diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5649 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 801,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.050421697822149,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.1544,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.4522491247185525,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.1683,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.209953771699016,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.1475,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.2116476352040837,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.1549,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.0260957864359623,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.1239,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.1512192994944326,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.1077,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8065571184144953,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.0738,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8266262708390884,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.0561,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9759219949909417,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 1.0237,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.052405545359168,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.9659,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.0087710253035542,
+      "learning_rate": 8.8e-06,
+      "loss": 1.044,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6850246498093417,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.9758,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.1615561768753055,
+      "learning_rate": 1.04e-05,
+      "loss": 0.9934,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.7564042577770114,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 1.0311,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3272069105236235,
+      "learning_rate": 1.2e-05,
+      "loss": 0.9553,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.688411223891557,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.9442,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.528134950351535,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.9862,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5328973752393094,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.9502,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2847554516079198,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 0.9205,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4221662366255885,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.9591,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.255367628510996,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.9001,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.148047804296119,
+      "learning_rate": 1.76e-05,
+      "loss": 0.9824,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9917836271027393,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 0.8932,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0823142111067903,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.9214,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0569825941940303,
+      "learning_rate": 2e-05,
+      "loss": 0.8773,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1512649834469049,
+      "learning_rate": 1.999991805061211e-05,
+      "loss": 0.8734,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.877680370813094,
+      "learning_rate": 1.9999672203791564e-05,
+      "loss": 0.8979,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9089963370162624,
+      "learning_rate": 1.9999262463567772e-05,
+      "loss": 0.9686,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9454114965461798,
+      "learning_rate": 1.9998688836656322e-05,
+      "loss": 0.8522,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0221925378438834,
+      "learning_rate": 1.999795133245889e-05,
+      "loss": 0.8927,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7560039515035071,
+      "learning_rate": 1.999704996306308e-05,
+      "loss": 0.8841,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8177802771211006,
+      "learning_rate": 1.999598474324223e-05,
+      "loss": 0.8383,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0409463245472501,
+      "learning_rate": 1.9994755690455154e-05,
+      "loss": 0.9304,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8314907519172736,
+      "learning_rate": 1.9993362824845878e-05,
+      "loss": 0.8596,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7233397253368891,
+      "learning_rate": 1.9991806169243302e-05,
+      "loss": 0.8474,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7250793663189042,
+      "learning_rate": 1.999008574916082e-05,
+      "loss": 0.8506,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8091084445628496,
+      "learning_rate": 1.998820159279591e-05,
+      "loss": 0.8244,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7578973831164427,
+      "learning_rate": 1.9986153731029657e-05,
+      "loss": 0.8457,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6638759236574057,
+      "learning_rate": 1.9983942197426272e-05,
+      "loss": 0.8406,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6896497026906081,
+      "learning_rate": 1.9981567028232514e-05,
+      "loss": 0.9,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6414383119060907,
+      "learning_rate": 1.997902826237712e-05,
+      "loss": 0.8667,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.728045432246043,
+      "learning_rate": 1.9976325941470147e-05,
+      "loss": 0.8454,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7288618347232488,
+      "learning_rate": 1.9973460109802306e-05,
+      "loss": 0.8109,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6374436993894802,
+      "learning_rate": 1.997043081434423e-05,
+      "loss": 0.8208,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6891585415522739,
+      "learning_rate": 1.9967238104745695e-05,
+      "loss": 0.8272,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6869606836622446,
+      "learning_rate": 1.9963882033334827e-05,
+      "loss": 0.869,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6909918132711272,
+      "learning_rate": 1.996036265511722e-05,
+      "loss": 0.8229,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5939328229579649,
+      "learning_rate": 1.9956680027775054e-05,
+      "loss": 0.8408,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.659777802804195,
+      "learning_rate": 1.995283421166614e-05,
+      "loss": 0.795,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.630173617841524,
+      "learning_rate": 1.9948825269822934e-05,
+      "loss": 0.8446,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8717985826540854,
+      "learning_rate": 1.9944653267951507e-05,
+      "loss": 0.7986,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.59493509124484,
+      "learning_rate": 1.994031827443045e-05,
+      "loss": 0.8134,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6740076785755504,
+      "learning_rate": 1.993582036030978e-05,
+      "loss": 0.8064,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6410642565408483,
+      "learning_rate": 1.9931159599309757e-05,
+      "loss": 0.8323,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6720779223299492,
+      "learning_rate": 1.9926336067819686e-05,
+      "loss": 0.8256,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.730476801871339,
+      "learning_rate": 1.9921349844896655e-05,
+      "loss": 0.789,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.595837869201607,
+      "learning_rate": 1.9916201012264255e-05,
+      "loss": 0.8144,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8127953128220716,
+      "learning_rate": 1.991088965431121e-05,
+      "loss": 0.8796,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6175756933005155,
+      "learning_rate": 1.9905415858090036e-05,
+      "loss": 0.8175,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6700936304713285,
+      "learning_rate": 1.9899779713315577e-05,
+      "loss": 0.8098,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6040877421514723,
+      "learning_rate": 1.9893981312363563e-05,
+      "loss": 0.7949,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.720386605678337,
+      "learning_rate": 1.9888020750269067e-05,
+      "loss": 0.774,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.609758482071129,
+      "learning_rate": 1.988189812472498e-05,
+      "loss": 0.8066,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7399468596495679,
+      "learning_rate": 1.987561353608038e-05,
+      "loss": 0.8534,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6288843677336958,
+      "learning_rate": 1.9869167087338908e-05,
+      "loss": 0.8093,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6832779153361084,
+      "learning_rate": 1.9862558884157067e-05,
+      "loss": 0.7917,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7330332971897311,
+      "learning_rate": 1.9855789034842504e-05,
+      "loss": 0.7978,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6972813519695797,
+      "learning_rate": 1.9848857650352213e-05,
+      "loss": 0.7741,
+      "step": 68
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6358450195025426,
+      "learning_rate": 1.9841764844290744e-05,
+      "loss": 0.8011,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6896477747547599,
+      "learning_rate": 1.9834510732908314e-05,
+      "loss": 0.7917,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6710283082370337,
+      "learning_rate": 1.9827095435098926e-05,
+      "loss": 0.7652,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7141801906145083,
+      "learning_rate": 1.9819519072398397e-05,
+      "loss": 0.8508,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6639436013137763,
+      "learning_rate": 1.9811781768982392e-05,
+      "loss": 0.7821,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5477095114515034,
+      "learning_rate": 1.980388365166436e-05,
+      "loss": 0.7874,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6945810654587365,
+      "learning_rate": 1.9795824849893483e-05,
+      "loss": 0.7965,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7217797206059553,
+      "learning_rate": 1.9787605495752528e-05,
+      "loss": 0.7692,
+      "step": 76
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6684061578707455,
+      "learning_rate": 1.977922572395571e-05,
+      "loss": 0.7969,
+      "step": 77
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6707150017766241,
+      "learning_rate": 1.977068567184646e-05,
+      "loss": 0.8307,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7717208943525561,
+      "learning_rate": 1.976198547939518e-05,
+      "loss": 0.757,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.58820913965947,
+      "learning_rate": 1.975312528919697e-05,
+      "loss": 0.8038,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7146796204632543,
+      "learning_rate": 1.9744105246469264e-05,
+      "loss": 0.797,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6743555560911234,
+      "learning_rate": 1.9734925499049446e-05,
+      "loss": 0.7683,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6195095279399262,
+      "learning_rate": 1.972558619739246e-05,
+      "loss": 0.851,
+      "step": 83
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6947888571089916,
+      "learning_rate": 1.9716087494568318e-05,
+      "loss": 0.7964,
+      "step": 84
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7498924442216305,
+      "learning_rate": 1.9706429546259592e-05,
+      "loss": 0.7475,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6477658851669421,
+      "learning_rate": 1.9696612510758878e-05,
+      "loss": 0.7832,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7473123140056833,
+      "learning_rate": 1.9686636548966177e-05,
+      "loss": 0.7764,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.749162547762247,
+      "learning_rate": 1.9676501824386295e-05,
+      "loss": 0.7457,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6956559440461865,
+      "learning_rate": 1.9666208503126115e-05,
+      "loss": 0.7895,
+      "step": 89
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7085425092435499,
+      "learning_rate": 1.9655756753891916e-05,
+      "loss": 0.8317,
+      "step": 90
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8582717663487707,
+      "learning_rate": 1.964514674798659e-05,
+      "loss": 0.7894,
+      "step": 91
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6621588825705549,
+      "learning_rate": 1.9634378659306834e-05,
+      "loss": 0.7851,
+      "step": 92
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7797275848414665,
+      "learning_rate": 1.9623452664340305e-05,
+      "loss": 0.7698,
+      "step": 93
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8540541196734044,
+      "learning_rate": 1.9612368942162717e-05,
+      "loss": 0.7454,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6855920721699429,
+      "learning_rate": 1.960112767443493e-05,
+      "loss": 0.7717,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6772325613360068,
+      "learning_rate": 1.9589729045399935e-05,
+      "loss": 0.7856,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.71783357181609,
+      "learning_rate": 1.957817324187987e-05,
+      "loss": 0.853,
+      "step": 97
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.651081876287226,
+      "learning_rate": 1.9566460453272945e-05,
+      "loss": 0.7758,
+      "step": 98
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7224314306320238,
+      "learning_rate": 1.955459087155033e-05,
+      "loss": 0.7698,
+      "step": 99
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7417366304140689,
+      "learning_rate": 1.954256469125301e-05,
+      "loss": 0.7565,
+      "step": 100
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.705860767192474,
+      "learning_rate": 1.953038210948861e-05,
+      "loss": 0.7521,
+      "step": 101
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6853121441171455,
+      "learning_rate": 1.9518043325928157e-05,
+      "loss": 0.7623,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8531911402501465,
+      "learning_rate": 1.9505548542802805e-05,
+      "loss": 0.7695,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7097227658031378,
+      "learning_rate": 1.9492897964900512e-05,
+      "loss": 0.7686,
+      "step": 104
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7305229333984748,
+      "learning_rate": 1.9480091799562706e-05,
+      "loss": 0.7594,
+      "step": 105
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6769047157703583,
+      "learning_rate": 1.9467130256680867e-05,
+      "loss": 0.7823,
+      "step": 106
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9227853066106919,
+      "learning_rate": 1.9454013548693103e-05,
+      "loss": 0.7604,
+      "step": 107
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6788203795678404,
+      "learning_rate": 1.9440741890580643e-05,
+      "loss": 0.7733,
+      "step": 108
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8101074316401654,
+      "learning_rate": 1.9427315499864345e-05,
+      "loss": 0.8122,
+      "step": 109
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7465255356027457,
+      "learning_rate": 1.9413734596601104e-05,
+      "loss": 0.7712,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.727253005664772,
+      "learning_rate": 1.9399999403380266e-05,
+      "loss": 0.769,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7445474864707267,
+      "learning_rate": 1.9386110145319962e-05,
+      "loss": 0.7253,
+      "step": 112
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7711000445972649,
+      "learning_rate": 1.937206705006344e-05,
+      "loss": 0.7745,
+      "step": 113
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7347966122270397,
+      "learning_rate": 1.93578703477753e-05,
+      "loss": 0.747,
+      "step": 114
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8007034594663927,
+      "learning_rate": 1.9343520271137764e-05,
+      "loss": 0.8104,
+      "step": 115
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6690335939490886,
+      "learning_rate": 1.932901705534683e-05,
+      "loss": 0.7703,
+      "step": 116
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6876843319988346,
+      "learning_rate": 1.9314360938108427e-05,
+      "loss": 0.7574,
+      "step": 117
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8026015331475426,
+      "learning_rate": 1.929955215963452e-05,
+      "loss": 0.7153,
+      "step": 118
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6083840676926519,
+      "learning_rate": 1.928459096263918e-05,
+      "loss": 0.7728,
+      "step": 119
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7793114985568895,
+      "learning_rate": 1.926947759233459e-05,
+      "loss": 0.7554,
+      "step": 120
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6471718444852702,
+      "learning_rate": 1.9254212296427043e-05,
+      "loss": 0.7561,
+      "step": 121
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5288366573811976,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 0.8227,
+      "step": 122
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7540586194531294,
+      "learning_rate": 1.922322693107434e-05,
+      "loss": 0.7607,
+      "step": 123
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8504085552569656,
+      "learning_rate": 1.920750736947553e-05,
+      "loss": 0.7451,
+      "step": 124
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7553248522606649,
+      "learning_rate": 1.9191636897958123e-05,
+      "loss": 0.7472,
+      "step": 125
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.761337788781711,
+      "learning_rate": 1.9175615776637212e-05,
+      "loss": 0.7485,
+      "step": 126
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7221698572987841,
+      "learning_rate": 1.9159444268097012e-05,
+      "loss": 0.7387,
+      "step": 127
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7417521945174339,
+      "learning_rate": 1.9143122637386567e-05,
+      "loss": 0.8169,
+      "step": 128
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7985821809817109,
+      "learning_rate": 1.9126651152015404e-05,
+      "loss": 0.7292,
+      "step": 129
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7182359164981328,
+      "learning_rate": 1.9110030081949157e-05,
+      "loss": 0.762,
+      "step": 130
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6823103865926027,
+      "learning_rate": 1.9093259699605125e-05,
+      "loss": 0.7707,
+      "step": 131
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7037103987162874,
+      "learning_rate": 1.907634027984782e-05,
+      "loss": 0.7543,
+      "step": 132
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6541431669524109,
+      "learning_rate": 1.905927209998447e-05,
+      "loss": 0.7865,
+      "step": 133
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6113930545486561,
+      "learning_rate": 1.9042055439760447e-05,
+      "loss": 0.7503,
+      "step": 134
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6734766407541498,
+      "learning_rate": 1.90246905813547e-05,
+      "loss": 0.7409,
+      "step": 135
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7038106427408382,
+      "learning_rate": 1.900717780937514e-05,
+      "loss": 0.7462,
+      "step": 136
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7009766460600779,
+      "learning_rate": 1.8989517410853956e-05,
+      "loss": 0.7209,
+      "step": 137
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6698192317539081,
+      "learning_rate": 1.897170967524291e-05,
+      "loss": 0.7489,
+      "step": 138
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7761639823087021,
+      "learning_rate": 1.8953754894408617e-05,
+      "loss": 0.7528,
+      "step": 139
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6938555197534098,
+      "learning_rate": 1.893565336262773e-05,
+      "loss": 0.7982,
+      "step": 140
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.780829308967089,
+      "learning_rate": 1.8917405376582144e-05,
+      "loss": 0.7441,
+      "step": 141
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6775910638531962,
+      "learning_rate": 1.8899011235354118e-05,
+      "loss": 0.7348,
+      "step": 142
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7887079664457872,
+      "learning_rate": 1.8880471240421365e-05,
+      "loss": 0.7391,
+      "step": 143
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7224368317100063,
+      "learning_rate": 1.8861785695652142e-05,
+      "loss": 0.7357,
+      "step": 144
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7193648740337865,
+      "learning_rate": 1.8842954907300236e-05,
+      "loss": 0.7436,
+      "step": 145
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6156201185600855,
+      "learning_rate": 1.8823979183999965e-05,
+      "loss": 0.7347,
+      "step": 146
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6564009178371437,
+      "learning_rate": 1.880485883676111e-05,
+      "loss": 0.7905,
+      "step": 147
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6295945305014009,
+      "learning_rate": 1.878559417896382e-05,
+      "loss": 0.7354,
+      "step": 148
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7473996781600849,
+      "learning_rate": 1.876618552635348e-05,
+      "loss": 0.7416,
+      "step": 149
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6864395075209073,
+      "learning_rate": 1.8746633197035525e-05,
+      "loss": 0.7144,
+      "step": 150
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7540407091733655,
+      "learning_rate": 1.8726937511470247e-05,
+      "loss": 0.7307,
+      "step": 151
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8221923969905752,
+      "learning_rate": 1.870709879246752e-05,
+      "loss": 0.728,
+      "step": 152
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.647639291571937,
+      "learning_rate": 1.8687117365181514e-05,
+      "loss": 0.7643,
+      "step": 153
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.876594778809403,
+      "learning_rate": 1.8666993557105377e-05,
+      "loss": 0.7066,
+      "step": 154
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6952076031723992,
+      "learning_rate": 1.8646727698065865e-05,
+      "loss": 0.7374,
+      "step": 155
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.800282441200813,
+      "learning_rate": 1.8626320120217922e-05,
+      "loss": 0.7442,
+      "step": 156
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7474869566385066,
+      "learning_rate": 1.8605771158039253e-05,
+      "loss": 0.7235,
+      "step": 157
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.722811246894062,
+      "learning_rate": 1.858508114832483e-05,
+      "loss": 0.7346,
+      "step": 158
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.824122704451794,
+      "learning_rate": 1.8564250430181387e-05,
+      "loss": 0.7828,
+      "step": 159
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7590804525040064,
+      "learning_rate": 1.8543279345021834e-05,
+      "loss": 0.7317,
+      "step": 160
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6772804496327979,
+      "learning_rate": 1.8522168236559693e-05,
+      "loss": 0.7236,
+      "step": 161
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7301531426117454,
+      "learning_rate": 1.850091745080345e-05,
+      "loss": 0.7086,
+      "step": 162
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6810509924599181,
+      "learning_rate": 1.847952733605088e-05,
+      "loss": 0.7349,
+      "step": 163
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7569415947786665,
+      "learning_rate": 1.8457998242883346e-05,
+      "loss": 0.7318,
+      "step": 164
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7064012926575857,
+      "learning_rate": 1.8436330524160048e-05,
+      "loss": 0.7957,
+      "step": 165
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.662694617080743,
+      "learning_rate": 1.8414524535012244e-05,
+      "loss": 0.709,
+      "step": 166
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8954259419770013,
+      "learning_rate": 1.8392580632837423e-05,
+      "loss": 0.7491,
+      "step": 167
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7024460813570171,
+      "learning_rate": 1.8370499177293463e-05,
+      "loss": 0.7115,
+      "step": 168
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8415609399672339,
+      "learning_rate": 1.8348280530292712e-05,
+      "loss": 0.7359,
+      "step": 169
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7254373543665715,
+      "learning_rate": 1.8325925055996076e-05,
+      "loss": 0.7299,
+      "step": 170
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7254597122440483,
+      "learning_rate": 1.8303433120807043e-05,
+      "loss": 0.733,
+      "step": 171
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7462576744539569,
+      "learning_rate": 1.8280805093365674e-05,
+      "loss": 0.7746,
+      "step": 172
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6748201934936428,
+      "learning_rate": 1.8258041344542567e-05,
+      "loss": 0.742,
+      "step": 173
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6451251599438318,
+      "learning_rate": 1.8235142247432784e-05,
+      "loss": 0.7047,
+      "step": 174
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6799131321350648,
+      "learning_rate": 1.8212108177349722e-05,
+      "loss": 0.7193,
+      "step": 175
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7972919554744684,
+      "learning_rate": 1.8188939511818965e-05,
+      "loss": 0.7381,
+      "step": 176
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7054502019474785,
+      "learning_rate": 1.816563663057211e-05,
+      "loss": 0.7109,
+      "step": 177
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5914644195974801,
+      "learning_rate": 1.814219991554053e-05,
+      "loss": 0.7697,
+      "step": 178
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7335647840733366,
+      "learning_rate": 1.8118629750849106e-05,
+      "loss": 0.7285,
+      "step": 179
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7183256424871337,
+      "learning_rate": 1.8094926522809958e-05,
+      "loss": 0.7131,
+      "step": 180
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6943282845084143,
+      "learning_rate": 1.8071090619916095e-05,
+      "loss": 0.7351,
+      "step": 181
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7377257711447108,
+      "learning_rate": 1.804712243283504e-05,
+      "loss": 0.7255,
+      "step": 182
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.724125878072718,
+      "learning_rate": 1.802302235440245e-05,
+      "loss": 0.7866,
+      "step": 183
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7493613491039653,
+      "learning_rate": 1.799879077961566e-05,
+      "loss": 0.7365,
+      "step": 184
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8908469645001831,
+      "learning_rate": 1.797442810562721e-05,
+      "loss": 0.71,
+      "step": 185
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7053338175952263,
+      "learning_rate": 1.7949934731738348e-05,
+      "loss": 0.734,
+      "step": 186
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9612976311357089,
+      "learning_rate": 1.7925311059392472e-05,
+      "loss": 0.7079,
+      "step": 187
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6836899111297493,
+      "learning_rate": 1.790055749216856e-05,
+      "loss": 0.7196,
+      "step": 188
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.9773072491586141,
+      "learning_rate": 1.7875674435774546e-05,
+      "loss": 0.706,
+      "step": 189
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6595513679823545,
+      "learning_rate": 1.7850662298040676e-05,
+      "loss": 0.7666,
+      "step": 190
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.8016137835161344,
+      "learning_rate": 1.7825521488912833e-05,
+      "loss": 0.7306,
+      "step": 191
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6859313585230287,
+      "learning_rate": 1.7800252420445788e-05,
+      "loss": 0.7209,
+      "step": 192
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6889799204445292,
+      "learning_rate": 1.7774855506796497e-05,
+      "loss": 0.7222,
+      "step": 193
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7147323195488424,
+      "learning_rate": 1.774933116421725e-05,
+      "loss": 0.7205,
+      "step": 194
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7572184994622753,
+      "learning_rate": 1.7723679811048904e-05,
+      "loss": 0.7313,
+      "step": 195
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7495623875878256,
+      "learning_rate": 1.7697901867713997e-05,
+      "loss": 0.696,
+      "step": 196
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8587732175016841,
+      "learning_rate": 1.767199775670986e-05,
+      "loss": 0.79,
+      "step": 197
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8226627175371956,
+      "learning_rate": 1.764596790260171e-05,
+      "loss": 0.7136,
+      "step": 198
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7742898660936148,
+      "learning_rate": 1.7619812732015664e-05,
+      "loss": 0.7106,
+      "step": 199
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7199718228477165,
+      "learning_rate": 1.7593532673631765e-05,
+      "loss": 0.7081,
+      "step": 200
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8389904161132045,
+      "learning_rate": 1.7567128158176955e-05,
+      "loss": 0.6976,
+      "step": 201
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6885626524548528,
+      "learning_rate": 1.754059961841801e-05,
+      "loss": 0.7166,
+      "step": 202
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.62855922965186,
+      "learning_rate": 1.7513947489154443e-05,
+      "loss": 0.7317,
+      "step": 203
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7645461252822144,
+      "learning_rate": 1.7487172207211395e-05,
+      "loss": 0.6871,
+      "step": 204
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7318185970621935,
+      "learning_rate": 1.7460274211432463e-05,
+      "loss": 0.7043,
+      "step": 205
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6593101915077957,
+      "learning_rate": 1.7433253942672497e-05,
+      "loss": 0.7422,
+      "step": 206
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.8233033841823448,
+      "learning_rate": 1.74061118437904e-05,
+      "loss": 0.6828,
+      "step": 207
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6191230026164649,
+      "learning_rate": 1.7378848359641846e-05,
+      "loss": 0.7255,
+      "step": 208
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7333042560107498,
+      "learning_rate": 1.7351463937072008e-05,
+      "loss": 0.7715,
+      "step": 209
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7489281525910614,
+      "learning_rate": 1.732395902490821e-05,
+      "loss": 0.7094,
+      "step": 210
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7997292555576989,
+      "learning_rate": 1.7296334073952606e-05,
+      "loss": 0.7097,
+      "step": 211
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7939984871333575,
+      "learning_rate": 1.726858953697475e-05,
+      "loss": 0.7051,
+      "step": 212
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8980147872239701,
+      "learning_rate": 1.7240725868704218e-05,
+      "loss": 0.7124,
+      "step": 213
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.9006092313288788,
+      "learning_rate": 1.721274352582311e-05,
+      "loss": 0.7164,
+      "step": 214
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7453391863645898,
+      "learning_rate": 1.718464296695861e-05,
+      "loss": 0.7834,
+      "step": 215
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7417223805399968,
+      "learning_rate": 1.7156424652675433e-05,
+      "loss": 0.7081,
+      "step": 216
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7534810498306946,
+      "learning_rate": 1.7128089045468294e-05,
+      "loss": 0.6968,
+      "step": 217
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7546517064189007,
+      "learning_rate": 1.709963660975433e-05,
+      "loss": 0.7126,
+      "step": 218
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.760563289758312,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.7091,
+      "step": 219
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7581693602018251,
+      "learning_rate": 1.7042383120040837e-05,
+      "loss": 0.7308,
+      "step": 220
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6977006533134363,
+      "learning_rate": 1.7013583004418994e-05,
+      "loss": 0.7031,
+      "step": 221
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8223212695782158,
+      "learning_rate": 1.698466793703032e-05,
+      "loss": 0.759,
+      "step": 222
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.9150130382964546,
+      "learning_rate": 1.695563839178923e-05,
+      "loss": 0.7349,
+      "step": 223
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7330338134515832,
+      "learning_rate": 1.6926494844486412e-05,
+      "loss": 0.6744,
+      "step": 224
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7739183269330984,
+      "learning_rate": 1.6897237772781046e-05,
+      "loss": 0.7137,
+      "step": 225
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.656468237101363,
+      "learning_rate": 1.6867867656192946e-05,
+      "loss": 0.7042,
+      "step": 226
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7454467369978334,
+      "learning_rate": 1.6838384976094738e-05,
+      "loss": 0.7106,
+      "step": 227
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8209720393413101,
+      "learning_rate": 1.6808790215703933e-05,
+      "loss": 0.7491,
+      "step": 228
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6902446956425804,
+      "learning_rate": 1.6779083860075032e-05,
+      "loss": 0.7054,
+      "step": 229
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7880210894086325,
+      "learning_rate": 1.674926639609157e-05,
+      "loss": 0.71,
+      "step": 230
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6652067545160919,
+      "learning_rate": 1.6719338312458123e-05,
+      "loss": 0.7269,
+      "step": 231
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6306135539192167,
+      "learning_rate": 1.6689300099692332e-05,
+      "loss": 0.7134,
+      "step": 232
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.8340897323013752,
+      "learning_rate": 1.665915225011681e-05,
+      "loss": 0.6879,
+      "step": 233
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6466154375134381,
+      "learning_rate": 1.6628895257851136e-05,
+      "loss": 0.7545,
+      "step": 234
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7618511659485634,
+      "learning_rate": 1.65985296188037e-05,
+      "loss": 0.6925,
+      "step": 235
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.8187703179003039,
+      "learning_rate": 1.656805583066361e-05,
+      "loss": 0.7176,
+      "step": 236
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7502264225276143,
+      "learning_rate": 1.6537474392892527e-05,
+      "loss": 0.7095,
+      "step": 237
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8097969662672855,
+      "learning_rate": 1.6506785806716464e-05,
+      "loss": 0.697,
+      "step": 238
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6906146305800908,
+      "learning_rate": 1.6475990575117603e-05,
+      "loss": 0.6869,
+      "step": 239
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8223197770677472,
+      "learning_rate": 1.644508920282601e-05,
+      "loss": 0.753,
+      "step": 240
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7024827757147297,
+      "learning_rate": 1.6414082196311402e-05,
+      "loss": 0.6904,
+      "step": 241
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.706997180401339,
+      "learning_rate": 1.638297006377481e-05,
+      "loss": 0.7003,
+      "step": 242
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.761095407493811,
+      "learning_rate": 1.6351753315140285e-05,
+      "loss": 0.698,
+      "step": 243
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7422167616921054,
+      "learning_rate": 1.6320432462046516e-05,
+      "loss": 0.7186,
+      "step": 244
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7695788913922835,
+      "learning_rate": 1.6289008017838447e-05,
+      "loss": 0.708,
+      "step": 245
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6786701728712313,
+      "learning_rate": 1.6257480497558873e-05,
+      "loss": 0.6793,
+      "step": 246
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7805136584093828,
+      "learning_rate": 1.622585041793999e-05,
+      "loss": 0.7804,
+      "step": 247
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.643788127329463,
+      "learning_rate": 1.6194118297394935e-05,
+      "loss": 0.6932,
+      "step": 248
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7779318756761856,
+      "learning_rate": 1.6162284656009276e-05,
+      "loss": 0.7118,
+      "step": 249
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7827061012286722,
+      "learning_rate": 1.6130350015532498e-05,
+      "loss": 0.7162,
+      "step": 250
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6252425539450749,
+      "learning_rate": 1.6098314899369446e-05,
+      "loss": 0.7002,
+      "step": 251
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6950977115396647,
+      "learning_rate": 1.6066179832571762e-05,
+      "loss": 0.6779,
+      "step": 252
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7564723629412936,
+      "learning_rate": 1.603394534182925e-05,
+      "loss": 0.7426,
+      "step": 253
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6662035677757666,
+      "learning_rate": 1.6001611955461265e-05,
+      "loss": 0.6738,
+      "step": 254
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7443885213732637,
+      "learning_rate": 1.5969180203408052e-05,
+      "loss": 0.6988,
+      "step": 255
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.725792857216776,
+      "learning_rate": 1.5936650617222063e-05,
+      "loss": 0.7183,
+      "step": 256
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.694790268487878,
+      "learning_rate": 1.5904023730059227e-05,
+      "loss": 0.6883,
+      "step": 257
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8056225558740404,
+      "learning_rate": 1.5871300076670236e-05,
+      "loss": 0.7136,
+      "step": 258
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.686737751223464,
+      "learning_rate": 1.5838480193391753e-05,
+      "loss": 0.7702,
+      "step": 259
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7293053389782644,
+      "learning_rate": 1.580556461813766e-05,
+      "loss": 0.7136,
+      "step": 260
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7681622413122055,
+      "learning_rate": 1.5772553890390196e-05,
+      "loss": 0.6989,
+      "step": 261
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6996249418785109,
+      "learning_rate": 1.573944855119115e-05,
+      "loss": 0.7092,
+      "step": 262
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7656196254260119,
+      "learning_rate": 1.5706249143132982e-05,
+      "loss": 0.6912,
+      "step": 263
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7935276488984404,
+      "learning_rate": 1.5672956210349923e-05,
+      "loss": 0.6909,
+      "step": 264
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7590965741522488,
+      "learning_rate": 1.5639570298509067e-05,
+      "loss": 0.7734,
+      "step": 265
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7369044883176309,
+      "learning_rate": 1.560609195480142e-05,
+      "loss": 0.6818,
+      "step": 266
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7879759693028896,
+      "learning_rate": 1.5572521727932937e-05,
+      "loss": 0.6902,
+      "step": 267
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6997745371270365,
+      "learning_rate": 1.5538860168115527e-05,
+      "loss": 0.6972,
+      "step": 268
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.8777129282885041,
+      "learning_rate": 1.5505107827058038e-05,
+      "loss": 0.6896,
+      "step": 269
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.830245073602301,
+      "learning_rate": 1.5471265257957202e-05,
+      "loss": 0.6856,
+      "step": 270
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6849413326975365,
+      "learning_rate": 1.5437333015488586e-05,
+      "loss": 0.7025,
+      "step": 271
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.863043013790657,
+      "learning_rate": 1.5403311655797494e-05,
+      "loss": 0.7513,
+      "step": 272
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.8187297413958853,
+      "learning_rate": 1.536920173648984e-05,
+      "loss": 0.7008,
+      "step": 273
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7844393722686187,
+      "learning_rate": 1.5335003816623027e-05,
+      "loss": 0.6828,
+      "step": 274
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6358387080819407,
+      "learning_rate": 1.530071845669678e-05,
+      "loss": 0.7122,
+      "step": 275
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.735870161613146,
+      "learning_rate": 1.526634621864395e-05,
+      "loss": 0.6833,
+      "step": 276
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7720874652707324,
+      "learning_rate": 1.52318876658213e-05,
+      "loss": 0.7016,
+      "step": 277
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.674584610026373,
+      "learning_rate": 1.5197343363000308e-05,
+      "loss": 0.717,
+      "step": 278
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7038084515462764,
+      "learning_rate": 1.516271387635786e-05,
+      "loss": 0.6707,
+      "step": 279
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6882612272149368,
+      "learning_rate": 1.5127999773467001e-05,
+      "loss": 0.7012,
+      "step": 280
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7024647865541672,
+      "learning_rate": 1.5093201623287631e-05,
+      "loss": 0.7181,
+      "step": 281
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6727023267211898,
+      "learning_rate": 1.5058319996157172e-05,
+      "loss": 0.6602,
+      "step": 282
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7465630000596386,
+      "learning_rate": 1.5023355463781221e-05,
+      "loss": 0.7163,
+      "step": 283
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6984225010098803,
+      "learning_rate": 1.4988308599224182e-05,
+      "loss": 0.7482,
+      "step": 284
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8619915515737648,
+      "learning_rate": 1.4953179976899878e-05,
+      "loss": 0.6766,
+      "step": 285
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7348774289117588,
+      "learning_rate": 1.4917970172562122e-05,
+      "loss": 0.6944,
+      "step": 286
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8366099612874214,
+      "learning_rate": 1.4882679763295307e-05,
+      "loss": 0.6723,
+      "step": 287
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5742961622865171,
+      "learning_rate": 1.484730932750491e-05,
+      "loss": 0.7052,
+      "step": 288
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8026662308952525,
+      "learning_rate": 1.4811859444908053e-05,
+      "loss": 0.6838,
+      "step": 289
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6448865575111775,
+      "learning_rate": 1.4776330696523964e-05,
+      "loss": 0.7157,
+      "step": 290
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.644596649858372,
+      "learning_rate": 1.4740723664664483e-05,
+      "loss": 0.6736,
+      "step": 291
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.668387681674303,
+      "learning_rate": 1.4705038932924502e-05,
+      "loss": 0.7109,
+      "step": 292
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6448680461747227,
+      "learning_rate": 1.4669277086172406e-05,
+      "loss": 0.6836,
+      "step": 293
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7448204272963629,
+      "learning_rate": 1.4633438710540488e-05,
+      "loss": 0.6842,
+      "step": 294
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6434274761308751,
+      "learning_rate": 1.4597524393415336e-05,
+      "loss": 0.703,
+      "step": 295
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7176591670030213,
+      "learning_rate": 1.4561534723428205e-05,
+      "loss": 0.6742,
+      "step": 296
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6284667275951265,
+      "learning_rate": 1.4525470290445392e-05,
+      "loss": 0.7568,
+      "step": 297
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6643417344803181,
+      "learning_rate": 1.4489331685558525e-05,
+      "loss": 0.698,
+      "step": 298
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5893887225885674,
+      "learning_rate": 1.4453119501074924e-05,
+      "loss": 0.6838,
+      "step": 299
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7644763131943855,
+      "learning_rate": 1.4416834330507857e-05,
+      "loss": 0.69,
+      "step": 300
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6363652856489801,
+      "learning_rate": 1.4380476768566825e-05,
+      "loss": 0.6928,
+      "step": 301
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7138641833174041,
+      "learning_rate": 1.434404741114782e-05,
+      "loss": 0.6661,
+      "step": 302
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.68262173814636,
+      "learning_rate": 1.4307546855323549e-05,
+      "loss": 0.7209,
+      "step": 303
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7231471005358049,
+      "learning_rate": 1.4270975699333653e-05,
+      "loss": 0.6663,
+      "step": 304
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6960566739420361,
+      "learning_rate": 1.4234334542574906e-05,
+      "loss": 0.6743,
+      "step": 305
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6939441970735452,
+      "learning_rate": 1.4197623985591373e-05,
+      "loss": 0.7015,
+      "step": 306
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7026781632309379,
+      "learning_rate": 1.4160844630064596e-05,
+      "loss": 0.6844,
+      "step": 307
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7622535513911972,
+      "learning_rate": 1.4123997078803708e-05,
+      "loss": 0.7098,
+      "step": 308
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.72349415439694,
+      "learning_rate": 1.4087081935735565e-05,
+      "loss": 0.7212,
+      "step": 309
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6413870716335239,
+      "learning_rate": 1.4050099805894837e-05,
+      "loss": 0.7123,
+      "step": 310
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6941497000214715,
+      "learning_rate": 1.4013051295414108e-05,
+      "loss": 0.6519,
+      "step": 311
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7075235355277671,
+      "learning_rate": 1.3975937011513931e-05,
+      "loss": 0.7026,
+      "step": 312
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.636952219522705,
+      "learning_rate": 1.3938757562492873e-05,
+      "loss": 0.6792,
+      "step": 313
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.669973911082459,
+      "learning_rate": 1.3901513557717554e-05,
+      "loss": 0.6937,
+      "step": 314
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.679186651508683,
+      "learning_rate": 1.3864205607612648e-05,
+      "loss": 0.7368,
+      "step": 315
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6881658251981118,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 0.6791,
+      "step": 316
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6201517167263211,
+      "learning_rate": 1.378940031834307e-05,
+      "loss": 0.7006,
+      "step": 317
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.697739152945046,
+      "learning_rate": 1.3751904205227922e-05,
+      "loss": 0.6644,
+      "step": 318
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.658040151246998,
+      "learning_rate": 1.3714346598862168e-05,
+      "loss": 0.6703,
+      "step": 319
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6360588523416896,
+      "learning_rate": 1.3676728114810367e-05,
+      "loss": 0.6858,
+      "step": 320
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6931092055073744,
+      "learning_rate": 1.3639049369634878e-05,
+      "loss": 0.7127,
+      "step": 321
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.604853285388519,
+      "learning_rate": 1.3601310980885714e-05,
+      "loss": 0.7177,
+      "step": 322
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6548628153153928,
+      "learning_rate": 1.356351356709045e-05,
+      "loss": 0.6955,
+      "step": 323
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6956475231891892,
+      "learning_rate": 1.3525657747744073e-05,
+      "loss": 0.6704,
+      "step": 324
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6376961636224463,
+      "learning_rate": 1.3487744143298822e-05,
+      "loss": 0.6851,
+      "step": 325
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6432910988957646,
+      "learning_rate": 1.344977337515404e-05,
+      "loss": 0.6687,
+      "step": 326
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7375306077254118,
+      "learning_rate": 1.3411746065645961e-05,
+      "loss": 0.6838,
+      "step": 327
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6490770364006783,
+      "learning_rate": 1.3373662838037538e-05,
+      "loss": 0.7153,
+      "step": 328
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7446844428783186,
+      "learning_rate": 1.3335524316508208e-05,
+      "loss": 0.6543,
+      "step": 329
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.766041681974227,
+      "learning_rate": 1.3297331126143667e-05,
+      "loss": 0.681,
+      "step": 330
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6935412449990047,
+      "learning_rate": 1.3259083892925633e-05,
+      "loss": 0.6915,
+      "step": 331
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6254311054575801,
+      "learning_rate": 1.3220783243721571e-05,
+      "loss": 0.6729,
+      "step": 332
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7762631783044686,
+      "learning_rate": 1.3182429806274442e-05,
+      "loss": 0.689,
+      "step": 333
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6881672074584007,
+      "learning_rate": 1.3144024209192378e-05,
+      "loss": 0.7411,
+      "step": 334
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6358753562338745,
+      "learning_rate": 1.3105567081938423e-05,
+      "loss": 0.6587,
+      "step": 335
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7470874264026129,
+      "learning_rate": 1.3067059054820184e-05,
+      "loss": 0.6764,
+      "step": 336
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7687355342474296,
+      "learning_rate": 1.3028500758979507e-05,
+      "loss": 0.6772,
+      "step": 337
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6627686750796015,
+      "learning_rate": 1.2989892826382144e-05,
+      "loss": 0.7011,
+      "step": 338
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7090967259878612,
+      "learning_rate": 1.2951235889807386e-05,
+      "loss": 0.678,
+      "step": 339
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7084210855766322,
+      "learning_rate": 1.2912530582837683e-05,
+      "loss": 0.7204,
+      "step": 340
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.663549702472892,
+      "learning_rate": 1.2873777539848284e-05,
+      "loss": 0.6595,
+      "step": 341
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6506360572049058,
+      "learning_rate": 1.2834977395996817e-05,
+      "loss": 0.7012,
+      "step": 342
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8430861012936159,
+      "learning_rate": 1.279613078721289e-05,
+      "loss": 0.6719,
+      "step": 343
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6853971357661781,
+      "learning_rate": 1.2757238350187669e-05,
+      "loss": 0.6706,
+      "step": 344
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6984152332324459,
+      "learning_rate": 1.2718300722363431e-05,
+      "loss": 0.6806,
+      "step": 345
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6840075820553941,
+      "learning_rate": 1.2679318541923131e-05,
+      "loss": 0.6747,
+      "step": 346
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6083802270169807,
+      "learning_rate": 1.2640292447779932e-05,
+      "loss": 0.7461,
+      "step": 347
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7286888062718239,
+      "learning_rate": 1.2601223079566745e-05,
+      "loss": 0.6981,
+      "step": 348
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6187807297641909,
+      "learning_rate": 1.2562111077625723e-05,
+      "loss": 0.6639,
+      "step": 349
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7224231721832342,
+      "learning_rate": 1.25229570829978e-05,
+      "loss": 0.6673,
+      "step": 350
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7247095699954746,
+      "learning_rate": 1.248376173741215e-05,
+      "loss": 0.674,
+      "step": 351
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.719452642263008,
+      "learning_rate": 1.2444525683275687e-05,
+      "loss": 0.6541,
+      "step": 352
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6325971772242212,
+      "learning_rate": 1.2405249563662539e-05,
+      "loss": 0.7007,
+      "step": 353
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7513157088192811,
+      "learning_rate": 1.2365934022303491e-05,
+      "loss": 0.6848,
+      "step": 354
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6708081453198337,
+      "learning_rate": 1.2326579703575464e-05,
+      "loss": 0.6696,
+      "step": 355
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6520575494208382,
+      "learning_rate": 1.2287187252490914e-05,
+      "loss": 0.6833,
+      "step": 356
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7203706269009875,
+      "learning_rate": 1.2247757314687296e-05,
+      "loss": 0.6715,
+      "step": 357
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6216048079860456,
+      "learning_rate": 1.2208290536416466e-05,
+      "loss": 0.6673,
+      "step": 358
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6383333297172197,
+      "learning_rate": 1.2168787564534078e-05,
+      "loss": 0.7199,
+      "step": 359
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6272184428646471,
+      "learning_rate": 1.212924904648902e-05,
+      "loss": 0.6811,
+      "step": 360
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6231867328925055,
+      "learning_rate": 1.2089675630312755e-05,
+      "loss": 0.6871,
+      "step": 361
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7010909078929145,
+      "learning_rate": 1.2050067964608725e-05,
+      "loss": 0.6613,
+      "step": 362
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6299035968697418,
+      "learning_rate": 1.2010426698541728e-05,
+      "loss": 0.6758,
+      "step": 363
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7157000181311906,
+      "learning_rate": 1.1970752481827261e-05,
+      "loss": 0.6627,
+      "step": 364
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6454302453667989,
+      "learning_rate": 1.1931045964720882e-05,
+      "loss": 0.726,
+      "step": 365
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7659639276679516,
+      "learning_rate": 1.1891307798007536e-05,
+      "loss": 0.6917,
+      "step": 366
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6755697345701616,
+      "learning_rate": 1.1851538632990922e-05,
+      "loss": 0.6818,
+      "step": 367
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6654538211659489,
+      "learning_rate": 1.1811739121482777e-05,
+      "loss": 0.6467,
+      "step": 368
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6950161014080175,
+      "learning_rate": 1.177190991579223e-05,
+      "loss": 0.6743,
+      "step": 369
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6794644177066992,
+      "learning_rate": 1.1732051668715082e-05,
+      "loss": 0.6829,
+      "step": 370
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6348118454316969,
+      "learning_rate": 1.1692165033523117e-05,
+      "loss": 0.6524,
+      "step": 371
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7090401791841606,
+      "learning_rate": 1.1652250663953415e-05,
+      "loss": 0.733,
+      "step": 372
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6926179531524889,
+      "learning_rate": 1.1612309214197599e-05,
+      "loss": 0.6448,
+      "step": 373
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7072241385311281,
+      "learning_rate": 1.1572341338891145e-05,
+      "loss": 0.6758,
+      "step": 374
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6675592242362295,
+      "learning_rate": 1.1532347693102632e-05,
+      "loss": 0.6936,
+      "step": 375
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6753398878253597,
+      "learning_rate": 1.1492328932323022e-05,
+      "loss": 0.669,
+      "step": 376
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6804988146949109,
+      "learning_rate": 1.1452285712454905e-05,
+      "loss": 0.6773,
+      "step": 377
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7440373259704037,
+      "learning_rate": 1.1412218689801748e-05,
+      "loss": 0.7477,
+      "step": 378
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6646052826352247,
+      "learning_rate": 1.1372128521057155e-05,
+      "loss": 0.6728,
+      "step": 379
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7709667240576361,
+      "learning_rate": 1.1332015863294078e-05,
+      "loss": 0.6343,
+      "step": 380
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7165327110839048,
+      "learning_rate": 1.1291881373954066e-05,
+      "loss": 0.695,
+      "step": 381
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7121053905605914,
+      "learning_rate": 1.125172571083649e-05,
+      "loss": 0.6701,
+      "step": 382
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7327494114274017,
+      "learning_rate": 1.1211549532087749e-05,
+      "loss": 0.695,
+      "step": 383
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7101126814293066,
+      "learning_rate": 1.1171353496190499e-05,
+      "loss": 0.7329,
+      "step": 384
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8069699735523975,
+      "learning_rate": 1.1131138261952845e-05,
+      "loss": 0.6599,
+      "step": 385
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8618469121997681,
+      "learning_rate": 1.109090448849755e-05,
+      "loss": 0.6456,
+      "step": 386
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8182088681876697,
+      "learning_rate": 1.105065283525124e-05,
+      "loss": 0.6772,
+      "step": 387
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6538529876543769,
+      "learning_rate": 1.1010383961933582e-05,
+      "loss": 0.6455,
+      "step": 388
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6448383225287607,
+      "learning_rate": 1.0970098528546482e-05,
+      "loss": 0.6735,
+      "step": 389
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6953896254306905,
+      "learning_rate": 1.0929797195363259e-05,
+      "loss": 0.7333,
+      "step": 390
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6721053785565456,
+      "learning_rate": 1.088948062291783e-05,
+      "loss": 0.6477,
+      "step": 391
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5904282967032243,
+      "learning_rate": 1.0849149471993883e-05,
+      "loss": 0.6787,
+      "step": 392
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6703922439817444,
+      "learning_rate": 1.0808804403614044e-05,
+      "loss": 0.6511,
+      "step": 393
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8705442920527945,
+      "learning_rate": 1.0768446079029044e-05,
+      "loss": 0.6733,
+      "step": 394
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8287230781701082,
+      "learning_rate": 1.0728075159706881e-05,
+      "loss": 0.6605,
+      "step": 395
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6189163109686591,
+      "learning_rate": 1.0687692307321984e-05,
+      "loss": 0.6753,
+      "step": 396
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6781933564255066,
+      "learning_rate": 1.0647298183744359e-05,
+      "loss": 0.7282,
+      "step": 397
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7957473649289525,
+      "learning_rate": 1.0606893451028743e-05,
+      "loss": 0.65,
+      "step": 398
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7485910605540942,
+      "learning_rate": 1.0566478771403763e-05,
+      "loss": 0.6776,
+      "step": 399
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7327909650878064,
+      "learning_rate": 1.0526054807261067e-05,
+      "loss": 0.6706,
+      "step": 400
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7991905466858001,
+      "learning_rate": 1.0485622221144485e-05,
+      "loss": 0.6758,
+      "step": 401
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6809947402682689,
+      "learning_rate": 1.0445181675739144e-05,
+      "loss": 0.6619,
+      "step": 402
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.20009020727991,
+      "learning_rate": 1.0404733833860639e-05,
+      "loss": 0.688,
+      "step": 403
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.903755672099354,
+      "learning_rate": 1.0364279358444144e-05,
+      "loss": 0.6817,
+      "step": 404
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7232572314336939,
+      "learning_rate": 1.0323818912533561e-05,
+      "loss": 0.656,
+      "step": 405
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7379740610126245,
+      "learning_rate": 1.0283353159270644e-05,
+      "loss": 0.6733,
+      "step": 406
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.86964222843065,
+      "learning_rate": 1.0242882761884132e-05,
+      "loss": 0.6601,
+      "step": 407
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6661044375368231,
+      "learning_rate": 1.0202408383678887e-05,
+      "loss": 0.6819,
+      "step": 408
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8253802655023462,
+      "learning_rate": 1.0161930688025018e-05,
+      "loss": 0.7143,
+      "step": 409
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8015741446733053,
+      "learning_rate": 1.012145033834699e-05,
+      "loss": 0.6719,
+      "step": 410
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5821933682049611,
+      "learning_rate": 1.0080967998112787e-05,
+      "loss": 0.6811,
+      "step": 411
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7690954493881962,
+      "learning_rate": 1.0040484330823006e-05,
+      "loss": 0.6565,
+      "step": 412
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7315445864189613,
+      "learning_rate": 1e-05,
+      "loss": 0.6754,
+      "step": 413
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7386622252432524,
+      "learning_rate": 9.959515669176997e-06,
+      "loss": 0.6603,
+      "step": 414
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6919812903907313,
+      "learning_rate": 9.919032001887215e-06,
+      "loss": 0.7076,
+      "step": 415
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8006599563405511,
+      "learning_rate": 9.878549661653013e-06,
+      "loss": 0.6364,
+      "step": 416
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6901988727816551,
+      "learning_rate": 9.838069311974986e-06,
+      "loss": 0.661,
+      "step": 417
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6678547489465795,
+      "learning_rate": 9.797591616321115e-06,
+      "loss": 0.6624,
+      "step": 418
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7748054049097298,
+      "learning_rate": 9.757117238115871e-06,
+      "loss": 0.6451,
+      "step": 419
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7553281040072153,
+      "learning_rate": 9.71664684072936e-06,
+      "loss": 0.6778,
+      "step": 420
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.641131867690177,
+      "learning_rate": 9.676181087466444e-06,
+      "loss": 0.6551,
+      "step": 421
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7291710749634952,
+      "learning_rate": 9.63572064155586e-06,
+      "loss": 0.7335,
+      "step": 422
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7163481736242343,
+      "learning_rate": 9.595266166139366e-06,
+      "loss": 0.6441,
+      "step": 423
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6572156520903942,
+      "learning_rate": 9.55481832426086e-06,
+      "loss": 0.6504,
+      "step": 424
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6559467876911665,
+      "learning_rate": 9.514377778855521e-06,
+      "loss": 0.657,
+      "step": 425
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.662909871152895,
+      "learning_rate": 9.473945192738933e-06,
+      "loss": 0.6645,
+      "step": 426
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7274810191049129,
+      "learning_rate": 9.433521228596237e-06,
+      "loss": 0.6467,
+      "step": 427
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.648261063314425,
+      "learning_rate": 9.393106548971257e-06,
+      "loss": 0.7467,
+      "step": 428
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.657451387169043,
+      "learning_rate": 9.352701816255643e-06,
+      "loss": 0.667,
+      "step": 429
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7124264275789814,
+      "learning_rate": 9.312307692678016e-06,
+      "loss": 0.6421,
+      "step": 430
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6478844634318383,
+      "learning_rate": 9.27192484029312e-06,
+      "loss": 0.688,
+      "step": 431
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6450618353291538,
+      "learning_rate": 9.231553920970958e-06,
+      "loss": 0.6645,
+      "step": 432
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6355390990355307,
+      "learning_rate": 9.19119559638596e-06,
+      "loss": 0.6626,
+      "step": 433
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6654110928484159,
+      "learning_rate": 9.150850528006118e-06,
+      "loss": 0.7181,
+      "step": 434
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.673116583781922,
+      "learning_rate": 9.110519377082174e-06,
+      "loss": 0.6683,
+      "step": 435
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6943712975431352,
+      "learning_rate": 9.070202804636745e-06,
+      "loss": 0.6355,
+      "step": 436
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7727459948269628,
+      "learning_rate": 9.02990147145352e-06,
+      "loss": 0.665,
+      "step": 437
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6426988150098589,
+      "learning_rate": 8.98961603806642e-06,
+      "loss": 0.65,
+      "step": 438
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.659509909598213,
+      "learning_rate": 8.949347164748761e-06,
+      "loss": 0.6709,
+      "step": 439
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7492054794995399,
+      "learning_rate": 8.909095511502452e-06,
+      "loss": 0.7386,
+      "step": 440
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6811842428516481,
+      "learning_rate": 8.868861738047158e-06,
+      "loss": 0.6432,
+      "step": 441
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5727608447671194,
+      "learning_rate": 8.828646503809505e-06,
+      "loss": 0.667,
+      "step": 442
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6886018587603037,
+      "learning_rate": 8.788450467912254e-06,
+      "loss": 0.654,
+      "step": 443
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7074868518593189,
+      "learning_rate": 8.748274289163514e-06,
+      "loss": 0.6586,
+      "step": 444
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7334493370583429,
+      "learning_rate": 8.708118626045939e-06,
+      "loss": 0.657,
+      "step": 445
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8201991046253926,
+      "learning_rate": 8.667984136705927e-06,
+      "loss": 0.6701,
+      "step": 446
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5969030852448933,
+      "learning_rate": 8.62787147894285e-06,
+      "loss": 0.6988,
+      "step": 447
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6576566732974023,
+      "learning_rate": 8.587781310198253e-06,
+      "loss": 0.656,
+      "step": 448
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6919291582377041,
+      "learning_rate": 8.5477142875451e-06,
+      "loss": 0.6261,
+      "step": 449
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7003569661536186,
+      "learning_rate": 8.50767106767698e-06,
+      "loss": 0.6501,
+      "step": 450
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7744163988155506,
+      "learning_rate": 8.46765230689737e-06,
+      "loss": 0.6383,
+      "step": 451
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6650808234702702,
+      "learning_rate": 8.427658661108857e-06,
+      "loss": 0.6645,
+      "step": 452
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6583815924553357,
+      "learning_rate": 8.387690785802403e-06,
+      "loss": 0.6552,
+      "step": 453
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6496240712606984,
+      "learning_rate": 8.347749336046587e-06,
+      "loss": 0.6678,
+      "step": 454
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6920289415492432,
+      "learning_rate": 8.307834966476885e-06,
+      "loss": 0.6852,
+      "step": 455
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.604457834334182,
+      "learning_rate": 8.267948331284923e-06,
+      "loss": 0.6591,
+      "step": 456
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6682155428303499,
+      "learning_rate": 8.228090084207773e-06,
+      "loss": 0.6561,
+      "step": 457
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7355914531635479,
+      "learning_rate": 8.188260878517224e-06,
+      "loss": 0.6581,
+      "step": 458
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6564122035284753,
+      "learning_rate": 8.148461367009081e-06,
+      "loss": 0.6918,
+      "step": 459
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6538360721528242,
+      "learning_rate": 8.108692201992466e-06,
+      "loss": 0.6542,
+      "step": 460
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.674957820323119,
+      "learning_rate": 8.068954035279121e-06,
+      "loss": 0.6692,
+      "step": 461
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6380977695912416,
+      "learning_rate": 8.02924751817274e-06,
+      "loss": 0.6391,
+      "step": 462
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.587744980514578,
+      "learning_rate": 7.989573301458274e-06,
+      "loss": 0.6637,
+      "step": 463
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6805705907418358,
+      "learning_rate": 7.949932035391279e-06,
+      "loss": 0.6667,
+      "step": 464
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6270081521977198,
+      "learning_rate": 7.91032436968725e-06,
+      "loss": 0.7029,
+      "step": 465
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6742300289735369,
+      "learning_rate": 7.870750953510983e-06,
+      "loss": 0.6603,
+      "step": 466
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8924071754646948,
+      "learning_rate": 7.831212435465925e-06,
+      "loss": 0.6694,
+      "step": 467
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5876531309244762,
+      "learning_rate": 7.791709463583541e-06,
+      "loss": 0.6478,
+      "step": 468
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7553792523453502,
+      "learning_rate": 7.752242685312709e-06,
+      "loss": 0.6335,
+      "step": 469
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5954675056029863,
+      "learning_rate": 7.712812747509091e-06,
+      "loss": 0.6599,
+      "step": 470
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.693082993076316,
+      "learning_rate": 7.673420296424541e-06,
+      "loss": 0.6537,
+      "step": 471
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6374690276629207,
+      "learning_rate": 7.63406597769651e-06,
+      "loss": 0.702,
+      "step": 472
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6350743256053588,
+      "learning_rate": 7.594750436337467e-06,
+      "loss": 0.6354,
+      "step": 473
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6142273470612479,
+      "learning_rate": 7.5554743167243135e-06,
+      "loss": 0.6545,
+      "step": 474
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7377267478799707,
+      "learning_rate": 7.516238262587851e-06,
+      "loss": 0.6664,
+      "step": 475
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6263530432538805,
+      "learning_rate": 7.4770429170022e-06,
+      "loss": 0.6484,
+      "step": 476
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6785776573913042,
+      "learning_rate": 7.4378889223742766e-06,
+      "loss": 0.631,
+      "step": 477
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6638092030648043,
+      "learning_rate": 7.398776920433257e-06,
+      "loss": 0.7161,
+      "step": 478
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7290380027719847,
+      "learning_rate": 7.35970755222007e-06,
+      "loss": 0.6701,
+      "step": 479
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7051357789907977,
+      "learning_rate": 7.320681458076871e-06,
+      "loss": 0.6357,
+      "step": 480
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6047614601388198,
+      "learning_rate": 7.2816992776365714e-06,
+      "loss": 0.6591,
+      "step": 481
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7301702851880185,
+      "learning_rate": 7.2427616498123356e-06,
+      "loss": 0.6502,
+      "step": 482
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.713717801487329,
+      "learning_rate": 7.203869212787112e-06,
+      "loss": 0.6391,
+      "step": 483
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7065239599437015,
+      "learning_rate": 7.165022604003187e-06,
+      "loss": 0.7297,
+      "step": 484
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6200614878322726,
+      "learning_rate": 7.126222460151719e-06,
+      "loss": 0.6432,
+      "step": 485
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6539748387323382,
+      "learning_rate": 7.08746941716232e-06,
+      "loss": 0.6165,
+      "step": 486
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7012548229981432,
+      "learning_rate": 7.048764110192618e-06,
+      "loss": 0.6583,
+      "step": 487
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.713792656054762,
+      "learning_rate": 7.010107173617857e-06,
+      "loss": 0.6277,
+      "step": 488
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6081705206738856,
+      "learning_rate": 6.971499241020495e-06,
+      "loss": 0.672,
+      "step": 489
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6784237349853492,
+      "learning_rate": 6.932940945179818e-06,
+      "loss": 0.7033,
+      "step": 490
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7241290103475653,
+      "learning_rate": 6.894432918061579e-06,
+      "loss": 0.669,
+      "step": 491
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5695616074610608,
+      "learning_rate": 6.855975790807623e-06,
+      "loss": 0.6577,
+      "step": 492
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7484769455712518,
+      "learning_rate": 6.8175701937255645e-06,
+      "loss": 0.6501,
+      "step": 493
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6614185441303015,
+      "learning_rate": 6.77921675627843e-06,
+      "loss": 0.6277,
+      "step": 494
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.672200149137363,
+      "learning_rate": 6.740916107074372e-06,
+      "loss": 0.6458,
+      "step": 495
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7200106295372684,
+      "learning_rate": 6.702668873856339e-06,
+      "loss": 0.6614,
+      "step": 496
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6641990003364376,
+      "learning_rate": 6.664475683491797e-06,
+      "loss": 0.7036,
+      "step": 497
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5879289278966713,
+      "learning_rate": 6.6263371619624615e-06,
+      "loss": 0.6549,
+      "step": 498
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7248226505864468,
+      "learning_rate": 6.588253934354039e-06,
+      "loss": 0.6248,
+      "step": 499
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7488352629327707,
+      "learning_rate": 6.550226624845961e-06,
+      "loss": 0.6592,
+      "step": 500
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7119848368674065,
+      "learning_rate": 6.5122558567011775e-06,
+      "loss": 0.6358,
+      "step": 501
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6187967774882688,
+      "learning_rate": 6.474342252255927e-06,
+      "loss": 0.662,
+      "step": 502
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6447575199279856,
+      "learning_rate": 6.43648643290955e-06,
+      "loss": 0.6374,
+      "step": 503
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.579615602622246,
+      "learning_rate": 6.398689019114289e-06,
+      "loss": 0.6393,
+      "step": 504
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6939331778963205,
+      "learning_rate": 6.360950630365126e-06,
+      "loss": 0.6488,
+      "step": 505
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6216844743810305,
+      "learning_rate": 6.323271885189636e-06,
+      "loss": 0.6771,
+      "step": 506
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6636839389062641,
+      "learning_rate": 6.2856534011378365e-06,
+      "loss": 0.6174,
+      "step": 507
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8288156576061698,
+      "learning_rate": 6.24809579477208e-06,
+      "loss": 0.6561,
+      "step": 508
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6360493953719707,
+      "learning_rate": 6.210599681656933e-06,
+      "loss": 0.6912,
+      "step": 509
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6021297745464462,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.6426,
+      "step": 510
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6152387963399792,
+      "learning_rate": 6.135794392387353e-06,
+      "loss": 0.6488,
+      "step": 511
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7904840795998466,
+      "learning_rate": 6.09848644228245e-06,
+      "loss": 0.6313,
+      "step": 512
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.593113398457472,
+      "learning_rate": 6.061242437507131e-06,
+      "loss": 0.6536,
+      "step": 513
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6838676801534896,
+      "learning_rate": 6.024062988486072e-06,
+      "loss": 0.6536,
+      "step": 514
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5852968494358796,
+      "learning_rate": 5.986948704585895e-06,
+      "loss": 0.6833,
+      "step": 515
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6880650442925168,
+      "learning_rate": 5.949900194105167e-06,
+      "loss": 0.615,
+      "step": 516
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6963184620091328,
+      "learning_rate": 5.912918064264441e-06,
+      "loss": 0.6802,
+      "step": 517
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.641163426385126,
+      "learning_rate": 5.876002921196296e-06,
+      "loss": 0.6252,
+      "step": 518
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6611089593608891,
+      "learning_rate": 5.839155369935407e-06,
+      "loss": 0.6215,
+      "step": 519
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5755346633727011,
+      "learning_rate": 5.802376014408632e-06,
+      "loss": 0.6653,
+      "step": 520
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.689385738981433,
+      "learning_rate": 5.765665457425102e-06,
+      "loss": 0.6449,
+      "step": 521
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5943498129532476,
+      "learning_rate": 5.729024300666349e-06,
+      "loss": 0.6927,
+      "step": 522
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.75720345607242,
+      "learning_rate": 5.692453144676451e-06,
+      "loss": 0.6395,
+      "step": 523
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6431721527289079,
+      "learning_rate": 5.655952588852181e-06,
+      "loss": 0.6454,
+      "step": 524
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5908943263514355,
+      "learning_rate": 5.619523231433177e-06,
+      "loss": 0.6473,
+      "step": 525
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7212254015768811,
+      "learning_rate": 5.5831656694921465e-06,
+      "loss": 0.6558,
+      "step": 526
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6081843034998314,
+      "learning_rate": 5.546880498925079e-06,
+      "loss": 0.6347,
+      "step": 527
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.578463789906364,
+      "learning_rate": 5.510668314441474e-06,
+      "loss": 0.6809,
+      "step": 528
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.615140245545869,
+      "learning_rate": 5.4745297095546125e-06,
+      "loss": 0.6416,
+      "step": 529
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6672837804788065,
+      "learning_rate": 5.438465276571796e-06,
+      "loss": 0.6237,
+      "step": 530
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6445792457327395,
+      "learning_rate": 5.40247560658467e-06,
+      "loss": 0.6536,
+      "step": 531
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6462547880358498,
+      "learning_rate": 5.366561289459512e-06,
+      "loss": 0.649,
+      "step": 532
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6221813752371448,
+      "learning_rate": 5.330722913827594e-06,
+      "loss": 0.6484,
+      "step": 533
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6205058787869572,
+      "learning_rate": 5.2949610670755e-06,
+      "loss": 0.7018,
+      "step": 534
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6968523348168115,
+      "learning_rate": 5.259276335335522e-06,
+      "loss": 0.6455,
+      "step": 535
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.630885368888865,
+      "learning_rate": 5.223669303476041e-06,
+      "loss": 0.6319,
+      "step": 536
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6781795525853452,
+      "learning_rate": 5.18814055509195e-06,
+      "loss": 0.6636,
+      "step": 537
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6848858498103306,
+      "learning_rate": 5.152690672495091e-06,
+      "loss": 0.6428,
+      "step": 538
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6468785611746863,
+      "learning_rate": 5.117320236704697e-06,
+      "loss": 0.6527,
+      "step": 539
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.4458831989646992,
+      "learning_rate": 5.08202982743788e-06,
+      "loss": 0.6941,
+      "step": 540
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6472389955425294,
+      "learning_rate": 5.046820023100129e-06,
+      "loss": 0.6284,
+      "step": 541
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6635680749476713,
+      "learning_rate": 5.01169140077582e-06,
+      "loss": 0.6492,
+      "step": 542
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6971370544968278,
+      "learning_rate": 4.976644536218783e-06,
+      "loss": 0.6266,
+      "step": 543
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7794954579271794,
+      "learning_rate": 4.9416800038428326e-06,
+      "loss": 0.6448,
+      "step": 544
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6799095465030572,
+      "learning_rate": 4.9067983767123736e-06,
+      "loss": 0.6514,
+      "step": 545
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6532432700542071,
+      "learning_rate": 4.872000226533001e-06,
+      "loss": 0.6464,
+      "step": 546
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6994117712449526,
+      "learning_rate": 4.837286123642141e-06,
+      "loss": 0.7218,
+      "step": 547
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5870973770863565,
+      "learning_rate": 4.802656636999693e-06,
+      "loss": 0.6633,
+      "step": 548
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8013741958726535,
+      "learning_rate": 4.7681123341787e-06,
+      "loss": 0.6072,
+      "step": 549
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6129642891356658,
+      "learning_rate": 4.733653781356055e-06,
+      "loss": 0.6532,
+      "step": 550
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7076801695534043,
+      "learning_rate": 4.699281543303222e-06,
+      "loss": 0.633,
+      "step": 551
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6236629866450143,
+      "learning_rate": 4.664996183376972e-06,
+      "loss": 0.645,
+      "step": 552
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.609684517349269,
+      "learning_rate": 4.630798263510162e-06,
+      "loss": 0.6458,
+      "step": 553
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5795979463866703,
+      "learning_rate": 4.596688344202509e-06,
+      "loss": 0.6523,
+      "step": 554
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6807246005682629,
+      "learning_rate": 4.562666984511416e-06,
+      "loss": 0.6105,
+      "step": 555
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5524238266816994,
+      "learning_rate": 4.528734742042803e-06,
+      "loss": 0.6605,
+      "step": 556
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.660370294991325,
+      "learning_rate": 4.494892172941965e-06,
+      "loss": 0.6241,
+      "step": 557
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7690865932483458,
+      "learning_rate": 4.461139831884475e-06,
+      "loss": 0.6614,
+      "step": 558
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6270039043389054,
+      "learning_rate": 4.427478272067066e-06,
+      "loss": 0.6937,
+      "step": 559
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6231361031902123,
+      "learning_rate": 4.393908045198585e-06,
+      "loss": 0.6427,
+      "step": 560
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5590129123246664,
+      "learning_rate": 4.360429701490935e-06,
+      "loss": 0.6442,
+      "step": 561
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6514974942052368,
+      "learning_rate": 4.327043789650078e-06,
+      "loss": 0.6377,
+      "step": 562
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.748539607647128,
+      "learning_rate": 4.2937508568670194e-06,
+      "loss": 0.6398,
+      "step": 563
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6361382498031051,
+      "learning_rate": 4.260551448808852e-06,
+      "loss": 0.6659,
+      "step": 564
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5969276356602796,
+      "learning_rate": 4.2274461096098085e-06,
+      "loss": 0.7093,
+      "step": 565
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6703106363143092,
+      "learning_rate": 4.194435381862343e-06,
+      "loss": 0.6194,
+      "step": 566
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6425376870707464,
+      "learning_rate": 4.1615198066082475e-06,
+      "loss": 0.6556,
+      "step": 567
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.591269622613119,
+      "learning_rate": 4.12869992332977e-06,
+      "loss": 0.6524,
+      "step": 568
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7260321894324775,
+      "learning_rate": 4.095976269940777e-06,
+      "loss": 0.6416,
+      "step": 569
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5583461510051794,
+      "learning_rate": 4.0633493827779425e-06,
+      "loss": 0.6466,
+      "step": 570
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7277733025288593,
+      "learning_rate": 4.03081979659195e-06,
+      "loss": 0.6579,
+      "step": 571
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6517350378102095,
+      "learning_rate": 3.998388044538737e-06,
+      "loss": 0.6547,
+      "step": 572
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7853172538242879,
+      "learning_rate": 3.966054658170754e-06,
+      "loss": 0.6917,
+      "step": 573
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5838898626379716,
+      "learning_rate": 3.933820167428241e-06,
+      "loss": 0.6378,
+      "step": 574
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6112863650448396,
+      "learning_rate": 3.901685100630554e-06,
+      "loss": 0.6607,
+      "step": 575
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6843097085761197,
+      "learning_rate": 3.869649984467504e-06,
+      "loss": 0.6462,
+      "step": 576
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5914223449212442,
+      "learning_rate": 3.837715343990727e-06,
+      "loss": 0.6177,
+      "step": 577
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5972948686649531,
+      "learning_rate": 3.8058817026050676e-06,
+      "loss": 0.7189,
+      "step": 578
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5915925626028785,
+      "learning_rate": 3.7741495820600128e-06,
+      "loss": 0.6428,
+      "step": 579
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7114421940143537,
+      "learning_rate": 3.742519502441132e-06,
+      "loss": 0.653,
+      "step": 580
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6512983655056599,
+      "learning_rate": 3.7109919821615546e-06,
+      "loss": 0.6562,
+      "step": 581
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5714596455670929,
+      "learning_rate": 3.6795675379534857e-06,
+      "loss": 0.6434,
+      "step": 582
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5773249093636545,
+      "learning_rate": 3.6482466848597164e-06,
+      "loss": 0.6445,
+      "step": 583
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.644524398284823,
+      "learning_rate": 3.6170299362251926e-06,
+      "loss": 0.7094,
+      "step": 584
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6432991090373537,
+      "learning_rate": 3.585917803688603e-06,
+      "loss": 0.6549,
+      "step": 585
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6193127063887414,
+      "learning_rate": 3.5549107971739905e-06,
+      "loss": 0.645,
+      "step": 586
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6556133044492171,
+      "learning_rate": 3.5240094248824e-06,
+      "loss": 0.6462,
+      "step": 587
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6534604203544768,
+      "learning_rate": 3.4932141932835362e-06,
+      "loss": 0.6377,
+      "step": 588
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.574479705465771,
+      "learning_rate": 3.4625256071074776e-06,
+      "loss": 0.6581,
+      "step": 589
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7167915390508185,
+      "learning_rate": 3.431944169336391e-06,
+      "loss": 0.6998,
+      "step": 590
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5753509546170014,
+      "learning_rate": 3.4014703811963024e-06,
+      "loss": 0.627,
+      "step": 591
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.547677827015996,
+      "learning_rate": 3.3711047421488676e-06,
+      "loss": 0.654,
+      "step": 592
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6849762180268518,
+      "learning_rate": 3.3408477498831917e-06,
+      "loss": 0.6191,
+      "step": 593
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7317642874060407,
+      "learning_rate": 3.3106999003076745e-06,
+      "loss": 0.6112,
+      "step": 594
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5689472355475474,
+      "learning_rate": 3.280661687541876e-06,
+      "loss": 0.6573,
+      "step": 595
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6342221541841166,
+      "learning_rate": 3.2507336039084315e-06,
+      "loss": 0.6377,
+      "step": 596
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6966057591584099,
+      "learning_rate": 3.2209161399249677e-06,
+      "loss": 0.71,
+      "step": 597
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6590124389889497,
+      "learning_rate": 3.1912097842960676e-06,
+      "loss": 0.6376,
+      "step": 598
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6238127220128191,
+      "learning_rate": 3.1616150239052647e-06,
+      "loss": 0.6216,
+      "step": 599
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6271365515219529,
+      "learning_rate": 3.132132343807056e-06,
+      "loss": 0.6394,
+      "step": 600
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6831564062808797,
+      "learning_rate": 3.1027622272189572e-06,
+      "loss": 0.6205,
+      "step": 601
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6404577568502366,
+      "learning_rate": 3.073505155513591e-06,
+      "loss": 0.6269,
+      "step": 602
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6402081470277372,
+      "learning_rate": 3.0443616082107753e-06,
+      "loss": 0.6381,
+      "step": 603
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5989734816151734,
+      "learning_rate": 3.015332062969685e-06,
+      "loss": 0.637,
+      "step": 604
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.695960785486385,
+      "learning_rate": 2.9864169955810085e-06,
+      "loss": 0.616,
+      "step": 605
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6023623426765436,
+      "learning_rate": 2.9576168799591663e-06,
+      "loss": 0.6553,
+      "step": 606
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5828529926677192,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.6286,
+      "step": 607
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6808201103441773,
+      "learning_rate": 2.900363390245674e-06,
+      "loss": 0.6264,
+      "step": 608
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6190621367944866,
+      "learning_rate": 2.8719109545317102e-06,
+      "loss": 0.7047,
+      "step": 609
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5908794901355795,
+      "learning_rate": 2.8435753473245697e-06,
+      "loss": 0.6349,
+      "step": 610
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.592429740136109,
+      "learning_rate": 2.8153570330413925e-06,
+      "loss": 0.6402,
+      "step": 611
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6963991041241643,
+      "learning_rate": 2.7872564741768917e-06,
+      "loss": 0.6456,
+      "step": 612
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5719804647487756,
+      "learning_rate": 2.759274131295787e-06,
+      "loss": 0.6349,
+      "step": 613
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6065837845826134,
+      "learning_rate": 2.7314104630252502e-06,
+      "loss": 0.6295,
+      "step": 614
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5997674456838561,
+      "learning_rate": 2.7036659260473973e-06,
+      "loss": 0.7026,
+      "step": 615
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6019462922224683,
+      "learning_rate": 2.6760409750917925e-06,
+      "loss": 0.6259,
+      "step": 616
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6341850666653158,
+      "learning_rate": 2.648536062927999e-06,
+      "loss": 0.6449,
+      "step": 617
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3001860073086675,
+      "learning_rate": 2.6211516403581585e-06,
+      "loss": 0.6439,
+      "step": 618
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5671409191707228,
+      "learning_rate": 2.593888156209603e-06,
+      "loss": 0.6429,
+      "step": 619
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6096302431352558,
+      "learning_rate": 2.5667460573275028e-06,
+      "loss": 0.6566,
+      "step": 620
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6524426099497527,
+      "learning_rate": 2.5397257885675396e-06,
+      "loss": 0.6372,
+      "step": 621
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5815091971020746,
+      "learning_rate": 2.512827792788606e-06,
+      "loss": 0.6302,
+      "step": 622
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6235790733836095,
+      "learning_rate": 2.48605251084556e-06,
+      "loss": 0.6964,
+      "step": 623
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6657262678609176,
+      "learning_rate": 2.459400381581997e-06,
+      "loss": 0.6425,
+      "step": 624
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5226966338040596,
+      "learning_rate": 2.432871841823047e-06,
+      "loss": 0.6515,
+      "step": 625
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6363516726370335,
+      "learning_rate": 2.406467326368237e-06,
+      "loss": 0.6308,
+      "step": 626
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6302163611631845,
+      "learning_rate": 2.3801872679843384e-06,
+      "loss": 0.6395,
+      "step": 627
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6404972641350611,
+      "learning_rate": 2.3540320973982924e-06,
+      "loss": 0.6441,
+      "step": 628
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5870759209949381,
+      "learning_rate": 2.328002243290138e-06,
+      "loss": 0.684,
+      "step": 629
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6477566587591898,
+      "learning_rate": 2.3020981322860057e-06,
+      "loss": 0.6065,
+      "step": 630
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7014352005135476,
+      "learning_rate": 2.2763201889510987e-06,
+      "loss": 0.6482,
+      "step": 631
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5847910108336893,
+      "learning_rate": 2.2506688357827546e-06,
+      "loss": 0.6379,
+      "step": 632
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8424888111482473,
+      "learning_rate": 2.2251444932035094e-06,
+      "loss": 0.6329,
+      "step": 633
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5889389063340384,
+      "learning_rate": 2.1997475795542113e-06,
+      "loss": 0.7132,
+      "step": 634
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6222285478118977,
+      "learning_rate": 2.1744785110871713e-06,
+      "loss": 0.6449,
+      "step": 635
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6184736487279212,
+      "learning_rate": 2.149337701959325e-06,
+      "loss": 0.6401,
+      "step": 636
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6721155524494571,
+      "learning_rate": 2.124325564225458e-06,
+      "loss": 0.6042,
+      "step": 637
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7082382392952378,
+      "learning_rate": 2.099442507831444e-06,
+      "loss": 0.6277,
+      "step": 638
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5746323943249898,
+      "learning_rate": 2.074688940607529e-06,
+      "loss": 0.6534,
+      "step": 639
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6976877229331887,
+      "learning_rate": 2.050065268261655e-06,
+      "loss": 0.6784,
+      "step": 640
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5627102288511716,
+      "learning_rate": 2.025571894372794e-06,
+      "loss": 0.6289,
+      "step": 641
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6172828112978435,
+      "learning_rate": 2.001209220384346e-06,
+      "loss": 0.6294,
+      "step": 642
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5662594447199433,
+      "learning_rate": 1.976977645597552e-06,
+      "loss": 0.6436,
+      "step": 643
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7074464895311687,
+      "learning_rate": 1.9528775671649593e-06,
+      "loss": 0.6173,
+      "step": 644
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6554966832321535,
+      "learning_rate": 1.9289093800839067e-06,
+      "loss": 0.6396,
+      "step": 645
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5872156699340129,
+      "learning_rate": 1.9050734771900414e-06,
+      "loss": 0.6371,
+      "step": 646
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6511462069083932,
+      "learning_rate": 1.8813702491508956e-06,
+      "loss": 0.7005,
+      "step": 647
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5779512437776096,
+      "learning_rate": 1.8578000844594746e-06,
+      "loss": 0.6329,
+      "step": 648
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.673722716371842,
+      "learning_rate": 1.8343633694278895e-06,
+      "loss": 0.6256,
+      "step": 649
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5551784609278542,
+      "learning_rate": 1.8110604881810357e-06,
+      "loss": 0.6267,
+      "step": 650
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8001123157526435,
+      "learning_rate": 1.7878918226502816e-06,
+      "loss": 0.6534,
+      "step": 651
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5597566957231792,
+      "learning_rate": 1.7648577525672195e-06,
+      "loss": 0.6202,
+      "step": 652
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5524746395213825,
+      "learning_rate": 1.7419586554574364e-06,
+      "loss": 0.6055,
+      "step": 653
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.6925034270245604,
+      "learning_rate": 1.7191949066343306e-06,
+      "loss": 0.6506,
+      "step": 654
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5982581710140573,
+      "learning_rate": 1.69656687919296e-06,
+      "loss": 0.6036,
+      "step": 655
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.625640893394704,
+      "learning_rate": 1.6740749440039262e-06,
+      "loss": 0.6613,
+      "step": 656
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5791710291449589,
+      "learning_rate": 1.6517194697072903e-06,
+      "loss": 0.6189,
+      "step": 657
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.6645440340948221,
+      "learning_rate": 1.6295008227065367e-06,
+      "loss": 0.6144,
+      "step": 658
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.668105235928013,
+      "learning_rate": 1.607419367162577e-06,
+      "loss": 0.7108,
+      "step": 659
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5385899476858674,
+      "learning_rate": 1.58547546498776e-06,
+      "loss": 0.6441,
+      "step": 660
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6843312572641347,
+      "learning_rate": 1.5636694758399563e-06,
+      "loss": 0.6146,
+      "step": 661
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6517691074145857,
+      "learning_rate": 1.542001757116658e-06,
+      "loss": 0.6446,
+      "step": 662
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6683044014030128,
+      "learning_rate": 1.520472663949122e-06,
+      "loss": 0.6259,
+      "step": 663
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5529928661133248,
+      "learning_rate": 1.4990825491965522e-06,
+      "loss": 0.6574,
+      "step": 664
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6354267241399787,
+      "learning_rate": 1.4778317634403082e-06,
+      "loss": 0.6859,
+      "step": 665
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6340161097980351,
+      "learning_rate": 1.4567206549781699e-06,
+      "loss": 0.6336,
+      "step": 666
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6857637368517085,
+      "learning_rate": 1.4357495698186186e-06,
+      "loss": 0.6634,
+      "step": 667
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6987918825152238,
+      "learning_rate": 1.41491885167517e-06,
+      "loss": 0.631,
+      "step": 668
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5905065070051041,
+      "learning_rate": 1.3942288419607476e-06,
+      "loss": 0.6318,
+      "step": 669
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5891202117275015,
+      "learning_rate": 1.3736798797820783e-06,
+      "loss": 0.6257,
+      "step": 670
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6389989609910032,
+      "learning_rate": 1.3532723019341376e-06,
+      "loss": 0.633,
+      "step": 671
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7481668062509925,
+      "learning_rate": 1.3330064428946255e-06,
+      "loss": 0.621,
+      "step": 672
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6201805967439965,
+      "learning_rate": 1.3128826348184886e-06,
+      "loss": 0.6891,
+      "step": 673
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.711558886670601,
+      "learning_rate": 1.2929012075324832e-06,
+      "loss": 0.6331,
+      "step": 674
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5601641885965722,
+      "learning_rate": 1.2730624885297537e-06,
+      "loss": 0.6402,
+      "step": 675
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7003049669889405,
+      "learning_rate": 1.2533668029644751e-06,
+      "loss": 0.6392,
+      "step": 676
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6512766895573538,
+      "learning_rate": 1.233814473646524e-06,
+      "loss": 0.6435,
+      "step": 677
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6304938936760185,
+      "learning_rate": 1.214405821036182e-06,
+      "loss": 0.6402,
+      "step": 678
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6453066742327469,
+      "learning_rate": 1.195141163238892e-06,
+      "loss": 0.6906,
+      "step": 679
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7791569651001938,
+      "learning_rate": 1.1760208160000364e-06,
+      "loss": 0.6327,
+      "step": 680
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6242645456514428,
+      "learning_rate": 1.1570450926997657e-06,
+      "loss": 0.6259,
+      "step": 681
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5852932502584727,
+      "learning_rate": 1.1382143043478599e-06,
+      "loss": 0.6558,
+      "step": 682
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7187172398967998,
+      "learning_rate": 1.1195287595786352e-06,
+      "loss": 0.633,
+      "step": 683
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.571420842271662,
+      "learning_rate": 1.1009887646458862e-06,
+      "loss": 0.6893,
+      "step": 684
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6286607515904005,
+      "learning_rate": 1.0825946234178575e-06,
+      "loss": 0.6251,
+      "step": 685
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5889490025349474,
+      "learning_rate": 1.064346637372271e-06,
+      "loss": 0.6456,
+      "step": 686
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6780378250552338,
+      "learning_rate": 1.0462451055913847e-06,
+      "loss": 0.6158,
+      "step": 687
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6488064822153722,
+      "learning_rate": 1.0282903247570908e-06,
+      "loss": 0.6014,
+      "step": 688
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5968658741723634,
+      "learning_rate": 1.010482589146048e-06,
+      "loss": 0.6446,
+      "step": 689
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6634490640695445,
+      "learning_rate": 9.928221906248614e-07,
+      "loss": 0.6842,
+      "step": 690
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6651655353815971,
+      "learning_rate": 9.753094186453028e-07,
+      "loss": 0.6141,
+      "step": 691
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5581435148846154,
+      "learning_rate": 9.579445602395577e-07,
+      "loss": 0.632,
+      "step": 692
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6312888479285492,
+      "learning_rate": 9.407279000155311e-07,
+      "loss": 0.6499,
+      "step": 693
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6905760011351594,
+      "learning_rate": 9.23659720152179e-07,
+      "loss": 0.619,
+      "step": 694
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6758689934885078,
+      "learning_rate": 9.067403003948783e-07,
+      "loss": 0.6512,
+      "step": 695
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5580070435331436,
+      "learning_rate": 8.89969918050847e-07,
+      "loss": 0.6426,
+      "step": 696
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6788609323246715,
+      "learning_rate": 8.733488479845997e-07,
+      "loss": 0.6926,
+      "step": 697
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7711818911159529,
+      "learning_rate": 8.568773626134363e-07,
+      "loss": 0.6431,
+      "step": 698
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5264819335472163,
+      "learning_rate": 8.405557319029911e-07,
+      "loss": 0.6346,
+      "step": 699
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6539755754682118,
+      "learning_rate": 8.243842233627897e-07,
+      "loss": 0.6088,
+      "step": 700
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6276884828047339,
+      "learning_rate": 8.083631020418792e-07,
+      "loss": 0.6392,
+      "step": 701
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6242556779335853,
+      "learning_rate": 7.924926305244729e-07,
+      "loss": 0.6303,
+      "step": 702
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6450261398029767,
+      "learning_rate": 7.767730689256614e-07,
+      "loss": 0.655,
+      "step": 703
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.704079501490032,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 0.627,
+      "step": 704
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6396858617866267,
+      "learning_rate": 7.457877035729588e-07,
+      "loss": 0.6174,
+      "step": 705
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6287790509868376,
+      "learning_rate": 7.305224076654127e-07,
+      "loss": 0.6635,
+      "step": 706
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.873048359207114,
+      "learning_rate": 7.154090373608236e-07,
+      "loss": 0.6388,
+      "step": 707
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5881138487876106,
+      "learning_rate": 7.004478403654835e-07,
+      "loss": 0.6062,
+      "step": 708
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.657388597051673,
+      "learning_rate": 6.856390618915775e-07,
+      "loss": 0.6897,
+      "step": 709
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5485546716949598,
+      "learning_rate": 6.709829446531734e-07,
+      "loss": 0.6541,
+      "step": 710
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6699766341713999,
+      "learning_rate": 6.564797288622371e-07,
+      "loss": 0.6084,
+      "step": 711
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6040708813799974,
+      "learning_rate": 6.421296522247012e-07,
+      "loss": 0.6341,
+      "step": 712
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7310239374931192,
+      "learning_rate": 6.279329499365649e-07,
+      "loss": 0.6353,
+      "step": 713
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5800753989859047,
+      "learning_rate": 6.138898546800398e-07,
+      "loss": 0.6628,
+      "step": 714
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6185190468285704,
+      "learning_rate": 6.000005966197387e-07,
+      "loss": 0.6831,
+      "step": 715
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6615470484183518,
+      "learning_rate": 5.86265403398899e-07,
+      "loss": 0.6292,
+      "step": 716
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6671132213915263,
+      "learning_rate": 5.726845001356573e-07,
+      "loss": 0.615,
+      "step": 717
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.710456836405862,
+      "learning_rate": 5.592581094193584e-07,
+      "loss": 0.6159,
+      "step": 718
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6052691264538014,
+      "learning_rate": 5.459864513068991e-07,
+      "loss": 0.6399,
+      "step": 719
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6425766862473259,
+      "learning_rate": 5.328697433191321e-07,
+      "loss": 0.6202,
+      "step": 720
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.581802218379288,
+      "learning_rate": 5.199082004372958e-07,
+      "loss": 0.6338,
+      "step": 721
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7207535136328385,
+      "learning_rate": 5.071020350994893e-07,
+      "loss": 0.6336,
+      "step": 722
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.606403956290431,
+      "learning_rate": 4.944514571971981e-07,
+      "loss": 0.6877,
+      "step": 723
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6890047165080336,
+      "learning_rate": 4.81956674071844e-07,
+      "loss": 0.6092,
+      "step": 724
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5361959776088648,
+      "learning_rate": 4.696178905113913e-07,
+      "loss": 0.6479,
+      "step": 725
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7361321178970542,
+      "learning_rate": 4.5743530874699293e-07,
+      "loss": 0.6283,
+      "step": 726
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6062549561428903,
+      "learning_rate": 4.454091284496731e-07,
+      "loss": 0.6327,
+      "step": 727
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6339383101378874,
+      "learning_rate": 4.3353954672705533e-07,
+      "loss": 0.6585,
+      "step": 728
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6510280022721718,
+      "learning_rate": 4.218267581201296e-07,
+      "loss": 0.673,
+      "step": 729
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6668925651315517,
+      "learning_rate": 4.1027095460006715e-07,
+      "loss": 0.6171,
+      "step": 730
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6387870452605278,
+      "learning_rate": 3.988723255650728e-07,
+      "loss": 0.6285,
+      "step": 731
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5570417151964138,
+      "learning_rate": 3.876310578372833e-07,
+      "loss": 0.6512,
+      "step": 732
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6804985201784207,
+      "learning_rate": 3.7654733565969826e-07,
+      "loss": 0.6208,
+      "step": 733
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6007759280722736,
+      "learning_rate": 3.6562134069316857e-07,
+      "loss": 0.6899,
+      "step": 734
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6069630554507705,
+      "learning_rate": 3.548532520134129e-07,
+      "loss": 0.6417,
+      "step": 735
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6020819730356862,
+      "learning_rate": 3.442432461080858e-07,
+      "loss": 0.6345,
+      "step": 736
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7599975053661953,
+      "learning_rate": 3.3379149687388866e-07,
+      "loss": 0.6241,
+      "step": 737
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7005319833124528,
+      "learning_rate": 3.23498175613709e-07,
+      "loss": 0.6116,
+      "step": 738
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6416299446222633,
+      "learning_rate": 3.133634510338235e-07,
+      "loss": 0.6489,
+      "step": 739
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5834401502468652,
+      "learning_rate": 3.0338748924112483e-07,
+      "loss": 0.6843,
+      "step": 740
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6922611041243157,
+      "learning_rate": 2.935704537404083e-07,
+      "loss": 0.6064,
+      "step": 741
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5686374086674281,
+      "learning_rate": 2.839125054316838e-07,
+      "loss": 0.6552,
+      "step": 742
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6062949563577924,
+      "learning_rate": 2.744138026075405e-07,
+      "loss": 0.647,
+      "step": 743
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6711008292078926,
+      "learning_rate": 2.650745009505562e-07,
+      "loss": 0.6079,
+      "step": 744
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6786978494337168,
+      "learning_rate": 2.5589475353073987e-07,
+      "loss": 0.6488,
+      "step": 745
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6081067481820261,
+      "learning_rate": 2.468747108030289e-07,
+      "loss": 0.6511,
+      "step": 746
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5725020341839012,
+      "learning_rate": 2.380145206048201e-07,
+      "loss": 0.6755,
+      "step": 747
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6861814398943483,
+      "learning_rate": 2.2931432815354593e-07,
+      "loss": 0.6373,
+      "step": 748
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6170554804449965,
+      "learning_rate": 2.2077427604429435e-07,
+      "loss": 0.6493,
+      "step": 749
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6825177064680187,
+      "learning_rate": 2.123945042474751e-07,
+      "loss": 0.6381,
+      "step": 750
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6226675880293764,
+      "learning_rate": 2.0417515010652032e-07,
+      "loss": 0.645,
+      "step": 751
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8098209287494129,
+      "learning_rate": 1.9611634833564096e-07,
+      "loss": 0.6268,
+      "step": 752
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5771913591950079,
+      "learning_rate": 1.8821823101760949e-07,
+      "loss": 0.6437,
+      "step": 753
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6693386449117495,
+      "learning_rate": 1.8048092760160286e-07,
+      "loss": 0.626,
+      "step": 754
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6563341557663969,
+      "learning_rate": 1.7290456490107522e-07,
+      "loss": 0.6453,
+      "step": 755
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6415636562126644,
+      "learning_rate": 1.6548926709168634e-07,
+      "loss": 0.6428,
+      "step": 756
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8492239870521986,
+      "learning_rate": 1.5823515570925763e-07,
+      "loss": 0.6353,
+      "step": 757
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.6107662239821435,
+      "learning_rate": 1.5114234964778707e-07,
+      "loss": 0.6306,
+      "step": 758
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.6807822568754512,
+      "learning_rate": 1.4421096515749855e-07,
+      "loss": 0.7021,
+      "step": 759
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5597939216807377,
+      "learning_rate": 1.374411158429323e-07,
+      "loss": 0.6446,
+      "step": 760
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7545145990379868,
+      "learning_rate": 1.30832912661093e-07,
+      "loss": 0.5991,
+      "step": 761
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.6443720022531878,
+      "learning_rate": 1.243864639196213e-07,
+      "loss": 0.6401,
+      "step": 762
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.6735630161306276,
+      "learning_rate": 1.1810187527502182e-07,
+      "loss": 0.6243,
+      "step": 763
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5527995931889649,
+      "learning_rate": 1.1197924973093466e-07,
+      "loss": 0.656,
+      "step": 764
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6954231846383898,
+      "learning_rate": 1.0601868763643997e-07,
+      "loss": 0.6995,
+      "step": 765
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5794149258910508,
+      "learning_rate": 1.0022028668442374e-07,
+      "loss": 0.6525,
+      "step": 766
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6763883420071272,
+      "learning_rate": 9.45841419099669e-08,
+      "loss": 0.6134,
+      "step": 767
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6163957803822115,
+      "learning_rate": 8.911034568879207e-08,
+      "loss": 0.6463,
+      "step": 768
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6872128172184772,
+      "learning_rate": 8.379898773574924e-08,
+      "loss": 0.6272,
+      "step": 769
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6969807884085863,
+      "learning_rate": 7.865015510334473e-08,
+      "loss": 0.6154,
+      "step": 770
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6067392208771799,
+      "learning_rate": 7.366393218031564e-08,
+      "loss": 0.6548,
+      "step": 771
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7326568974910563,
+      "learning_rate": 6.884040069024434e-08,
+      "loss": 0.6139,
+      "step": 772
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6271685352214629,
+      "learning_rate": 6.417963969022389e-08,
+      "loss": 0.6984,
+      "step": 773
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7231828207260607,
+      "learning_rate": 5.968172556955365e-08,
+      "loss": 0.6098,
+      "step": 774
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6886787137949122,
+      "learning_rate": 5.534673204849572e-08,
+      "loss": 0.6389,
+      "step": 775
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6968668645848208,
+      "learning_rate": 5.1174730177064866e-08,
+      "loss": 0.6127,
+      "step": 776
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5940420732274541,
+      "learning_rate": 4.716578833386054e-08,
+      "loss": 0.6101,
+      "step": 777
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5916204706340398,
+      "learning_rate": 4.331997222494777e-08,
+      "loss": 0.6435,
+      "step": 778
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6596010087216369,
+      "learning_rate": 3.963734488278248e-08,
+      "loss": 0.7019,
+      "step": 779
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.689525877668857,
+      "learning_rate": 3.6117966665175644e-08,
+      "loss": 0.6174,
+      "step": 780
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7061654652231126,
+      "learning_rate": 3.2761895254306285e-08,
+      "loss": 0.6326,
+      "step": 781
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5320778827729019,
+      "learning_rate": 2.9569185655773382e-08,
+      "loss": 0.6332,
+      "step": 782
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6484502555965168,
+      "learning_rate": 2.6539890197695428e-08,
+      "loss": 0.6075,
+      "step": 783
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5932090460175373,
+      "learning_rate": 2.3674058529855603e-08,
+      "loss": 0.7004,
+      "step": 784
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6351269850447796,
+      "learning_rate": 2.0971737622883515e-08,
+      "loss": 0.6203,
+      "step": 785
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6469454549653973,
+      "learning_rate": 1.8432971767488038e-08,
+      "loss": 0.6311,
+      "step": 786
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.730675850136464,
+      "learning_rate": 1.605780257373124e-08,
+      "loss": 0.6379,
+      "step": 787
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8638624065133594,
+      "learning_rate": 1.3846268970344467e-08,
+      "loss": 0.6049,
+      "step": 788
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6053294417590919,
+      "learning_rate": 1.179840720409331e-08,
+      "loss": 0.6473,
+      "step": 789
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6000098616667142,
+      "learning_rate": 9.914250839180296e-09,
+      "loss": 0.6451,
+      "step": 790
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7050234381418286,
+      "learning_rate": 8.193830756699773e-09,
+      "loss": 0.6587,
+      "step": 791
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5633463982321911,
+      "learning_rate": 6.6371751541249865e-09,
+      "loss": 0.6258,
+      "step": 792
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5930904755005525,
+      "learning_rate": 5.2443095448506674e-09,
+      "loss": 0.6397,
+      "step": 793
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6519111229301998,
+      "learning_rate": 4.015256757774477e-09,
+      "loss": 0.6157,
+      "step": 794
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6669790713557369,
+      "learning_rate": 2.9500369369195313e-09,
+      "loss": 0.6331,
+      "step": 795
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6697417881781613,
+      "learning_rate": 2.0486675411102165e-09,
+      "loss": 0.6671,
+      "step": 796
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6523758350111415,
+      "learning_rate": 1.3111633436779792e-09,
+      "loss": 0.6909,
+      "step": 797
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6926259144477184,
+      "learning_rate": 7.375364322292911e-10,
+      "loss": 0.6272,
+      "step": 798
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.542270719932054,
+      "learning_rate": 3.277962084369257e-10,
+      "loss": 0.662,
+      "step": 799
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6419827339376735,
+      "learning_rate": 8.19493878945199e-11,
+      "loss": 0.5797,
+      "step": 800
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.637833002177059,
+      "learning_rate": 0.0,
+      "loss": 0.6405,
+      "step": 801
+    },
+    {
+      "epoch": 1.0,
+      "step": 801,
+      "total_flos": 0.0,
+      "train_loss": 0.032679110430599595,
+      "train_runtime": 10037.2401,
+      "train_samples_per_second": 164.437,
+      "train_steps_per_second": 0.08
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 801,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}