diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5187 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.985163204747774,
+  "eval_steps": 500,
+  "global_step": 735,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.006782534972445952,
+      "grad_norm": 7.135720261444181,
+      "learning_rate": 1.0810810810810812e-06,
+      "loss": 1.1988,
+      "step": 1
+    },
+    {
+      "epoch": 0.013565069944891903,
+      "grad_norm": 7.10061152713567,
+      "learning_rate": 2.1621621621621623e-06,
+      "loss": 1.1926,
+      "step": 2
+    },
+    {
+      "epoch": 0.020347604917337857,
+      "grad_norm": 7.012644871929241,
+      "learning_rate": 3.2432432432432437e-06,
+      "loss": 1.1873,
+      "step": 3
+    },
+    {
+      "epoch": 0.027130139889783807,
+      "grad_norm": 6.4062690637507425,
+      "learning_rate": 4.324324324324325e-06,
+      "loss": 1.1664,
+      "step": 4
+    },
+    {
+      "epoch": 0.03391267486222976,
+      "grad_norm": 4.819533337520507,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 1.1316,
+      "step": 5
+    },
+    {
+      "epoch": 0.040695209834675714,
+      "grad_norm": 3.0815069633752463,
+      "learning_rate": 6.486486486486487e-06,
+      "loss": 1.0897,
+      "step": 6
+    },
+    {
+      "epoch": 0.04747774480712166,
+      "grad_norm": 4.965671310290254,
+      "learning_rate": 7.567567567567569e-06,
+      "loss": 1.0541,
+      "step": 7
+    },
+    {
+      "epoch": 0.05426027977956761,
+      "grad_norm": 5.523597681999487,
+      "learning_rate": 8.64864864864865e-06,
+      "loss": 1.0812,
+      "step": 8
+    },
+    {
+      "epoch": 0.06104281475201356,
+      "grad_norm": 4.7758478538296245,
+      "learning_rate": 9.729729729729732e-06,
+      "loss": 1.0615,
+      "step": 9
+    },
+    {
+      "epoch": 0.06782534972445951,
+      "grad_norm": 4.37471758206583,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 1.0034,
+      "step": 10
+    },
+    {
+      "epoch": 0.07460788469690546,
+      "grad_norm": 3.415263530374433,
+      "learning_rate": 1.1891891891891894e-05,
+      "loss": 0.9965,
+      "step": 11
+    },
+    {
+      "epoch": 0.08139041966935143,
+      "grad_norm": 2.1249671206986167,
+      "learning_rate": 1.2972972972972975e-05,
+      "loss": 0.9672,
+      "step": 12
+    },
+    {
+      "epoch": 0.08817295464179738,
+      "grad_norm": 2.804510150071328,
+      "learning_rate": 1.4054054054054055e-05,
+      "loss": 0.9248,
+      "step": 13
+    },
+    {
+      "epoch": 0.09495548961424333,
+      "grad_norm": 2.227059944094318,
+      "learning_rate": 1.5135135135135138e-05,
+      "loss": 0.9174,
+      "step": 14
+    },
+    {
+      "epoch": 0.10173802458668928,
+      "grad_norm": 1.9376154207127252,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 0.9111,
+      "step": 15
+    },
+    {
+      "epoch": 0.10852055955913523,
+      "grad_norm": 1.5867898298795111,
+      "learning_rate": 1.72972972972973e-05,
+      "loss": 0.8742,
+      "step": 16
+    },
+    {
+      "epoch": 0.11530309453158118,
+      "grad_norm": 1.824405438277367,
+      "learning_rate": 1.8378378378378383e-05,
+      "loss": 0.8663,
+      "step": 17
+    },
+    {
+      "epoch": 0.12208562950402713,
+      "grad_norm": 1.2484939184792223,
+      "learning_rate": 1.9459459459459463e-05,
+      "loss": 0.8685,
+      "step": 18
+    },
+    {
+      "epoch": 0.1288681644764731,
+      "grad_norm": 1.6950608297204033,
+      "learning_rate": 2.054054054054054e-05,
+      "loss": 0.8643,
+      "step": 19
+    },
+    {
+      "epoch": 0.13565069944891903,
+      "grad_norm": 1.4257834623155268,
+      "learning_rate": 2.1621621621621624e-05,
+      "loss": 0.8589,
+      "step": 20
+    },
+    {
+      "epoch": 0.142433234421365,
+      "grad_norm": 1.2788529677027458,
+      "learning_rate": 2.2702702702702705e-05,
+      "loss": 0.832,
+      "step": 21
+    },
+    {
+      "epoch": 0.14921576939381093,
+      "grad_norm": 1.4653330199646428,
+      "learning_rate": 2.378378378378379e-05,
+      "loss": 0.8257,
+      "step": 22
+    },
+    {
+      "epoch": 0.1559983043662569,
+      "grad_norm": 1.1706491825282224,
+      "learning_rate": 2.4864864864864866e-05,
+      "loss": 0.8154,
+      "step": 23
+    },
+    {
+      "epoch": 0.16278083933870285,
+      "grad_norm": 1.575724350793329,
+      "learning_rate": 2.594594594594595e-05,
+      "loss": 0.83,
+      "step": 24
+    },
+    {
+      "epoch": 0.1695633743111488,
+      "grad_norm": 1.273387962120446,
+      "learning_rate": 2.702702702702703e-05,
+      "loss": 0.8327,
+      "step": 25
+    },
+    {
+      "epoch": 0.17634590928359475,
+      "grad_norm": 1.5867731702030288,
+      "learning_rate": 2.810810810810811e-05,
+      "loss": 0.8204,
+      "step": 26
+    },
+    {
+      "epoch": 0.1831284442560407,
+      "grad_norm": 0.8564117760497798,
+      "learning_rate": 2.918918918918919e-05,
+      "loss": 0.8056,
+      "step": 27
+    },
+    {
+      "epoch": 0.18991097922848665,
+      "grad_norm": 1.4715395766174262,
+      "learning_rate": 3.0270270270270275e-05,
+      "loss": 0.8055,
+      "step": 28
+    },
+    {
+      "epoch": 0.1966935142009326,
+      "grad_norm": 0.8375306490548892,
+      "learning_rate": 3.135135135135135e-05,
+      "loss": 0.7937,
+      "step": 29
+    },
+    {
+      "epoch": 0.20347604917337855,
+      "grad_norm": 1.620124215802785,
+      "learning_rate": 3.2432432432432436e-05,
+      "loss": 0.8001,
+      "step": 30
+    },
+    {
+      "epoch": 0.2102585841458245,
+      "grad_norm": 1.553317483743022,
+      "learning_rate": 3.351351351351351e-05,
+      "loss": 0.7876,
+      "step": 31
+    },
+    {
+      "epoch": 0.21704111911827045,
+      "grad_norm": 1.658909246335399,
+      "learning_rate": 3.45945945945946e-05,
+      "loss": 0.7891,
+      "step": 32
+    },
+    {
+      "epoch": 0.22382365409071642,
+      "grad_norm": 1.1987131034054377,
+      "learning_rate": 3.567567567567568e-05,
+      "loss": 0.7979,
+      "step": 33
+    },
+    {
+      "epoch": 0.23060618906316235,
+      "grad_norm": 2.2091757491329065,
+      "learning_rate": 3.6756756756756765e-05,
+      "loss": 0.7921,
+      "step": 34
+    },
+    {
+      "epoch": 0.23738872403560832,
+      "grad_norm": 2.173619464441479,
+      "learning_rate": 3.783783783783784e-05,
+      "loss": 0.7997,
+      "step": 35
+    },
+    {
+      "epoch": 0.24417125900805425,
+      "grad_norm": 1.3174649366078477,
+      "learning_rate": 3.8918918918918926e-05,
+      "loss": 0.768,
+      "step": 36
+    },
+    {
+      "epoch": 0.2509537939805002,
+      "grad_norm": 2.560545381997849,
+      "learning_rate": 4e-05,
+      "loss": 0.7938,
+      "step": 37
+    },
+    {
+      "epoch": 0.2577363289529462,
+      "grad_norm": 1.6460978741311298,
+      "learning_rate": 4.108108108108108e-05,
+      "loss": 0.7873,
+      "step": 38
+    },
+    {
+      "epoch": 0.2645188639253921,
+      "grad_norm": 2.1426591768978938,
+      "learning_rate": 4.2162162162162164e-05,
+      "loss": 0.781,
+      "step": 39
+    },
+    {
+      "epoch": 0.27130139889783805,
+      "grad_norm": 1.8160361307625077,
+      "learning_rate": 4.324324324324325e-05,
+      "loss": 0.7809,
+      "step": 40
+    },
+    {
+      "epoch": 0.278083933870284,
+      "grad_norm": 1.6910805765281172,
+      "learning_rate": 4.4324324324324325e-05,
+      "loss": 0.7744,
+      "step": 41
+    },
+    {
+      "epoch": 0.28486646884273,
+      "grad_norm": 1.9801882908067985,
+      "learning_rate": 4.540540540540541e-05,
+      "loss": 0.7732,
+      "step": 42
+    },
+    {
+      "epoch": 0.29164900381517594,
+      "grad_norm": 1.3912495139596244,
+      "learning_rate": 4.6486486486486486e-05,
+      "loss": 0.778,
+      "step": 43
+    },
+    {
+      "epoch": 0.29843153878762185,
+      "grad_norm": 1.9510858316270183,
+      "learning_rate": 4.756756756756758e-05,
+      "loss": 0.7628,
+      "step": 44
+    },
+    {
+      "epoch": 0.3052140737600678,
+      "grad_norm": 1.9755444838845442,
+      "learning_rate": 4.8648648648648654e-05,
+      "loss": 0.77,
+      "step": 45
+    },
+    {
+      "epoch": 0.3119966087325138,
+      "grad_norm": 1.7136629431095296,
+      "learning_rate": 4.972972972972973e-05,
+      "loss": 0.7776,
+      "step": 46
+    },
+    {
+      "epoch": 0.31877914370495974,
+      "grad_norm": 1.9262891096799022,
+      "learning_rate": 5.081081081081081e-05,
+      "loss": 0.7591,
+      "step": 47
+    },
+    {
+      "epoch": 0.3255616786774057,
+      "grad_norm": 1.8644045069770525,
+      "learning_rate": 5.18918918918919e-05,
+      "loss": 0.7564,
+      "step": 48
+    },
+    {
+      "epoch": 0.3323442136498516,
+      "grad_norm": 1.273771704881583,
+      "learning_rate": 5.2972972972972976e-05,
+      "loss": 0.7615,
+      "step": 49
+    },
+    {
+      "epoch": 0.3391267486222976,
+      "grad_norm": 2.4421063866094843,
+      "learning_rate": 5.405405405405406e-05,
+      "loss": 0.7803,
+      "step": 50
+    },
+    {
+      "epoch": 0.34590928359474354,
+      "grad_norm": 2.8264810278747197,
+      "learning_rate": 5.513513513513514e-05,
+      "loss": 0.7614,
+      "step": 51
+    },
+    {
+      "epoch": 0.3526918185671895,
+      "grad_norm": 2.2601831362375036,
+      "learning_rate": 5.621621621621622e-05,
+      "loss": 0.7727,
+      "step": 52
+    },
+    {
+      "epoch": 0.3594743535396354,
+      "grad_norm": 3.49645659835363,
+      "learning_rate": 5.7297297297297305e-05,
+      "loss": 0.7589,
+      "step": 53
+    },
+    {
+      "epoch": 0.3662568885120814,
+      "grad_norm": 2.8843912596082983,
+      "learning_rate": 5.837837837837838e-05,
+      "loss": 0.7561,
+      "step": 54
+    },
+    {
+      "epoch": 0.37303942348452734,
+      "grad_norm": 2.7371626553055965,
+      "learning_rate": 5.945945945945946e-05,
+      "loss": 0.7468,
+      "step": 55
+    },
+    {
+      "epoch": 0.3798219584569733,
+      "grad_norm": 2.4650676398708997,
+      "learning_rate": 6.054054054054055e-05,
+      "loss": 0.7501,
+      "step": 56
+    },
+    {
+      "epoch": 0.38660449342941927,
+      "grad_norm": 2.617962450345242,
+      "learning_rate": 6.162162162162163e-05,
+      "loss": 0.7504,
+      "step": 57
+    },
+    {
+      "epoch": 0.3933870284018652,
+      "grad_norm": 1.9045313378628999,
+      "learning_rate": 6.27027027027027e-05,
+      "loss": 0.7443,
+      "step": 58
+    },
+    {
+      "epoch": 0.40016956337431114,
+      "grad_norm": 3.214875277973139,
+      "learning_rate": 6.378378378378379e-05,
+      "loss": 0.756,
+      "step": 59
+    },
+    {
+      "epoch": 0.4069520983467571,
+      "grad_norm": 2.607985787054137,
+      "learning_rate": 6.486486486486487e-05,
+      "loss": 0.7575,
+      "step": 60
+    },
+    {
+      "epoch": 0.41373463331920307,
+      "grad_norm": 2.462297061139689,
+      "learning_rate": 6.594594594594596e-05,
+      "loss": 0.7395,
+      "step": 61
+    },
+    {
+      "epoch": 0.420517168291649,
+      "grad_norm": 1.7928802854683021,
+      "learning_rate": 6.702702702702703e-05,
+      "loss": 0.7443,
+      "step": 62
+    },
+    {
+      "epoch": 0.42729970326409494,
+      "grad_norm": 2.8782764835293015,
+      "learning_rate": 6.810810810810811e-05,
+      "loss": 0.7574,
+      "step": 63
+    },
+    {
+      "epoch": 0.4340822382365409,
+      "grad_norm": 2.107005343355103,
+      "learning_rate": 6.91891891891892e-05,
+      "loss": 0.7524,
+      "step": 64
+    },
+    {
+      "epoch": 0.44086477320898687,
+      "grad_norm": 2.400586779165952,
+      "learning_rate": 7.027027027027028e-05,
+      "loss": 0.7539,
+      "step": 65
+    },
+    {
+      "epoch": 0.44764730818143283,
+      "grad_norm": 1.8984648015553665,
+      "learning_rate": 7.135135135135136e-05,
+      "loss": 0.7661,
+      "step": 66
+    },
+    {
+      "epoch": 0.45442984315387874,
+      "grad_norm": 2.701931984195687,
+      "learning_rate": 7.243243243243243e-05,
+      "loss": 0.7423,
+      "step": 67
+    },
+    {
+      "epoch": 0.4612123781263247,
+      "grad_norm": 3.1216702832116807,
+      "learning_rate": 7.351351351351353e-05,
+      "loss": 0.7558,
+      "step": 68
+    },
+    {
+      "epoch": 0.46799491309877067,
+      "grad_norm": 1.7791622339381934,
+      "learning_rate": 7.45945945945946e-05,
+      "loss": 0.7338,
+      "step": 69
+    },
+    {
+      "epoch": 0.47477744807121663,
+      "grad_norm": 2.283223713003048,
+      "learning_rate": 7.567567567567568e-05,
+      "loss": 0.7512,
+      "step": 70
+    },
+    {
+      "epoch": 0.48155998304366254,
+      "grad_norm": 2.518755689574504,
+      "learning_rate": 7.675675675675675e-05,
+      "loss": 0.7371,
+      "step": 71
+    },
+    {
+      "epoch": 0.4883425180161085,
+      "grad_norm": 1.99481044416433,
+      "learning_rate": 7.783783783783785e-05,
+      "loss": 0.7478,
+      "step": 72
+    },
+    {
+      "epoch": 0.49512505298855447,
+      "grad_norm": 2.7463149548347183,
+      "learning_rate": 7.891891891891892e-05,
+      "loss": 0.7365,
+      "step": 73
+    },
+    {
+      "epoch": 0.5019075879610004,
+      "grad_norm": 2.138101333819343,
+      "learning_rate": 8e-05,
+      "loss": 0.7254,
+      "step": 74
+    },
+    {
+      "epoch": 0.5086901229334464,
+      "grad_norm": 2.640544503377941,
+      "learning_rate": 7.999954822103665e-05,
+      "loss": 0.7417,
+      "step": 75
+    },
+    {
+      "epoch": 0.5154726579058924,
+      "grad_norm": 2.2402410683129688,
+      "learning_rate": 7.999819289435179e-05,
+      "loss": 0.7354,
+      "step": 76
+    },
+    {
+      "epoch": 0.5222551928783383,
+      "grad_norm": 2.49341837988937,
+      "learning_rate": 7.999593405056084e-05,
+      "loss": 0.738,
+      "step": 77
+    },
+    {
+      "epoch": 0.5290377278507842,
+      "grad_norm": 2.3287753265703253,
+      "learning_rate": 7.999277174068872e-05,
+      "loss": 0.7247,
+      "step": 78
+    },
+    {
+      "epoch": 0.5358202628232301,
+      "grad_norm": 2.2524400032207716,
+      "learning_rate": 7.998870603616864e-05,
+      "loss": 0.7368,
+      "step": 79
+    },
+    {
+      "epoch": 0.5426027977956761,
+      "grad_norm": 2.686780411560504,
+      "learning_rate": 7.998373702884062e-05,
+      "loss": 0.7315,
+      "step": 80
+    },
+    {
+      "epoch": 0.5493853327681221,
+      "grad_norm": 1.7256862069861159,
+      "learning_rate": 7.997786483094931e-05,
+      "loss": 0.7492,
+      "step": 81
+    },
+    {
+      "epoch": 0.556167867740568,
+      "grad_norm": 2.6274112743201523,
+      "learning_rate": 7.997108957514146e-05,
+      "loss": 0.7229,
+      "step": 82
+    },
+    {
+      "epoch": 0.562950402713014,
+      "grad_norm": 2.0731276463095596,
+      "learning_rate": 7.9963411414463e-05,
+      "loss": 0.7375,
+      "step": 83
+    },
+    {
+      "epoch": 0.56973293768546,
+      "grad_norm": 2.6126411934480123,
+      "learning_rate": 7.99548305223555e-05,
+      "loss": 0.7284,
+      "step": 84
+    },
+    {
+      "epoch": 0.5765154726579059,
+      "grad_norm": 1.4633346753767393,
+      "learning_rate": 7.994534709265226e-05,
+      "loss": 0.7248,
+      "step": 85
+    },
+    {
+      "epoch": 0.5832980076303519,
+      "grad_norm": 3.4984880607635307,
+      "learning_rate": 7.993496133957401e-05,
+      "loss": 0.7517,
+      "step": 86
+    },
+    {
+      "epoch": 0.5900805426027977,
+      "grad_norm": 2.4422440886664147,
+      "learning_rate": 7.992367349772398e-05,
+      "loss": 0.7378,
+      "step": 87
+    },
+    {
+      "epoch": 0.5968630775752437,
+      "grad_norm": 1.8295430254229907,
+      "learning_rate": 7.991148382208265e-05,
+      "loss": 0.7241,
+      "step": 88
+    },
+    {
+      "epoch": 0.6036456125476897,
+      "grad_norm": 2.9619629036982755,
+      "learning_rate": 7.989839258800196e-05,
+      "loss": 0.7422,
+      "step": 89
+    },
+    {
+      "epoch": 0.6104281475201356,
+      "grad_norm": 1.9664968756669656,
+      "learning_rate": 7.988440009119911e-05,
+      "loss": 0.751,
+      "step": 90
+    },
+    {
+      "epoch": 0.6172106824925816,
+      "grad_norm": 1.6194583900993504,
+      "learning_rate": 7.986950664774992e-05,
+      "loss": 0.7346,
+      "step": 91
+    },
+    {
+      "epoch": 0.6239932174650276,
+      "grad_norm": 2.8148837821592796,
+      "learning_rate": 7.985371259408157e-05,
+      "loss": 0.7413,
+      "step": 92
+    },
+    {
+      "epoch": 0.6307757524374735,
+      "grad_norm": 1.486533554565982,
+      "learning_rate": 7.983701828696515e-05,
+      "loss": 0.7296,
+      "step": 93
+    },
+    {
+      "epoch": 0.6375582874099195,
+      "grad_norm": 3.229587748852655,
+      "learning_rate": 7.98194241035075e-05,
+      "loss": 0.7483,
+      "step": 94
+    },
+    {
+      "epoch": 0.6443408223823655,
+      "grad_norm": 2.518619876391423,
+      "learning_rate": 7.980093044114269e-05,
+      "loss": 0.7524,
+      "step": 95
+    },
+    {
+      "epoch": 0.6511233573548114,
+      "grad_norm": 2.6606747542222915,
+      "learning_rate": 7.978153771762311e-05,
+      "loss": 0.7531,
+      "step": 96
+    },
+    {
+      "epoch": 0.6579058923272573,
+      "grad_norm": 1.7367257209193254,
+      "learning_rate": 7.976124637101e-05,
+      "loss": 0.736,
+      "step": 97
+    },
+    {
+      "epoch": 0.6646884272997032,
+      "grad_norm": 1.8350871328487093,
+      "learning_rate": 7.974005685966354e-05,
+      "loss": 0.7248,
+      "step": 98
+    },
+    {
+      "epoch": 0.6714709622721492,
+      "grad_norm": 1.5007282529505472,
+      "learning_rate": 7.971796966223248e-05,
+      "loss": 0.7498,
+      "step": 99
+    },
+    {
+      "epoch": 0.6782534972445952,
+      "grad_norm": 1.4248528231943807,
+      "learning_rate": 7.969498527764341e-05,
+      "loss": 0.7208,
+      "step": 100
+    },
+    {
+      "epoch": 0.6850360322170411,
+      "grad_norm": 2.217134309648403,
+      "learning_rate": 7.967110422508936e-05,
+      "loss": 0.7295,
+      "step": 101
+    },
+    {
+      "epoch": 0.6918185671894871,
+      "grad_norm": 2.2429893368943277,
+      "learning_rate": 7.964632704401823e-05,
+      "loss": 0.7429,
+      "step": 102
+    },
+    {
+      "epoch": 0.698601102161933,
+      "grad_norm": 1.317977635899577,
+      "learning_rate": 7.962065429412046e-05,
+      "loss": 0.7311,
+      "step": 103
+    },
+    {
+      "epoch": 0.705383637134379,
+      "grad_norm": 3.1540990631312718,
+      "learning_rate": 7.959408655531646e-05,
+      "loss": 0.7331,
+      "step": 104
+    },
+    {
+      "epoch": 0.712166172106825,
+      "grad_norm": 2.0946781410468382,
+      "learning_rate": 7.956662442774351e-05,
+      "loss": 0.7357,
+      "step": 105
+    },
+    {
+      "epoch": 0.7189487070792708,
+      "grad_norm": 3.6514929315006834,
+      "learning_rate": 7.953826853174218e-05,
+      "loss": 0.724,
+      "step": 106
+    },
+    {
+      "epoch": 0.7257312420517168,
+      "grad_norm": 2.743797555116305,
+      "learning_rate": 7.950901950784236e-05,
+      "loss": 0.7337,
+      "step": 107
+    },
+    {
+      "epoch": 0.7325137770241628,
+      "grad_norm": 2.6025703631899364,
+      "learning_rate": 7.947887801674872e-05,
+      "loss": 0.7231,
+      "step": 108
+    },
+    {
+      "epoch": 0.7392963119966087,
+      "grad_norm": 2.33134578145788,
+      "learning_rate": 7.944784473932583e-05,
+      "loss": 0.7268,
+      "step": 109
+    },
+    {
+      "epoch": 0.7460788469690547,
+      "grad_norm": 2.369166475084867,
+      "learning_rate": 7.941592037658279e-05,
+      "loss": 0.7175,
+      "step": 110
+    },
+    {
+      "epoch": 0.7528613819415007,
+      "grad_norm": 1.5983852946337034,
+      "learning_rate": 7.93831056496574e-05,
+      "loss": 0.7215,
+      "step": 111
+    },
+    {
+      "epoch": 0.7596439169139466,
+      "grad_norm": 2.3446201220972696,
+      "learning_rate": 7.934940129979979e-05,
+      "loss": 0.7314,
+      "step": 112
+    },
+    {
+      "epoch": 0.7664264518863926,
+      "grad_norm": 1.872428876067427,
+      "learning_rate": 7.931480808835577e-05,
+      "loss": 0.7287,
+      "step": 113
+    },
+    {
+      "epoch": 0.7732089868588385,
+      "grad_norm": 1.982878649441548,
+      "learning_rate": 7.927932679674964e-05,
+      "loss": 0.7193,
+      "step": 114
+    },
+    {
+      "epoch": 0.7799915218312844,
+      "grad_norm": 1.5101882188042894,
+      "learning_rate": 7.924295822646643e-05,
+      "loss": 0.7217,
+      "step": 115
+    },
+    {
+      "epoch": 0.7867740568037304,
+      "grad_norm": 1.5820041393952022,
+      "learning_rate": 7.92057031990339e-05,
+      "loss": 0.7199,
+      "step": 116
+    },
+    {
+      "epoch": 0.7935565917761763,
+      "grad_norm": 1.5630207257272397,
+      "learning_rate": 7.91675625560039e-05,
+      "loss": 0.7141,
+      "step": 117
+    },
+    {
+      "epoch": 0.8003391267486223,
+      "grad_norm": 2.457059696195913,
+      "learning_rate": 7.91285371589335e-05,
+      "loss": 0.7082,
+      "step": 118
+    },
+    {
+      "epoch": 0.8071216617210683,
+      "grad_norm": 1.6028005358543407,
+      "learning_rate": 7.908862788936532e-05,
+      "loss": 0.7149,
+      "step": 119
+    },
+    {
+      "epoch": 0.8139041966935142,
+      "grad_norm": 1.8116395322587184,
+      "learning_rate": 7.904783564880779e-05,
+      "loss": 0.7123,
+      "step": 120
+    },
+    {
+      "epoch": 0.8206867316659602,
+      "grad_norm": 2.0950388250383996,
+      "learning_rate": 7.900616135871474e-05,
+      "loss": 0.7354,
+      "step": 121
+    },
+    {
+      "epoch": 0.8274692666384061,
+      "grad_norm": 1.8624393412855933,
+      "learning_rate": 7.896360596046453e-05,
+      "loss": 0.7156,
+      "step": 122
+    },
+    {
+      "epoch": 0.8342518016108521,
+      "grad_norm": 1.4851216429547307,
+      "learning_rate": 7.892017041533886e-05,
+      "loss": 0.7114,
+      "step": 123
+    },
+    {
+      "epoch": 0.841034336583298,
+      "grad_norm": 2.9385975867591765,
+      "learning_rate": 7.887585570450098e-05,
+      "loss": 0.7127,
+      "step": 124
+    },
+    {
+      "epoch": 0.8478168715557439,
+      "grad_norm": 1.8479656443451273,
+      "learning_rate": 7.883066282897362e-05,
+      "loss": 0.7173,
+      "step": 125
+    },
+    {
+      "epoch": 0.8545994065281899,
+      "grad_norm": 2.894143467483064,
+      "learning_rate": 7.878459280961629e-05,
+      "loss": 0.7183,
+      "step": 126
+    },
+    {
+      "epoch": 0.8613819415006359,
+      "grad_norm": 2.167706950028001,
+      "learning_rate": 7.873764668710228e-05,
+      "loss": 0.7237,
+      "step": 127
+    },
+    {
+      "epoch": 0.8681644764730818,
+      "grad_norm": 2.646470616726372,
+      "learning_rate": 7.868982552189514e-05,
+      "loss": 0.7132,
+      "step": 128
+    },
+    {
+      "epoch": 0.8749470114455278,
+      "grad_norm": 2.242758826547572,
+      "learning_rate": 7.864113039422464e-05,
+      "loss": 0.7296,
+      "step": 129
+    },
+    {
+      "epoch": 0.8817295464179737,
+      "grad_norm": 2.493893447513006,
+      "learning_rate": 7.859156240406252e-05,
+      "loss": 0.7185,
+      "step": 130
+    },
+    {
+      "epoch": 0.8885120813904197,
+      "grad_norm": 1.721360057791857,
+      "learning_rate": 7.854112267109756e-05,
+      "loss": 0.7244,
+      "step": 131
+    },
+    {
+      "epoch": 0.8952946163628657,
+      "grad_norm": 2.6683450004265583,
+      "learning_rate": 7.848981233471024e-05,
+      "loss": 0.7149,
+      "step": 132
+    },
+    {
+      "epoch": 0.9020771513353115,
+      "grad_norm": 2.061224678516105,
+      "learning_rate": 7.843763255394711e-05,
+      "loss": 0.7118,
+      "step": 133
+    },
+    {
+      "epoch": 0.9088596863077575,
+      "grad_norm": 1.9936485418092913,
+      "learning_rate": 7.838458450749452e-05,
+      "loss": 0.7057,
+      "step": 134
+    },
+    {
+      "epoch": 0.9156422212802034,
+      "grad_norm": 1.5127328421814061,
+      "learning_rate": 7.833066939365206e-05,
+      "loss": 0.7043,
+      "step": 135
+    },
+    {
+      "epoch": 0.9224247562526494,
+      "grad_norm": 2.040325487116571,
+      "learning_rate": 7.827588843030543e-05,
+      "loss": 0.7087,
+      "step": 136
+    },
+    {
+      "epoch": 0.9292072912250954,
+      "grad_norm": 1.6268257121266954,
+      "learning_rate": 7.822024285489896e-05,
+      "loss": 0.7105,
+      "step": 137
+    },
+    {
+      "epoch": 0.9359898261975413,
+      "grad_norm": 2.3150904956674263,
+      "learning_rate": 7.81637339244077e-05,
+      "loss": 0.7139,
+      "step": 138
+    },
+    {
+      "epoch": 0.9427723611699873,
+      "grad_norm": 1.841068832751663,
+      "learning_rate": 7.810636291530893e-05,
+      "loss": 0.7052,
+      "step": 139
+    },
+    {
+      "epoch": 0.9495548961424333,
+      "grad_norm": 1.9983783549939655,
+      "learning_rate": 7.804813112355339e-05,
+      "loss": 0.7115,
+      "step": 140
+    },
+    {
+      "epoch": 0.9563374311148792,
+      "grad_norm": 1.9754428939529725,
+      "learning_rate": 7.798903986453603e-05,
+      "loss": 0.7142,
+      "step": 141
+    },
+    {
+      "epoch": 0.9631199660873251,
+      "grad_norm": 1.735226023700741,
+      "learning_rate": 7.792909047306623e-05,
+      "loss": 0.7205,
+      "step": 142
+    },
+    {
+      "epoch": 0.969902501059771,
+      "grad_norm": 2.184380091482261,
+      "learning_rate": 7.786828430333769e-05,
+      "loss": 0.7107,
+      "step": 143
+    },
+    {
+      "epoch": 0.976685036032217,
+      "grad_norm": 1.8757020725047386,
+      "learning_rate": 7.78066227288978e-05,
+      "loss": 0.7198,
+      "step": 144
+    },
+    {
+      "epoch": 0.983467571004663,
+      "grad_norm": 1.0079051823953802,
+      "learning_rate": 7.77441071426167e-05,
+      "loss": 0.714,
+      "step": 145
+    },
+    {
+      "epoch": 0.9902501059771089,
+      "grad_norm": 1.5288723958450625,
+      "learning_rate": 7.768073895665573e-05,
+      "loss": 0.7087,
+      "step": 146
+    },
+    {
+      "epoch": 0.9970326409495549,
+      "grad_norm": 2.3013059467419996,
+      "learning_rate": 7.761651960243554e-05,
+      "loss": 0.7366,
+      "step": 147
+    },
+    {
+      "epoch": 1.0038151759220009,
+      "grad_norm": 2.269978916078414,
+      "learning_rate": 7.755145053060378e-05,
+      "loss": 1.1363,
+      "step": 148
+    },
+    {
+      "epoch": 1.0105977108944468,
+      "grad_norm": 1.6332374526993498,
+      "learning_rate": 7.748553321100238e-05,
+      "loss": 0.697,
+      "step": 149
+    },
+    {
+      "epoch": 1.0173802458668928,
+      "grad_norm": 2.0144072844842054,
+      "learning_rate": 7.741876913263422e-05,
+      "loss": 0.7169,
+      "step": 150
+    },
+    {
+      "epoch": 1.0241627808393388,
+      "grad_norm": 1.9963737399863932,
+      "learning_rate": 7.735115980362964e-05,
+      "loss": 0.6942,
+      "step": 151
+    },
+    {
+      "epoch": 1.0309453158117847,
+      "grad_norm": 2.010956402670031,
+      "learning_rate": 7.728270675121224e-05,
+      "loss": 0.7108,
+      "step": 152
+    },
+    {
+      "epoch": 1.0377278507842307,
+      "grad_norm": 1.1256640161683924,
+      "learning_rate": 7.721341152166448e-05,
+      "loss": 0.706,
+      "step": 153
+    },
+    {
+      "epoch": 1.0445103857566767,
+      "grad_norm": 2.7360045649430287,
+      "learning_rate": 7.714327568029272e-05,
+      "loss": 0.7035,
+      "step": 154
+    },
+    {
+      "epoch": 1.0512929207291226,
+      "grad_norm": 1.8273909265523915,
+      "learning_rate": 7.707230081139184e-05,
+      "loss": 0.7064,
+      "step": 155
+    },
+    {
+      "epoch": 1.0580754557015684,
+      "grad_norm": 2.155560932773446,
+      "learning_rate": 7.700048851820946e-05,
+      "loss": 0.6991,
+      "step": 156
+    },
+    {
+      "epoch": 1.0648579906740143,
+      "grad_norm": 2.1669039610273493,
+      "learning_rate": 7.692784042290976e-05,
+      "loss": 0.7099,
+      "step": 157
+    },
+    {
+      "epoch": 1.0716405256464603,
+      "grad_norm": 1.503612527062841,
+      "learning_rate": 7.685435816653681e-05,
+      "loss": 0.6973,
+      "step": 158
+    },
+    {
+      "epoch": 1.0784230606189062,
+      "grad_norm": 1.937309301259521,
+      "learning_rate": 7.678004340897747e-05,
+      "loss": 0.7049,
+      "step": 159
+    },
+    {
+      "epoch": 1.0852055955913522,
+      "grad_norm": 1.3969091003754739,
+      "learning_rate": 7.670489782892396e-05,
+      "loss": 0.6886,
+      "step": 160
+    },
+    {
+      "epoch": 1.0919881305637982,
+      "grad_norm": 1.9964168232948958,
+      "learning_rate": 7.662892312383592e-05,
+      "loss": 0.6883,
+      "step": 161
+    },
+    {
+      "epoch": 1.0987706655362441,
+      "grad_norm": 1.7569755198383514,
+      "learning_rate": 7.655212100990195e-05,
+      "loss": 0.6987,
+      "step": 162
+    },
+    {
+      "epoch": 1.10555320050869,
+      "grad_norm": 1.4447704185089303,
+      "learning_rate": 7.647449322200108e-05,
+      "loss": 0.6906,
+      "step": 163
+    },
+    {
+      "epoch": 1.112335735481136,
+      "grad_norm": 1.2497557944098203,
+      "learning_rate": 7.639604151366339e-05,
+      "loss": 0.7009,
+      "step": 164
+    },
+    {
+      "epoch": 1.119118270453582,
+      "grad_norm": 2.3116148247793267,
+      "learning_rate": 7.631676765703042e-05,
+      "loss": 0.6895,
+      "step": 165
+    },
+    {
+      "epoch": 1.125900805426028,
+      "grad_norm": 1.330812583383646,
+      "learning_rate": 7.623667344281522e-05,
+      "loss": 0.7004,
+      "step": 166
+    },
+    {
+      "epoch": 1.132683340398474,
+      "grad_norm": 1.8610464265540336,
+      "learning_rate": 7.615576068026187e-05,
+      "loss": 0.7077,
+      "step": 167
+    },
+    {
+      "epoch": 1.13946587537092,
+      "grad_norm": 1.7667897999541395,
+      "learning_rate": 7.607403119710453e-05,
+      "loss": 0.6913,
+      "step": 168
+    },
+    {
+      "epoch": 1.1462484103433659,
+      "grad_norm": 1.208336959762333,
+      "learning_rate": 7.599148683952628e-05,
+      "loss": 0.6839,
+      "step": 169
+    },
+    {
+      "epoch": 1.1530309453158119,
+      "grad_norm": 1.9033319484816924,
+      "learning_rate": 7.590812947211733e-05,
+      "loss": 0.6898,
+      "step": 170
+    },
+    {
+      "epoch": 1.1598134802882578,
+      "grad_norm": 1.213830580906014,
+      "learning_rate": 7.582396097783294e-05,
+      "loss": 0.6866,
+      "step": 171
+    },
+    {
+      "epoch": 1.1665960152607038,
+      "grad_norm": 2.309488724064315,
+      "learning_rate": 7.573898325795086e-05,
+      "loss": 0.697,
+      "step": 172
+    },
+    {
+      "epoch": 1.1733785502331497,
+      "grad_norm": 1.3337697012970882,
+      "learning_rate": 7.565319823202838e-05,
+      "loss": 0.7083,
+      "step": 173
+    },
+    {
+      "epoch": 1.1801610852055955,
+      "grad_norm": 1.644453886623139,
+      "learning_rate": 7.556660783785904e-05,
+      "loss": 0.6905,
+      "step": 174
+    },
+    {
+      "epoch": 1.1869436201780414,
+      "grad_norm": 1.3203501723312958,
+      "learning_rate": 7.547921403142874e-05,
+      "loss": 0.7048,
+      "step": 175
+    },
+    {
+      "epoch": 1.1937261551504874,
+      "grad_norm": 1.4545383798679439,
+      "learning_rate": 7.539101878687165e-05,
+      "loss": 0.6865,
+      "step": 176
+    },
+    {
+      "epoch": 1.2005086901229334,
+      "grad_norm": 2.1879709481996987,
+      "learning_rate": 7.53020240964256e-05,
+      "loss": 0.6964,
+      "step": 177
+    },
+    {
+      "epoch": 1.2072912250953793,
+      "grad_norm": 1.8322469465167428,
+      "learning_rate": 7.5212231970387e-05,
+      "loss": 0.6921,
+      "step": 178
+    },
+    {
+      "epoch": 1.2140737600678253,
+      "grad_norm": 1.2841938775963582,
+      "learning_rate": 7.512164443706555e-05,
+      "loss": 0.6816,
+      "step": 179
+    },
+    {
+      "epoch": 1.2208562950402713,
+      "grad_norm": 3.1832942767833003,
+      "learning_rate": 7.503026354273834e-05,
+      "loss": 0.7107,
+      "step": 180
+    },
+    {
+      "epoch": 1.2276388300127172,
+      "grad_norm": 2.223671181736338,
+      "learning_rate": 7.493809135160367e-05,
+      "loss": 0.7068,
+      "step": 181
+    },
+    {
+      "epoch": 1.2344213649851632,
+      "grad_norm": 2.5365710286141727,
+      "learning_rate": 7.484512994573438e-05,
+      "loss": 0.6997,
+      "step": 182
+    },
+    {
+      "epoch": 1.2412038999576092,
+      "grad_norm": 1.7033409104571895,
+      "learning_rate": 7.475138142503083e-05,
+      "loss": 0.6973,
+      "step": 183
+    },
+    {
+      "epoch": 1.2479864349300551,
+      "grad_norm": 2.666703064771071,
+      "learning_rate": 7.465684790717354e-05,
+      "loss": 0.6904,
+      "step": 184
+    },
+    {
+      "epoch": 1.254768969902501,
+      "grad_norm": 1.6449288965298707,
+      "learning_rate": 7.45615315275752e-05,
+      "loss": 0.7008,
+      "step": 185
+    },
+    {
+      "epoch": 1.261551504874947,
+      "grad_norm": 2.670948821208354,
+      "learning_rate": 7.446543443933258e-05,
+      "loss": 0.7187,
+      "step": 186
+    },
+    {
+      "epoch": 1.268334039847393,
+      "grad_norm": 2.052845388192151,
+      "learning_rate": 7.436855881317784e-05,
+      "loss": 0.7154,
+      "step": 187
+    },
+    {
+      "epoch": 1.275116574819839,
+      "grad_norm": 2.3638246999276054,
+      "learning_rate": 7.427090683742947e-05,
+      "loss": 0.6942,
+      "step": 188
+    },
+    {
+      "epoch": 1.281899109792285,
+      "grad_norm": 2.367916777331452,
+      "learning_rate": 7.417248071794288e-05,
+      "loss": 0.6978,
+      "step": 189
+    },
+    {
+      "epoch": 1.288681644764731,
+      "grad_norm": 1.919070978827847,
+      "learning_rate": 7.40732826780606e-05,
+      "loss": 0.6778,
+      "step": 190
+    },
+    {
+      "epoch": 1.2954641797371766,
+      "grad_norm": 1.5807447282945328,
+      "learning_rate": 7.397331495856199e-05,
+      "loss": 0.6878,
+      "step": 191
+    },
+    {
+      "epoch": 1.3022467147096228,
+      "grad_norm": 1.5725379589793347,
+      "learning_rate": 7.38725798176127e-05,
+      "loss": 0.6942,
+      "step": 192
+    },
+    {
+      "epoch": 1.3090292496820686,
+      "grad_norm": 1.5854647888888829,
+      "learning_rate": 7.37710795307136e-05,
+      "loss": 0.6942,
+      "step": 193
+    },
+    {
+      "epoch": 1.3158117846545148,
+      "grad_norm": 0.9840396505987244,
+      "learning_rate": 7.36688163906494e-05,
+      "loss": 0.6822,
+      "step": 194
+    },
+    {
+      "epoch": 1.3225943196269605,
+      "grad_norm": 1.9293846538238884,
+      "learning_rate": 7.356579270743689e-05,
+      "loss": 0.707,
+      "step": 195
+    },
+    {
+      "epoch": 1.3293768545994065,
+      "grad_norm": 1.297177627346237,
+      "learning_rate": 7.346201080827272e-05,
+      "loss": 0.6966,
+      "step": 196
+    },
+    {
+      "epoch": 1.3361593895718524,
+      "grad_norm": 2.3709366791443456,
+      "learning_rate": 7.335747303748079e-05,
+      "loss": 0.7001,
+      "step": 197
+    },
+    {
+      "epoch": 1.3429419245442984,
+      "grad_norm": 1.7251163867182135,
+      "learning_rate": 7.325218175645942e-05,
+      "loss": 0.694,
+      "step": 198
+    },
+    {
+      "epoch": 1.3497244595167444,
+      "grad_norm": 2.088301626825016,
+      "learning_rate": 7.314613934362788e-05,
+      "loss": 0.6879,
+      "step": 199
+    },
+    {
+      "epoch": 1.3565069944891903,
+      "grad_norm": 1.9529188107557567,
+      "learning_rate": 7.303934819437277e-05,
+      "loss": 0.6944,
+      "step": 200
+    },
+    {
+      "epoch": 1.3632895294616363,
+      "grad_norm": 1.6962527949805855,
+      "learning_rate": 7.293181072099377e-05,
+      "loss": 0.6917,
+      "step": 201
+    },
+    {
+      "epoch": 1.3700720644340822,
+      "grad_norm": 1.6169230360470628,
+      "learning_rate": 7.282352935264934e-05,
+      "loss": 0.684,
+      "step": 202
+    },
+    {
+      "epoch": 1.3768545994065282,
+      "grad_norm": 1.485732940861678,
+      "learning_rate": 7.271450653530167e-05,
+      "loss": 0.6862,
+      "step": 203
+    },
+    {
+      "epoch": 1.3836371343789742,
+      "grad_norm": 1.6704659351121771,
+      "learning_rate": 7.260474473166154e-05,
+      "loss": 0.6932,
+      "step": 204
+    },
+    {
+      "epoch": 1.3904196693514201,
+      "grad_norm": 1.5799837137385815,
+      "learning_rate": 7.249424642113266e-05,
+      "loss": 0.697,
+      "step": 205
+    },
+    {
+      "epoch": 1.397202204323866,
+      "grad_norm": 1.2977311485642662,
+      "learning_rate": 7.238301409975561e-05,
+      "loss": 0.6919,
+      "step": 206
+    },
+    {
+      "epoch": 1.403984739296312,
+      "grad_norm": 1.451392232428334,
+      "learning_rate": 7.227105028015156e-05,
+      "loss": 0.699,
+      "step": 207
+    },
+    {
+      "epoch": 1.410767274268758,
+      "grad_norm": 1.2941406989928823,
+      "learning_rate": 7.21583574914654e-05,
+      "loss": 0.6708,
+      "step": 208
+    },
+    {
+      "epoch": 1.417549809241204,
+      "grad_norm": 1.6593482637225496,
+      "learning_rate": 7.204493827930869e-05,
+      "loss": 0.6871,
+      "step": 209
+    },
+    {
+      "epoch": 1.4243323442136497,
+      "grad_norm": 1.528014842634458,
+      "learning_rate": 7.193079520570217e-05,
+      "loss": 0.6777,
+      "step": 210
+    },
+    {
+      "epoch": 1.431114879186096,
+      "grad_norm": 1.2836973493312132,
+      "learning_rate": 7.18159308490178e-05,
+      "loss": 0.6741,
+      "step": 211
+    },
+    {
+      "epoch": 1.4378974141585417,
+      "grad_norm": 1.2069873065957852,
+      "learning_rate": 7.170034780392055e-05,
+      "loss": 0.6852,
+      "step": 212
+    },
+    {
+      "epoch": 1.4446799491309876,
+      "grad_norm": 1.6652484448685945,
+      "learning_rate": 7.158404868130988e-05,
+      "loss": 0.6711,
+      "step": 213
+    },
+    {
+      "epoch": 1.4514624841034336,
+      "grad_norm": 1.2576599603912495,
+      "learning_rate": 7.14670361082606e-05,
+      "loss": 0.6849,
+      "step": 214
+    },
+    {
+      "epoch": 1.4582450190758796,
+      "grad_norm": 1.5405832213240351,
+      "learning_rate": 7.134931272796368e-05,
+      "loss": 0.691,
+      "step": 215
+    },
+    {
+      "epoch": 1.4650275540483255,
+      "grad_norm": 1.684185898740359,
+      "learning_rate": 7.123088119966647e-05,
+      "loss": 0.6868,
+      "step": 216
+    },
+    {
+      "epoch": 1.4718100890207715,
+      "grad_norm": 1.0827904024479826,
+      "learning_rate": 7.11117441986126e-05,
+      "loss": 0.6878,
+      "step": 217
+    },
+    {
+      "epoch": 1.4785926239932174,
+      "grad_norm": 1.5453825081747208,
+      "learning_rate": 7.099190441598161e-05,
+      "loss": 0.6838,
+      "step": 218
+    },
+    {
+      "epoch": 1.4853751589656634,
+      "grad_norm": 1.0691158300615973,
+      "learning_rate": 7.087136455882816e-05,
+      "loss": 0.6738,
+      "step": 219
+    },
+    {
+      "epoch": 1.4921576939381094,
+      "grad_norm": 2.077881844593604,
+      "learning_rate": 7.07501273500208e-05,
+      "loss": 0.6912,
+      "step": 220
+    },
+    {
+      "epoch": 1.4989402289105553,
+      "grad_norm": 0.9909091986869899,
+      "learning_rate": 7.06281955281806e-05,
+      "loss": 0.6875,
+      "step": 221
+    },
+    {
+      "epoch": 1.5057227638830013,
+      "grad_norm": 1.826572971343745,
+      "learning_rate": 7.050557184761915e-05,
+      "loss": 0.6855,
+      "step": 222
+    },
+    {
+      "epoch": 1.5125052988554473,
+      "grad_norm": 1.279113591847188,
+      "learning_rate": 7.038225907827639e-05,
+      "loss": 0.6961,
+      "step": 223
+    },
+    {
+      "epoch": 1.5192878338278932,
+      "grad_norm": 1.229774591769691,
+      "learning_rate": 7.02582600056581e-05,
+      "loss": 0.6802,
+      "step": 224
+    },
+    {
+      "epoch": 1.526070368800339,
+      "grad_norm": 1.3203007597562126,
+      "learning_rate": 7.013357743077289e-05,
+      "loss": 0.6965,
+      "step": 225
+    },
+    {
+      "epoch": 1.5328529037727852,
+      "grad_norm": 1.2862135502175833,
+      "learning_rate": 7.000821417006898e-05,
+      "loss": 0.6815,
+      "step": 226
+    },
+    {
+      "epoch": 1.539635438745231,
+      "grad_norm": 1.0628503441369677,
+      "learning_rate": 6.988217305537058e-05,
+      "loss": 0.6768,
+      "step": 227
+    },
+    {
+      "epoch": 1.546417973717677,
+      "grad_norm": 1.3535468021784238,
+      "learning_rate": 6.97554569338139e-05,
+      "loss": 0.6959,
+      "step": 228
+    },
+    {
+      "epoch": 1.5532005086901228,
+      "grad_norm": 1.1587176301112725,
+      "learning_rate": 6.962806866778284e-05,
+      "loss": 0.6821,
+      "step": 229
+    },
+    {
+      "epoch": 1.559983043662569,
+      "grad_norm": 1.6895118703555423,
+      "learning_rate": 6.950001113484432e-05,
+      "loss": 0.6832,
+      "step": 230
+    },
+    {
+      "epoch": 1.5667655786350148,
+      "grad_norm": 1.3111534840956083,
+      "learning_rate": 6.937128722768333e-05,
+      "loss": 0.6747,
+      "step": 231
+    },
+    {
+      "epoch": 1.573548113607461,
+      "grad_norm": 1.2243130191258123,
+      "learning_rate": 6.924189985403753e-05,
+      "loss": 0.69,
+      "step": 232
+    },
+    {
+      "epoch": 1.5803306485799067,
+      "grad_norm": 1.4287675639720405,
+      "learning_rate": 6.91118519366316e-05,
+      "loss": 0.6801,
+      "step": 233
+    },
+    {
+      "epoch": 1.5871131835523526,
+      "grad_norm": 0.9267438225948489,
+      "learning_rate": 6.898114641311122e-05,
+      "loss": 0.6845,
+      "step": 234
+    },
+    {
+      "epoch": 1.5938957185247986,
+      "grad_norm": 1.3352877429277719,
+      "learning_rate": 6.884978623597665e-05,
+      "loss": 0.6993,
+      "step": 235
+    },
+    {
+      "epoch": 1.6006782534972446,
+      "grad_norm": 1.334254924822497,
+      "learning_rate": 6.871777437251617e-05,
+      "loss": 0.6817,
+      "step": 236
+    },
+    {
+      "epoch": 1.6074607884696905,
+      "grad_norm": 1.2977369297336303,
+      "learning_rate": 6.858511380473887e-05,
+      "loss": 0.6856,
+      "step": 237
+    },
+    {
+      "epoch": 1.6142433234421365,
+      "grad_norm": 1.1900674515806826,
+      "learning_rate": 6.845180752930749e-05,
+      "loss": 0.6947,
+      "step": 238
+    },
+    {
+      "epoch": 1.6210258584145825,
+      "grad_norm": 2.7317108581757057,
+      "learning_rate": 6.831785855747054e-05,
+      "loss": 0.6985,
+      "step": 239
+    },
+    {
+      "epoch": 1.6278083933870284,
+      "grad_norm": 1.2267690972200642,
+      "learning_rate": 6.81832699149944e-05,
+      "loss": 0.6773,
+      "step": 240
+    },
+    {
+      "epoch": 1.6345909283594744,
+      "grad_norm": 3.8760882746190255,
+      "learning_rate": 6.804804464209498e-05,
+      "loss": 0.6895,
+      "step": 241
+    },
+    {
+      "epoch": 1.6413734633319204,
+      "grad_norm": 3.5268150192285157,
+      "learning_rate": 6.791218579336891e-05,
+      "loss": 0.6994,
+      "step": 242
+    },
+    {
+      "epoch": 1.6481559983043663,
+      "grad_norm": 1.7590143767536321,
+      "learning_rate": 6.777569643772472e-05,
+      "loss": 0.683,
+      "step": 243
+    },
+    {
+      "epoch": 1.654938533276812,
+      "grad_norm": 1.7753106738494688,
+      "learning_rate": 6.763857965831337e-05,
+      "loss": 0.6811,
+      "step": 244
+    },
+    {
+      "epoch": 1.6617210682492582,
+      "grad_norm": 1.7854517795067268,
+      "learning_rate": 6.750083855245869e-05,
+      "loss": 0.6818,
+      "step": 245
+    },
+    {
+      "epoch": 1.668503603221704,
+      "grad_norm": 1.2937361964461016,
+      "learning_rate": 6.736247623158738e-05,
+      "loss": 0.6825,
+      "step": 246
+    },
+    {
+      "epoch": 1.6752861381941502,
+      "grad_norm": 1.2452039091600213,
+      "learning_rate": 6.722349582115872e-05,
+      "loss": 0.6779,
+      "step": 247
+    },
+    {
+      "epoch": 1.682068673166596,
+      "grad_norm": 1.6654925410640917,
+      "learning_rate": 6.708390046059402e-05,
+      "loss": 0.6897,
+      "step": 248
+    },
+    {
+      "epoch": 1.688851208139042,
+      "grad_norm": 1.1695940537064082,
+      "learning_rate": 6.694369330320563e-05,
+      "loss": 0.6894,
+      "step": 249
+    },
+    {
+      "epoch": 1.6956337431114878,
+      "grad_norm": 1.790433077118171,
+      "learning_rate": 6.680287751612576e-05,
+      "loss": 0.6965,
+      "step": 250
+    },
+    {
+      "epoch": 1.702416278083934,
+      "grad_norm": 1.3940506254821923,
+      "learning_rate": 6.666145628023495e-05,
+      "loss": 0.6883,
+      "step": 251
+    },
+    {
+      "epoch": 1.7091988130563798,
+      "grad_norm": 1.5788853280407766,
+      "learning_rate": 6.651943279009015e-05,
+      "loss": 0.674,
+      "step": 252
+    },
+    {
+      "epoch": 1.7159813480288257,
+      "grad_norm": 1.135684283004857,
+      "learning_rate": 6.637681025385261e-05,
+      "loss": 0.6944,
+      "step": 253
+    },
+    {
+      "epoch": 1.7227638830012717,
+      "grad_norm": 1.4839512076861858,
+      "learning_rate": 6.62335918932154e-05,
+      "loss": 0.6752,
+      "step": 254
+    },
+    {
+      "epoch": 1.7295464179737177,
+      "grad_norm": 1.3150087814531235,
+      "learning_rate": 6.608978094333068e-05,
+      "loss": 0.6826,
+      "step": 255
+    },
+    {
+      "epoch": 1.7363289529461636,
+      "grad_norm": 1.0720984172971046,
+      "learning_rate": 6.594538065273652e-05,
+      "loss": 0.683,
+      "step": 256
+    },
+    {
+      "epoch": 1.7431114879186096,
+      "grad_norm": 1.46543732407952,
+      "learning_rate": 6.58003942832836e-05,
+      "loss": 0.6796,
+      "step": 257
+    },
+    {
+      "epoch": 1.7498940228910556,
+      "grad_norm": 0.9292343800735161,
+      "learning_rate": 6.56548251100615e-05,
+      "loss": 0.6686,
+      "step": 258
+    },
+    {
+      "epoch": 1.7566765578635015,
+      "grad_norm": 1.5664491391186621,
+      "learning_rate": 6.550867642132474e-05,
+      "loss": 0.6912,
+      "step": 259
+    },
+    {
+      "epoch": 1.7634590928359475,
+      "grad_norm": 0.9821764593726072,
+      "learning_rate": 6.536195151841847e-05,
+      "loss": 0.6702,
+      "step": 260
+    },
+    {
+      "epoch": 1.7702416278083932,
+      "grad_norm": 1.4298526249936745,
+      "learning_rate": 6.521465371570393e-05,
+      "loss": 0.6871,
+      "step": 261
+    },
+    {
+      "epoch": 1.7770241627808394,
+      "grad_norm": 1.2452682220878895,
+      "learning_rate": 6.506678634048353e-05,
+      "loss": 0.6677,
+      "step": 262
+    },
+    {
+      "epoch": 1.7838066977532852,
+      "grad_norm": 1.4920349233263195,
+      "learning_rate": 6.491835273292574e-05,
+      "loss": 0.6809,
+      "step": 263
+    },
+    {
+      "epoch": 1.7905892327257313,
+      "grad_norm": 1.212833170547237,
+      "learning_rate": 6.476935624598966e-05,
+      "loss": 0.6809,
+      "step": 264
+    },
+    {
+      "epoch": 1.797371767698177,
+      "grad_norm": 1.250429726318394,
+      "learning_rate": 6.461980024534918e-05,
+      "loss": 0.674,
+      "step": 265
+    },
+    {
+      "epoch": 1.8041543026706233,
+      "grad_norm": 1.0681520589021245,
+      "learning_rate": 6.446968810931707e-05,
+      "loss": 0.6767,
+      "step": 266
+    },
+    {
+      "epoch": 1.810936837643069,
+      "grad_norm": 1.470191760394408,
+      "learning_rate": 6.431902322876855e-05,
+      "loss": 0.6737,
+      "step": 267
+    },
+    {
+      "epoch": 1.8177193726155152,
+      "grad_norm": 1.1749343585137977,
+      "learning_rate": 6.416780900706484e-05,
+      "loss": 0.683,
+      "step": 268
+    },
+    {
+      "epoch": 1.824501907587961,
+      "grad_norm": 1.1727141312870375,
+      "learning_rate": 6.401604885997614e-05,
+      "loss": 0.6808,
+      "step": 269
+    },
+    {
+      "epoch": 1.831284442560407,
+      "grad_norm": 1.3495790960778822,
+      "learning_rate": 6.386374621560455e-05,
+      "loss": 0.6847,
+      "step": 270
+    },
+    {
+      "epoch": 1.8380669775328529,
+      "grad_norm": 1.075708125802829,
+      "learning_rate": 6.37109045143066e-05,
+      "loss": 0.676,
+      "step": 271
+    },
+    {
+      "epoch": 1.8448495125052988,
+      "grad_norm": 1.1555696102312405,
+      "learning_rate": 6.355752720861559e-05,
+      "loss": 0.6661,
+      "step": 272
+    },
+    {
+      "epoch": 1.8516320474777448,
+      "grad_norm": 1.383594134484124,
+      "learning_rate": 6.340361776316349e-05,
+      "loss": 0.6784,
+      "step": 273
+    },
+    {
+      "epoch": 1.8584145824501908,
+      "grad_norm": 1.063837234136454,
+      "learning_rate": 6.324917965460279e-05,
+      "loss": 0.6781,
+      "step": 274
+    },
+    {
+      "epoch": 1.8651971174226367,
+      "grad_norm": 1.173609581411256,
+      "learning_rate": 6.309421637152794e-05,
+      "loss": 0.6682,
+      "step": 275
+    },
+    {
+      "epoch": 1.8719796523950827,
+      "grad_norm": 1.0658200919149257,
+      "learning_rate": 6.29387314143965e-05,
+      "loss": 0.6878,
+      "step": 276
+    },
+    {
+      "epoch": 1.8787621873675286,
+      "grad_norm": 1.7400399121218273,
+      "learning_rate": 6.278272829545011e-05,
+      "loss": 0.6706,
+      "step": 277
+    },
+    {
+      "epoch": 1.8855447223399746,
+      "grad_norm": 0.7229472871914281,
+      "learning_rate": 6.26262105386351e-05,
+      "loss": 0.6847,
+      "step": 278
+    },
+    {
+      "epoch": 1.8923272573124206,
+      "grad_norm": 1.0742091230131698,
+      "learning_rate": 6.246918167952304e-05,
+      "loss": 0.6756,
+      "step": 279
+    },
+    {
+      "epoch": 1.8991097922848663,
+      "grad_norm": 2.3846707594542456,
+      "learning_rate": 6.231164526523063e-05,
+      "loss": 0.6776,
+      "step": 280
+    },
+    {
+      "epoch": 1.9058923272573125,
+      "grad_norm": 1.181568933400366,
+      "learning_rate": 6.21536048543398e-05,
+      "loss": 0.686,
+      "step": 281
+    },
+    {
+      "epoch": 1.9126748622297582,
+      "grad_norm": 3.197777220254654,
+      "learning_rate": 6.199506401681718e-05,
+      "loss": 0.7007,
+      "step": 282
+    },
+    {
+      "epoch": 1.9194573972022044,
+      "grad_norm": 2.538862401423035,
+      "learning_rate": 6.183602633393352e-05,
+      "loss": 0.7031,
+      "step": 283
+    },
+    {
+      "epoch": 1.9262399321746502,
+      "grad_norm": 1.8353914421208335,
+      "learning_rate": 6.167649539818284e-05,
+      "loss": 0.6663,
+      "step": 284
+    },
+    {
+      "epoch": 1.9330224671470964,
+      "grad_norm": 1.8245299177531569,
+      "learning_rate": 6.151647481320114e-05,
+      "loss": 0.7011,
+      "step": 285
+    },
+    {
+      "epoch": 1.939805002119542,
+      "grad_norm": 2.284062594732862,
+      "learning_rate": 6.135596819368512e-05,
+      "loss": 0.701,
+      "step": 286
+    },
+    {
+      "epoch": 1.9465875370919883,
+      "grad_norm": 1.5939451814656143,
+      "learning_rate": 6.119497916531053e-05,
+      "loss": 0.6898,
+      "step": 287
+    },
+    {
+      "epoch": 1.953370072064434,
+      "grad_norm": 2.382128196264195,
+      "learning_rate": 6.103351136465014e-05,
+      "loss": 0.689,
+      "step": 288
+    },
+    {
+      "epoch": 1.96015260703688,
+      "grad_norm": 2.255809021386828,
+      "learning_rate": 6.0871568439091743e-05,
+      "loss": 0.689,
+      "step": 289
+    },
+    {
+      "epoch": 1.966935142009326,
+      "grad_norm": 1.192524289628154,
+      "learning_rate": 6.070915404675571e-05,
+      "loss": 0.6709,
+      "step": 290
+    },
+    {
+      "epoch": 1.973717676981772,
+      "grad_norm": 1.2397061535202811,
+      "learning_rate": 6.0546271856412306e-05,
+      "loss": 0.687,
+      "step": 291
+    },
+    {
+      "epoch": 1.9805002119542179,
+      "grad_norm": 0.9812283662135952,
+      "learning_rate": 6.03829255473989e-05,
+      "loss": 0.677,
+      "step": 292
+    },
+    {
+      "epoch": 1.9872827469266638,
+      "grad_norm": 1.410310313066498,
+      "learning_rate": 6.0219118809536794e-05,
+      "loss": 0.6882,
+      "step": 293
+    },
+    {
+      "epoch": 1.9940652818991098,
+      "grad_norm": 0.7865238012992424,
+      "learning_rate": 6.0054855343047914e-05,
+      "loss": 0.6769,
+      "step": 294
+    },
+    {
+      "epoch": 2.0008478168715556,
+      "grad_norm": 1.9901508305957583,
+      "learning_rate": 5.989013885847117e-05,
+      "loss": 1.0693,
+      "step": 295
+    },
+    {
+      "epoch": 2.0076303518440017,
+      "grad_norm": 1.3663813100431552,
+      "learning_rate": 5.972497307657869e-05,
+      "loss": 0.6673,
+      "step": 296
+    },
+    {
+      "epoch": 2.0144128868164475,
+      "grad_norm": 1.1381480345523374,
+      "learning_rate": 5.955936172829179e-05,
+      "loss": 0.6676,
+      "step": 297
+    },
+    {
+      "epoch": 2.0211954217888937,
+      "grad_norm": 1.496250854193103,
+      "learning_rate": 5.939330855459661e-05,
+      "loss": 0.6551,
+      "step": 298
+    },
+    {
+      "epoch": 2.0279779567613394,
+      "grad_norm": 0.9716458462062207,
+      "learning_rate": 5.922681730645968e-05,
+      "loss": 0.6723,
+      "step": 299
+    },
+    {
+      "epoch": 2.0347604917337856,
+      "grad_norm": 1.501595368602735,
+      "learning_rate": 5.905989174474319e-05,
+      "loss": 0.6509,
+      "step": 300
+    },
+    {
+      "epoch": 2.0415430267062313,
+      "grad_norm": 0.849974962385986,
+      "learning_rate": 5.889253564011999e-05,
+      "loss": 0.6577,
+      "step": 301
+    },
+    {
+      "epoch": 2.0483255616786775,
+      "grad_norm": 1.418902591957157,
+      "learning_rate": 5.872475277298847e-05,
+      "loss": 0.6665,
+      "step": 302
+    },
+    {
+      "epoch": 2.0551080966511233,
+      "grad_norm": 0.8446815446152741,
+      "learning_rate": 5.855654693338711e-05,
+      "loss": 0.6556,
+      "step": 303
+    },
+    {
+      "epoch": 2.0618906316235694,
+      "grad_norm": 1.3364133345670484,
+      "learning_rate": 5.838792192090889e-05,
+      "loss": 0.6604,
+      "step": 304
+    },
+    {
+      "epoch": 2.068673166596015,
+      "grad_norm": 0.8691215260428371,
+      "learning_rate": 5.821888154461549e-05,
+      "loss": 0.6593,
+      "step": 305
+    },
+    {
+      "epoch": 2.0754557015684614,
+      "grad_norm": 1.2207875769882877,
+      "learning_rate": 5.8049429622951194e-05,
+      "loss": 0.6637,
+      "step": 306
+    },
+    {
+      "epoch": 2.082238236540907,
+      "grad_norm": 0.7775512997267359,
+      "learning_rate": 5.7879569983656694e-05,
+      "loss": 0.6433,
+      "step": 307
+    },
+    {
+      "epoch": 2.0890207715133533,
+      "grad_norm": 1.2237758983601073,
+      "learning_rate": 5.770930646368257e-05,
+      "loss": 0.6681,
+      "step": 308
+    },
+    {
+      "epoch": 2.095803306485799,
+      "grad_norm": 0.6935518063946065,
+      "learning_rate": 5.7538642909102654e-05,
+      "loss": 0.6571,
+      "step": 309
+    },
+    {
+      "epoch": 2.1025858414582452,
+      "grad_norm": 1.0509844665628039,
+      "learning_rate": 5.736758317502714e-05,
+      "loss": 0.6462,
+      "step": 310
+    },
+    {
+      "epoch": 2.109368376430691,
+      "grad_norm": 0.6501336655556583,
+      "learning_rate": 5.719613112551546e-05,
+      "loss": 0.642,
+      "step": 311
+    },
+    {
+      "epoch": 2.1161509114031367,
+      "grad_norm": 0.791518345366507,
+      "learning_rate": 5.702429063348912e-05,
+      "loss": 0.6523,
+      "step": 312
+    },
+    {
+      "epoch": 2.122933446375583,
+      "grad_norm": 0.7216839703720191,
+      "learning_rate": 5.685206558064407e-05,
+      "loss": 0.6485,
+      "step": 313
+    },
+    {
+      "epoch": 2.1297159813480286,
+      "grad_norm": 1.3490427513390442,
+      "learning_rate": 5.667945985736308e-05,
+      "loss": 0.6622,
+      "step": 314
+    },
+    {
+      "epoch": 2.136498516320475,
+      "grad_norm": 1.036563799894598,
+      "learning_rate": 5.6506477362627926e-05,
+      "loss": 0.6598,
+      "step": 315
+    },
+    {
+      "epoch": 2.1432810512929206,
+      "grad_norm": 0.8799949127384937,
+      "learning_rate": 5.6333122003931186e-05,
+      "loss": 0.6631,
+      "step": 316
+    },
+    {
+      "epoch": 2.1500635862653668,
+      "grad_norm": 0.9027804053204243,
+      "learning_rate": 5.615939769718809e-05,
+      "loss": 0.6543,
+      "step": 317
+    },
+    {
+      "epoch": 2.1568461212378125,
+      "grad_norm": 1.08723379469273,
+      "learning_rate": 5.5985308366647985e-05,
+      "loss": 0.6453,
+      "step": 318
+    },
+    {
+      "epoch": 2.1636286562102587,
+      "grad_norm": 1.198788894798635,
+      "learning_rate": 5.5810857944805744e-05,
+      "loss": 0.6466,
+      "step": 319
+    },
+    {
+      "epoch": 2.1704111911827044,
+      "grad_norm": 1.2287340016478474,
+      "learning_rate": 5.5636050372312896e-05,
+      "loss": 0.6776,
+      "step": 320
+    },
+    {
+      "epoch": 2.1771937261551506,
+      "grad_norm": 0.676307375412045,
+      "learning_rate": 5.546088959788862e-05,
+      "loss": 0.6517,
+      "step": 321
+    },
+    {
+      "epoch": 2.1839762611275964,
+      "grad_norm": 0.8023647253320252,
+      "learning_rate": 5.528537957823061e-05,
+      "loss": 0.6565,
+      "step": 322
+    },
+    {
+      "epoch": 2.1907587961000425,
+      "grad_norm": 1.4391382554036933,
+      "learning_rate": 5.510952427792559e-05,
+      "loss": 0.6587,
+      "step": 323
+    },
+    {
+      "epoch": 2.1975413310724883,
+      "grad_norm": 0.7633580340639281,
+      "learning_rate": 5.493332766935981e-05,
+      "loss": 0.6433,
+      "step": 324
+    },
+    {
+      "epoch": 2.2043238660449345,
+      "grad_norm": 1.0303353819004128,
+      "learning_rate": 5.4756793732629335e-05,
+      "loss": 0.6578,
+      "step": 325
+    },
+    {
+      "epoch": 2.21110640101738,
+      "grad_norm": 1.3526690779961053,
+      "learning_rate": 5.45799264554501e-05,
+      "loss": 0.6499,
+      "step": 326
+    },
+    {
+      "epoch": 2.2178889359898264,
+      "grad_norm": 0.5228459526464206,
+      "learning_rate": 5.440272983306789e-05,
+      "loss": 0.6394,
+      "step": 327
+    },
+    {
+      "epoch": 2.224671470962272,
+      "grad_norm": 1.4252351086818837,
+      "learning_rate": 5.4225207868167994e-05,
+      "loss": 0.6554,
+      "step": 328
+    },
+    {
+      "epoch": 2.231454005934718,
+      "grad_norm": 0.6950731562608328,
+      "learning_rate": 5.404736457078489e-05,
+      "loss": 0.6482,
+      "step": 329
+    },
+    {
+      "epoch": 2.238236540907164,
+      "grad_norm": 0.7644651875362564,
+      "learning_rate": 5.38692039582116e-05,
+      "loss": 0.6503,
+      "step": 330
+    },
+    {
+      "epoch": 2.24501907587961,
+      "grad_norm": 0.9023571857971306,
+      "learning_rate": 5.3690730054908985e-05,
+      "loss": 0.6468,
+      "step": 331
+    },
+    {
+      "epoch": 2.251801610852056,
+      "grad_norm": 0.9107802448508358,
+      "learning_rate": 5.3511946892414775e-05,
+      "loss": 0.6518,
+      "step": 332
+    },
+    {
+      "epoch": 2.2585841458245017,
+      "grad_norm": 1.1127754837574242,
+      "learning_rate": 5.333285850925256e-05,
+      "loss": 0.6565,
+      "step": 333
+    },
+    {
+      "epoch": 2.265366680796948,
+      "grad_norm": 0.8282253306938299,
+      "learning_rate": 5.315346895084056e-05,
+      "loss": 0.6564,
+      "step": 334
+    },
+    {
+      "epoch": 2.2721492157693937,
+      "grad_norm": 1.14130477616649,
+      "learning_rate": 5.297378226940019e-05,
+      "loss": 0.6516,
+      "step": 335
+    },
+    {
+      "epoch": 2.27893175074184,
+      "grad_norm": 1.0103971093744055,
+      "learning_rate": 5.279380252386461e-05,
+      "loss": 0.6626,
+      "step": 336
+    },
+    {
+      "epoch": 2.2857142857142856,
+      "grad_norm": 0.9536394369627204,
+      "learning_rate": 5.2613533779786945e-05,
+      "loss": 0.6663,
+      "step": 337
+    },
+    {
+      "epoch": 2.2924968206867318,
+      "grad_norm": 0.942528066117481,
+      "learning_rate": 5.243298010924852e-05,
+      "loss": 0.6408,
+      "step": 338
+    },
+    {
+      "epoch": 2.2992793556591775,
+      "grad_norm": 1.2117876187133245,
+      "learning_rate": 5.225214559076683e-05,
+      "loss": 0.6435,
+      "step": 339
+    },
+    {
+      "epoch": 2.3060618906316237,
+      "grad_norm": 0.8633776831955887,
+      "learning_rate": 5.207103430920345e-05,
+      "loss": 0.6395,
+      "step": 340
+    },
+    {
+      "epoch": 2.3128444256040694,
+      "grad_norm": 0.727531307349217,
+      "learning_rate": 5.1889650355671725e-05,
+      "loss": 0.654,
+      "step": 341
+    },
+    {
+      "epoch": 2.3196269605765156,
+      "grad_norm": 0.669365998637684,
+      "learning_rate": 5.1707997827444394e-05,
+      "loss": 0.6546,
+      "step": 342
+    },
+    {
+      "epoch": 2.3264094955489614,
+      "grad_norm": 0.8312703348077676,
+      "learning_rate": 5.152608082786098e-05,
+      "loss": 0.6646,
+      "step": 343
+    },
+    {
+      "epoch": 2.3331920305214076,
+      "grad_norm": 1.0187489492222126,
+      "learning_rate": 5.1343903466235174e-05,
+      "loss": 0.648,
+      "step": 344
+    },
+    {
+      "epoch": 2.3399745654938533,
+      "grad_norm": 1.034867027111884,
+      "learning_rate": 5.116146985776194e-05,
+      "loss": 0.6593,
+      "step": 345
+    },
+    {
+      "epoch": 2.3467571004662995,
+      "grad_norm": 1.1577447007405788,
+      "learning_rate": 5.0978784123424617e-05,
+      "loss": 0.6516,
+      "step": 346
+    },
+    {
+      "epoch": 2.3535396354387452,
+      "grad_norm": 1.0011140671455747,
+      "learning_rate": 5.0795850389901784e-05,
+      "loss": 0.6638,
+      "step": 347
+    },
+    {
+      "epoch": 2.360322170411191,
+      "grad_norm": 0.9415943625257757,
+      "learning_rate": 5.061267278947408e-05,
+      "loss": 0.6457,
+      "step": 348
+    },
+    {
+      "epoch": 2.367104705383637,
+      "grad_norm": 0.6919138650974885,
+      "learning_rate": 5.042925545993079e-05,
+      "loss": 0.6584,
+      "step": 349
+    },
+    {
+      "epoch": 2.373887240356083,
+      "grad_norm": 0.7266901542517143,
+      "learning_rate": 5.02456025444765e-05,
+      "loss": 0.6348,
+      "step": 350
+    },
+    {
+      "epoch": 2.380669775328529,
+      "grad_norm": 0.6635241827908179,
+      "learning_rate": 5.0061718191637394e-05,
+      "loss": 0.6574,
+      "step": 351
+    },
+    {
+      "epoch": 2.387452310300975,
+      "grad_norm": 0.5653141179051638,
+      "learning_rate": 4.987760655516757e-05,
+      "loss": 0.6502,
+      "step": 352
+    },
+    {
+      "epoch": 2.394234845273421,
+      "grad_norm": 0.7007173012785425,
+      "learning_rate": 4.9693271793955255e-05,
+      "loss": 0.6468,
+      "step": 353
+    },
+    {
+      "epoch": 2.4010173802458668,
+      "grad_norm": 0.5821543619857267,
+      "learning_rate": 4.95087180719288e-05,
+      "loss": 0.6486,
+      "step": 354
+    },
+    {
+      "epoch": 2.407799915218313,
+      "grad_norm": 0.4265984966323706,
+      "learning_rate": 4.9323949557962684e-05,
+      "loss": 0.6477,
+      "step": 355
+    },
+    {
+      "epoch": 2.4145824501907587,
+      "grad_norm": 0.6454773425594047,
+      "learning_rate": 4.913897042578327e-05,
+      "loss": 0.6479,
+      "step": 356
+    },
+    {
+      "epoch": 2.421364985163205,
+      "grad_norm": 0.615481868155692,
+      "learning_rate": 4.8953784853874624e-05,
+      "loss": 0.6605,
+      "step": 357
+    },
+    {
+      "epoch": 2.4281475201356506,
+      "grad_norm": 0.4923342474878763,
+      "learning_rate": 4.8768397025383996e-05,
+      "loss": 0.6414,
+      "step": 358
+    },
+    {
+      "epoch": 2.434930055108097,
+      "grad_norm": 0.4236872576222672,
+      "learning_rate": 4.858281112802745e-05,
+      "loss": 0.6526,
+      "step": 359
+    },
+    {
+      "epoch": 2.4417125900805425,
+      "grad_norm": 0.4376255004390753,
+      "learning_rate": 4.83970313539952e-05,
+      "loss": 0.6503,
+      "step": 360
+    },
+    {
+      "epoch": 2.4484951250529887,
+      "grad_norm": 0.7271305316785697,
+      "learning_rate": 4.821106189985693e-05,
+      "loss": 0.6447,
+      "step": 361
+    },
+    {
+      "epoch": 2.4552776600254345,
+      "grad_norm": 1.001429290369445,
+      "learning_rate": 4.8024906966467e-05,
+      "loss": 0.6566,
+      "step": 362
+    },
+    {
+      "epoch": 2.4620601949978806,
+      "grad_norm": 1.168304354121221,
+      "learning_rate": 4.783857075886956e-05,
+      "loss": 0.6614,
+      "step": 363
+    },
+    {
+      "epoch": 2.4688427299703264,
+      "grad_norm": 0.8021141929873027,
+      "learning_rate": 4.7652057486203525e-05,
+      "loss": 0.6446,
+      "step": 364
+    },
+    {
+      "epoch": 2.4756252649427726,
+      "grad_norm": 0.7075159940006046,
+      "learning_rate": 4.746537136160757e-05,
+      "loss": 0.6563,
+      "step": 365
+    },
+    {
+      "epoch": 2.4824077999152183,
+      "grad_norm": 0.7183111983007646,
+      "learning_rate": 4.727851660212487e-05,
+      "loss": 0.6502,
+      "step": 366
+    },
+    {
+      "epoch": 2.489190334887664,
+      "grad_norm": 0.755055422137936,
+      "learning_rate": 4.709149742860792e-05,
+      "loss": 0.6424,
+      "step": 367
+    },
+    {
+      "epoch": 2.4959728698601102,
+      "grad_norm": 0.7253802127810576,
+      "learning_rate": 4.690431806562311e-05,
+      "loss": 0.6496,
+      "step": 368
+    },
+    {
+      "epoch": 2.5027554048325564,
+      "grad_norm": 0.5125324963207099,
+      "learning_rate": 4.6716982741355386e-05,
+      "loss": 0.6533,
+      "step": 369
+    },
+    {
+      "epoch": 2.509537939805002,
+      "grad_norm": 0.5597998347086172,
+      "learning_rate": 4.652949568751267e-05,
+      "loss": 0.6402,
+      "step": 370
+    },
+    {
+      "epoch": 2.516320474777448,
+      "grad_norm": 0.7015467668995325,
+      "learning_rate": 4.63418611392303e-05,
+      "loss": 0.6482,
+      "step": 371
+    },
+    {
+      "epoch": 2.523103009749894,
+      "grad_norm": 0.511973221694181,
+      "learning_rate": 4.615408333497538e-05,
+      "loss": 0.6522,
+      "step": 372
+    },
+    {
+      "epoch": 2.52988554472234,
+      "grad_norm": 0.4401280437938751,
+      "learning_rate": 4.5966166516450985e-05,
+      "loss": 0.6573,
+      "step": 373
+    },
+    {
+      "epoch": 2.536668079694786,
+      "grad_norm": 0.614469648497891,
+      "learning_rate": 4.577811492850039e-05,
+      "loss": 0.6598,
+      "step": 374
+    },
+    {
+      "epoch": 2.5434506146672318,
+      "grad_norm": 0.7173941689992674,
+      "learning_rate": 4.558993281901116e-05,
+      "loss": 0.6465,
+      "step": 375
+    },
+    {
+      "epoch": 2.550233149639678,
+      "grad_norm": 0.6690409730578648,
+      "learning_rate": 4.540162443881922e-05,
+      "loss": 0.6488,
+      "step": 376
+    },
+    {
+      "epoch": 2.5570156846121237,
+      "grad_norm": 0.5848162533642309,
+      "learning_rate": 4.5213194041612824e-05,
+      "loss": 0.648,
+      "step": 377
+    },
+    {
+      "epoch": 2.56379821958457,
+      "grad_norm": 0.549981546670745,
+      "learning_rate": 4.5024645883836426e-05,
+      "loss": 0.6579,
+      "step": 378
+    },
+    {
+      "epoch": 2.5705807545570156,
+      "grad_norm": 0.556140263499365,
+      "learning_rate": 4.4835984224594586e-05,
+      "loss": 0.6621,
+      "step": 379
+    },
+    {
+      "epoch": 2.577363289529462,
+      "grad_norm": 0.5195224064491745,
+      "learning_rate": 4.464721332555577e-05,
+      "loss": 0.6512,
+      "step": 380
+    },
+    {
+      "epoch": 2.5841458245019076,
+      "grad_norm": 0.38259318048535124,
+      "learning_rate": 4.445833745085602e-05,
+      "loss": 0.6492,
+      "step": 381
+    },
+    {
+      "epoch": 2.5909283594743533,
+      "grad_norm": 0.3534037726943067,
+      "learning_rate": 4.4269360867002675e-05,
+      "loss": 0.6589,
+      "step": 382
+    },
+    {
+      "epoch": 2.5977108944467995,
+      "grad_norm": 0.315508715276319,
+      "learning_rate": 4.408028784277799e-05,
+      "loss": 0.6493,
+      "step": 383
+    },
+    {
+      "epoch": 2.6044934294192457,
+      "grad_norm": 0.3298656413250479,
+      "learning_rate": 4.389112264914273e-05,
+      "loss": 0.6589,
+      "step": 384
+    },
+    {
+      "epoch": 2.6112759643916914,
+      "grad_norm": 0.3016047037119661,
+      "learning_rate": 4.370186955913962e-05,
+      "loss": 0.6561,
+      "step": 385
+    },
+    {
+      "epoch": 2.618058499364137,
+      "grad_norm": 0.34845242594318765,
+      "learning_rate": 4.351253284779692e-05,
+      "loss": 0.6552,
+      "step": 386
+    },
+    {
+      "epoch": 2.6248410343365833,
+      "grad_norm": 0.38736846664169255,
+      "learning_rate": 4.332311679203177e-05,
+      "loss": 0.6364,
+      "step": 387
+    },
+    {
+      "epoch": 2.6316235693090295,
+      "grad_norm": 0.3951534285172612,
+      "learning_rate": 4.313362567055367e-05,
+      "loss": 0.6502,
+      "step": 388
+    },
+    {
+      "epoch": 2.6384061042814753,
+      "grad_norm": 0.4657313097001591,
+      "learning_rate": 4.294406376376771e-05,
+      "loss": 0.6561,
+      "step": 389
+    },
+    {
+      "epoch": 2.645188639253921,
+      "grad_norm": 0.4131171546046392,
+      "learning_rate": 4.2754435353677985e-05,
+      "loss": 0.6567,
+      "step": 390
+    },
+    {
+      "epoch": 2.651971174226367,
+      "grad_norm": 0.5305329556333895,
+      "learning_rate": 4.2564744723790835e-05,
+      "loss": 0.6568,
+      "step": 391
+    },
+    {
+      "epoch": 2.658753709198813,
+      "grad_norm": 0.7002749870897172,
+      "learning_rate": 4.237499615901805e-05,
+      "loss": 0.6552,
+      "step": 392
+    },
+    {
+      "epoch": 2.665536244171259,
+      "grad_norm": 0.7839360975679,
+      "learning_rate": 4.218519394558013e-05,
+      "loss": 0.6333,
+      "step": 393
+    },
+    {
+      "epoch": 2.672318779143705,
+      "grad_norm": 0.7621551938800204,
+      "learning_rate": 4.199534237090943e-05,
+      "loss": 0.6501,
+      "step": 394
+    },
+    {
+      "epoch": 2.679101314116151,
+      "grad_norm": 0.6652202397867919,
+      "learning_rate": 4.1805445723553346e-05,
+      "loss": 0.648,
+      "step": 395
+    },
+    {
+      "epoch": 2.685883849088597,
+      "grad_norm": 0.4811941000709009,
+      "learning_rate": 4.1615508293077394e-05,
+      "loss": 0.6544,
+      "step": 396
+    },
+    {
+      "epoch": 2.692666384061043,
+      "grad_norm": 0.3387872066498026,
+      "learning_rate": 4.142553436996834e-05,
+      "loss": 0.6558,
+      "step": 397
+    },
+    {
+      "epoch": 2.6994489190334887,
+      "grad_norm": 0.28441864482912343,
+      "learning_rate": 4.12355282455373e-05,
+      "loss": 0.6354,
+      "step": 398
+    },
+    {
+      "epoch": 2.706231454005935,
+      "grad_norm": 0.285244776358457,
+      "learning_rate": 4.1045494211822756e-05,
+      "loss": 0.6493,
+      "step": 399
+    },
+    {
+      "epoch": 2.7130139889783806,
+      "grad_norm": 0.3573803066995761,
+      "learning_rate": 4.085543656149365e-05,
+      "loss": 0.6397,
+      "step": 400
+    },
+    {
+      "epoch": 2.7197965239508264,
+      "grad_norm": 0.4278142282008419,
+      "learning_rate": 4.0665359587752394e-05,
+      "loss": 0.6537,
+      "step": 401
+    },
+    {
+      "epoch": 2.7265790589232726,
+      "grad_norm": 0.49944968397215816,
+      "learning_rate": 4.04752675842379e-05,
+      "loss": 0.656,
+      "step": 402
+    },
+    {
+      "epoch": 2.7333615938957188,
+      "grad_norm": 0.4642912024112797,
+      "learning_rate": 4.028516484492857e-05,
+      "loss": 0.6424,
+      "step": 403
+    },
+    {
+      "epoch": 2.7401441288681645,
+      "grad_norm": 0.40098112979297407,
+      "learning_rate": 4.009505566404535e-05,
+      "loss": 0.6377,
+      "step": 404
+    },
+    {
+      "epoch": 2.7469266638406102,
+      "grad_norm": 0.3708879898016676,
+      "learning_rate": 3.990494433595466e-05,
+      "loss": 0.6518,
+      "step": 405
+    },
+    {
+      "epoch": 2.7537091988130564,
+      "grad_norm": 0.40008123571174364,
+      "learning_rate": 3.9714835155071435e-05,
+      "loss": 0.6536,
+      "step": 406
+    },
+    {
+      "epoch": 2.760491733785502,
+      "grad_norm": 0.29605943254539613,
+      "learning_rate": 3.952473241576211e-05,
+      "loss": 0.6583,
+      "step": 407
+    },
+    {
+      "epoch": 2.7672742687579484,
+      "grad_norm": 0.3406525853896961,
+      "learning_rate": 3.933464041224761e-05,
+      "loss": 0.6545,
+      "step": 408
+    },
+    {
+      "epoch": 2.774056803730394,
+      "grad_norm": 0.4722405867001987,
+      "learning_rate": 3.914456343850637e-05,
+      "loss": 0.6513,
+      "step": 409
+    },
+    {
+      "epoch": 2.7808393387028403,
+      "grad_norm": 0.5263658831108335,
+      "learning_rate": 3.895450578817727e-05,
+      "loss": 0.6548,
+      "step": 410
+    },
+    {
+      "epoch": 2.787621873675286,
+      "grad_norm": 0.5203524120767699,
+      "learning_rate": 3.8764471754462714e-05,
+      "loss": 0.654,
+      "step": 411
+    },
+    {
+      "epoch": 2.794404408647732,
+      "grad_norm": 0.3902260359882798,
+      "learning_rate": 3.857446563003167e-05,
+      "loss": 0.651,
+      "step": 412
+    },
+    {
+      "epoch": 2.801186943620178,
+      "grad_norm": 0.3931366323184314,
+      "learning_rate": 3.838449170692262e-05,
+      "loss": 0.6418,
+      "step": 413
+    },
+    {
+      "epoch": 2.807969478592624,
+      "grad_norm": 0.4496897165040769,
+      "learning_rate": 3.819455427644666e-05,
+      "loss": 0.6566,
+      "step": 414
+    },
+    {
+      "epoch": 2.81475201356507,
+      "grad_norm": 0.4843275111384109,
+      "learning_rate": 3.800465762909057e-05,
+      "loss": 0.6494,
+      "step": 415
+    },
+    {
+      "epoch": 2.821534548537516,
+      "grad_norm": 0.4284479617853188,
+      "learning_rate": 3.781480605441989e-05,
+      "loss": 0.6504,
+      "step": 416
+    },
+    {
+      "epoch": 2.828317083509962,
+      "grad_norm": 0.3505752717784012,
+      "learning_rate": 3.7625003840981976e-05,
+      "loss": 0.6515,
+      "step": 417
+    },
+    {
+      "epoch": 2.835099618482408,
+      "grad_norm": 0.2739600638831855,
+      "learning_rate": 3.743525527620918e-05,
+      "loss": 0.6482,
+      "step": 418
+    },
+    {
+      "epoch": 2.8418821534548537,
+      "grad_norm": 0.2821715826512274,
+      "learning_rate": 3.724556464632203e-05,
+      "loss": 0.6558,
+      "step": 419
+    },
+    {
+      "epoch": 2.8486646884272995,
+      "grad_norm": 0.3001235929581994,
+      "learning_rate": 3.7055936236232296e-05,
+      "loss": 0.6283,
+      "step": 420
+    },
+    {
+      "epoch": 2.8554472233997457,
+      "grad_norm": 0.34084577684820244,
+      "learning_rate": 3.686637432944634e-05,
+      "loss": 0.6524,
+      "step": 421
+    },
+    {
+      "epoch": 2.862229758372192,
+      "grad_norm": 0.31354392380727714,
+      "learning_rate": 3.6676883207968226e-05,
+      "loss": 0.6407,
+      "step": 422
+    },
+    {
+      "epoch": 2.8690122933446376,
+      "grad_norm": 0.29938503730372684,
+      "learning_rate": 3.648746715220309e-05,
+      "loss": 0.6431,
+      "step": 423
+    },
+    {
+      "epoch": 2.8757948283170833,
+      "grad_norm": 0.292588496703024,
+      "learning_rate": 3.6298130440860394e-05,
+      "loss": 0.6457,
+      "step": 424
+    },
+    {
+      "epoch": 2.8825773632895295,
+      "grad_norm": 0.34211158982130174,
+      "learning_rate": 3.6108877350857296e-05,
+      "loss": 0.6413,
+      "step": 425
+    },
+    {
+      "epoch": 2.8893598982619753,
+      "grad_norm": 0.28453013315503256,
+      "learning_rate": 3.5919712157222014e-05,
+      "loss": 0.6493,
+      "step": 426
+    },
+    {
+      "epoch": 2.8961424332344214,
+      "grad_norm": 0.25000620848294364,
+      "learning_rate": 3.573063913299733e-05,
+      "loss": 0.6471,
+      "step": 427
+    },
+    {
+      "epoch": 2.902924968206867,
+      "grad_norm": 0.32984361513817445,
+      "learning_rate": 3.554166254914399e-05,
+      "loss": 0.6525,
+      "step": 428
+    },
+    {
+      "epoch": 2.9097075031793134,
+      "grad_norm": 0.34282163750459993,
+      "learning_rate": 3.535278667444423e-05,
+      "loss": 0.6435,
+      "step": 429
+    },
+    {
+      "epoch": 2.916490038151759,
+      "grad_norm": 0.3376932266610861,
+      "learning_rate": 3.5164015775405414e-05,
+      "loss": 0.6537,
+      "step": 430
+    },
+    {
+      "epoch": 2.9232725731242053,
+      "grad_norm": 0.2736761009463742,
+      "learning_rate": 3.4975354116163594e-05,
+      "loss": 0.6528,
+      "step": 431
+    },
+    {
+      "epoch": 2.930055108096651,
+      "grad_norm": 0.28401811879767763,
+      "learning_rate": 3.478680595838719e-05,
+      "loss": 0.6552,
+      "step": 432
+    },
+    {
+      "epoch": 2.9368376430690972,
+      "grad_norm": 0.21207447278197997,
+      "learning_rate": 3.4598375561180783e-05,
+      "loss": 0.6477,
+      "step": 433
+    },
+    {
+      "epoch": 2.943620178041543,
+      "grad_norm": 0.3273666184303206,
+      "learning_rate": 3.441006718098885e-05,
+      "loss": 0.6415,
+      "step": 434
+    },
+    {
+      "epoch": 2.9504027130139887,
+      "grad_norm": 0.3922599048141651,
+      "learning_rate": 3.422188507149962e-05,
+      "loss": 0.66,
+      "step": 435
+    },
+    {
+      "epoch": 2.957185247986435,
+      "grad_norm": 0.33853238933941526,
+      "learning_rate": 3.403383348354902e-05,
+      "loss": 0.6392,
+      "step": 436
+    },
+    {
+      "epoch": 2.963967782958881,
+      "grad_norm": 0.2904409798403406,
+      "learning_rate": 3.3845916665024626e-05,
+      "loss": 0.6571,
+      "step": 437
+    },
+    {
+      "epoch": 2.970750317931327,
+      "grad_norm": 0.25650378002806623,
+      "learning_rate": 3.36581388607697e-05,
+      "loss": 0.658,
+      "step": 438
+    },
+    {
+      "epoch": 2.9775328529037726,
+      "grad_norm": 0.30516122417008495,
+      "learning_rate": 3.347050431248735e-05,
+      "loss": 0.6445,
+      "step": 439
+    },
+    {
+      "epoch": 2.9843153878762188,
+      "grad_norm": 0.35193441095679195,
+      "learning_rate": 3.3283017258644634e-05,
+      "loss": 0.6514,
+      "step": 440
+    },
+    {
+      "epoch": 2.991097922848665,
+      "grad_norm": 0.26810523993687996,
+      "learning_rate": 3.30956819343769e-05,
+      "loss": 0.6479,
+      "step": 441
+    },
+    {
+      "epoch": 2.9978804578211107,
+      "grad_norm": 0.40136478388949387,
+      "learning_rate": 3.290850257139209e-05,
+      "loss": 0.7446,
+      "step": 442
+    },
+    {
+      "epoch": 3.0046629927935564,
+      "grad_norm": 0.6219962321725231,
+      "learning_rate": 3.2721483397875135e-05,
+      "loss": 0.9011,
+      "step": 443
+    },
+    {
+      "epoch": 3.0114455277660026,
+      "grad_norm": 0.6301524386828496,
+      "learning_rate": 3.253462863839243e-05,
+      "loss": 0.6098,
+      "step": 444
+    },
+    {
+      "epoch": 3.0182280627384483,
+      "grad_norm": 0.5440271202067795,
+      "learning_rate": 3.2347942513796475e-05,
+      "loss": 0.6367,
+      "step": 445
+    },
+    {
+      "epoch": 3.0250105977108945,
+      "grad_norm": 0.5386656826635444,
+      "learning_rate": 3.216142924113046e-05,
+      "loss": 0.6285,
+      "step": 446
+    },
+    {
+      "epoch": 3.0317931326833403,
+      "grad_norm": 0.4074410048009017,
+      "learning_rate": 3.1975093033533003e-05,
+      "loss": 0.6332,
+      "step": 447
+    },
+    {
+      "epoch": 3.0385756676557865,
+      "grad_norm": 0.25928533106144735,
+      "learning_rate": 3.1788938100143086e-05,
+      "loss": 0.6246,
+      "step": 448
+    },
+    {
+      "epoch": 3.045358202628232,
+      "grad_norm": 0.445439070033727,
+      "learning_rate": 3.160296864600482e-05,
+      "loss": 0.6193,
+      "step": 449
+    },
+    {
+      "epoch": 3.0521407376006784,
+      "grad_norm": 0.46469489323619156,
+      "learning_rate": 3.1417188871972554e-05,
+      "loss": 0.6264,
+      "step": 450
+    },
+    {
+      "epoch": 3.058923272573124,
+      "grad_norm": 0.3272979337949089,
+      "learning_rate": 3.123160297461601e-05,
+      "loss": 0.631,
+      "step": 451
+    },
+    {
+      "epoch": 3.0657058075455703,
+      "grad_norm": 0.3178424540888323,
+      "learning_rate": 3.104621514612538e-05,
+      "loss": 0.6344,
+      "step": 452
+    },
+    {
+      "epoch": 3.072488342518016,
+      "grad_norm": 0.2754042730785633,
+      "learning_rate": 3.086102957421672e-05,
+      "loss": 0.6129,
+      "step": 453
+    },
+    {
+      "epoch": 3.0792708774904622,
+      "grad_norm": 0.35442538475012997,
+      "learning_rate": 3.0676050442037336e-05,
+      "loss": 0.6269,
+      "step": 454
+    },
+    {
+      "epoch": 3.086053412462908,
+      "grad_norm": 0.2797121425291907,
+      "learning_rate": 3.0491281928071217e-05,
+      "loss": 0.6221,
+      "step": 455
+    },
+    {
+      "epoch": 3.092835947435354,
+      "grad_norm": 0.23765468818900978,
+      "learning_rate": 3.030672820604476e-05,
+      "loss": 0.6256,
+      "step": 456
+    },
+    {
+      "epoch": 3.0996184824078,
+      "grad_norm": 0.26572260092618777,
+      "learning_rate": 3.012239344483244e-05,
+      "loss": 0.6177,
+      "step": 457
+    },
+    {
+      "epoch": 3.1064010173802457,
+      "grad_norm": 0.33390446968478465,
+      "learning_rate": 2.993828180836262e-05,
+      "loss": 0.6113,
+      "step": 458
+    },
+    {
+      "epoch": 3.113183552352692,
+      "grad_norm": 0.2666734723327277,
+      "learning_rate": 2.9754397455523497e-05,
+      "loss": 0.6297,
+      "step": 459
+    },
+    {
+      "epoch": 3.1199660873251376,
+      "grad_norm": 0.29084625313932944,
+      "learning_rate": 2.957074454006921e-05,
+      "loss": 0.6196,
+      "step": 460
+    },
+    {
+      "epoch": 3.1267486222975838,
+      "grad_norm": 0.19046989657100505,
+      "learning_rate": 2.9387327210525942e-05,
+      "loss": 0.6188,
+      "step": 461
+    },
+    {
+      "epoch": 3.1335311572700295,
+      "grad_norm": 0.338123523854364,
+      "learning_rate": 2.9204149610098223e-05,
+      "loss": 0.6281,
+      "step": 462
+    },
+    {
+      "epoch": 3.1403136922424757,
+      "grad_norm": 0.22618376500160342,
+      "learning_rate": 2.902121587657539e-05,
+      "loss": 0.6157,
+      "step": 463
+    },
+    {
+      "epoch": 3.1470962272149214,
+      "grad_norm": 0.35386494541824426,
+      "learning_rate": 2.8838530142238076e-05,
+      "loss": 0.626,
+      "step": 464
+    },
+    {
+      "epoch": 3.1538787621873676,
+      "grad_norm": 0.2840090224796156,
+      "learning_rate": 2.865609653376484e-05,
+      "loss": 0.6199,
+      "step": 465
+    },
+    {
+      "epoch": 3.1606612971598134,
+      "grad_norm": 0.2835278774397419,
+      "learning_rate": 2.8473919172139034e-05,
+      "loss": 0.635,
+      "step": 466
+    },
+    {
+      "epoch": 3.1674438321322596,
+      "grad_norm": 0.2932289544670752,
+      "learning_rate": 2.8292002172555616e-05,
+      "loss": 0.6181,
+      "step": 467
+    },
+    {
+      "epoch": 3.1742263671047053,
+      "grad_norm": 0.21388735557547445,
+      "learning_rate": 2.8110349644328275e-05,
+      "loss": 0.6304,
+      "step": 468
+    },
+    {
+      "epoch": 3.1810089020771515,
+      "grad_norm": 0.25484830423076227,
+      "learning_rate": 2.7928965690796562e-05,
+      "loss": 0.6351,
+      "step": 469
+    },
+    {
+      "epoch": 3.187791437049597,
+      "grad_norm": 0.24843127490388153,
+      "learning_rate": 2.7747854409233187e-05,
+      "loss": 0.6301,
+      "step": 470
+    },
+    {
+      "epoch": 3.1945739720220434,
+      "grad_norm": 0.2451683485321557,
+      "learning_rate": 2.7567019890751493e-05,
+      "loss": 0.6207,
+      "step": 471
+    },
+    {
+      "epoch": 3.201356506994489,
+      "grad_norm": 0.3164960123899796,
+      "learning_rate": 2.7386466220213065e-05,
+      "loss": 0.6213,
+      "step": 472
+    },
+    {
+      "epoch": 3.2081390419669353,
+      "grad_norm": 0.2697409403652239,
+      "learning_rate": 2.7206197476135403e-05,
+      "loss": 0.6212,
+      "step": 473
+    },
+    {
+      "epoch": 3.214921576939381,
+      "grad_norm": 0.19669860705803766,
+      "learning_rate": 2.7026217730599814e-05,
+      "loss": 0.6284,
+      "step": 474
+    },
+    {
+      "epoch": 3.2217041119118273,
+      "grad_norm": 0.23674235246224137,
+      "learning_rate": 2.6846531049159454e-05,
+      "loss": 0.627,
+      "step": 475
+    },
+    {
+      "epoch": 3.228486646884273,
+      "grad_norm": 0.18982523269613027,
+      "learning_rate": 2.6667141490747454e-05,
+      "loss": 0.6236,
+      "step": 476
+    },
+    {
+      "epoch": 3.2352691818567187,
+      "grad_norm": 0.20626545109123062,
+      "learning_rate": 2.648805310758524e-05,
+      "loss": 0.6242,
+      "step": 477
+    },
+    {
+      "epoch": 3.242051716829165,
+      "grad_norm": 0.25273617130117126,
+      "learning_rate": 2.6309269945091025e-05,
+      "loss": 0.6135,
+      "step": 478
+    },
+    {
+      "epoch": 3.2488342518016107,
+      "grad_norm": 0.19396420036714845,
+      "learning_rate": 2.6130796041788404e-05,
+      "loss": 0.6401,
+      "step": 479
+    },
+    {
+      "epoch": 3.255616786774057,
+      "grad_norm": 0.26449583344475813,
+      "learning_rate": 2.5952635429215117e-05,
+      "loss": 0.6303,
+      "step": 480
+    },
+    {
+      "epoch": 3.2623993217465026,
+      "grad_norm": 0.2880771386228117,
+      "learning_rate": 2.5774792131832012e-05,
+      "loss": 0.6213,
+      "step": 481
+    },
+    {
+      "epoch": 3.269181856718949,
+      "grad_norm": 0.19301524096425413,
+      "learning_rate": 2.559727016693212e-05,
+      "loss": 0.6368,
+      "step": 482
+    },
+    {
+      "epoch": 3.2759643916913945,
+      "grad_norm": 0.31002080783316865,
+      "learning_rate": 2.54200735445499e-05,
+      "loss": 0.641,
+      "step": 483
+    },
+    {
+      "epoch": 3.2827469266638407,
+      "grad_norm": 0.24500950182263187,
+      "learning_rate": 2.5243206267370685e-05,
+      "loss": 0.6204,
+      "step": 484
+    },
+    {
+      "epoch": 3.2895294616362865,
+      "grad_norm": 0.25623143489960465,
+      "learning_rate": 2.506667233064021e-05,
+      "loss": 0.6207,
+      "step": 485
+    },
+    {
+      "epoch": 3.2963119966087326,
+      "grad_norm": 0.2918652621084026,
+      "learning_rate": 2.489047572207442e-05,
+      "loss": 0.6371,
+      "step": 486
+    },
+    {
+      "epoch": 3.3030945315811784,
+      "grad_norm": 0.23337516028968874,
+      "learning_rate": 2.4714620421769394e-05,
+      "loss": 0.6285,
+      "step": 487
+    },
+    {
+      "epoch": 3.3098770665536246,
+      "grad_norm": 0.23619697217842073,
+      "learning_rate": 2.4539110402111383e-05,
+      "loss": 0.631,
+      "step": 488
+    },
+    {
+      "epoch": 3.3166596015260703,
+      "grad_norm": 0.2448512796334409,
+      "learning_rate": 2.436394962768712e-05,
+      "loss": 0.6276,
+      "step": 489
+    },
+    {
+      "epoch": 3.3234421364985165,
+      "grad_norm": 0.1983649853769521,
+      "learning_rate": 2.4189142055194276e-05,
+      "loss": 0.6244,
+      "step": 490
+    },
+    {
+      "epoch": 3.3302246714709622,
+      "grad_norm": 0.21190165986074258,
+      "learning_rate": 2.4014691633352038e-05,
+      "loss": 0.6101,
+      "step": 491
+    },
+    {
+      "epoch": 3.337007206443408,
+      "grad_norm": 0.25804827992388285,
+      "learning_rate": 2.3840602302811923e-05,
+      "loss": 0.6135,
+      "step": 492
+    },
+    {
+      "epoch": 3.343789741415854,
+      "grad_norm": 0.18856304220278483,
+      "learning_rate": 2.3666877996068824e-05,
+      "loss": 0.6288,
+      "step": 493
+    },
+    {
+      "epoch": 3.3505722763883004,
+      "grad_norm": 0.3301373974007395,
+      "learning_rate": 2.3493522637372087e-05,
+      "loss": 0.6326,
+      "step": 494
+    },
+    {
+      "epoch": 3.357354811360746,
+      "grad_norm": 0.30315622915732504,
+      "learning_rate": 2.332054014263692e-05,
+      "loss": 0.6259,
+      "step": 495
+    },
+    {
+      "epoch": 3.364137346333192,
+      "grad_norm": 0.22646576230809926,
+      "learning_rate": 2.3147934419355935e-05,
+      "loss": 0.6245,
+      "step": 496
+    },
+    {
+      "epoch": 3.370919881305638,
+      "grad_norm": 0.30590058038547674,
+      "learning_rate": 2.2975709366510887e-05,
+      "loss": 0.6316,
+      "step": 497
+    },
+    {
+      "epoch": 3.3777024162780838,
+      "grad_norm": 0.26486089548607467,
+      "learning_rate": 2.2803868874484536e-05,
+      "loss": 0.6383,
+      "step": 498
+    },
+    {
+      "epoch": 3.38448495125053,
+      "grad_norm": 0.209806274707419,
+      "learning_rate": 2.2632416824972886e-05,
+      "loss": 0.614,
+      "step": 499
+    },
+    {
+      "epoch": 3.3912674862229757,
+      "grad_norm": 0.3076979703172019,
+      "learning_rate": 2.2461357090897352e-05,
+      "loss": 0.6317,
+      "step": 500
+    },
+    {
+      "epoch": 3.398050021195422,
+      "grad_norm": 0.27851643544183735,
+      "learning_rate": 2.2290693536317444e-05,
+      "loss": 0.6226,
+      "step": 501
+    },
+    {
+      "epoch": 3.4048325561678676,
+      "grad_norm": 0.26720054263757376,
+      "learning_rate": 2.212043001634332e-05,
+      "loss": 0.6306,
+      "step": 502
+    },
+    {
+      "epoch": 3.411615091140314,
+      "grad_norm": 0.3264075780455984,
+      "learning_rate": 2.1950570377048813e-05,
+      "loss": 0.6144,
+      "step": 503
+    },
+    {
+      "epoch": 3.4183976261127595,
+      "grad_norm": 0.24476904503391836,
+      "learning_rate": 2.178111845538453e-05,
+      "loss": 0.6317,
+      "step": 504
+    },
+    {
+      "epoch": 3.4251801610852057,
+      "grad_norm": 0.3445192467182261,
+      "learning_rate": 2.1612078079091125e-05,
+      "loss": 0.6227,
+      "step": 505
+    },
+    {
+      "epoch": 3.4319626960576515,
+      "grad_norm": 0.2248390312896755,
+      "learning_rate": 2.144345306661291e-05,
+      "loss": 0.6179,
+      "step": 506
+    },
+    {
+      "epoch": 3.4387452310300977,
+      "grad_norm": 0.24580333907022087,
+      "learning_rate": 2.1275247227011536e-05,
+      "loss": 0.6218,
+      "step": 507
+    },
+    {
+      "epoch": 3.4455277660025434,
+      "grad_norm": 0.2333810028522329,
+      "learning_rate": 2.1107464359880022e-05,
+      "loss": 0.6313,
+      "step": 508
+    },
+    {
+      "epoch": 3.4523103009749896,
+      "grad_norm": 0.20038699343598237,
+      "learning_rate": 2.0940108255256823e-05,
+      "loss": 0.6309,
+      "step": 509
+    },
+    {
+      "epoch": 3.4590928359474353,
+      "grad_norm": 0.19291801763604194,
+      "learning_rate": 2.0773182693540325e-05,
+      "loss": 0.6076,
+      "step": 510
+    },
+    {
+      "epoch": 3.465875370919881,
+      "grad_norm": 0.21138321490252843,
+      "learning_rate": 2.0606691445403398e-05,
+      "loss": 0.6238,
+      "step": 511
+    },
+    {
+      "epoch": 3.4726579058923273,
+      "grad_norm": 0.2081459000808386,
+      "learning_rate": 2.0440638271708226e-05,
+      "loss": 0.6135,
+      "step": 512
+    },
+    {
+      "epoch": 3.4794404408647734,
+      "grad_norm": 0.21359018594893273,
+      "learning_rate": 2.0275026923421315e-05,
+      "loss": 0.6347,
+      "step": 513
+    },
+    {
+      "epoch": 3.486222975837219,
+      "grad_norm": 0.262459399100271,
+      "learning_rate": 2.0109861141528853e-05,
+      "loss": 0.6289,
+      "step": 514
+    },
+    {
+      "epoch": 3.493005510809665,
+      "grad_norm": 0.24248389999156902,
+      "learning_rate": 1.9945144656952103e-05,
+      "loss": 0.6128,
+      "step": 515
+    },
+    {
+      "epoch": 3.499788045782111,
+      "grad_norm": 0.22920015918246003,
+      "learning_rate": 1.9780881190463222e-05,
+      "loss": 0.634,
+      "step": 516
+    },
+    {
+      "epoch": 3.506570580754557,
+      "grad_norm": 0.2824421315404272,
+      "learning_rate": 1.9617074452601113e-05,
+      "loss": 0.6322,
+      "step": 517
+    },
+    {
+      "epoch": 3.513353115727003,
+      "grad_norm": 0.19432350726010572,
+      "learning_rate": 1.9453728143587697e-05,
+      "loss": 0.6231,
+      "step": 518
+    },
+    {
+      "epoch": 3.520135650699449,
+      "grad_norm": 0.2846932920722668,
+      "learning_rate": 1.9290845953244307e-05,
+      "loss": 0.625,
+      "step": 519
+    },
+    {
+      "epoch": 3.526918185671895,
+      "grad_norm": 0.16416357169798437,
+      "learning_rate": 1.9128431560908263e-05,
+      "loss": 0.6191,
+      "step": 520
+    },
+    {
+      "epoch": 3.5337007206443407,
+      "grad_norm": 0.2747654462569946,
+      "learning_rate": 1.896648863534988e-05,
+      "loss": 0.622,
+      "step": 521
+    },
+    {
+      "epoch": 3.540483255616787,
+      "grad_norm": 0.15903407751400953,
+      "learning_rate": 1.8805020834689487e-05,
+      "loss": 0.6194,
+      "step": 522
+    },
+    {
+      "epoch": 3.5472657905892326,
+      "grad_norm": 0.20365263598678185,
+      "learning_rate": 1.864403180631489e-05,
+      "loss": 0.6255,
+      "step": 523
+    },
+    {
+      "epoch": 3.554048325561679,
+      "grad_norm": 0.16301528679144703,
+      "learning_rate": 1.8483525186798875e-05,
+      "loss": 0.6253,
+      "step": 524
+    },
+    {
+      "epoch": 3.5608308605341246,
+      "grad_norm": 0.23098070191792286,
+      "learning_rate": 1.832350460181717e-05,
+      "loss": 0.62,
+      "step": 525
+    },
+    {
+      "epoch": 3.5676133955065703,
+      "grad_norm": 0.16673057537934008,
+      "learning_rate": 1.8163973666066473e-05,
+      "loss": 0.629,
+      "step": 526
+    },
+    {
+      "epoch": 3.5743959304790165,
+      "grad_norm": 0.19378211117923547,
+      "learning_rate": 1.8004935983182835e-05,
+      "loss": 0.6344,
+      "step": 527
+    },
+    {
+      "epoch": 3.5811784654514627,
+      "grad_norm": 0.17643236928174555,
+      "learning_rate": 1.784639514566021e-05,
+      "loss": 0.6284,
+      "step": 528
+    },
+    {
+      "epoch": 3.5879610004239084,
+      "grad_norm": 0.19805537148789543,
+      "learning_rate": 1.7688354734769378e-05,
+      "loss": 0.6245,
+      "step": 529
+    },
+    {
+      "epoch": 3.594743535396354,
+      "grad_norm": 0.17444243321183842,
+      "learning_rate": 1.753081832047697e-05,
+      "loss": 0.6241,
+      "step": 530
+    },
+    {
+      "epoch": 3.6015260703688003,
+      "grad_norm": 0.22629631276174578,
+      "learning_rate": 1.7373789461364904e-05,
+      "loss": 0.6296,
+      "step": 531
+    },
+    {
+      "epoch": 3.6083086053412465,
+      "grad_norm": 0.2322655718660276,
+      "learning_rate": 1.7217271704549913e-05,
+      "loss": 0.6335,
+      "step": 532
+    },
+    {
+      "epoch": 3.6150911403136923,
+      "grad_norm": 0.20701067571130669,
+      "learning_rate": 1.7061268585603507e-05,
+      "loss": 0.6307,
+      "step": 533
+    },
+    {
+      "epoch": 3.621873675286138,
+      "grad_norm": 0.2206959741176191,
+      "learning_rate": 1.6905783628472074e-05,
+      "loss": 0.6347,
+      "step": 534
+    },
+    {
+      "epoch": 3.628656210258584,
+      "grad_norm": 0.15861912044210094,
+      "learning_rate": 1.6750820345397217e-05,
+      "loss": 0.6265,
+      "step": 535
+    },
+    {
+      "epoch": 3.63543874523103,
+      "grad_norm": 0.2001185331683702,
+      "learning_rate": 1.659638223683653e-05,
+      "loss": 0.6287,
+      "step": 536
+    },
+    {
+      "epoch": 3.642221280203476,
+      "grad_norm": 0.1746493458635906,
+      "learning_rate": 1.644247279138442e-05,
+      "loss": 0.6361,
+      "step": 537
+    },
+    {
+      "epoch": 3.649003815175922,
+      "grad_norm": 0.17095450468813203,
+      "learning_rate": 1.628909548569339e-05,
+      "loss": 0.6251,
+      "step": 538
+    },
+    {
+      "epoch": 3.655786350148368,
+      "grad_norm": 0.15588769920603499,
+      "learning_rate": 1.6136253784395455e-05,
+      "loss": 0.629,
+      "step": 539
+    },
+    {
+      "epoch": 3.662568885120814,
+      "grad_norm": 0.14759859028229927,
+      "learning_rate": 1.5983951140023864e-05,
+      "loss": 0.6174,
+      "step": 540
+    },
+    {
+      "epoch": 3.66935142009326,
+      "grad_norm": 0.16662793387428904,
+      "learning_rate": 1.583219099293516e-05,
+      "loss": 0.6192,
+      "step": 541
+    },
+    {
+      "epoch": 3.6761339550657057,
+      "grad_norm": 0.16325631134586066,
+      "learning_rate": 1.568097677123146e-05,
+      "loss": 0.6193,
+      "step": 542
+    },
+    {
+      "epoch": 3.682916490038152,
+      "grad_norm": 0.1590144709587067,
+      "learning_rate": 1.5530311890682946e-05,
+      "loss": 0.6126,
+      "step": 543
+    },
+    {
+      "epoch": 3.6896990250105977,
+      "grad_norm": 0.16971629601446409,
+      "learning_rate": 1.5380199754650838e-05,
+      "loss": 0.6204,
+      "step": 544
+    },
+    {
+      "epoch": 3.6964815599830434,
+      "grad_norm": 0.17653934494907608,
+      "learning_rate": 1.5230643754010355e-05,
+      "loss": 0.634,
+      "step": 545
+    },
+    {
+      "epoch": 3.7032640949554896,
+      "grad_norm": 0.1602687042337523,
+      "learning_rate": 1.508164726707427e-05,
+      "loss": 0.6132,
+      "step": 546
+    },
+    {
+      "epoch": 3.7100466299279358,
+      "grad_norm": 0.1853250131338786,
+      "learning_rate": 1.4933213659516485e-05,
+      "loss": 0.6263,
+      "step": 547
+    },
+    {
+      "epoch": 3.7168291649003815,
+      "grad_norm": 0.16166905381502983,
+      "learning_rate": 1.4785346284296078e-05,
+      "loss": 0.6158,
+      "step": 548
+    },
+    {
+      "epoch": 3.7236116998728273,
+      "grad_norm": 0.19092582748564702,
+      "learning_rate": 1.4638048481581537e-05,
+      "loss": 0.6235,
+      "step": 549
+    },
+    {
+      "epoch": 3.7303942348452734,
+      "grad_norm": 0.14148685680959427,
+      "learning_rate": 1.4491323578675265e-05,
+      "loss": 0.6137,
+      "step": 550
+    },
+    {
+      "epoch": 3.7371767698177196,
+      "grad_norm": 0.18984864298487686,
+      "learning_rate": 1.4345174889938514e-05,
+      "loss": 0.6329,
+      "step": 551
+    },
+    {
+      "epoch": 3.7439593047901654,
+      "grad_norm": 0.14327461316917306,
+      "learning_rate": 1.4199605716716414e-05,
+      "loss": 0.6272,
+      "step": 552
+    },
+    {
+      "epoch": 3.750741839762611,
+      "grad_norm": 0.18666689301215467,
+      "learning_rate": 1.4054619347263487e-05,
+      "loss": 0.6363,
+      "step": 553
+    },
+    {
+      "epoch": 3.7575243747350573,
+      "grad_norm": 0.14946115608698116,
+      "learning_rate": 1.3910219056669335e-05,
+      "loss": 0.6218,
+      "step": 554
+    },
+    {
+      "epoch": 3.764306909707503,
+      "grad_norm": 0.17777432794193568,
+      "learning_rate": 1.3766408106784601e-05,
+      "loss": 0.6278,
+      "step": 555
+    },
+    {
+      "epoch": 3.771089444679949,
+      "grad_norm": 0.16174854488646243,
+      "learning_rate": 1.3623189746147398e-05,
+      "loss": 0.6227,
+      "step": 556
+    },
+    {
+      "epoch": 3.777871979652395,
+      "grad_norm": 0.15621831808343503,
+      "learning_rate": 1.3480567209909863e-05,
+      "loss": 0.6296,
+      "step": 557
+    },
+    {
+      "epoch": 3.784654514624841,
+      "grad_norm": 0.17960424688042897,
+      "learning_rate": 1.333854371976505e-05,
+      "loss": 0.6175,
+      "step": 558
+    },
+    {
+      "epoch": 3.791437049597287,
+      "grad_norm": 0.161165540194335,
+      "learning_rate": 1.319712248387424e-05,
+      "loss": 0.6259,
+      "step": 559
+    },
+    {
+      "epoch": 3.798219584569733,
+      "grad_norm": 0.18195946037024635,
+      "learning_rate": 1.3056306696794377e-05,
+      "loss": 0.6259,
+      "step": 560
+    },
+    {
+      "epoch": 3.805002119542179,
+      "grad_norm": 0.15299855099029444,
+      "learning_rate": 1.2916099539405997e-05,
+      "loss": 0.623,
+      "step": 561
+    },
+    {
+      "epoch": 3.811784654514625,
+      "grad_norm": 0.19532294279356358,
+      "learning_rate": 1.2776504178841291e-05,
+      "loss": 0.629,
+      "step": 562
+    },
+    {
+      "epoch": 3.8185671894870707,
+      "grad_norm": 0.1505952319707712,
+      "learning_rate": 1.2637523768412633e-05,
+      "loss": 0.6239,
+      "step": 563
+    },
+    {
+      "epoch": 3.8253497244595165,
+      "grad_norm": 0.15785614162676143,
+      "learning_rate": 1.249916144754133e-05,
+      "loss": 0.6213,
+      "step": 564
+    },
+    {
+      "epoch": 3.8321322594319627,
+      "grad_norm": 0.17665393013957048,
+      "learning_rate": 1.2361420341686641e-05,
+      "loss": 0.6205,
+      "step": 565
+    },
+    {
+      "epoch": 3.838914794404409,
+      "grad_norm": 0.16420702224755918,
+      "learning_rate": 1.22243035622753e-05,
+      "loss": 0.6221,
+      "step": 566
+    },
+    {
+      "epoch": 3.8456973293768546,
+      "grad_norm": 0.17607498450226258,
+      "learning_rate": 1.2087814206631095e-05,
+      "loss": 0.6212,
+      "step": 567
+    },
+    {
+      "epoch": 3.8524798643493003,
+      "grad_norm": 0.19592268500165855,
+      "learning_rate": 1.1951955357905033e-05,
+      "loss": 0.6199,
+      "step": 568
+    },
+    {
+      "epoch": 3.8592623993217465,
+      "grad_norm": 0.1402346539755977,
+      "learning_rate": 1.1816730085005603e-05,
+      "loss": 0.6198,
+      "step": 569
+    },
+    {
+      "epoch": 3.8660449342941927,
+      "grad_norm": 0.18099131317071607,
+      "learning_rate": 1.1682141442529473e-05,
+      "loss": 0.6256,
+      "step": 570
+    },
+    {
+      "epoch": 3.8728274692666385,
+      "grad_norm": 0.19545074620930133,
+      "learning_rate": 1.1548192470692516e-05,
+      "loss": 0.6157,
+      "step": 571
+    },
+    {
+      "epoch": 3.879610004239084,
+      "grad_norm": 0.15514995332077897,
+      "learning_rate": 1.1414886195261135e-05,
+      "loss": 0.6262,
+      "step": 572
+    },
+    {
+      "epoch": 3.8863925392115304,
+      "grad_norm": 0.1534732088259275,
+      "learning_rate": 1.128222562748384e-05,
+      "loss": 0.6218,
+      "step": 573
+    },
+    {
+      "epoch": 3.893175074183976,
+      "grad_norm": 0.16016994591208966,
+      "learning_rate": 1.1150213764023357e-05,
+      "loss": 0.6264,
+      "step": 574
+    },
+    {
+      "epoch": 3.8999576091564223,
+      "grad_norm": 0.1393505701166259,
+      "learning_rate": 1.1018853586888794e-05,
+      "loss": 0.6217,
+      "step": 575
+    },
+    {
+      "epoch": 3.906740144128868,
+      "grad_norm": 0.13613614708020336,
+      "learning_rate": 1.0888148063368411e-05,
+      "loss": 0.6271,
+      "step": 576
+    },
+    {
+      "epoch": 3.9135226791013142,
+      "grad_norm": 0.13989845925554537,
+      "learning_rate": 1.075810014596248e-05,
+      "loss": 0.6246,
+      "step": 577
+    },
+    {
+      "epoch": 3.92030521407376,
+      "grad_norm": 0.14295575937313854,
+      "learning_rate": 1.0628712772316682e-05,
+      "loss": 0.6186,
+      "step": 578
+    },
+    {
+      "epoch": 3.927087749046206,
+      "grad_norm": 0.14578310176151307,
+      "learning_rate": 1.04999888651557e-05,
+      "loss": 0.6237,
+      "step": 579
+    },
+    {
+      "epoch": 3.933870284018652,
+      "grad_norm": 0.1418828430912816,
+      "learning_rate": 1.0371931332217175e-05,
+      "loss": 0.6247,
+      "step": 580
+    },
+    {
+      "epoch": 3.940652818991098,
+      "grad_norm": 0.1325595548114853,
+      "learning_rate": 1.0244543066186102e-05,
+      "loss": 0.623,
+      "step": 581
+    },
+    {
+      "epoch": 3.947435353963544,
+      "grad_norm": 0.1458164336431073,
+      "learning_rate": 1.0117826944629425e-05,
+      "loss": 0.6311,
+      "step": 582
+    },
+    {
+      "epoch": 3.9542178889359896,
+      "grad_norm": 0.1369850394553834,
+      "learning_rate": 9.991785829931024e-06,
+      "loss": 0.6352,
+      "step": 583
+    },
+    {
+      "epoch": 3.9610004239084358,
+      "grad_norm": 0.11861951638152013,
+      "learning_rate": 9.866422569227133e-06,
+      "loss": 0.616,
+      "step": 584
+    },
+    {
+      "epoch": 3.967782958880882,
+      "grad_norm": 0.12198529312439654,
+      "learning_rate": 9.74173999434192e-06,
+      "loss": 0.6267,
+      "step": 585
+    },
+    {
+      "epoch": 3.9745654938533277,
+      "grad_norm": 0.13835745132234695,
+      "learning_rate": 9.61774092172362e-06,
+      "loss": 0.6228,
+      "step": 586
+    },
+    {
+      "epoch": 3.9813480288257734,
+      "grad_norm": 0.14469581512913757,
+      "learning_rate": 9.494428152380868e-06,
+      "loss": 0.6249,
+      "step": 587
+    },
+    {
+      "epoch": 3.9881305637982196,
+      "grad_norm": 0.14538533970008014,
+      "learning_rate": 9.371804471819401e-06,
+      "loss": 0.6327,
+      "step": 588
+    },
+    {
+      "epoch": 3.994913098770666,
+      "grad_norm": 0.12842111782896357,
+      "learning_rate": 9.249872649979203e-06,
+      "loss": 0.6303,
+      "step": 589
+    },
+    {
+      "epoch": 4.001695633743111,
+      "grad_norm": 0.26900932028906,
+      "learning_rate": 9.128635441171854e-06,
+      "loss": 0.947,
+      "step": 590
+    },
+    {
+      "epoch": 4.008478168715557,
+      "grad_norm": 0.1546620990829906,
+      "learning_rate": 9.008095584018406e-06,
+      "loss": 0.6138,
+      "step": 591
+    },
+    {
+      "epoch": 4.0152607036880035,
+      "grad_norm": 0.15420921751444336,
+      "learning_rate": 8.888255801387417e-06,
+      "loss": 0.6084,
+      "step": 592
+    },
+    {
+      "epoch": 4.02204323866045,
+      "grad_norm": 0.15369650428751042,
+      "learning_rate": 8.76911880033354e-06,
+      "loss": 0.6196,
+      "step": 593
+    },
+    {
+      "epoch": 4.028825773632895,
+      "grad_norm": 0.1491321631121837,
+      "learning_rate": 8.65068727203633e-06,
+      "loss": 0.6193,
+      "step": 594
+    },
+    {
+      "epoch": 4.035608308605341,
+      "grad_norm": 0.14296040181465852,
+      "learning_rate": 8.53296389173941e-06,
+      "loss": 0.5971,
+      "step": 595
+    },
+    {
+      "epoch": 4.042390843577787,
+      "grad_norm": 0.17332216986843843,
+      "learning_rate": 8.415951318690134e-06,
+      "loss": 0.6085,
+      "step": 596
+    },
+    {
+      "epoch": 4.0491733785502335,
+      "grad_norm": 0.1392839623319017,
+      "learning_rate": 8.299652196079462e-06,
+      "loss": 0.6001,
+      "step": 597
+    },
+    {
+      "epoch": 4.055955913522679,
+      "grad_norm": 0.13152438443257536,
+      "learning_rate": 8.184069150982217e-06,
+      "loss": 0.6105,
+      "step": 598
+    },
+    {
+      "epoch": 4.062738448495125,
+      "grad_norm": 0.1534567228015376,
+      "learning_rate": 8.069204794297843e-06,
+      "loss": 0.6055,
+      "step": 599
+    },
+    {
+      "epoch": 4.069520983467571,
+      "grad_norm": 0.13360654544275674,
+      "learning_rate": 7.955061720691315e-06,
+      "loss": 0.6032,
+      "step": 600
+    },
+    {
+      "epoch": 4.076303518440017,
+      "grad_norm": 0.12240252460009761,
+      "learning_rate": 7.841642508534617e-06,
+      "loss": 0.6123,
+      "step": 601
+    },
+    {
+      "epoch": 4.083086053412463,
+      "grad_norm": 0.13532197093706563,
+      "learning_rate": 7.728949719848464e-06,
+      "loss": 0.6153,
+      "step": 602
+    },
+    {
+      "epoch": 4.089868588384909,
+      "grad_norm": 0.14193336659405498,
+      "learning_rate": 7.6169859002443954e-06,
+      "loss": 0.6168,
+      "step": 603
+    },
+    {
+      "epoch": 4.096651123357355,
+      "grad_norm": 0.12702143209359149,
+      "learning_rate": 7.505753578867354e-06,
+      "loss": 0.6151,
+      "step": 604
+    },
+    {
+      "epoch": 4.1034336583298,
+      "grad_norm": 0.13369676075093645,
+      "learning_rate": 7.395255268338459e-06,
+      "loss": 0.6166,
+      "step": 605
+    },
+    {
+      "epoch": 4.1102161933022465,
+      "grad_norm": 0.1334402641156527,
+      "learning_rate": 7.285493464698343e-06,
+      "loss": 0.6103,
+      "step": 606
+    },
+    {
+      "epoch": 4.116998728274693,
+      "grad_norm": 0.11322677082673437,
+      "learning_rate": 7.176470647350675e-06,
+      "loss": 0.6213,
+      "step": 607
+    },
+    {
+      "epoch": 4.123781263247139,
+      "grad_norm": 0.1301924506080975,
+      "learning_rate": 7.068189279006237e-06,
+      "loss": 0.6099,
+      "step": 608
+    },
+    {
+      "epoch": 4.130563798219584,
+      "grad_norm": 0.13846251217535666,
+      "learning_rate": 6.960651805627248e-06,
+      "loss": 0.5961,
+      "step": 609
+    },
+    {
+      "epoch": 4.13734633319203,
+      "grad_norm": 0.11426017628960868,
+      "learning_rate": 6.853860656372125e-06,
+      "loss": 0.6001,
+      "step": 610
+    },
+    {
+      "epoch": 4.144128868164477,
+      "grad_norm": 0.11173851423251971,
+      "learning_rate": 6.747818243540587e-06,
+      "loss": 0.6068,
+      "step": 611
+    },
+    {
+      "epoch": 4.150911403136923,
+      "grad_norm": 0.13703419453889504,
+      "learning_rate": 6.642526962519218e-06,
+      "loss": 0.6083,
+      "step": 612
+    },
+    {
+      "epoch": 4.157693938109368,
+      "grad_norm": 0.12103577637533855,
+      "learning_rate": 6.537989191727292e-06,
+      "loss": 0.6026,
+      "step": 613
+    },
+    {
+      "epoch": 4.164476473081814,
+      "grad_norm": 0.12350144156787667,
+      "learning_rate": 6.4342072925631125e-06,
+      "loss": 0.6019,
+      "step": 614
+    },
+    {
+      "epoch": 4.17125900805426,
+      "grad_norm": 0.11640263763641474,
+      "learning_rate": 6.331183609350602e-06,
+      "loss": 0.5954,
+      "step": 615
+    },
+    {
+      "epoch": 4.178041543026707,
+      "grad_norm": 0.10545213840827801,
+      "learning_rate": 6.228920469286408e-06,
+      "loss": 0.6117,
+      "step": 616
+    },
+    {
+      "epoch": 4.184824077999152,
+      "grad_norm": 0.11495651066805103,
+      "learning_rate": 6.127420182387314e-06,
+      "loss": 0.5946,
+      "step": 617
+    },
+    {
+      "epoch": 4.191606612971598,
+      "grad_norm": 0.11620572636315642,
+      "learning_rate": 6.026685041438018e-06,
+      "loss": 0.6034,
+      "step": 618
+    },
+    {
+      "epoch": 4.198389147944044,
+      "grad_norm": 0.10866699027698076,
+      "learning_rate": 5.926717321939417e-06,
+      "loss": 0.6048,
+      "step": 619
+    },
+    {
+      "epoch": 4.2051716829164905,
+      "grad_norm": 0.1245636539367363,
+      "learning_rate": 5.827519282057128e-06,
+      "loss": 0.6037,
+      "step": 620
+    },
+    {
+      "epoch": 4.211954217888936,
+      "grad_norm": 0.10644196388055245,
+      "learning_rate": 5.729093162570545e-06,
+      "loss": 0.607,
+      "step": 621
+    },
+    {
+      "epoch": 4.218736752861382,
+      "grad_norm": 0.10894266432537414,
+      "learning_rate": 5.631441186822168e-06,
+      "loss": 0.6135,
+      "step": 622
+    },
+    {
+      "epoch": 4.225519287833828,
+      "grad_norm": 0.12336549859525484,
+      "learning_rate": 5.534565560667426e-06,
+      "loss": 0.5981,
+      "step": 623
+    },
+    {
+      "epoch": 4.232301822806273,
+      "grad_norm": 0.10948607920606619,
+      "learning_rate": 5.4384684724248096e-06,
+      "loss": 0.6118,
+      "step": 624
+    },
+    {
+      "epoch": 4.23908435777872,
+      "grad_norm": 0.10785742485131251,
+      "learning_rate": 5.343152092826476e-06,
+      "loss": 0.6184,
+      "step": 625
+    },
+    {
+      "epoch": 4.245866892751166,
+      "grad_norm": 0.1346812509747976,
+      "learning_rate": 5.2486185749691665e-06,
+      "loss": 0.6058,
+      "step": 626
+    },
+    {
+      "epoch": 4.252649427723612,
+      "grad_norm": 0.10711676011900324,
+      "learning_rate": 5.154870054265635e-06,
+      "loss": 0.6071,
+      "step": 627
+    },
+    {
+      "epoch": 4.259431962696057,
+      "grad_norm": 0.09686690911531301,
+      "learning_rate": 5.061908648396338e-06,
+      "loss": 0.6019,
+      "step": 628
+    },
+    {
+      "epoch": 4.2662144976685035,
+      "grad_norm": 0.11286945799474804,
+      "learning_rate": 4.969736457261669e-06,
+      "loss": 0.604,
+      "step": 629
+    },
+    {
+      "epoch": 4.27299703264095,
+      "grad_norm": 0.12539528001284939,
+      "learning_rate": 4.878355562934465e-06,
+      "loss": 0.6103,
+      "step": 630
+    },
+    {
+      "epoch": 4.279779567613396,
+      "grad_norm": 0.11846976963374484,
+      "learning_rate": 4.787768029613014e-06,
+      "loss": 0.6102,
+      "step": 631
+    },
+    {
+      "epoch": 4.286562102585841,
+      "grad_norm": 0.11455949347213798,
+      "learning_rate": 4.697975903574419e-06,
+      "loss": 0.6044,
+      "step": 632
+    },
+    {
+      "epoch": 4.293344637558287,
+      "grad_norm": 0.11423245618436771,
+      "learning_rate": 4.608981213128347e-06,
+      "loss": 0.6104,
+      "step": 633
+    },
+    {
+      "epoch": 4.3001271725307335,
+      "grad_norm": 0.11763066458751498,
+      "learning_rate": 4.520785968571266e-06,
+      "loss": 0.6103,
+      "step": 634
+    },
+    {
+      "epoch": 4.30690970750318,
+      "grad_norm": 0.12407775475836062,
+      "learning_rate": 4.433392162140968e-06,
+      "loss": 0.6127,
+      "step": 635
+    },
+    {
+      "epoch": 4.313692242475625,
+      "grad_norm": 0.1177975105922243,
+      "learning_rate": 4.3468017679716245e-06,
+      "loss": 0.6258,
+      "step": 636
+    },
+    {
+      "epoch": 4.320474777448071,
+      "grad_norm": 0.11062581278413729,
+      "learning_rate": 4.261016742049155e-06,
+      "loss": 0.617,
+      "step": 637
+    },
+    {
+      "epoch": 4.327257312420517,
+      "grad_norm": 0.11328064594157465,
+      "learning_rate": 4.176039022167064e-06,
+      "loss": 0.6086,
+      "step": 638
+    },
+    {
+      "epoch": 4.3340398473929636,
+      "grad_norm": 0.12286256944458376,
+      "learning_rate": 4.0918705278826685e-06,
+      "loss": 0.6108,
+      "step": 639
+    },
+    {
+      "epoch": 4.340822382365409,
+      "grad_norm": 0.11726282791596607,
+      "learning_rate": 4.008513160473726e-06,
+      "loss": 0.6122,
+      "step": 640
+    },
+    {
+      "epoch": 4.347604917337855,
+      "grad_norm": 0.13243402516995725,
+      "learning_rate": 3.925968802895477e-06,
+      "loss": 0.6059,
+      "step": 641
+    },
+    {
+      "epoch": 4.354387452310301,
+      "grad_norm": 0.11732415982857679,
+      "learning_rate": 3.8442393197381456e-06,
+      "loss": 0.6159,
+      "step": 642
+    },
+    {
+      "epoch": 4.3611699872827465,
+      "grad_norm": 0.11432499142874704,
+      "learning_rate": 3.7633265571847787e-06,
+      "loss": 0.6053,
+      "step": 643
+    },
+    {
+      "epoch": 4.367952522255193,
+      "grad_norm": 0.12438661313712396,
+      "learning_rate": 3.683232342969594e-06,
+      "loss": 0.6185,
+      "step": 644
+    },
+    {
+      "epoch": 4.374735057227639,
+      "grad_norm": 0.12003558611356997,
+      "learning_rate": 3.603958486336625e-06,
+      "loss": 0.6311,
+      "step": 645
+    },
+    {
+      "epoch": 4.381517592200085,
+      "grad_norm": 0.10916503150131288,
+      "learning_rate": 3.5255067779989173e-06,
+      "loss": 0.6158,
+      "step": 646
+    },
+    {
+      "epoch": 4.38830012717253,
+      "grad_norm": 0.11467284610857227,
+      "learning_rate": 3.4478789900980545e-06,
+      "loss": 0.6176,
+      "step": 647
+    },
+    {
+      "epoch": 4.395082662144977,
+      "grad_norm": 0.11307569011740914,
+      "learning_rate": 3.371076876164101e-06,
+      "loss": 0.614,
+      "step": 648
+    },
+    {
+      "epoch": 4.401865197117423,
+      "grad_norm": 0.11385086584182105,
+      "learning_rate": 3.295102171076039e-06,
+      "loss": 0.6098,
+      "step": 649
+    },
+    {
+      "epoch": 4.408647732089869,
+      "grad_norm": 0.13858694628446455,
+      "learning_rate": 3.2199565910225305e-06,
+      "loss": 0.5976,
+      "step": 650
+    },
+    {
+      "epoch": 4.415430267062314,
+      "grad_norm": 0.12042166145746055,
+      "learning_rate": 3.145641833463198e-06,
+      "loss": 0.6164,
+      "step": 651
+    },
+    {
+      "epoch": 4.42221280203476,
+      "grad_norm": 0.11519596457203823,
+      "learning_rate": 3.0721595770902435e-06,
+      "loss": 0.6156,
+      "step": 652
+    },
+    {
+      "epoch": 4.428995337007207,
+      "grad_norm": 0.09905226368345378,
+      "learning_rate": 2.9995114817905493e-06,
+      "loss": 0.6001,
+      "step": 653
+    },
+    {
+      "epoch": 4.435777871979653,
+      "grad_norm": 0.11732702729180129,
+      "learning_rate": 2.927699188608171e-06,
+      "loss": 0.607,
+      "step": 654
+    },
+    {
+      "epoch": 4.442560406952098,
+      "grad_norm": 0.1179451846081634,
+      "learning_rate": 2.856724319707289e-06,
+      "loss": 0.6169,
+      "step": 655
+    },
+    {
+      "epoch": 4.449342941924544,
+      "grad_norm": 0.10442323225441018,
+      "learning_rate": 2.7865884783355234e-06,
+      "loss": 0.6179,
+      "step": 656
+    },
+    {
+      "epoch": 4.4561254768969905,
+      "grad_norm": 0.09435389572080079,
+      "learning_rate": 2.717293248787769e-06,
+      "loss": 0.6201,
+      "step": 657
+    },
+    {
+      "epoch": 4.462908011869436,
+      "grad_norm": 0.11295536763791658,
+      "learning_rate": 2.648840196370368e-06,
+      "loss": 0.6019,
+      "step": 658
+    },
+    {
+      "epoch": 4.469690546841882,
+      "grad_norm": 0.11594274422462209,
+      "learning_rate": 2.5812308673657825e-06,
+      "loss": 0.5977,
+      "step": 659
+    },
+    {
+      "epoch": 4.476473081814328,
+      "grad_norm": 0.10182885233809727,
+      "learning_rate": 2.5144667889976316e-06,
+      "loss": 0.6017,
+      "step": 660
+    },
+    {
+      "epoch": 4.483255616786774,
+      "grad_norm": 0.10348633393081054,
+      "learning_rate": 2.4485494693962197e-06,
+      "loss": 0.6086,
+      "step": 661
+    },
+    {
+      "epoch": 4.49003815175922,
+      "grad_norm": 0.10025894496329021,
+      "learning_rate": 2.3834803975644772e-06,
+      "loss": 0.6182,
+      "step": 662
+    },
+    {
+      "epoch": 4.496820686731666,
+      "grad_norm": 0.10187054293463914,
+      "learning_rate": 2.3192610433442784e-06,
+      "loss": 0.6074,
+      "step": 663
+    },
+    {
+      "epoch": 4.503603221704112,
+      "grad_norm": 0.09715916393818302,
+      "learning_rate": 2.2558928573833016e-06,
+      "loss": 0.6076,
+      "step": 664
+    },
+    {
+      "epoch": 4.510385756676558,
+      "grad_norm": 0.09756303138919038,
+      "learning_rate": 2.1933772711021995e-06,
+      "loss": 0.6019,
+      "step": 665
+    },
+    {
+      "epoch": 4.5171682916490035,
+      "grad_norm": 0.09525032911781027,
+      "learning_rate": 2.1317156966623243e-06,
+      "loss": 0.5995,
+      "step": 666
+    },
+    {
+      "epoch": 4.52395082662145,
+      "grad_norm": 0.09487242097788966,
+      "learning_rate": 2.0709095269337755e-06,
+      "loss": 0.6023,
+      "step": 667
+    },
+    {
+      "epoch": 4.530733361593896,
+      "grad_norm": 0.11476206706719008,
+      "learning_rate": 2.0109601354639706e-06,
+      "loss": 0.6143,
+      "step": 668
+    },
+    {
+      "epoch": 4.537515896566342,
+      "grad_norm": 0.09904862194550892,
+      "learning_rate": 1.9518688764466096e-06,
+      "loss": 0.6143,
+      "step": 669
+    },
+    {
+      "epoch": 4.544298431538787,
+      "grad_norm": 0.09278301353904947,
+      "learning_rate": 1.8936370846910845e-06,
+      "loss": 0.6092,
+      "step": 670
+    },
+    {
+      "epoch": 4.5510809665112335,
+      "grad_norm": 0.10688691601881584,
+      "learning_rate": 1.8362660755923079e-06,
+      "loss": 0.6186,
+      "step": 671
+    },
+    {
+      "epoch": 4.55786350148368,
+      "grad_norm": 0.09697916608330605,
+      "learning_rate": 1.7797571451010442e-06,
+      "loss": 0.5983,
+      "step": 672
+    },
+    {
+      "epoch": 4.564646036456125,
+      "grad_norm": 0.10471706981067595,
+      "learning_rate": 1.7241115696945821e-06,
+      "loss": 0.6132,
+      "step": 673
+    },
+    {
+      "epoch": 4.571428571428571,
+      "grad_norm": 0.09406802929392503,
+      "learning_rate": 1.669330606347952e-06,
+      "loss": 0.5962,
+      "step": 674
+    },
+    {
+      "epoch": 4.578211106401017,
+      "grad_norm": 0.0954037644840885,
+      "learning_rate": 1.6154154925054878e-06,
+      "loss": 0.6141,
+      "step": 675
+    },
+    {
+      "epoch": 4.5849936413734635,
+      "grad_norm": 0.10094861062769968,
+      "learning_rate": 1.5623674460529014e-06,
+      "loss": 0.6024,
+      "step": 676
+    },
+    {
+      "epoch": 4.59177617634591,
+      "grad_norm": 0.09573172404295381,
+      "learning_rate": 1.510187665289773e-06,
+      "loss": 0.6086,
+      "step": 677
+    },
+    {
+      "epoch": 4.598558711318355,
+      "grad_norm": 0.09995771347446304,
+      "learning_rate": 1.458877328902455e-06,
+      "loss": 0.6052,
+      "step": 678
+    },
+    {
+      "epoch": 4.605341246290801,
+      "grad_norm": 0.09656263063599439,
+      "learning_rate": 1.4084375959374864e-06,
+      "loss": 0.6072,
+      "step": 679
+    },
+    {
+      "epoch": 4.612123781263247,
+      "grad_norm": 0.09822686472765482,
+      "learning_rate": 1.3588696057753725e-06,
+      "loss": 0.6089,
+      "step": 680
+    },
+    {
+      "epoch": 4.618906316235693,
+      "grad_norm": 0.09676277658643694,
+      "learning_rate": 1.3101744781048774e-06,
+      "loss": 0.6016,
+      "step": 681
+    },
+    {
+      "epoch": 4.625688851208139,
+      "grad_norm": 0.0933744691144583,
+      "learning_rate": 1.262353312897715e-06,
+      "loss": 0.6075,
+      "step": 682
+    },
+    {
+      "epoch": 4.632471386180585,
+      "grad_norm": 0.0996442423520143,
+      "learning_rate": 1.2154071903837107e-06,
+      "loss": 0.6157,
+      "step": 683
+    },
+    {
+      "epoch": 4.639253921153031,
+      "grad_norm": 0.10109190980066803,
+      "learning_rate": 1.1693371710263857e-06,
+      "loss": 0.6049,
+      "step": 684
+    },
+    {
+      "epoch": 4.6460364561254766,
+      "grad_norm": 0.08703236124988406,
+      "learning_rate": 1.1241442954990301e-06,
+      "loss": 0.6048,
+      "step": 685
+    },
+    {
+      "epoch": 4.652818991097923,
+      "grad_norm": 0.09280509544326945,
+      "learning_rate": 1.0798295846611562e-06,
+      "loss": 0.5989,
+      "step": 686
+    },
+    {
+      "epoch": 4.659601526070369,
+      "grad_norm": 0.09785736700280491,
+      "learning_rate": 1.0363940395354777e-06,
+      "loss": 0.6098,
+      "step": 687
+    },
+    {
+      "epoch": 4.666384061042815,
+      "grad_norm": 0.09508110174195833,
+      "learning_rate": 9.938386412852652e-07,
+      "loss": 0.6149,
+      "step": 688
+    },
+    {
+      "epoch": 4.67316659601526,
+      "grad_norm": 0.09065950330259104,
+      "learning_rate": 9.52164351192213e-07,
+      "loss": 0.6088,
+      "step": 689
+    },
+    {
+      "epoch": 4.679949130987707,
+      "grad_norm": 0.09551867620930947,
+      "learning_rate": 9.113721106346918e-07,
+      "loss": 0.5989,
+      "step": 690
+    },
+    {
+      "epoch": 4.686731665960153,
+      "grad_norm": 0.09832290856026324,
+      "learning_rate": 8.714628410665082e-07,
+      "loss": 0.6172,
+      "step": 691
+    },
+    {
+      "epoch": 4.693514200932599,
+      "grad_norm": 0.09371892802996978,
+      "learning_rate": 8.324374439960947e-07,
+      "loss": 0.6067,
+      "step": 692
+    },
+    {
+      "epoch": 4.700296735905044,
+      "grad_norm": 0.09014551117205812,
+      "learning_rate": 7.942968009661123e-07,
+      "loss": 0.6142,
+      "step": 693
+    },
+    {
+      "epoch": 4.7070792708774905,
+      "grad_norm": 0.09724706571681141,
+      "learning_rate": 7.570417735335733e-07,
+      "loss": 0.614,
+      "step": 694
+    },
+    {
+      "epoch": 4.713861805849937,
+      "grad_norm": 0.08494611369686998,
+      "learning_rate": 7.206732032503638e-07,
+      "loss": 0.5969,
+      "step": 695
+    },
+    {
+      "epoch": 4.720644340822382,
+      "grad_norm": 0.0927927564675014,
+      "learning_rate": 6.851919116442274e-07,
+      "loss": 0.6122,
+      "step": 696
+    },
+    {
+      "epoch": 4.727426875794828,
+      "grad_norm": 0.09273724936533222,
+      "learning_rate": 6.505987002002245e-07,
+      "loss": 0.6092,
+      "step": 697
+    },
+    {
+      "epoch": 4.734209410767274,
+      "grad_norm": 0.08796818616185165,
+      "learning_rate": 6.168943503426139e-07,
+      "loss": 0.6038,
+      "step": 698
+    },
+    {
+      "epoch": 4.7409919457397205,
+      "grad_norm": 0.08686636687902749,
+      "learning_rate": 5.840796234172085e-07,
+      "loss": 0.5943,
+      "step": 699
+    },
+    {
+      "epoch": 4.747774480712166,
+      "grad_norm": 0.09236449250012803,
+      "learning_rate": 5.521552606741765e-07,
+      "loss": 0.6066,
+      "step": 700
+    },
+    {
+      "epoch": 4.754557015684612,
+      "grad_norm": 0.0926749240950783,
+      "learning_rate": 5.211219832512893e-07,
+      "loss": 0.6114,
+      "step": 701
+    },
+    {
+      "epoch": 4.761339550657058,
+      "grad_norm": 0.10499405025384424,
+      "learning_rate": 4.909804921576466e-07,
+      "loss": 0.6185,
+      "step": 702
+    },
+    {
+      "epoch": 4.768122085629504,
+      "grad_norm": 0.08848832101586154,
+      "learning_rate": 4.6173146825782224e-07,
+      "loss": 0.6067,
+      "step": 703
+    },
+    {
+      "epoch": 4.77490462060195,
+      "grad_norm": 0.09319395321366955,
+      "learning_rate": 4.3337557225650695e-07,
+      "loss": 0.6047,
+      "step": 704
+    },
+    {
+      "epoch": 4.781687155574396,
+      "grad_norm": 0.0915059840709739,
+      "learning_rate": 4.0591344468355666e-07,
+      "loss": 0.6118,
+      "step": 705
+    },
+    {
+      "epoch": 4.788469690546842,
+      "grad_norm": 0.10729723170818359,
+      "learning_rate": 3.793457058795591e-07,
+      "loss": 0.6189,
+      "step": 706
+    },
+    {
+      "epoch": 4.795252225519288,
+      "grad_norm": 0.09285587112058538,
+      "learning_rate": 3.5367295598178307e-07,
+      "loss": 0.6085,
+      "step": 707
+    },
+    {
+      "epoch": 4.8020347604917335,
+      "grad_norm": 0.08196537866623092,
+      "learning_rate": 3.2889577491064693e-07,
+      "loss": 0.6045,
+      "step": 708
+    },
+    {
+      "epoch": 4.80881729546418,
+      "grad_norm": 0.08720837958758754,
+      "learning_rate": 3.050147223566091e-07,
+      "loss": 0.6122,
+      "step": 709
+    },
+    {
+      "epoch": 4.815599830436626,
+      "grad_norm": 0.09158067797742139,
+      "learning_rate": 2.8203033776752487e-07,
+      "loss": 0.6033,
+      "step": 710
+    },
+    {
+      "epoch": 4.822382365409071,
+      "grad_norm": 0.09043295389558649,
+      "learning_rate": 2.599431403364694e-07,
+      "loss": 0.6094,
+      "step": 711
+    },
+    {
+      "epoch": 4.829164900381517,
+      "grad_norm": 0.09180388224113892,
+      "learning_rate": 2.3875362899000054e-07,
+      "loss": 0.604,
+      "step": 712
+    },
+    {
+      "epoch": 4.8359474353539635,
+      "grad_norm": 0.08651605692378189,
+      "learning_rate": 2.1846228237689226e-07,
+      "loss": 0.6087,
+      "step": 713
+    },
+    {
+      "epoch": 4.84272997032641,
+      "grad_norm": 0.08511470109771488,
+      "learning_rate": 1.9906955885732104e-07,
+      "loss": 0.6091,
+      "step": 714
+    },
+    {
+      "epoch": 4.849512505298856,
+      "grad_norm": 0.08454739481192287,
+      "learning_rate": 1.8057589649251862e-07,
+      "loss": 0.6146,
+      "step": 715
+    },
+    {
+      "epoch": 4.856295040271301,
+      "grad_norm": 0.08724704005095565,
+      "learning_rate": 1.62981713034851e-07,
+      "loss": 0.6042,
+      "step": 716
+    },
+    {
+      "epoch": 4.863077575243747,
+      "grad_norm": 0.08396574948424247,
+      "learning_rate": 1.462874059184305e-07,
+      "loss": 0.6086,
+      "step": 717
+    },
+    {
+      "epoch": 4.869860110216194,
+      "grad_norm": 0.09082130018441244,
+      "learning_rate": 1.3049335225009175e-07,
+      "loss": 0.6295,
+      "step": 718
+    },
+    {
+      "epoch": 4.876642645188639,
+      "grad_norm": 0.08722605095693269,
+      "learning_rate": 1.1559990880089189e-07,
+      "loss": 0.6085,
+      "step": 719
+    },
+    {
+      "epoch": 4.883425180161085,
+      "grad_norm": 0.0878159462681937,
+      "learning_rate": 1.0160741199805036e-07,
+      "loss": 0.5998,
+      "step": 720
+    },
+    {
+      "epoch": 4.890207715133531,
+      "grad_norm": 0.0878736816907958,
+      "learning_rate": 8.851617791735933e-08,
+      "loss": 0.6145,
+      "step": 721
+    },
+    {
+      "epoch": 4.896990250105977,
+      "grad_norm": 0.08847166155449576,
+      "learning_rate": 7.632650227602511e-08,
+      "loss": 0.604,
+      "step": 722
+    },
+    {
+      "epoch": 4.903772785078423,
+      "grad_norm": 0.08650107449834574,
+      "learning_rate": 6.503866042599338e-08,
+      "loss": 0.6187,
+      "step": 723
+    },
+    {
+      "epoch": 4.910555320050869,
+      "grad_norm": 0.09963600285093287,
+      "learning_rate": 5.4652907347745307e-08,
+      "loss": 0.6166,
+      "step": 724
+    },
+    {
+      "epoch": 4.917337855023315,
+      "grad_norm": 0.08660963081265105,
+      "learning_rate": 4.516947764451107e-08,
+      "loss": 0.6156,
+      "step": 725
+    },
+    {
+      "epoch": 4.924120389995761,
+      "grad_norm": 0.09162932265931903,
+      "learning_rate": 3.6588585537002955e-08,
+      "loss": 0.6085,
+      "step": 726
+    },
+    {
+      "epoch": 4.930902924968207,
+      "grad_norm": 0.08533741638096395,
+      "learning_rate": 2.8910424858543673e-08,
+      "loss": 0.6026,
+      "step": 727
+    },
+    {
+      "epoch": 4.937685459940653,
+      "grad_norm": 0.08431534489847897,
+      "learning_rate": 2.213516905070101e-08,
+      "loss": 0.6012,
+      "step": 728
+    },
+    {
+      "epoch": 4.944467994913099,
+      "grad_norm": 0.08633616914365788,
+      "learning_rate": 1.6262971159384243e-08,
+      "loss": 0.6146,
+      "step": 729
+    },
+    {
+      "epoch": 4.951250529885545,
+      "grad_norm": 0.0931259062674402,
+      "learning_rate": 1.1293963831366939e-08,
+      "loss": 0.6116,
+      "step": 730
+    },
+    {
+      "epoch": 4.9580330648579904,
+      "grad_norm": 0.08678440176862204,
+      "learning_rate": 7.228259311293784e-09,
+      "loss": 0.602,
+      "step": 731
+    },
+    {
+      "epoch": 4.964815599830437,
+      "grad_norm": 0.0942499520961841,
+      "learning_rate": 4.065949439158168e-09,
+      "loss": 0.6099,
+      "step": 732
+    },
+    {
+      "epoch": 4.971598134802883,
+      "grad_norm": 0.07935089647201646,
+      "learning_rate": 1.8071056482105166e-09,
+      "loss": 0.615,
+      "step": 733
+    },
+    {
+      "epoch": 4.978380669775328,
+      "grad_norm": 0.08485220331271215,
+      "learning_rate": 4.517789633551317e-10,
+      "loss": 0.6104,
+      "step": 734
+    },
+    {
+      "epoch": 4.985163204747774,
+      "grad_norm": 0.08704984136629952,
+      "learning_rate": 0.0,
+      "loss": 0.6111,
+      "step": 735
+    },
+    {
+      "epoch": 4.985163204747774,
+      "step": 735,
+      "total_flos": 1.8913738317545603e+19,
+      "train_loss": 0.673498132682982,
+      "train_runtime": 162938.9767,
+      "train_samples_per_second": 2.316,
+      "train_steps_per_second": 0.005
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 735,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.8913738317545603e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}