diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9235 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 100000,
+  "global_step": 1315,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.573443006537114,
+      "learning_rate": 5e-06,
+      "loss": 1.9387,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.897721646038031,
+      "learning_rate": 1e-05,
+      "loss": 2.0156,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6833924792734236,
+      "learning_rate": 1.5e-05,
+      "loss": 2.0546,
+      "step": 3
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.468088694638523,
+      "learning_rate": 2e-05,
+      "loss": 2.0645,
+      "step": 4
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.406008119826161,
+      "learning_rate": 2.5e-05,
+      "loss": 2.0242,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.312818519185682,
+      "learning_rate": 3e-05,
+      "loss": 1.9849,
+      "step": 6
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 30.436909570000097,
+      "learning_rate": 3.5e-05,
+      "loss": 2.1739,
+      "step": 7
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 12.523913728413042,
+      "learning_rate": 4e-05,
+      "loss": 2.2418,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 15.2606455120682,
+      "learning_rate": 4.5e-05,
+      "loss": 2.3105,
+      "step": 9
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.643041426875583,
+      "learning_rate": 5e-05,
+      "loss": 2.2914,
+      "step": 10
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 8.397314715185384,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 2.1375,
+      "step": 11
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 9.57043786725823,
+      "learning_rate": 6e-05,
+      "loss": 2.0965,
+      "step": 12
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.42438448291741,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 2.2145,
+      "step": 13
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 12.175282793363975,
+      "learning_rate": 7e-05,
+      "loss": 2.101,
+      "step": 14
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 14.717379244980794,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 2.1007,
+      "step": 15
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 8.933052336812903,
+      "learning_rate": 8e-05,
+      "loss": 2.1711,
+      "step": 16
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 10.74175580631161,
+      "learning_rate": 8.5e-05,
+      "loss": 2.1341,
+      "step": 17
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 7.55818119352462,
+      "learning_rate": 9e-05,
+      "loss": 2.1282,
+      "step": 18
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.92381389753304,
+      "learning_rate": 9.5e-05,
+      "loss": 2.147,
+      "step": 19
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.517398583520854,
+      "learning_rate": 0.0001,
+      "loss": 1.9388,
+      "step": 20
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.940433361030487,
+      "learning_rate": 0.000105,
+      "loss": 1.926,
+      "step": 21
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.27198011754989,
+      "learning_rate": 0.00011000000000000002,
+      "loss": 1.8269,
+      "step": 22
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.678665597854372,
+      "learning_rate": 0.00011499999999999999,
+      "loss": 1.8219,
+      "step": 23
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.26781501215367,
+      "learning_rate": 0.00012,
+      "loss": 1.8097,
+      "step": 24
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.9136661974692695,
+      "learning_rate": 0.000125,
+      "loss": 1.8466,
+      "step": 25
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.207740676409568,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 1.7851,
+      "step": 26
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 11.853977462764831,
+      "learning_rate": 0.00013500000000000003,
+      "loss": 1.9357,
+      "step": 27
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 16.965081796682217,
+      "learning_rate": 0.00014,
+      "loss": 1.8313,
+      "step": 28
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 20.772549016630467,
+      "learning_rate": 0.000145,
+      "loss": 2.0554,
+      "step": 29
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 13.141880936993415,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 2.0323,
+      "step": 30
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 7.52536019151535,
+      "learning_rate": 0.000155,
+      "loss": 1.9211,
+      "step": 31
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.818360373749082,
+      "learning_rate": 0.00016,
+      "loss": 1.8557,
+      "step": 32
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 10.06456231077085,
+      "learning_rate": 0.000165,
+      "loss": 1.8338,
+      "step": 33
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 10.064567501603403,
+      "learning_rate": 0.00017,
+      "loss": 1.888,
+      "step": 34
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.888601202002441,
+      "learning_rate": 0.000175,
+      "loss": 1.652,
+      "step": 35
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 7.246119453405153,
+      "learning_rate": 0.00018,
+      "loss": 1.6953,
+      "step": 36
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.920004161375261,
+      "learning_rate": 0.00018500000000000002,
+      "loss": 1.6515,
+      "step": 37
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.428107781433948,
+      "learning_rate": 0.00019,
+      "loss": 1.6564,
+      "step": 38
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.512979341247617,
+      "learning_rate": 0.000195,
+      "loss": 1.5937,
+      "step": 39
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.335692865326675,
+      "learning_rate": 0.0002,
+      "loss": 1.5949,
+      "step": 40
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.281191433350451,
+      "learning_rate": 0.00019999969643677332,
+      "loss": 1.5979,
+      "step": 41
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.405645077013346,
+      "learning_rate": 0.00019999878574893627,
+      "loss": 1.6188,
+      "step": 42
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.21476029175686,
+      "learning_rate": 0.0001999972679420179,
+      "loss": 1.6717,
+      "step": 43
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.845798262400613,
+      "learning_rate": 0.0001999951430252332,
+      "loss": 1.4979,
+      "step": 44
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.239927360704172,
+      "learning_rate": 0.00019999241101148306,
+      "loss": 1.6265,
+      "step": 45
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.358619845268801,
+      "learning_rate": 0.00019998907191735434,
+      "loss": 1.4591,
+      "step": 46
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.8267490249600646,
+      "learning_rate": 0.00019998512576311953,
+      "loss": 1.4208,
+      "step": 47
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.147399221041491,
+      "learning_rate": 0.00019998057257273675,
+      "loss": 1.4566,
+      "step": 48
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.338625677579742,
+      "learning_rate": 0.00019997541237384966,
+      "loss": 1.4881,
+      "step": 49
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.7187095457569357,
+      "learning_rate": 0.0001999696451977872,
+      "loss": 1.4533,
+      "step": 50
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.9607219261179027,
+      "learning_rate": 0.00019996327107956333,
+      "loss": 1.2318,
+      "step": 51
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.0362900916282465,
+      "learning_rate": 0.00019995629005787713,
+      "loss": 1.3823,
+      "step": 52
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.7969234114178305,
+      "learning_rate": 0.00019994870217511217,
+      "loss": 1.446,
+      "step": 53
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.7080254422467696,
+      "learning_rate": 0.0001999405074773365,
+      "loss": 1.4316,
+      "step": 54
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.063585122695151,
+      "learning_rate": 0.0001999317060143023,
+      "loss": 1.3645,
+      "step": 55
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.7402372325358417,
+      "learning_rate": 0.00019992229783944557,
+      "loss": 1.2407,
+      "step": 56
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.8207270537749243,
+      "learning_rate": 0.00019991228300988585,
+      "loss": 1.205,
+      "step": 57
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.7094628455808234,
+      "learning_rate": 0.0001999016615864258,
+      "loss": 1.2771,
+      "step": 58
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.3113813659466667,
+      "learning_rate": 0.0001998904336335509,
+      "loss": 1.1836,
+      "step": 59
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.5196053817850674,
+      "learning_rate": 0.00019987859921942903,
+      "loss": 1.1159,
+      "step": 60
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.5709450941788736,
+      "learning_rate": 0.00019986615841591002,
+      "loss": 1.3317,
+      "step": 61
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.463104653992305,
+      "learning_rate": 0.0001998531112985253,
+      "loss": 1.2463,
+      "step": 62
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.320500671538088,
+      "learning_rate": 0.00019983945794648734,
+      "loss": 1.283,
+      "step": 63
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.5510630018459257,
+      "learning_rate": 0.00019982519844268933,
+      "loss": 1.3017,
+      "step": 64
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.987149471239435,
+      "learning_rate": 0.00019981033287370443,
+      "loss": 1.2435,
+      "step": 65
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.325157554670091,
+      "learning_rate": 0.00019979486132978545,
+      "loss": 1.1995,
+      "step": 66
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.126118197654148,
+      "learning_rate": 0.0001997787839048642,
+      "loss": 1.0614,
+      "step": 67
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.9570989160504375,
+      "learning_rate": 0.00019976210069655104,
+      "loss": 1.0633,
+      "step": 68
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.915825313814494,
+      "learning_rate": 0.0001997448118061341,
+      "loss": 1.085,
+      "step": 69
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.225235751109955,
+      "learning_rate": 0.00019972691733857883,
+      "loss": 1.0231,
+      "step": 70
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.7935769859340955,
+      "learning_rate": 0.00019970841740252725,
+      "loss": 0.9455,
+      "step": 71
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.246320561634353,
+      "learning_rate": 0.00019968931211029734,
+      "loss": 1.087,
+      "step": 72
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.3768860350828667,
+      "learning_rate": 0.00019966960157788248,
+      "loss": 1.0763,
+      "step": 73
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.1859762187782703,
+      "learning_rate": 0.00019964928592495045,
+      "loss": 0.9622,
+      "step": 74
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.1292454634488265,
+      "learning_rate": 0.00019962836527484296,
+      "loss": 1.0875,
+      "step": 75
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.21054438460834,
+      "learning_rate": 0.0001996068397545748,
+      "loss": 1.0167,
+      "step": 76
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.9203885888759236,
+      "learning_rate": 0.00019958470949483318,
+      "loss": 0.915,
+      "step": 77
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.6991729975190246,
+      "learning_rate": 0.00019956197462997667,
+      "loss": 0.9995,
+      "step": 78
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.882625643559257,
+      "learning_rate": 0.00019953863529803466,
+      "loss": 0.8967,
+      "step": 79
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.32109766820597,
+      "learning_rate": 0.00019951469164070646,
+      "loss": 0.9709,
+      "step": 80
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9448477760401173,
+      "learning_rate": 0.00019949014380336028,
+      "loss": 0.836,
+      "step": 81
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.140016495650312,
+      "learning_rate": 0.00019946499193503262,
+      "loss": 0.8402,
+      "step": 82
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.3882833576921483,
+      "learning_rate": 0.000199439236188427,
+      "loss": 1.0349,
+      "step": 83
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.567878310709308,
+      "learning_rate": 0.0001994128767199135,
+      "loss": 0.939,
+      "step": 84
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0240169932097123,
+      "learning_rate": 0.0001993859136895274,
+      "loss": 0.8708,
+      "step": 85
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.038200660930203,
+      "learning_rate": 0.0001993583472609683,
+      "loss": 0.811,
+      "step": 86
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.7834481586541524,
+      "learning_rate": 0.00019933017760159937,
+      "loss": 0.8912,
+      "step": 87
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.832443917561836,
+      "learning_rate": 0.00019930140488244602,
+      "loss": 0.8894,
+      "step": 88
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6261617642028856,
+      "learning_rate": 0.0001992720292781951,
+      "loss": 0.798,
+      "step": 89
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.5554599480024014,
+      "learning_rate": 0.0001992420509671936,
+      "loss": 0.8266,
+      "step": 90
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.0427567116752514,
+      "learning_rate": 0.0001992114701314478,
+      "loss": 1.0053,
+      "step": 91
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8997664630924755,
+      "learning_rate": 0.00019918028695662207,
+      "loss": 0.7694,
+      "step": 92
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7578258142609196,
+      "learning_rate": 0.00019914850163203768,
+      "loss": 0.7539,
+      "step": 93
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.701930931031041,
+      "learning_rate": 0.00019911611435067172,
+      "loss": 0.7927,
+      "step": 94
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.846507608699195,
+      "learning_rate": 0.00019908312530915603,
+      "loss": 0.832,
+      "step": 95
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.074392118739785,
+      "learning_rate": 0.00019904953470777575,
+      "loss": 0.7953,
+      "step": 96
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.96006198183252,
+      "learning_rate": 0.0001990153427504683,
+      "loss": 0.7536,
+      "step": 97
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.8230073271013865,
+      "learning_rate": 0.00019898054964482214,
+      "loss": 0.7038,
+      "step": 98
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.800682109344343,
+      "learning_rate": 0.00019894515560207537,
+      "loss": 0.8174,
+      "step": 99
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.688544083040805,
+      "learning_rate": 0.0001989091608371146,
+      "loss": 0.7013,
+      "step": 100
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.6859086746646312,
+      "learning_rate": 0.0001988725655684736,
+      "loss": 0.6977,
+      "step": 101
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.755085287594568,
+      "learning_rate": 0.00019883537001833188,
+      "loss": 0.7442,
+      "step": 102
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.78513624349955,
+      "learning_rate": 0.0001987975744125135,
+      "loss": 0.827,
+      "step": 103
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.7853358198811975,
+      "learning_rate": 0.00019875917898048558,
+      "loss": 0.6047,
+      "step": 104
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.588991832718384,
+      "learning_rate": 0.0001987201839553569,
+      "loss": 0.7334,
+      "step": 105
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6967663216490827,
+      "learning_rate": 0.00019868058957387663,
+      "loss": 0.7101,
+      "step": 106
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.61951928384871,
+      "learning_rate": 0.00019864039607643273,
+      "loss": 0.634,
+      "step": 107
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.5557457011108884,
+      "learning_rate": 0.0001985996037070505,
+      "loss": 0.6746,
+      "step": 108
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.777194501324582,
+      "learning_rate": 0.00019855821271339125,
+      "loss": 0.7021,
+      "step": 109
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.693078451741438,
+      "learning_rate": 0.00019851622334675066,
+      "loss": 0.6598,
+      "step": 110
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8037567214655654,
+      "learning_rate": 0.00019847363586205727,
+      "loss": 0.6819,
+      "step": 111
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.406011250652929,
+      "learning_rate": 0.00019843045051787096,
+      "loss": 0.7199,
+      "step": 112
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.244855646918106,
+      "learning_rate": 0.00019838666757638135,
+      "loss": 0.6318,
+      "step": 113
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2926770122966893,
+      "learning_rate": 0.0001983422873034063,
+      "loss": 0.5791,
+      "step": 114
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.0672980925025386,
+      "learning_rate": 0.0001982973099683902,
+      "loss": 0.7036,
+      "step": 115
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.562291226953939,
+      "learning_rate": 0.00019825173584440232,
+      "loss": 0.5725,
+      "step": 116
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.491767406932237,
+      "learning_rate": 0.0001982055652081352,
+      "loss": 0.5714,
+      "step": 117
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4791900034448577,
+      "learning_rate": 0.00019815879833990304,
+      "loss": 0.591,
+      "step": 118
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.239738976629645,
+      "learning_rate": 0.00019811143552363983,
+      "loss": 0.5859,
+      "step": 119
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.104415969417008,
+      "learning_rate": 0.00019806347704689778,
+      "loss": 0.5879,
+      "step": 120
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3716966512186946,
+      "learning_rate": 0.00019801492320084546,
+      "loss": 0.6932,
+      "step": 121
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.036984019743635,
+      "learning_rate": 0.00019796577428026616,
+      "loss": 0.7244,
+      "step": 122
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3063256794057327,
+      "learning_rate": 0.00019791603058355595,
+      "loss": 0.5494,
+      "step": 123
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.762955758826247,
+      "learning_rate": 0.00019786569241272197,
+      "loss": 0.6619,
+      "step": 124
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.454205809449977,
+      "learning_rate": 0.00019781476007338058,
+      "loss": 0.6419,
+      "step": 125
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.399439371027546,
+      "learning_rate": 0.00019776323387475547,
+      "loss": 0.5749,
+      "step": 126
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.403739746505696,
+      "learning_rate": 0.00019771111412967583,
+      "loss": 0.6307,
+      "step": 127
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4509378321493114,
+      "learning_rate": 0.0001976584011545744,
+      "loss": 0.6253,
+      "step": 128
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4053656637435505,
+      "learning_rate": 0.00019760509526948566,
+      "loss": 0.589,
+      "step": 129
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.647225523908514,
+      "learning_rate": 0.00019755119679804367,
+      "loss": 0.6761,
+      "step": 130
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3368812810604775,
+      "learning_rate": 0.00019749670606748033,
+      "loss": 0.5658,
+      "step": 131
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2854986239979476,
+      "learning_rate": 0.0001974416234086233,
+      "loss": 0.5664,
+      "step": 132
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.286850783114172,
+      "learning_rate": 0.00019738594915589397,
+      "loss": 0.5727,
+      "step": 133
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.2951572163622855,
+      "learning_rate": 0.00019732968364730545,
+      "loss": 0.6668,
+      "step": 134
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.1662078751465077,
+      "learning_rate": 0.00019727282722446047,
+      "loss": 0.4999,
+      "step": 135
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.15554212745671,
+      "learning_rate": 0.0001972153802325495,
+      "loss": 0.5444,
+      "step": 136
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.100795429925952,
+      "learning_rate": 0.0001971573430203484,
+      "loss": 0.5202,
+      "step": 137
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.183280156295744,
+      "learning_rate": 0.00019709871594021642,
+      "loss": 0.57,
+      "step": 138
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.298967977449057,
+      "learning_rate": 0.00019703949934809408,
+      "loss": 0.5468,
+      "step": 139
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.296062633171164,
+      "learning_rate": 0.00019697969360350098,
+      "loss": 0.5473,
+      "step": 140
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.4744569442457474,
+      "learning_rate": 0.00019691929906953356,
+      "loss": 0.5548,
+      "step": 141
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1844460699726653,
+      "learning_rate": 0.0001968583161128631,
+      "loss": 0.5618,
+      "step": 142
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.4679672865517612,
+      "learning_rate": 0.00019679674510373325,
+      "loss": 0.5642,
+      "step": 143
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2204656868236317,
+      "learning_rate": 0.00019673458641595784,
+      "loss": 0.5073,
+      "step": 144
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1450429498725727,
+      "learning_rate": 0.00019667184042691875,
+      "loss": 0.5578,
+      "step": 145
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1813702731700464,
+      "learning_rate": 0.00019660850751756348,
+      "loss": 0.4392,
+      "step": 146
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4335842086457995,
+      "learning_rate": 0.00019654458807240283,
+      "loss": 0.634,
+      "step": 147
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0571985859980013,
+      "learning_rate": 0.0001964800824795087,
+      "loss": 0.5107,
+      "step": 148
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3947861455659716,
+      "learning_rate": 0.00019641499113051157,
+      "loss": 0.5923,
+      "step": 149
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0627153272464143,
+      "learning_rate": 0.00019634931442059832,
+      "loss": 0.4836,
+      "step": 150
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.310761873008071,
+      "learning_rate": 0.00019628305274850956,
+      "loss": 0.6365,
+      "step": 151
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2345616651970714,
+      "learning_rate": 0.00019621620651653744,
+      "loss": 0.5477,
+      "step": 152
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2484276550142384,
+      "learning_rate": 0.00019614877613052312,
+      "loss": 0.552,
+      "step": 153
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.114200465281353,
+      "learning_rate": 0.00019608076199985433,
+      "loss": 0.4484,
+      "step": 154
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3246486222209275,
+      "learning_rate": 0.00019601216453746283,
+      "loss": 0.6015,
+      "step": 155
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.133410973736004,
+      "learning_rate": 0.00019594298415982194,
+      "loss": 0.4857,
+      "step": 156
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0782984494483534,
+      "learning_rate": 0.0001958732212869441,
+      "loss": 0.4369,
+      "step": 157
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.427115943165033,
+      "learning_rate": 0.00019580287634237808,
+      "loss": 0.4362,
+      "step": 158
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0650253968872208,
+      "learning_rate": 0.00019573194975320673,
+      "loss": 0.396,
+      "step": 159
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7456009362916207,
+      "learning_rate": 0.0001956604419500441,
+      "loss": 0.6388,
+      "step": 160
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1579314898092243,
+      "learning_rate": 0.00019558835336703294,
+      "loss": 0.4831,
+      "step": 161
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2010068941611722,
+      "learning_rate": 0.00019551568444184215,
+      "loss": 0.4436,
+      "step": 162
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2411037681238386,
+      "learning_rate": 0.00019544243561566403,
+      "loss": 0.5172,
+      "step": 163
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9827146652184773,
+      "learning_rate": 0.00019536860733321152,
+      "loss": 0.4691,
+      "step": 164
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.986046670798816,
+      "learning_rate": 0.00019529420004271567,
+      "loss": 0.3806,
+      "step": 165
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.207670827624158,
+      "learning_rate": 0.00019521921419592283,
+      "loss": 0.4448,
+      "step": 166
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.040401349419082,
+      "learning_rate": 0.0001951436502480919,
+      "loss": 0.4723,
+      "step": 167
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2497924420864415,
+      "learning_rate": 0.00019506750865799162,
+      "loss": 0.4816,
+      "step": 168
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2615716784265816,
+      "learning_rate": 0.0001949907898878977,
+      "loss": 0.484,
+      "step": 169
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0673566712072318,
+      "learning_rate": 0.00019491349440359015,
+      "loss": 0.4573,
+      "step": 170
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.872899547407347,
+      "learning_rate": 0.00019483562267435018,
+      "loss": 0.3883,
+      "step": 171
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9564604457400563,
+      "learning_rate": 0.00019475717517295778,
+      "loss": 0.4273,
+      "step": 172
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2645017346910663,
+      "learning_rate": 0.00019467815237568842,
+      "loss": 0.3864,
+      "step": 173
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.123298251563089,
+      "learning_rate": 0.00019459855476231043,
+      "loss": 0.4613,
+      "step": 174
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.7460621885886454,
+      "learning_rate": 0.00019451838281608197,
+      "loss": 0.3458,
+      "step": 175
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.7500541851050797,
+      "learning_rate": 0.00019443763702374812,
+      "loss": 0.3574,
+      "step": 176
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.003069201799193,
+      "learning_rate": 0.00019435631787553795,
+      "loss": 0.3795,
+      "step": 177
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1436562781275876,
+      "learning_rate": 0.00019427442586516155,
+      "loss": 0.424,
+      "step": 178
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8825800725517832,
+      "learning_rate": 0.00019419196148980693,
+      "loss": 0.4252,
+      "step": 179
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.082574210393827,
+      "learning_rate": 0.0001941089252501372,
+      "loss": 0.4048,
+      "step": 180
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1287353040149557,
+      "learning_rate": 0.00019402531765028722,
+      "loss": 0.4976,
+      "step": 181
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8406972894616151,
+      "learning_rate": 0.00019394113919786094,
+      "loss": 0.3865,
+      "step": 182
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.6692876537227375,
+      "learning_rate": 0.00019385639040392803,
+      "loss": 0.3458,
+      "step": 183
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2642453588760043,
+      "learning_rate": 0.00019377107178302074,
+      "loss": 0.4659,
+      "step": 184
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.8777392881522392,
+      "learning_rate": 0.00019368518385313107,
+      "loss": 0.4092,
+      "step": 185
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9664784548256118,
+      "learning_rate": 0.00019359872713570732,
+      "loss": 0.4032,
+      "step": 186
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.157056753435417,
+      "learning_rate": 0.00019351170215565114,
+      "loss": 0.4422,
+      "step": 187
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.079509251222586,
+      "learning_rate": 0.00019342410944131415,
+      "loss": 0.3961,
+      "step": 188
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8751660854571803,
+      "learning_rate": 0.00019333594952449488,
+      "loss": 0.3586,
+      "step": 189
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.871774405314273,
+      "learning_rate": 0.00019324722294043558,
+      "loss": 0.3636,
+      "step": 190
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.013665074393391,
+      "learning_rate": 0.00019315793022781877,
+      "loss": 0.4558,
+      "step": 191
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8037850367821293,
+      "learning_rate": 0.00019306807192876412,
+      "loss": 0.3672,
+      "step": 192
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8044331924116956,
+      "learning_rate": 0.00019297764858882514,
+      "loss": 0.3496,
+      "step": 193
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.065884631579252,
+      "learning_rate": 0.00019288666075698588,
+      "loss": 0.4294,
+      "step": 194
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.034369512940989,
+      "learning_rate": 0.0001927951089856575,
+      "loss": 0.4782,
+      "step": 195
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0596831007308407,
+      "learning_rate": 0.00019270299383067498,
+      "loss": 0.4376,
+      "step": 196
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.8871813025947615,
+      "learning_rate": 0.00019261031585129386,
+      "loss": 0.3832,
+      "step": 197
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9637422763777594,
+      "learning_rate": 0.0001925170756101867,
+      "loss": 0.369,
+      "step": 198
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.195471186858808,
+      "learning_rate": 0.0001924232736734396,
+      "loss": 0.4647,
+      "step": 199
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.9070157338882112,
+      "learning_rate": 0.00019232891061054895,
+      "loss": 0.4075,
+      "step": 200
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.03759870796533,
+      "learning_rate": 0.00019223398699441785,
+      "loss": 0.4311,
+      "step": 201
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.8912950950862146,
+      "learning_rate": 0.00019213850340135276,
+      "loss": 0.3558,
+      "step": 202
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.9955716140746471,
+      "learning_rate": 0.00019204246041105974,
+      "loss": 0.4405,
+      "step": 203
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.788404674601641,
+      "learning_rate": 0.0001919458586066412,
+      "loss": 0.3861,
+      "step": 204
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.9421281839467572,
+      "learning_rate": 0.00019184869857459232,
+      "loss": 0.4108,
+      "step": 205
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.9775231254783163,
+      "learning_rate": 0.00019175098090479727,
+      "loss": 0.4302,
+      "step": 206
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.723743168807481,
+      "learning_rate": 0.00019165270619052595,
+      "loss": 0.3128,
+      "step": 207
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.7171199625308375,
+      "learning_rate": 0.00019155387502843013,
+      "loss": 0.3149,
+      "step": 208
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.8555278186430686,
+      "learning_rate": 0.00019145448801853989,
+      "loss": 0.3384,
+      "step": 209
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.9668352266339795,
+      "learning_rate": 0.0001913545457642601,
+      "loss": 0.4025,
+      "step": 210
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.5353356294808531,
+      "learning_rate": 0.00019125404887236663,
+      "loss": 0.2486,
+      "step": 211
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.817006922943714,
+      "learning_rate": 0.00019115299795300267,
+      "loss": 0.3376,
+      "step": 212
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.050710002478977,
+      "learning_rate": 0.00019105139361967507,
+      "loss": 0.4536,
+      "step": 213
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.0100046470759474,
+      "learning_rate": 0.00019094923648925067,
+      "loss": 0.3895,
+      "step": 214
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.927936818702207,
+      "learning_rate": 0.00019084652718195238,
+      "loss": 0.3563,
+      "step": 215
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.9001834296749816,
+      "learning_rate": 0.00019074326632135562,
+      "loss": 0.3633,
+      "step": 216
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.9830006097613737,
+      "learning_rate": 0.00019063945453438432,
+      "loss": 0.4182,
+      "step": 217
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.7266888874279112,
+      "learning_rate": 0.0001905350924513074,
+      "loss": 0.2989,
+      "step": 218
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.8960491026139463,
+      "learning_rate": 0.0001904301807057346,
+      "loss": 0.3713,
+      "step": 219
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.7523519084458183,
+      "learning_rate": 0.0001903247199346129,
+      "loss": 0.3446,
+      "step": 220
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.7810889773096152,
+      "learning_rate": 0.00019021871077822255,
+      "loss": 0.3047,
+      "step": 221
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.8702517060188693,
+      "learning_rate": 0.00019011215388017316,
+      "loss": 0.3721,
+      "step": 222
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.832942805257814,
+      "learning_rate": 0.00019000504988739986,
+      "loss": 0.3966,
+      "step": 223
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.9030634365181744,
+      "learning_rate": 0.00018989739945015933,
+      "loss": 0.3685,
+      "step": 224
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.1489366593071484,
+      "learning_rate": 0.00018978920322202582,
+      "loss": 0.4271,
+      "step": 225
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.7756711735029047,
+      "learning_rate": 0.00018968046185988732,
+      "loss": 0.3151,
+      "step": 226
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.7219521476789044,
+      "learning_rate": 0.0001895711760239413,
+      "loss": 0.3406,
+      "step": 227
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.7708485686240514,
+      "learning_rate": 0.00018946134637769105,
+      "loss": 0.3126,
+      "step": 228
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.9821238740081968,
+      "learning_rate": 0.00018935097358794144,
+      "loss": 0.3398,
+      "step": 229
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.898214854275105,
+      "learning_rate": 0.00018924005832479478,
+      "loss": 0.3856,
+      "step": 230
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.910529136341384,
+      "learning_rate": 0.00018912860126164707,
+      "loss": 0.3398,
+      "step": 231
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.610319520690533,
+      "learning_rate": 0.00018901660307518354,
+      "loss": 0.3019,
+      "step": 232
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.5714183763645118,
+      "learning_rate": 0.00018890406444537486,
+      "loss": 0.2889,
+      "step": 233
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.8635695347404497,
+      "learning_rate": 0.0001887909860554728,
+      "loss": 0.3574,
+      "step": 234
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.85065613238587,
+      "learning_rate": 0.0001886773685920062,
+      "loss": 0.4121,
+      "step": 235
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.4768932179673513,
+      "learning_rate": 0.00018856321274477673,
+      "loss": 0.2201,
+      "step": 236
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.5854047199815071,
+      "learning_rate": 0.0001884485192068547,
+      "loss": 0.327,
+      "step": 237
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.753421354277276,
+      "learning_rate": 0.00018833328867457497,
+      "loss": 0.2798,
+      "step": 238
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.8414313680807404,
+      "learning_rate": 0.00018821752184753252,
+      "loss": 0.3317,
+      "step": 239
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.8011554349427403,
+      "learning_rate": 0.00018810121942857845,
+      "loss": 0.3009,
+      "step": 240
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.8352862922832573,
+      "learning_rate": 0.0001879843821238155,
+      "loss": 0.3436,
+      "step": 241
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.5865888388501308,
+      "learning_rate": 0.00018786701064259383,
+      "loss": 0.2498,
+      "step": 242
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.8245533160559124,
+      "learning_rate": 0.00018774910569750673,
+      "loss": 0.3069,
+      "step": 243
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.717462638965888,
+      "learning_rate": 0.00018763066800438636,
+      "loss": 0.2707,
+      "step": 244
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.7475920032079937,
+      "learning_rate": 0.00018751169828229927,
+      "loss": 0.299,
+      "step": 245
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.6185981574561485,
+      "learning_rate": 0.00018739219725354212,
+      "loss": 0.3118,
+      "step": 246
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.6108301489955423,
+      "learning_rate": 0.00018727216564363723,
+      "loss": 0.3018,
+      "step": 247
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.667675984259701,
+      "learning_rate": 0.00018715160418132832,
+      "loss": 0.294,
+      "step": 248
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.6379108608409592,
+      "learning_rate": 0.00018703051359857586,
+      "loss": 0.2273,
+      "step": 249
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.749729874850288,
+      "learning_rate": 0.00018690889463055283,
+      "loss": 0.3203,
+      "step": 250
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.6719571451590929,
+      "learning_rate": 0.0001867867480156402,
+      "loss": 0.2754,
+      "step": 251
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.5568430085763993,
+      "learning_rate": 0.00018666407449542232,
+      "loss": 0.2568,
+      "step": 252
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.9357912672439357,
+      "learning_rate": 0.0001865408748146826,
+      "loss": 0.3533,
+      "step": 253
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.6677836160743433,
+      "learning_rate": 0.0001864171497213989,
+      "loss": 0.3009,
+      "step": 254
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.4806397494675019,
+      "learning_rate": 0.00018629289996673897,
+      "loss": 0.2685,
+      "step": 255
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.3443639482079255,
+      "learning_rate": 0.00018616812630505597,
+      "loss": 0.2146,
+      "step": 256
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.8641957489501426,
+      "learning_rate": 0.0001860428294938838,
+      "loss": 0.4054,
+      "step": 257
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.484638648868682,
+      "learning_rate": 0.00018591701029393255,
+      "loss": 0.2774,
+      "step": 258
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.6970545560127506,
+      "learning_rate": 0.00018579066946908384,
+      "loss": 0.3135,
+      "step": 259
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.7723909332701073,
+      "learning_rate": 0.00018566380778638628,
+      "loss": 0.2624,
+      "step": 260
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.362583291370308,
+      "learning_rate": 0.00018553642601605068,
+      "loss": 0.2358,
+      "step": 261
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.6330213475929671,
+      "learning_rate": 0.00018540852493144545,
+      "loss": 0.3205,
+      "step": 262
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.6361661549458946,
+      "learning_rate": 0.00018528010530909192,
+      "loss": 0.2967,
+      "step": 263
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.2674463487202237,
+      "learning_rate": 0.00018515116792865957,
+      "loss": 0.1865,
+      "step": 264
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.5824522976951254,
+      "learning_rate": 0.00018502171357296144,
+      "loss": 0.1597,
+      "step": 265
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.2028223497326778,
+      "learning_rate": 0.00018489174302794905,
+      "loss": 0.1379,
+      "step": 266
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.3508526606958497,
+      "learning_rate": 0.000184761257082708,
+      "loss": 0.1545,
+      "step": 267
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.5490325110995469,
+      "learning_rate": 0.000184630256529453,
+      "loss": 0.1631,
+      "step": 268
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.1474450566836425,
+      "learning_rate": 0.00018449874216352306,
+      "loss": 0.1185,
+      "step": 269
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.2781114669925107,
+      "learning_rate": 0.00018436671478337666,
+      "loss": 0.1514,
+      "step": 270
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.2810004870165943,
+      "learning_rate": 0.00018423417519058694,
+      "loss": 0.1522,
+      "step": 271
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.3746220478342002,
+      "learning_rate": 0.0001841011241898369,
+      "loss": 0.1563,
+      "step": 272
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.1722419831721185,
+      "learning_rate": 0.0001839675625889143,
+      "loss": 0.1219,
+      "step": 273
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.318793059118971,
+      "learning_rate": 0.00018383349119870695,
+      "loss": 0.1853,
+      "step": 274
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.464801561065941,
+      "learning_rate": 0.00018369891083319778,
+      "loss": 0.1695,
+      "step": 275
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.213635872968982,
+      "learning_rate": 0.00018356382230945976,
+      "loss": 0.149,
+      "step": 276
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.2514902868374735,
+      "learning_rate": 0.00018342822644765104,
+      "loss": 0.1422,
+      "step": 277
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.3264287421045853,
+      "learning_rate": 0.00018329212407100994,
+      "loss": 0.1658,
+      "step": 278
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.3681833233862792,
+      "learning_rate": 0.00018315551600585009,
+      "loss": 0.1583,
+      "step": 279
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.2741885360217686,
+      "learning_rate": 0.00018301840308155507,
+      "loss": 0.1544,
+      "step": 280
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.154701575063112,
+      "learning_rate": 0.0001828807861305738,
+      "loss": 0.112,
+      "step": 281
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.1859724085916519,
+      "learning_rate": 0.00018274266598841517,
+      "loss": 0.1287,
+      "step": 282
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.2001889021464058,
+      "learning_rate": 0.0001826040434936431,
+      "loss": 0.1305,
+      "step": 283
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.1636645037528681,
+      "learning_rate": 0.0001824649194878714,
+      "loss": 0.1091,
+      "step": 284
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.279551381173061,
+      "learning_rate": 0.00018232529481575872,
+      "loss": 0.1488,
+      "step": 285
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.3213439025691391,
+      "learning_rate": 0.00018218517032500344,
+      "loss": 0.1405,
+      "step": 286
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.1545434339904947,
+      "learning_rate": 0.00018204454686633834,
+      "loss": 0.1288,
+      "step": 287
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.1801584415716582,
+      "learning_rate": 0.00018190342529352565,
+      "loss": 0.1349,
+      "step": 288
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.4340039957081578,
+      "learning_rate": 0.0001817618064633518,
+      "loss": 0.1809,
+      "step": 289
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.2084487167858948,
+      "learning_rate": 0.0001816196912356222,
+      "loss": 0.1241,
+      "step": 290
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.1778367464211388,
+      "learning_rate": 0.00018147708047315587,
+      "loss": 0.1176,
+      "step": 291
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.4038456890346582,
+      "learning_rate": 0.00018133397504178057,
+      "loss": 0.1595,
+      "step": 292
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.2282066098715116,
+      "learning_rate": 0.00018119037581032724,
+      "loss": 0.135,
+      "step": 293
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.1491948662557472,
+      "learning_rate": 0.00018104628365062477,
+      "loss": 0.1367,
+      "step": 294
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.1350853852329246,
+      "learning_rate": 0.00018090169943749476,
+      "loss": 0.1201,
+      "step": 295
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.1130893159099848,
+      "learning_rate": 0.00018075662404874626,
+      "loss": 0.1181,
+      "step": 296
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.3501229026875887,
+      "learning_rate": 0.00018061105836517024,
+      "loss": 0.1652,
+      "step": 297
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.0995499735027865,
+      "learning_rate": 0.00018046500327053463,
+      "loss": 0.114,
+      "step": 298
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.3881775922642186,
+      "learning_rate": 0.0001803184596515784,
+      "loss": 0.1705,
+      "step": 299
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.054095947636947,
+      "learning_rate": 0.00018017142839800668,
+      "loss": 0.0961,
+      "step": 300
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.0216970696389942,
+      "learning_rate": 0.0001800239104024851,
+      "loss": 0.1094,
+      "step": 301
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.0412573323551204,
+      "learning_rate": 0.0001798759065606345,
+      "loss": 0.1042,
+      "step": 302
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.265029933085797,
+      "learning_rate": 0.00017972741777102523,
+      "loss": 0.1491,
+      "step": 303
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.1936092690625473,
+      "learning_rate": 0.00017957844493517213,
+      "loss": 0.1411,
+      "step": 304
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.1965288754867318,
+      "learning_rate": 0.0001794289889575286,
+      "loss": 0.1246,
+      "step": 305
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.12658571000799,
+      "learning_rate": 0.0001792790507454815,
+      "loss": 0.1246,
+      "step": 306
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.2685560666194993,
+      "learning_rate": 0.00017912863120934534,
+      "loss": 0.1532,
+      "step": 307
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.2337975635857688,
+      "learning_rate": 0.00017897773126235688,
+      "loss": 0.1182,
+      "step": 308
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.157003238914916,
+      "learning_rate": 0.0001788263518206697,
+      "loss": 0.1163,
+      "step": 309
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.0886799416108037,
+      "learning_rate": 0.00017867449380334834,
+      "loss": 0.1116,
+      "step": 310
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.2201648706122956,
+      "learning_rate": 0.00017852215813236305,
+      "loss": 0.1083,
+      "step": 311
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.3514577133841812,
+      "learning_rate": 0.000178369345732584,
+      "loss": 0.1457,
+      "step": 312
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.0992243013770466,
+      "learning_rate": 0.00017821605753177562,
+      "loss": 0.111,
+      "step": 313
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.9739082414598146,
+      "learning_rate": 0.00017806229446059124,
+      "loss": 0.115,
+      "step": 314
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.1378737903139309,
+      "learning_rate": 0.00017790805745256704,
+      "loss": 0.1069,
+      "step": 315
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.258174163049199,
+      "learning_rate": 0.00017775334744411678,
+      "loss": 0.1253,
+      "step": 316
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.1065825287192146,
+      "learning_rate": 0.00017759816537452574,
+      "loss": 0.1229,
+      "step": 317
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.0340137749344727,
+      "learning_rate": 0.00017744251218594542,
+      "loss": 0.115,
+      "step": 318
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.3144508191041717,
+      "learning_rate": 0.00017728638882338746,
+      "loss": 0.1466,
+      "step": 319
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.312979421872984,
+      "learning_rate": 0.00017712979623471807,
+      "loss": 0.1504,
+      "step": 320
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.3919156770726522,
+      "learning_rate": 0.00017697273537065232,
+      "loss": 0.1699,
+      "step": 321
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.2876324148047005,
+      "learning_rate": 0.00017681520718474823,
+      "loss": 0.1427,
+      "step": 322
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.89431688366937,
+      "learning_rate": 0.00017665721263340113,
+      "loss": 0.0852,
+      "step": 323
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.1410335040332307,
+      "learning_rate": 0.0001764987526758377,
+      "loss": 0.137,
+      "step": 324
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.2225398004015087,
+      "learning_rate": 0.00017633982827411032,
+      "loss": 0.1268,
+      "step": 325
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.3502645939476032,
+      "learning_rate": 0.00017618044039309098,
+      "loss": 0.1686,
+      "step": 326
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.217572179910748,
+      "learning_rate": 0.0001760205900004657,
+      "loss": 0.133,
+      "step": 327
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.0646382700165697,
+      "learning_rate": 0.00017586027806672857,
+      "loss": 0.1249,
+      "step": 328
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.525321068685934,
+      "learning_rate": 0.00017569950556517566,
+      "loss": 0.1938,
+      "step": 329
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.2037571428378924,
+      "learning_rate": 0.00017553827347189938,
+      "loss": 0.1388,
+      "step": 330
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.0639005210904018,
+      "learning_rate": 0.00017537658276578247,
+      "loss": 0.1096,
+      "step": 331
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.270051737424283,
+      "learning_rate": 0.00017521443442849188,
+      "loss": 0.1358,
+      "step": 332
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.172655557910847,
+      "learning_rate": 0.00017505182944447316,
+      "loss": 0.14,
+      "step": 333
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.9666237830295965,
+      "learning_rate": 0.00017488876880094413,
+      "loss": 0.0958,
+      "step": 334
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.010180641447394,
+      "learning_rate": 0.0001747252534878891,
+      "loss": 0.1134,
+      "step": 335
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.1275360299833992,
+      "learning_rate": 0.0001745612844980528,
+      "loss": 0.1267,
+      "step": 336
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.1537718093035536,
+      "learning_rate": 0.00017439686282693436,
+      "loss": 0.1166,
+      "step": 337
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.0975401668917748,
+      "learning_rate": 0.00017423198947278117,
+      "loss": 0.1142,
+      "step": 338
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.049318074845895,
+      "learning_rate": 0.00017406666543658304,
+      "loss": 0.1085,
+      "step": 339
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.0963786009952383,
+      "learning_rate": 0.00017390089172206592,
+      "loss": 0.1188,
+      "step": 340
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.1047878683301895,
+      "learning_rate": 0.00017373466933568588,
+      "loss": 0.1212,
+      "step": 341
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.1236657137656456,
+      "learning_rate": 0.00017356799928662297,
+      "loss": 0.1269,
+      "step": 342
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.1118580861042129,
+      "learning_rate": 0.00017340088258677522,
+      "loss": 0.1374,
+      "step": 343
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.0736382042768697,
+      "learning_rate": 0.00017323332025075223,
+      "loss": 0.1074,
+      "step": 344
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.1426906416177507,
+      "learning_rate": 0.00017306531329586933,
+      "loss": 0.1055,
+      "step": 345
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.213068713099822,
+      "learning_rate": 0.00017289686274214118,
+      "loss": 0.1322,
+      "step": 346
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.1949359428837787,
+      "learning_rate": 0.00017272796961227563,
+      "loss": 0.1219,
+      "step": 347
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.2025317641122422,
+      "learning_rate": 0.00017255863493166756,
+      "loss": 0.1262,
+      "step": 348
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.1368816480634238,
+      "learning_rate": 0.0001723888597283926,
+      "loss": 0.1261,
+      "step": 349
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.1613667950557833,
+      "learning_rate": 0.00017221864503320092,
+      "loss": 0.1433,
+      "step": 350
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.0329044603091726,
+      "learning_rate": 0.00017204799187951105,
+      "loss": 0.1023,
+      "step": 351
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.3245595469562248,
+      "learning_rate": 0.00017187690130340328,
+      "loss": 0.1545,
+      "step": 352
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.203479193942691,
+      "learning_rate": 0.00017170537434361386,
+      "loss": 0.1288,
+      "step": 353
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.9863759258225825,
+      "learning_rate": 0.0001715334120415283,
+      "loss": 0.098,
+      "step": 354
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.1591024768779974,
+      "learning_rate": 0.00017136101544117525,
+      "loss": 0.1216,
+      "step": 355
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.4759832550397851,
+      "learning_rate": 0.00017118818558922003,
+      "loss": 0.1497,
+      "step": 356
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.1241359664722508,
+      "learning_rate": 0.00017101492353495845,
+      "loss": 0.1173,
+      "step": 357
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.2552790869577835,
+      "learning_rate": 0.00017084123033031024,
+      "loss": 0.1284,
+      "step": 358
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.1684885710218835,
+      "learning_rate": 0.0001706671070298128,
+      "loss": 0.1213,
+      "step": 359
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.1466310714307573,
+      "learning_rate": 0.00017049255469061474,
+      "loss": 0.1236,
+      "step": 360
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.16556796629183,
+      "learning_rate": 0.00017031757437246947,
+      "loss": 0.1251,
+      "step": 361
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.1867147077622797,
+      "learning_rate": 0.00017014216713772884,
+      "loss": 0.1143,
+      "step": 362
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.185771561059137,
+      "learning_rate": 0.00016996633405133655,
+      "loss": 0.1229,
+      "step": 363
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.0969793888066444,
+      "learning_rate": 0.00016979007618082175,
+      "loss": 0.1228,
+      "step": 364
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.0805661037287781,
+      "learning_rate": 0.0001696133945962927,
+      "loss": 0.1082,
+      "step": 365
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.134417411097424,
+      "learning_rate": 0.0001694362903704299,
+      "loss": 0.1124,
+      "step": 366
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.2507033744013867,
+      "learning_rate": 0.00016925876457848,
+      "loss": 0.1353,
+      "step": 367
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.181144430276729,
+      "learning_rate": 0.00016908081829824912,
+      "loss": 0.1181,
+      "step": 368
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.284578406692821,
+      "learning_rate": 0.0001689024526100961,
+      "loss": 0.1562,
+      "step": 369
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.3636964562598302,
+      "learning_rate": 0.00016872366859692627,
+      "loss": 0.1364,
+      "step": 370
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.352735117877114,
+      "learning_rate": 0.00016854446734418466,
+      "loss": 0.1499,
+      "step": 371
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.1111973080463173,
+      "learning_rate": 0.0001683648499398495,
+      "loss": 0.1328,
+      "step": 372
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.0299877082401738,
+      "learning_rate": 0.00016818481747442554,
+      "loss": 0.1204,
+      "step": 373
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.151789080552945,
+      "learning_rate": 0.0001680043710409375,
+      "loss": 0.1098,
+      "step": 374
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.3276600739198514,
+      "learning_rate": 0.00016782351173492342,
+      "loss": 0.1823,
+      "step": 375
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.0843113771148096,
+      "learning_rate": 0.00016764224065442796,
+      "loss": 0.1094,
+      "step": 376
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.179487545194111,
+      "learning_rate": 0.0001674605588999959,
+      "loss": 0.1328,
+      "step": 377
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.112577076574857,
+      "learning_rate": 0.0001672784675746651,
+      "loss": 0.1181,
+      "step": 378
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.1518263574635146,
+      "learning_rate": 0.00016709596778396026,
+      "loss": 0.1227,
+      "step": 379
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.3138990247523017,
+      "learning_rate": 0.00016691306063588583,
+      "loss": 0.171,
+      "step": 380
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.128245021866168,
+      "learning_rate": 0.00016672974724091954,
+      "loss": 0.1208,
+      "step": 381
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.172069477792324,
+      "learning_rate": 0.00016654602871200546,
+      "loss": 0.1193,
+      "step": 382
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.1056948230974584,
+      "learning_rate": 0.0001663619061645474,
+      "loss": 0.1279,
+      "step": 383
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.1732592080737412,
+      "learning_rate": 0.00016617738071640208,
+      "loss": 0.1123,
+      "step": 384
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.2218179628317245,
+      "learning_rate": 0.0001659924534878723,
+      "loss": 0.1231,
+      "step": 385
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.2416664251211538,
+      "learning_rate": 0.0001658071256017001,
+      "loss": 0.1468,
+      "step": 386
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.051278225463692,
+      "learning_rate": 0.0001656213981830602,
+      "loss": 0.0988,
+      "step": 387
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.248345297594324,
+      "learning_rate": 0.00016543527235955282,
+      "loss": 0.13,
+      "step": 388
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.1787185698327693,
+      "learning_rate": 0.00016524874926119717,
+      "loss": 0.1286,
+      "step": 389
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.1593153833153118,
+      "learning_rate": 0.0001650618300204242,
+      "loss": 0.1208,
+      "step": 390
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.0249734191002977,
+      "learning_rate": 0.00016487451577207018,
+      "loss": 0.0874,
+      "step": 391
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.1245900841960161,
+      "learning_rate": 0.00016468680765336936,
+      "loss": 0.1239,
+      "step": 392
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.2960395006762668,
+      "learning_rate": 0.00016449870680394747,
+      "loss": 0.1471,
+      "step": 393
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.2579126899367365,
+      "learning_rate": 0.0001643102143658145,
+      "loss": 0.1297,
+      "step": 394
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.0336259475221405,
+      "learning_rate": 0.00016412133148335784,
+      "loss": 0.098,
+      "step": 395
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.0962316724437684,
+      "learning_rate": 0.0001639320593033355,
+      "loss": 0.1116,
+      "step": 396
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.1711925860128423,
+      "learning_rate": 0.000163742398974869,
+      "loss": 0.1198,
+      "step": 397
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.2377519228297558,
+      "learning_rate": 0.00016355235164943626,
+      "loss": 0.1486,
+      "step": 398
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.1213992145878255,
+      "learning_rate": 0.0001633619184808649,
+      "loss": 0.1218,
+      "step": 399
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.3643395185067257,
+      "learning_rate": 0.0001631711006253251,
+      "loss": 0.1364,
+      "step": 400
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.2789762227954533,
+      "learning_rate": 0.00016297989924132252,
+      "loss": 0.1424,
+      "step": 401
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.0613192629084378,
+      "learning_rate": 0.00016278831548969134,
+      "loss": 0.1117,
+      "step": 402
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.1899872719351117,
+      "learning_rate": 0.00016259635053358717,
+      "loss": 0.1213,
+      "step": 403
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.2074253594697786,
+      "learning_rate": 0.00016240400553848007,
+      "loss": 0.14,
+      "step": 404
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.1730989432314962,
+      "learning_rate": 0.0001622112816721474,
+      "loss": 0.1148,
+      "step": 405
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.1355850520210402,
+      "learning_rate": 0.0001620181801046667,
+      "loss": 0.1206,
+      "step": 406
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.1229573029581186,
+      "learning_rate": 0.00016182470200840868,
+      "loss": 0.1133,
+      "step": 407
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.0290790219642958,
+      "learning_rate": 0.00016163084855803006,
+      "loss": 0.1092,
+      "step": 408
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.1077377854374164,
+      "learning_rate": 0.00016143662093046638,
+      "loss": 0.1037,
+      "step": 409
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.173768664042554,
+      "learning_rate": 0.000161242020304925,
+      "loss": 0.1413,
+      "step": 410
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.9856109567170077,
+      "learning_rate": 0.0001610470478628778,
+      "loss": 0.0886,
+      "step": 411
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.3044417561730586,
+      "learning_rate": 0.00016085170478805395,
+      "loss": 0.1271,
+      "step": 412
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.1533657015784773,
+      "learning_rate": 0.00016065599226643303,
+      "loss": 0.1345,
+      "step": 413
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.0224756186960617,
+      "learning_rate": 0.0001604599114862375,
+      "loss": 0.0982,
+      "step": 414
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.0703060311362116,
+      "learning_rate": 0.00016026346363792567,
+      "loss": 0.1147,
+      "step": 415
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.1645710557496303,
+      "learning_rate": 0.00016006664991418434,
+      "loss": 0.1122,
+      "step": 416
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.250459760812092,
+      "learning_rate": 0.00015986947150992172,
+      "loss": 0.1396,
+      "step": 417
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.10298779531962,
+      "learning_rate": 0.0001596719296222601,
+      "loss": 0.1144,
+      "step": 418
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.2258824732511358,
+      "learning_rate": 0.0001594740254505285,
+      "loss": 0.1343,
+      "step": 419
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.2466155659176525,
+      "learning_rate": 0.0001592757601962555,
+      "loss": 0.1108,
+      "step": 420
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.2734326544622663,
+      "learning_rate": 0.00015907713506316192,
+      "loss": 0.1328,
+      "step": 421
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.2002353100905563,
+      "learning_rate": 0.00015887815125715344,
+      "loss": 0.1164,
+      "step": 422
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.0524824697871311,
+      "learning_rate": 0.00015867880998631347,
+      "loss": 0.0927,
+      "step": 423
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.2418587130674024,
+      "learning_rate": 0.0001584791124608955,
+      "loss": 0.129,
+      "step": 424
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.0944433620136427,
+      "learning_rate": 0.0001582790598933161,
+      "loss": 0.1018,
+      "step": 425
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.10661332213394,
+      "learning_rate": 0.00015807865349814733,
+      "loss": 0.1041,
+      "step": 426
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.0834117975231323,
+      "learning_rate": 0.00015787789449210938,
+      "loss": 0.1109,
+      "step": 427
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.2300031734853023,
+      "learning_rate": 0.0001576767840940633,
+      "loss": 0.1022,
+      "step": 428
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.0831394975438198,
+      "learning_rate": 0.00015747532352500357,
+      "loss": 0.1065,
+      "step": 429
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.0813138766521708,
+      "learning_rate": 0.00015727351400805052,
+      "loss": 0.0955,
+      "step": 430
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.0659954453103273,
+      "learning_rate": 0.0001570713567684432,
+      "loss": 0.157,
+      "step": 431
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.1902389357892282,
+      "learning_rate": 0.0001568688530335316,
+      "loss": 0.1232,
+      "step": 432
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.9151560322880303,
+      "learning_rate": 0.0001566660040327695,
+      "loss": 0.0828,
+      "step": 433
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.0383042558328932,
+      "learning_rate": 0.00015646281099770682,
+      "loss": 0.1008,
+      "step": 434
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.9547139373716885,
+      "learning_rate": 0.00015625927516198232,
+      "loss": 0.0831,
+      "step": 435
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.0170470337229764,
+      "learning_rate": 0.0001560553977613158,
+      "loss": 0.1041,
+      "step": 436
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.2610928445236638,
+      "learning_rate": 0.00015585118003350092,
+      "loss": 0.1193,
+      "step": 437
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.260778908043354,
+      "learning_rate": 0.00015564662321839755,
+      "loss": 0.1445,
+      "step": 438
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.3109463484039794,
+      "learning_rate": 0.00015544172855792423,
+      "loss": 0.1314,
+      "step": 439
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.0403866139042997,
+      "learning_rate": 0.0001552364972960506,
+      "loss": 0.0938,
+      "step": 440
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.9832537369636843,
+      "learning_rate": 0.00015503093067878996,
+      "loss": 0.1027,
+      "step": 441
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.03937905126318,
+      "learning_rate": 0.00015482502995419167,
+      "loss": 0.0997,
+      "step": 442
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.0509647801608448,
+      "learning_rate": 0.0001546187963723334,
+      "loss": 0.1041,
+      "step": 443
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.145340634526724,
+      "learning_rate": 0.00015441223118531388,
+      "loss": 0.1191,
+      "step": 444
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.1580189133026946,
+      "learning_rate": 0.00015420533564724495,
+      "loss": 0.1329,
+      "step": 445
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.9480958437964944,
+      "learning_rate": 0.00015399811101424418,
+      "loss": 0.0901,
+      "step": 446
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.1863829610028518,
+      "learning_rate": 0.00015379055854442708,
+      "loss": 0.1351,
+      "step": 447
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.9188746172784557,
+      "learning_rate": 0.00015358267949789966,
+      "loss": 0.0943,
+      "step": 448
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.026039753668433,
+      "learning_rate": 0.0001533744751367506,
+      "loss": 0.1068,
+      "step": 449
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.0688305797232243,
+      "learning_rate": 0.0001531659467250436,
+      "loss": 0.1005,
+      "step": 450
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.0987439506265773,
+      "learning_rate": 0.0001529570955288099,
+      "loss": 0.1163,
+      "step": 451
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.1086460712850783,
+      "learning_rate": 0.00015274792281604028,
+      "loss": 0.0941,
+      "step": 452
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.1444827639510378,
+      "learning_rate": 0.00015253842985667762,
+      "loss": 0.1146,
+      "step": 453
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.1422862703126007,
+      "learning_rate": 0.0001523286179226091,
+      "loss": 0.1244,
+      "step": 454
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.0302331489441574,
+      "learning_rate": 0.0001521184882876585,
+      "loss": 0.0898,
+      "step": 455
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.2365591624658727,
+      "learning_rate": 0.00015190804222757845,
+      "loss": 0.1226,
+      "step": 456
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.129977628641513,
+      "learning_rate": 0.00015169728102004256,
+      "loss": 0.1077,
+      "step": 457
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.9607184856381425,
+      "learning_rate": 0.00015148620594463794,
+      "loss": 0.086,
+      "step": 458
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.217876117969091,
+      "learning_rate": 0.00015127481828285718,
+      "loss": 0.1246,
+      "step": 459
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.0697772676497657,
+      "learning_rate": 0.0001510631193180907,
+      "loss": 0.0898,
+      "step": 460
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.9861575897486603,
+      "learning_rate": 0.00015085111033561895,
+      "loss": 0.0781,
+      "step": 461
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.0879037831005531,
+      "learning_rate": 0.00015063879262260446,
+      "loss": 0.1012,
+      "step": 462
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.109366800553283,
+      "learning_rate": 0.00015042616746808435,
+      "loss": 0.1051,
+      "step": 463
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.2591651189421935,
+      "learning_rate": 0.00015021323616296213,
+      "loss": 0.127,
+      "step": 464
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.0760753249908668,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.1019,
+      "step": 465
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.9280169398003503,
+      "learning_rate": 0.00014978646027381123,
+      "loss": 0.0874,
+      "step": 466
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.9005299186843474,
+      "learning_rate": 0.00014957261828085191,
+      "loss": 0.0754,
+      "step": 467
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.9937813944513447,
+      "learning_rate": 0.0001493584753194134,
+      "loss": 0.0901,
+      "step": 468
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.0735895453397655,
+      "learning_rate": 0.00014914403268961426,
+      "loss": 0.0873,
+      "step": 469
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.1030985841715417,
+      "learning_rate": 0.00014892929169339235,
+      "loss": 0.1072,
+      "step": 470
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.030478622809127,
+      "learning_rate": 0.00014871425363449718,
+      "loss": 0.1002,
+      "step": 471
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.0663000622043026,
+      "learning_rate": 0.00014849891981848158,
+      "loss": 0.1126,
+      "step": 472
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.8962296554379596,
+      "learning_rate": 0.0001482832915526942,
+      "loss": 0.0759,
+      "step": 473
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.0066612977416853,
+      "learning_rate": 0.00014806737014627124,
+      "loss": 0.1071,
+      "step": 474
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.9387323811602087,
+      "learning_rate": 0.00014785115691012864,
+      "loss": 0.0845,
+      "step": 475
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.1919524131091321,
+      "learning_rate": 0.00014763465315695425,
+      "loss": 0.1236,
+      "step": 476
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.979838586951276,
+      "learning_rate": 0.00014741786020119955,
+      "loss": 0.0957,
+      "step": 477
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.245695783405674,
+      "learning_rate": 0.00014720077935907196,
+      "loss": 0.1492,
+      "step": 478
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.052120827921073,
+      "learning_rate": 0.00014698341194852664,
+      "loss": 0.0848,
+      "step": 479
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.1123400310799008,
+      "learning_rate": 0.00014676575928925867,
+      "loss": 0.1174,
+      "step": 480
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.9815137018507386,
+      "learning_rate": 0.00014654782270269497,
+      "loss": 0.0934,
+      "step": 481
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.0236793306971954,
+      "learning_rate": 0.00014632960351198618,
+      "loss": 0.0968,
+      "step": 482
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.8569472766935837,
+      "learning_rate": 0.00014611110304199872,
+      "loss": 0.0872,
+      "step": 483
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.0757911454552826,
+      "learning_rate": 0.00014589232261930674,
+      "loss": 0.1105,
+      "step": 484
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.0394925161188941,
+      "learning_rate": 0.00014567326357218407,
+      "loss": 0.1021,
+      "step": 485
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.0649292412578595,
+      "learning_rate": 0.00014545392723059616,
+      "loss": 0.1121,
+      "step": 486
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.101539460661583,
+      "learning_rate": 0.0001452343149261919,
+      "loss": 0.1175,
+      "step": 487
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.0358871929257938,
+      "learning_rate": 0.00014501442799229572,
+      "loss": 0.0888,
+      "step": 488
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.7890107874401794,
+      "learning_rate": 0.00014479426776389936,
+      "loss": 0.0561,
+      "step": 489
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.0147429482810784,
+      "learning_rate": 0.00014457383557765386,
+      "loss": 0.0986,
+      "step": 490
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.1168392985496396,
+      "learning_rate": 0.00014435313277186125,
+      "loss": 0.1218,
+      "step": 491
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.9765951771037416,
+      "learning_rate": 0.00014413216068646668,
+      "loss": 0.0872,
+      "step": 492
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.1324426001862569,
+      "learning_rate": 0.0001439109206630501,
+      "loss": 0.1295,
+      "step": 493
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.065775501780017,
+      "learning_rate": 0.0001436894140448183,
+      "loss": 0.1085,
+      "step": 494
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.1022589987121634,
+      "learning_rate": 0.00014346764217659653,
+      "loss": 0.1185,
+      "step": 495
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.0775235088101212,
+      "learning_rate": 0.0001432456064048204,
+      "loss": 0.0981,
+      "step": 496
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.1963150327160164,
+      "learning_rate": 0.00014302330807752786,
+      "loss": 0.1253,
+      "step": 497
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.0399771365234602,
+      "learning_rate": 0.0001428007485443509,
+      "loss": 0.0773,
+      "step": 498
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.1108878636722963,
+      "learning_rate": 0.00014257792915650728,
+      "loss": 0.0953,
+      "step": 499
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.108845946848657,
+      "learning_rate": 0.00014235485126679243,
+      "loss": 0.126,
+      "step": 500
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.1619040746623464,
+      "learning_rate": 0.00014213151622957128,
+      "loss": 0.1407,
+      "step": 501
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.2162486710138671,
+      "learning_rate": 0.00014190792540076986,
+      "loss": 0.1417,
+      "step": 502
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.0277984588739848,
+      "learning_rate": 0.00014168408013786728,
+      "loss": 0.0769,
+      "step": 503
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.1037858661652986,
+      "learning_rate": 0.00014145998179988735,
+      "loss": 0.0889,
+      "step": 504
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.9955897970776222,
+      "learning_rate": 0.00014123563174739037,
+      "loss": 0.0913,
+      "step": 505
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.004603808737871,
+      "learning_rate": 0.0001410110313424648,
+      "loss": 0.0945,
+      "step": 506
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.9521375009935686,
+      "learning_rate": 0.00014078618194871914,
+      "loss": 0.0849,
+      "step": 507
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.2056722387889085,
+      "learning_rate": 0.0001405610849312736,
+      "loss": 0.1252,
+      "step": 508
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.0034574243348113,
+      "learning_rate": 0.00014033574165675164,
+      "loss": 0.1045,
+      "step": 509
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.936242860709272,
+      "learning_rate": 0.00014011015349327187,
+      "loss": 0.1073,
+      "step": 510
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.9539554959023054,
+      "learning_rate": 0.00013988432181043982,
+      "loss": 0.0772,
+      "step": 511
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.8594050537619622,
+      "learning_rate": 0.00013965824797933926,
+      "loss": 0.0724,
+      "step": 512
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.9084347357804663,
+      "learning_rate": 0.0001394319333725243,
+      "loss": 0.0783,
+      "step": 513
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.9791758004678371,
+      "learning_rate": 0.00013920537936401077,
+      "loss": 0.0929,
+      "step": 514
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.8685870183555011,
+      "learning_rate": 0.00013897858732926793,
+      "loss": 0.0694,
+      "step": 515
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.9348385116099472,
+      "learning_rate": 0.0001387515586452103,
+      "loss": 0.0911,
+      "step": 516
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.9225167849249485,
+      "learning_rate": 0.000138524294690189,
+      "loss": 0.0732,
+      "step": 517
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.8245676748975493,
+      "learning_rate": 0.00013829679684398375,
+      "loss": 0.0709,
+      "step": 518
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.2206670471582055,
+      "learning_rate": 0.000138069066487794,
+      "loss": 0.1186,
+      "step": 519
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.8766127308851455,
+      "learning_rate": 0.00013784110500423104,
+      "loss": 0.0931,
+      "step": 520
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.1134359097549715,
+      "learning_rate": 0.00013761291377730936,
+      "loss": 0.1237,
+      "step": 521
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.8213784062077529,
+      "learning_rate": 0.00013738449419243827,
+      "loss": 0.0646,
+      "step": 522
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.9481138543739505,
+      "learning_rate": 0.00013715584763641345,
+      "loss": 0.0919,
+      "step": 523
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.854411363730186,
+      "learning_rate": 0.0001369269754974087,
+      "loss": 0.0618,
+      "step": 524
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.180010872060639,
+      "learning_rate": 0.00013669787916496722,
+      "loss": 0.1078,
+      "step": 525
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.9496822361295773,
+      "learning_rate": 0.00013646856002999354,
+      "loss": 0.0822,
+      "step": 526
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.6858015228524464,
+      "learning_rate": 0.00013623901948474473,
+      "loss": 0.0405,
+      "step": 527
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.6571394073795536,
+      "learning_rate": 0.00013600925892282218,
+      "loss": 0.0511,
+      "step": 528
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.762961035308753,
+      "learning_rate": 0.00013577927973916306,
+      "loss": 0.0528,
+      "step": 529
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.7520548361567521,
+      "learning_rate": 0.0001355490833300318,
+      "loss": 0.042,
+      "step": 530
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.7333475928961776,
+      "learning_rate": 0.00013531867109301175,
+      "loss": 0.0469,
+      "step": 531
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.7676491123660488,
+      "learning_rate": 0.00013508804442699648,
+      "loss": 0.0533,
+      "step": 532
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.6967119199236123,
+      "learning_rate": 0.00013485720473218154,
+      "loss": 0.0408,
+      "step": 533
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.6609643484866947,
+      "learning_rate": 0.00013462615341005573,
+      "loss": 0.045,
+      "step": 534
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.6480544956263923,
+      "learning_rate": 0.00013439489186339282,
+      "loss": 0.0331,
+      "step": 535
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.7091472463448646,
+      "learning_rate": 0.0001341634214962428,
+      "loss": 0.0517,
+      "step": 536
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.6904985583702147,
+      "learning_rate": 0.00013393174371392348,
+      "loss": 0.0392,
+      "step": 537
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.8055534074393317,
+      "learning_rate": 0.00013369985992301198,
+      "loss": 0.0538,
+      "step": 538
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.5951010282873163,
+      "learning_rate": 0.00013346777153133615,
+      "loss": 0.0277,
+      "step": 539
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.7084094048077838,
+      "learning_rate": 0.00013323547994796597,
+      "loss": 0.0419,
+      "step": 540
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.5594336269324913,
+      "learning_rate": 0.00013300298658320517,
+      "loss": 0.0341,
+      "step": 541
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.7569860647877042,
+      "learning_rate": 0.00013277029284858237,
+      "loss": 0.0358,
+      "step": 542
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.6912671078138903,
+      "learning_rate": 0.00013253740015684284,
+      "loss": 0.0462,
+      "step": 543
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.5122227668617941,
+      "learning_rate": 0.00013230430992193973,
+      "loss": 0.0256,
+      "step": 544
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.5828532655266476,
+      "learning_rate": 0.00013207102355902552,
+      "loss": 0.031,
+      "step": 545
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.5394379778357576,
+      "learning_rate": 0.00013183754248444343,
+      "loss": 0.0292,
+      "step": 546
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.6536904045509789,
+      "learning_rate": 0.00013160386811571876,
+      "loss": 0.0378,
+      "step": 547
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.6912014052991422,
+      "learning_rate": 0.0001313700018715505,
+      "loss": 0.0366,
+      "step": 548
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.778069251795465,
+      "learning_rate": 0.00013113594517180242,
+      "loss": 0.0489,
+      "step": 549
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.6541634109474361,
+      "learning_rate": 0.00013090169943749476,
+      "loss": 0.0382,
+      "step": 550
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.7562685180743578,
+      "learning_rate": 0.00013066726609079526,
+      "loss": 0.0375,
+      "step": 551
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.6064868860346022,
+      "learning_rate": 0.00013043264655501074,
+      "loss": 0.0382,
+      "step": 552
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.5522536847261518,
+      "learning_rate": 0.00013019784225457855,
+      "loss": 0.0241,
+      "step": 553
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.6457456584832749,
+      "learning_rate": 0.0001299628546150577,
+      "loss": 0.0379,
+      "step": 554
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.5210727769955851,
+      "learning_rate": 0.00012972768506312027,
+      "loss": 0.0239,
+      "step": 555
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.6046357834548366,
+      "learning_rate": 0.00012949233502654284,
+      "loss": 0.0281,
+      "step": 556
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.5904117725405653,
+      "learning_rate": 0.00012925680593419778,
+      "loss": 0.0372,
+      "step": 557
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.5246939436151645,
+      "learning_rate": 0.00012902109921604448,
+      "loss": 0.0272,
+      "step": 558
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.5595638490514965,
+      "learning_rate": 0.00012878521630312078,
+      "loss": 0.0247,
+      "step": 559
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.5337591015992911,
+      "learning_rate": 0.00012854915862753422,
+      "loss": 0.0245,
+      "step": 560
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.6692944468313586,
+      "learning_rate": 0.0001283129276224534,
+      "loss": 0.0295,
+      "step": 561
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.6427986679099493,
+      "learning_rate": 0.0001280765247220993,
+      "loss": 0.0401,
+      "step": 562
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.6115096607448915,
+      "learning_rate": 0.0001278399513617364,
+      "loss": 0.037,
+      "step": 563
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.4660536346159077,
+      "learning_rate": 0.0001276032089776642,
+      "loss": 0.0211,
+      "step": 564
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.5649495300419388,
+      "learning_rate": 0.0001273662990072083,
+      "loss": 0.0314,
+      "step": 565
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.6414870475846763,
+      "learning_rate": 0.0001271292228887118,
+      "loss": 0.0257,
+      "step": 566
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.5313420070317386,
+      "learning_rate": 0.00012689198206152657,
+      "loss": 0.02,
+      "step": 567
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.5316969086531306,
+      "learning_rate": 0.00012665457796600443,
+      "loss": 0.0289,
+      "step": 568
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.5030722888722443,
+      "learning_rate": 0.0001264170120434884,
+      "loss": 0.0247,
+      "step": 569
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.6050404714396409,
+      "learning_rate": 0.00012617928573630406,
+      "loss": 0.0283,
+      "step": 570
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.5851526476600509,
+      "learning_rate": 0.0001259414004877507,
+      "loss": 0.0257,
+      "step": 571
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.6599791646555999,
+      "learning_rate": 0.0001257033577420926,
+      "loss": 0.0323,
+      "step": 572
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.6724313553982461,
+      "learning_rate": 0.00012546515894455026,
+      "loss": 0.0351,
+      "step": 573
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.5324688904410898,
+      "learning_rate": 0.00012522680554129156,
+      "loss": 0.03,
+      "step": 574
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.620516231976994,
+      "learning_rate": 0.0001249882989794231,
+      "loss": 0.0412,
+      "step": 575
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.6781119292449547,
+      "learning_rate": 0.00012474964070698127,
+      "loss": 0.0385,
+      "step": 576
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.6277887422000257,
+      "learning_rate": 0.00012451083217292357,
+      "loss": 0.0333,
+      "step": 577
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.6445125916699169,
+      "learning_rate": 0.00012427187482711986,
+      "loss": 0.0388,
+      "step": 578
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.6138662551863093,
+      "learning_rate": 0.0001240327701203433,
+      "loss": 0.0342,
+      "step": 579
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.4640076126065093,
+      "learning_rate": 0.00012379351950426187,
+      "loss": 0.0213,
+      "step": 580
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.572798081700149,
+      "learning_rate": 0.00012355412443142936,
+      "loss": 0.0344,
+      "step": 581
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.5360388196785951,
+      "learning_rate": 0.00012331458635527658,
+      "loss": 0.0334,
+      "step": 582
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.6008317385274303,
+      "learning_rate": 0.0001230749067301025,
+      "loss": 0.0309,
+      "step": 583
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.5760839680110084,
+      "learning_rate": 0.00012283508701106557,
+      "loss": 0.0303,
+      "step": 584
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.5286488276263329,
+      "learning_rate": 0.00012259512865417477,
+      "loss": 0.0246,
+      "step": 585
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.5438221627469808,
+      "learning_rate": 0.00012235503311628073,
+      "loss": 0.0326,
+      "step": 586
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.4297787949314061,
+      "learning_rate": 0.00012211480185506698,
+      "loss": 0.0208,
+      "step": 587
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.6424723090750846,
+      "learning_rate": 0.00012187443632904105,
+      "loss": 0.0281,
+      "step": 588
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.5257966002392329,
+      "learning_rate": 0.00012163393799752565,
+      "loss": 0.0252,
+      "step": 589
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.4095673255602295,
+      "learning_rate": 0.00012139330832064974,
+      "loss": 0.0166,
+      "step": 590
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.5583434593510831,
+      "learning_rate": 0.00012115254875933979,
+      "loss": 0.0292,
+      "step": 591
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.5168891474822531,
+      "learning_rate": 0.00012091166077531075,
+      "loss": 0.0249,
+      "step": 592
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.6728944038128665,
+      "learning_rate": 0.00012067064583105729,
+      "loss": 0.0469,
+      "step": 593
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.4769244081553537,
+      "learning_rate": 0.00012042950538984492,
+      "loss": 0.0241,
+      "step": 594
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.5034822969563644,
+      "learning_rate": 0.00012018824091570103,
+      "loss": 0.0312,
+      "step": 595
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.563073461578303,
+      "learning_rate": 0.00011994685387340607,
+      "loss": 0.026,
+      "step": 596
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.6519903358677784,
+      "learning_rate": 0.00011970534572848464,
+      "loss": 0.0407,
+      "step": 597
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.5116804349501632,
+      "learning_rate": 0.00011946371794719656,
+      "loss": 0.0201,
+      "step": 598
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.5181796932487033,
+      "learning_rate": 0.000119221971996528,
+      "loss": 0.0318,
+      "step": 599
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.6057285281257341,
+      "learning_rate": 0.0001189801093441826,
+      "loss": 0.032,
+      "step": 600
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.5298306948783188,
+      "learning_rate": 0.00011873813145857249,
+      "loss": 0.0203,
+      "step": 601
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.5712849485420385,
+      "learning_rate": 0.0001184960398088094,
+      "loss": 0.0307,
+      "step": 602
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.4811744746646079,
+      "learning_rate": 0.00011825383586469583,
+      "loss": 0.0155,
+      "step": 603
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.6558421666825484,
+      "learning_rate": 0.00011801152109671595,
+      "loss": 0.0374,
+      "step": 604
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4376023286758574,
+      "learning_rate": 0.00011776909697602689,
+      "loss": 0.0187,
+      "step": 605
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.49801437853935393,
+      "learning_rate": 0.00011752656497444952,
+      "loss": 0.0182,
+      "step": 606
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.5790325116792634,
+      "learning_rate": 0.00011728392656445981,
+      "loss": 0.0262,
+      "step": 607
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.5128280483717645,
+      "learning_rate": 0.00011704118321917976,
+      "loss": 0.0292,
+      "step": 608
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.2830429198481699,
+      "learning_rate": 0.00011679833641236844,
+      "loss": 0.0086,
+      "step": 609
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.501953513125815,
+      "learning_rate": 0.000116555387618413,
+      "loss": 0.0178,
+      "step": 610
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.5062182874132735,
+      "learning_rate": 0.00011631233831231991,
+      "loss": 0.0262,
+      "step": 611
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.6154714955243902,
+      "learning_rate": 0.00011606918996970573,
+      "loss": 0.0337,
+      "step": 612
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.6020232162987171,
+      "learning_rate": 0.00011582594406678839,
+      "loss": 0.0333,
+      "step": 613
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.44498590690673834,
+      "learning_rate": 0.00011558260208037817,
+      "loss": 0.0203,
+      "step": 614
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.5952890614278376,
+      "learning_rate": 0.00011533916548786857,
+      "loss": 0.0262,
+      "step": 615
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.6071787838184781,
+      "learning_rate": 0.00011509563576722753,
+      "loss": 0.0252,
+      "step": 616
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.47234449802232037,
+      "learning_rate": 0.00011485201439698847,
+      "loss": 0.0169,
+      "step": 617
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4933328172799777,
+      "learning_rate": 0.00011460830285624118,
+      "loss": 0.0213,
+      "step": 618
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.6372253732306842,
+      "learning_rate": 0.00011436450262462284,
+      "loss": 0.0333,
+      "step": 619
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.519219720648446,
+      "learning_rate": 0.00011412061518230914,
+      "loss": 0.0253,
+      "step": 620
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.6196689800360667,
+      "learning_rate": 0.00011387664201000532,
+      "loss": 0.0292,
+      "step": 621
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.5775458817874398,
+      "learning_rate": 0.00011363258458893699,
+      "loss": 0.0311,
+      "step": 622
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.5697725668136054,
+      "learning_rate": 0.00011338844440084138,
+      "loss": 0.0284,
+      "step": 623
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.5911597846051566,
+      "learning_rate": 0.0001131442229279581,
+      "loss": 0.0275,
+      "step": 624
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.5279414911914028,
+      "learning_rate": 0.00011289992165302035,
+      "loss": 0.0195,
+      "step": 625
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.608273796586728,
+      "learning_rate": 0.00011265554205924575,
+      "loss": 0.0413,
+      "step": 626
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.5856245679475895,
+      "learning_rate": 0.0001124110856303275,
+      "loss": 0.0273,
+      "step": 627
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.7124940869986345,
+      "learning_rate": 0.00011216655385042525,
+      "loss": 0.031,
+      "step": 628
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.6123250240389259,
+      "learning_rate": 0.00011192194820415609,
+      "loss": 0.0331,
+      "step": 629
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.5358901150929475,
+      "learning_rate": 0.00011167727017658562,
+      "loss": 0.0324,
+      "step": 630
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.4544843163631446,
+      "learning_rate": 0.00011143252125321892,
+      "loss": 0.0174,
+      "step": 631
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.5805280562873458,
+      "learning_rate": 0.00011118770291999137,
+      "loss": 0.037,
+      "step": 632
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.4472561452883386,
+      "learning_rate": 0.00011094281666325988,
+      "loss": 0.0172,
+      "step": 633
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.5188702473567809,
+      "learning_rate": 0.00011069786396979367,
+      "loss": 0.0212,
+      "step": 634
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.5400244435354565,
+      "learning_rate": 0.00011045284632676536,
+      "loss": 0.0249,
+      "step": 635
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.5420587374236245,
+      "learning_rate": 0.00011020776522174186,
+      "loss": 0.0238,
+      "step": 636
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.6196179121250547,
+      "learning_rate": 0.0001099626221426754,
+      "loss": 0.0293,
+      "step": 637
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.5383719118779018,
+      "learning_rate": 0.0001097174185778945,
+      "loss": 0.0358,
+      "step": 638
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.6762263205904572,
+      "learning_rate": 0.00010947215601609479,
+      "loss": 0.0409,
+      "step": 639
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.6256688216263538,
+      "learning_rate": 0.00010922683594633021,
+      "loss": 0.0339,
+      "step": 640
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.49551381751762735,
+      "learning_rate": 0.00010898145985800381,
+      "loss": 0.0191,
+      "step": 641
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.5246286239646697,
+      "learning_rate": 0.00010873602924085869,
+      "loss": 0.0256,
+      "step": 642
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.524115852815209,
+      "learning_rate": 0.00010849054558496905,
+      "loss": 0.0254,
+      "step": 643
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.5420709441073354,
+      "learning_rate": 0.00010824501038073116,
+      "loss": 0.0232,
+      "step": 644
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.5701070050530599,
+      "learning_rate": 0.00010799942511885418,
+      "loss": 0.025,
+      "step": 645
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.5080457887402549,
+      "learning_rate": 0.00010775379129035116,
+      "loss": 0.0244,
+      "step": 646
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.45654050525469825,
+      "learning_rate": 0.00010750811038653008,
+      "loss": 0.0188,
+      "step": 647
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.5189413569394382,
+      "learning_rate": 0.00010726238389898471,
+      "loss": 0.0225,
+      "step": 648
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.5058080435911045,
+      "learning_rate": 0.00010701661331958553,
+      "loss": 0.0296,
+      "step": 649
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.519250316789339,
+      "learning_rate": 0.00010677080014047076,
+      "loss": 0.0258,
+      "step": 650
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.5290400761444146,
+      "learning_rate": 0.00010652494585403725,
+      "loss": 0.0215,
+      "step": 651
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.6615093590815386,
+      "learning_rate": 0.00010627905195293135,
+      "loss": 0.0266,
+      "step": 652
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.5354939767954683,
+      "learning_rate": 0.00010603311993004004,
+      "loss": 0.0245,
+      "step": 653
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.5351407511539458,
+      "learning_rate": 0.00010578715127848167,
+      "loss": 0.0272,
+      "step": 654
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.5698775113909572,
+      "learning_rate": 0.000105541147491597,
+      "loss": 0.0269,
+      "step": 655
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.6130353474867041,
+      "learning_rate": 0.00010529511006294009,
+      "loss": 0.0297,
+      "step": 656
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.4895569517150817,
+      "learning_rate": 0.00010504904048626925,
+      "loss": 0.0199,
+      "step": 657
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.47257677049891267,
+      "learning_rate": 0.00010480294025553798,
+      "loss": 0.0222,
+      "step": 658
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4672637211308872,
+      "learning_rate": 0.00010455681086488586,
+      "loss": 0.0174,
+      "step": 659
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.5602667058319084,
+      "learning_rate": 0.00010431065380862959,
+      "loss": 0.025,
+      "step": 660
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.5802731549365994,
+      "learning_rate": 0.00010406447058125368,
+      "loss": 0.032,
+      "step": 661
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.5518267842664953,
+      "learning_rate": 0.00010381826267740171,
+      "loss": 0.0235,
+      "step": 662
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.44442501419740804,
+      "learning_rate": 0.00010357203159186694,
+      "loss": 0.0228,
+      "step": 663
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.5484365074883688,
+      "learning_rate": 0.0001033257788195835,
+      "loss": 0.0304,
+      "step": 664
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.45203868868231645,
+      "learning_rate": 0.00010307950585561706,
+      "loss": 0.0175,
+      "step": 665
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.5528661702202621,
+      "learning_rate": 0.0001028332141951559,
+      "loss": 0.0344,
+      "step": 666
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.565833520307177,
+      "learning_rate": 0.0001025869053335019,
+      "loss": 0.0245,
+      "step": 667
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.46308852726806676,
+      "learning_rate": 0.0001023405807660613,
+      "loss": 0.0205,
+      "step": 668
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.5041398287444961,
+      "learning_rate": 0.0001020942419883357,
+      "loss": 0.017,
+      "step": 669
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.5738142635147863,
+      "learning_rate": 0.00010184789049591299,
+      "loss": 0.0227,
+      "step": 670
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.46258284664566995,
+      "learning_rate": 0.00010160152778445829,
+      "loss": 0.0147,
+      "step": 671
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.5813069642397575,
+      "learning_rate": 0.0001013551553497047,
+      "loss": 0.0276,
+      "step": 672
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.5222449914794433,
+      "learning_rate": 0.0001011087746874445,
+      "loss": 0.0266,
+      "step": 673
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.6134613380703298,
+      "learning_rate": 0.00010086238729351988,
+      "loss": 0.0299,
+      "step": 674
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.6211598871101578,
+      "learning_rate": 0.00010061599466381389,
+      "loss": 0.0321,
+      "step": 675
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.5468825861285473,
+      "learning_rate": 0.00010036959829424131,
+      "loss": 0.0304,
+      "step": 676
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.42605364461242434,
+      "learning_rate": 0.0001001231996807397,
+      "loss": 0.0155,
+      "step": 677
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.5727505561847714,
+      "learning_rate": 9.987680031926032e-05,
+      "loss": 0.0259,
+      "step": 678
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.4473190431236514,
+      "learning_rate": 9.96304017057587e-05,
+      "loss": 0.0217,
+      "step": 679
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.4415294808959684,
+      "learning_rate": 9.938400533618615e-05,
+      "loss": 0.0261,
+      "step": 680
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.48466603607124437,
+      "learning_rate": 9.913761270648015e-05,
+      "loss": 0.0196,
+      "step": 681
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.7337997382440585,
+      "learning_rate": 9.889122531255552e-05,
+      "loss": 0.0494,
+      "step": 682
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.47049961987293065,
+      "learning_rate": 9.864484465029536e-05,
+      "loss": 0.0195,
+      "step": 683
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.49256398291908615,
+      "learning_rate": 9.839847221554175e-05,
+      "loss": 0.0235,
+      "step": 684
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.7528810950813937,
+      "learning_rate": 9.815210950408704e-05,
+      "loss": 0.0334,
+      "step": 685
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.5952646132589948,
+      "learning_rate": 9.790575801166432e-05,
+      "loss": 0.0302,
+      "step": 686
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.399457185881344,
+      "learning_rate": 9.765941923393874e-05,
+      "loss": 0.0139,
+      "step": 687
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.5657337873742687,
+      "learning_rate": 9.741309466649813e-05,
+      "loss": 0.0258,
+      "step": 688
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.600437499908102,
+      "learning_rate": 9.716678580484411e-05,
+      "loss": 0.0255,
+      "step": 689
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.522442346647649,
+      "learning_rate": 9.692049414438299e-05,
+      "loss": 0.0261,
+      "step": 690
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.5016041589897849,
+      "learning_rate": 9.667422118041651e-05,
+      "loss": 0.0194,
+      "step": 691
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.5056510062574688,
+      "learning_rate": 9.642796840813308e-05,
+      "loss": 0.0198,
+      "step": 692
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.5682714181610791,
+      "learning_rate": 9.61817373225983e-05,
+      "loss": 0.0231,
+      "step": 693
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.47200147810919907,
+      "learning_rate": 9.593552941874635e-05,
+      "loss": 0.0193,
+      "step": 694
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.5610337017702816,
+      "learning_rate": 9.568934619137046e-05,
+      "loss": 0.0196,
+      "step": 695
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.5595816223930552,
+      "learning_rate": 9.544318913511416e-05,
+      "loss": 0.0211,
+      "step": 696
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.6385155240889565,
+      "learning_rate": 9.519705974446207e-05,
+      "loss": 0.0341,
+      "step": 697
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.44784328882578667,
+      "learning_rate": 9.495095951373076e-05,
+      "loss": 0.0145,
+      "step": 698
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.5275303714621343,
+      "learning_rate": 9.470488993705992e-05,
+      "loss": 0.0254,
+      "step": 699
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.6131637499215741,
+      "learning_rate": 9.4458852508403e-05,
+      "loss": 0.0202,
+      "step": 700
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.5679057525626448,
+      "learning_rate": 9.421284872151836e-05,
+      "loss": 0.0267,
+      "step": 701
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.5529403069118796,
+      "learning_rate": 9.396688006996e-05,
+      "loss": 0.024,
+      "step": 702
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43085141841166075,
+      "learning_rate": 9.372094804706867e-05,
+      "loss": 0.0172,
+      "step": 703
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.7317250890330461,
+      "learning_rate": 9.34750541459628e-05,
+      "loss": 0.0312,
+      "step": 704
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.5842960023180133,
+      "learning_rate": 9.322919985952926e-05,
+      "loss": 0.0294,
+      "step": 705
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.4786962319157736,
+      "learning_rate": 9.298338668041451e-05,
+      "loss": 0.0181,
+      "step": 706
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.4727243696288956,
+      "learning_rate": 9.27376161010153e-05,
+      "loss": 0.0203,
+      "step": 707
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.4942805196210144,
+      "learning_rate": 9.249188961346993e-05,
+      "loss": 0.0221,
+      "step": 708
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.6185755663215672,
+      "learning_rate": 9.224620870964886e-05,
+      "loss": 0.0296,
+      "step": 709
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.47283254490235566,
+      "learning_rate": 9.200057488114585e-05,
+      "loss": 0.0203,
+      "step": 710
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.4553843934598605,
+      "learning_rate": 9.175498961926886e-05,
+      "loss": 0.0212,
+      "step": 711
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.4217347170131017,
+      "learning_rate": 9.150945441503093e-05,
+      "loss": 0.0145,
+      "step": 712
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.5171269187612787,
+      "learning_rate": 9.126397075914135e-05,
+      "loss": 0.0199,
+      "step": 713
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.4956330175447685,
+      "learning_rate": 9.101854014199622e-05,
+      "loss": 0.0222,
+      "step": 714
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.5084160944545737,
+      "learning_rate": 9.077316405366981e-05,
+      "loss": 0.0198,
+      "step": 715
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.6214001353662084,
+      "learning_rate": 9.052784398390525e-05,
+      "loss": 0.0352,
+      "step": 716
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39498527455529264,
+      "learning_rate": 9.028258142210552e-05,
+      "loss": 0.015,
+      "step": 717
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.47055833355962373,
+      "learning_rate": 9.00373778573246e-05,
+      "loss": 0.0189,
+      "step": 718
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.5220024286179519,
+      "learning_rate": 8.979223477825814e-05,
+      "loss": 0.0204,
+      "step": 719
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.4963380691750351,
+      "learning_rate": 8.954715367323468e-05,
+      "loss": 0.0213,
+      "step": 720
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.584006429597866,
+      "learning_rate": 8.930213603020638e-05,
+      "loss": 0.031,
+      "step": 721
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.4424997488579047,
+      "learning_rate": 8.905718333674013e-05,
+      "loss": 0.0244,
+      "step": 722
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.4664664224836227,
+      "learning_rate": 8.881229708000865e-05,
+      "loss": 0.0172,
+      "step": 723
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.44862845194363976,
+      "learning_rate": 8.85674787467811e-05,
+      "loss": 0.0195,
+      "step": 724
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.5254680842782607,
+      "learning_rate": 8.832272982341439e-05,
+      "loss": 0.022,
+      "step": 725
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.5108846998739364,
+      "learning_rate": 8.80780517958439e-05,
+      "loss": 0.0319,
+      "step": 726
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.4775708906243602,
+      "learning_rate": 8.783344614957477e-05,
+      "loss": 0.0262,
+      "step": 727
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.49699514969294745,
+      "learning_rate": 8.758891436967252e-05,
+      "loss": 0.025,
+      "step": 728
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.47591805097451106,
+      "learning_rate": 8.734445794075428e-05,
+      "loss": 0.0116,
+      "step": 729
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.5219573868667244,
+      "learning_rate": 8.710007834697969e-05,
+      "loss": 0.0238,
+      "step": 730
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.435410140260886,
+      "learning_rate": 8.68557770720419e-05,
+      "loss": 0.0126,
+      "step": 731
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.47180477921084735,
+      "learning_rate": 8.661155559915863e-05,
+      "loss": 0.0173,
+      "step": 732
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.3968967293138565,
+      "learning_rate": 8.636741541106299e-05,
+      "loss": 0.0134,
+      "step": 733
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.4774917401618593,
+      "learning_rate": 8.61233579899947e-05,
+      "loss": 0.0175,
+      "step": 734
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.5300371367359242,
+      "learning_rate": 8.587938481769089e-05,
+      "loss": 0.0268,
+      "step": 735
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.5958398604390659,
+      "learning_rate": 8.563549737537719e-05,
+      "loss": 0.0291,
+      "step": 736
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4478812589747157,
+      "learning_rate": 8.539169714375885e-05,
+      "loss": 0.0224,
+      "step": 737
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.3987992897699387,
+      "learning_rate": 8.514798560301152e-05,
+      "loss": 0.0163,
+      "step": 738
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.4592665533473802,
+      "learning_rate": 8.490436423277248e-05,
+      "loss": 0.0178,
+      "step": 739
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.5701086588589518,
+      "learning_rate": 8.466083451213144e-05,
+      "loss": 0.0322,
+      "step": 740
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.6016849009736805,
+      "learning_rate": 8.441739791962187e-05,
+      "loss": 0.0353,
+      "step": 741
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.4222993539793022,
+      "learning_rate": 8.417405593321163e-05,
+      "loss": 0.0163,
+      "step": 742
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.6198480681990343,
+      "learning_rate": 8.393081003029431e-05,
+      "loss": 0.0269,
+      "step": 743
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.5805070041998106,
+      "learning_rate": 8.368766168768014e-05,
+      "loss": 0.0291,
+      "step": 744
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.5163843354184551,
+      "learning_rate": 8.344461238158699e-05,
+      "loss": 0.0275,
+      "step": 745
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.4880144278941018,
+      "learning_rate": 8.320166358763159e-05,
+      "loss": 0.0206,
+      "step": 746
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.4747850164588571,
+      "learning_rate": 8.295881678082024e-05,
+      "loss": 0.0205,
+      "step": 747
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.4583198689345464,
+      "learning_rate": 8.271607343554021e-05,
+      "loss": 0.0249,
+      "step": 748
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.6101699826029817,
+      "learning_rate": 8.247343502555053e-05,
+      "loss": 0.0285,
+      "step": 749
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.4638191159335852,
+      "learning_rate": 8.223090302397313e-05,
+      "loss": 0.0155,
+      "step": 750
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.5115783827396744,
+      "learning_rate": 8.198847890328406e-05,
+      "loss": 0.0205,
+      "step": 751
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.6852725507845391,
+      "learning_rate": 8.174616413530418e-05,
+      "loss": 0.0371,
+      "step": 752
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.46987909241574577,
+      "learning_rate": 8.150396019119062e-05,
+      "loss": 0.0156,
+      "step": 753
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.4730730637586314,
+      "learning_rate": 8.126186854142752e-05,
+      "loss": 0.0169,
+      "step": 754
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.3214359224084894,
+      "learning_rate": 8.101989065581743e-05,
+      "loss": 0.0097,
+      "step": 755
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.45972279363925644,
+      "learning_rate": 8.077802800347205e-05,
+      "loss": 0.0182,
+      "step": 756
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.4375311226290829,
+      "learning_rate": 8.053628205280347e-05,
+      "loss": 0.0106,
+      "step": 757
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.4619297633945344,
+      "learning_rate": 8.029465427151538e-05,
+      "loss": 0.0253,
+      "step": 758
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.5150844075096656,
+      "learning_rate": 8.005314612659393e-05,
+      "loss": 0.0163,
+      "step": 759
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.4491135226495127,
+      "learning_rate": 7.9811759084299e-05,
+      "loss": 0.0259,
+      "step": 760
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.3536423066285945,
+      "learning_rate": 7.957049461015512e-05,
+      "loss": 0.0107,
+      "step": 761
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.4438880503615582,
+      "learning_rate": 7.932935416894272e-05,
+      "loss": 0.0178,
+      "step": 762
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.37623621834666027,
+      "learning_rate": 7.908833922468927e-05,
+      "loss": 0.0097,
+      "step": 763
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.5105633683603313,
+      "learning_rate": 7.884745124066023e-05,
+      "loss": 0.019,
+      "step": 764
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.6485848108682821,
+      "learning_rate": 7.860669167935028e-05,
+      "loss": 0.0252,
+      "step": 765
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.4159732578619667,
+      "learning_rate": 7.836606200247436e-05,
+      "loss": 0.0168,
+      "step": 766
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.41576212399541784,
+      "learning_rate": 7.812556367095896e-05,
+      "loss": 0.0187,
+      "step": 767
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.3847390982789357,
+      "learning_rate": 7.788519814493304e-05,
+      "loss": 0.0122,
+      "step": 768
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.43858998005042715,
+      "learning_rate": 7.764496688371929e-05,
+      "loss": 0.014,
+      "step": 769
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.3685823051128708,
+      "learning_rate": 7.740487134582525e-05,
+      "loss": 0.018,
+      "step": 770
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.3581858393041895,
+      "learning_rate": 7.716491298893442e-05,
+      "loss": 0.0128,
+      "step": 771
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.4889802363054539,
+      "learning_rate": 7.692509326989753e-05,
+      "loss": 0.0257,
+      "step": 772
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.4206872214966915,
+      "learning_rate": 7.668541364472346e-05,
+      "loss": 0.0177,
+      "step": 773
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.5330372219745317,
+      "learning_rate": 7.644587556857065e-05,
+      "loss": 0.0213,
+      "step": 774
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.5588861524191437,
+      "learning_rate": 7.620648049573815e-05,
+      "loss": 0.0234,
+      "step": 775
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.5217099791716121,
+      "learning_rate": 7.596722987965669e-05,
+      "loss": 0.0264,
+      "step": 776
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.39884362679267354,
+      "learning_rate": 7.572812517288018e-05,
+      "loss": 0.0169,
+      "step": 777
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.5497737744764293,
+      "learning_rate": 7.548916782707642e-05,
+      "loss": 0.0217,
+      "step": 778
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.5049492193512928,
+      "learning_rate": 7.525035929301877e-05,
+      "loss": 0.0182,
+      "step": 779
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.4834305773030485,
+      "learning_rate": 7.50117010205769e-05,
+      "loss": 0.0163,
+      "step": 780
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.5066733381415807,
+      "learning_rate": 7.477319445870845e-05,
+      "loss": 0.0232,
+      "step": 781
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.5191587450653661,
+      "learning_rate": 7.453484105544976e-05,
+      "loss": 0.0229,
+      "step": 782
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.41513313973294463,
+      "learning_rate": 7.429664225790743e-05,
+      "loss": 0.0225,
+      "step": 783
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.5159517105682486,
+      "learning_rate": 7.405859951224933e-05,
+      "loss": 0.0178,
+      "step": 784
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.46806676407919134,
+      "learning_rate": 7.382071426369597e-05,
+      "loss": 0.0195,
+      "step": 785
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.38907075322280127,
+      "learning_rate": 7.358298795651165e-05,
+      "loss": 0.0211,
+      "step": 786
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.6253105120526593,
+      "learning_rate": 7.33454220339956e-05,
+      "loss": 0.0255,
+      "step": 787
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.5214375224166684,
+      "learning_rate": 7.310801793847344e-05,
+      "loss": 0.021,
+      "step": 788
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.40072634334330276,
+      "learning_rate": 7.287077711128823e-05,
+      "loss": 0.0121,
+      "step": 789
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.3872714563316075,
+      "learning_rate": 7.263370099279172e-05,
+      "loss": 0.0114,
+      "step": 790
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.21056682906748808,
+      "learning_rate": 7.239679102233582e-05,
+      "loss": 0.0037,
+      "step": 791
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.30784351872518456,
+      "learning_rate": 7.21600486382636e-05,
+      "loss": 0.0108,
+      "step": 792
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 0.36448557916568775,
+      "learning_rate": 7.192347527790073e-05,
+      "loss": 0.0138,
+      "step": 793
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 0.4208207403069888,
+      "learning_rate": 7.168707237754658e-05,
+      "loss": 0.0185,
+      "step": 794
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 0.2708962378411728,
+      "learning_rate": 7.14508413724658e-05,
+      "loss": 0.0085,
+      "step": 795
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.2960738228440382,
+      "learning_rate": 7.121478369687926e-05,
+      "loss": 0.0189,
+      "step": 796
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.28229585398379153,
+      "learning_rate": 7.097890078395553e-05,
+      "loss": 0.0055,
+      "step": 797
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.32244547156861053,
+      "learning_rate": 7.074319406580224e-05,
+      "loss": 0.009,
+      "step": 798
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.415492701781414,
+      "learning_rate": 7.050766497345714e-05,
+      "loss": 0.0161,
+      "step": 799
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.33912882122938637,
+      "learning_rate": 7.027231493687974e-05,
+      "loss": 0.0113,
+      "step": 800
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.325988432281054,
+      "learning_rate": 7.003714538494233e-05,
+      "loss": 0.0072,
+      "step": 801
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.3675796449121323,
+      "learning_rate": 6.980215774542147e-05,
+      "loss": 0.012,
+      "step": 802
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.3137932486186184,
+      "learning_rate": 6.95673534449893e-05,
+      "loss": 0.0075,
+      "step": 803
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.33318096922741336,
+      "learning_rate": 6.933273390920478e-05,
+      "loss": 0.0073,
+      "step": 804
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.3495871838860547,
+      "learning_rate": 6.909830056250527e-05,
+      "loss": 0.0104,
+      "step": 805
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.4296719571304371,
+      "learning_rate": 6.886405482819756e-05,
+      "loss": 0.0112,
+      "step": 806
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.2832552108674304,
+      "learning_rate": 6.862999812844953e-05,
+      "loss": 0.0078,
+      "step": 807
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.24353373409742712,
+      "learning_rate": 6.839613188428126e-05,
+      "loss": 0.0115,
+      "step": 808
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.23306303194729303,
+      "learning_rate": 6.81624575155566e-05,
+      "loss": 0.0047,
+      "step": 809
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.2633732297072212,
+      "learning_rate": 6.792897644097451e-05,
+      "loss": 0.007,
+      "step": 810
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.35004308706229786,
+      "learning_rate": 6.769569007806027e-05,
+      "loss": 0.0096,
+      "step": 811
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.33162754511706916,
+      "learning_rate": 6.746259984315717e-05,
+      "loss": 0.0096,
+      "step": 812
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.2486494220940083,
+      "learning_rate": 6.722970715141763e-05,
+      "loss": 0.0076,
+      "step": 813
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.2060761460321788,
+      "learning_rate": 6.699701341679488e-05,
+      "loss": 0.005,
+      "step": 814
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.29686564162916673,
+      "learning_rate": 6.676452005203406e-05,
+      "loss": 0.0059,
+      "step": 815
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.27468433547209364,
+      "learning_rate": 6.653222846866389e-05,
+      "loss": 0.0083,
+      "step": 816
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.2240834476254555,
+      "learning_rate": 6.630014007698807e-05,
+      "loss": 0.0071,
+      "step": 817
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.1891084078380572,
+      "learning_rate": 6.606825628607654e-05,
+      "loss": 0.0036,
+      "step": 818
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.29557066258392495,
+      "learning_rate": 6.583657850375723e-05,
+      "loss": 0.0068,
+      "step": 819
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.3226297439272109,
+      "learning_rate": 6.560510813660719e-05,
+      "loss": 0.0164,
+      "step": 820
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.30987812721664565,
+      "learning_rate": 6.537384658994428e-05,
+      "loss": 0.009,
+      "step": 821
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.2605920375598695,
+      "learning_rate": 6.51427952678185e-05,
+      "loss": 0.0076,
+      "step": 822
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.31288606849124107,
+      "learning_rate": 6.491195557300353e-05,
+      "loss": 0.0107,
+      "step": 823
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.3018846731272461,
+      "learning_rate": 6.468132890698829e-05,
+      "loss": 0.007,
+      "step": 824
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.21814291192582394,
+      "learning_rate": 6.44509166699682e-05,
+      "loss": 0.005,
+      "step": 825
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.28364072746502833,
+      "learning_rate": 6.422072026083697e-05,
+      "loss": 0.008,
+      "step": 826
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.24688175782460228,
+      "learning_rate": 6.399074107717782e-05,
+      "loss": 0.0042,
+      "step": 827
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.2876525409202033,
+      "learning_rate": 6.376098051525529e-05,
+      "loss": 0.0104,
+      "step": 828
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.24783790760390528,
+      "learning_rate": 6.35314399700065e-05,
+      "loss": 0.005,
+      "step": 829
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.24473135761339637,
+      "learning_rate": 6.33021208350328e-05,
+      "loss": 0.0142,
+      "step": 830
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.24254397673673195,
+      "learning_rate": 6.307302450259136e-05,
+      "loss": 0.0095,
+      "step": 831
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.28662638165877835,
+      "learning_rate": 6.284415236358653e-05,
+      "loss": 0.0119,
+      "step": 832
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.27261288238299247,
+      "learning_rate": 6.261550580756175e-05,
+      "loss": 0.0064,
+      "step": 833
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.3141339322837964,
+      "learning_rate": 6.238708622269065e-05,
+      "loss": 0.0083,
+      "step": 834
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.22293609936862782,
+      "learning_rate": 6.215889499576898e-05,
+      "loss": 0.0067,
+      "step": 835
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 0.22908761406258643,
+      "learning_rate": 6.193093351220605e-05,
+      "loss": 0.0062,
+      "step": 836
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 0.31064282609153193,
+      "learning_rate": 6.170320315601628e-05,
+      "loss": 0.0147,
+      "step": 837
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.2965572881610421,
+      "learning_rate": 6.147570530981099e-05,
+      "loss": 0.0067,
+      "step": 838
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.2507362214536929,
+      "learning_rate": 6.12484413547897e-05,
+      "loss": 0.0066,
+      "step": 839
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.21721434624800082,
+      "learning_rate": 6.102141267073207e-05,
+      "loss": 0.0103,
+      "step": 840
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.31167165837580324,
+      "learning_rate": 6.0794620635989244e-05,
+      "loss": 0.0056,
+      "step": 841
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.2958740863777161,
+      "learning_rate": 6.056806662747572e-05,
+      "loss": 0.0061,
+      "step": 842
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.24710195244200553,
+      "learning_rate": 6.034175202066077e-05,
+      "loss": 0.006,
+      "step": 843
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.3027224086023083,
+      "learning_rate": 6.011567818956021e-05,
+      "loss": 0.0073,
+      "step": 844
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.35652628992891383,
+      "learning_rate": 5.988984650672813e-05,
+      "loss": 0.0141,
+      "step": 845
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.2564692867189945,
+      "learning_rate": 5.96642583432484e-05,
+      "loss": 0.0137,
+      "step": 846
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.3494670856158618,
+      "learning_rate": 5.943891506872645e-05,
+      "loss": 0.0106,
+      "step": 847
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.3229875785438857,
+      "learning_rate": 5.921381805128088e-05,
+      "loss": 0.0139,
+      "step": 848
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.24728114721366126,
+      "learning_rate": 5.898896865753522e-05,
+      "loss": 0.0097,
+      "step": 849
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.27897698046035213,
+      "learning_rate": 5.876436825260967e-05,
+      "loss": 0.0052,
+      "step": 850
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.26878080374740176,
+      "learning_rate": 5.854001820011265e-05,
+      "loss": 0.0044,
+      "step": 851
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.349425725603305,
+      "learning_rate": 5.831591986213274e-05,
+      "loss": 0.0075,
+      "step": 852
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.3553442628962405,
+      "learning_rate": 5.809207459923016e-05,
+      "loss": 0.0117,
+      "step": 853
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.16517368884881764,
+      "learning_rate": 5.786848377042875e-05,
+      "loss": 0.004,
+      "step": 854
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.2968422099485621,
+      "learning_rate": 5.764514873320761e-05,
+      "loss": 0.0065,
+      "step": 855
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.34348198394081064,
+      "learning_rate": 5.7422070843492734e-05,
+      "loss": 0.0084,
+      "step": 856
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.19397093181793817,
+      "learning_rate": 5.719925145564913e-05,
+      "loss": 0.0056,
+      "step": 857
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.21307662220492118,
+      "learning_rate": 5.697669192247215e-05,
+      "loss": 0.0062,
+      "step": 858
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 0.3474592800865426,
+      "learning_rate": 5.675439359517962e-05,
+      "loss": 0.0088,
+      "step": 859
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 0.29870118509204274,
+      "learning_rate": 5.6532357823403517e-05,
+      "loss": 0.0087,
+      "step": 860
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 0.27828720292152737,
+      "learning_rate": 5.63105859551817e-05,
+      "loss": 0.0057,
+      "step": 861
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.21827817531205063,
+      "learning_rate": 5.608907933694994e-05,
+      "loss": 0.0076,
+      "step": 862
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.3111088594215299,
+      "learning_rate": 5.586783931353338e-05,
+      "loss": 0.0102,
+      "step": 863
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.22923346790242627,
+      "learning_rate": 5.56468672281388e-05,
+      "loss": 0.0074,
+      "step": 864
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.24241133851093505,
+      "learning_rate": 5.542616442234618e-05,
+      "loss": 0.0069,
+      "step": 865
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.36328179004652117,
+      "learning_rate": 5.5205732236100635e-05,
+      "loss": 0.0174,
+      "step": 866
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.32016591871948663,
+      "learning_rate": 5.498557200770429e-05,
+      "loss": 0.0053,
+      "step": 867
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.2996734240922401,
+      "learning_rate": 5.476568507380815e-05,
+      "loss": 0.0097,
+      "step": 868
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.1969391039103401,
+      "learning_rate": 5.454607276940389e-05,
+      "loss": 0.0082,
+      "step": 869
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 0.28123836179709116,
+      "learning_rate": 5.4326736427815946e-05,
+      "loss": 0.0069,
+      "step": 870
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 0.13638110070331544,
+      "learning_rate": 5.410767738069328e-05,
+      "loss": 0.0021,
+      "step": 871
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 0.1351774894217744,
+      "learning_rate": 5.388889695800129e-05,
+      "loss": 0.0029,
+      "step": 872
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 0.22835602750194167,
+      "learning_rate": 5.3670396488013854e-05,
+      "loss": 0.0038,
+      "step": 873
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 0.1300902200167527,
+      "learning_rate": 5.345217729730501e-05,
+      "loss": 0.0025,
+      "step": 874
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.28857829548728386,
+      "learning_rate": 5.3234240710741337e-05,
+      "loss": 0.0081,
+      "step": 875
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.3061372458634303,
+      "learning_rate": 5.301658805147338e-05,
+      "loss": 0.0083,
+      "step": 876
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.1899830664005816,
+      "learning_rate": 5.279922064092808e-05,
+      "loss": 0.0053,
+      "step": 877
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.1174482384781678,
+      "learning_rate": 5.25821397988005e-05,
+      "loss": 0.0022,
+      "step": 878
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.2722808386220444,
+      "learning_rate": 5.236534684304575e-05,
+      "loss": 0.007,
+      "step": 879
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.18861460282477344,
+      "learning_rate": 5.214884308987136e-05,
+      "loss": 0.0062,
+      "step": 880
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.22907254405660735,
+      "learning_rate": 5.193262985372879e-05,
+      "loss": 0.0033,
+      "step": 881
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.33932601780333094,
+      "learning_rate": 5.171670844730581e-05,
+      "loss": 0.008,
+      "step": 882
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.22954496242274092,
+      "learning_rate": 5.150108018151845e-05,
+      "loss": 0.003,
+      "step": 883
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.25309398425206814,
+      "learning_rate": 5.128574636550283e-05,
+      "loss": 0.0059,
+      "step": 884
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.14746350829795857,
+      "learning_rate": 5.107070830660765e-05,
+      "loss": 0.0033,
+      "step": 885
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.28419435718654534,
+      "learning_rate": 5.0855967310385776e-05,
+      "loss": 0.006,
+      "step": 886
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.2466096124650076,
+      "learning_rate": 5.064152468058661e-05,
+      "loss": 0.0128,
+      "step": 887
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.2332366893895638,
+      "learning_rate": 5.0427381719148115e-05,
+      "loss": 0.005,
+      "step": 888
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.22151985687385028,
+      "learning_rate": 5.021353972618877e-05,
+      "loss": 0.0046,
+      "step": 889
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.27594725391990216,
+      "learning_rate": 5.000000000000002e-05,
+      "loss": 0.0054,
+      "step": 890
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.23798418447034836,
+      "learning_rate": 4.978676383703792e-05,
+      "loss": 0.0041,
+      "step": 891
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.3058094151897891,
+      "learning_rate": 4.957383253191567e-05,
+      "loss": 0.0068,
+      "step": 892
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.2143215086556055,
+      "learning_rate": 4.9361207377395526e-05,
+      "loss": 0.0042,
+      "step": 893
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.1385084250600042,
+      "learning_rate": 4.914888966438107e-05,
+      "loss": 0.0021,
+      "step": 894
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.2675591985507158,
+      "learning_rate": 4.893688068190932e-05,
+      "loss": 0.0068,
+      "step": 895
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.2699101034578167,
+      "learning_rate": 4.872518171714285e-05,
+      "loss": 0.0053,
+      "step": 896
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.0987773614832064,
+      "learning_rate": 4.8513794055362094e-05,
+      "loss": 0.0017,
+      "step": 897
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.2867768357409853,
+      "learning_rate": 4.8302718979957465e-05,
+      "loss": 0.0055,
+      "step": 898
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.2692023055775869,
+      "learning_rate": 4.809195777242157e-05,
+      "loss": 0.01,
+      "step": 899
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.0919200107888228,
+      "learning_rate": 4.7881511712341484e-05,
+      "loss": 0.0021,
+      "step": 900
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.17821533920098714,
+      "learning_rate": 4.7671382077390923e-05,
+      "loss": 0.0064,
+      "step": 901
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.14481351930897424,
+      "learning_rate": 4.746157014332242e-05,
+      "loss": 0.0053,
+      "step": 902
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.40620740766951424,
+      "learning_rate": 4.7252077183959766e-05,
+      "loss": 0.0073,
+      "step": 903
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.2318035659082105,
+      "learning_rate": 4.704290447119013e-05,
+      "loss": 0.0039,
+      "step": 904
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.28180304468632866,
+      "learning_rate": 4.683405327495638e-05,
+      "loss": 0.0089,
+      "step": 905
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.25257580920854156,
+      "learning_rate": 4.6625524863249435e-05,
+      "loss": 0.0068,
+      "step": 906
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.2756381942170026,
+      "learning_rate": 4.6417320502100316e-05,
+      "loss": 0.0062,
+      "step": 907
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.11886082249100405,
+      "learning_rate": 4.6209441455572934e-05,
+      "loss": 0.002,
+      "step": 908
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.3367644855622204,
+      "learning_rate": 4.600188898575585e-05,
+      "loss": 0.0061,
+      "step": 909
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.22436921001498406,
+      "learning_rate": 4.5794664352755055e-05,
+      "loss": 0.0044,
+      "step": 910
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.26716386589388574,
+      "learning_rate": 4.558776881468616e-05,
+      "loss": 0.0052,
+      "step": 911
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 0.2836890776070014,
+      "learning_rate": 4.538120362766659e-05,
+      "loss": 0.0035,
+      "step": 912
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 0.1719159907041175,
+      "learning_rate": 4.5174970045808373e-05,
+      "loss": 0.0029,
+      "step": 913
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.3978665296675174,
+      "learning_rate": 4.496906932121006e-05,
+      "loss": 0.0061,
+      "step": 914
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.28698458932941184,
+      "learning_rate": 4.476350270394942e-05,
+      "loss": 0.0047,
+      "step": 915
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.2541722328992379,
+      "learning_rate": 4.4558271442075817e-05,
+      "loss": 0.0057,
+      "step": 916
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.3605245472466042,
+      "learning_rate": 4.435337678160244e-05,
+      "loss": 0.0148,
+      "step": 917
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.3139032675968191,
+      "learning_rate": 4.414881996649909e-05,
+      "loss": 0.0135,
+      "step": 918
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.2330292194278837,
+      "learning_rate": 4.394460223868422e-05,
+      "loss": 0.0036,
+      "step": 919
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.19083600536418646,
+      "learning_rate": 4.374072483801769e-05,
+      "loss": 0.0047,
+      "step": 920
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.13582457795092648,
+      "learning_rate": 4.353718900229315e-05,
+      "loss": 0.0021,
+      "step": 921
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.3107549872345709,
+      "learning_rate": 4.333399596723054e-05,
+      "loss": 0.0066,
+      "step": 922
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.19103361601132318,
+      "learning_rate": 4.313114696646844e-05,
+      "loss": 0.0031,
+      "step": 923
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.3452916169554124,
+      "learning_rate": 4.2928643231556844e-05,
+      "loss": 0.0088,
+      "step": 924
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.2920554443502079,
+      "learning_rate": 4.272648599194948e-05,
+      "loss": 0.005,
+      "step": 925
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.20660570216502214,
+      "learning_rate": 4.2524676474996436e-05,
+      "loss": 0.0033,
+      "step": 926
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.09635691456815476,
+      "learning_rate": 4.232321590593672e-05,
+      "loss": 0.0016,
+      "step": 927
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.1684216014603991,
+      "learning_rate": 4.212210550789066e-05,
+      "loss": 0.0029,
+      "step": 928
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.24697826100983358,
+      "learning_rate": 4.192134650185271e-05,
+      "loss": 0.0047,
+      "step": 929
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.2818555533606619,
+      "learning_rate": 4.172094010668391e-05,
+      "loss": 0.0124,
+      "step": 930
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.2020956710004044,
+      "learning_rate": 4.1520887539104516e-05,
+      "loss": 0.0053,
+      "step": 931
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.2764539735060777,
+      "learning_rate": 4.132119001368658e-05,
+      "loss": 0.006,
+      "step": 932
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.27812618927964217,
+      "learning_rate": 4.112184874284655e-05,
+      "loss": 0.0083,
+      "step": 933
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.26091617484641083,
+      "learning_rate": 4.092286493683812e-05,
+      "loss": 0.0046,
+      "step": 934
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.2742087518432973,
+      "learning_rate": 4.072423980374452e-05,
+      "loss": 0.0091,
+      "step": 935
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.2186654584563193,
+      "learning_rate": 4.052597454947151e-05,
+      "loss": 0.0049,
+      "step": 936
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.06343741169546462,
+      "learning_rate": 4.0328070377739936e-05,
+      "loss": 0.0013,
+      "step": 937
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.23409572834724834,
+      "learning_rate": 4.0130528490078255e-05,
+      "loss": 0.0042,
+      "step": 938
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.09603505018094859,
+      "learning_rate": 3.993335008581569e-05,
+      "loss": 0.0028,
+      "step": 939
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.23943575857886956,
+      "learning_rate": 3.973653636207437e-05,
+      "loss": 0.0052,
+      "step": 940
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.19799328432575966,
+      "learning_rate": 3.954008851376252e-05,
+      "loss": 0.0045,
+      "step": 941
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.2930375477359168,
+      "learning_rate": 3.934400773356702e-05,
+      "loss": 0.0055,
+      "step": 942
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.205356759535838,
+      "learning_rate": 3.914829521194606e-05,
+      "loss": 0.0043,
+      "step": 943
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.2568633644839911,
+      "learning_rate": 3.895295213712227e-05,
+      "loss": 0.0053,
+      "step": 944
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.32917761715356114,
+      "learning_rate": 3.875797969507502e-05,
+      "loss": 0.0085,
+      "step": 945
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.13988987297896957,
+      "learning_rate": 3.8563379069533626e-05,
+      "loss": 0.002,
+      "step": 946
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.3024185505983371,
+      "learning_rate": 3.836915144196995e-05,
+      "loss": 0.0128,
+      "step": 947
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.24180662340794465,
+      "learning_rate": 3.8175297991591316e-05,
+      "loss": 0.0045,
+      "step": 948
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 0.3651297540235093,
+      "learning_rate": 3.7981819895333336e-05,
+      "loss": 0.0091,
+      "step": 949
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 0.20379244963238086,
+      "learning_rate": 3.778871832785262e-05,
+      "loss": 0.0036,
+      "step": 950
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.3231664452524896,
+      "learning_rate": 3.759599446151994e-05,
+      "loss": 0.0145,
+      "step": 951
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.1838658695322691,
+      "learning_rate": 3.740364946641284e-05,
+      "loss": 0.0032,
+      "step": 952
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.2764949641154661,
+      "learning_rate": 3.721168451030868e-05,
+      "loss": 0.0087,
+      "step": 953
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.18860779307165287,
+      "learning_rate": 3.702010075867748e-05,
+      "loss": 0.0111,
+      "step": 954
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.1479063723958804,
+      "learning_rate": 3.682889937467493e-05,
+      "loss": 0.0018,
+      "step": 955
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.12334004238243676,
+      "learning_rate": 3.6638081519135115e-05,
+      "loss": 0.0052,
+      "step": 956
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.27038734932845937,
+      "learning_rate": 3.6447648350563767e-05,
+      "loss": 0.007,
+      "step": 957
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.21179985562905868,
+      "learning_rate": 3.6257601025131026e-05,
+      "loss": 0.0028,
+      "step": 958
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.22706406176986857,
+      "learning_rate": 3.6067940696664484e-05,
+      "loss": 0.0037,
+      "step": 959
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.16829930822470549,
+      "learning_rate": 3.587866851664219e-05,
+      "loss": 0.0041,
+      "step": 960
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.16999112011791642,
+      "learning_rate": 3.568978563418551e-05,
+      "loss": 0.004,
+      "step": 961
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.19308465605231837,
+      "learning_rate": 3.5501293196052544e-05,
+      "loss": 0.0048,
+      "step": 962
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.22101325826502366,
+      "learning_rate": 3.531319234663063e-05,
+      "loss": 0.0034,
+      "step": 963
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.2804545091652832,
+      "learning_rate": 3.512548422792983e-05,
+      "loss": 0.0089,
+      "step": 964
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.13808941403940983,
+      "learning_rate": 3.493816997957582e-05,
+      "loss": 0.0023,
+      "step": 965
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.23340717845723516,
+      "learning_rate": 3.4751250738802835e-05,
+      "loss": 0.0025,
+      "step": 966
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.2515812835729633,
+      "learning_rate": 3.456472764044718e-05,
+      "loss": 0.0069,
+      "step": 967
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.15818288659483654,
+      "learning_rate": 3.4378601816939824e-05,
+      "loss": 0.0021,
+      "step": 968
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.104553276450037,
+      "learning_rate": 3.4192874398299915e-05,
+      "loss": 0.0016,
+      "step": 969
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.22999984507474122,
+      "learning_rate": 3.400754651212776e-05,
+      "loss": 0.0046,
+      "step": 970
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.29845371871356685,
+      "learning_rate": 3.382261928359791e-05,
+      "loss": 0.0035,
+      "step": 971
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.2376077365816816,
+      "learning_rate": 3.36380938354526e-05,
+      "loss": 0.0065,
+      "step": 972
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.21647700919177892,
+      "learning_rate": 3.3453971287994545e-05,
+      "loss": 0.0055,
+      "step": 973
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.327425190153311,
+      "learning_rate": 3.3270252759080476e-05,
+      "loss": 0.0095,
+      "step": 974
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.13837128288987777,
+      "learning_rate": 3.308693936411421e-05,
+      "loss": 0.002,
+      "step": 975
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.09141043599534644,
+      "learning_rate": 3.290403221603976e-05,
+      "loss": 0.0013,
+      "step": 976
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.1702227021725679,
+      "learning_rate": 3.2721532425334934e-05,
+      "loss": 0.0022,
+      "step": 977
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 0.29115611328990826,
+      "learning_rate": 3.253944110000415e-05,
+      "loss": 0.0051,
+      "step": 978
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 0.12243175957878329,
+      "learning_rate": 3.235775934557204e-05,
+      "loss": 0.0018,
+      "step": 979
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.12994952503561974,
+      "learning_rate": 3.2176488265076596e-05,
+      "loss": 0.0016,
+      "step": 980
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.23928266029490827,
+      "learning_rate": 3.199562895906252e-05,
+      "loss": 0.0063,
+      "step": 981
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.23193004825040406,
+      "learning_rate": 3.1815182525574495e-05,
+      "loss": 0.0028,
+      "step": 982
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.2924505926099492,
+      "learning_rate": 3.163515006015052e-05,
+      "loss": 0.0068,
+      "step": 983
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.1158181822607879,
+      "learning_rate": 3.1455532655815346e-05,
+      "loss": 0.0018,
+      "step": 984
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.12282675856255618,
+      "learning_rate": 3.1276331403073735e-05,
+      "loss": 0.0022,
+      "step": 985
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.1505372867214304,
+      "learning_rate": 3.10975473899039e-05,
+      "loss": 0.0023,
+      "step": 986
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.16083085511116196,
+      "learning_rate": 3.09191817017509e-05,
+      "loss": 0.0031,
+      "step": 987
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.26000400377883015,
+      "learning_rate": 3.074123542152001e-05,
+      "loss": 0.0049,
+      "step": 988
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.17051507592485188,
+      "learning_rate": 3.056370962957014e-05,
+      "loss": 0.0025,
+      "step": 989
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.22080550671419563,
+      "learning_rate": 3.0386605403707346e-05,
+      "loss": 0.004,
+      "step": 990
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.23630885615098962,
+      "learning_rate": 3.020992381917823e-05,
+      "loss": 0.0191,
+      "step": 991
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.24217233158494697,
+      "learning_rate": 3.0033665948663448e-05,
+      "loss": 0.0065,
+      "step": 992
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.18799988580004937,
+      "learning_rate": 2.9857832862271183e-05,
+      "loss": 0.003,
+      "step": 993
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.27229833092835276,
+      "learning_rate": 2.968242562753051e-05,
+      "loss": 0.0044,
+      "step": 994
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.1636657614372744,
+      "learning_rate": 2.9507445309385294e-05,
+      "loss": 0.0061,
+      "step": 995
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.25572416828302935,
+      "learning_rate": 2.9332892970187255e-05,
+      "loss": 0.0066,
+      "step": 996
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.1690761688898268,
+      "learning_rate": 2.915876966968978e-05,
+      "loss": 0.0047,
+      "step": 997
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.10287534610456749,
+      "learning_rate": 2.8985076465041582e-05,
+      "loss": 0.0013,
+      "step": 998
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.21014688832253814,
+      "learning_rate": 2.8811814410779957e-05,
+      "loss": 0.0065,
+      "step": 999
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.2230263083890269,
+      "learning_rate": 2.8638984558824777e-05,
+      "loss": 0.0057,
+      "step": 1000
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.16327223371346988,
+      "learning_rate": 2.8466587958471713e-05,
+      "loss": 0.0036,
+      "step": 1001
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.31853794312849976,
+      "learning_rate": 2.8294625656386153e-05,
+      "loss": 0.0075,
+      "step": 1002
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.16611663588887882,
+      "learning_rate": 2.812309869659675e-05,
+      "loss": 0.0043,
+      "step": 1003
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.10524374179837746,
+      "learning_rate": 2.7952008120489005e-05,
+      "loss": 0.0014,
+      "step": 1004
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.12770625799454124,
+      "learning_rate": 2.7781354966799078e-05,
+      "loss": 0.0025,
+      "step": 1005
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.31588571578573565,
+      "learning_rate": 2.7611140271607417e-05,
+      "loss": 0.0049,
+      "step": 1006
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.1032087924536394,
+      "learning_rate": 2.744136506833247e-05,
+      "loss": 0.0019,
+      "step": 1007
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.20770741403826715,
+      "learning_rate": 2.7272030387724423e-05,
+      "loss": 0.0062,
+      "step": 1008
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.11248942074914868,
+      "learning_rate": 2.7103137257858868e-05,
+      "loss": 0.0016,
+      "step": 1009
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.2876952639483102,
+      "learning_rate": 2.6934686704130696e-05,
+      "loss": 0.0119,
+      "step": 1010
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.09503155597393347,
+      "learning_rate": 2.6766679749247793e-05,
+      "loss": 0.0026,
+      "step": 1011
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.08594085714064408,
+      "learning_rate": 2.6599117413224817e-05,
+      "loss": 0.0013,
+      "step": 1012
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.23044026820632793,
+      "learning_rate": 2.6432000713377027e-05,
+      "loss": 0.0067,
+      "step": 1013
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.23103635490697938,
+      "learning_rate": 2.6265330664314157e-05,
+      "loss": 0.0074,
+      "step": 1014
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.21567016421107285,
+      "learning_rate": 2.6099108277934103e-05,
+      "loss": 0.0034,
+      "step": 1015
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.21227717462458343,
+      "learning_rate": 2.5933334563416976e-05,
+      "loss": 0.0034,
+      "step": 1016
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.3030082384414693,
+      "learning_rate": 2.5768010527218845e-05,
+      "loss": 0.0083,
+      "step": 1017
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.19256399417436143,
+      "learning_rate": 2.5603137173065674e-05,
+      "loss": 0.003,
+      "step": 1018
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.1585123261207571,
+      "learning_rate": 2.543871550194723e-05,
+      "loss": 0.0032,
+      "step": 1019
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.16701803350284344,
+      "learning_rate": 2.527474651211089e-05,
+      "loss": 0.0025,
+      "step": 1020
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.22564417026931996,
+      "learning_rate": 2.5111231199055896e-05,
+      "loss": 0.007,
+      "step": 1021
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.16803748707764324,
+      "learning_rate": 2.494817055552686e-05,
+      "loss": 0.0031,
+      "step": 1022
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.3328056531118439,
+      "learning_rate": 2.4785565571508118e-05,
+      "loss": 0.0078,
+      "step": 1023
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.06972035346534146,
+      "learning_rate": 2.462341723421758e-05,
+      "loss": 0.0015,
+      "step": 1024
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.16262192118180727,
+      "learning_rate": 2.4461726528100615e-05,
+      "loss": 0.0059,
+      "step": 1025
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.09681113954612036,
+      "learning_rate": 2.4300494434824373e-05,
+      "loss": 0.0013,
+      "step": 1026
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.16033745713919884,
+      "learning_rate": 2.4139721933271465e-05,
+      "loss": 0.002,
+      "step": 1027
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.06889500336319135,
+      "learning_rate": 2.3979409999534298e-05,
+      "loss": 0.0013,
+      "step": 1028
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.3234323528532147,
+      "learning_rate": 2.381955960690906e-05,
+      "loss": 0.0093,
+      "step": 1029
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.13398288212015214,
+      "learning_rate": 2.36601717258897e-05,
+      "loss": 0.0024,
+      "step": 1030
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.2945809357430045,
+      "learning_rate": 2.35012473241623e-05,
+      "loss": 0.0058,
+      "step": 1031
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.08303003772920295,
+      "learning_rate": 2.3342787366598872e-05,
+      "loss": 0.0017,
+      "step": 1032
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.15570044533854507,
+      "learning_rate": 2.3184792815251766e-05,
+      "loss": 0.0024,
+      "step": 1033
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.07621876270230488,
+      "learning_rate": 2.302726462934769e-05,
+      "loss": 0.0011,
+      "step": 1034
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.2407120806363898,
+      "learning_rate": 2.2870203765281926e-05,
+      "loss": 0.0045,
+      "step": 1035
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.21917712522459296,
+      "learning_rate": 2.2713611176612582e-05,
+      "loss": 0.0035,
+      "step": 1036
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.18322208195958656,
+      "learning_rate": 2.25574878140546e-05,
+      "loss": 0.0041,
+      "step": 1037
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.24206434480839525,
+      "learning_rate": 2.240183462547427e-05,
+      "loss": 0.003,
+      "step": 1038
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.07137116220112984,
+      "learning_rate": 2.224665255588325e-05,
+      "loss": 0.0014,
+      "step": 1039
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.24616444458442913,
+      "learning_rate": 2.2091942547432955e-05,
+      "loss": 0.0061,
+      "step": 1040
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 0.26571080554404275,
+      "learning_rate": 2.193770553940876e-05,
+      "loss": 0.0041,
+      "step": 1041
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 0.12936681612846132,
+      "learning_rate": 2.1783942468224382e-05,
+      "loss": 0.002,
+      "step": 1042
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 0.1468591636277515,
+      "learning_rate": 2.163065426741603e-05,
+      "loss": 0.0018,
+      "step": 1043
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 0.22930625678224004,
+      "learning_rate": 2.147784186763696e-05,
+      "loss": 0.0045,
+      "step": 1044
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 0.11688640761027856,
+      "learning_rate": 2.132550619665168e-05,
+      "loss": 0.0021,
+      "step": 1045
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.2834291902096678,
+      "learning_rate": 2.117364817933033e-05,
+      "loss": 0.0088,
+      "step": 1046
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.09770994113011587,
+      "learning_rate": 2.1022268737643138e-05,
+      "loss": 0.0012,
+      "step": 1047
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.21179115225847997,
+      "learning_rate": 2.08713687906547e-05,
+      "loss": 0.0091,
+      "step": 1048
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.19582537804772682,
+      "learning_rate": 2.0720949254518517e-05,
+      "loss": 0.0061,
+      "step": 1049
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.11467862009629635,
+      "learning_rate": 2.05710110424714e-05,
+      "loss": 0.0053,
+      "step": 1050
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.21350029916775803,
+      "learning_rate": 2.0421555064827878e-05,
+      "loss": 0.0061,
+      "step": 1051
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.20538455731062802,
+      "learning_rate": 2.0272582228974792e-05,
+      "loss": 0.0075,
+      "step": 1052
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.10140087343257974,
+      "learning_rate": 2.012409343936551e-05,
+      "loss": 0.0013,
+      "step": 1053
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.01919076553851994,
+      "learning_rate": 1.9976089597514903e-05,
+      "loss": 0.0008,
+      "step": 1054
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.03218145870583934,
+      "learning_rate": 1.982857160199334e-05,
+      "loss": 0.0008,
+      "step": 1055
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.03739635894441978,
+      "learning_rate": 1.9681540348421623e-05,
+      "loss": 0.0008,
+      "step": 1056
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.13613219870973856,
+      "learning_rate": 1.9534996729465426e-05,
+      "loss": 0.0021,
+      "step": 1057
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.054086187632928044,
+      "learning_rate": 1.938894163482974e-05,
+      "loss": 0.0011,
+      "step": 1058
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.046127873214146596,
+      "learning_rate": 1.9243375951253796e-05,
+      "loss": 0.001,
+      "step": 1059
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.11613815329191052,
+      "learning_rate": 1.9098300562505266e-05,
+      "loss": 0.0025,
+      "step": 1060
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.03726995517623346,
+      "learning_rate": 1.895371634937525e-05,
+      "loss": 0.0008,
+      "step": 1061
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.08420377805157721,
+      "learning_rate": 1.880962418967279e-05,
+      "loss": 0.0016,
+      "step": 1062
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.10452756767726376,
+      "learning_rate": 1.8666024958219408e-05,
+      "loss": 0.0019,
+      "step": 1063
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.01383662809848538,
+      "learning_rate": 1.852291952684414e-05,
+      "loss": 0.0007,
+      "step": 1064
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.022577848984948395,
+      "learning_rate": 1.8380308764377842e-05,
+      "loss": 0.0007,
+      "step": 1065
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.07871408754105688,
+      "learning_rate": 1.8238193536648195e-05,
+      "loss": 0.0011,
+      "step": 1066
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.08545473681241783,
+      "learning_rate": 1.8096574706474333e-05,
+      "loss": 0.0012,
+      "step": 1067
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.1353741514334004,
+      "learning_rate": 1.795545313366166e-05,
+      "loss": 0.0036,
+      "step": 1068
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.12439647561416785,
+      "learning_rate": 1.7814829674996592e-05,
+      "loss": 0.0011,
+      "step": 1069
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.08429751975272234,
+      "learning_rate": 1.767470518424129e-05,
+      "loss": 0.0031,
+      "step": 1070
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.1280720043756421,
+      "learning_rate": 1.7535080512128632e-05,
+      "loss": 0.0013,
+      "step": 1071
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.12450764824136708,
+      "learning_rate": 1.7395956506356937e-05,
+      "loss": 0.0033,
+      "step": 1072
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.02410438266254552,
+      "learning_rate": 1.7257334011584847e-05,
+      "loss": 0.0008,
+      "step": 1073
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.12120251299516387,
+      "learning_rate": 1.7119213869426197e-05,
+      "loss": 0.0015,
+      "step": 1074
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.046851269481934375,
+      "learning_rate": 1.6981596918444953e-05,
+      "loss": 0.0009,
+      "step": 1075
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.019763526247537133,
+      "learning_rate": 1.684448399414994e-05,
+      "loss": 0.0007,
+      "step": 1076
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.03726169930633526,
+      "learning_rate": 1.6707875928990058e-05,
+      "loss": 0.0008,
+      "step": 1077
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.017895000481248813,
+      "learning_rate": 1.6571773552349e-05,
+      "loss": 0.0008,
+      "step": 1078
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.06382125990818506,
+      "learning_rate": 1.6436177690540243e-05,
+      "loss": 0.0014,
+      "step": 1079
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 0.07676229627868801,
+      "learning_rate": 1.630108916680223e-05,
+      "loss": 0.0012,
+      "step": 1080
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 0.14409684160283778,
+      "learning_rate": 1.6166508801293013e-05,
+      "loss": 0.0016,
+      "step": 1081
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 0.1024034052457828,
+      "learning_rate": 1.603243741108571e-05,
+      "loss": 0.005,
+      "step": 1082
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.10474202722991857,
+      "learning_rate": 1.5898875810163137e-05,
+      "loss": 0.0026,
+      "step": 1083
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.02073090773880473,
+      "learning_rate": 1.5765824809413056e-05,
+      "loss": 0.0008,
+      "step": 1084
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.029262793325674667,
+      "learning_rate": 1.5633285216623385e-05,
+      "loss": 0.001,
+      "step": 1085
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.05250326546898917,
+      "learning_rate": 1.5501257836476978e-05,
+      "loss": 0.0009,
+      "step": 1086
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.08895119486604487,
+      "learning_rate": 1.5369743470547027e-05,
+      "loss": 0.0011,
+      "step": 1087
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.03674723337027439,
+      "learning_rate": 1.5238742917292015e-05,
+      "loss": 0.0009,
+      "step": 1088
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.10396161935118733,
+      "learning_rate": 1.5108256972050972e-05,
+      "loss": 0.0017,
+      "step": 1089
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.2002544848798124,
+      "learning_rate": 1.4978286427038601e-05,
+      "loss": 0.002,
+      "step": 1090
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.1435954143257352,
+      "learning_rate": 1.4848832071340423e-05,
+      "loss": 0.0015,
+      "step": 1091
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.12829196489598618,
+      "learning_rate": 1.4719894690908098e-05,
+      "loss": 0.0016,
+      "step": 1092
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.0913935380506535,
+      "learning_rate": 1.4591475068554572e-05,
+      "loss": 0.0019,
+      "step": 1093
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.12299032858319592,
+      "learning_rate": 1.4463573983949341e-05,
+      "loss": 0.0022,
+      "step": 1094
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.08199866482225956,
+      "learning_rate": 1.4336192213613742e-05,
+      "loss": 0.0013,
+      "step": 1095
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.19204946938484016,
+      "learning_rate": 1.4209330530916165e-05,
+      "loss": 0.0028,
+      "step": 1096
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.014720881385259075,
+      "learning_rate": 1.4082989706067461e-05,
+      "loss": 0.0007,
+      "step": 1097
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.018505393949536906,
+      "learning_rate": 1.3957170506116201e-05,
+      "loss": 0.0007,
+      "step": 1098
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 0.01384413643030154,
+      "learning_rate": 1.3831873694944031e-05,
+      "loss": 0.0007,
+      "step": 1099
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 0.036346693638824895,
+      "learning_rate": 1.3707100033261034e-05,
+      "loss": 0.0008,
+      "step": 1100
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.01765844361926089,
+      "learning_rate": 1.3582850278601134e-05,
+      "loss": 0.0007,
+      "step": 1101
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.04518964321008793,
+      "learning_rate": 1.3459125185317434e-05,
+      "loss": 0.001,
+      "step": 1102
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.10733551347246247,
+      "learning_rate": 1.3335925504577717e-05,
+      "loss": 0.002,
+      "step": 1103
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.023185767020048136,
+      "learning_rate": 1.3213251984359831e-05,
+      "loss": 0.0008,
+      "step": 1104
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.08880950865679069,
+      "learning_rate": 1.3091105369447165e-05,
+      "loss": 0.0011,
+      "step": 1105
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 0.1083682533280248,
+      "learning_rate": 1.2969486401424169e-05,
+      "loss": 0.002,
+      "step": 1106
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 0.026544746900945333,
+      "learning_rate": 1.2848395818671687e-05,
+      "loss": 0.0008,
+      "step": 1107
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 0.11970788222992348,
+      "learning_rate": 1.2727834356362778e-05,
+      "loss": 0.0028,
+      "step": 1108
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 0.03716895455467463,
+      "learning_rate": 1.2607802746457897e-05,
+      "loss": 0.0009,
+      "step": 1109
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 0.030824964937963995,
+      "learning_rate": 1.2488301717700735e-05,
+      "loss": 0.0008,
+      "step": 1110
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 0.06854610979395971,
+      "learning_rate": 1.2369331995613665e-05,
+      "loss": 0.0015,
+      "step": 1111
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.06501895775582617,
+      "learning_rate": 1.2250894302493265e-05,
+      "loss": 0.001,
+      "step": 1112
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.05234778746681681,
+      "learning_rate": 1.21329893574062e-05,
+      "loss": 0.001,
+      "step": 1113
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.03259327320159292,
+      "learning_rate": 1.2015617876184527e-05,
+      "loss": 0.0008,
+      "step": 1114
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.015279019434158323,
+      "learning_rate": 1.1898780571421552e-05,
+      "loss": 0.0007,
+      "step": 1115
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.07246380911324947,
+      "learning_rate": 1.1782478152467502e-05,
+      "loss": 0.001,
+      "step": 1116
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.02273785085702488,
+      "learning_rate": 1.166671132542505e-05,
+      "loss": 0.0007,
+      "step": 1117
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.1035323351629705,
+      "learning_rate": 1.1551480793145331e-05,
+      "loss": 0.0015,
+      "step": 1118
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.05127935195746323,
+      "learning_rate": 1.1436787255223302e-05,
+      "loss": 0.0008,
+      "step": 1119
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.016800993811507427,
+      "learning_rate": 1.1322631407993811e-05,
+      "loss": 0.0007,
+      "step": 1120
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.024119629962673438,
+      "learning_rate": 1.1209013944527203e-05,
+      "loss": 0.0007,
+      "step": 1121
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.010667077214656819,
+      "learning_rate": 1.1095935554625148e-05,
+      "loss": 0.0006,
+      "step": 1122
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.09773204278110174,
+      "learning_rate": 1.098339692481648e-05,
+      "loss": 0.0029,
+      "step": 1123
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.24316870994224807,
+      "learning_rate": 1.0871398738352955e-05,
+      "loss": 0.0022,
+      "step": 1124
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 0.023028795994797544,
+      "learning_rate": 1.0759941675205221e-05,
+      "loss": 0.0008,
+      "step": 1125
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 0.03889689044698488,
+      "learning_rate": 1.0649026412058583e-05,
+      "loss": 0.0009,
+      "step": 1126
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.04029309332427343,
+      "learning_rate": 1.0538653622308948e-05,
+      "loss": 0.0008,
+      "step": 1127
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.015697786114817317,
+      "learning_rate": 1.042882397605871e-05,
+      "loss": 0.0006,
+      "step": 1128
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.0567106388225468,
+      "learning_rate": 1.0319538140112728e-05,
+      "loss": 0.0016,
+      "step": 1129
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.1784512351442451,
+      "learning_rate": 1.0210796777974197e-05,
+      "loss": 0.0023,
+      "step": 1130
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.08772494322055742,
+      "learning_rate": 1.0102600549840701e-05,
+      "loss": 0.0009,
+      "step": 1131
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.20955932841208275,
+      "learning_rate": 9.994950112600154e-06,
+      "loss": 0.002,
+      "step": 1132
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 0.14727917652607253,
+      "learning_rate": 9.887846119826849e-06,
+      "loss": 0.0014,
+      "step": 1133
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 0.12857837430265331,
+      "learning_rate": 9.781289221777478e-06,
+      "loss": 0.0011,
+      "step": 1134
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.08938293238751545,
+      "learning_rate": 9.675280065387116e-06,
+      "loss": 0.0021,
+      "step": 1135
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.050856917690314946,
+      "learning_rate": 9.569819294265414e-06,
+      "loss": 0.0008,
+      "step": 1136
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.06534382410281618,
+      "learning_rate": 9.464907548692614e-06,
+      "loss": 0.0028,
+      "step": 1137
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.0332557205580419,
+      "learning_rate": 9.360545465615667e-06,
+      "loss": 0.0008,
+      "step": 1138
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.07027197782202682,
+      "learning_rate": 9.256733678644414e-06,
+      "loss": 0.0011,
+      "step": 1139
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.03885998911344652,
+      "learning_rate": 9.153472818047625e-06,
+      "loss": 0.0008,
+      "step": 1140
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.017220404269721257,
+      "learning_rate": 9.05076351074936e-06,
+      "loss": 0.0006,
+      "step": 1141
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.0966940128891765,
+      "learning_rate": 8.948606380324941e-06,
+      "loss": 0.0011,
+      "step": 1142
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.03136176008776217,
+      "learning_rate": 8.847002046997354e-06,
+      "loss": 0.0007,
+      "step": 1143
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.0701507716252104,
+      "learning_rate": 8.745951127633411e-06,
+      "loss": 0.001,
+      "step": 1144
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.022917394797858304,
+      "learning_rate": 8.645454235739903e-06,
+      "loss": 0.0008,
+      "step": 1145
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.050858668069693864,
+      "learning_rate": 8.54551198146013e-06,
+      "loss": 0.0009,
+      "step": 1146
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.011726337626123934,
+      "learning_rate": 8.44612497156989e-06,
+      "loss": 0.0006,
+      "step": 1147
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.01978894522286886,
+      "learning_rate": 8.347293809474054e-06,
+      "loss": 0.0007,
+      "step": 1148
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.08302124640943384,
+      "learning_rate": 8.249019095202736e-06,
+      "loss": 0.0015,
+      "step": 1149
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.05186518424971129,
+      "learning_rate": 8.151301425407699e-06,
+      "loss": 0.0007,
+      "step": 1150
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.15471092053538033,
+      "learning_rate": 8.054141393358816e-06,
+      "loss": 0.0018,
+      "step": 1151
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.06084785437540428,
+      "learning_rate": 7.957539588940299e-06,
+      "loss": 0.001,
+      "step": 1152
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.026927491764069434,
+      "learning_rate": 7.861496598647278e-06,
+      "loss": 0.0007,
+      "step": 1153
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 0.1048450022792098,
+      "learning_rate": 7.76601300558214e-06,
+      "loss": 0.0031,
+      "step": 1154
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 0.023020896230008622,
+      "learning_rate": 7.671089389451058e-06,
+      "loss": 0.0007,
+      "step": 1155
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.06283017596763896,
+      "learning_rate": 7.576726326560424e-06,
+      "loss": 0.0013,
+      "step": 1156
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.014730074518723359,
+      "learning_rate": 7.482924389813317e-06,
+      "loss": 0.0007,
+      "step": 1157
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.028101057337962233,
+      "learning_rate": 7.389684148706122e-06,
+      "loss": 0.0008,
+      "step": 1158
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.05285940970870937,
+      "learning_rate": 7.2970061693250154e-06,
+      "loss": 0.001,
+      "step": 1159
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.13191030206091162,
+      "learning_rate": 7.204891014342552e-06,
+      "loss": 0.0033,
+      "step": 1160
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.02356702168837524,
+      "learning_rate": 7.113339243014139e-06,
+      "loss": 0.0008,
+      "step": 1161
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.0818664659307645,
+      "learning_rate": 7.022351411174866e-06,
+      "loss": 0.0019,
+      "step": 1162
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.07673173856957743,
+      "learning_rate": 6.931928071235894e-06,
+      "loss": 0.0017,
+      "step": 1163
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.05858257501923769,
+      "learning_rate": 6.842069772181236e-06,
+      "loss": 0.001,
+      "step": 1164
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.05475450894175683,
+      "learning_rate": 6.75277705956443e-06,
+      "loss": 0.0011,
+      "step": 1165
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.11394691130455702,
+      "learning_rate": 6.664050475505101e-06,
+      "loss": 0.001,
+      "step": 1166
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.08548426384173068,
+      "learning_rate": 6.575890558685882e-06,
+      "loss": 0.0013,
+      "step": 1167
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.017118906762927648,
+      "learning_rate": 6.48829784434889e-06,
+      "loss": 0.0007,
+      "step": 1168
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.10346357029151644,
+      "learning_rate": 6.4012728642926845e-06,
+      "loss": 0.0017,
+      "step": 1169
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 0.03887979797166313,
+      "learning_rate": 6.314816146868952e-06,
+      "loss": 0.0008,
+      "step": 1170
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 0.19091091665474808,
+      "learning_rate": 6.228928216979257e-06,
+      "loss": 0.0038,
+      "step": 1171
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.013500867024626814,
+      "learning_rate": 6.143609596072008e-06,
+      "loss": 0.0006,
+      "step": 1172
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.203206132798381,
+      "learning_rate": 6.0588608021390655e-06,
+      "loss": 0.0013,
+      "step": 1173
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.09932338220339965,
+      "learning_rate": 5.97468234971279e-06,
+      "loss": 0.0011,
+      "step": 1174
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.018530307689402132,
+      "learning_rate": 5.891074749862857e-06,
+      "loss": 0.0006,
+      "step": 1175
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.022634026716283052,
+      "learning_rate": 5.80803851019307e-06,
+      "loss": 0.0006,
+      "step": 1176
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.030076579884974423,
+      "learning_rate": 5.725574134838474e-06,
+      "loss": 0.0009,
+      "step": 1177
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.053602431560927276,
+      "learning_rate": 5.643682124462057e-06,
+      "loss": 0.001,
+      "step": 1178
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.13540720692271374,
+      "learning_rate": 5.562362976251901e-06,
+      "loss": 0.0032,
+      "step": 1179
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.049268095531138596,
+      "learning_rate": 5.481617183918053e-06,
+      "loss": 0.001,
+      "step": 1180
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.14435454285355126,
+      "learning_rate": 5.401445237689573e-06,
+      "loss": 0.0011,
+      "step": 1181
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.0500720841341082,
+      "learning_rate": 5.321847624311593e-06,
+      "loss": 0.0011,
+      "step": 1182
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.10757510934121438,
+      "learning_rate": 5.242824827042237e-06,
+      "loss": 0.0013,
+      "step": 1183
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.03921020553142721,
+      "learning_rate": 5.1643773256498164e-06,
+      "loss": 0.0007,
+      "step": 1184
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 0.06759469016444933,
+      "learning_rate": 5.086505596409885e-06,
+      "loss": 0.0055,
+      "step": 1185
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 0.13664786763659992,
+      "learning_rate": 5.009210112102292e-06,
+      "loss": 0.002,
+      "step": 1186
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 0.15003565816164469,
+      "learning_rate": 4.932491342008383e-06,
+      "loss": 0.0019,
+      "step": 1187
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.10973268691019623,
+      "learning_rate": 4.856349751908107e-06,
+      "loss": 0.0022,
+      "step": 1188
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.012812765317198193,
+      "learning_rate": 4.7807858040771924e-06,
+      "loss": 0.0006,
+      "step": 1189
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.02090168761288768,
+      "learning_rate": 4.705799957284351e-06,
+      "loss": 0.0006,
+      "step": 1190
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.03913651192715962,
+      "learning_rate": 4.6313926667885035e-06,
+      "loss": 0.0008,
+      "step": 1191
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.01822403460355446,
+      "learning_rate": 4.557564384335977e-06,
+      "loss": 0.0007,
+      "step": 1192
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.02767231226389733,
+      "learning_rate": 4.4843155581578474e-06,
+      "loss": 0.0008,
+      "step": 1193
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.08509749300948648,
+      "learning_rate": 4.411646632967059e-06,
+      "loss": 0.001,
+      "step": 1194
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.12718622349967054,
+      "learning_rate": 4.339558049955927e-06,
+      "loss": 0.0015,
+      "step": 1195
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.11791588852754038,
+      "learning_rate": 4.268050246793276e-06,
+      "loss": 0.0066,
+      "step": 1196
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.09709032334450204,
+      "learning_rate": 4.197123657621915e-06,
+      "loss": 0.0011,
+      "step": 1197
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 0.13311027377720236,
+      "learning_rate": 4.126778713055923e-06,
+      "loss": 0.0017,
+      "step": 1198
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 0.014722359842611643,
+      "learning_rate": 4.0570158401780465e-06,
+      "loss": 0.0006,
+      "step": 1199
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 0.06724692866367353,
+      "learning_rate": 3.987835462537193e-06,
+      "loss": 0.0009,
+      "step": 1200
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.18749678465885222,
+      "learning_rate": 3.919238000145687e-06,
+      "loss": 0.0025,
+      "step": 1201
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.02711478710481096,
+      "learning_rate": 3.8512238694768835e-06,
+      "loss": 0.0008,
+      "step": 1202
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.15344305249791065,
+      "learning_rate": 3.783793483462583e-06,
+      "loss": 0.0016,
+      "step": 1203
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.024577349084646084,
+      "learning_rate": 3.7169472514904544e-06,
+      "loss": 0.0007,
+      "step": 1204
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.017793645227127574,
+      "learning_rate": 3.6506855794016913e-06,
+      "loss": 0.0007,
+      "step": 1205
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.021060684859901417,
+      "learning_rate": 3.5850088694884156e-06,
+      "loss": 0.0007,
+      "step": 1206
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.01298124449888585,
+      "learning_rate": 3.5199175204913117e-06,
+      "loss": 0.0006,
+      "step": 1207
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.10120842722667384,
+      "learning_rate": 3.455411927597185e-06,
+      "loss": 0.0018,
+      "step": 1208
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.03487705779457792,
+      "learning_rate": 3.3914924824365422e-06,
+      "loss": 0.0007,
+      "step": 1209
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.029008071748288324,
+      "learning_rate": 3.3281595730812575e-06,
+      "loss": 0.0007,
+      "step": 1210
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.0322519541930152,
+      "learning_rate": 3.265413584042165e-06,
+      "loss": 0.0008,
+      "step": 1211
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.022006655515884467,
+      "learning_rate": 3.203254896266761e-06,
+      "loss": 0.0007,
+      "step": 1212
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.028931490349953615,
+      "learning_rate": 3.1416838871368924e-06,
+      "loss": 0.0008,
+      "step": 1213
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.10097559997911604,
+      "learning_rate": 3.080700930466429e-06,
+      "loss": 0.0016,
+      "step": 1214
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.0783156199159443,
+      "learning_rate": 3.0203063964990617e-06,
+      "loss": 0.0012,
+      "step": 1215
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.03423453699339226,
+      "learning_rate": 2.9605006519059507e-06,
+      "loss": 0.0008,
+      "step": 1216
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.05255356449793071,
+      "learning_rate": 2.9012840597836045e-06,
+      "loss": 0.0017,
+      "step": 1217
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.15091037601685472,
+      "learning_rate": 2.8426569796516146e-06,
+      "loss": 0.0014,
+      "step": 1218
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.028357881496545626,
+      "learning_rate": 2.7846197674504913e-06,
+      "loss": 0.0007,
+      "step": 1219
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.11315719752635794,
+      "learning_rate": 2.7271727755395214e-06,
+      "loss": 0.0021,
+      "step": 1220
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.08798526393143069,
+      "learning_rate": 2.6703163526945794e-06,
+      "loss": 0.0013,
+      "step": 1221
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 0.026352572645490915,
+      "learning_rate": 2.614050844106042e-06,
+      "loss": 0.0007,
+      "step": 1222
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 0.03567197006948489,
+      "learning_rate": 2.5583765913766987e-06,
+      "loss": 0.0008,
+      "step": 1223
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 0.02168815624645082,
+      "learning_rate": 2.503293932519668e-06,
+      "loss": 0.0008,
+      "step": 1224
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 0.0872434290107731,
+      "learning_rate": 2.4488032019563402e-06,
+      "loss": 0.0037,
+      "step": 1225
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 0.12548586283201357,
+      "learning_rate": 2.394904730514358e-06,
+      "loss": 0.0013,
+      "step": 1226
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.13838774983602323,
+      "learning_rate": 2.3415988454255855e-06,
+      "loss": 0.003,
+      "step": 1227
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.018445083621635758,
+      "learning_rate": 2.2888858703241866e-06,
+      "loss": 0.0006,
+      "step": 1228
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.1396676556275959,
+      "learning_rate": 2.236766125244549e-06,
+      "loss": 0.0026,
+      "step": 1229
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 0.021233616634377974,
+      "learning_rate": 2.1852399266194314e-06,
+      "loss": 0.0007,
+      "step": 1230
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 0.10351195169315834,
+      "learning_rate": 2.1343075872780396e-06,
+      "loss": 0.0021,
+      "step": 1231
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 0.06293925964774824,
+      "learning_rate": 2.0839694164440425e-06,
+      "loss": 0.001,
+      "step": 1232
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 0.015730748883720166,
+      "learning_rate": 2.03422571973384e-06,
+      "loss": 0.0007,
+      "step": 1233
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 0.035928677992341104,
+      "learning_rate": 1.985076799154528e-06,
+      "loss": 0.0008,
+      "step": 1234
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 0.07636815076824766,
+      "learning_rate": 1.9365229531022264e-06,
+      "loss": 0.0009,
+      "step": 1235
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 0.04640569562427641,
+      "learning_rate": 1.8885644763601774e-06,
+      "loss": 0.0011,
+      "step": 1236
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 0.03086735995397519,
+      "learning_rate": 1.8412016600969695e-06,
+      "loss": 0.0007,
+      "step": 1237
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 0.016452317137242538,
+      "learning_rate": 1.7944347918648185e-06,
+      "loss": 0.0007,
+      "step": 1238
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 0.030068581883872736,
+      "learning_rate": 1.748264155597712e-06,
+      "loss": 0.0008,
+      "step": 1239
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 0.023978134292451384,
+      "learning_rate": 1.7026900316098215e-06,
+      "loss": 0.0007,
+      "step": 1240
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 0.03424361711767579,
+      "learning_rate": 1.657712696593705e-06,
+      "loss": 0.0009,
+      "step": 1241
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 0.01392480564301655,
+      "learning_rate": 1.6133324236186742e-06,
+      "loss": 0.0006,
+      "step": 1242
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 0.18774342633677002,
+      "learning_rate": 1.5695494821290735e-06,
+      "loss": 0.0033,
+      "step": 1243
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 0.021531321530652542,
+      "learning_rate": 1.5263641379427595e-06,
+      "loss": 0.0007,
+      "step": 1244
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 0.03724728912473665,
+      "learning_rate": 1.4837766532493468e-06,
+      "loss": 0.0009,
+      "step": 1245
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 0.010372650892936815,
+      "learning_rate": 1.4417872866087534e-06,
+      "loss": 0.0005,
+      "step": 1246
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 0.11171948389703704,
+      "learning_rate": 1.400396292949513e-06,
+      "loss": 0.0016,
+      "step": 1247
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.020327483983931985,
+      "learning_rate": 1.3596039235672874e-06,
+      "loss": 0.0006,
+      "step": 1248
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.023877471746245795,
+      "learning_rate": 1.3194104261233798e-06,
+      "loss": 0.0007,
+      "step": 1249
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.10971201191562108,
+      "learning_rate": 1.2798160446431006e-06,
+      "loss": 0.0014,
+      "step": 1250
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 0.08883583735718313,
+      "learning_rate": 1.2408210195144376e-06,
+      "loss": 0.0009,
+      "step": 1251
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 0.14775663281036278,
+      "learning_rate": 1.2024255874865108e-06,
+      "loss": 0.0018,
+      "step": 1252
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 0.07076694182371907,
+      "learning_rate": 1.1646299816681195e-06,
+      "loss": 0.0023,
+      "step": 1253
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 0.07816028502462304,
+      "learning_rate": 1.1274344315264196e-06,
+      "loss": 0.0019,
+      "step": 1254
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 0.14809768533907064,
+      "learning_rate": 1.0908391628854041e-06,
+      "loss": 0.0066,
+      "step": 1255
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 0.09661952428836978,
+      "learning_rate": 1.0548443979246481e-06,
+      "loss": 0.0013,
+      "step": 1256
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 0.10290659291537475,
+      "learning_rate": 1.0194503551778866e-06,
+      "loss": 0.0013,
+      "step": 1257
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 0.020365140491938917,
+      "learning_rate": 9.846572495316952e-07,
+      "loss": 0.0007,
+      "step": 1258
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 0.04336920909838203,
+      "learning_rate": 9.504652922242562e-07,
+      "loss": 0.0008,
+      "step": 1259
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 0.08591413241609115,
+      "learning_rate": 9.168746908439718e-07,
+      "loss": 0.0024,
+      "step": 1260
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 0.012560733981925342,
+      "learning_rate": 8.838856493282754e-07,
+      "loss": 0.0006,
+      "step": 1261
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.03008555789324348,
+      "learning_rate": 8.514983679623556e-07,
+      "loss": 0.0007,
+      "step": 1262
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.10052078557617806,
+      "learning_rate": 8.197130433779565e-07,
+      "loss": 0.0019,
+      "step": 1263
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 0.10075177742486086,
+      "learning_rate": 7.885298685522235e-07,
+      "loss": 0.0019,
+      "step": 1264
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 0.07691576123408798,
+      "learning_rate": 7.579490328064265e-07,
+      "loss": 0.0019,
+      "step": 1265
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 0.011988181730916309,
+      "learning_rate": 7.27970721804927e-07,
+      "loss": 0.0006,
+      "step": 1266
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 0.059827650653866,
+      "learning_rate": 6.985951175539796e-07,
+      "loss": 0.0009,
+      "step": 1267
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 0.18626835726566712,
+      "learning_rate": 6.698223984006436e-07,
+      "loss": 0.0011,
+      "step": 1268
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 0.017337958889351607,
+      "learning_rate": 6.416527390317173e-07,
+      "loss": 0.0006,
+      "step": 1269
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 0.08562635575051426,
+      "learning_rate": 6.140863104726391e-07,
+      "loss": 0.0025,
+      "step": 1270
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 0.17579938063234693,
+      "learning_rate": 5.87123280086499e-07,
+      "loss": 0.005,
+      "step": 1271
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 0.11891100432575016,
+      "learning_rate": 5.607638115729841e-07,
+      "loss": 0.0015,
+      "step": 1272
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 0.02350445020180528,
+      "learning_rate": 5.350080649674127e-07,
+      "loss": 0.0008,
+      "step": 1273
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 0.14229582193472826,
+      "learning_rate": 5.098561966397131e-07,
+      "loss": 0.0014,
+      "step": 1274
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 0.021076889914199266,
+      "learning_rate": 4.85308359293557e-07,
+      "loss": 0.0007,
+      "step": 1275
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 0.027962840089651737,
+      "learning_rate": 4.613647019653389e-07,
+      "loss": 0.0007,
+      "step": 1276
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 0.023540717007848738,
+      "learning_rate": 4.3802537002335386e-07,
+      "loss": 0.0007,
+      "step": 1277
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 0.07783290011045198,
+      "learning_rate": 4.152905051668321e-07,
+      "loss": 0.0021,
+      "step": 1278
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 0.1477536814845743,
+      "learning_rate": 3.931602454251837e-07,
+      "loss": 0.0024,
+      "step": 1279
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 0.05415129686700728,
+      "learning_rate": 3.716347251570551e-07,
+      "loss": 0.0009,
+      "step": 1280
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 0.10688029228040345,
+      "learning_rate": 3.50714075049563e-07,
+      "loss": 0.0014,
+      "step": 1281
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 0.011211447437895239,
+      "learning_rate": 3.3039842211752824e-07,
+      "loss": 0.0006,
+      "step": 1282
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 0.034187950445730836,
+      "learning_rate": 3.106878897026544e-07,
+      "loss": 0.0008,
+      "step": 1283
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 0.010258436895645257,
+      "learning_rate": 2.915825974727726e-07,
+      "loss": 0.0006,
+      "step": 1284
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 0.11601700752977635,
+      "learning_rate": 2.7308266142119785e-07,
+      "loss": 0.0016,
+      "step": 1285
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 0.15218388627244026,
+      "learning_rate": 2.5518819386590734e-07,
+      "loss": 0.0015,
+      "step": 1286
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 0.07445685913573166,
+      "learning_rate": 2.3789930344897404e-07,
+      "loss": 0.0018,
+      "step": 1287
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.030128524156390737,
+      "learning_rate": 2.212160951358011e-07,
+      "loss": 0.0008,
+      "step": 1288
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.018684805487496823,
+      "learning_rate": 2.0513867021457744e-07,
+      "loss": 0.0007,
+      "step": 1289
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.031701118491831055,
+      "learning_rate": 1.8966712629558957e-07,
+      "loss": 0.0008,
+      "step": 1290
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 0.026168419899834938,
+      "learning_rate": 1.748015573106887e-07,
+      "loss": 0.0007,
+      "step": 1291
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 0.21056544758807397,
+      "learning_rate": 1.6054205351265784e-07,
+      "loss": 0.0069,
+      "step": 1292
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 0.19522623097769917,
+      "learning_rate": 1.4688870147473443e-07,
+      "loss": 0.0032,
+      "step": 1293
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 0.059157612594969565,
+      "learning_rate": 1.338415840900109e-07,
+      "loss": 0.0007,
+      "step": 1294
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 0.053940392355256016,
+      "learning_rate": 1.2140078057101266e-07,
+      "loss": 0.0009,
+      "step": 1295
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 0.05247469755440444,
+      "learning_rate": 1.0956636644912088e-07,
+      "loss": 0.001,
+      "step": 1296
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 0.022720476558229258,
+      "learning_rate": 9.833841357421714e-08,
+      "loss": 0.0007,
+      "step": 1297
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 0.21363185980192992,
+      "learning_rate": 8.771699011416168e-08,
+      "loss": 0.0038,
+      "step": 1298
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 0.024773998866581916,
+      "learning_rate": 7.770216055443814e-08,
+      "loss": 0.0008,
+      "step": 1299
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 0.021976256993898078,
+      "learning_rate": 6.829398569770939e-08,
+      "loss": 0.0006,
+      "step": 1300
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 0.023154625605837812,
+      "learning_rate": 5.94925226635068e-08,
+      "loss": 0.0008,
+      "step": 1301
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 0.06416291587351453,
+      "learning_rate": 5.1297824887841516e-08,
+      "loss": 0.0026,
+      "step": 1302
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 0.011642582291543501,
+      "learning_rate": 4.370994212287149e-08,
+      "loss": 0.0006,
+      "step": 1303
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.047289961192521615,
+      "learning_rate": 3.672892043666831e-08,
+      "loss": 0.0011,
+      "step": 1304
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.061956823412063274,
+      "learning_rate": 3.0354802212839705e-08,
+      "loss": 0.0009,
+      "step": 1305
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 0.06736415132709304,
+      "learning_rate": 2.4587626150351926e-08,
+      "loss": 0.0007,
+      "step": 1306
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 0.05022755639062635,
+      "learning_rate": 1.94274272632633e-08,
+      "loss": 0.0013,
+      "step": 1307
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 0.061308536967494456,
+      "learning_rate": 1.4874236880491055e-08,
+      "loss": 0.0012,
+      "step": 1308
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 0.022252197910701644,
+      "learning_rate": 1.0928082645667025e-08,
+      "loss": 0.0007,
+      "step": 1309
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 0.17108281033359857,
+      "learning_rate": 7.58898851693779e-09,
+      "loss": 0.001,
+      "step": 1310
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 0.050067756307481906,
+      "learning_rate": 4.856974766831446e-09,
+      "loss": 0.001,
+      "step": 1311
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 0.01600682663664517,
+      "learning_rate": 2.732057982124392e-09,
+      "loss": 0.0006,
+      "step": 1312
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 0.04081956925949677,
+      "learning_rate": 1.2142510637414006e-09,
+      "loss": 0.0007,
+      "step": 1313
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.09239950302890293,
+      "learning_rate": 3.035632266890076e-10,
+      "loss": 0.0015,
+      "step": 1314
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.03247669105406384,
+      "learning_rate": 0.0,
+      "loss": 0.0008,
+      "step": 1315
+    },
+    {
+      "epoch": 5.0,
+      "step": 1315,
+      "total_flos": 1.7248353720152883e+17,
+      "train_loss": 0.19667112988795193,
+      "train_runtime": 4217.7385,
+      "train_samples_per_second": 79.758,
+      "train_steps_per_second": 0.312
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1315,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 1000,
+  "total_flos": 1.7248353720152883e+17,
+  "train_batch_size": 64,
+  "trial_name": null,
+  "trial_params": null
+}