,
+      "step": 2505
+    },
+    {
+      "epoch": 3.3864864864864863,
+      "grad_norm": 0.23478037118911743,
+      "learning_rate": 0.0001078158850898883,
+      "loss": 0.8649934530258179,
+      "step": 2506
+    },
+    {
+      "epoch": 3.387837837837838,
+      "grad_norm": 0.26104670763015747,
+      "learning_rate": 0.00010737971551165043,
+      "loss": 0.8714638948440552,
+      "step": 2507
+    },
+    {
+      "epoch": 3.389189189189189,
+      "grad_norm": 0.25487908720970154,
+      "learning_rate": 0.00010694418855816557,
+      "loss": 0.8773446679115295,
+      "step": 2508
+    },
+    {
+      "epoch": 3.3905405405405404,
+      "grad_norm": 0.44605663418769836,
+      "learning_rate": 0.00010650930619185009,
+      "loss": 0.8915648460388184,
+      "step": 2509
+    },
+    {
+      "epoch": 3.391891891891892,
+      "grad_norm": 0.25735628604888916,
+      "learning_rate": 0.0001060750703722164,
+      "loss": 0.8720310926437378,
+      "step": 2510
+    },
+    {
+      "epoch": 3.3932432432432433,
+      "grad_norm": 0.25383830070495605,
+      "learning_rate": 0.00010564148305586297,
+      "loss": 0.8670020699501038,
+      "step": 2511
+    },
+    {
+      "epoch": 3.3945945945945946,
+      "grad_norm": 0.2715550661087036,
+      "learning_rate": 0.00010520854619646689,
+      "loss": 0.8608202934265137,
+      "step": 2512
+    },
+    {
+      "epoch": 3.395945945945946,
+      "grad_norm": 0.23813371360301971,
+      "learning_rate": 0.00010477626174477404,
+      "loss": 0.8706060647964478,
+      "step": 2513
+    },
+    {
+      "epoch": 3.3972972972972975,
+      "grad_norm": 0.25104638934135437,
+      "learning_rate": 0.0001043446316485906,
+      "loss": 0.8686371445655823,
+      "step": 2514
+    },
+    {
+      "epoch": 3.3986486486486487,
+      "grad_norm": 0.22809019684791565,
+      "learning_rate": 0.00010391365785277473,
+      "loss": 0.8601222038269043,
+      "step": 2515
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.26542654633522034,
+      "learning_rate": 0.00010348334229922676,
+      "loss": 0.872133731842041,
+      "step": 2516
+    },
+    {
+      "epoch": 3.4013513513513516,
+      "grad_norm": 0.22266772389411926,
+      "learning_rate": 0.00010305368692688174,
+      "loss": 0.8589619398117065,
+      "step": 2517
+    },
+    {
+      "epoch": 3.402702702702703,
+      "grad_norm": 0.2277406007051468,
+      "learning_rate": 0.00010262469367169963,
+      "loss": 0.8674957752227783,
+      "step": 2518
+    },
+    {
+      "epoch": 3.404054054054054,
+      "grad_norm": 0.2762044668197632,
+      "learning_rate": 0.00010219636446665703,
+      "loss": 0.8712243437767029,
+      "step": 2519
+    },
+    {
+      "epoch": 3.4054054054054053,
+      "grad_norm": 0.2423914074897766,
+      "learning_rate": 0.00010176870124173878,
+      "loss": 0.8674852252006531,
+      "step": 2520
+    },
+    {
+      "epoch": 3.406756756756757,
+      "grad_norm": 0.22598566114902496,
+      "learning_rate": 0.00010134170592392835,
+      "loss": 0.8506726026535034,
+      "step": 2521
+    },
+    {
+      "epoch": 3.408108108108108,
+      "grad_norm": 0.23059645295143127,
+      "learning_rate": 0.00010091538043720022,
+      "loss": 0.8580012917518616,
+      "step": 2522
+    },
+    {
+      "epoch": 3.4094594594594594,
+      "grad_norm": 0.22932514548301697,
+      "learning_rate": 0.0001004897267025105,
+      "loss": 0.8587660789489746,
+      "step": 2523
+    },
+    {
+      "epoch": 3.410810810810811,
+      "grad_norm": 0.2765813171863556,
+      "learning_rate": 0.00010006474663778847,
+      "loss": 0.8724586367607117,
+      "step": 2524
+    },
+    {
+      "epoch": 3.4121621621621623,
+      "grad_norm": 0.20693416893482208,
+      "learning_rate": 9.964044215792795e-05,
+      "loss": 0.8451635837554932,
+      "step": 2525
+    },
+    {
+      "epoch": 3.4135135135135135,
+      "grad_norm": 0.2136680781841278,
+      "learning_rate": 9.921681517477899e-05,
+      "loss": 0.8518712520599365,
+      "step": 2526
+    },
+    {
+      "epoch": 3.4148648648648647,
+      "grad_norm": 0.2166067510843277,
+      "learning_rate": 9.879386759713833e-05,
+      "loss": 0.8521533012390137,
+      "step": 2527
+    },
+    {
+      "epoch": 3.4162162162162164,
+      "grad_norm": 0.2293974757194519,
+      "learning_rate": 9.837160133074202e-05,
+      "loss": 0.8693077564239502,
+      "step": 2528
+    },
+    {
+      "epoch": 3.4175675675675676,
+      "grad_norm": 0.24341222643852234,
+      "learning_rate": 9.795001827825595e-05,
+      "loss": 0.8518421649932861,
+      "step": 2529
+    },
+    {
+      "epoch": 3.418918918918919,
+      "grad_norm": 0.25018471479415894,
+      "learning_rate": 9.752912033926742e-05,
+      "loss": 0.8699618577957153,
+      "step": 2530
+    },
+    {
+      "epoch": 3.42027027027027,
+      "grad_norm": 0.21593692898750305,
+      "learning_rate": 9.710890941027722e-05,
+      "loss": 0.8573335409164429,
+      "step": 2531
+    },
+    {
+      "epoch": 3.4216216216216218,
+      "grad_norm": 0.21938760578632355,
+      "learning_rate": 9.668938738468989e-05,
+      "loss": 0.8534834384918213,
+      "step": 2532
+    },
+    {
+      "epoch": 3.422972972972973,
+      "grad_norm": 0.23638474941253662,
+      "learning_rate": 9.627055615280641e-05,
+      "loss": 0.8613983988761902,
+      "step": 2533
+    },
+    {
+      "epoch": 3.424324324324324,
+      "grad_norm": 0.2804933488368988,
+      "learning_rate": 9.585241760181499e-05,
+      "loss": 0.8659054040908813,
+      "step": 2534
+    },
+    {
+      "epoch": 3.4256756756756754,
+      "grad_norm": 0.27525392174720764,
+      "learning_rate": 9.543497361578254e-05,
+      "loss": 0.8719008564949036,
+      "step": 2535
+    },
+    {
+      "epoch": 3.427027027027027,
+      "grad_norm": 0.2713029980659485,
+      "learning_rate": 9.501822607564677e-05,
+      "loss": 0.8632816076278687,
+      "step": 2536
+    },
+    {
+      "epoch": 3.4283783783783783,
+      "grad_norm": 0.241143599152565,
+      "learning_rate": 9.460217685920697e-05,
+      "loss": 0.8596347570419312,
+      "step": 2537
+    },
+    {
+      "epoch": 3.4297297297297296,
+      "grad_norm": 0.24229653179645538,
+      "learning_rate": 9.418682784111601e-05,
+      "loss": 0.8691472411155701,
+      "step": 2538
+    },
+    {
+      "epoch": 3.4310810810810812,
+      "grad_norm": 0.25162750482559204,
+      "learning_rate": 9.377218089287179e-05,
+      "loss": 0.8597682118415833,
+      "step": 2539
+    },
+    {
+      "epoch": 3.4324324324324325,
+      "grad_norm": 0.2768096327781677,
+      "learning_rate": 9.335823788280873e-05,
+      "loss": 0.8701770305633545,
+      "step": 2540
+    },
+    {
+      "epoch": 3.4337837837837837,
+      "grad_norm": 0.2619352638721466,
+      "learning_rate": 9.29450006760894e-05,
+      "loss": 0.8702062368392944,
+      "step": 2541
+    },
+    {
+      "epoch": 3.435135135135135,
+      "grad_norm": 0.23935765027999878,
+      "learning_rate": 9.253247113469646e-05,
+      "loss": 0.8617749214172363,
+      "step": 2542
+    },
+    {
+      "epoch": 3.4364864864864866,
+      "grad_norm": 0.28716346621513367,
+      "learning_rate": 9.212065111742326e-05,
+      "loss": 0.8681508898735046,
+      "step": 2543
+    },
+    {
+      "epoch": 3.437837837837838,
+      "grad_norm": 0.22078992426395416,
+      "learning_rate": 9.170954247986691e-05,
+      "loss": 0.8613612651824951,
+      "step": 2544
+    },
+    {
+      "epoch": 3.439189189189189,
+      "grad_norm": 0.24942998588085175,
+      "learning_rate": 9.129914707441864e-05,
+      "loss": 0.8649256229400635,
+      "step": 2545
+    },
+    {
+      "epoch": 3.4405405405405407,
+      "grad_norm": 0.31352952122688293,
+      "learning_rate": 9.088946675025606e-05,
+      "loss": 0.8734772205352783,
+      "step": 2546
+    },
+    {
+      "epoch": 3.441891891891892,
+      "grad_norm": 0.2678256630897522,
+      "learning_rate": 9.048050335333505e-05,
+      "loss": 0.8649768233299255,
+      "step": 2547
+    },
+    {
+      "epoch": 3.443243243243243,
+      "grad_norm": 0.24439330399036407,
+      "learning_rate": 9.007225872638053e-05,
+      "loss": 0.860141396522522,
+      "step": 2548
+    },
+    {
+      "epoch": 3.4445945945945944,
+      "grad_norm": 0.25117260217666626,
+      "learning_rate": 8.96647347088794e-05,
+      "loss": 0.8666969537734985,
+      "step": 2549
+    },
+    {
+      "epoch": 3.445945945945946,
+      "grad_norm": 0.24473243951797485,
+      "learning_rate": 8.925793313707117e-05,
+      "loss": 0.8499115705490112,
+      "step": 2550
+    },
+    {
+      "epoch": 3.4472972972972973,
+      "grad_norm": 0.297563761472702,
+      "learning_rate": 8.88518558439402e-05,
+      "loss": 0.8896875977516174,
+      "step": 2551
+    },
+    {
+      "epoch": 3.4486486486486485,
+      "grad_norm": 0.26814550161361694,
+      "learning_rate": 8.844650465920761e-05,
+      "loss": 0.8771052956581116,
+      "step": 2552
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.22546331584453583,
+      "learning_rate": 8.804188140932252e-05,
+      "loss": 0.8462048172950745,
+      "step": 2553
+    },
+    {
+      "epoch": 3.4513513513513514,
+      "grad_norm": 0.2648927569389343,
+      "learning_rate": 8.763798791745412e-05,
+      "loss": 0.8692748546600342,
+      "step": 2554
+    },
+    {
+      "epoch": 3.4527027027027026,
+      "grad_norm": 0.2391367256641388,
+      "learning_rate": 8.72348260034835e-05,
+      "loss": 0.8738073110580444,
+      "step": 2555
+    },
+    {
+      "epoch": 3.454054054054054,
+      "grad_norm": 0.3582954406738281,
+      "learning_rate": 8.683239748399538e-05,
+      "loss": 0.8740545511245728,
+      "step": 2556
+    },
+    {
+      "epoch": 3.4554054054054055,
+      "grad_norm": 0.24431242048740387,
+      "learning_rate": 8.643070417226978e-05,
+      "loss": 0.8610081672668457,
+      "step": 2557
+    },
+    {
+      "epoch": 3.4567567567567568,
+      "grad_norm": 0.24968132376670837,
+      "learning_rate": 8.602974787827436e-05,
+      "loss": 0.8607233762741089,
+      "step": 2558
+    },
+    {
+      "epoch": 3.458108108108108,
+      "grad_norm": 0.264494389295578,
+      "learning_rate": 8.562953040865531e-05,
+      "loss": 0.8698164224624634,
+      "step": 2559
+    },
+    {
+      "epoch": 3.4594594594594597,
+      "grad_norm": 0.26323434710502625,
+      "learning_rate": 8.523005356673032e-05,
+      "loss": 0.8745170831680298,
+      "step": 2560
+    },
+    {
+      "epoch": 3.460810810810811,
+      "grad_norm": 0.238840714097023,
+      "learning_rate": 8.483131915247969e-05,
+      "loss": 0.8532969951629639,
+      "step": 2561
+    },
+    {
+      "epoch": 3.462162162162162,
+      "grad_norm": 0.2173490822315216,
+      "learning_rate": 8.443332896253836e-05,
+      "loss": 0.8567383885383606,
+      "step": 2562
+    },
+    {
+      "epoch": 3.4635135135135133,
+      "grad_norm": 0.25553181767463684,
+      "learning_rate": 8.403608479018832e-05,
+      "loss": 0.863215446472168,
+      "step": 2563
+    },
+    {
+      "epoch": 3.464864864864865,
+      "grad_norm": 0.2383936494588852,
+      "learning_rate": 8.363958842534949e-05,
+      "loss": 0.8681496381759644,
+      "step": 2564
+    },
+    {
+      "epoch": 3.4662162162162162,
+      "grad_norm": 0.24160626530647278,
+      "learning_rate": 8.324384165457289e-05,
+      "loss": 0.8705663084983826,
+      "step": 2565
+    },
+    {
+      "epoch": 3.4675675675675675,
+      "grad_norm": 0.25379857420921326,
+      "learning_rate": 8.284884626103165e-05,
+      "loss": 0.8539824485778809,
+      "step": 2566
+    },
+    {
+      "epoch": 3.468918918918919,
+      "grad_norm": 0.26047998666763306,
+      "learning_rate": 8.24546040245133e-05,
+      "loss": 0.8691788911819458,
+      "step": 2567
+    },
+    {
+      "epoch": 3.4702702702702704,
+      "grad_norm": 0.21852974593639374,
+      "learning_rate": 8.206111672141204e-05,
+      "loss": 0.8594878911972046,
+      "step": 2568
+    },
+    {
+      "epoch": 3.4716216216216216,
+      "grad_norm": 0.2360246181488037,
+      "learning_rate": 8.166838612472019e-05,
+      "loss": 0.8519748449325562,
+      "step": 2569
+    },
+    {
+      "epoch": 3.472972972972973,
+      "grad_norm": 0.22827701270580292,
+      "learning_rate": 8.127641400402053e-05,
+      "loss": 0.8597067594528198,
+      "step": 2570
+    },
+    {
+      "epoch": 3.4743243243243245,
+      "grad_norm": 0.24562479555606842,
+      "learning_rate": 8.088520212547831e-05,
+      "loss": 0.8567913770675659,
+      "step": 2571
+    },
+    {
+      "epoch": 3.4756756756756757,
+      "grad_norm": 0.22964097559452057,
+      "learning_rate": 8.049475225183323e-05,
+      "loss": 0.8639025688171387,
+      "step": 2572
+    },
+    {
+      "epoch": 3.477027027027027,
+      "grad_norm": 0.25030308961868286,
+      "learning_rate": 8.010506614239138e-05,
+      "loss": 0.8562885522842407,
+      "step": 2573
+    },
+    {
+      "epoch": 3.4783783783783786,
+      "grad_norm": 0.2492954283952713,
+      "learning_rate": 7.97161455530179e-05,
+      "loss": 0.8655052781105042,
+      "step": 2574
+    },
+    {
+      "epoch": 3.47972972972973,
+      "grad_norm": 0.23418836295604706,
+      "learning_rate": 7.932799223612788e-05,
+      "loss": 0.8626313805580139,
+      "step": 2575
+    },
+    {
+      "epoch": 3.481081081081081,
+      "grad_norm": 0.2748301327228546,
+      "learning_rate": 7.894060794067987e-05,
+      "loss": 0.8559788465499878,
+      "step": 2576
+    },
+    {
+      "epoch": 3.4824324324324323,
+      "grad_norm": 0.2487088292837143,
+      "learning_rate": 7.855399441216716e-05,
+      "loss": 0.8609334826469421,
+      "step": 2577
+    },
+    {
+      "epoch": 3.483783783783784,
+      "grad_norm": 0.22755350172519684,
+      "learning_rate": 7.816815339260972e-05,
+      "loss": 0.8608084917068481,
+      "step": 2578
+    },
+    {
+      "epoch": 3.485135135135135,
+      "grad_norm": 0.26166096329689026,
+      "learning_rate": 7.77830866205472e-05,
+      "loss": 0.8581379652023315,
+      "step": 2579
+    },
+    {
+      "epoch": 3.4864864864864864,
+      "grad_norm": 0.2274831235408783,
+      "learning_rate": 7.739879583103033e-05,
+      "loss": 0.8526138663291931,
+      "step": 2580
+    },
+    {
+      "epoch": 3.487837837837838,
+      "grad_norm": 0.20184828341007233,
+      "learning_rate": 7.701528275561349e-05,
+      "loss": 0.849195122718811,
+      "step": 2581
+    },
+    {
+      "epoch": 3.4891891891891893,
+      "grad_norm": 0.2836126983165741,
+      "learning_rate": 7.663254912234671e-05,
+      "loss": 0.8729177117347717,
+      "step": 2582
+    },
+    {
+      "epoch": 3.4905405405405405,
+      "grad_norm": 0.23263707756996155,
+      "learning_rate": 7.625059665576803e-05,
+      "loss": 0.8629276156425476,
+      "step": 2583
+    },
+    {
+      "epoch": 3.4918918918918918,
+      "grad_norm": 0.2604852616786957,
+      "learning_rate": 7.586942707689578e-05,
+      "loss": 0.8726860880851746,
+      "step": 2584
+    },
+    {
+      "epoch": 3.4932432432432434,
+      "grad_norm": 0.23029673099517822,
+      "learning_rate": 7.548904210322058e-05,
+      "loss": 0.8571387529373169,
+      "step": 2585
+    },
+    {
+      "epoch": 3.4945945945945946,
+      "grad_norm": 0.25274819135665894,
+      "learning_rate": 7.510944344869774e-05,
+      "loss": 0.8565754294395447,
+      "step": 2586
+    },
+    {
+      "epoch": 3.495945945945946,
+      "grad_norm": 0.23141194880008698,
+      "learning_rate": 7.47306328237396e-05,
+      "loss": 0.8714017868041992,
+      "step": 2587
+    },
+    {
+      "epoch": 3.4972972972972975,
+      "grad_norm": 0.28116753697395325,
+      "learning_rate": 7.435261193520773e-05,
+      "loss": 0.8711215257644653,
+      "step": 2588
+    },
+    {
+      "epoch": 3.4986486486486488,
+      "grad_norm": 0.22799846529960632,
+      "learning_rate": 7.397538248640526e-05,
+      "loss": 0.853801965713501,
+      "step": 2589
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.2770105302333832,
+      "learning_rate": 7.359894617706939e-05,
+      "loss": 0.8680166006088257,
+      "step": 2590
+    },
+    {
+      "epoch": 3.5013513513513512,
+      "grad_norm": 0.2404516190290451,
+      "learning_rate": 7.322330470336314e-05,
+      "loss": 0.8583754301071167,
+      "step": 2591
+    },
+    {
+      "epoch": 3.5027027027027025,
+      "grad_norm": 0.23749877512454987,
+      "learning_rate": 7.284845975786853e-05,
+      "loss": 0.8582051992416382,
+      "step": 2592
+    },
+    {
+      "epoch": 3.504054054054054,
+      "grad_norm": 0.26335328817367554,
+      "learning_rate": 7.247441302957858e-05,
+      "loss": 0.8642014861106873,
+      "step": 2593
+    },
+    {
+      "epoch": 3.5054054054054054,
+      "grad_norm": 0.23426513373851776,
+      "learning_rate": 7.210116620388917e-05,
+      "loss": 0.8664218187332153,
+      "step": 2594
+    },
+    {
+      "epoch": 3.506756756756757,
+      "grad_norm": 0.282230943441391,
+      "learning_rate": 7.172872096259253e-05,
+      "loss": 0.8702486753463745,
+      "step": 2595
+    },
+    {
+      "epoch": 3.5081081081081082,
+      "grad_norm": 0.22521956264972687,
+      "learning_rate": 7.13570789838687e-05,
+      "loss": 0.8527774810791016,
+      "step": 2596
+    },
+    {
+      "epoch": 3.5094594594594595,
+      "grad_norm": 0.32148319482803345,
+      "learning_rate": 7.098624194227845e-05,
+      "loss": 0.867697536945343,
+      "step": 2597
+    },
+    {
+      "epoch": 3.5108108108108107,
+      "grad_norm": 0.22331486642360687,
+      "learning_rate": 7.061621150875569e-05,
+      "loss": 0.855079174041748,
+      "step": 2598
+    },
+    {
+      "epoch": 3.512162162162162,
+      "grad_norm": 0.2249329686164856,
+      "learning_rate": 7.024698935059981e-05,
+      "loss": 0.8634952306747437,
+      "step": 2599
+    },
+    {
+      "epoch": 3.5135135135135136,
+      "grad_norm": 0.24223804473876953,
+      "learning_rate": 6.987857713146817e-05,
+      "loss": 0.8509316444396973,
+      "step": 2600
+    },
+    {
+      "epoch": 3.514864864864865,
+      "grad_norm": 0.304464191198349,
+      "learning_rate": 6.95109765113689e-05,
+      "loss": 0.8602609038352966,
+      "step": 2601
+    },
+    {
+      "epoch": 3.516216216216216,
+      "grad_norm": 0.2513377070426941,
+      "learning_rate": 6.914418914665299e-05,
+      "loss": 0.8676443099975586,
+      "step": 2602
+    },
+    {
+      "epoch": 3.5175675675675677,
+      "grad_norm": 0.25975245237350464,
+      "learning_rate": 6.877821669000705e-05,
+      "loss": 0.8784201145172119,
+      "step": 2603
+    },
+    {
+      "epoch": 3.518918918918919,
+      "grad_norm": 0.2122911959886551,
+      "learning_rate": 6.841306079044596e-05,
+      "loss": 0.8546497821807861,
+      "step": 2604
+    },
+    {
+      "epoch": 3.52027027027027,
+      "grad_norm": 0.2480059117078781,
+      "learning_rate": 6.804872309330506e-05,
+      "loss": 0.8510614037513733,
+      "step": 2605
+    },
+    {
+      "epoch": 3.5216216216216214,
+      "grad_norm": 0.21087580919265747,
+      "learning_rate": 6.768520524023347e-05,
+      "loss": 0.8464944958686829,
+      "step": 2606
+    },
+    {
+      "epoch": 3.522972972972973,
+      "grad_norm": 0.2324579954147339,
+      "learning_rate": 6.732250886918562e-05,
+      "loss": 0.8623895049095154,
+      "step": 2607
+    },
+    {
+      "epoch": 3.5243243243243243,
+      "grad_norm": 0.24997439980506897,
+      "learning_rate": 6.69606356144149e-05,
+      "loss": 0.8558083176612854,
+      "step": 2608
+    },
+    {
+      "epoch": 3.5256756756756755,
+      "grad_norm": 0.2600928544998169,
+      "learning_rate": 6.65995871064659e-05,
+      "loss": 0.863627552986145,
+      "step": 2609
+    },
+    {
+      "epoch": 3.527027027027027,
+      "grad_norm": 0.2585203945636749,
+      "learning_rate": 6.623936497216663e-05,
+      "loss": 0.8711979389190674,
+      "step": 2610
+    },
+    {
+      "epoch": 3.5283783783783784,
+      "grad_norm": 0.23898598551750183,
+      "learning_rate": 6.587997083462196e-05,
+      "loss": 0.858191967010498,
+      "step": 2611
+    },
+    {
+      "epoch": 3.5297297297297296,
+      "grad_norm": 0.24870900809764862,
+      "learning_rate": 6.55214063132058e-05,
+      "loss": 0.8516863584518433,
+      "step": 2612
+    },
+    {
+      "epoch": 3.531081081081081,
+      "grad_norm": 0.24700656533241272,
+      "learning_rate": 6.516367302355392e-05,
+      "loss": 0.864006519317627,
+      "step": 2613
+    },
+    {
+      "epoch": 3.5324324324324325,
+      "grad_norm": 0.22042971849441528,
+      "learning_rate": 6.480677257755671e-05,
+      "loss": 0.8637567758560181,
+      "step": 2614
+    },
+    {
+      "epoch": 3.5337837837837838,
+      "grad_norm": 0.23715583980083466,
+      "learning_rate": 6.445070658335195e-05,
+      "loss": 0.8699110746383667,
+      "step": 2615
+    },
+    {
+      "epoch": 3.535135135135135,
+      "grad_norm": 0.22728890180587769,
+      "learning_rate": 6.409547664531735e-05,
+      "loss": 0.8660768270492554,
+      "step": 2616
+    },
+    {
+      "epoch": 3.5364864864864867,
+      "grad_norm": 0.27524542808532715,
+      "learning_rate": 6.374108436406373e-05,
+      "loss": 0.8789306879043579,
+      "step": 2617
+    },
+    {
+      "epoch": 3.537837837837838,
+      "grad_norm": 0.24115294218063354,
+      "learning_rate": 6.338753133642738e-05,
+      "loss": 0.8569843173027039,
+      "step": 2618
+    },
+    {
+      "epoch": 3.539189189189189,
+      "grad_norm": 0.26138654351234436,
+      "learning_rate": 6.303481915546299e-05,
+      "loss": 0.8615937232971191,
+      "step": 2619
+    },
+    {
+      "epoch": 3.5405405405405403,
+      "grad_norm": 0.1937621831893921,
+      "learning_rate": 6.268294941043687e-05,
+      "loss": 0.8515458106994629,
+      "step": 2620
+    },
+    {
+      "epoch": 3.541891891891892,
+      "grad_norm": 0.2621687352657318,
+      "learning_rate": 6.23319236868189e-05,
+      "loss": 0.8604860901832581,
+      "step": 2621
+    },
+    {
+      "epoch": 3.5432432432432432,
+      "grad_norm": 0.2583198845386505,
+      "learning_rate": 6.198174356627645e-05,
+      "loss": 0.872991681098938,
+      "step": 2622
+    },
+    {
+      "epoch": 3.5445945945945945,
+      "grad_norm": 0.2025819569826126,
+      "learning_rate": 6.163241062666641e-05,
+      "loss": 0.8512885570526123,
+      "step": 2623
+    },
+    {
+      "epoch": 3.545945945945946,
+      "grad_norm": 0.2649320662021637,
+      "learning_rate": 6.128392644202848e-05,
+      "loss": 0.8598507046699524,
+      "step": 2624
+    },
+    {
+      "epoch": 3.5472972972972974,
+      "grad_norm": 0.24546624720096588,
+      "learning_rate": 6.0936292582578215e-05,
+      "loss": 0.8533722162246704,
+      "step": 2625
+    },
+    {
+      "epoch": 3.5486486486486486,
+      "grad_norm": 0.20967160165309906,
+      "learning_rate": 6.0589510614699305e-05,
+      "loss": 0.8488725423812866,
+      "step": 2626
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.22892723977565765,
+      "learning_rate": 6.024358210093736e-05,
+      "loss": 0.8559744358062744,
+      "step": 2627
+    },
+    {
+      "epoch": 3.5513513513513515,
+      "grad_norm": 0.2703656256198883,
+      "learning_rate": 5.989850859999227e-05,
+      "loss": 0.8714434504508972,
+      "step": 2628
+    },
+    {
+      "epoch": 3.5527027027027027,
+      "grad_norm": 0.263184130191803,
+      "learning_rate": 5.955429166671139e-05,
+      "loss": 0.8625983595848083,
+      "step": 2629
+    },
+    {
+      "epoch": 3.554054054054054,
+      "grad_norm": 0.30168768763542175,
+      "learning_rate": 5.921093285208254e-05,
+      "loss": 0.874972939491272,
+      "step": 2630
+    },
+    {
+      "epoch": 3.5554054054054056,
+      "grad_norm": 0.25695356726646423,
+      "learning_rate": 5.886843370322692e-05,
+      "loss": 0.8618142604827881,
+      "step": 2631
+    },
+    {
+      "epoch": 3.556756756756757,
+      "grad_norm": 0.23423981666564941,
+      "learning_rate": 5.8526795763392235e-05,
+      "loss": 0.8641743063926697,
+      "step": 2632
+    },
+    {
+      "epoch": 3.558108108108108,
+      "grad_norm": 0.26692166924476624,
+      "learning_rate": 5.818602057194589e-05,
+      "loss": 0.8648234605789185,
+      "step": 2633
+    },
+    {
+      "epoch": 3.5594594594594593,
+      "grad_norm": 0.24087733030319214,
+      "learning_rate": 5.784610966436765e-05,
+      "loss": 0.8565750122070312,
+      "step": 2634
+    },
+    {
+      "epoch": 3.560810810810811,
+      "grad_norm": 0.2534603774547577,
+      "learning_rate": 5.7507064572242976e-05,
+      "loss": 0.8695438504219055,
+      "step": 2635
+    },
+    {
+      "epoch": 3.562162162162162,
+      "grad_norm": 0.21048572659492493,
+      "learning_rate": 5.7168886823256355e-05,
+      "loss": 0.8557475805282593,
+      "step": 2636
+    },
+    {
+      "epoch": 3.5635135135135134,
+      "grad_norm": 0.2161572426557541,
+      "learning_rate": 5.683157794118371e-05,
+      "loss": 0.8584903478622437,
+      "step": 2637
+    },
+    {
+      "epoch": 3.564864864864865,
+      "grad_norm": 0.3246324062347412,
+      "learning_rate": 5.6495139445886465e-05,
+      "loss": 0.8811219334602356,
+      "step": 2638
+    },
+    {
+      "epoch": 3.5662162162162163,
+      "grad_norm": 0.2212764024734497,
+      "learning_rate": 5.6159572853303864e-05,
+      "loss": 0.8536820411682129,
+      "step": 2639
+    },
+    {
+      "epoch": 3.5675675675675675,
+      "grad_norm": 0.2634066641330719,
+      "learning_rate": 5.582487967544664e-05,
+      "loss": 0.8679936528205872,
+      "step": 2640
+    },
+    {
+      "epoch": 3.5689189189189188,
+      "grad_norm": 0.24105414748191833,
+      "learning_rate": 5.549106142039018e-05,
+      "loss": 0.8691647052764893,
+      "step": 2641
+    },
+    {
+      "epoch": 3.57027027027027,
+      "grad_norm": 0.21220137178897858,
+      "learning_rate": 5.515811959226722e-05,
+      "loss": 0.8521804809570312,
+      "step": 2642
+    },
+    {
+      "epoch": 3.5716216216216217,
+      "grad_norm": 0.2234511822462082,
+      "learning_rate": 5.4826055691261864e-05,
+      "loss": 0.8620285391807556,
+      "step": 2643
+    },
+    {
+      "epoch": 3.572972972972973,
+      "grad_norm": 0.23759201169013977,
+      "learning_rate": 5.449487121360225e-05,
+      "loss": 0.8628519773483276,
+      "step": 2644
+    },
+    {
+      "epoch": 3.5743243243243246,
+      "grad_norm": 0.19822268187999725,
+      "learning_rate": 5.416456765155392e-05,
+      "loss": 0.8529448509216309,
+      "step": 2645
+    },
+    {
+      "epoch": 3.575675675675676,
+      "grad_norm": 0.27974170446395874,
+      "learning_rate": 5.38351464934132e-05,
+      "loss": 0.8585306406021118,
+      "step": 2646
+    },
+    {
+      "epoch": 3.577027027027027,
+      "grad_norm": 0.2654901444911957,
+      "learning_rate": 5.3506609223500477e-05,
+      "loss": 0.8569751977920532,
+      "step": 2647
+    },
+    {
+      "epoch": 3.5783783783783782,
+      "grad_norm": 0.2615439295768738,
+      "learning_rate": 5.3178957322153304e-05,
+      "loss": 0.8549971580505371,
+      "step": 2648
+    },
+    {
+      "epoch": 3.5797297297297295,
+      "grad_norm": 0.27609607577323914,
+      "learning_rate": 5.285219226572022e-05,
+      "loss": 0.8636960983276367,
+      "step": 2649
+    },
+    {
+      "epoch": 3.581081081081081,
+      "grad_norm": 0.230943500995636,
+      "learning_rate": 5.252631552655351e-05,
+      "loss": 0.8581464290618896,
+      "step": 2650
+    },
+    {
+      "epoch": 3.5824324324324324,
+      "grad_norm": 0.2452688217163086,
+      "learning_rate": 5.220132857300286e-05,
+      "loss": 0.8674389123916626,
+      "step": 2651
+    },
+    {
+      "epoch": 3.583783783783784,
+      "grad_norm": 0.259328156709671,
+      "learning_rate": 5.187723286940899e-05,
+      "loss": 0.8854425549507141,
+      "step": 2652
+    },
+    {
+      "epoch": 3.5851351351351353,
+      "grad_norm": 0.2271302491426468,
+      "learning_rate": 5.155402987609628e-05,
+      "loss": 0.859516441822052,
+      "step": 2653
+    },
+    {
+      "epoch": 3.5864864864864865,
+      "grad_norm": 0.27370166778564453,
+      "learning_rate": 5.123172104936724e-05,
+      "loss": 0.8780825138092041,
+      "step": 2654
+    },
+    {
+      "epoch": 3.5878378378378377,
+      "grad_norm": 0.2331799417734146,
+      "learning_rate": 5.0910307841495085e-05,
+      "loss": 0.8595250844955444,
+      "step": 2655
+    },
+    {
+      "epoch": 3.589189189189189,
+      "grad_norm": 0.25112366676330566,
+      "learning_rate": 5.0589791700717537e-05,
+      "loss": 0.8619415760040283,
+      "step": 2656
+    },
+    {
+      "epoch": 3.5905405405405406,
+      "grad_norm": 0.26083987951278687,
+      "learning_rate": 5.027017407123047e-05,
+      "loss": 0.8716884851455688,
+      "step": 2657
+    },
+    {
+      "epoch": 3.591891891891892,
+      "grad_norm": 0.24041712284088135,
+      "learning_rate": 4.995145639318085e-05,
+      "loss": 0.8697226643562317,
+      "step": 2658
+    },
+    {
+      "epoch": 3.5932432432432435,
+      "grad_norm": 0.23056897521018982,
+      "learning_rate": 4.963364010266097e-05,
+      "loss": 0.8591373562812805,
+      "step": 2659
+    },
+    {
+      "epoch": 3.5945945945945947,
+      "grad_norm": 0.24147137999534607,
+      "learning_rate": 4.931672663170145e-05,
+      "loss": 0.866375744342804,
+      "step": 2660
+    },
+    {
+      "epoch": 3.595945945945946,
+      "grad_norm": 0.22505447268486023,
+      "learning_rate": 4.900071740826489e-05,
+      "loss": 0.8449146151542664,
+      "step": 2661
+    },
+    {
+      "epoch": 3.597297297297297,
+      "grad_norm": 0.24323715269565582,
+      "learning_rate": 4.8685613856239513e-05,
+      "loss": 0.8640052080154419,
+      "step": 2662
+    },
+    {
+      "epoch": 3.5986486486486484,
+      "grad_norm": 0.21837688982486725,
+      "learning_rate": 4.837141739543299e-05,
+      "loss": 0.8565473556518555,
+      "step": 2663
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.24551840126514435,
+      "learning_rate": 4.805812944156526e-05,
+      "loss": 0.86920166015625,
+      "step": 2664
+    },
+    {
+      "epoch": 3.6013513513513513,
+      "grad_norm": 0.3077377378940582,
+      "learning_rate": 4.7745751406263163e-05,
+      "loss": 0.8786932229995728,
+      "step": 2665
+    },
+    {
+      "epoch": 3.6027027027027025,
+      "grad_norm": 0.2570001482963562,
+      "learning_rate": 4.7434284697053354e-05,
+      "loss": 0.8658260703086853,
+      "step": 2666
+    },
+    {
+      "epoch": 3.604054054054054,
+      "grad_norm": 0.27508968114852905,
+      "learning_rate": 4.712373071735615e-05,
+      "loss": 0.8708093166351318,
+      "step": 2667
+    },
+    {
+      "epoch": 3.6054054054054054,
+      "grad_norm": 0.22837017476558685,
+      "learning_rate": 4.681409086647956e-05,
+      "loss": 0.8568570613861084,
+      "step": 2668
+    },
+    {
+      "epoch": 3.6067567567567567,
+      "grad_norm": 0.2694655954837799,
+      "learning_rate": 4.6505366539612155e-05,
+      "loss": 0.8813218474388123,
+      "step": 2669
+    },
+    {
+      "epoch": 3.608108108108108,
+      "grad_norm": 0.24789981544017792,
+      "learning_rate": 4.619755912781779e-05,
+      "loss": 0.8616808652877808,
+      "step": 2670
+    },
+    {
+      "epoch": 3.6094594594594596,
+      "grad_norm": 0.23303234577178955,
+      "learning_rate": 4.5890670018028604e-05,
+      "loss": 0.8573690056800842,
+      "step": 2671
+    },
+    {
+      "epoch": 3.610810810810811,
+      "grad_norm": 0.2952406108379364,
+      "learning_rate": 4.5584700593038955e-05,
+      "loss": 0.8747352361679077,
+      "step": 2672
+    },
+    {
+      "epoch": 3.612162162162162,
+      "grad_norm": 0.24640585482120514,
+      "learning_rate": 4.5279652231499574e-05,
+      "loss": 0.8728247880935669,
+      "step": 2673
+    },
+    {
+      "epoch": 3.6135135135135137,
+      "grad_norm": 0.2441546767950058,
+      "learning_rate": 4.497552630791049e-05,
+      "loss": 0.8605356216430664,
+      "step": 2674
+    },
+    {
+      "epoch": 3.614864864864865,
+      "grad_norm": 0.22988544404506683,
+      "learning_rate": 4.467232419261591e-05,
+      "loss": 0.860270082950592,
+      "step": 2675
+    },
+    {
+      "epoch": 3.616216216216216,
+      "grad_norm": 0.24679914116859436,
+      "learning_rate": 4.437004725179714e-05,
+      "loss": 0.8752086162567139,
+      "step": 2676
+    },
+    {
+      "epoch": 3.6175675675675674,
+      "grad_norm": 0.29532942175865173,
+      "learning_rate": 4.4068696847466977e-05,
+      "loss": 0.8690840005874634,
+      "step": 2677
+    },
+    {
+      "epoch": 3.618918918918919,
+      "grad_norm": 0.2281295508146286,
+      "learning_rate": 4.376827433746322e-05,
+      "loss": 0.858695387840271,
+      "step": 2678
+    },
+    {
+      "epoch": 3.6202702702702703,
+      "grad_norm": 0.23479512333869934,
+      "learning_rate": 4.3468781075443084e-05,
+      "loss": 0.8602907657623291,
+      "step": 2679
+    },
+    {
+      "epoch": 3.6216216216216215,
+      "grad_norm": 0.2949625849723816,
+      "learning_rate": 4.317021841087615e-05,
+      "loss": 0.8709598779678345,
+      "step": 2680
+    },
+    {
+      "epoch": 3.622972972972973,
+      "grad_norm": 0.26125529408454895,
+      "learning_rate": 4.287258768903948e-05,
+      "loss": 0.8609077334403992,
+      "step": 2681
+    },
+    {
+      "epoch": 3.6243243243243244,
+      "grad_norm": 0.22168302536010742,
+      "learning_rate": 4.2575890251010576e-05,
+      "loss": 0.8539755344390869,
+      "step": 2682
+    },
+    {
+      "epoch": 3.6256756756756756,
+      "grad_norm": 0.2341485619544983,
+      "learning_rate": 4.228012743366175e-05,
+      "loss": 0.8634673953056335,
+      "step": 2683
+    },
+    {
+      "epoch": 3.627027027027027,
+      "grad_norm": 0.21290335059165955,
+      "learning_rate": 4.19853005696543e-05,
+      "loss": 0.8512980937957764,
+      "step": 2684
+    },
+    {
+      "epoch": 3.6283783783783785,
+      "grad_norm": 0.2628742456436157,
+      "learning_rate": 4.1691410987431815e-05,
+      "loss": 0.8658726811408997,
+      "step": 2685
+    },
+    {
+      "epoch": 3.6297297297297297,
+      "grad_norm": 0.23015610873699188,
+      "learning_rate": 4.139846001121514e-05,
+      "loss": 0.861553430557251,
+      "step": 2686
+    },
+    {
+      "epoch": 3.631081081081081,
+      "grad_norm": 0.23172658681869507,
+      "learning_rate": 4.110644896099558e-05,
+      "loss": 0.8558934330940247,
+      "step": 2687
+    },
+    {
+      "epoch": 3.6324324324324326,
+      "grad_norm": 0.23428742587566376,
+      "learning_rate": 4.081537915252931e-05,
+      "loss": 0.8593648076057434,
+      "step": 2688
+    },
+    {
+      "epoch": 3.633783783783784,
+      "grad_norm": 0.2737089693546295,
+      "learning_rate": 4.0525251897331663e-05,
+      "loss": 0.8653750419616699,
+      "step": 2689
+    },
+    {
+      "epoch": 3.635135135135135,
+      "grad_norm": 0.2624095380306244,
+      "learning_rate": 4.0236068502670556e-05,
+      "loss": 0.8590494394302368,
+      "step": 2690
+    },
+    {
+      "epoch": 3.6364864864864863,
+      "grad_norm": 0.2543865740299225,
+      "learning_rate": 3.994783027156143e-05,
+      "loss": 0.8626692295074463,
+      "step": 2691
+    },
+    {
+      "epoch": 3.637837837837838,
+      "grad_norm": 0.24927565455436707,
+      "learning_rate": 3.966053850276072e-05,
+      "loss": 0.8612467050552368,
+      "step": 2692
+    },
+    {
+      "epoch": 3.639189189189189,
+      "grad_norm": 0.24631288647651672,
+      "learning_rate": 3.937419449076035e-05,
+      "loss": 0.8659945130348206,
+      "step": 2693
+    },
+    {
+      "epoch": 3.6405405405405404,
+      "grad_norm": 0.2660532593727112,
+      "learning_rate": 3.908879952578168e-05,
+      "loss": 0.8647119998931885,
+      "step": 2694
+    },
+    {
+      "epoch": 3.641891891891892,
+      "grad_norm": 0.22153107821941376,
+      "learning_rate": 3.880435489377007e-05,
+      "loss": 0.8549377918243408,
+      "step": 2695
+    },
+    {
+      "epoch": 3.6432432432432433,
+      "grad_norm": 0.2550741732120514,
+      "learning_rate": 3.852086187638845e-05,
+      "loss": 0.8823832869529724,
+      "step": 2696
+    },
+    {
+      "epoch": 3.6445945945945946,
+      "grad_norm": 0.233835369348526,
+      "learning_rate": 3.82383217510123e-05,
+      "loss": 0.8523450493812561,
+      "step": 2697
+    },
+    {
+      "epoch": 3.645945945945946,
+      "grad_norm": 0.21111544966697693,
+      "learning_rate": 3.7956735790723286e-05,
+      "loss": 0.848663330078125,
+      "step": 2698
+    },
+    {
+      "epoch": 3.6472972972972975,
+      "grad_norm": 0.21229945123195648,
+      "learning_rate": 3.767610526430373e-05,
+      "loss": 0.8477547764778137,
+      "step": 2699
+    },
+    {
+      "epoch": 3.6486486486486487,
+      "grad_norm": 0.24837704002857208,
+      "learning_rate": 3.7396431436231256e-05,
+      "loss": 0.8652114868164062,
+      "step": 2700
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.2717590630054474,
+      "learning_rate": 3.711771556667218e-05,
+      "loss": 0.8773582577705383,
+      "step": 2701
+    },
+    {
+      "epoch": 3.6513513513513516,
+      "grad_norm": 0.2289438396692276,
+      "learning_rate": 3.6839958911476953e-05,
+      "loss": 0.8608661890029907,
+      "step": 2702
+    },
+    {
+      "epoch": 3.652702702702703,
+      "grad_norm": 0.2610377371311188,
+      "learning_rate": 3.6563162722173696e-05,
+      "loss": 0.8754374384880066,
+      "step": 2703
+    },
+    {
+      "epoch": 3.654054054054054,
+      "grad_norm": 0.19842903316020966,
+      "learning_rate": 3.628732824596273e-05,
+      "loss": 0.8406215906143188,
+      "step": 2704
+    },
+    {
+      "epoch": 3.6554054054054053,
+      "grad_norm": 0.2513572573661804,
+      "learning_rate": 3.6012456725711437e-05,
+      "loss": 0.8662301301956177,
+      "step": 2705
+    },
+    {
+      "epoch": 3.6567567567567565,
+      "grad_norm": 0.2591841220855713,
+      "learning_rate": 3.57385493999477e-05,
+      "loss": 0.8577200174331665,
+      "step": 2706
+    },
+    {
+      "epoch": 3.658108108108108,
+      "grad_norm": 0.22577618062496185,
+      "learning_rate": 3.546560750285541e-05,
+      "loss": 0.8704397678375244,
+      "step": 2707
+    },
+    {
+      "epoch": 3.6594594594594594,
+      "grad_norm": 0.2133210301399231,
+      "learning_rate": 3.519363226426808e-05,
+      "loss": 0.8453714847564697,
+      "step": 2708
+    },
+    {
+      "epoch": 3.660810810810811,
+      "grad_norm": 0.28403595089912415,
+      "learning_rate": 3.492262490966377e-05,
+      "loss": 0.8821121454238892,
+      "step": 2709
+    },
+    {
+      "epoch": 3.6621621621621623,
+      "grad_norm": 0.2064274102449417,
+      "learning_rate": 3.465258666015925e-05,
+      "loss": 0.8492337465286255,
+      "step": 2710
+    },
+    {
+      "epoch": 3.6635135135135135,
+      "grad_norm": 0.22381411492824554,
+      "learning_rate": 3.438351873250492e-05,
+      "loss": 0.848127543926239,
+      "step": 2711
+    },
+    {
+      "epoch": 3.6648648648648647,
+      "grad_norm": 0.24247613549232483,
+      "learning_rate": 3.411542233907866e-05,
+      "loss": 0.8648490905761719,
+      "step": 2712
+    },
+    {
+      "epoch": 3.666216216216216,
+      "grad_norm": 0.25872495770454407,
+      "learning_rate": 3.384829868788114e-05,
+      "loss": 0.8583657741546631,
+      "step": 2713
+    },
+    {
+      "epoch": 3.6675675675675676,
+      "grad_norm": 0.23001796007156372,
+      "learning_rate": 3.358214898252987e-05,
+      "loss": 0.8616563081741333,
+      "step": 2714
+    },
+    {
+      "epoch": 3.668918918918919,
+      "grad_norm": 0.22298818826675415,
+      "learning_rate": 3.331697442225376e-05,
+      "loss": 0.8544929623603821,
+      "step": 2715
+    },
+    {
+      "epoch": 3.6702702702702705,
+      "grad_norm": 0.24740351736545563,
+      "learning_rate": 3.305277620188826e-05,
+      "loss": 0.8743737936019897,
+      "step": 2716
+    },
+    {
+      "epoch": 3.6716216216216218,
+      "grad_norm": 0.22889117896556854,
+      "learning_rate": 3.278955551186904e-05,
+      "loss": 0.8573654294013977,
+      "step": 2717
+    },
+    {
+      "epoch": 3.672972972972973,
+      "grad_norm": 0.2719807028770447,
+      "learning_rate": 3.2527313538227684e-05,
+      "loss": 0.8691202998161316,
+      "step": 2718
+    },
+    {
+      "epoch": 3.674324324324324,
+      "grad_norm": 0.22987744212150574,
+      "learning_rate": 3.226605146258557e-05,
+      "loss": 0.8668116331100464,
+      "step": 2719
+    },
+    {
+      "epoch": 3.6756756756756754,
+      "grad_norm": 0.24414892494678497,
+      "learning_rate": 3.2005770462148754e-05,
+      "loss": 0.852127194404602,
+      "step": 2720
+    },
+    {
+      "epoch": 3.677027027027027,
+      "grad_norm": 0.2162362039089203,
+      "learning_rate": 3.174647170970296e-05,
+      "loss": 0.8519914150238037,
+      "step": 2721
+    },
+    {
+      "epoch": 3.6783783783783783,
+      "grad_norm": 0.22545860707759857,
+      "learning_rate": 3.148815637360783e-05,
+      "loss": 0.8449630737304688,
+      "step": 2722
+    },
+    {
+      "epoch": 3.67972972972973,
+      "grad_norm": 0.2253289669752121,
+      "learning_rate": 3.1230825617792006e-05,
+      "loss": 0.8637599945068359,
+      "step": 2723
+    },
+    {
+      "epoch": 3.6810810810810812,
+      "grad_norm": 0.2501432001590729,
+      "learning_rate": 3.097448060174771e-05,
+      "loss": 0.8892085552215576,
+      "step": 2724
+    },
+    {
+      "epoch": 3.6824324324324325,
+      "grad_norm": 0.2204691469669342,
+      "learning_rate": 3.071912248052561e-05,
+      "loss": 0.8576047420501709,
+      "step": 2725
+    },
+    {
+      "epoch": 3.6837837837837837,
+      "grad_norm": 0.24868755042552948,
+      "learning_rate": 3.0464752404729485e-05,
+      "loss": 0.8642221689224243,
+      "step": 2726
+    },
+    {
+      "epoch": 3.685135135135135,
+      "grad_norm": 0.2559647858142853,
+      "learning_rate": 3.02113715205114e-05,
+      "loss": 0.8698571920394897,
+      "step": 2727
+    },
+    {
+      "epoch": 3.6864864864864866,
+      "grad_norm": 0.22164326906204224,
+      "learning_rate": 2.995898096956587e-05,
+      "loss": 0.8496715426445007,
+      "step": 2728
+    },
+    {
+      "epoch": 3.687837837837838,
+      "grad_norm": 0.27178990840911865,
+      "learning_rate": 2.9707581889125506e-05,
+      "loss": 0.8691734075546265,
+      "step": 2729
+    },
+    {
+      "epoch": 3.689189189189189,
+      "grad_norm": 0.27556541562080383,
+      "learning_rate": 2.945717541195528e-05,
+      "loss": 0.8541702032089233,
+      "step": 2730
+    },
+    {
+      "epoch": 3.6905405405405407,
+      "grad_norm": 0.24379758536815643,
+      "learning_rate": 2.920776266634767e-05,
+      "loss": 0.8749319314956665,
+      "step": 2731
+    },
+    {
+      "epoch": 3.691891891891892,
+      "grad_norm": 0.2389015108346939,
+      "learning_rate": 2.8959344776117752e-05,
+      "loss": 0.8695369958877563,
+      "step": 2732
+    },
+    {
+      "epoch": 3.693243243243243,
+      "grad_norm": 0.24034276604652405,
+      "learning_rate": 2.8711922860597596e-05,
+      "loss": 0.8664259910583496,
+      "step": 2733
+    },
+    {
+      "epoch": 3.6945945945945944,
+      "grad_norm": 0.19041305780410767,
+      "learning_rate": 2.8465498034631886e-05,
+      "loss": 0.8453043699264526,
+      "step": 2734
+    },
+    {
+      "epoch": 3.695945945945946,
+      "grad_norm": 0.26550906896591187,
+      "learning_rate": 2.8220071408572412e-05,
+      "loss": 0.8827742338180542,
+      "step": 2735
+    },
+    {
+      "epoch": 3.6972972972972973,
+      "grad_norm": 0.2834649980068207,
+      "learning_rate": 2.797564408827319e-05,
+      "loss": 0.8661600351333618,
+      "step": 2736
+    },
+    {
+      "epoch": 3.6986486486486485,
+      "grad_norm": 0.25028884410858154,
+      "learning_rate": 2.7732217175085727e-05,
+      "loss": 0.8621430993080139,
+      "step": 2737
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.2350788563489914,
+      "learning_rate": 2.7489791765853646e-05,
+      "loss": 0.850670337677002,
+      "step": 2738
+    },
+    {
+      "epoch": 3.7013513513513514,
+      "grad_norm": 0.20619527995586395,
+      "learning_rate": 2.7248368952908055e-05,
+      "loss": 0.847693145275116,
+      "step": 2739
+    },
+    {
+      "epoch": 3.7027027027027026,
+      "grad_norm": 0.2342010736465454,
+      "learning_rate": 2.7007949824062434e-05,
+      "loss": 0.8660016655921936,
+      "step": 2740
+    },
+    {
+      "epoch": 3.704054054054054,
+      "grad_norm": 0.2735770642757416,
+      "learning_rate": 2.6768535462607907e-05,
+      "loss": 0.8743529915809631,
+      "step": 2741
+    },
+    {
+      "epoch": 3.7054054054054055,
+      "grad_norm": 0.2652524411678314,
+      "learning_rate": 2.653012694730819e-05,
+      "loss": 0.8573848009109497,
+      "step": 2742
+    },
+    {
+      "epoch": 3.7067567567567568,
+      "grad_norm": 0.23533514142036438,
+      "learning_rate": 2.629272535239499e-05,
+      "loss": 0.8643157482147217,
+      "step": 2743
+    },
+    {
+      "epoch": 3.708108108108108,
+      "grad_norm": 0.24755093455314636,
+      "learning_rate": 2.6056331747562668e-05,
+      "loss": 0.8689997792243958,
+      "step": 2744
+    },
+    {
+      "epoch": 3.7094594594594597,
+      "grad_norm": 0.2998432517051697,
+      "learning_rate": 2.58209471979641e-05,
+      "loss": 0.8679542541503906,
+      "step": 2745
+    },
+    {
+      "epoch": 3.710810810810811,
+      "grad_norm": 0.272763192653656,
+      "learning_rate": 2.5586572764205258e-05,
+      "loss": 0.8646060824394226,
+      "step": 2746
+    },
+    {
+      "epoch": 3.712162162162162,
+      "grad_norm": 0.24997828900814056,
+      "learning_rate": 2.535320950234074e-05,
+      "loss": 0.8668996095657349,
+      "step": 2747
+    },
+    {
+      "epoch": 3.7135135135135133,
+      "grad_norm": 0.2154756635427475,
+      "learning_rate": 2.5120858463869188e-05,
+      "loss": 0.8601770401000977,
+      "step": 2748
+    },
+    {
+      "epoch": 3.714864864864865,
+      "grad_norm": 0.28154465556144714,
+      "learning_rate": 2.488952069572789e-05,
+      "loss": 0.8662179708480835,
+      "step": 2749
+    },
+    {
+      "epoch": 3.7162162162162162,
+      "grad_norm": 0.22393879294395447,
+      "learning_rate": 2.4659197240288893e-05,
+      "loss": 0.8553083539009094,
+      "step": 2750
+    },
+    {
+      "epoch": 3.7175675675675675,
+      "grad_norm": 0.18406270444393158,
+      "learning_rate": 2.442988913535374e-05,
+      "loss": 0.8561198115348816,
+      "step": 2751
+    },
+    {
+      "epoch": 3.718918918918919,
+      "grad_norm": 0.2355114370584488,
+      "learning_rate": 2.4201597414148873e-05,
+      "loss": 0.8534416556358337,
+      "step": 2752
+    },
+    {
+      "epoch": 3.7202702702702704,
+      "grad_norm": 0.2732534110546112,
+      "learning_rate": 2.397432310532133e-05,
+      "loss": 0.8569135665893555,
+      "step": 2753
+    },
+    {
+      "epoch": 3.7216216216216216,
+      "grad_norm": 0.1892644762992859,
+      "learning_rate": 2.374806723293363e-05,
+      "loss": 0.8432250022888184,
+      "step": 2754
+    },
+    {
+      "epoch": 3.722972972972973,
+      "grad_norm": 0.2560158371925354,
+      "learning_rate": 2.3522830816459395e-05,
+      "loss": 0.8666621446609497,
+      "step": 2755
+    },
+    {
+      "epoch": 3.7243243243243245,
+      "grad_norm": 0.22973574697971344,
+      "learning_rate": 2.3298614870778834e-05,
+      "loss": 0.8591421842575073,
+      "step": 2756
+    },
+    {
+      "epoch": 3.7256756756756757,
+      "grad_norm": 0.25027477741241455,
+      "learning_rate": 2.3075420406173997e-05,
+      "loss": 0.8655200600624084,
+      "step": 2757
+    },
+    {
+      "epoch": 3.727027027027027,
+      "grad_norm": 0.24963104724884033,
+      "learning_rate": 2.2853248428324258e-05,
+      "loss": 0.8650161623954773,
+      "step": 2758
+    },
+    {
+      "epoch": 3.7283783783783786,
+      "grad_norm": 0.2229507714509964,
+      "learning_rate": 2.2632099938302093e-05,
+      "loss": 0.8645890951156616,
+      "step": 2759
+    },
+    {
+      "epoch": 3.72972972972973,
+      "grad_norm": 0.2523234486579895,
+      "learning_rate": 2.2411975932567828e-05,
+      "loss": 0.8679872751235962,
+      "step": 2760
+    },
+    {
+      "epoch": 3.731081081081081,
+      "grad_norm": 0.26563960313796997,
+      "learning_rate": 2.219287740296605e-05,
+      "loss": 0.8705397844314575,
+      "step": 2761
+    },
+    {
+      "epoch": 3.7324324324324323,
+      "grad_norm": 0.20702576637268066,
+      "learning_rate": 2.197480533672061e-05,
+      "loss": 0.8491113781929016,
+      "step": 2762
+    },
+    {
+      "epoch": 3.733783783783784,
+      "grad_norm": 0.21841701865196228,
+      "learning_rate": 2.175776071643007e-05,
+      "loss": 0.8551273345947266,
+      "step": 2763
+    },
+    {
+      "epoch": 3.735135135135135,
+      "grad_norm": 0.2349194437265396,
+      "learning_rate": 2.1541744520063716e-05,
+      "loss": 0.8688463568687439,
+      "step": 2764
+    },
+    {
+      "epoch": 3.7364864864864864,
+      "grad_norm": 0.22457386553287506,
+      "learning_rate": 2.1326757720956826e-05,
+      "loss": 0.8569310903549194,
+      "step": 2765
+    },
+    {
+      "epoch": 3.737837837837838,
+      "grad_norm": 0.21774189174175262,
+      "learning_rate": 2.1112801287806378e-05,
+      "loss": 0.8630508184432983,
+      "step": 2766
+    },
+    {
+      "epoch": 3.7391891891891893,
+      "grad_norm": 0.23934215307235718,
+      "learning_rate": 2.0899876184666654e-05,
+      "loss": 0.8613144159317017,
+      "step": 2767
+    },
+    {
+      "epoch": 3.7405405405405405,
+      "grad_norm": 0.2800068259239197,
+      "learning_rate": 2.0687983370944936e-05,
+      "loss": 0.8543647527694702,
+      "step": 2768
+    },
+    {
+      "epoch": 3.7418918918918918,
+      "grad_norm": 0.30000585317611694,
+      "learning_rate": 2.047712380139727e-05,
+      "loss": 0.8862980604171753,
+      "step": 2769
+    },
+    {
+      "epoch": 3.743243243243243,
+      "grad_norm": 0.20037543773651123,
+      "learning_rate": 2.0267298426123932e-05,
+      "loss": 0.8533813953399658,
+      "step": 2770
+    },
+    {
+      "epoch": 3.7445945945945946,
+      "grad_norm": 0.23713862895965576,
+      "learning_rate": 2.0058508190565316e-05,
+      "loss": 0.8569244146347046,
+      "step": 2771
+    },
+    {
+      "epoch": 3.745945945945946,
+      "grad_norm": 0.25092074275016785,
+      "learning_rate": 1.9850754035497688e-05,
+      "loss": 0.8782235980033875,
+      "step": 2772
+    },
+    {
+      "epoch": 3.7472972972972975,
+      "grad_norm": 0.22953690588474274,
+      "learning_rate": 1.9644036897028815e-05,
+      "loss": 0.8558734655380249,
+      "step": 2773
+    },
+    {
+      "epoch": 3.7486486486486488,
+      "grad_norm": 0.2744849920272827,
+      "learning_rate": 1.9438357706593884e-05,
+      "loss": 0.8691932559013367,
+      "step": 2774
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.25697776675224304,
+      "learning_rate": 1.923371739095131e-05,
+      "loss": 0.8631426095962524,
+      "step": 2775
+    },
+    {
+      "epoch": 3.7513513513513512,
+      "grad_norm": 0.23103508353233337,
+      "learning_rate": 1.9030116872178316e-05,
+      "loss": 0.8542687296867371,
+      "step": 2776
+    },
+    {
+      "epoch": 3.7527027027027025,
+      "grad_norm": 0.22856485843658447,
+      "learning_rate": 1.8827557067667146e-05,
+      "loss": 0.8566356301307678,
+      "step": 2777
+    },
+    {
+      "epoch": 3.754054054054054,
+      "grad_norm": 0.23556697368621826,
+      "learning_rate": 1.862603889012074e-05,
+      "loss": 0.8536059856414795,
+      "step": 2778
+    },
+    {
+      "epoch": 3.7554054054054054,
+      "grad_norm": 0.20096305012702942,
+      "learning_rate": 1.8425563247548405e-05,
+      "loss": 0.8609071969985962,
+      "step": 2779
+    },
+    {
+      "epoch": 3.756756756756757,
+      "grad_norm": 0.2830381989479065,
+      "learning_rate": 1.822613104326229e-05,
+      "loss": 0.864898681640625,
+      "step": 2780
+    },
+    {
+      "epoch": 3.7581081081081082,
+      "grad_norm": 0.2208063304424286,
+      "learning_rate": 1.8027743175872664e-05,
+      "loss": 0.8496872782707214,
+      "step": 2781
+    },
+    {
+      "epoch": 3.7594594594594595,
+      "grad_norm": 0.2165670245885849,
+      "learning_rate": 1.783040053928439e-05,
+      "loss": 0.8585920333862305,
+      "step": 2782
+    },
+    {
+      "epoch": 3.7608108108108107,
+      "grad_norm": 0.27349063754081726,
+      "learning_rate": 1.763410402269258e-05,
+      "loss": 0.8529818058013916,
+      "step": 2783
+    },
+    {
+      "epoch": 3.762162162162162,
+      "grad_norm": 0.2482490837574005,
+      "learning_rate": 1.7438854510578695e-05,
+      "loss": 0.8618587851524353,
+      "step": 2784
+    },
+    {
+      "epoch": 3.7635135135135136,
+      "grad_norm": 0.2905270755290985,
+      "learning_rate": 1.7244652882706546e-05,
+      "loss": 0.872962474822998,
+      "step": 2785
+    },
+    {
+      "epoch": 3.764864864864865,
+      "grad_norm": 0.2592305541038513,
+      "learning_rate": 1.7051500014118455e-05,
+      "loss": 0.8734329342842102,
+      "step": 2786
+    },
+    {
+      "epoch": 3.766216216216216,
+      "grad_norm": 0.26556992530822754,
+      "learning_rate": 1.6859396775131098e-05,
+      "loss": 0.8548123836517334,
+      "step": 2787
+    },
+    {
+      "epoch": 3.7675675675675677,
+      "grad_norm": 0.2424938827753067,
+      "learning_rate": 1.6668344031331662e-05,
+      "loss": 0.8570354580879211,
+      "step": 2788
+    },
+    {
+      "epoch": 3.768918918918919,
+      "grad_norm": 0.2466237097978592,
+      "learning_rate": 1.6478342643574006e-05,
+      "loss": 0.8733378648757935,
+      "step": 2789
+    },
+    {
+      "epoch": 3.77027027027027,
+      "grad_norm": 0.23004792630672455,
+      "learning_rate": 1.6289393467974757e-05,
+      "loss": 0.8548240065574646,
+      "step": 2790
+    },
+    {
+      "epoch": 3.7716216216216214,
+      "grad_norm": 0.2692965567111969,
+      "learning_rate": 1.610149735590949e-05,
+      "loss": 0.8787268996238708,
+      "step": 2791
+    },
+    {
+      "epoch": 3.772972972972973,
+      "grad_norm": 0.25896963477134705,
+      "learning_rate": 1.591465515400864e-05,
+      "loss": 0.8674949407577515,
+      "step": 2792
+    },
+    {
+      "epoch": 3.7743243243243243,
+      "grad_norm": 0.24642837047576904,
+      "learning_rate": 1.5728867704154075e-05,
+      "loss": 0.8575776219367981,
+      "step": 2793
+    },
+    {
+      "epoch": 3.7756756756756755,
+      "grad_norm": 0.2558622658252716,
+      "learning_rate": 1.5544135843475194e-05,
+      "loss": 0.8656233549118042,
+      "step": 2794
+    },
+    {
+      "epoch": 3.777027027027027,
+      "grad_norm": 0.2278105765581131,
+      "learning_rate": 1.5360460404344717e-05,
+      "loss": 0.8500628471374512,
+      "step": 2795
+    },
+    {
+      "epoch": 3.7783783783783784,
+      "grad_norm": 0.3172438442707062,
+      "learning_rate": 1.5177842214375681e-05,
+      "loss": 0.8649758696556091,
+      "step": 2796
+    },
+    {
+      "epoch": 3.7797297297297296,
+      "grad_norm": 0.2600136399269104,
+      "learning_rate": 1.4996282096417125e-05,
+      "loss": 0.8730435371398926,
+      "step": 2797
+    },
+    {
+      "epoch": 3.781081081081081,
+      "grad_norm": 0.23770098388195038,
+      "learning_rate": 1.4815780868550593e-05,
+      "loss": 0.8571155071258545,
+      "step": 2798
+    },
+    {
+      "epoch": 3.7824324324324325,
+      "grad_norm": 0.24373097717761993,
+      "learning_rate": 1.4636339344086453e-05,
+      "loss": 0.8798297643661499,
+      "step": 2799
+    },
+    {
+      "epoch": 3.7837837837837838,
+      "grad_norm": 0.2035806030035019,
+      "learning_rate": 1.4457958331560245e-05,
+      "loss": 0.850256085395813,
+      "step": 2800
+    },
+    {
+      "epoch": 3.785135135135135,
+      "grad_norm": 0.23062920570373535,
+      "learning_rate": 1.428063863472895e-05,
+      "loss": 0.8492347598075867,
+      "step": 2801
+    },
+    {
+      "epoch": 3.7864864864864867,
+      "grad_norm": 0.22495591640472412,
+      "learning_rate": 1.4104381052567534e-05,
+      "loss": 0.8537865281105042,
+      "step": 2802
+    },
+    {
+      "epoch": 3.787837837837838,
+      "grad_norm": 0.2518802285194397,
+      "learning_rate": 1.3929186379265101e-05,
+      "loss": 0.864928126335144,
+      "step": 2803
+    },
+    {
+      "epoch": 3.789189189189189,
+      "grad_norm": 0.2503625452518463,
+      "learning_rate": 1.3755055404221505e-05,
+      "loss": 0.8563740253448486,
+      "step": 2804
+    },
+    {
+      "epoch": 3.7905405405405403,
+      "grad_norm": 0.25038284063339233,
+      "learning_rate": 1.3581988912043846e-05,
+      "loss": 0.8681548237800598,
+      "step": 2805
+    },
+    {
+      "epoch": 3.791891891891892,
+      "grad_norm": 0.24879653751850128,
+      "learning_rate": 1.3409987682542618e-05,
+      "loss": 0.8668654561042786,
+      "step": 2806
+    },
+    {
+      "epoch": 3.7932432432432432,
+      "grad_norm": 0.2378370463848114,
+      "learning_rate": 1.3239052490728626e-05,
+      "loss": 0.858367919921875,
+      "step": 2807
+    },
+    {
+      "epoch": 3.7945945945945945,
+      "grad_norm": 0.25542181730270386,
+      "learning_rate": 1.3069184106809128e-05,
+      "loss": 0.8682346940040588,
+      "step": 2808
+    },
+    {
+      "epoch": 3.795945945945946,
+      "grad_norm": 0.23329246044158936,
+      "learning_rate": 1.2900383296184538e-05,
+      "loss": 0.8600069284439087,
+      "step": 2809
+    },
+    {
+      "epoch": 3.7972972972972974,
+      "grad_norm": 0.2525200843811035,
+      "learning_rate": 1.2732650819445107e-05,
+      "loss": 0.8694725036621094,
+      "step": 2810
+    },
+    {
+      "epoch": 3.7986486486486486,
+      "grad_norm": 0.2242072969675064,
+      "learning_rate": 1.2565987432367032e-05,
+      "loss": 0.8501420617103577,
+      "step": 2811
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.2664201855659485,
+      "learning_rate": 1.2400393885909633e-05,
+      "loss": 0.8701269626617432,
+      "step": 2812
+    },
+    {
+      "epoch": 3.8013513513513515,
+      "grad_norm": 0.2550244927406311,
+      "learning_rate": 1.2235870926211617e-05,
+      "loss": 0.8684715628623962,
+      "step": 2813
+    },
+    {
+      "epoch": 3.8027027027027027,
+      "grad_norm": 0.30465394258499146,
+      "learning_rate": 1.2072419294587717e-05,
+      "loss": 0.875123918056488,
+      "step": 2814
+    },
+    {
+      "epoch": 3.804054054054054,
+      "grad_norm": 0.2691992223262787,
+      "learning_rate": 1.1910039727525524e-05,
+      "loss": 0.8566411733627319,
+      "step": 2815
+    },
+    {
+      "epoch": 3.8054054054054056,
+      "grad_norm": 0.19907744228839874,
+      "learning_rate": 1.1748732956682024e-05,
+      "loss": 0.8509312272071838,
+      "step": 2816
+    },
+    {
+      "epoch": 3.806756756756757,
+      "grad_norm": 0.23717203736305237,
+      "learning_rate": 1.1588499708880318e-05,
+      "loss": 0.8594992756843567,
+      "step": 2817
+    },
+    {
+      "epoch": 3.808108108108108,
+      "grad_norm": 0.23352646827697754,
+      "learning_rate": 1.1429340706106516e-05,
+      "loss": 0.8603246212005615,
+      "step": 2818
+    },
+    {
+      "epoch": 3.8094594594594593,
+      "grad_norm": 0.2384115308523178,
+      "learning_rate": 1.1271256665506185e-05,
+      "loss": 0.8515311479568481,
+      "step": 2819
+    },
+    {
+      "epoch": 3.810810810810811,
+      "grad_norm": 0.2404327541589737,
+      "learning_rate": 1.1114248299381346e-05,
+      "loss": 0.8697001934051514,
+      "step": 2820
+    },
+    {
+      "epoch": 3.812162162162162,
+      "grad_norm": 0.2199287712574005,
+      "learning_rate": 1.0958316315187289e-05,
+      "loss": 0.8627071976661682,
+      "step": 2821
+    },
+    {
+      "epoch": 3.8135135135135134,
+      "grad_norm": 0.22231224179267883,
+      "learning_rate": 1.0803461415529098e-05,
+      "loss": 0.861302375793457,
+      "step": 2822
+    },
+    {
+      "epoch": 3.814864864864865,
+      "grad_norm": 0.2521754801273346,
+      "learning_rate": 1.0649684298158852e-05,
+      "loss": 0.8574410676956177,
+      "step": 2823
+    },
+    {
+      "epoch": 3.8162162162162163,
+      "grad_norm": 0.2559994161128998,
+      "learning_rate": 1.0496985655972264e-05,
+      "loss": 0.8574870824813843,
+      "step": 2824
+    },
+    {
+      "epoch": 3.8175675675675675,
+      "grad_norm": 0.24460476636886597,
+      "learning_rate": 1.0345366177005544e-05,
+      "loss": 0.8575687408447266,
+      "step": 2825
+    },
+    {
+      "epoch": 3.8189189189189188,
+      "grad_norm": 0.21930775046348572,
+      "learning_rate": 1.0194826544432518e-05,
+      "loss": 0.8522089719772339,
+      "step": 2826
+    },
+    {
+      "epoch": 3.82027027027027,
+      "grad_norm": 0.23672562837600708,
+      "learning_rate": 1.004536743656112e-05,
+      "loss": 0.8605261445045471,
+      "step": 2827
+    },
+    {
+      "epoch": 3.8216216216216217,
+      "grad_norm": 0.18512049317359924,
+      "learning_rate": 9.896989526830907e-06,
+      "loss": 0.8484898209571838,
+      "step": 2828
+    },
+    {
+      "epoch": 3.822972972972973,
+      "grad_norm": 0.2269180715084076,
+      "learning_rate": 9.749693483809551e-06,
+      "loss": 0.8571679592132568,
+      "step": 2829
+    },
+    {
+      "epoch": 3.8243243243243246,
+      "grad_norm": 0.2687380611896515,
+      "learning_rate": 9.603479971190044e-06,
+      "loss": 0.8725071549415588,
+      "step": 2830
+    },
+    {
+      "epoch": 3.825675675675676,
+      "grad_norm": 0.24502205848693848,
+      "learning_rate": 9.458349647787662e-06,
+      "loss": 0.8464882373809814,
+      "step": 2831
+    },
+    {
+      "epoch": 3.827027027027027,
+      "grad_norm": 0.28032153844833374,
+      "learning_rate": 9.314303167537008e-06,
+      "loss": 0.8714008331298828,
+      "step": 2832
+    },
+    {
+      "epoch": 3.8283783783783782,
+      "grad_norm": 0.25088298320770264,
+      "learning_rate": 9.171341179489035e-06,
+      "loss": 0.8726781010627747,
+      "step": 2833
+    },
+    {
+      "epoch": 3.8297297297297295,
+      "grad_norm": 0.27858805656433105,
+      "learning_rate": 9.029464327808185e-06,
+      "loss": 0.8720171451568604,
+      "step": 2834
+    },
+    {
+      "epoch": 3.831081081081081,
+      "grad_norm": 0.1901162564754486,
+      "learning_rate": 8.888673251769396e-06,
+      "loss": 0.8366218209266663,
+      "step": 2835
+    },
+    {
+      "epoch": 3.8324324324324324,
+      "grad_norm": 0.20379436016082764,
+      "learning_rate": 8.748968585755273e-06,
+      "loss": 0.8485777378082275,
+      "step": 2836
+    },
+    {
+      "epoch": 3.833783783783784,
+      "grad_norm": 0.25718504190444946,
+      "learning_rate": 8.610350959253332e-06,
+      "loss": 0.8726571202278137,
+      "step": 2837
+    },
+    {
+      "epoch": 3.8351351351351353,
+      "grad_norm": 0.24544653296470642,
+      "learning_rate": 8.47282099685287e-06,
+      "loss": 0.857839822769165,
+      "step": 2838
+    },
+    {
+      "epoch": 3.8364864864864865,
+      "grad_norm": 0.2843606770038605,
+      "learning_rate": 8.336379318242521e-06,
+      "loss": 0.8794070482254028,
+      "step": 2839
+    },
+    {
+      "epoch": 3.8378378378378377,
+      "grad_norm": 0.2375718504190445,
+      "learning_rate": 8.201026538207146e-06,
+      "loss": 0.8644974231719971,
+      "step": 2840
+    },
+    {
+      "epoch": 3.839189189189189,
+      "grad_norm": 0.3052612245082855,
+      "learning_rate": 8.066763266625281e-06,
+      "loss": 0.874138355255127,
+      "step": 2841
+    },
+    {
+      "epoch": 3.8405405405405406,
+      "grad_norm": 0.25995439291000366,
+      "learning_rate": 7.933590108466337e-06,
+      "loss": 0.8522234559059143,
+      "step": 2842
+    },
+    {
+      "epoch": 3.841891891891892,
+      "grad_norm": 0.22818626463413239,
+      "learning_rate": 7.801507663787676e-06,
+      "loss": 0.8528287410736084,
+      "step": 2843
+    },
+    {
+      "epoch": 3.8432432432432435,
+      "grad_norm": 0.23279845714569092,
+      "learning_rate": 7.670516527732263e-06,
+      "loss": 0.8556399345397949,
+      "step": 2844
+    },
+    {
+      "epoch": 3.8445945945945947,
+      "grad_norm": 0.22167368233203888,
+      "learning_rate": 7.540617290525742e-06,
+      "loss": 0.8527496457099915,
+      "step": 2845
+    },
+    {
+      "epoch": 3.845945945945946,
+      "grad_norm": 0.20148922502994537,
+      "learning_rate": 7.411810537473751e-06,
+      "loss": 0.8480883836746216,
+      "step": 2846
+    },
+    {
+      "epoch": 3.847297297297297,
+      "grad_norm": 0.21240970492362976,
+      "learning_rate": 7.2840968489594205e-06,
+      "loss": 0.8456725478172302,
+      "step": 2847
+    },
+    {
+      "epoch": 3.8486486486486484,
+      "grad_norm": 0.2286936342716217,
+      "learning_rate": 7.157476800440821e-06,
+      "loss": 0.8468947410583496,
+      "step": 2848
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.20458172261714935,
+      "learning_rate": 7.031950962447992e-06,
+      "loss": 0.8532216548919678,
+      "step": 2849
+    },
+    {
+      "epoch": 3.8513513513513513,
+      "grad_norm": 0.23595719039440155,
+      "learning_rate": 6.907519900580861e-06,
+      "loss": 0.8615290522575378,
+      "step": 2850
+    },
+    {
+      "epoch": 3.8527027027027025,
+      "grad_norm": 0.29322656989097595,
+      "learning_rate": 6.784184175506358e-06,
+      "loss": 0.878531277179718,
+      "step": 2851
+    },
+    {
+      "epoch": 3.854054054054054,
+      "grad_norm": 0.23042002320289612,
+      "learning_rate": 6.661944342956e-06,
+      "loss": 0.8543225526809692,
+      "step": 2852
+    },
+    {
+      "epoch": 3.8554054054054054,
+      "grad_norm": 0.24946138262748718,
+      "learning_rate": 6.540800953723502e-06,
+      "loss": 0.8669743537902832,
+      "step": 2853
+    },
+    {
+      "epoch": 3.8567567567567567,
+      "grad_norm": 0.2301298975944519,
+      "learning_rate": 6.4207545536619226e-06,
+      "loss": 0.867255687713623,
+      "step": 2854
+    },
+    {
+      "epoch": 3.858108108108108,
+      "grad_norm": 0.2571972906589508,
+      "learning_rate": 6.301805683681744e-06,
+      "loss": 0.8615138530731201,
+      "step": 2855
+    },
+    {
+      "epoch": 3.8594594594594596,
+      "grad_norm": 0.27989479899406433,
+      "learning_rate": 6.1839548797479605e-06,
+      "loss": 0.8640744090080261,
+      "step": 2856
+    },
+    {
+      "epoch": 3.860810810810811,
+      "grad_norm": 0.22122475504875183,
+      "learning_rate": 6.067202672877886e-06,
+      "loss": 0.8585621118545532,
+      "step": 2857
+    },
+    {
+      "epoch": 3.862162162162162,
+      "grad_norm": 0.30201807618141174,
+      "learning_rate": 5.951549589138822e-06,
+      "loss": 0.8669102191925049,
+      "step": 2858
+    },
+    {
+      "epoch": 3.8635135135135137,
+      "grad_norm": 0.22130148112773895,
+      "learning_rate": 5.836996149645446e-06,
+      "loss": 0.8550734519958496,
+      "step": 2859
+    },
+    {
+      "epoch": 3.864864864864865,
+      "grad_norm": 0.2372446060180664,
+      "learning_rate": 5.723542870557735e-06,
+      "loss": 0.8580465316772461,
+      "step": 2860
+    },
+    {
+      "epoch": 3.866216216216216,
+      "grad_norm": 0.25958284735679626,
+      "learning_rate": 5.611190263078464e-06,
+      "loss": 0.8778803944587708,
+      "step": 2861
+    },
+    {
+      "epoch": 3.8675675675675674,
+      "grad_norm": 0.22226625680923462,
+      "learning_rate": 5.499938833450929e-06,
+      "loss": 0.8565491437911987,
+      "step": 2862
+    },
+    {
+      "epoch": 3.868918918918919,
+      "grad_norm": 0.2549592852592468,
+      "learning_rate": 5.389789082956731e-06,
+      "loss": 0.8609598875045776,
+      "step": 2863
+    },
+    {
+      "epoch": 3.8702702702702703,
+      "grad_norm": 0.24860826134681702,
+      "learning_rate": 5.2807415079134946e-06,
+      "loss": 0.8758200407028198,
+      "step": 2864
+    },
+    {
+      "epoch": 3.8716216216216215,
+      "grad_norm": 0.23083719611167908,
+      "learning_rate": 5.172796599672486e-06,
+      "loss": 0.8743436336517334,
+      "step": 2865
+    },
+    {
+      "epoch": 3.872972972972973,
+      "grad_norm": 0.27522334456443787,
+      "learning_rate": 5.065954844616722e-06,
+      "loss": 0.8609395027160645,
+      "step": 2866
+    },
+    {
+      "epoch": 3.8743243243243244,
+      "grad_norm": 0.2365145981311798,
+      "learning_rate": 4.960216724158445e-06,
+      "loss": 0.8670240640640259,
+      "step": 2867
+    },
+    {
+      "epoch": 3.8756756756756756,
+      "grad_norm": 0.21584731340408325,
+      "learning_rate": 4.855582714737128e-06,
+      "loss": 0.8518607020378113,
+      "step": 2868
+    },
+    {
+      "epoch": 3.877027027027027,
+      "grad_norm": 0.24992552399635315,
+      "learning_rate": 4.752053287817332e-06,
+      "loss": 0.8643487691879272,
+      "step": 2869
+    },
+    {
+      "epoch": 3.8783783783783785,
+      "grad_norm": 0.22549884021282196,
+      "learning_rate": 4.649628909886406e-06,
+      "loss": 0.8618836402893066,
+      "step": 2870
+    },
+    {
+      "epoch": 3.8797297297297297,
+      "grad_norm": 0.2593865394592285,
+      "learning_rate": 4.548310042452736e-06,
+      "loss": 0.8707873821258545,
+      "step": 2871
+    },
+    {
+      "epoch": 3.881081081081081,
+      "grad_norm": 0.22364111244678497,
+      "learning_rate": 4.44809714204325e-06,
+      "loss": 0.8529161810874939,
+      "step": 2872
+    },
+    {
+      "epoch": 3.8824324324324326,
+      "grad_norm": 0.23583835363388062,
+      "learning_rate": 4.348990660201668e-06,
+      "loss": 0.8568888902664185,
+      "step": 2873
+    },
+    {
+      "epoch": 3.883783783783784,
+      "grad_norm": 0.2454340159893036,
+      "learning_rate": 4.250991043486391e-06,
+      "loss": 0.8617950677871704,
+      "step": 2874
+    },
+    {
+      "epoch": 3.885135135135135,
+      "grad_norm": 0.2031455636024475,
+      "learning_rate": 4.1540987334683086e-06,
+      "loss": 0.8665478229522705,
+      "step": 2875
+    },
+    {
+      "epoch": 3.8864864864864863,
+      "grad_norm": 0.24109362065792084,
+      "learning_rate": 4.05831416672911e-06,
+      "loss": 0.8545305728912354,
+      "step": 2876
+    },
+    {
+      "epoch": 3.887837837837838,
+      "grad_norm": 0.22278422117233276,
+      "learning_rate": 3.963637774859114e-06,
+      "loss": 0.8548373579978943,
+      "step": 2877
+    },
+    {
+      "epoch": 3.889189189189189,
+      "grad_norm": 0.2461424022912979,
+      "learning_rate": 3.870069984455355e-06,
+      "loss": 0.8656236529350281,
+      "step": 2878
+    },
+    {
+      "epoch": 3.8905405405405404,
+      "grad_norm": 0.25374677777290344,
+      "learning_rate": 3.7776112171196976e-06,
+      "loss": 0.8495681881904602,
+      "step": 2879
+    },
+    {
+      "epoch": 3.891891891891892,
+      "grad_norm": 0.21826878190040588,
+      "learning_rate": 3.6862618894569488e-06,
+      "loss": 0.8626958727836609,
+      "step": 2880
+    },
+    {
+      "epoch": 3.8932432432432433,
+      "grad_norm": 0.2294096201658249,
+      "learning_rate": 3.5960224130728858e-06,
+      "loss": 0.857250452041626,
+      "step": 2881
+    },
+    {
+      "epoch": 3.8945945945945946,
+      "grad_norm": 0.26183387637138367,
+      "learning_rate": 3.5068931945725637e-06,
+      "loss": 0.8719149827957153,
+      "step": 2882
+    },
+    {
+      "epoch": 3.895945945945946,
+      "grad_norm": 0.21493947505950928,
+      "learning_rate": 3.4188746355582887e-06,
+      "loss": 0.8482574224472046,
+      "step": 2883
+    },
+    {
+      "epoch": 3.8972972972972975,
+      "grad_norm": 0.24976293742656708,
+      "learning_rate": 3.3319671326279833e-06,
+      "loss": 0.8704991340637207,
+      "step": 2884
+    },
+    {
+      "epoch": 3.8986486486486487,
+      "grad_norm": 0.21327799558639526,
+      "learning_rate": 3.2461710773732946e-06,
+      "loss": 0.8553286194801331,
+      "step": 2885
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.23919163644313812,
+      "learning_rate": 3.1614868563778486e-06,
+      "loss": 0.8708356022834778,
+      "step": 2886
+    },
+    {
+      "epoch": 3.9013513513513516,
+      "grad_norm": 0.23134875297546387,
+      "learning_rate": 3.077914851215585e-06,
+      "loss": 0.8613228797912598,
+      "step": 2887
+    },
+    {
+      "epoch": 3.902702702702703,
+      "grad_norm": 0.26010072231292725,
+      "learning_rate": 2.995455438448841e-06,
+      "loss": 0.8557901382446289,
+      "step": 2888
+    },
+    {
+      "epoch": 3.904054054054054,
+      "grad_norm": 0.23856641352176666,
+      "learning_rate": 2.9141089896269345e-06,
+      "loss": 0.8675954341888428,
+      "step": 2889
+    },
+    {
+      "epoch": 3.9054054054054053,
+      "grad_norm": 0.23363174498081207,
+      "learning_rate": 2.8338758712842527e-06,
+      "loss": 0.8512870073318481,
+      "step": 2890
+    },
+    {
+      "epoch": 3.9067567567567565,
+      "grad_norm": 0.2461332380771637,
+      "learning_rate": 2.7547564449386665e-06,
+      "loss": 0.8564362525939941,
+      "step": 2891
+    },
+    {
+      "epoch": 3.908108108108108,
+      "grad_norm": 0.24351440370082855,
+      "learning_rate": 2.6767510670900608e-06,
+      "loss": 0.8598102331161499,
+      "step": 2892
+    },
+    {
+      "epoch": 3.9094594594594594,
+      "grad_norm": 0.23332273960113525,
+      "learning_rate": 2.59986008921842e-06,
+      "loss": 0.8609124422073364,
+      "step": 2893
+    },
+    {
+      "epoch": 3.910810810810811,
+      "grad_norm": 0.2473081797361374,
+      "learning_rate": 2.524083857782522e-06,
+      "loss": 0.8569774031639099,
+      "step": 2894
+    },
+    {
+      "epoch": 3.9121621621621623,
+      "grad_norm": 0.2862650156021118,
+      "learning_rate": 2.4494227142182467e-06,
+      "loss": 0.8548111319541931,
+      "step": 2895
+    },
+    {
+      "epoch": 3.9135135135135135,
+      "grad_norm": 0.2562488913536072,
+      "learning_rate": 2.375876994937076e-06,
+      "loss": 0.8595067262649536,
+      "step": 2896
+    },
+    {
+      "epoch": 3.9148648648648647,
+      "grad_norm": 0.21320299804210663,
+      "learning_rate": 2.303447031324485e-06,
+      "loss": 0.8570829629898071,
+      "step": 2897
+    },
+    {
+      "epoch": 3.916216216216216,
+      "grad_norm": 0.23596172034740448,
+      "learning_rate": 2.232133149738663e-06,
+      "loss": 0.8540196418762207,
+      "step": 2898
+    },
+    {
+      "epoch": 3.9175675675675676,
+      "grad_norm": 0.27923157811164856,
+      "learning_rate": 2.1619356715088245e-06,
+      "loss": 0.8845282793045044,
+      "step": 2899
+    },
+    {
+      "epoch": 3.918918918918919,
+      "grad_norm": 0.26145535707473755,
+      "learning_rate": 2.0928549129338172e-06,
+      "loss": 0.8634969592094421,
+      "step": 2900
+    },
+    {
+      "epoch": 3.9202702702702705,
+      "grad_norm": 0.2393837422132492,
+      "learning_rate": 2.024891185280792e-06,
+      "loss": 0.8510044813156128,
+      "step": 2901
+    },
+    {
+      "epoch": 3.9216216216216218,
+      "grad_norm": 0.21927346289157867,
+      "learning_rate": 1.9580447947836755e-06,
+      "loss": 0.8499420881271362,
+      "step": 2902
+    },
+    {
+      "epoch": 3.922972972972973,
+      "grad_norm": 0.25582394003868103,
+      "learning_rate": 1.892316042641923e-06,
+      "loss": 0.8587778210639954,
+      "step": 2903
+    },
+    {
+      "epoch": 3.924324324324324,
+      "grad_norm": 0.2509669065475464,
+      "learning_rate": 1.8277052250189885e-06,
+      "loss": 0.848979115486145,
+      "step": 2904
+    },
+    {
+      "epoch": 3.9256756756756754,
+      "grad_norm": 0.26379695534706116,
+      "learning_rate": 1.7642126330411624e-06,
+      "loss": 0.8582193851470947,
+      "step": 2905
+    },
+    {
+      "epoch": 3.927027027027027,
+      "grad_norm": 0.2308712899684906,
+      "learning_rate": 1.7018385527961532e-06,
+      "loss": 0.8632068634033203,
+      "step": 2906
+    },
+    {
+      "epoch": 3.9283783783783783,
+      "grad_norm": 0.286137193441391,
+      "learning_rate": 1.6405832653318408e-06,
+      "loss": 0.8845812082290649,
+      "step": 2907
+    },
+    {
+      "epoch": 3.92972972972973,
+      "grad_norm": 0.27754345536231995,
+      "learning_rate": 1.580447046654998e-06,
+      "loss": 0.8794586062431335,
+      "step": 2908
+    },
+    {
+      "epoch": 3.9310810810810812,
+      "grad_norm": 0.2311629205942154,
+      "learning_rate": 1.5214301677300425e-06,
+      "loss": 0.8643574714660645,
+      "step": 2909
+    },
+    {
+      "epoch": 3.9324324324324325,
+      "grad_norm": 0.23007115721702576,
+      "learning_rate": 1.4635328944778148e-06,
+      "loss": 0.8586368560791016,
+      "step": 2910
+    },
+    {
+      "epoch": 3.9337837837837837,
+      "grad_norm": 0.22602854669094086,
+      "learning_rate": 1.406755487774386e-06,
+      "loss": 0.8601138591766357,
+      "step": 2911
+    },
+    {
+      "epoch": 3.935135135135135,
+      "grad_norm": 0.24954913556575775,
+      "learning_rate": 1.3510982034499187e-06,
+      "loss": 0.8694870471954346,
+      "step": 2912
+    },
+    {
+      "epoch": 3.9364864864864866,
+      "grad_norm": 0.24480272829532623,
+      "learning_rate": 1.296561292287446e-06,
+      "loss": 0.8698195815086365,
+      "step": 2913
+    },
+    {
+      "epoch": 3.937837837837838,
+      "grad_norm": 0.30221325159072876,
+      "learning_rate": 1.2431450000217615e-06,
+      "loss": 0.8674991726875305,
+      "step": 2914
+    },
+    {
+      "epoch": 3.939189189189189,
+      "grad_norm": 0.2306823879480362,
+      "learning_rate": 1.1908495673383924e-06,
+      "loss": 0.857147216796875,
+      "step": 2915
+    },
+    {
+      "epoch": 3.9405405405405407,
+      "grad_norm": 0.20528508722782135,
+      "learning_rate": 1.13967522987235e-06,
+      "loss": 0.8533166646957397,
+      "step": 2916
+    },
+    {
+      "epoch": 3.941891891891892,
+      "grad_norm": 0.2680051624774933,
+      "learning_rate": 1.0896222182072423e-06,
+      "loss": 0.8767186403274536,
+      "step": 2917
+    },
+    {
+      "epoch": 3.943243243243243,
+      "grad_norm": 0.25400981307029724,
+      "learning_rate": 1.0406907578741353e-06,
+      "loss": 0.8706134557723999,
+      "step": 2918
+    },
+    {
+      "epoch": 3.9445945945945944,
+      "grad_norm": 0.24701231718063354,
+      "learning_rate": 9.928810693505263e-07,
+      "loss": 0.8686222434043884,
+      "step": 2919
+    },
+    {
+      "epoch": 3.945945945945946,
+      "grad_norm": 0.22947126626968384,
+      "learning_rate": 9.461933680594559e-07,
+      "loss": 0.854644775390625,
+      "step": 2920
+    },
+    {
+      "epoch": 3.9472972972972973,
+      "grad_norm": 0.2354872077703476,
+      "learning_rate": 9.006278643683696e-07,
+      "loss": 0.8536649942398071,
+      "step": 2921
+    },
+    {
+      "epoch": 3.9486486486486485,
+      "grad_norm": 0.2668014168739319,
+      "learning_rate": 8.56184763588369e-07,
+      "loss": 0.8655614256858826,
+      "step": 2922
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.25601038336753845,
+      "learning_rate": 8.128642659731289e-07,
+      "loss": 0.8667013645172119,
+      "step": 2923
+    },
+    {
+      "epoch": 3.9513513513513514,
+      "grad_norm": 0.21539060771465302,
+      "learning_rate": 7.70666566718009e-07,
+      "loss": 0.8637655377388,
+      "step": 2924
+    },
+    {
+      "epoch": 3.9527027027027026,
+      "grad_norm": 0.2828670144081116,
+      "learning_rate": 7.295918559593051e-07,
+      "loss": 0.8717225193977356,
+      "step": 2925
+    },
+    {
+      "epoch": 3.954054054054054,
+      "grad_norm": 0.2704210877418518,
+      "learning_rate": 6.896403187732214e-07,
+      "loss": 0.8806003928184509,
+      "step": 2926
+    },
+    {
+      "epoch": 3.9554054054054055,
+      "grad_norm": 0.23779624700546265,
+      "learning_rate": 6.508121351751773e-07,
+      "loss": 0.8602455854415894,
+      "step": 2927
+    },
+    {
+      "epoch": 3.9567567567567568,
+      "grad_norm": 0.24474170804023743,
+      "learning_rate": 6.131074801189185e-07,
+      "loss": 0.8482333421707153,
+      "step": 2928
+    },
+    {
+      "epoch": 3.958108108108108,
+      "grad_norm": 0.1818297654390335,
+      "learning_rate": 5.765265234957129e-07,
+      "loss": 0.8410011529922485,
+      "step": 2929
+    },
+    {
+      "epoch": 3.9594594594594597,
+      "grad_norm": 0.2445749044418335,
+      "learning_rate": 5.41069430133656e-07,
+      "loss": 0.8658677339553833,
+      "step": 2930
+    },
+    {
+      "epoch": 3.960810810810811,
+      "grad_norm": 0.2458166629076004,
+      "learning_rate": 5.067363597968666e-07,
+      "loss": 0.8565350770950317,
+      "step": 2931
+    },
+    {
+      "epoch": 3.962162162162162,
+      "grad_norm": 0.24617113173007965,
+      "learning_rate": 4.7352746718482e-07,
+      "loss": 0.87021803855896,
+      "step": 2932
+    },
+    {
+      "epoch": 3.9635135135135133,
+      "grad_norm": 0.2413380891084671,
+      "learning_rate": 4.414429019316268e-07,
+      "loss": 0.8703094720840454,
+      "step": 2933
+    },
+    {
+      "epoch": 3.964864864864865,
+      "grad_norm": 0.2646099925041199,
+      "learning_rate": 4.1048280860528363e-07,
+      "loss": 0.878566324710846,
+      "step": 2934
+    },
+    {
+      "epoch": 3.9662162162162162,
+      "grad_norm": 0.29184049367904663,
+      "learning_rate": 3.8064732670717304e-07,
+      "loss": 0.8857858180999756,
+      "step": 2935
+    },
+    {
+      "epoch": 3.9675675675675675,
+      "grad_norm": 0.2625546157360077,
+      "learning_rate": 3.5193659067131456e-07,
+      "loss": 0.8623246550559998,
+      "step": 2936
+    },
+    {
+      "epoch": 3.968918918918919,
+      "grad_norm": 0.23265071213245392,
+      "learning_rate": 3.2435072986378155e-07,
+      "loss": 0.8656038641929626,
+      "step": 2937
+    },
+    {
+      "epoch": 3.9702702702702704,
+      "grad_norm": 0.20124533772468567,
+      "learning_rate": 2.9788986858220177e-07,
+      "loss": 0.8510494232177734,
+      "step": 2938
+    },
+    {
+      "epoch": 3.9716216216216216,
+      "grad_norm": 0.20958954095840454,
+      "learning_rate": 2.7255412605506347e-07,
+      "loss": 0.8537824153900146,
+      "step": 2939
+    },
+    {
+      "epoch": 3.972972972972973,
+      "grad_norm": 0.22538672387599945,
+      "learning_rate": 2.4834361644129887e-07,
+      "loss": 0.8653491735458374,
+      "step": 2940
+    },
+    {
+      "epoch": 3.9743243243243245,
+      "grad_norm": 0.22078093886375427,
+      "learning_rate": 2.252584488296461e-07,
+      "loss": 0.859553337097168,
+      "step": 2941
+    },
+    {
+      "epoch": 3.9756756756756757,
+      "grad_norm": 0.21707819402217865,
+      "learning_rate": 2.03298727238288e-07,
+      "loss": 0.8572168946266174,
+      "step": 2942
+    },
+    {
+      "epoch": 3.977027027027027,
+      "grad_norm": 0.2555243968963623,
+      "learning_rate": 1.8246455061429746e-07,
+      "loss": 0.8721463680267334,
+      "step": 2943
+    },
+    {
+      "epoch": 3.9783783783783786,
+      "grad_norm": 0.22064903378486633,
+      "learning_rate": 1.6275601283322061e-07,
+      "loss": 0.8565831184387207,
+      "step": 2944
+    },
+    {
+      "epoch": 3.97972972972973,
+      "grad_norm": 0.21320542693138123,
+      "learning_rate": 1.4417320269868862e-07,
+      "loss": 0.8568220138549805,
+      "step": 2945
+    },
+    {
+      "epoch": 3.981081081081081,
+      "grad_norm": 0.2409137636423111,
+      "learning_rate": 1.267162039418901e-07,
+      "loss": 0.8560389280319214,
+      "step": 2946
+    },
+    {
+      "epoch": 3.9824324324324323,
+      "grad_norm": 0.24712640047073364,
+      "learning_rate": 1.1038509522140472e-07,
+      "loss": 0.8680664300918579,
+      "step": 2947
+    },
+    {
+      "epoch": 3.983783783783784,
+      "grad_norm": 0.22115212678909302,
+      "learning_rate": 9.517995012259251e-08,
+      "loss": 0.8586644530296326,
+      "step": 2948
+    },
+    {
+      "epoch": 3.985135135135135,
+      "grad_norm": 0.2543070614337921,
+      "learning_rate": 8.11008371575106e-08,
+      "loss": 0.8786360025405884,
+      "step": 2949
+    },
+    {
+      "epoch": 3.9864864864864864,
+      "grad_norm": 0.21990635991096497,
+      "learning_rate": 6.81478197644414e-08,
+      "loss": 0.861034631729126,
+      "step": 2950
+    },
+    {
+      "epoch": 3.987837837837838,
+      "grad_norm": 0.28812140226364136,
+      "learning_rate": 5.632095630764278e-08,
+      "loss": 0.8687256574630737,
+      "step": 2951
+    },
+    {
+      "epoch": 3.9891891891891893,
+      "grad_norm": 0.2501102387905121,
+      "learning_rate": 4.562030007712603e-08,
+      "loss": 0.8508026599884033,
+      "step": 2952
+    },
+    {
+      "epoch": 3.9905405405405405,
+      "grad_norm": 0.23981282114982605,
+      "learning_rate": 3.604589928837832e-08,
+      "loss": 0.8689833879470825,
+      "step": 2953
+    },
+    {
+      "epoch": 3.9918918918918918,
+      "grad_norm": 0.26858052611351013,
+      "learning_rate": 2.7597797082168365e-08,
+      "loss": 0.86488938331604,
+      "step": 2954
+    },
+    {
+      "epoch": 3.993243243243243,
+      "grad_norm": 0.23143915832042694,
+      "learning_rate": 2.027603152429669e-08,
+      "loss": 0.8513102531433105,
+      "step": 2955
+    },
+    {
+      "epoch": 3.9945945945945946,
+      "grad_norm": 0.2745286822319031,
+      "learning_rate": 1.4080635605512315e-08,
+      "loss": 0.8803550004959106,
+      "step": 2956
+    },
+    {
+      "epoch": 3.995945945945946,
+      "grad_norm": 0.24288560450077057,
+      "learning_rate": 9.011637241318482e-09,
+      "loss": 0.8742020130157471,
+      "step": 2957
+    },
+    {
+      "epoch": 3.9972972972972975,
+      "grad_norm": 0.27016910910606384,
+      "learning_rate": 5.06905927180612e-09,
+      "loss": 0.8689770102500916,
+      "step": 2958
+    },
+    {
+      "epoch": 3.9986486486486488,
+      "grad_norm": 0.24802911281585693,
+      "learning_rate": 2.2529194616260905e-09,
+      "loss": 0.860700786113739,
+      "step": 2959
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.22847764194011688,
+      "learning_rate": 5.63230499850409e-10,
+      "loss": 0.8538284301757812,
+      "step": 2960
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2960,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 740,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.452010024785674e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}