diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,19888 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 2835,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0010582010582010583,
+      "grad_norm": 49.71796324571249,
+      "learning_rate": 0.0,
+      "loss": 11.3407,
+      "step": 1
+    },
+    {
+      "epoch": 0.0021164021164021165,
+      "grad_norm": 47.138920498887764,
+      "learning_rate": 1.7605633802816901e-07,
+      "loss": 11.3367,
+      "step": 2
+    },
+    {
+      "epoch": 0.0031746031746031746,
+      "grad_norm": 45.86427493870416,
+      "learning_rate": 3.5211267605633803e-07,
+      "loss": 11.3782,
+      "step": 3
+    },
+    {
+      "epoch": 0.004232804232804233,
+      "grad_norm": 46.592216421647926,
+      "learning_rate": 5.28169014084507e-07,
+      "loss": 11.4448,
+      "step": 4
+    },
+    {
+      "epoch": 0.005291005291005291,
+      "grad_norm": 51.27861968325324,
+      "learning_rate": 7.042253521126761e-07,
+      "loss": 11.1976,
+      "step": 5
+    },
+    {
+      "epoch": 0.006349206349206349,
+      "grad_norm": 48.68604904156757,
+      "learning_rate": 8.802816901408452e-07,
+      "loss": 11.3589,
+      "step": 6
+    },
+    {
+      "epoch": 0.007407407407407408,
+      "grad_norm": 49.04868842468927,
+      "learning_rate": 1.056338028169014e-06,
+      "loss": 11.216,
+      "step": 7
+    },
+    {
+      "epoch": 0.008465608465608466,
+      "grad_norm": 45.78424778978211,
+      "learning_rate": 1.232394366197183e-06,
+      "loss": 11.2126,
+      "step": 8
+    },
+    {
+      "epoch": 0.009523809523809525,
+      "grad_norm": 48.288808560697824,
+      "learning_rate": 1.4084507042253521e-06,
+      "loss": 11.2473,
+      "step": 9
+    },
+    {
+      "epoch": 0.010582010582010581,
+      "grad_norm": 47.00529127513613,
+      "learning_rate": 1.5845070422535212e-06,
+      "loss": 11.181,
+      "step": 10
+    },
+    {
+      "epoch": 0.01164021164021164,
+      "grad_norm": 69.96523123803429,
+      "learning_rate": 1.7605633802816904e-06,
+      "loss": 10.1444,
+      "step": 11
+    },
+    {
+      "epoch": 0.012698412698412698,
+      "grad_norm": 67.99361968629948,
+      "learning_rate": 1.936619718309859e-06,
+      "loss": 10.2666,
+      "step": 12
+    },
+    {
+      "epoch": 0.013756613756613757,
+      "grad_norm": 82.7361163900619,
+      "learning_rate": 2.112676056338028e-06,
+      "loss": 9.6636,
+      "step": 13
+    },
+    {
+      "epoch": 0.014814814814814815,
+      "grad_norm": 77.91883219725891,
+      "learning_rate": 2.2887323943661975e-06,
+      "loss": 9.7129,
+      "step": 14
+    },
+    {
+      "epoch": 0.015873015873015872,
+      "grad_norm": 80.69870504421503,
+      "learning_rate": 2.464788732394366e-06,
+      "loss": 5.0932,
+      "step": 15
+    },
+    {
+      "epoch": 0.016931216931216932,
+      "grad_norm": 73.85506610645764,
+      "learning_rate": 2.640845070422535e-06,
+      "loss": 4.7543,
+      "step": 16
+    },
+    {
+      "epoch": 0.01798941798941799,
+      "grad_norm": 72.44122211654165,
+      "learning_rate": 2.8169014084507042e-06,
+      "loss": 4.7132,
+      "step": 17
+    },
+    {
+      "epoch": 0.01904761904761905,
+      "grad_norm": 54.80640084894919,
+      "learning_rate": 2.9929577464788733e-06,
+      "loss": 3.8029,
+      "step": 18
+    },
+    {
+      "epoch": 0.020105820105820106,
+      "grad_norm": 51.95247253108577,
+      "learning_rate": 3.1690140845070423e-06,
+      "loss": 3.7085,
+      "step": 19
+    },
+    {
+      "epoch": 0.021164021164021163,
+      "grad_norm": 14.24491850890378,
+      "learning_rate": 3.3450704225352113e-06,
+      "loss": 2.1113,
+      "step": 20
+    },
+    {
+      "epoch": 0.022222222222222223,
+      "grad_norm": 6.987157752616172,
+      "learning_rate": 3.521126760563381e-06,
+      "loss": 1.7438,
+      "step": 21
+    },
+    {
+      "epoch": 0.02328042328042328,
+      "grad_norm": 5.830944797295927,
+      "learning_rate": 3.6971830985915494e-06,
+      "loss": 1.7295,
+      "step": 22
+    },
+    {
+      "epoch": 0.02433862433862434,
+      "grad_norm": 5.138674915616038,
+      "learning_rate": 3.873239436619718e-06,
+      "loss": 1.6184,
+      "step": 23
+    },
+    {
+      "epoch": 0.025396825396825397,
+      "grad_norm": 3.8763256924499605,
+      "learning_rate": 4.0492957746478875e-06,
+      "loss": 1.5636,
+      "step": 24
+    },
+    {
+      "epoch": 0.026455026455026454,
+      "grad_norm": 4.336003458578358,
+      "learning_rate": 4.225352112676056e-06,
+      "loss": 1.5242,
+      "step": 25
+    },
+    {
+      "epoch": 0.027513227513227514,
+      "grad_norm": 3.002308914115821,
+      "learning_rate": 4.401408450704226e-06,
+      "loss": 1.5592,
+      "step": 26
+    },
+    {
+      "epoch": 0.02857142857142857,
+      "grad_norm": 2.211668797457168,
+      "learning_rate": 4.577464788732395e-06,
+      "loss": 1.3246,
+      "step": 27
+    },
+    {
+      "epoch": 0.02962962962962963,
+      "grad_norm": 1.856090631854003,
+      "learning_rate": 4.753521126760564e-06,
+      "loss": 1.3504,
+      "step": 28
+    },
+    {
+      "epoch": 0.030687830687830688,
+      "grad_norm": 1.730441467550159,
+      "learning_rate": 4.929577464788732e-06,
+      "loss": 1.1383,
+      "step": 29
+    },
+    {
+      "epoch": 0.031746031746031744,
+      "grad_norm": 3.037278309026151,
+      "learning_rate": 5.105633802816902e-06,
+      "loss": 1.0615,
+      "step": 30
+    },
+    {
+      "epoch": 0.0328042328042328,
+      "grad_norm": 20.463734102356668,
+      "learning_rate": 5.28169014084507e-06,
+      "loss": 1.1804,
+      "step": 31
+    },
+    {
+      "epoch": 0.033862433862433865,
+      "grad_norm": 1.1822365397348313,
+      "learning_rate": 5.457746478873239e-06,
+      "loss": 1.1662,
+      "step": 32
+    },
+    {
+      "epoch": 0.03492063492063492,
+      "grad_norm": 0.9353301639240665,
+      "learning_rate": 5.6338028169014084e-06,
+      "loss": 1.0131,
+      "step": 33
+    },
+    {
+      "epoch": 0.03597883597883598,
+      "grad_norm": 0.9206265704946479,
+      "learning_rate": 5.809859154929578e-06,
+      "loss": 1.0732,
+      "step": 34
+    },
+    {
+      "epoch": 0.037037037037037035,
+      "grad_norm": 11.424200349684694,
+      "learning_rate": 5.9859154929577465e-06,
+      "loss": 1.1348,
+      "step": 35
+    },
+    {
+      "epoch": 0.0380952380952381,
+      "grad_norm": 0.8604419013240469,
+      "learning_rate": 6.161971830985916e-06,
+      "loss": 1.0904,
+      "step": 36
+    },
+    {
+      "epoch": 0.039153439153439155,
+      "grad_norm": 0.779260466106171,
+      "learning_rate": 6.338028169014085e-06,
+      "loss": 0.9949,
+      "step": 37
+    },
+    {
+      "epoch": 0.04021164021164021,
+      "grad_norm": 1.053891911809051,
+      "learning_rate": 6.514084507042253e-06,
+      "loss": 0.9232,
+      "step": 38
+    },
+    {
+      "epoch": 0.04126984126984127,
+      "grad_norm": 0.8065280665981377,
+      "learning_rate": 6.690140845070423e-06,
+      "loss": 1.0275,
+      "step": 39
+    },
+    {
+      "epoch": 0.042328042328042326,
+      "grad_norm": 0.5932376581872476,
+      "learning_rate": 6.866197183098592e-06,
+      "loss": 0.9951,
+      "step": 40
+    },
+    {
+      "epoch": 0.04338624338624339,
+      "grad_norm": 0.7470084393654567,
+      "learning_rate": 7.042253521126762e-06,
+      "loss": 0.9576,
+      "step": 41
+    },
+    {
+      "epoch": 0.044444444444444446,
+      "grad_norm": 0.7190983046762137,
+      "learning_rate": 7.21830985915493e-06,
+      "loss": 1.0888,
+      "step": 42
+    },
+    {
+      "epoch": 0.0455026455026455,
+      "grad_norm": 0.6401594584722508,
+      "learning_rate": 7.394366197183099e-06,
+      "loss": 0.9013,
+      "step": 43
+    },
+    {
+      "epoch": 0.04656084656084656,
+      "grad_norm": 0.5484291413984935,
+      "learning_rate": 7.5704225352112675e-06,
+      "loss": 1.0244,
+      "step": 44
+    },
+    {
+      "epoch": 0.047619047619047616,
+      "grad_norm": 0.5552253491849386,
+      "learning_rate": 7.746478873239436e-06,
+      "loss": 0.9532,
+      "step": 45
+    },
+    {
+      "epoch": 0.04867724867724868,
+      "grad_norm": 0.5652331360674767,
+      "learning_rate": 7.922535211267606e-06,
+      "loss": 0.9989,
+      "step": 46
+    },
+    {
+      "epoch": 0.04973544973544974,
+      "grad_norm": 0.5293118764740489,
+      "learning_rate": 8.098591549295775e-06,
+      "loss": 0.9298,
+      "step": 47
+    },
+    {
+      "epoch": 0.050793650793650794,
+      "grad_norm": 0.5285967447293929,
+      "learning_rate": 8.274647887323944e-06,
+      "loss": 0.9737,
+      "step": 48
+    },
+    {
+      "epoch": 0.05185185185185185,
+      "grad_norm": 0.5178212820525971,
+      "learning_rate": 8.450704225352112e-06,
+      "loss": 0.9258,
+      "step": 49
+    },
+    {
+      "epoch": 0.05291005291005291,
+      "grad_norm": 0.4733170168194918,
+      "learning_rate": 8.626760563380283e-06,
+      "loss": 0.9534,
+      "step": 50
+    },
+    {
+      "epoch": 0.05396825396825397,
+      "grad_norm": 0.5149256098520258,
+      "learning_rate": 8.802816901408451e-06,
+      "loss": 1.0816,
+      "step": 51
+    },
+    {
+      "epoch": 0.05502645502645503,
+      "grad_norm": 0.4685693575510743,
+      "learning_rate": 8.978873239436621e-06,
+      "loss": 1.027,
+      "step": 52
+    },
+    {
+      "epoch": 0.056084656084656084,
+      "grad_norm": 0.39387712556840015,
+      "learning_rate": 9.15492957746479e-06,
+      "loss": 0.8691,
+      "step": 53
+    },
+    {
+      "epoch": 0.05714285714285714,
+      "grad_norm": 0.420846981768801,
+      "learning_rate": 9.330985915492959e-06,
+      "loss": 0.8751,
+      "step": 54
+    },
+    {
+      "epoch": 0.0582010582010582,
+      "grad_norm": 0.3985823470678549,
+      "learning_rate": 9.507042253521127e-06,
+      "loss": 0.7835,
+      "step": 55
+    },
+    {
+      "epoch": 0.05925925925925926,
+      "grad_norm": 0.4015984425355995,
+      "learning_rate": 9.683098591549296e-06,
+      "loss": 0.8231,
+      "step": 56
+    },
+    {
+      "epoch": 0.06031746031746032,
+      "grad_norm": 0.41302581392299265,
+      "learning_rate": 9.859154929577465e-06,
+      "loss": 0.9619,
+      "step": 57
+    },
+    {
+      "epoch": 0.061375661375661375,
+      "grad_norm": 0.3973364065159126,
+      "learning_rate": 1.0035211267605635e-05,
+      "loss": 0.8624,
+      "step": 58
+    },
+    {
+      "epoch": 0.06243386243386243,
+      "grad_norm": 0.3600376955888632,
+      "learning_rate": 1.0211267605633803e-05,
+      "loss": 0.9197,
+      "step": 59
+    },
+    {
+      "epoch": 0.06349206349206349,
+      "grad_norm": 0.38943169766010316,
+      "learning_rate": 1.0387323943661972e-05,
+      "loss": 0.7723,
+      "step": 60
+    },
+    {
+      "epoch": 0.06455026455026455,
+      "grad_norm": 0.40177295774598387,
+      "learning_rate": 1.056338028169014e-05,
+      "loss": 0.8586,
+      "step": 61
+    },
+    {
+      "epoch": 0.0656084656084656,
+      "grad_norm": 0.36974015962677353,
+      "learning_rate": 1.073943661971831e-05,
+      "loss": 0.8905,
+      "step": 62
+    },
+    {
+      "epoch": 0.06666666666666667,
+      "grad_norm": 0.3669768884030218,
+      "learning_rate": 1.0915492957746478e-05,
+      "loss": 0.8458,
+      "step": 63
+    },
+    {
+      "epoch": 0.06772486772486773,
+      "grad_norm": 0.4194352772922352,
+      "learning_rate": 1.1091549295774648e-05,
+      "loss": 0.92,
+      "step": 64
+    },
+    {
+      "epoch": 0.06878306878306878,
+      "grad_norm": 0.33216123239758316,
+      "learning_rate": 1.1267605633802817e-05,
+      "loss": 0.7894,
+      "step": 65
+    },
+    {
+      "epoch": 0.06984126984126984,
+      "grad_norm": 0.32639923159440226,
+      "learning_rate": 1.1443661971830987e-05,
+      "loss": 0.8095,
+      "step": 66
+    },
+    {
+      "epoch": 0.07089947089947089,
+      "grad_norm": 0.3829927205982677,
+      "learning_rate": 1.1619718309859156e-05,
+      "loss": 0.8557,
+      "step": 67
+    },
+    {
+      "epoch": 0.07195767195767196,
+      "grad_norm": 0.3222649310951342,
+      "learning_rate": 1.1795774647887324e-05,
+      "loss": 0.8743,
+      "step": 68
+    },
+    {
+      "epoch": 0.07301587301587302,
+      "grad_norm": 0.30020466526689116,
+      "learning_rate": 1.1971830985915493e-05,
+      "loss": 0.7349,
+      "step": 69
+    },
+    {
+      "epoch": 0.07407407407407407,
+      "grad_norm": 0.3519372592943403,
+      "learning_rate": 1.2147887323943663e-05,
+      "loss": 0.7388,
+      "step": 70
+    },
+    {
+      "epoch": 0.07513227513227513,
+      "grad_norm": 0.318183168764709,
+      "learning_rate": 1.2323943661971832e-05,
+      "loss": 0.7519,
+      "step": 71
+    },
+    {
+      "epoch": 0.0761904761904762,
+      "grad_norm": 0.310753241584739,
+      "learning_rate": 1.25e-05,
+      "loss": 0.7459,
+      "step": 72
+    },
+    {
+      "epoch": 0.07724867724867725,
+      "grad_norm": 0.3517552226109988,
+      "learning_rate": 1.267605633802817e-05,
+      "loss": 0.9719,
+      "step": 73
+    },
+    {
+      "epoch": 0.07830687830687831,
+      "grad_norm": 0.3213372230642392,
+      "learning_rate": 1.2852112676056338e-05,
+      "loss": 0.7786,
+      "step": 74
+    },
+    {
+      "epoch": 0.07936507936507936,
+      "grad_norm": 1.014002688907997,
+      "learning_rate": 1.3028169014084506e-05,
+      "loss": 0.8212,
+      "step": 75
+    },
+    {
+      "epoch": 0.08042328042328042,
+      "grad_norm": 0.2970169734383055,
+      "learning_rate": 1.3204225352112675e-05,
+      "loss": 0.8132,
+      "step": 76
+    },
+    {
+      "epoch": 0.08148148148148149,
+      "grad_norm": 0.305476164829128,
+      "learning_rate": 1.3380281690140845e-05,
+      "loss": 0.792,
+      "step": 77
+    },
+    {
+      "epoch": 0.08253968253968254,
+      "grad_norm": 0.30966098583703827,
+      "learning_rate": 1.3556338028169016e-05,
+      "loss": 0.7983,
+      "step": 78
+    },
+    {
+      "epoch": 0.0835978835978836,
+      "grad_norm": 0.35996391292327967,
+      "learning_rate": 1.3732394366197184e-05,
+      "loss": 0.8742,
+      "step": 79
+    },
+    {
+      "epoch": 0.08465608465608465,
+      "grad_norm": 0.30968125317223627,
+      "learning_rate": 1.3908450704225353e-05,
+      "loss": 0.748,
+      "step": 80
+    },
+    {
+      "epoch": 0.08571428571428572,
+      "grad_norm": 0.35611081185640775,
+      "learning_rate": 1.4084507042253523e-05,
+      "loss": 0.853,
+      "step": 81
+    },
+    {
+      "epoch": 0.08677248677248678,
+      "grad_norm": 0.2966375377032811,
+      "learning_rate": 1.4260563380281692e-05,
+      "loss": 0.7151,
+      "step": 82
+    },
+    {
+      "epoch": 0.08783068783068783,
+      "grad_norm": 0.7068384588765272,
+      "learning_rate": 1.443661971830986e-05,
+      "loss": 0.7145,
+      "step": 83
+    },
+    {
+      "epoch": 0.08888888888888889,
+      "grad_norm": 0.356149749502437,
+      "learning_rate": 1.4612676056338029e-05,
+      "loss": 0.7752,
+      "step": 84
+    },
+    {
+      "epoch": 0.08994708994708994,
+      "grad_norm": 0.30485993113060483,
+      "learning_rate": 1.4788732394366198e-05,
+      "loss": 0.6923,
+      "step": 85
+    },
+    {
+      "epoch": 0.091005291005291,
+      "grad_norm": 0.3283908018328014,
+      "learning_rate": 1.4964788732394366e-05,
+      "loss": 0.8052,
+      "step": 86
+    },
+    {
+      "epoch": 0.09206349206349207,
+      "grad_norm": 0.3272519816451936,
+      "learning_rate": 1.5140845070422535e-05,
+      "loss": 0.6502,
+      "step": 87
+    },
+    {
+      "epoch": 0.09312169312169312,
+      "grad_norm": 0.3540380713478706,
+      "learning_rate": 1.5316901408450704e-05,
+      "loss": 0.8252,
+      "step": 88
+    },
+    {
+      "epoch": 0.09417989417989418,
+      "grad_norm": 0.32708774961615394,
+      "learning_rate": 1.5492957746478872e-05,
+      "loss": 0.8256,
+      "step": 89
+    },
+    {
+      "epoch": 0.09523809523809523,
+      "grad_norm": 0.3078761207400959,
+      "learning_rate": 1.5669014084507044e-05,
+      "loss": 0.7787,
+      "step": 90
+    },
+    {
+      "epoch": 0.0962962962962963,
+      "grad_norm": 0.3085651879842953,
+      "learning_rate": 1.5845070422535213e-05,
+      "loss": 0.7873,
+      "step": 91
+    },
+    {
+      "epoch": 0.09735449735449736,
+      "grad_norm": 0.3411588858236989,
+      "learning_rate": 1.602112676056338e-05,
+      "loss": 0.7641,
+      "step": 92
+    },
+    {
+      "epoch": 0.09841269841269841,
+      "grad_norm": 0.30194570638829077,
+      "learning_rate": 1.619718309859155e-05,
+      "loss": 0.6853,
+      "step": 93
+    },
+    {
+      "epoch": 0.09947089947089947,
+      "grad_norm": 0.3417269654477643,
+      "learning_rate": 1.637323943661972e-05,
+      "loss": 0.7916,
+      "step": 94
+    },
+    {
+      "epoch": 0.10052910052910052,
+      "grad_norm": 0.31589650880434206,
+      "learning_rate": 1.6549295774647887e-05,
+      "loss": 0.7737,
+      "step": 95
+    },
+    {
+      "epoch": 0.10158730158730159,
+      "grad_norm": 0.31367036126327846,
+      "learning_rate": 1.6725352112676056e-05,
+      "loss": 0.7542,
+      "step": 96
+    },
+    {
+      "epoch": 0.10264550264550265,
+      "grad_norm": 0.31057819203477377,
+      "learning_rate": 1.6901408450704224e-05,
+      "loss": 0.6593,
+      "step": 97
+    },
+    {
+      "epoch": 0.1037037037037037,
+      "grad_norm": 0.34793873952604554,
+      "learning_rate": 1.7077464788732393e-05,
+      "loss": 0.8028,
+      "step": 98
+    },
+    {
+      "epoch": 0.10476190476190476,
+      "grad_norm": 0.3480073285637551,
+      "learning_rate": 1.7253521126760565e-05,
+      "loss": 0.7468,
+      "step": 99
+    },
+    {
+      "epoch": 0.10582010582010581,
+      "grad_norm": 0.28880176690327597,
+      "learning_rate": 1.7429577464788734e-05,
+      "loss": 0.692,
+      "step": 100
+    },
+    {
+      "epoch": 0.10687830687830688,
+      "grad_norm": 0.3138056785283419,
+      "learning_rate": 1.7605633802816902e-05,
+      "loss": 0.6864,
+      "step": 101
+    },
+    {
+      "epoch": 0.10793650793650794,
+      "grad_norm": 0.32366719651656767,
+      "learning_rate": 1.778169014084507e-05,
+      "loss": 0.7089,
+      "step": 102
+    },
+    {
+      "epoch": 0.10899470899470899,
+      "grad_norm": 0.33175345049430516,
+      "learning_rate": 1.7957746478873243e-05,
+      "loss": 0.8091,
+      "step": 103
+    },
+    {
+      "epoch": 0.11005291005291006,
+      "grad_norm": 0.3133523678106677,
+      "learning_rate": 1.813380281690141e-05,
+      "loss": 0.7224,
+      "step": 104
+    },
+    {
+      "epoch": 0.1111111111111111,
+      "grad_norm": 0.3061911187313938,
+      "learning_rate": 1.830985915492958e-05,
+      "loss": 0.7376,
+      "step": 105
+    },
+    {
+      "epoch": 0.11216931216931217,
+      "grad_norm": 0.3006228205551271,
+      "learning_rate": 1.848591549295775e-05,
+      "loss": 0.7126,
+      "step": 106
+    },
+    {
+      "epoch": 0.11322751322751323,
+      "grad_norm": 0.30813453357630666,
+      "learning_rate": 1.8661971830985917e-05,
+      "loss": 0.6727,
+      "step": 107
+    },
+    {
+      "epoch": 0.11428571428571428,
+      "grad_norm": 0.2877128391886516,
+      "learning_rate": 1.8838028169014086e-05,
+      "loss": 0.661,
+      "step": 108
+    },
+    {
+      "epoch": 0.11534391534391535,
+      "grad_norm": 0.3638361745005654,
+      "learning_rate": 1.9014084507042255e-05,
+      "loss": 0.7072,
+      "step": 109
+    },
+    {
+      "epoch": 0.1164021164021164,
+      "grad_norm": 0.30920630591593606,
+      "learning_rate": 1.9190140845070423e-05,
+      "loss": 0.6213,
+      "step": 110
+    },
+    {
+      "epoch": 0.11746031746031746,
+      "grad_norm": 0.5492672965501992,
+      "learning_rate": 1.9366197183098592e-05,
+      "loss": 0.7353,
+      "step": 111
+    },
+    {
+      "epoch": 0.11851851851851852,
+      "grad_norm": 0.3885248761167444,
+      "learning_rate": 1.954225352112676e-05,
+      "loss": 0.8189,
+      "step": 112
+    },
+    {
+      "epoch": 0.11957671957671957,
+      "grad_norm": 0.31766936558876047,
+      "learning_rate": 1.971830985915493e-05,
+      "loss": 0.7251,
+      "step": 113
+    },
+    {
+      "epoch": 0.12063492063492064,
+      "grad_norm": 0.3878825328465268,
+      "learning_rate": 1.98943661971831e-05,
+      "loss": 0.7747,
+      "step": 114
+    },
+    {
+      "epoch": 0.12169312169312169,
+      "grad_norm": 0.3260125841556687,
+      "learning_rate": 2.007042253521127e-05,
+      "loss": 0.6867,
+      "step": 115
+    },
+    {
+      "epoch": 0.12275132275132275,
+      "grad_norm": 0.3582626153310118,
+      "learning_rate": 2.024647887323944e-05,
+      "loss": 0.6981,
+      "step": 116
+    },
+    {
+      "epoch": 0.12380952380952381,
+      "grad_norm": 0.3585469784947143,
+      "learning_rate": 2.0422535211267607e-05,
+      "loss": 0.7825,
+      "step": 117
+    },
+    {
+      "epoch": 0.12486772486772486,
+      "grad_norm": 0.3442159652247866,
+      "learning_rate": 2.0598591549295776e-05,
+      "loss": 0.7701,
+      "step": 118
+    },
+    {
+      "epoch": 0.1259259259259259,
+      "grad_norm": 0.3403066620592321,
+      "learning_rate": 2.0774647887323944e-05,
+      "loss": 0.6793,
+      "step": 119
+    },
+    {
+      "epoch": 0.12698412698412698,
+      "grad_norm": 0.33853441823411223,
+      "learning_rate": 2.0950704225352113e-05,
+      "loss": 0.6876,
+      "step": 120
+    },
+    {
+      "epoch": 0.12804232804232804,
+      "grad_norm": 0.3153321491578789,
+      "learning_rate": 2.112676056338028e-05,
+      "loss": 0.5972,
+      "step": 121
+    },
+    {
+      "epoch": 0.1291005291005291,
+      "grad_norm": 0.41035577764059683,
+      "learning_rate": 2.130281690140845e-05,
+      "loss": 0.7687,
+      "step": 122
+    },
+    {
+      "epoch": 0.13015873015873017,
+      "grad_norm": 0.34224845150721145,
+      "learning_rate": 2.147887323943662e-05,
+      "loss": 0.6678,
+      "step": 123
+    },
+    {
+      "epoch": 0.1312169312169312,
+      "grad_norm": 0.37838653308297,
+      "learning_rate": 2.1654929577464787e-05,
+      "loss": 0.664,
+      "step": 124
+    },
+    {
+      "epoch": 0.13227513227513227,
+      "grad_norm": 0.356256058295494,
+      "learning_rate": 2.1830985915492956e-05,
+      "loss": 0.7416,
+      "step": 125
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "grad_norm": 0.3670688477705132,
+      "learning_rate": 2.2007042253521128e-05,
+      "loss": 0.6603,
+      "step": 126
+    },
+    {
+      "epoch": 0.1343915343915344,
+      "grad_norm": 0.3696864614307807,
+      "learning_rate": 2.2183098591549297e-05,
+      "loss": 0.6723,
+      "step": 127
+    },
+    {
+      "epoch": 0.13544973544973546,
+      "grad_norm": 0.305168106942622,
+      "learning_rate": 2.2359154929577465e-05,
+      "loss": 0.6295,
+      "step": 128
+    },
+    {
+      "epoch": 0.1365079365079365,
+      "grad_norm": 1.5741882840916825,
+      "learning_rate": 2.2535211267605634e-05,
+      "loss": 0.738,
+      "step": 129
+    },
+    {
+      "epoch": 0.13756613756613756,
+      "grad_norm": 0.40676802977095083,
+      "learning_rate": 2.2711267605633806e-05,
+      "loss": 0.6888,
+      "step": 130
+    },
+    {
+      "epoch": 0.13862433862433862,
+      "grad_norm": 0.40640312190176336,
+      "learning_rate": 2.2887323943661974e-05,
+      "loss": 0.8524,
+      "step": 131
+    },
+    {
+      "epoch": 0.13968253968253969,
+      "grad_norm": 0.3996709103146616,
+      "learning_rate": 2.3063380281690143e-05,
+      "loss": 0.6707,
+      "step": 132
+    },
+    {
+      "epoch": 0.14074074074074075,
+      "grad_norm": 0.3626239173606667,
+      "learning_rate": 2.323943661971831e-05,
+      "loss": 0.7601,
+      "step": 133
+    },
+    {
+      "epoch": 0.14179894179894179,
+      "grad_norm": 0.3889512118071032,
+      "learning_rate": 2.341549295774648e-05,
+      "loss": 0.6534,
+      "step": 134
+    },
+    {
+      "epoch": 0.14285714285714285,
+      "grad_norm": 0.439542393481695,
+      "learning_rate": 2.359154929577465e-05,
+      "loss": 0.7289,
+      "step": 135
+    },
+    {
+      "epoch": 0.1439153439153439,
+      "grad_norm": 0.36264558696354626,
+      "learning_rate": 2.3767605633802817e-05,
+      "loss": 0.6388,
+      "step": 136
+    },
+    {
+      "epoch": 0.14497354497354498,
+      "grad_norm": 0.3630801750617645,
+      "learning_rate": 2.3943661971830986e-05,
+      "loss": 0.7319,
+      "step": 137
+    },
+    {
+      "epoch": 0.14603174603174604,
+      "grad_norm": 0.32798334573994153,
+      "learning_rate": 2.4119718309859158e-05,
+      "loss": 0.7374,
+      "step": 138
+    },
+    {
+      "epoch": 0.14708994708994708,
+      "grad_norm": 0.3595935453918023,
+      "learning_rate": 2.4295774647887327e-05,
+      "loss": 0.6395,
+      "step": 139
+    },
+    {
+      "epoch": 0.14814814814814814,
+      "grad_norm": 0.40173981052190955,
+      "learning_rate": 2.4471830985915495e-05,
+      "loss": 0.6469,
+      "step": 140
+    },
+    {
+      "epoch": 0.1492063492063492,
+      "grad_norm": 0.3567439766311727,
+      "learning_rate": 2.4647887323943664e-05,
+      "loss": 0.641,
+      "step": 141
+    },
+    {
+      "epoch": 0.15026455026455027,
+      "grad_norm": 0.44153899522902645,
+      "learning_rate": 2.4823943661971833e-05,
+      "loss": 0.6945,
+      "step": 142
+    },
+    {
+      "epoch": 0.15132275132275133,
+      "grad_norm": 0.41760121811058754,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6683,
+      "step": 143
+    },
+    {
+      "epoch": 0.1523809523809524,
+      "grad_norm": 0.3532367100024593,
+      "learning_rate": 2.517605633802817e-05,
+      "loss": 0.6531,
+      "step": 144
+    },
+    {
+      "epoch": 0.15343915343915343,
+      "grad_norm": 0.36386964811751354,
+      "learning_rate": 2.535211267605634e-05,
+      "loss": 0.7372,
+      "step": 145
+    },
+    {
+      "epoch": 0.1544973544973545,
+      "grad_norm": 0.3667487681947238,
+      "learning_rate": 2.5528169014084507e-05,
+      "loss": 0.6541,
+      "step": 146
+    },
+    {
+      "epoch": 0.15555555555555556,
+      "grad_norm": 0.32814632317286685,
+      "learning_rate": 2.5704225352112676e-05,
+      "loss": 0.7053,
+      "step": 147
+    },
+    {
+      "epoch": 0.15661375661375662,
+      "grad_norm": 0.36213420149359243,
+      "learning_rate": 2.5880281690140844e-05,
+      "loss": 0.7516,
+      "step": 148
+    },
+    {
+      "epoch": 0.15767195767195769,
+      "grad_norm": 0.3388384477751442,
+      "learning_rate": 2.6056338028169013e-05,
+      "loss": 0.6326,
+      "step": 149
+    },
+    {
+      "epoch": 0.15873015873015872,
+      "grad_norm": 0.3739254654391917,
+      "learning_rate": 2.623239436619718e-05,
+      "loss": 0.6875,
+      "step": 150
+    },
+    {
+      "epoch": 0.15978835978835979,
+      "grad_norm": 0.37488629646986044,
+      "learning_rate": 2.640845070422535e-05,
+      "loss": 0.6842,
+      "step": 151
+    },
+    {
+      "epoch": 0.16084656084656085,
+      "grad_norm": 0.3315805811522645,
+      "learning_rate": 2.658450704225352e-05,
+      "loss": 0.7401,
+      "step": 152
+    },
+    {
+      "epoch": 0.1619047619047619,
+      "grad_norm": 0.32245214995851174,
+      "learning_rate": 2.676056338028169e-05,
+      "loss": 0.6606,
+      "step": 153
+    },
+    {
+      "epoch": 0.16296296296296298,
+      "grad_norm": 0.29179792377833336,
+      "learning_rate": 2.693661971830986e-05,
+      "loss": 0.6626,
+      "step": 154
+    },
+    {
+      "epoch": 0.164021164021164,
+      "grad_norm": 0.3645068699975821,
+      "learning_rate": 2.711267605633803e-05,
+      "loss": 0.7877,
+      "step": 155
+    },
+    {
+      "epoch": 0.16507936507936508,
+      "grad_norm": 0.3120322106769627,
+      "learning_rate": 2.72887323943662e-05,
+      "loss": 0.6277,
+      "step": 156
+    },
+    {
+      "epoch": 0.16613756613756614,
+      "grad_norm": 0.31281154861735205,
+      "learning_rate": 2.746478873239437e-05,
+      "loss": 0.6771,
+      "step": 157
+    },
+    {
+      "epoch": 0.1671957671957672,
+      "grad_norm": 0.379934869718354,
+      "learning_rate": 2.7640845070422537e-05,
+      "loss": 0.6179,
+      "step": 158
+    },
+    {
+      "epoch": 0.16825396825396827,
+      "grad_norm": 0.30999752344647463,
+      "learning_rate": 2.7816901408450706e-05,
+      "loss": 0.6421,
+      "step": 159
+    },
+    {
+      "epoch": 0.1693121693121693,
+      "grad_norm": 0.3743307355482593,
+      "learning_rate": 2.7992957746478874e-05,
+      "loss": 0.795,
+      "step": 160
+    },
+    {
+      "epoch": 0.17037037037037037,
+      "grad_norm": 0.38423928807562113,
+      "learning_rate": 2.8169014084507046e-05,
+      "loss": 0.7782,
+      "step": 161
+    },
+    {
+      "epoch": 0.17142857142857143,
+      "grad_norm": 0.4125835959551516,
+      "learning_rate": 2.8345070422535215e-05,
+      "loss": 0.6243,
+      "step": 162
+    },
+    {
+      "epoch": 0.1724867724867725,
+      "grad_norm": 0.33067536279547244,
+      "learning_rate": 2.8521126760563384e-05,
+      "loss": 0.6739,
+      "step": 163
+    },
+    {
+      "epoch": 0.17354497354497356,
+      "grad_norm": 0.39452863321092113,
+      "learning_rate": 2.8697183098591552e-05,
+      "loss": 0.6706,
+      "step": 164
+    },
+    {
+      "epoch": 0.1746031746031746,
+      "grad_norm": 0.33341809821413604,
+      "learning_rate": 2.887323943661972e-05,
+      "loss": 0.6624,
+      "step": 165
+    },
+    {
+      "epoch": 0.17566137566137566,
+      "grad_norm": 0.3999598309398766,
+      "learning_rate": 2.904929577464789e-05,
+      "loss": 0.7741,
+      "step": 166
+    },
+    {
+      "epoch": 0.17671957671957672,
+      "grad_norm": 0.348362123360907,
+      "learning_rate": 2.9225352112676058e-05,
+      "loss": 0.6416,
+      "step": 167
+    },
+    {
+      "epoch": 0.17777777777777778,
+      "grad_norm": 0.38051950011821134,
+      "learning_rate": 2.9401408450704227e-05,
+      "loss": 0.7498,
+      "step": 168
+    },
+    {
+      "epoch": 0.17883597883597885,
+      "grad_norm": 0.3398083012362421,
+      "learning_rate": 2.9577464788732395e-05,
+      "loss": 0.6188,
+      "step": 169
+    },
+    {
+      "epoch": 0.17989417989417988,
+      "grad_norm": 0.3263197338387835,
+      "learning_rate": 2.9753521126760564e-05,
+      "loss": 0.6523,
+      "step": 170
+    },
+    {
+      "epoch": 0.18095238095238095,
+      "grad_norm": 0.29768829451839607,
+      "learning_rate": 2.9929577464788733e-05,
+      "loss": 0.5983,
+      "step": 171
+    },
+    {
+      "epoch": 0.182010582010582,
+      "grad_norm": 0.3616844593383965,
+      "learning_rate": 3.01056338028169e-05,
+      "loss": 0.6778,
+      "step": 172
+    },
+    {
+      "epoch": 0.18306878306878308,
+      "grad_norm": 0.32445494168670697,
+      "learning_rate": 3.028169014084507e-05,
+      "loss": 0.5868,
+      "step": 173
+    },
+    {
+      "epoch": 0.18412698412698414,
+      "grad_norm": 0.3744637982463626,
+      "learning_rate": 3.045774647887324e-05,
+      "loss": 0.5966,
+      "step": 174
+    },
+    {
+      "epoch": 0.18518518518518517,
+      "grad_norm": 0.34514886886805735,
+      "learning_rate": 3.063380281690141e-05,
+      "loss": 0.735,
+      "step": 175
+    },
+    {
+      "epoch": 0.18624338624338624,
+      "grad_norm": 0.3614006619260275,
+      "learning_rate": 3.0809859154929576e-05,
+      "loss": 0.6083,
+      "step": 176
+    },
+    {
+      "epoch": 0.1873015873015873,
+      "grad_norm": 0.34851465059010306,
+      "learning_rate": 3.0985915492957744e-05,
+      "loss": 0.6449,
+      "step": 177
+    },
+    {
+      "epoch": 0.18835978835978837,
+      "grad_norm": 0.3083459665743903,
+      "learning_rate": 3.116197183098591e-05,
+      "loss": 0.6277,
+      "step": 178
+    },
+    {
+      "epoch": 0.18941798941798943,
+      "grad_norm": 0.34720965786750074,
+      "learning_rate": 3.133802816901409e-05,
+      "loss": 0.6329,
+      "step": 179
+    },
+    {
+      "epoch": 0.19047619047619047,
+      "grad_norm": 0.358285604869434,
+      "learning_rate": 3.151408450704226e-05,
+      "loss": 0.6505,
+      "step": 180
+    },
+    {
+      "epoch": 0.19153439153439153,
+      "grad_norm": 0.34474320283746784,
+      "learning_rate": 3.1690140845070426e-05,
+      "loss": 0.6319,
+      "step": 181
+    },
+    {
+      "epoch": 0.1925925925925926,
+      "grad_norm": 0.3895257488548781,
+      "learning_rate": 3.1866197183098594e-05,
+      "loss": 0.8262,
+      "step": 182
+    },
+    {
+      "epoch": 0.19365079365079366,
+      "grad_norm": 0.3457383269254632,
+      "learning_rate": 3.204225352112676e-05,
+      "loss": 0.6699,
+      "step": 183
+    },
+    {
+      "epoch": 0.19470899470899472,
+      "grad_norm": 0.33258526228041074,
+      "learning_rate": 3.221830985915493e-05,
+      "loss": 0.6082,
+      "step": 184
+    },
+    {
+      "epoch": 0.19576719576719576,
+      "grad_norm": 0.352563011531979,
+      "learning_rate": 3.23943661971831e-05,
+      "loss": 0.6628,
+      "step": 185
+    },
+    {
+      "epoch": 0.19682539682539682,
+      "grad_norm": 0.3320170771366527,
+      "learning_rate": 3.257042253521127e-05,
+      "loss": 0.6525,
+      "step": 186
+    },
+    {
+      "epoch": 0.19788359788359788,
+      "grad_norm": 0.32318375497210705,
+      "learning_rate": 3.274647887323944e-05,
+      "loss": 0.5735,
+      "step": 187
+    },
+    {
+      "epoch": 0.19894179894179895,
+      "grad_norm": 0.3270203777340338,
+      "learning_rate": 3.2922535211267606e-05,
+      "loss": 0.5571,
+      "step": 188
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3469279421040238,
+      "learning_rate": 3.3098591549295775e-05,
+      "loss": 0.6787,
+      "step": 189
+    },
+    {
+      "epoch": 0.20105820105820105,
+      "grad_norm": 0.36765912576607546,
+      "learning_rate": 3.327464788732394e-05,
+      "loss": 0.7363,
+      "step": 190
+    },
+    {
+      "epoch": 0.2021164021164021,
+      "grad_norm": 0.34672109468153317,
+      "learning_rate": 3.345070422535211e-05,
+      "loss": 0.637,
+      "step": 191
+    },
+    {
+      "epoch": 0.20317460317460317,
+      "grad_norm": 0.38433667329288207,
+      "learning_rate": 3.362676056338028e-05,
+      "loss": 0.5852,
+      "step": 192
+    },
+    {
+      "epoch": 0.20423280423280424,
+      "grad_norm": 0.38599947669557544,
+      "learning_rate": 3.380281690140845e-05,
+      "loss": 0.6301,
+      "step": 193
+    },
+    {
+      "epoch": 0.2052910052910053,
+      "grad_norm": 0.40046349554165817,
+      "learning_rate": 3.397887323943662e-05,
+      "loss": 0.6748,
+      "step": 194
+    },
+    {
+      "epoch": 0.20634920634920634,
+      "grad_norm": 0.3979623426535507,
+      "learning_rate": 3.4154929577464786e-05,
+      "loss": 0.671,
+      "step": 195
+    },
+    {
+      "epoch": 0.2074074074074074,
+      "grad_norm": 0.40709279738188864,
+      "learning_rate": 3.4330985915492955e-05,
+      "loss": 0.6654,
+      "step": 196
+    },
+    {
+      "epoch": 0.20846560846560847,
+      "grad_norm": 0.36359801057653723,
+      "learning_rate": 3.450704225352113e-05,
+      "loss": 0.6099,
+      "step": 197
+    },
+    {
+      "epoch": 0.20952380952380953,
+      "grad_norm": 0.38002929487320863,
+      "learning_rate": 3.46830985915493e-05,
+      "loss": 0.7158,
+      "step": 198
+    },
+    {
+      "epoch": 0.2105820105820106,
+      "grad_norm": 0.34947905555847264,
+      "learning_rate": 3.485915492957747e-05,
+      "loss": 0.5947,
+      "step": 199
+    },
+    {
+      "epoch": 0.21164021164021163,
+      "grad_norm": 0.3671022867710289,
+      "learning_rate": 3.5035211267605636e-05,
+      "loss": 0.5809,
+      "step": 200
+    },
+    {
+      "epoch": 0.2126984126984127,
+      "grad_norm": 0.4001809161609547,
+      "learning_rate": 3.5211267605633805e-05,
+      "loss": 0.6605,
+      "step": 201
+    },
+    {
+      "epoch": 0.21375661375661376,
+      "grad_norm": 0.39279284072815546,
+      "learning_rate": 3.538732394366197e-05,
+      "loss": 0.6068,
+      "step": 202
+    },
+    {
+      "epoch": 0.21481481481481482,
+      "grad_norm": 0.33271126101318277,
+      "learning_rate": 3.556338028169014e-05,
+      "loss": 0.5607,
+      "step": 203
+    },
+    {
+      "epoch": 0.21587301587301588,
+      "grad_norm": 0.443314082948306,
+      "learning_rate": 3.573943661971831e-05,
+      "loss": 0.6927,
+      "step": 204
+    },
+    {
+      "epoch": 0.21693121693121692,
+      "grad_norm": 0.31502780054757173,
+      "learning_rate": 3.5915492957746486e-05,
+      "loss": 0.6896,
+      "step": 205
+    },
+    {
+      "epoch": 0.21798941798941798,
+      "grad_norm": 0.38688454724207,
+      "learning_rate": 3.6091549295774655e-05,
+      "loss": 0.604,
+      "step": 206
+    },
+    {
+      "epoch": 0.21904761904761905,
+      "grad_norm": 0.3312260584034657,
+      "learning_rate": 3.626760563380282e-05,
+      "loss": 0.6305,
+      "step": 207
+    },
+    {
+      "epoch": 0.2201058201058201,
+      "grad_norm": 0.34099970700466303,
+      "learning_rate": 3.644366197183099e-05,
+      "loss": 0.6235,
+      "step": 208
+    },
+    {
+      "epoch": 0.22116402116402117,
+      "grad_norm": 0.35741901452146296,
+      "learning_rate": 3.661971830985916e-05,
+      "loss": 0.7602,
+      "step": 209
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 0.3650578278726511,
+      "learning_rate": 3.679577464788733e-05,
+      "loss": 0.6775,
+      "step": 210
+    },
+    {
+      "epoch": 0.22328042328042327,
+      "grad_norm": 0.32198145647198445,
+      "learning_rate": 3.69718309859155e-05,
+      "loss": 0.6514,
+      "step": 211
+    },
+    {
+      "epoch": 0.22433862433862434,
+      "grad_norm": 0.34345547766950246,
+      "learning_rate": 3.7147887323943666e-05,
+      "loss": 0.5875,
+      "step": 212
+    },
+    {
+      "epoch": 0.2253968253968254,
+      "grad_norm": 0.274703336883991,
+      "learning_rate": 3.7323943661971835e-05,
+      "loss": 0.5498,
+      "step": 213
+    },
+    {
+      "epoch": 0.22645502645502646,
+      "grad_norm": 0.34787089300375457,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.634,
+      "step": 214
+    },
+    {
+      "epoch": 0.2275132275132275,
+      "grad_norm": 0.32858837285549813,
+      "learning_rate": 3.767605633802817e-05,
+      "loss": 0.7155,
+      "step": 215
+    },
+    {
+      "epoch": 0.22857142857142856,
+      "grad_norm": 0.3549541250892209,
+      "learning_rate": 3.785211267605634e-05,
+      "loss": 0.658,
+      "step": 216
+    },
+    {
+      "epoch": 0.22962962962962963,
+      "grad_norm": 0.3726458256752988,
+      "learning_rate": 3.802816901408451e-05,
+      "loss": 0.6604,
+      "step": 217
+    },
+    {
+      "epoch": 0.2306878306878307,
+      "grad_norm": 0.3698559943959681,
+      "learning_rate": 3.820422535211268e-05,
+      "loss": 0.7083,
+      "step": 218
+    },
+    {
+      "epoch": 0.23174603174603176,
+      "grad_norm": 0.3461402136268809,
+      "learning_rate": 3.8380281690140847e-05,
+      "loss": 0.6028,
+      "step": 219
+    },
+    {
+      "epoch": 0.2328042328042328,
+      "grad_norm": 0.36882537497597867,
+      "learning_rate": 3.8556338028169015e-05,
+      "loss": 0.5933,
+      "step": 220
+    },
+    {
+      "epoch": 0.23386243386243386,
+      "grad_norm": 0.32269501706477743,
+      "learning_rate": 3.8732394366197184e-05,
+      "loss": 0.5564,
+      "step": 221
+    },
+    {
+      "epoch": 0.23492063492063492,
+      "grad_norm": 0.40676474989986633,
+      "learning_rate": 3.890845070422535e-05,
+      "loss": 0.5672,
+      "step": 222
+    },
+    {
+      "epoch": 0.23597883597883598,
+      "grad_norm": 0.31466605195623004,
+      "learning_rate": 3.908450704225352e-05,
+      "loss": 0.6466,
+      "step": 223
+    },
+    {
+      "epoch": 0.23703703703703705,
+      "grad_norm": 0.4176582469387856,
+      "learning_rate": 3.926056338028169e-05,
+      "loss": 0.588,
+      "step": 224
+    },
+    {
+      "epoch": 0.23809523809523808,
+      "grad_norm": 0.3300399938594373,
+      "learning_rate": 3.943661971830986e-05,
+      "loss": 0.653,
+      "step": 225
+    },
+    {
+      "epoch": 0.23915343915343915,
+      "grad_norm": 0.37714569544063986,
+      "learning_rate": 3.9612676056338034e-05,
+      "loss": 0.5,
+      "step": 226
+    },
+    {
+      "epoch": 0.2402116402116402,
+      "grad_norm": 0.40552770928374765,
+      "learning_rate": 3.97887323943662e-05,
+      "loss": 0.6434,
+      "step": 227
+    },
+    {
+      "epoch": 0.24126984126984127,
+      "grad_norm": 0.4636367646766144,
+      "learning_rate": 3.996478873239437e-05,
+      "loss": 0.675,
+      "step": 228
+    },
+    {
+      "epoch": 0.24232804232804234,
+      "grad_norm": 0.4172347857566623,
+      "learning_rate": 4.014084507042254e-05,
+      "loss": 0.6065,
+      "step": 229
+    },
+    {
+      "epoch": 0.24338624338624337,
+      "grad_norm": 0.5288156764553862,
+      "learning_rate": 4.031690140845071e-05,
+      "loss": 0.6875,
+      "step": 230
+    },
+    {
+      "epoch": 0.24444444444444444,
+      "grad_norm": 0.36944896552596407,
+      "learning_rate": 4.049295774647888e-05,
+      "loss": 0.6154,
+      "step": 231
+    },
+    {
+      "epoch": 0.2455026455026455,
+      "grad_norm": 0.4691395217338354,
+      "learning_rate": 4.0669014084507045e-05,
+      "loss": 0.7015,
+      "step": 232
+    },
+    {
+      "epoch": 0.24656084656084656,
+      "grad_norm": 0.35238179145757065,
+      "learning_rate": 4.0845070422535214e-05,
+      "loss": 0.6848,
+      "step": 233
+    },
+    {
+      "epoch": 0.24761904761904763,
+      "grad_norm": 0.483662014218935,
+      "learning_rate": 4.102112676056338e-05,
+      "loss": 0.6265,
+      "step": 234
+    },
+    {
+      "epoch": 0.24867724867724866,
+      "grad_norm": 0.3765557351707901,
+      "learning_rate": 4.119718309859155e-05,
+      "loss": 0.685,
+      "step": 235
+    },
+    {
+      "epoch": 0.24973544973544973,
+      "grad_norm": 0.43161674678571094,
+      "learning_rate": 4.137323943661972e-05,
+      "loss": 0.5812,
+      "step": 236
+    },
+    {
+      "epoch": 0.2507936507936508,
+      "grad_norm": 0.3554164355775738,
+      "learning_rate": 4.154929577464789e-05,
+      "loss": 0.5958,
+      "step": 237
+    },
+    {
+      "epoch": 0.2518518518518518,
+      "grad_norm": 0.48752122898965977,
+      "learning_rate": 4.172535211267606e-05,
+      "loss": 0.6589,
+      "step": 238
+    },
+    {
+      "epoch": 0.2529100529100529,
+      "grad_norm": 0.41293508091850323,
+      "learning_rate": 4.1901408450704226e-05,
+      "loss": 0.6455,
+      "step": 239
+    },
+    {
+      "epoch": 0.25396825396825395,
+      "grad_norm": 0.4459574033198097,
+      "learning_rate": 4.2077464788732394e-05,
+      "loss": 0.696,
+      "step": 240
+    },
+    {
+      "epoch": 0.25502645502645505,
+      "grad_norm": 0.5261242261127165,
+      "learning_rate": 4.225352112676056e-05,
+      "loss": 0.612,
+      "step": 241
+    },
+    {
+      "epoch": 0.2560846560846561,
+      "grad_norm": 0.4137102192090692,
+      "learning_rate": 4.242957746478873e-05,
+      "loss": 0.5578,
+      "step": 242
+    },
+    {
+      "epoch": 0.2571428571428571,
+      "grad_norm": 0.570279904177202,
+      "learning_rate": 4.26056338028169e-05,
+      "loss": 0.6889,
+      "step": 243
+    },
+    {
+      "epoch": 0.2582010582010582,
+      "grad_norm": 0.5430310414509273,
+      "learning_rate": 4.278169014084507e-05,
+      "loss": 0.6589,
+      "step": 244
+    },
+    {
+      "epoch": 0.25925925925925924,
+      "grad_norm": 0.441503758930117,
+      "learning_rate": 4.295774647887324e-05,
+      "loss": 0.6545,
+      "step": 245
+    },
+    {
+      "epoch": 0.26031746031746034,
+      "grad_norm": 0.5384182838717864,
+      "learning_rate": 4.3133802816901406e-05,
+      "loss": 0.6461,
+      "step": 246
+    },
+    {
+      "epoch": 0.2613756613756614,
+      "grad_norm": 0.3999205740390739,
+      "learning_rate": 4.3309859154929575e-05,
+      "loss": 0.589,
+      "step": 247
+    },
+    {
+      "epoch": 0.2624338624338624,
+      "grad_norm": 0.4747833562137057,
+      "learning_rate": 4.348591549295774e-05,
+      "loss": 0.6146,
+      "step": 248
+    },
+    {
+      "epoch": 0.2634920634920635,
+      "grad_norm": 0.576119057299997,
+      "learning_rate": 4.366197183098591e-05,
+      "loss": 0.6782,
+      "step": 249
+    },
+    {
+      "epoch": 0.26455026455026454,
+      "grad_norm": 0.5322525365860533,
+      "learning_rate": 4.383802816901409e-05,
+      "loss": 0.5898,
+      "step": 250
+    },
+    {
+      "epoch": 0.2656084656084656,
+      "grad_norm": 0.5225758045153239,
+      "learning_rate": 4.4014084507042256e-05,
+      "loss": 0.5973,
+      "step": 251
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 1.1554341236749963,
+      "learning_rate": 4.4190140845070424e-05,
+      "loss": 0.4985,
+      "step": 252
+    },
+    {
+      "epoch": 0.2677248677248677,
+      "grad_norm": 0.47852793764541357,
+      "learning_rate": 4.436619718309859e-05,
+      "loss": 0.5225,
+      "step": 253
+    },
+    {
+      "epoch": 0.2687830687830688,
+      "grad_norm": 0.41663440733827256,
+      "learning_rate": 4.454225352112676e-05,
+      "loss": 0.6353,
+      "step": 254
+    },
+    {
+      "epoch": 0.2698412698412698,
+      "grad_norm": 0.5546294618558951,
+      "learning_rate": 4.471830985915493e-05,
+      "loss": 0.6392,
+      "step": 255
+    },
+    {
+      "epoch": 0.2708994708994709,
+      "grad_norm": 0.37390131385634334,
+      "learning_rate": 4.48943661971831e-05,
+      "loss": 0.69,
+      "step": 256
+    },
+    {
+      "epoch": 0.27195767195767195,
+      "grad_norm": 1.3249897376641209,
+      "learning_rate": 4.507042253521127e-05,
+      "loss": 0.5528,
+      "step": 257
+    },
+    {
+      "epoch": 0.273015873015873,
+      "grad_norm": 0.4769685334018541,
+      "learning_rate": 4.5246478873239436e-05,
+      "loss": 0.5465,
+      "step": 258
+    },
+    {
+      "epoch": 0.2740740740740741,
+      "grad_norm": 0.639538521986945,
+      "learning_rate": 4.542253521126761e-05,
+      "loss": 0.6834,
+      "step": 259
+    },
+    {
+      "epoch": 0.2751322751322751,
+      "grad_norm": 0.5219205657799278,
+      "learning_rate": 4.559859154929578e-05,
+      "loss": 0.5925,
+      "step": 260
+    },
+    {
+      "epoch": 0.2761904761904762,
+      "grad_norm": 0.47303904730740937,
+      "learning_rate": 4.577464788732395e-05,
+      "loss": 0.6291,
+      "step": 261
+    },
+    {
+      "epoch": 0.27724867724867724,
+      "grad_norm": 0.4124587679203123,
+      "learning_rate": 4.595070422535212e-05,
+      "loss": 0.5333,
+      "step": 262
+    },
+    {
+      "epoch": 0.2783068783068783,
+      "grad_norm": 0.39574829022247976,
+      "learning_rate": 4.6126760563380286e-05,
+      "loss": 0.6118,
+      "step": 263
+    },
+    {
+      "epoch": 0.27936507936507937,
+      "grad_norm": 0.4906299930682528,
+      "learning_rate": 4.6302816901408455e-05,
+      "loss": 0.5492,
+      "step": 264
+    },
+    {
+      "epoch": 0.2804232804232804,
+      "grad_norm": 0.45541184091560766,
+      "learning_rate": 4.647887323943662e-05,
+      "loss": 0.6135,
+      "step": 265
+    },
+    {
+      "epoch": 0.2814814814814815,
+      "grad_norm": 0.3569305414446637,
+      "learning_rate": 4.665492957746479e-05,
+      "loss": 0.5374,
+      "step": 266
+    },
+    {
+      "epoch": 0.28253968253968254,
+      "grad_norm": 0.43109469741112555,
+      "learning_rate": 4.683098591549296e-05,
+      "loss": 0.6969,
+      "step": 267
+    },
+    {
+      "epoch": 0.28359788359788357,
+      "grad_norm": 0.34850457997668804,
+      "learning_rate": 4.700704225352113e-05,
+      "loss": 0.5812,
+      "step": 268
+    },
+    {
+      "epoch": 0.28465608465608466,
+      "grad_norm": 0.43017148890786167,
+      "learning_rate": 4.71830985915493e-05,
+      "loss": 0.6669,
+      "step": 269
+    },
+    {
+      "epoch": 0.2857142857142857,
+      "grad_norm": 0.3979891127651658,
+      "learning_rate": 4.7359154929577466e-05,
+      "loss": 0.5677,
+      "step": 270
+    },
+    {
+      "epoch": 0.2867724867724868,
+      "grad_norm": 0.5535072160058837,
+      "learning_rate": 4.7535211267605635e-05,
+      "loss": 0.6401,
+      "step": 271
+    },
+    {
+      "epoch": 0.2878306878306878,
+      "grad_norm": 0.4788371326408822,
+      "learning_rate": 4.7711267605633804e-05,
+      "loss": 0.6775,
+      "step": 272
+    },
+    {
+      "epoch": 0.28888888888888886,
+      "grad_norm": 0.37053522739872213,
+      "learning_rate": 4.788732394366197e-05,
+      "loss": 0.6266,
+      "step": 273
+    },
+    {
+      "epoch": 0.28994708994708995,
+      "grad_norm": 0.44280982619784764,
+      "learning_rate": 4.806338028169015e-05,
+      "loss": 0.6108,
+      "step": 274
+    },
+    {
+      "epoch": 0.291005291005291,
+      "grad_norm": 0.34292181255238274,
+      "learning_rate": 4.8239436619718316e-05,
+      "loss": 0.5563,
+      "step": 275
+    },
+    {
+      "epoch": 0.2920634920634921,
+      "grad_norm": 0.4678353671213705,
+      "learning_rate": 4.8415492957746485e-05,
+      "loss": 0.6338,
+      "step": 276
+    },
+    {
+      "epoch": 0.2931216931216931,
+      "grad_norm": 0.3603311974292169,
+      "learning_rate": 4.8591549295774653e-05,
+      "loss": 0.565,
+      "step": 277
+    },
+    {
+      "epoch": 0.29417989417989415,
+      "grad_norm": 0.4616974029069646,
+      "learning_rate": 4.876760563380282e-05,
+      "loss": 0.5827,
+      "step": 278
+    },
+    {
+      "epoch": 0.29523809523809524,
+      "grad_norm": 0.4417612006792997,
+      "learning_rate": 4.894366197183099e-05,
+      "loss": 0.7125,
+      "step": 279
+    },
+    {
+      "epoch": 0.2962962962962963,
+      "grad_norm": 0.6134549574544247,
+      "learning_rate": 4.911971830985916e-05,
+      "loss": 0.6591,
+      "step": 280
+    },
+    {
+      "epoch": 0.29735449735449737,
+      "grad_norm": 0.37720738345155475,
+      "learning_rate": 4.929577464788733e-05,
+      "loss": 0.5585,
+      "step": 281
+    },
+    {
+      "epoch": 0.2984126984126984,
+      "grad_norm": 0.4237444553725601,
+      "learning_rate": 4.9471830985915497e-05,
+      "loss": 0.6572,
+      "step": 282
+    },
+    {
+      "epoch": 0.29947089947089944,
+      "grad_norm": 0.36275418041559837,
+      "learning_rate": 4.9647887323943665e-05,
+      "loss": 0.6876,
+      "step": 283
+    },
+    {
+      "epoch": 0.30052910052910053,
+      "grad_norm": 0.5213976463565436,
+      "learning_rate": 4.9823943661971834e-05,
+      "loss": 0.6989,
+      "step": 284
+    },
+    {
+      "epoch": 0.30158730158730157,
+      "grad_norm": 0.3570263641303588,
+      "learning_rate": 5e-05,
+      "loss": 0.6558,
+      "step": 285
+    },
+    {
+      "epoch": 0.30264550264550266,
+      "grad_norm": 0.4185098108573085,
+      "learning_rate": 4.998039984319875e-05,
+      "loss": 0.5007,
+      "step": 286
+    },
+    {
+      "epoch": 0.3037037037037037,
+      "grad_norm": 0.3850834324831248,
+      "learning_rate": 4.9960799686397494e-05,
+      "loss": 0.5663,
+      "step": 287
+    },
+    {
+      "epoch": 0.3047619047619048,
+      "grad_norm": 0.42577227525237116,
+      "learning_rate": 4.994119952959624e-05,
+      "loss": 0.6345,
+      "step": 288
+    },
+    {
+      "epoch": 0.3058201058201058,
+      "grad_norm": 0.365037307220747,
+      "learning_rate": 4.9921599372794986e-05,
+      "loss": 0.6379,
+      "step": 289
+    },
+    {
+      "epoch": 0.30687830687830686,
+      "grad_norm": 0.4118237945823454,
+      "learning_rate": 4.990199921599373e-05,
+      "loss": 0.634,
+      "step": 290
+    },
+    {
+      "epoch": 0.30793650793650795,
+      "grad_norm": 0.33860288653166326,
+      "learning_rate": 4.988239905919248e-05,
+      "loss": 0.5576,
+      "step": 291
+    },
+    {
+      "epoch": 0.308994708994709,
+      "grad_norm": 0.4301618112703962,
+      "learning_rate": 4.986279890239122e-05,
+      "loss": 0.5168,
+      "step": 292
+    },
+    {
+      "epoch": 0.3100529100529101,
+      "grad_norm": 0.3153318738893075,
+      "learning_rate": 4.984319874558997e-05,
+      "loss": 0.5811,
+      "step": 293
+    },
+    {
+      "epoch": 0.3111111111111111,
+      "grad_norm": 0.363675255321121,
+      "learning_rate": 4.982359858878871e-05,
+      "loss": 0.5799,
+      "step": 294
+    },
+    {
+      "epoch": 0.31216931216931215,
+      "grad_norm": 0.34691937369358705,
+      "learning_rate": 4.980399843198746e-05,
+      "loss": 0.5285,
+      "step": 295
+    },
+    {
+      "epoch": 0.31322751322751324,
+      "grad_norm": 0.36271401144515686,
+      "learning_rate": 4.9784398275186204e-05,
+      "loss": 0.5956,
+      "step": 296
+    },
+    {
+      "epoch": 0.3142857142857143,
+      "grad_norm": 0.39889753761155844,
+      "learning_rate": 4.9764798118384946e-05,
+      "loss": 0.6149,
+      "step": 297
+    },
+    {
+      "epoch": 0.31534391534391537,
+      "grad_norm": 1.4954635983258886,
+      "learning_rate": 4.9745197961583695e-05,
+      "loss": 0.6453,
+      "step": 298
+    },
+    {
+      "epoch": 0.3164021164021164,
+      "grad_norm": 0.453710769398266,
+      "learning_rate": 4.9725597804782445e-05,
+      "loss": 0.6233,
+      "step": 299
+    },
+    {
+      "epoch": 0.31746031746031744,
+      "grad_norm": 0.3324927531881442,
+      "learning_rate": 4.970599764798119e-05,
+      "loss": 0.5543,
+      "step": 300
+    },
+    {
+      "epoch": 0.31851851851851853,
+      "grad_norm": 0.3623203424678947,
+      "learning_rate": 4.968639749117993e-05,
+      "loss": 0.6053,
+      "step": 301
+    },
+    {
+      "epoch": 0.31957671957671957,
+      "grad_norm": 0.37974058426507107,
+      "learning_rate": 4.966679733437868e-05,
+      "loss": 0.6463,
+      "step": 302
+    },
+    {
+      "epoch": 0.32063492063492066,
+      "grad_norm": 0.37093076432402033,
+      "learning_rate": 4.964719717757742e-05,
+      "loss": 0.6168,
+      "step": 303
+    },
+    {
+      "epoch": 0.3216931216931217,
+      "grad_norm": 0.4056043538201469,
+      "learning_rate": 4.962759702077617e-05,
+      "loss": 0.7495,
+      "step": 304
+    },
+    {
+      "epoch": 0.32275132275132273,
+      "grad_norm": 0.328152358631735,
+      "learning_rate": 4.960799686397491e-05,
+      "loss": 0.5899,
+      "step": 305
+    },
+    {
+      "epoch": 0.3238095238095238,
+      "grad_norm": 0.35175655000013734,
+      "learning_rate": 4.958839670717366e-05,
+      "loss": 0.5318,
+      "step": 306
+    },
+    {
+      "epoch": 0.32486772486772486,
+      "grad_norm": 0.3569933392052981,
+      "learning_rate": 4.9568796550372405e-05,
+      "loss": 0.6631,
+      "step": 307
+    },
+    {
+      "epoch": 0.32592592592592595,
+      "grad_norm": 0.37456493527809115,
+      "learning_rate": 4.9549196393571154e-05,
+      "loss": 0.7267,
+      "step": 308
+    },
+    {
+      "epoch": 0.326984126984127,
+      "grad_norm": 6.48230463042188,
+      "learning_rate": 4.9529596236769897e-05,
+      "loss": 0.9326,
+      "step": 309
+    },
+    {
+      "epoch": 0.328042328042328,
+      "grad_norm": 0.597786693643111,
+      "learning_rate": 4.950999607996864e-05,
+      "loss": 0.6035,
+      "step": 310
+    },
+    {
+      "epoch": 0.3291005291005291,
+      "grad_norm": 1.0072574874604938,
+      "learning_rate": 4.949039592316739e-05,
+      "loss": 0.6678,
+      "step": 311
+    },
+    {
+      "epoch": 0.33015873015873015,
+      "grad_norm": 2.015193754844906,
+      "learning_rate": 4.947079576636614e-05,
+      "loss": 0.6074,
+      "step": 312
+    },
+    {
+      "epoch": 0.33121693121693124,
+      "grad_norm": 0.9965912669595596,
+      "learning_rate": 4.945119560956487e-05,
+      "loss": 0.5941,
+      "step": 313
+    },
+    {
+      "epoch": 0.3322751322751323,
+      "grad_norm": 0.6804805860670657,
+      "learning_rate": 4.943159545276362e-05,
+      "loss": 0.6363,
+      "step": 314
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.6017533613165538,
+      "learning_rate": 4.941199529596237e-05,
+      "loss": 0.6276,
+      "step": 315
+    },
+    {
+      "epoch": 0.3343915343915344,
+      "grad_norm": 0.704554658779125,
+      "learning_rate": 4.939239513916112e-05,
+      "loss": 0.6388,
+      "step": 316
+    },
+    {
+      "epoch": 0.33544973544973544,
+      "grad_norm": 0.5032673111541713,
+      "learning_rate": 4.937279498235986e-05,
+      "loss": 0.5389,
+      "step": 317
+    },
+    {
+      "epoch": 0.33650793650793653,
+      "grad_norm": 0.5015885516895849,
+      "learning_rate": 4.9353194825558606e-05,
+      "loss": 0.5515,
+      "step": 318
+    },
+    {
+      "epoch": 0.33756613756613757,
+      "grad_norm": 0.5097653076876837,
+      "learning_rate": 4.9333594668757355e-05,
+      "loss": 0.5483,
+      "step": 319
+    },
+    {
+      "epoch": 0.3386243386243386,
+      "grad_norm": 0.689407061137572,
+      "learning_rate": 4.93139945119561e-05,
+      "loss": 0.6595,
+      "step": 320
+    },
+    {
+      "epoch": 0.3396825396825397,
+      "grad_norm": 1.3956902752361948,
+      "learning_rate": 4.929439435515485e-05,
+      "loss": 0.6863,
+      "step": 321
+    },
+    {
+      "epoch": 0.34074074074074073,
+      "grad_norm": 0.6477224984591293,
+      "learning_rate": 4.927479419835359e-05,
+      "loss": 0.6719,
+      "step": 322
+    },
+    {
+      "epoch": 0.3417989417989418,
+      "grad_norm": 0.4791685078247184,
+      "learning_rate": 4.925519404155233e-05,
+      "loss": 0.6532,
+      "step": 323
+    },
+    {
+      "epoch": 0.34285714285714286,
+      "grad_norm": 0.7647455686764629,
+      "learning_rate": 4.923559388475108e-05,
+      "loss": 0.5633,
+      "step": 324
+    },
+    {
+      "epoch": 0.3439153439153439,
+      "grad_norm": 0.540162982463785,
+      "learning_rate": 4.921599372794983e-05,
+      "loss": 0.6154,
+      "step": 325
+    },
+    {
+      "epoch": 0.344973544973545,
+      "grad_norm": 0.5987926929939865,
+      "learning_rate": 4.9196393571148566e-05,
+      "loss": 0.5485,
+      "step": 326
+    },
+    {
+      "epoch": 0.346031746031746,
+      "grad_norm": 0.47012692994553495,
+      "learning_rate": 4.9176793414347316e-05,
+      "loss": 0.6356,
+      "step": 327
+    },
+    {
+      "epoch": 0.3470899470899471,
+      "grad_norm": 0.6190282717604556,
+      "learning_rate": 4.9157193257546065e-05,
+      "loss": 0.737,
+      "step": 328
+    },
+    {
+      "epoch": 0.34814814814814815,
+      "grad_norm": 0.48951022131560773,
+      "learning_rate": 4.9137593100744814e-05,
+      "loss": 0.6171,
+      "step": 329
+    },
+    {
+      "epoch": 0.3492063492063492,
+      "grad_norm": 0.5704216168233464,
+      "learning_rate": 4.911799294394355e-05,
+      "loss": 0.6843,
+      "step": 330
+    },
+    {
+      "epoch": 0.3502645502645503,
+      "grad_norm": 0.3828305070852499,
+      "learning_rate": 4.90983927871423e-05,
+      "loss": 0.6277,
+      "step": 331
+    },
+    {
+      "epoch": 0.3513227513227513,
+      "grad_norm": 0.4899467959752788,
+      "learning_rate": 4.907879263034105e-05,
+      "loss": 0.5997,
+      "step": 332
+    },
+    {
+      "epoch": 0.3523809523809524,
+      "grad_norm": 0.4491699388666299,
+      "learning_rate": 4.905919247353979e-05,
+      "loss": 0.5925,
+      "step": 333
+    },
+    {
+      "epoch": 0.35343915343915344,
+      "grad_norm": 0.425441565694508,
+      "learning_rate": 4.903959231673853e-05,
+      "loss": 0.5301,
+      "step": 334
+    },
+    {
+      "epoch": 0.3544973544973545,
+      "grad_norm": 0.45512477300228416,
+      "learning_rate": 4.901999215993728e-05,
+      "loss": 0.5845,
+      "step": 335
+    },
+    {
+      "epoch": 0.35555555555555557,
+      "grad_norm": 10.638459759457483,
+      "learning_rate": 4.9000392003136025e-05,
+      "loss": 0.7328,
+      "step": 336
+    },
+    {
+      "epoch": 0.3566137566137566,
+      "grad_norm": 0.6506851562077983,
+      "learning_rate": 4.8980791846334774e-05,
+      "loss": 0.5733,
+      "step": 337
+    },
+    {
+      "epoch": 0.3576719576719577,
+      "grad_norm": 0.3597461888788251,
+      "learning_rate": 4.896119168953352e-05,
+      "loss": 0.5873,
+      "step": 338
+    },
+    {
+      "epoch": 0.35873015873015873,
+      "grad_norm": 0.489755609131464,
+      "learning_rate": 4.894159153273226e-05,
+      "loss": 0.5659,
+      "step": 339
+    },
+    {
+      "epoch": 0.35978835978835977,
+      "grad_norm": 0.4683550804241493,
+      "learning_rate": 4.892199137593101e-05,
+      "loss": 0.62,
+      "step": 340
+    },
+    {
+      "epoch": 0.36084656084656086,
+      "grad_norm": 0.5994458890657961,
+      "learning_rate": 4.890239121912976e-05,
+      "loss": 0.6241,
+      "step": 341
+    },
+    {
+      "epoch": 0.3619047619047619,
+      "grad_norm": 0.36417066783567376,
+      "learning_rate": 4.88827910623285e-05,
+      "loss": 0.5937,
+      "step": 342
+    },
+    {
+      "epoch": 0.362962962962963,
+      "grad_norm": 0.5283081961653806,
+      "learning_rate": 4.886319090552724e-05,
+      "loss": 0.6184,
+      "step": 343
+    },
+    {
+      "epoch": 0.364021164021164,
+      "grad_norm": 0.39305815236656894,
+      "learning_rate": 4.884359074872599e-05,
+      "loss": 0.6288,
+      "step": 344
+    },
+    {
+      "epoch": 0.36507936507936506,
+      "grad_norm": 0.45312073034904926,
+      "learning_rate": 4.882399059192474e-05,
+      "loss": 0.6075,
+      "step": 345
+    },
+    {
+      "epoch": 0.36613756613756615,
+      "grad_norm": 0.4058520594663374,
+      "learning_rate": 4.8804390435123484e-05,
+      "loss": 0.694,
+      "step": 346
+    },
+    {
+      "epoch": 0.3671957671957672,
+      "grad_norm": 0.5002449512089666,
+      "learning_rate": 4.8784790278322226e-05,
+      "loss": 0.5748,
+      "step": 347
+    },
+    {
+      "epoch": 0.3682539682539683,
+      "grad_norm": 0.4117436134851183,
+      "learning_rate": 4.8765190121520976e-05,
+      "loss": 0.6376,
+      "step": 348
+    },
+    {
+      "epoch": 0.3693121693121693,
+      "grad_norm": 0.502404236336507,
+      "learning_rate": 4.874558996471972e-05,
+      "loss": 0.6094,
+      "step": 349
+    },
+    {
+      "epoch": 0.37037037037037035,
+      "grad_norm": 0.40455105073614056,
+      "learning_rate": 4.872598980791847e-05,
+      "loss": 0.6217,
+      "step": 350
+    },
+    {
+      "epoch": 0.37142857142857144,
+      "grad_norm": 0.35704360652012157,
+      "learning_rate": 4.870638965111721e-05,
+      "loss": 0.5024,
+      "step": 351
+    },
+    {
+      "epoch": 0.3724867724867725,
+      "grad_norm": 0.4264916243948178,
+      "learning_rate": 4.868678949431595e-05,
+      "loss": 0.5388,
+      "step": 352
+    },
+    {
+      "epoch": 0.37354497354497357,
+      "grad_norm": 0.3327240004352966,
+      "learning_rate": 4.86671893375147e-05,
+      "loss": 0.5935,
+      "step": 353
+    },
+    {
+      "epoch": 0.3746031746031746,
+      "grad_norm": 0.41835670277009634,
+      "learning_rate": 4.864758918071345e-05,
+      "loss": 0.635,
+      "step": 354
+    },
+    {
+      "epoch": 0.37566137566137564,
+      "grad_norm": 0.3307152093175963,
+      "learning_rate": 4.862798902391219e-05,
+      "loss": 0.618,
+      "step": 355
+    },
+    {
+      "epoch": 0.37671957671957673,
+      "grad_norm": 0.36724207232476874,
+      "learning_rate": 4.8608388867110936e-05,
+      "loss": 0.5644,
+      "step": 356
+    },
+    {
+      "epoch": 0.37777777777777777,
+      "grad_norm": 0.33194305741881575,
+      "learning_rate": 4.8588788710309685e-05,
+      "loss": 0.5842,
+      "step": 357
+    },
+    {
+      "epoch": 0.37883597883597886,
+      "grad_norm": 0.35340969186734694,
+      "learning_rate": 4.8569188553508434e-05,
+      "loss": 0.6276,
+      "step": 358
+    },
+    {
+      "epoch": 0.3798941798941799,
+      "grad_norm": 0.3326126922758559,
+      "learning_rate": 4.854958839670718e-05,
+      "loss": 0.6179,
+      "step": 359
+    },
+    {
+      "epoch": 0.38095238095238093,
+      "grad_norm": 0.3405811498157609,
+      "learning_rate": 4.852998823990592e-05,
+      "loss": 0.6401,
+      "step": 360
+    },
+    {
+      "epoch": 0.382010582010582,
+      "grad_norm": 0.3226096419931992,
+      "learning_rate": 4.851038808310467e-05,
+      "loss": 0.605,
+      "step": 361
+    },
+    {
+      "epoch": 0.38306878306878306,
+      "grad_norm": 0.32239461661768787,
+      "learning_rate": 4.849078792630341e-05,
+      "loss": 0.6507,
+      "step": 362
+    },
+    {
+      "epoch": 0.38412698412698415,
+      "grad_norm": 0.3484471583643884,
+      "learning_rate": 4.847118776950216e-05,
+      "loss": 0.605,
+      "step": 363
+    },
+    {
+      "epoch": 0.3851851851851852,
+      "grad_norm": 0.3112894013568265,
+      "learning_rate": 4.84515876127009e-05,
+      "loss": 0.5174,
+      "step": 364
+    },
+    {
+      "epoch": 0.3862433862433862,
+      "grad_norm": 0.2899220938284599,
+      "learning_rate": 4.843198745589965e-05,
+      "loss": 0.5094,
+      "step": 365
+    },
+    {
+      "epoch": 0.3873015873015873,
+      "grad_norm": 0.28917708457385394,
+      "learning_rate": 4.8412387299098395e-05,
+      "loss": 0.5609,
+      "step": 366
+    },
+    {
+      "epoch": 0.38835978835978835,
+      "grad_norm": 0.3538788286573949,
+      "learning_rate": 4.8392787142297144e-05,
+      "loss": 0.6225,
+      "step": 367
+    },
+    {
+      "epoch": 0.38941798941798944,
+      "grad_norm": 0.3012030881102022,
+      "learning_rate": 4.8373186985495886e-05,
+      "loss": 0.6549,
+      "step": 368
+    },
+    {
+      "epoch": 0.3904761904761905,
+      "grad_norm": 0.3265061631766346,
+      "learning_rate": 4.835358682869463e-05,
+      "loss": 0.5666,
+      "step": 369
+    },
+    {
+      "epoch": 0.3915343915343915,
+      "grad_norm": 0.2683583970944115,
+      "learning_rate": 4.833398667189338e-05,
+      "loss": 0.5693,
+      "step": 370
+    },
+    {
+      "epoch": 0.3925925925925926,
+      "grad_norm": 0.3579867664845553,
+      "learning_rate": 4.831438651509213e-05,
+      "loss": 0.6117,
+      "step": 371
+    },
+    {
+      "epoch": 0.39365079365079364,
+      "grad_norm": 0.29023429064286466,
+      "learning_rate": 4.829478635829087e-05,
+      "loss": 0.6243,
+      "step": 372
+    },
+    {
+      "epoch": 0.39470899470899473,
+      "grad_norm": 0.30720865653544066,
+      "learning_rate": 4.827518620148961e-05,
+      "loss": 0.6137,
+      "step": 373
+    },
+    {
+      "epoch": 0.39576719576719577,
+      "grad_norm": 0.28341688168565876,
+      "learning_rate": 4.825558604468836e-05,
+      "loss": 0.6066,
+      "step": 374
+    },
+    {
+      "epoch": 0.3968253968253968,
+      "grad_norm": 0.27817987141750344,
+      "learning_rate": 4.8235985887887104e-05,
+      "loss": 0.523,
+      "step": 375
+    },
+    {
+      "epoch": 0.3978835978835979,
+      "grad_norm": 0.3145876613453268,
+      "learning_rate": 4.821638573108585e-05,
+      "loss": 0.6418,
+      "step": 376
+    },
+    {
+      "epoch": 0.39894179894179893,
+      "grad_norm": 0.3096978266574846,
+      "learning_rate": 4.8196785574284596e-05,
+      "loss": 0.6102,
+      "step": 377
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.29312324806193996,
+      "learning_rate": 4.8177185417483345e-05,
+      "loss": 0.6043,
+      "step": 378
+    },
+    {
+      "epoch": 0.40105820105820106,
+      "grad_norm": 0.3215404710144968,
+      "learning_rate": 4.815758526068209e-05,
+      "loss": 0.5813,
+      "step": 379
+    },
+    {
+      "epoch": 0.4021164021164021,
+      "grad_norm": 0.2843918875735979,
+      "learning_rate": 4.813798510388084e-05,
+      "loss": 0.6212,
+      "step": 380
+    },
+    {
+      "epoch": 0.4031746031746032,
+      "grad_norm": 0.30037403275169,
+      "learning_rate": 4.811838494707958e-05,
+      "loss": 0.6729,
+      "step": 381
+    },
+    {
+      "epoch": 0.4042328042328042,
+      "grad_norm": 0.3207760435283416,
+      "learning_rate": 4.809878479027832e-05,
+      "loss": 0.55,
+      "step": 382
+    },
+    {
+      "epoch": 0.4052910052910053,
+      "grad_norm": 0.26503156107162107,
+      "learning_rate": 4.807918463347707e-05,
+      "loss": 0.5419,
+      "step": 383
+    },
+    {
+      "epoch": 0.40634920634920635,
+      "grad_norm": 0.2882065071584723,
+      "learning_rate": 4.805958447667582e-05,
+      "loss": 0.569,
+      "step": 384
+    },
+    {
+      "epoch": 0.4074074074074074,
+      "grad_norm": 0.29455047481787927,
+      "learning_rate": 4.8039984319874556e-05,
+      "loss": 0.5042,
+      "step": 385
+    },
+    {
+      "epoch": 0.4084656084656085,
+      "grad_norm": 0.29685194902902406,
+      "learning_rate": 4.8020384163073305e-05,
+      "loss": 0.5383,
+      "step": 386
+    },
+    {
+      "epoch": 0.4095238095238095,
+      "grad_norm": 0.376863384403299,
+      "learning_rate": 4.8000784006272054e-05,
+      "loss": 0.6392,
+      "step": 387
+    },
+    {
+      "epoch": 0.4105820105820106,
+      "grad_norm": 0.3371594913183494,
+      "learning_rate": 4.79811838494708e-05,
+      "loss": 0.661,
+      "step": 388
+    },
+    {
+      "epoch": 0.41164021164021164,
+      "grad_norm": 0.34032184447977354,
+      "learning_rate": 4.796158369266954e-05,
+      "loss": 0.5174,
+      "step": 389
+    },
+    {
+      "epoch": 0.4126984126984127,
+      "grad_norm": 0.3887738951246669,
+      "learning_rate": 4.794198353586829e-05,
+      "loss": 0.6675,
+      "step": 390
+    },
+    {
+      "epoch": 0.41375661375661377,
+      "grad_norm": 0.29671883772859686,
+      "learning_rate": 4.792238337906704e-05,
+      "loss": 0.5057,
+      "step": 391
+    },
+    {
+      "epoch": 0.4148148148148148,
+      "grad_norm": 0.35504708481760416,
+      "learning_rate": 4.790278322226578e-05,
+      "loss": 0.5723,
+      "step": 392
+    },
+    {
+      "epoch": 0.4158730158730159,
+      "grad_norm": 0.3187145224838173,
+      "learning_rate": 4.788318306546453e-05,
+      "loss": 0.5921,
+      "step": 393
+    },
+    {
+      "epoch": 0.41693121693121693,
+      "grad_norm": 0.296017951892389,
+      "learning_rate": 4.786358290866327e-05,
+      "loss": 0.5281,
+      "step": 394
+    },
+    {
+      "epoch": 0.41798941798941797,
+      "grad_norm": 0.39656850609570043,
+      "learning_rate": 4.7843982751862015e-05,
+      "loss": 0.6467,
+      "step": 395
+    },
+    {
+      "epoch": 0.41904761904761906,
+      "grad_norm": 0.32082406571324745,
+      "learning_rate": 4.7824382595060764e-05,
+      "loss": 0.5867,
+      "step": 396
+    },
+    {
+      "epoch": 0.4201058201058201,
+      "grad_norm": 0.4041020790707597,
+      "learning_rate": 4.780478243825951e-05,
+      "loss": 0.533,
+      "step": 397
+    },
+    {
+      "epoch": 0.4211640211640212,
+      "grad_norm": 0.38193752458595226,
+      "learning_rate": 4.778518228145825e-05,
+      "loss": 0.5414,
+      "step": 398
+    },
+    {
+      "epoch": 0.4222222222222222,
+      "grad_norm": 0.30383562921606366,
+      "learning_rate": 4.7765582124657e-05,
+      "loss": 0.5696,
+      "step": 399
+    },
+    {
+      "epoch": 0.42328042328042326,
+      "grad_norm": 0.3922882135971307,
+      "learning_rate": 4.774598196785575e-05,
+      "loss": 0.5608,
+      "step": 400
+    },
+    {
+      "epoch": 0.42433862433862435,
+      "grad_norm": 0.43007597049774227,
+      "learning_rate": 4.772638181105449e-05,
+      "loss": 0.5647,
+      "step": 401
+    },
+    {
+      "epoch": 0.4253968253968254,
+      "grad_norm": 0.5686439579885741,
+      "learning_rate": 4.770678165425323e-05,
+      "loss": 0.4939,
+      "step": 402
+    },
+    {
+      "epoch": 0.4264550264550265,
+      "grad_norm": 0.36392633625996323,
+      "learning_rate": 4.768718149745198e-05,
+      "loss": 0.5443,
+      "step": 403
+    },
+    {
+      "epoch": 0.4275132275132275,
+      "grad_norm": 0.3272462847928772,
+      "learning_rate": 4.766758134065073e-05,
+      "loss": 0.5693,
+      "step": 404
+    },
+    {
+      "epoch": 0.42857142857142855,
+      "grad_norm": 0.3124002758364203,
+      "learning_rate": 4.7647981183849473e-05,
+      "loss": 0.6362,
+      "step": 405
+    },
+    {
+      "epoch": 0.42962962962962964,
+      "grad_norm": 0.29871765035322945,
+      "learning_rate": 4.7628381027048216e-05,
+      "loss": 0.5147,
+      "step": 406
+    },
+    {
+      "epoch": 0.4306878306878307,
+      "grad_norm": 0.3962457157308558,
+      "learning_rate": 4.7608780870246965e-05,
+      "loss": 0.5617,
+      "step": 407
+    },
+    {
+      "epoch": 0.43174603174603177,
+      "grad_norm": 0.334309491928329,
+      "learning_rate": 4.758918071344571e-05,
+      "loss": 0.5507,
+      "step": 408
+    },
+    {
+      "epoch": 0.4328042328042328,
+      "grad_norm": 0.3494500717283481,
+      "learning_rate": 4.756958055664446e-05,
+      "loss": 0.5789,
+      "step": 409
+    },
+    {
+      "epoch": 0.43386243386243384,
+      "grad_norm": 0.9776571014024243,
+      "learning_rate": 4.75499803998432e-05,
+      "loss": 0.5971,
+      "step": 410
+    },
+    {
+      "epoch": 0.43492063492063493,
+      "grad_norm": 0.36178604692099614,
+      "learning_rate": 4.753038024304194e-05,
+      "loss": 0.5824,
+      "step": 411
+    },
+    {
+      "epoch": 0.43597883597883597,
+      "grad_norm": 0.3701183639060816,
+      "learning_rate": 4.751078008624069e-05,
+      "loss": 0.6046,
+      "step": 412
+    },
+    {
+      "epoch": 0.43703703703703706,
+      "grad_norm": 0.37912182400136274,
+      "learning_rate": 4.749117992943944e-05,
+      "loss": 0.5887,
+      "step": 413
+    },
+    {
+      "epoch": 0.4380952380952381,
+      "grad_norm": 0.34637123340056375,
+      "learning_rate": 4.747157977263818e-05,
+      "loss": 0.5773,
+      "step": 414
+    },
+    {
+      "epoch": 0.43915343915343913,
+      "grad_norm": 0.4073065757199422,
+      "learning_rate": 4.7451979615836925e-05,
+      "loss": 0.5738,
+      "step": 415
+    },
+    {
+      "epoch": 0.4402116402116402,
+      "grad_norm": 0.3344005038216227,
+      "learning_rate": 4.7432379459035675e-05,
+      "loss": 0.5061,
+      "step": 416
+    },
+    {
+      "epoch": 0.44126984126984126,
+      "grad_norm": 0.3739197194136766,
+      "learning_rate": 4.7412779302234424e-05,
+      "loss": 0.585,
+      "step": 417
+    },
+    {
+      "epoch": 0.44232804232804235,
+      "grad_norm": 0.6438115315588397,
+      "learning_rate": 4.7393179145433166e-05,
+      "loss": 0.6128,
+      "step": 418
+    },
+    {
+      "epoch": 0.4433862433862434,
+      "grad_norm": 0.4877567334363792,
+      "learning_rate": 4.737357898863191e-05,
+      "loss": 0.6998,
+      "step": 419
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.33503924359280984,
+      "learning_rate": 4.735397883183066e-05,
+      "loss": 0.6582,
+      "step": 420
+    },
+    {
+      "epoch": 0.4455026455026455,
+      "grad_norm": 0.35412692478679975,
+      "learning_rate": 4.73343786750294e-05,
+      "loss": 0.5902,
+      "step": 421
+    },
+    {
+      "epoch": 0.44656084656084655,
+      "grad_norm": 0.32185468036018056,
+      "learning_rate": 4.731477851822815e-05,
+      "loss": 0.5647,
+      "step": 422
+    },
+    {
+      "epoch": 0.44761904761904764,
+      "grad_norm": 0.33447079814754294,
+      "learning_rate": 4.729517836142689e-05,
+      "loss": 0.5391,
+      "step": 423
+    },
+    {
+      "epoch": 0.4486772486772487,
+      "grad_norm": 5.115754194500417,
+      "learning_rate": 4.7275578204625635e-05,
+      "loss": 0.6673,
+      "step": 424
+    },
+    {
+      "epoch": 0.4497354497354497,
+      "grad_norm": 1.328092342445049,
+      "learning_rate": 4.7255978047824384e-05,
+      "loss": 0.7066,
+      "step": 425
+    },
+    {
+      "epoch": 0.4507936507936508,
+      "grad_norm": 0.40429664376112756,
+      "learning_rate": 4.7236377891023133e-05,
+      "loss": 0.6268,
+      "step": 426
+    },
+    {
+      "epoch": 0.45185185185185184,
+      "grad_norm": 0.35636211087696995,
+      "learning_rate": 4.7216777734221876e-05,
+      "loss": 0.5591,
+      "step": 427
+    },
+    {
+      "epoch": 0.45291005291005293,
+      "grad_norm": 0.8540612192195892,
+      "learning_rate": 4.719717757742062e-05,
+      "loss": 0.5389,
+      "step": 428
+    },
+    {
+      "epoch": 0.45396825396825397,
+      "grad_norm": 0.40756442415791055,
+      "learning_rate": 4.717757742061937e-05,
+      "loss": 0.5483,
+      "step": 429
+    },
+    {
+      "epoch": 0.455026455026455,
+      "grad_norm": 0.3970491214306376,
+      "learning_rate": 4.715797726381812e-05,
+      "loss": 0.6473,
+      "step": 430
+    },
+    {
+      "epoch": 0.4560846560846561,
+      "grad_norm": 0.33077010193886147,
+      "learning_rate": 4.713837710701686e-05,
+      "loss": 0.5264,
+      "step": 431
+    },
+    {
+      "epoch": 0.45714285714285713,
+      "grad_norm": 0.37580554982943354,
+      "learning_rate": 4.71187769502156e-05,
+      "loss": 0.5222,
+      "step": 432
+    },
+    {
+      "epoch": 0.4582010582010582,
+      "grad_norm": 0.4135755906441715,
+      "learning_rate": 4.709917679341435e-05,
+      "loss": 0.5681,
+      "step": 433
+    },
+    {
+      "epoch": 0.45925925925925926,
+      "grad_norm": 0.3556795364307751,
+      "learning_rate": 4.7079576636613094e-05,
+      "loss": 0.616,
+      "step": 434
+    },
+    {
+      "epoch": 0.4603174603174603,
+      "grad_norm": 0.3255431792109049,
+      "learning_rate": 4.705997647981184e-05,
+      "loss": 0.6155,
+      "step": 435
+    },
+    {
+      "epoch": 0.4613756613756614,
+      "grad_norm": 0.3621995346427838,
+      "learning_rate": 4.7040376323010585e-05,
+      "loss": 0.5851,
+      "step": 436
+    },
+    {
+      "epoch": 0.4624338624338624,
+      "grad_norm": 0.31078411014873397,
+      "learning_rate": 4.702077616620933e-05,
+      "loss": 0.5034,
+      "step": 437
+    },
+    {
+      "epoch": 0.4634920634920635,
+      "grad_norm": 0.27503733967112304,
+      "learning_rate": 4.700117600940808e-05,
+      "loss": 0.5851,
+      "step": 438
+    },
+    {
+      "epoch": 0.46455026455026455,
+      "grad_norm": 0.34224618740297685,
+      "learning_rate": 4.6981575852606826e-05,
+      "loss": 0.6285,
+      "step": 439
+    },
+    {
+      "epoch": 0.4656084656084656,
+      "grad_norm": 0.27425336801037914,
+      "learning_rate": 4.696197569580557e-05,
+      "loss": 0.5225,
+      "step": 440
+    },
+    {
+      "epoch": 0.4666666666666667,
+      "grad_norm": 0.3603257769316038,
+      "learning_rate": 4.694237553900431e-05,
+      "loss": 0.5921,
+      "step": 441
+    },
+    {
+      "epoch": 0.4677248677248677,
+      "grad_norm": 0.32429639783867814,
+      "learning_rate": 4.692277538220306e-05,
+      "loss": 0.6941,
+      "step": 442
+    },
+    {
+      "epoch": 0.4687830687830688,
+      "grad_norm": 0.34481993099571034,
+      "learning_rate": 4.690317522540181e-05,
+      "loss": 0.5203,
+      "step": 443
+    },
+    {
+      "epoch": 0.46984126984126984,
+      "grad_norm": 0.3193736572025401,
+      "learning_rate": 4.688357506860055e-05,
+      "loss": 0.6302,
+      "step": 444
+    },
+    {
+      "epoch": 0.4708994708994709,
+      "grad_norm": 1.6187717657921294,
+      "learning_rate": 4.6863974911799295e-05,
+      "loss": 0.6413,
+      "step": 445
+    },
+    {
+      "epoch": 0.47195767195767196,
+      "grad_norm": 0.45961110444236447,
+      "learning_rate": 4.6844374754998044e-05,
+      "loss": 0.6542,
+      "step": 446
+    },
+    {
+      "epoch": 0.473015873015873,
+      "grad_norm": 0.34842912443778445,
+      "learning_rate": 4.682477459819679e-05,
+      "loss": 0.5353,
+      "step": 447
+    },
+    {
+      "epoch": 0.4740740740740741,
+      "grad_norm": 0.6192864098610441,
+      "learning_rate": 4.6805174441395536e-05,
+      "loss": 0.4736,
+      "step": 448
+    },
+    {
+      "epoch": 0.47513227513227513,
+      "grad_norm": 0.43463315211498527,
+      "learning_rate": 4.678557428459428e-05,
+      "loss": 0.6634,
+      "step": 449
+    },
+    {
+      "epoch": 0.47619047619047616,
+      "grad_norm": 0.4140965336433174,
+      "learning_rate": 4.676597412779302e-05,
+      "loss": 0.6842,
+      "step": 450
+    },
+    {
+      "epoch": 0.47724867724867726,
+      "grad_norm": 0.3907863076094101,
+      "learning_rate": 4.674637397099177e-05,
+      "loss": 0.5236,
+      "step": 451
+    },
+    {
+      "epoch": 0.4783068783068783,
+      "grad_norm": 0.3553573532226626,
+      "learning_rate": 4.672677381419052e-05,
+      "loss": 0.6322,
+      "step": 452
+    },
+    {
+      "epoch": 0.4793650793650794,
+      "grad_norm": 0.32855146366281446,
+      "learning_rate": 4.670717365738926e-05,
+      "loss": 0.5676,
+      "step": 453
+    },
+    {
+      "epoch": 0.4804232804232804,
+      "grad_norm": 0.3355219316367844,
+      "learning_rate": 4.6687573500588004e-05,
+      "loss": 0.6183,
+      "step": 454
+    },
+    {
+      "epoch": 0.48148148148148145,
+      "grad_norm": 0.33915222521124233,
+      "learning_rate": 4.6667973343786754e-05,
+      "loss": 0.5957,
+      "step": 455
+    },
+    {
+      "epoch": 0.48253968253968255,
+      "grad_norm": 0.2999900043336735,
+      "learning_rate": 4.66483731869855e-05,
+      "loss": 0.6239,
+      "step": 456
+    },
+    {
+      "epoch": 0.4835978835978836,
+      "grad_norm": 1.932322840844984,
+      "learning_rate": 4.662877303018424e-05,
+      "loss": 0.6079,
+      "step": 457
+    },
+    {
+      "epoch": 0.4846560846560847,
+      "grad_norm": 0.3841930370997371,
+      "learning_rate": 4.660917287338299e-05,
+      "loss": 0.5339,
+      "step": 458
+    },
+    {
+      "epoch": 0.4857142857142857,
+      "grad_norm": 0.29299834519404916,
+      "learning_rate": 4.658957271658174e-05,
+      "loss": 0.5155,
+      "step": 459
+    },
+    {
+      "epoch": 0.48677248677248675,
+      "grad_norm": 0.3508678158463962,
+      "learning_rate": 4.656997255978048e-05,
+      "loss": 0.5569,
+      "step": 460
+    },
+    {
+      "epoch": 0.48783068783068784,
+      "grad_norm": 0.370161022664498,
+      "learning_rate": 4.655037240297922e-05,
+      "loss": 0.6227,
+      "step": 461
+    },
+    {
+      "epoch": 0.4888888888888889,
+      "grad_norm": 0.336334840849176,
+      "learning_rate": 4.653077224617797e-05,
+      "loss": 0.5896,
+      "step": 462
+    },
+    {
+      "epoch": 0.48994708994708996,
+      "grad_norm": 0.3387385164468496,
+      "learning_rate": 4.651117208937672e-05,
+      "loss": 0.5575,
+      "step": 463
+    },
+    {
+      "epoch": 0.491005291005291,
+      "grad_norm": 0.29356754801120044,
+      "learning_rate": 4.649157193257546e-05,
+      "loss": 0.5692,
+      "step": 464
+    },
+    {
+      "epoch": 0.49206349206349204,
+      "grad_norm": 0.32417509452640764,
+      "learning_rate": 4.647197177577421e-05,
+      "loss": 0.542,
+      "step": 465
+    },
+    {
+      "epoch": 0.4931216931216931,
+      "grad_norm": 0.26738840774433187,
+      "learning_rate": 4.6452371618972955e-05,
+      "loss": 0.5007,
+      "step": 466
+    },
+    {
+      "epoch": 0.49417989417989416,
+      "grad_norm": 0.3639897702647601,
+      "learning_rate": 4.64327714621717e-05,
+      "loss": 0.5834,
+      "step": 467
+    },
+    {
+      "epoch": 0.49523809523809526,
+      "grad_norm": 7.352056250865767,
+      "learning_rate": 4.641317130537045e-05,
+      "loss": 0.7152,
+      "step": 468
+    },
+    {
+      "epoch": 0.4962962962962963,
+      "grad_norm": 0.35916204859142065,
+      "learning_rate": 4.6393571148569196e-05,
+      "loss": 0.508,
+      "step": 469
+    },
+    {
+      "epoch": 0.4973544973544973,
+      "grad_norm": 1.1916607882832544,
+      "learning_rate": 4.637397099176793e-05,
+      "loss": 0.716,
+      "step": 470
+    },
+    {
+      "epoch": 0.4984126984126984,
+      "grad_norm": 0.2879923677665714,
+      "learning_rate": 4.635437083496668e-05,
+      "loss": 0.4927,
+      "step": 471
+    },
+    {
+      "epoch": 0.49947089947089945,
+      "grad_norm": 0.3327630140013385,
+      "learning_rate": 4.633477067816543e-05,
+      "loss": 0.567,
+      "step": 472
+    },
+    {
+      "epoch": 0.5005291005291005,
+      "grad_norm": 0.47964078282416894,
+      "learning_rate": 4.631517052136417e-05,
+      "loss": 0.5215,
+      "step": 473
+    },
+    {
+      "epoch": 0.5015873015873016,
+      "grad_norm": 0.9360332803048105,
+      "learning_rate": 4.6295570364562915e-05,
+      "loss": 0.5973,
+      "step": 474
+    },
+    {
+      "epoch": 0.5026455026455027,
+      "grad_norm": 0.39106581876004903,
+      "learning_rate": 4.6275970207761664e-05,
+      "loss": 0.5984,
+      "step": 475
+    },
+    {
+      "epoch": 0.5037037037037037,
+      "grad_norm": 1.6307210187055834,
+      "learning_rate": 4.6256370050960414e-05,
+      "loss": 0.557,
+      "step": 476
+    },
+    {
+      "epoch": 0.5047619047619047,
+      "grad_norm": 0.29838204099866084,
+      "learning_rate": 4.6236769894159156e-05,
+      "loss": 0.584,
+      "step": 477
+    },
+    {
+      "epoch": 0.5058201058201058,
+      "grad_norm": 0.3257371970902655,
+      "learning_rate": 4.62171697373579e-05,
+      "loss": 0.5222,
+      "step": 478
+    },
+    {
+      "epoch": 0.5068783068783069,
+      "grad_norm": 0.2892518749067453,
+      "learning_rate": 4.619756958055665e-05,
+      "loss": 0.6022,
+      "step": 479
+    },
+    {
+      "epoch": 0.5079365079365079,
+      "grad_norm": 0.3395163891295951,
+      "learning_rate": 4.617796942375539e-05,
+      "loss": 0.6507,
+      "step": 480
+    },
+    {
+      "epoch": 0.508994708994709,
+      "grad_norm": 0.29784589758599245,
+      "learning_rate": 4.615836926695414e-05,
+      "loss": 0.5763,
+      "step": 481
+    },
+    {
+      "epoch": 0.5100529100529101,
+      "grad_norm": 0.28702137011485396,
+      "learning_rate": 4.613876911015288e-05,
+      "loss": 0.5317,
+      "step": 482
+    },
+    {
+      "epoch": 0.5111111111111111,
+      "grad_norm": 1.920530223820589,
+      "learning_rate": 4.6119168953351625e-05,
+      "loss": 0.5852,
+      "step": 483
+    },
+    {
+      "epoch": 0.5121693121693122,
+      "grad_norm": 0.482240940409115,
+      "learning_rate": 4.6099568796550374e-05,
+      "loss": 0.5854,
+      "step": 484
+    },
+    {
+      "epoch": 0.5132275132275133,
+      "grad_norm": 0.39713425911160416,
+      "learning_rate": 4.607996863974912e-05,
+      "loss": 0.5813,
+      "step": 485
+    },
+    {
+      "epoch": 0.5142857142857142,
+      "grad_norm": 0.36739851508814947,
+      "learning_rate": 4.6060368482947866e-05,
+      "loss": 0.6217,
+      "step": 486
+    },
+    {
+      "epoch": 0.5153439153439153,
+      "grad_norm": 0.4261831012022294,
+      "learning_rate": 4.604076832614661e-05,
+      "loss": 0.577,
+      "step": 487
+    },
+    {
+      "epoch": 0.5164021164021164,
+      "grad_norm": 4.16824575872653,
+      "learning_rate": 4.602116816934536e-05,
+      "loss": 0.8018,
+      "step": 488
+    },
+    {
+      "epoch": 0.5174603174603175,
+      "grad_norm": 0.4804692579466171,
+      "learning_rate": 4.6001568012544107e-05,
+      "loss": 0.5141,
+      "step": 489
+    },
+    {
+      "epoch": 0.5185185185185185,
+      "grad_norm": 0.31014473785231145,
+      "learning_rate": 4.598196785574285e-05,
+      "loss": 0.5435,
+      "step": 490
+    },
+    {
+      "epoch": 0.5195767195767196,
+      "grad_norm": 1.2690098499967961,
+      "learning_rate": 4.596236769894159e-05,
+      "loss": 0.5384,
+      "step": 491
+    },
+    {
+      "epoch": 0.5206349206349207,
+      "grad_norm": 0.3480203912635704,
+      "learning_rate": 4.594276754214034e-05,
+      "loss": 0.5376,
+      "step": 492
+    },
+    {
+      "epoch": 0.5216931216931217,
+      "grad_norm": 0.35152029139562335,
+      "learning_rate": 4.592316738533908e-05,
+      "loss": 0.5837,
+      "step": 493
+    },
+    {
+      "epoch": 0.5227513227513227,
+      "grad_norm": 0.33161605035498015,
+      "learning_rate": 4.590356722853783e-05,
+      "loss": 0.5044,
+      "step": 494
+    },
+    {
+      "epoch": 0.5238095238095238,
+      "grad_norm": 0.36352262970279386,
+      "learning_rate": 4.5883967071736575e-05,
+      "loss": 0.5796,
+      "step": 495
+    },
+    {
+      "epoch": 0.5248677248677248,
+      "grad_norm": 0.3237400201076874,
+      "learning_rate": 4.586436691493532e-05,
+      "loss": 0.5769,
+      "step": 496
+    },
+    {
+      "epoch": 0.5259259259259259,
+      "grad_norm": 2.9866473245279956,
+      "learning_rate": 4.584476675813407e-05,
+      "loss": 0.7518,
+      "step": 497
+    },
+    {
+      "epoch": 0.526984126984127,
+      "grad_norm": 0.6522215400738879,
+      "learning_rate": 4.5825166601332816e-05,
+      "loss": 0.5486,
+      "step": 498
+    },
+    {
+      "epoch": 0.5280423280423281,
+      "grad_norm": 0.42283674583260256,
+      "learning_rate": 4.580556644453156e-05,
+      "loss": 0.5667,
+      "step": 499
+    },
+    {
+      "epoch": 0.5291005291005291,
+      "grad_norm": 0.584791999049459,
+      "learning_rate": 4.57859662877303e-05,
+      "loss": 0.5344,
+      "step": 500
+    },
+    {
+      "epoch": 0.5301587301587302,
+      "grad_norm": 0.40242508358942275,
+      "learning_rate": 4.576636613092905e-05,
+      "loss": 0.5313,
+      "step": 501
+    },
+    {
+      "epoch": 0.5312169312169313,
+      "grad_norm": 0.745312348825906,
+      "learning_rate": 4.57467659741278e-05,
+      "loss": 0.5376,
+      "step": 502
+    },
+    {
+      "epoch": 0.5322751322751322,
+      "grad_norm": 0.49986791711058115,
+      "learning_rate": 4.572716581732654e-05,
+      "loss": 0.6753,
+      "step": 503
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 0.42496269751477106,
+      "learning_rate": 4.5707565660525285e-05,
+      "loss": 0.5841,
+      "step": 504
+    },
+    {
+      "epoch": 0.5343915343915344,
+      "grad_norm": 0.4181393322777193,
+      "learning_rate": 4.5687965503724034e-05,
+      "loss": 0.5555,
+      "step": 505
+    },
+    {
+      "epoch": 0.5354497354497354,
+      "grad_norm": 0.3771184252890316,
+      "learning_rate": 4.5668365346922776e-05,
+      "loss": 0.5982,
+      "step": 506
+    },
+    {
+      "epoch": 0.5365079365079365,
+      "grad_norm": 0.47820059656385455,
+      "learning_rate": 4.5648765190121526e-05,
+      "loss": 0.633,
+      "step": 507
+    },
+    {
+      "epoch": 0.5375661375661376,
+      "grad_norm": 0.31449523301220705,
+      "learning_rate": 4.562916503332027e-05,
+      "loss": 0.5999,
+      "step": 508
+    },
+    {
+      "epoch": 0.5386243386243387,
+      "grad_norm": 5.245441857198801,
+      "learning_rate": 4.560956487651901e-05,
+      "loss": 0.6226,
+      "step": 509
+    },
+    {
+      "epoch": 0.5396825396825397,
+      "grad_norm": 0.64012171457468,
+      "learning_rate": 4.558996471971776e-05,
+      "loss": 0.6185,
+      "step": 510
+    },
+    {
+      "epoch": 0.5407407407407407,
+      "grad_norm": 0.354568866388443,
+      "learning_rate": 4.557036456291651e-05,
+      "loss": 0.6311,
+      "step": 511
+    },
+    {
+      "epoch": 0.5417989417989418,
+      "grad_norm": 1.9487919605291462,
+      "learning_rate": 4.555076440611525e-05,
+      "loss": 0.5987,
+      "step": 512
+    },
+    {
+      "epoch": 0.5428571428571428,
+      "grad_norm": 0.6933984738129275,
+      "learning_rate": 4.5531164249313994e-05,
+      "loss": 0.6377,
+      "step": 513
+    },
+    {
+      "epoch": 0.5439153439153439,
+      "grad_norm": 0.4755759149794088,
+      "learning_rate": 4.551156409251274e-05,
+      "loss": 0.5982,
+      "step": 514
+    },
+    {
+      "epoch": 0.544973544973545,
+      "grad_norm": 0.650297043746455,
+      "learning_rate": 4.549196393571149e-05,
+      "loss": 0.5998,
+      "step": 515
+    },
+    {
+      "epoch": 0.546031746031746,
+      "grad_norm": 0.5279264918555835,
+      "learning_rate": 4.5472363778910235e-05,
+      "loss": 0.5869,
+      "step": 516
+    },
+    {
+      "epoch": 0.5470899470899471,
+      "grad_norm": 0.6614038657390391,
+      "learning_rate": 4.545276362210898e-05,
+      "loss": 0.5718,
+      "step": 517
+    },
+    {
+      "epoch": 0.5481481481481482,
+      "grad_norm": 0.4181458689972182,
+      "learning_rate": 4.543316346530773e-05,
+      "loss": 0.5577,
+      "step": 518
+    },
+    {
+      "epoch": 0.5492063492063493,
+      "grad_norm": 4.552268896504232,
+      "learning_rate": 4.541356330850647e-05,
+      "loss": 0.7083,
+      "step": 519
+    },
+    {
+      "epoch": 0.5502645502645502,
+      "grad_norm": 33.061004547207546,
+      "learning_rate": 4.539396315170522e-05,
+      "loss": 0.8391,
+      "step": 520
+    },
+    {
+      "epoch": 0.5513227513227513,
+      "grad_norm": 1.2664930712290081,
+      "learning_rate": 4.537436299490396e-05,
+      "loss": 0.5705,
+      "step": 521
+    },
+    {
+      "epoch": 0.5523809523809524,
+      "grad_norm": 0.42354787993108955,
+      "learning_rate": 4.5354762838102704e-05,
+      "loss": 0.5052,
+      "step": 522
+    },
+    {
+      "epoch": 0.5534391534391534,
+      "grad_norm": 0.8388212731855088,
+      "learning_rate": 4.533516268130145e-05,
+      "loss": 0.5349,
+      "step": 523
+    },
+    {
+      "epoch": 0.5544973544973545,
+      "grad_norm": 0.7949298334624874,
+      "learning_rate": 4.53155625245002e-05,
+      "loss": 0.5497,
+      "step": 524
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 0.44871889970887846,
+      "learning_rate": 4.5295962367698945e-05,
+      "loss": 0.5978,
+      "step": 525
+    },
+    {
+      "epoch": 0.5566137566137566,
+      "grad_norm": 0.7676280541833576,
+      "learning_rate": 4.527636221089769e-05,
+      "loss": 0.551,
+      "step": 526
+    },
+    {
+      "epoch": 0.5576719576719577,
+      "grad_norm": 0.7002659965437515,
+      "learning_rate": 4.5256762054096436e-05,
+      "loss": 0.5808,
+      "step": 527
+    },
+    {
+      "epoch": 0.5587301587301587,
+      "grad_norm": 1.2074663454141952,
+      "learning_rate": 4.5237161897295186e-05,
+      "loss": 0.6605,
+      "step": 528
+    },
+    {
+      "epoch": 0.5597883597883598,
+      "grad_norm": 0.712483552007316,
+      "learning_rate": 4.521756174049392e-05,
+      "loss": 0.618,
+      "step": 529
+    },
+    {
+      "epoch": 0.5608465608465608,
+      "grad_norm": 0.331920911665764,
+      "learning_rate": 4.519796158369267e-05,
+      "loss": 0.5375,
+      "step": 530
+    },
+    {
+      "epoch": 0.5619047619047619,
+      "grad_norm": 0.598217747836947,
+      "learning_rate": 4.517836142689142e-05,
+      "loss": 0.5801,
+      "step": 531
+    },
+    {
+      "epoch": 0.562962962962963,
+      "grad_norm": 0.39707437203429746,
+      "learning_rate": 4.515876127009016e-05,
+      "loss": 0.5298,
+      "step": 532
+    },
+    {
+      "epoch": 0.564021164021164,
+      "grad_norm": 0.33682693857579143,
+      "learning_rate": 4.5139161113288905e-05,
+      "loss": 0.5137,
+      "step": 533
+    },
+    {
+      "epoch": 0.5650793650793651,
+      "grad_norm": 0.48164899152667084,
+      "learning_rate": 4.5119560956487654e-05,
+      "loss": 0.5781,
+      "step": 534
+    },
+    {
+      "epoch": 0.5661375661375662,
+      "grad_norm": 0.3219468021100436,
+      "learning_rate": 4.5099960799686397e-05,
+      "loss": 0.4857,
+      "step": 535
+    },
+    {
+      "epoch": 0.5671957671957671,
+      "grad_norm": 0.3320548497953154,
+      "learning_rate": 4.5080360642885146e-05,
+      "loss": 0.5725,
+      "step": 536
+    },
+    {
+      "epoch": 0.5682539682539682,
+      "grad_norm": 0.4483913400362615,
+      "learning_rate": 4.5060760486083895e-05,
+      "loss": 0.4854,
+      "step": 537
+    },
+    {
+      "epoch": 0.5693121693121693,
+      "grad_norm": 0.36430948857681866,
+      "learning_rate": 4.504116032928264e-05,
+      "loss": 0.662,
+      "step": 538
+    },
+    {
+      "epoch": 0.5703703703703704,
+      "grad_norm": 0.342825639833936,
+      "learning_rate": 4.502156017248138e-05,
+      "loss": 0.5921,
+      "step": 539
+    },
+    {
+      "epoch": 0.5714285714285714,
+      "grad_norm": 0.3734526413757713,
+      "learning_rate": 4.500196001568013e-05,
+      "loss": 0.5297,
+      "step": 540
+    },
+    {
+      "epoch": 0.5724867724867725,
+      "grad_norm": 0.3388581653822819,
+      "learning_rate": 4.498235985887888e-05,
+      "loss": 0.6147,
+      "step": 541
+    },
+    {
+      "epoch": 0.5735449735449736,
+      "grad_norm": 0.3205452586395781,
+      "learning_rate": 4.4962759702077614e-05,
+      "loss": 0.5175,
+      "step": 542
+    },
+    {
+      "epoch": 0.5746031746031746,
+      "grad_norm": 0.34747287406127814,
+      "learning_rate": 4.4943159545276363e-05,
+      "loss": 0.588,
+      "step": 543
+    },
+    {
+      "epoch": 0.5756613756613757,
+      "grad_norm": 0.3293870457786716,
+      "learning_rate": 4.492355938847511e-05,
+      "loss": 0.6207,
+      "step": 544
+    },
+    {
+      "epoch": 0.5767195767195767,
+      "grad_norm": 1.5911037019280505,
+      "learning_rate": 4.4903959231673855e-05,
+      "loss": 0.6216,
+      "step": 545
+    },
+    {
+      "epoch": 0.5777777777777777,
+      "grad_norm": 0.2970675826508983,
+      "learning_rate": 4.48843590748726e-05,
+      "loss": 0.543,
+      "step": 546
+    },
+    {
+      "epoch": 0.5788359788359788,
+      "grad_norm": 0.3347900655603031,
+      "learning_rate": 4.486475891807135e-05,
+      "loss": 0.5524,
+      "step": 547
+    },
+    {
+      "epoch": 0.5798941798941799,
+      "grad_norm": 0.3297134029351915,
+      "learning_rate": 4.484515876127009e-05,
+      "loss": 0.5289,
+      "step": 548
+    },
+    {
+      "epoch": 0.580952380952381,
+      "grad_norm": 0.33203980488440166,
+      "learning_rate": 4.482555860446884e-05,
+      "loss": 0.5628,
+      "step": 549
+    },
+    {
+      "epoch": 0.582010582010582,
+      "grad_norm": 0.3675901822117606,
+      "learning_rate": 4.480595844766758e-05,
+      "loss": 0.6431,
+      "step": 550
+    },
+    {
+      "epoch": 0.5830687830687831,
+      "grad_norm": 8.35298256158783,
+      "learning_rate": 4.478635829086633e-05,
+      "loss": 0.6803,
+      "step": 551
+    },
+    {
+      "epoch": 0.5841269841269842,
+      "grad_norm": 0.38944467670316385,
+      "learning_rate": 4.476675813406507e-05,
+      "loss": 0.6053,
+      "step": 552
+    },
+    {
+      "epoch": 0.5851851851851851,
+      "grad_norm": 1.8390547635610588,
+      "learning_rate": 4.474715797726382e-05,
+      "loss": 0.5739,
+      "step": 553
+    },
+    {
+      "epoch": 0.5862433862433862,
+      "grad_norm": 0.3840415785962572,
+      "learning_rate": 4.4727557820462565e-05,
+      "loss": 0.6275,
+      "step": 554
+    },
+    {
+      "epoch": 0.5873015873015873,
+      "grad_norm": 0.29418089397003,
+      "learning_rate": 4.470795766366131e-05,
+      "loss": 0.535,
+      "step": 555
+    },
+    {
+      "epoch": 0.5883597883597883,
+      "grad_norm": 0.3736504054747309,
+      "learning_rate": 4.4688357506860056e-05,
+      "loss": 0.5943,
+      "step": 556
+    },
+    {
+      "epoch": 0.5894179894179894,
+      "grad_norm": 0.33285236147227376,
+      "learning_rate": 4.4668757350058806e-05,
+      "loss": 0.5586,
+      "step": 557
+    },
+    {
+      "epoch": 0.5904761904761905,
+      "grad_norm": 0.2903484444203814,
+      "learning_rate": 4.464915719325755e-05,
+      "loss": 0.5365,
+      "step": 558
+    },
+    {
+      "epoch": 0.5915343915343916,
+      "grad_norm": 5.46404606246135,
+      "learning_rate": 4.462955703645629e-05,
+      "loss": 0.6439,
+      "step": 559
+    },
+    {
+      "epoch": 0.5925925925925926,
+      "grad_norm": 0.4599968468845947,
+      "learning_rate": 4.460995687965504e-05,
+      "loss": 0.6976,
+      "step": 560
+    },
+    {
+      "epoch": 0.5936507936507937,
+      "grad_norm": 0.2982395341967786,
+      "learning_rate": 4.459035672285379e-05,
+      "loss": 0.5041,
+      "step": 561
+    },
+    {
+      "epoch": 0.5947089947089947,
+      "grad_norm": 0.30282795685306324,
+      "learning_rate": 4.457075656605253e-05,
+      "loss": 0.5909,
+      "step": 562
+    },
+    {
+      "epoch": 0.5957671957671957,
+      "grad_norm": 0.39972890415638074,
+      "learning_rate": 4.4551156409251274e-05,
+      "loss": 0.6235,
+      "step": 563
+    },
+    {
+      "epoch": 0.5968253968253968,
+      "grad_norm": 0.2873678600174425,
+      "learning_rate": 4.4531556252450023e-05,
+      "loss": 0.5421,
+      "step": 564
+    },
+    {
+      "epoch": 0.5978835978835979,
+      "grad_norm": 0.3559983769537934,
+      "learning_rate": 4.4511956095648766e-05,
+      "loss": 0.5561,
+      "step": 565
+    },
+    {
+      "epoch": 0.5989417989417989,
+      "grad_norm": 0.3147975278256888,
+      "learning_rate": 4.4492355938847515e-05,
+      "loss": 0.4711,
+      "step": 566
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2607641173456458,
+      "learning_rate": 4.447275578204626e-05,
+      "loss": 0.5823,
+      "step": 567
+    },
+    {
+      "epoch": 0.6010582010582011,
+      "grad_norm": 0.2926523078468275,
+      "learning_rate": 4.4453155625245e-05,
+      "loss": 0.5406,
+      "step": 568
+    },
+    {
+      "epoch": 0.6021164021164022,
+      "grad_norm": 0.3452463175268165,
+      "learning_rate": 4.443355546844375e-05,
+      "loss": 0.5894,
+      "step": 569
+    },
+    {
+      "epoch": 0.6031746031746031,
+      "grad_norm": 0.24630431433470665,
+      "learning_rate": 4.44139553116425e-05,
+      "loss": 0.517,
+      "step": 570
+    },
+    {
+      "epoch": 0.6042328042328042,
+      "grad_norm": 0.3401893647581997,
+      "learning_rate": 4.439435515484124e-05,
+      "loss": 0.5647,
+      "step": 571
+    },
+    {
+      "epoch": 0.6052910052910053,
+      "grad_norm": 0.3208322781574765,
+      "learning_rate": 4.4374754998039984e-05,
+      "loss": 0.5116,
+      "step": 572
+    },
+    {
+      "epoch": 0.6063492063492063,
+      "grad_norm": 0.28383141803835954,
+      "learning_rate": 4.435515484123873e-05,
+      "loss": 0.5441,
+      "step": 573
+    },
+    {
+      "epoch": 0.6074074074074074,
+      "grad_norm": 0.34262017652008814,
+      "learning_rate": 4.433555468443748e-05,
+      "loss": 0.5972,
+      "step": 574
+    },
+    {
+      "epoch": 0.6084656084656085,
+      "grad_norm": 0.31557046384544746,
+      "learning_rate": 4.4315954527636225e-05,
+      "loss": 0.6478,
+      "step": 575
+    },
+    {
+      "epoch": 0.6095238095238096,
+      "grad_norm": 0.31340622501168497,
+      "learning_rate": 4.429635437083497e-05,
+      "loss": 0.561,
+      "step": 576
+    },
+    {
+      "epoch": 0.6105820105820106,
+      "grad_norm": 0.27478250517189123,
+      "learning_rate": 4.4276754214033716e-05,
+      "loss": 0.6372,
+      "step": 577
+    },
+    {
+      "epoch": 0.6116402116402117,
+      "grad_norm": 0.3096835013791637,
+      "learning_rate": 4.425715405723246e-05,
+      "loss": 0.5873,
+      "step": 578
+    },
+    {
+      "epoch": 0.6126984126984127,
+      "grad_norm": 0.29052274812803913,
+      "learning_rate": 4.423755390043121e-05,
+      "loss": 0.6673,
+      "step": 579
+    },
+    {
+      "epoch": 0.6137566137566137,
+      "grad_norm": 0.2726481497564029,
+      "learning_rate": 4.421795374362995e-05,
+      "loss": 0.5688,
+      "step": 580
+    },
+    {
+      "epoch": 0.6148148148148148,
+      "grad_norm": 0.30868513597771813,
+      "learning_rate": 4.419835358682869e-05,
+      "loss": 0.5413,
+      "step": 581
+    },
+    {
+      "epoch": 0.6158730158730159,
+      "grad_norm": 0.2931107306494861,
+      "learning_rate": 4.417875343002744e-05,
+      "loss": 0.6191,
+      "step": 582
+    },
+    {
+      "epoch": 0.6169312169312169,
+      "grad_norm": 0.27359852790549766,
+      "learning_rate": 4.415915327322619e-05,
+      "loss": 0.5976,
+      "step": 583
+    },
+    {
+      "epoch": 0.617989417989418,
+      "grad_norm": 0.300451446217763,
+      "learning_rate": 4.4139553116424934e-05,
+      "loss": 0.5958,
+      "step": 584
+    },
+    {
+      "epoch": 0.6190476190476191,
+      "grad_norm": 0.2739263908226725,
+      "learning_rate": 4.411995295962368e-05,
+      "loss": 0.4819,
+      "step": 585
+    },
+    {
+      "epoch": 0.6201058201058202,
+      "grad_norm": 0.3629370235736209,
+      "learning_rate": 4.4100352802822426e-05,
+      "loss": 0.5969,
+      "step": 586
+    },
+    {
+      "epoch": 0.6211640211640211,
+      "grad_norm": 0.2546483909460872,
+      "learning_rate": 4.4080752646021175e-05,
+      "loss": 0.5637,
+      "step": 587
+    },
+    {
+      "epoch": 0.6222222222222222,
+      "grad_norm": 0.2900274901311552,
+      "learning_rate": 4.406115248921992e-05,
+      "loss": 0.5981,
+      "step": 588
+    },
+    {
+      "epoch": 0.6232804232804233,
+      "grad_norm": 0.30997705577212653,
+      "learning_rate": 4.404155233241866e-05,
+      "loss": 0.5763,
+      "step": 589
+    },
+    {
+      "epoch": 0.6243386243386243,
+      "grad_norm": 0.27235932084700465,
+      "learning_rate": 4.402195217561741e-05,
+      "loss": 0.6027,
+      "step": 590
+    },
+    {
+      "epoch": 0.6253968253968254,
+      "grad_norm": 0.27424029005807476,
+      "learning_rate": 4.400235201881615e-05,
+      "loss": 0.528,
+      "step": 591
+    },
+    {
+      "epoch": 0.6264550264550265,
+      "grad_norm": 0.8024320046005814,
+      "learning_rate": 4.39827518620149e-05,
+      "loss": 0.5073,
+      "step": 592
+    },
+    {
+      "epoch": 0.6275132275132275,
+      "grad_norm": 0.25330323543588595,
+      "learning_rate": 4.3963151705213644e-05,
+      "loss": 0.5903,
+      "step": 593
+    },
+    {
+      "epoch": 0.6285714285714286,
+      "grad_norm": 0.2786349566048451,
+      "learning_rate": 4.3943551548412386e-05,
+      "loss": 0.5646,
+      "step": 594
+    },
+    {
+      "epoch": 0.6296296296296297,
+      "grad_norm": 0.27111077124213456,
+      "learning_rate": 4.3923951391611135e-05,
+      "loss": 0.5166,
+      "step": 595
+    },
+    {
+      "epoch": 0.6306878306878307,
+      "grad_norm": 0.2537581235144252,
+      "learning_rate": 4.3904351234809885e-05,
+      "loss": 0.5475,
+      "step": 596
+    },
+    {
+      "epoch": 0.6317460317460317,
+      "grad_norm": 0.2940667859193826,
+      "learning_rate": 4.388475107800862e-05,
+      "loss": 0.5437,
+      "step": 597
+    },
+    {
+      "epoch": 0.6328042328042328,
+      "grad_norm": 0.26691327126042796,
+      "learning_rate": 4.386515092120737e-05,
+      "loss": 0.5508,
+      "step": 598
+    },
+    {
+      "epoch": 0.6338624338624339,
+      "grad_norm": 3.125987633190478,
+      "learning_rate": 4.384555076440612e-05,
+      "loss": 0.5517,
+      "step": 599
+    },
+    {
+      "epoch": 0.6349206349206349,
+      "grad_norm": 0.3051459051560689,
+      "learning_rate": 4.382595060760487e-05,
+      "loss": 0.5032,
+      "step": 600
+    },
+    {
+      "epoch": 0.635978835978836,
+      "grad_norm": 0.2820838021225074,
+      "learning_rate": 4.3806350450803604e-05,
+      "loss": 0.5407,
+      "step": 601
+    },
+    {
+      "epoch": 0.6370370370370371,
+      "grad_norm": 1.3393856684804768,
+      "learning_rate": 4.378675029400235e-05,
+      "loss": 0.6108,
+      "step": 602
+    },
+    {
+      "epoch": 0.638095238095238,
+      "grad_norm": 0.3228594095981327,
+      "learning_rate": 4.37671501372011e-05,
+      "loss": 0.5025,
+      "step": 603
+    },
+    {
+      "epoch": 0.6391534391534391,
+      "grad_norm": 0.3351209221453693,
+      "learning_rate": 4.3747549980399845e-05,
+      "loss": 0.5659,
+      "step": 604
+    },
+    {
+      "epoch": 0.6402116402116402,
+      "grad_norm": 0.31230131092186864,
+      "learning_rate": 4.372794982359859e-05,
+      "loss": 0.6007,
+      "step": 605
+    },
+    {
+      "epoch": 0.6412698412698413,
+      "grad_norm": 0.2828867288582573,
+      "learning_rate": 4.370834966679734e-05,
+      "loss": 0.527,
+      "step": 606
+    },
+    {
+      "epoch": 0.6423280423280423,
+      "grad_norm": 0.34529541782733586,
+      "learning_rate": 4.368874950999608e-05,
+      "loss": 0.5486,
+      "step": 607
+    },
+    {
+      "epoch": 0.6433862433862434,
+      "grad_norm": 0.2951068794541986,
+      "learning_rate": 4.366914935319483e-05,
+      "loss": 0.5267,
+      "step": 608
+    },
+    {
+      "epoch": 0.6444444444444445,
+      "grad_norm": 0.40815008679176695,
+      "learning_rate": 4.364954919639358e-05,
+      "loss": 0.5177,
+      "step": 609
+    },
+    {
+      "epoch": 0.6455026455026455,
+      "grad_norm": 0.32045696467121443,
+      "learning_rate": 4.362994903959231e-05,
+      "loss": 0.5586,
+      "step": 610
+    },
+    {
+      "epoch": 0.6465608465608466,
+      "grad_norm": 0.32938732342590915,
+      "learning_rate": 4.361034888279106e-05,
+      "loss": 0.5711,
+      "step": 611
+    },
+    {
+      "epoch": 0.6476190476190476,
+      "grad_norm": 0.2914985462165065,
+      "learning_rate": 4.359074872598981e-05,
+      "loss": 0.6356,
+      "step": 612
+    },
+    {
+      "epoch": 0.6486772486772486,
+      "grad_norm": 0.31957339334532875,
+      "learning_rate": 4.357114856918856e-05,
+      "loss": 0.5775,
+      "step": 613
+    },
+    {
+      "epoch": 0.6497354497354497,
+      "grad_norm": 0.3380048318234734,
+      "learning_rate": 4.35515484123873e-05,
+      "loss": 0.5668,
+      "step": 614
+    },
+    {
+      "epoch": 0.6507936507936508,
+      "grad_norm": 0.2677052222743914,
+      "learning_rate": 4.3531948255586046e-05,
+      "loss": 0.5867,
+      "step": 615
+    },
+    {
+      "epoch": 0.6518518518518519,
+      "grad_norm": 0.2714050773104208,
+      "learning_rate": 4.3512348098784795e-05,
+      "loss": 0.569,
+      "step": 616
+    },
+    {
+      "epoch": 0.6529100529100529,
+      "grad_norm": 0.322108580723527,
+      "learning_rate": 4.349274794198354e-05,
+      "loss": 0.5379,
+      "step": 617
+    },
+    {
+      "epoch": 0.653968253968254,
+      "grad_norm": 0.2569245838676028,
+      "learning_rate": 4.347314778518228e-05,
+      "loss": 0.5118,
+      "step": 618
+    },
+    {
+      "epoch": 0.6550264550264551,
+      "grad_norm": 0.2530435498061503,
+      "learning_rate": 4.345354762838103e-05,
+      "loss": 0.5139,
+      "step": 619
+    },
+    {
+      "epoch": 0.656084656084656,
+      "grad_norm": 0.29585162017581107,
+      "learning_rate": 4.343394747157977e-05,
+      "loss": 0.4954,
+      "step": 620
+    },
+    {
+      "epoch": 0.6571428571428571,
+      "grad_norm": 0.2924074793370025,
+      "learning_rate": 4.341434731477852e-05,
+      "loss": 0.6042,
+      "step": 621
+    },
+    {
+      "epoch": 0.6582010582010582,
+      "grad_norm": 0.2855302831026707,
+      "learning_rate": 4.3394747157977264e-05,
+      "loss": 0.514,
+      "step": 622
+    },
+    {
+      "epoch": 0.6592592592592592,
+      "grad_norm": 0.7366169291367959,
+      "learning_rate": 4.337514700117601e-05,
+      "loss": 0.4895,
+      "step": 623
+    },
+    {
+      "epoch": 0.6603174603174603,
+      "grad_norm": 0.27063474014038635,
+      "learning_rate": 4.3355546844374756e-05,
+      "loss": 0.5592,
+      "step": 624
+    },
+    {
+      "epoch": 0.6613756613756614,
+      "grad_norm": 0.2539651508587892,
+      "learning_rate": 4.3335946687573505e-05,
+      "loss": 0.5301,
+      "step": 625
+    },
+    {
+      "epoch": 0.6624338624338625,
+      "grad_norm": 0.2458381741039525,
+      "learning_rate": 4.331634653077225e-05,
+      "loss": 0.5762,
+      "step": 626
+    },
+    {
+      "epoch": 0.6634920634920635,
+      "grad_norm": 0.25275919533490976,
+      "learning_rate": 4.329674637397099e-05,
+      "loss": 0.5422,
+      "step": 627
+    },
+    {
+      "epoch": 0.6645502645502646,
+      "grad_norm": 0.2626626352002252,
+      "learning_rate": 4.327714621716974e-05,
+      "loss": 0.5175,
+      "step": 628
+    },
+    {
+      "epoch": 0.6656084656084656,
+      "grad_norm": 0.2646967802621259,
+      "learning_rate": 4.325754606036849e-05,
+      "loss": 0.4619,
+      "step": 629
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.24368769535748608,
+      "learning_rate": 4.323794590356723e-05,
+      "loss": 0.5964,
+      "step": 630
+    },
+    {
+      "epoch": 0.6677248677248677,
+      "grad_norm": 0.2959718652590851,
+      "learning_rate": 4.321834574676597e-05,
+      "loss": 0.5691,
+      "step": 631
+    },
+    {
+      "epoch": 0.6687830687830688,
+      "grad_norm": 0.2871293046686386,
+      "learning_rate": 4.319874558996472e-05,
+      "loss": 0.5696,
+      "step": 632
+    },
+    {
+      "epoch": 0.6698412698412698,
+      "grad_norm": 0.2905598264986975,
+      "learning_rate": 4.3179145433163465e-05,
+      "loss": 0.629,
+      "step": 633
+    },
+    {
+      "epoch": 0.6708994708994709,
+      "grad_norm": 0.2528097775183016,
+      "learning_rate": 4.3159545276362214e-05,
+      "loss": 0.4849,
+      "step": 634
+    },
+    {
+      "epoch": 0.671957671957672,
+      "grad_norm": 0.33621938773228494,
+      "learning_rate": 4.313994511956096e-05,
+      "loss": 0.5879,
+      "step": 635
+    },
+    {
+      "epoch": 0.6730158730158731,
+      "grad_norm": 0.2394966771033253,
+      "learning_rate": 4.3120344962759706e-05,
+      "loss": 0.439,
+      "step": 636
+    },
+    {
+      "epoch": 0.674074074074074,
+      "grad_norm": 0.3150605848513219,
+      "learning_rate": 4.310074480595845e-05,
+      "loss": 0.5989,
+      "step": 637
+    },
+    {
+      "epoch": 0.6751322751322751,
+      "grad_norm": 0.2981707132882546,
+      "learning_rate": 4.30811446491572e-05,
+      "loss": 0.6225,
+      "step": 638
+    },
+    {
+      "epoch": 0.6761904761904762,
+      "grad_norm": 0.30492112978736213,
+      "learning_rate": 4.306154449235594e-05,
+      "loss": 0.5404,
+      "step": 639
+    },
+    {
+      "epoch": 0.6772486772486772,
+      "grad_norm": 0.2814127526303401,
+      "learning_rate": 4.304194433555468e-05,
+      "loss": 0.5233,
+      "step": 640
+    },
+    {
+      "epoch": 0.6783068783068783,
+      "grad_norm": 0.30173716166671577,
+      "learning_rate": 4.302234417875343e-05,
+      "loss": 0.6427,
+      "step": 641
+    },
+    {
+      "epoch": 0.6793650793650794,
+      "grad_norm": 0.25137410162980395,
+      "learning_rate": 4.300274402195218e-05,
+      "loss": 0.5607,
+      "step": 642
+    },
+    {
+      "epoch": 0.6804232804232804,
+      "grad_norm": 0.30267125046888244,
+      "learning_rate": 4.2983143865150924e-05,
+      "loss": 0.6308,
+      "step": 643
+    },
+    {
+      "epoch": 0.6814814814814815,
+      "grad_norm": 0.2570686438542522,
+      "learning_rate": 4.2963543708349666e-05,
+      "loss": 0.5708,
+      "step": 644
+    },
+    {
+      "epoch": 0.6825396825396826,
+      "grad_norm": 0.47196520519510343,
+      "learning_rate": 4.2943943551548416e-05,
+      "loss": 0.4772,
+      "step": 645
+    },
+    {
+      "epoch": 0.6835978835978836,
+      "grad_norm": 0.2733048973472466,
+      "learning_rate": 4.292434339474716e-05,
+      "loss": 0.5569,
+      "step": 646
+    },
+    {
+      "epoch": 0.6846560846560846,
+      "grad_norm": 0.29886485226274184,
+      "learning_rate": 4.290474323794591e-05,
+      "loss": 0.5785,
+      "step": 647
+    },
+    {
+      "epoch": 0.6857142857142857,
+      "grad_norm": 0.2403402937990743,
+      "learning_rate": 4.288514308114465e-05,
+      "loss": 0.5595,
+      "step": 648
+    },
+    {
+      "epoch": 0.6867724867724868,
+      "grad_norm": 0.2794008545700776,
+      "learning_rate": 4.28655429243434e-05,
+      "loss": 0.5396,
+      "step": 649
+    },
+    {
+      "epoch": 0.6878306878306878,
+      "grad_norm": 0.3041316437038585,
+      "learning_rate": 4.284594276754214e-05,
+      "loss": 0.5601,
+      "step": 650
+    },
+    {
+      "epoch": 0.6888888888888889,
+      "grad_norm": 0.24855481366370547,
+      "learning_rate": 4.282634261074089e-05,
+      "loss": 0.5405,
+      "step": 651
+    },
+    {
+      "epoch": 0.68994708994709,
+      "grad_norm": 0.3040097458165702,
+      "learning_rate": 4.280674245393963e-05,
+      "loss": 0.64,
+      "step": 652
+    },
+    {
+      "epoch": 0.691005291005291,
+      "grad_norm": 0.3069869635860471,
+      "learning_rate": 4.2787142297138376e-05,
+      "loss": 0.4583,
+      "step": 653
+    },
+    {
+      "epoch": 0.692063492063492,
+      "grad_norm": 0.2936008162342032,
+      "learning_rate": 4.2767542140337125e-05,
+      "loss": 0.5027,
+      "step": 654
+    },
+    {
+      "epoch": 0.6931216931216931,
+      "grad_norm": 0.25492661711077735,
+      "learning_rate": 4.2747941983535874e-05,
+      "loss": 0.5225,
+      "step": 655
+    },
+    {
+      "epoch": 0.6941798941798942,
+      "grad_norm": 2.5144220055291275,
+      "learning_rate": 4.272834182673462e-05,
+      "loss": 0.6846,
+      "step": 656
+    },
+    {
+      "epoch": 0.6952380952380952,
+      "grad_norm": 0.28990557153471347,
+      "learning_rate": 4.270874166993336e-05,
+      "loss": 0.5752,
+      "step": 657
+    },
+    {
+      "epoch": 0.6962962962962963,
+      "grad_norm": 0.26542534568407355,
+      "learning_rate": 4.268914151313211e-05,
+      "loss": 0.4701,
+      "step": 658
+    },
+    {
+      "epoch": 0.6973544973544974,
+      "grad_norm": 0.3193855535773065,
+      "learning_rate": 4.266954135633085e-05,
+      "loss": 0.4586,
+      "step": 659
+    },
+    {
+      "epoch": 0.6984126984126984,
+      "grad_norm": 0.27477559170920174,
+      "learning_rate": 4.26499411995296e-05,
+      "loss": 0.5044,
+      "step": 660
+    },
+    {
+      "epoch": 0.6994708994708995,
+      "grad_norm": 0.2691283100685374,
+      "learning_rate": 4.263034104272834e-05,
+      "loss": 0.6087,
+      "step": 661
+    },
+    {
+      "epoch": 0.7005291005291006,
+      "grad_norm": 0.3830270820991914,
+      "learning_rate": 4.261074088592709e-05,
+      "loss": 0.5775,
+      "step": 662
+    },
+    {
+      "epoch": 0.7015873015873015,
+      "grad_norm": 0.2752894226111952,
+      "learning_rate": 4.2591140729125835e-05,
+      "loss": 0.5079,
+      "step": 663
+    },
+    {
+      "epoch": 0.7026455026455026,
+      "grad_norm": 0.28397687271151834,
+      "learning_rate": 4.2571540572324584e-05,
+      "loss": 0.5804,
+      "step": 664
+    },
+    {
+      "epoch": 0.7037037037037037,
+      "grad_norm": 3.367695352363058,
+      "learning_rate": 4.2551940415523326e-05,
+      "loss": 0.7196,
+      "step": 665
+    },
+    {
+      "epoch": 0.7047619047619048,
+      "grad_norm": 0.38941063877467563,
+      "learning_rate": 4.253234025872207e-05,
+      "loss": 0.5676,
+      "step": 666
+    },
+    {
+      "epoch": 0.7058201058201058,
+      "grad_norm": 0.299525285575676,
+      "learning_rate": 4.251274010192082e-05,
+      "loss": 0.5809,
+      "step": 667
+    },
+    {
+      "epoch": 0.7068783068783069,
+      "grad_norm": 0.24410082905969419,
+      "learning_rate": 4.249313994511957e-05,
+      "loss": 0.5784,
+      "step": 668
+    },
+    {
+      "epoch": 0.707936507936508,
+      "grad_norm": 0.817819426083752,
+      "learning_rate": 4.24735397883183e-05,
+      "loss": 0.5015,
+      "step": 669
+    },
+    {
+      "epoch": 0.708994708994709,
+      "grad_norm": 0.35760232367472883,
+      "learning_rate": 4.245393963151705e-05,
+      "loss": 0.6156,
+      "step": 670
+    },
+    {
+      "epoch": 0.71005291005291,
+      "grad_norm": 0.2827406205547459,
+      "learning_rate": 4.24343394747158e-05,
+      "loss": 0.5757,
+      "step": 671
+    },
+    {
+      "epoch": 0.7111111111111111,
+      "grad_norm": 0.28008993326048687,
+      "learning_rate": 4.241473931791455e-05,
+      "loss": 0.5112,
+      "step": 672
+    },
+    {
+      "epoch": 0.7121693121693121,
+      "grad_norm": 0.2924001411666532,
+      "learning_rate": 4.2395139161113287e-05,
+      "loss": 0.5523,
+      "step": 673
+    },
+    {
+      "epoch": 0.7132275132275132,
+      "grad_norm": 0.27305443841057714,
+      "learning_rate": 4.2375539004312036e-05,
+      "loss": 0.5126,
+      "step": 674
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 0.2736723087303262,
+      "learning_rate": 4.2355938847510785e-05,
+      "loss": 0.5388,
+      "step": 675
+    },
+    {
+      "epoch": 0.7153439153439154,
+      "grad_norm": 0.27719351183152063,
+      "learning_rate": 4.233633869070953e-05,
+      "loss": 0.5331,
+      "step": 676
+    },
+    {
+      "epoch": 0.7164021164021164,
+      "grad_norm": 0.29024073675556417,
+      "learning_rate": 4.231673853390827e-05,
+      "loss": 0.5324,
+      "step": 677
+    },
+    {
+      "epoch": 0.7174603174603175,
+      "grad_norm": 0.28459105874291735,
+      "learning_rate": 4.229713837710702e-05,
+      "loss": 0.5723,
+      "step": 678
+    },
+    {
+      "epoch": 0.7185185185185186,
+      "grad_norm": 0.27008204854468826,
+      "learning_rate": 4.227753822030576e-05,
+      "loss": 0.5183,
+      "step": 679
+    },
+    {
+      "epoch": 0.7195767195767195,
+      "grad_norm": 0.32531200458934517,
+      "learning_rate": 4.225793806350451e-05,
+      "loss": 0.5417,
+      "step": 680
+    },
+    {
+      "epoch": 0.7206349206349206,
+      "grad_norm": 0.2827590448071415,
+      "learning_rate": 4.223833790670326e-05,
+      "loss": 0.5688,
+      "step": 681
+    },
+    {
+      "epoch": 0.7216931216931217,
+      "grad_norm": 0.32843041348454105,
+      "learning_rate": 4.2218737749901996e-05,
+      "loss": 0.4738,
+      "step": 682
+    },
+    {
+      "epoch": 0.7227513227513227,
+      "grad_norm": 0.29203603535869277,
+      "learning_rate": 4.2199137593100745e-05,
+      "loss": 0.571,
+      "step": 683
+    },
+    {
+      "epoch": 0.7238095238095238,
+      "grad_norm": 0.26235624444883343,
+      "learning_rate": 4.2179537436299495e-05,
+      "loss": 0.5339,
+      "step": 684
+    },
+    {
+      "epoch": 0.7248677248677249,
+      "grad_norm": 0.2743770991741004,
+      "learning_rate": 4.2159937279498244e-05,
+      "loss": 0.5938,
+      "step": 685
+    },
+    {
+      "epoch": 0.725925925925926,
+      "grad_norm": 0.28706543540832447,
+      "learning_rate": 4.214033712269698e-05,
+      "loss": 0.5088,
+      "step": 686
+    },
+    {
+      "epoch": 0.726984126984127,
+      "grad_norm": 0.2831332896571073,
+      "learning_rate": 4.212073696589573e-05,
+      "loss": 0.606,
+      "step": 687
+    },
+    {
+      "epoch": 0.728042328042328,
+      "grad_norm": 0.262186187511753,
+      "learning_rate": 4.210113680909448e-05,
+      "loss": 0.5341,
+      "step": 688
+    },
+    {
+      "epoch": 0.7291005291005291,
+      "grad_norm": 0.3096610015904226,
+      "learning_rate": 4.208153665229322e-05,
+      "loss": 0.5336,
+      "step": 689
+    },
+    {
+      "epoch": 0.7301587301587301,
+      "grad_norm": 0.2733332314767257,
+      "learning_rate": 4.206193649549196e-05,
+      "loss": 0.5145,
+      "step": 690
+    },
+    {
+      "epoch": 0.7312169312169312,
+      "grad_norm": 0.2438987308317826,
+      "learning_rate": 4.204233633869071e-05,
+      "loss": 0.5574,
+      "step": 691
+    },
+    {
+      "epoch": 0.7322751322751323,
+      "grad_norm": 0.29975311338331406,
+      "learning_rate": 4.2022736181889455e-05,
+      "loss": 0.5676,
+      "step": 692
+    },
+    {
+      "epoch": 0.7333333333333333,
+      "grad_norm": 0.3159755996646559,
+      "learning_rate": 4.2003136025088204e-05,
+      "loss": 0.6154,
+      "step": 693
+    },
+    {
+      "epoch": 0.7343915343915344,
+      "grad_norm": 0.24416951903546183,
+      "learning_rate": 4.1983535868286947e-05,
+      "loss": 0.5248,
+      "step": 694
+    },
+    {
+      "epoch": 0.7354497354497355,
+      "grad_norm": 0.24947362261992206,
+      "learning_rate": 4.196393571148569e-05,
+      "loss": 0.5121,
+      "step": 695
+    },
+    {
+      "epoch": 0.7365079365079366,
+      "grad_norm": 0.30381658421133123,
+      "learning_rate": 4.194433555468444e-05,
+      "loss": 0.583,
+      "step": 696
+    },
+    {
+      "epoch": 0.7375661375661375,
+      "grad_norm": 0.23096151371713622,
+      "learning_rate": 4.192473539788319e-05,
+      "loss": 0.5245,
+      "step": 697
+    },
+    {
+      "epoch": 0.7386243386243386,
+      "grad_norm": 0.3162192673911441,
+      "learning_rate": 4.190513524108193e-05,
+      "loss": 0.5468,
+      "step": 698
+    },
+    {
+      "epoch": 0.7396825396825397,
+      "grad_norm": 0.25182522028970394,
+      "learning_rate": 4.188553508428067e-05,
+      "loss": 0.597,
+      "step": 699
+    },
+    {
+      "epoch": 0.7407407407407407,
+      "grad_norm": 0.2684334525549805,
+      "learning_rate": 4.186593492747942e-05,
+      "loss": 0.576,
+      "step": 700
+    },
+    {
+      "epoch": 0.7417989417989418,
+      "grad_norm": 1.523561308125903,
+      "learning_rate": 4.184633477067817e-05,
+      "loss": 0.5329,
+      "step": 701
+    },
+    {
+      "epoch": 0.7428571428571429,
+      "grad_norm": 0.28565441595265484,
+      "learning_rate": 4.1826734613876914e-05,
+      "loss": 0.5196,
+      "step": 702
+    },
+    {
+      "epoch": 0.7439153439153439,
+      "grad_norm": 0.2888862478367132,
+      "learning_rate": 4.1807134457075656e-05,
+      "loss": 0.6255,
+      "step": 703
+    },
+    {
+      "epoch": 0.744973544973545,
+      "grad_norm": 0.3007510062268032,
+      "learning_rate": 4.1787534300274405e-05,
+      "loss": 0.5565,
+      "step": 704
+    },
+    {
+      "epoch": 0.746031746031746,
+      "grad_norm": 0.27427995824764073,
+      "learning_rate": 4.176793414347315e-05,
+      "loss": 0.5533,
+      "step": 705
+    },
+    {
+      "epoch": 0.7470899470899471,
+      "grad_norm": 0.24178542578986884,
+      "learning_rate": 4.17483339866719e-05,
+      "loss": 0.4839,
+      "step": 706
+    },
+    {
+      "epoch": 0.7481481481481481,
+      "grad_norm": 0.3447762627838107,
+      "learning_rate": 4.172873382987064e-05,
+      "loss": 0.5418,
+      "step": 707
+    },
+    {
+      "epoch": 0.7492063492063492,
+      "grad_norm": 0.28512746124111127,
+      "learning_rate": 4.170913367306938e-05,
+      "loss": 0.5593,
+      "step": 708
+    },
+    {
+      "epoch": 0.7502645502645503,
+      "grad_norm": 0.25654137696146834,
+      "learning_rate": 4.168953351626813e-05,
+      "loss": 0.5308,
+      "step": 709
+    },
+    {
+      "epoch": 0.7513227513227513,
+      "grad_norm": 0.3162447256848906,
+      "learning_rate": 4.166993335946688e-05,
+      "loss": 0.5092,
+      "step": 710
+    },
+    {
+      "epoch": 0.7523809523809524,
+      "grad_norm": 0.2706709395661285,
+      "learning_rate": 4.165033320266562e-05,
+      "loss": 0.549,
+      "step": 711
+    },
+    {
+      "epoch": 0.7534391534391535,
+      "grad_norm": 0.27951783492610943,
+      "learning_rate": 4.1630733045864365e-05,
+      "loss": 0.5039,
+      "step": 712
+    },
+    {
+      "epoch": 0.7544973544973544,
+      "grad_norm": 0.34542932566254875,
+      "learning_rate": 4.1611132889063115e-05,
+      "loss": 0.5269,
+      "step": 713
+    },
+    {
+      "epoch": 0.7555555555555555,
+      "grad_norm": 0.2666243377680122,
+      "learning_rate": 4.1591532732261864e-05,
+      "loss": 0.5833,
+      "step": 714
+    },
+    {
+      "epoch": 0.7566137566137566,
+      "grad_norm": 0.3245207820226863,
+      "learning_rate": 4.1571932575460606e-05,
+      "loss": 0.5764,
+      "step": 715
+    },
+    {
+      "epoch": 0.7576719576719577,
+      "grad_norm": 0.26147609233769936,
+      "learning_rate": 4.155233241865935e-05,
+      "loss": 0.5527,
+      "step": 716
+    },
+    {
+      "epoch": 0.7587301587301587,
+      "grad_norm": 0.2584431704263216,
+      "learning_rate": 4.15327322618581e-05,
+      "loss": 0.5637,
+      "step": 717
+    },
+    {
+      "epoch": 0.7597883597883598,
+      "grad_norm": 0.2731600454527648,
+      "learning_rate": 4.151313210505684e-05,
+      "loss": 0.6257,
+      "step": 718
+    },
+    {
+      "epoch": 0.7608465608465609,
+      "grad_norm": 0.2430610169832162,
+      "learning_rate": 4.149353194825559e-05,
+      "loss": 0.5118,
+      "step": 719
+    },
+    {
+      "epoch": 0.7619047619047619,
+      "grad_norm": 0.46705609114743374,
+      "learning_rate": 4.147393179145433e-05,
+      "loss": 0.5657,
+      "step": 720
+    },
+    {
+      "epoch": 0.762962962962963,
+      "grad_norm": 0.2765619748725685,
+      "learning_rate": 4.145433163465308e-05,
+      "loss": 0.5653,
+      "step": 721
+    },
+    {
+      "epoch": 0.764021164021164,
+      "grad_norm": 0.2445879143109297,
+      "learning_rate": 4.1434731477851824e-05,
+      "loss": 0.5549,
+      "step": 722
+    },
+    {
+      "epoch": 0.765079365079365,
+      "grad_norm": 0.29444717176229945,
+      "learning_rate": 4.1415131321050573e-05,
+      "loss": 0.5159,
+      "step": 723
+    },
+    {
+      "epoch": 0.7661375661375661,
+      "grad_norm": 0.23188418245940617,
+      "learning_rate": 4.1395531164249316e-05,
+      "loss": 0.5596,
+      "step": 724
+    },
+    {
+      "epoch": 0.7671957671957672,
+      "grad_norm": 0.2726530916672444,
+      "learning_rate": 4.137593100744806e-05,
+      "loss": 0.5596,
+      "step": 725
+    },
+    {
+      "epoch": 0.7682539682539683,
+      "grad_norm": 0.26947956374326953,
+      "learning_rate": 4.135633085064681e-05,
+      "loss": 0.5813,
+      "step": 726
+    },
+    {
+      "epoch": 0.7693121693121693,
+      "grad_norm": 0.23506890162797972,
+      "learning_rate": 4.133673069384556e-05,
+      "loss": 0.5437,
+      "step": 727
+    },
+    {
+      "epoch": 0.7703703703703704,
+      "grad_norm": 2.972237291768401,
+      "learning_rate": 4.13171305370443e-05,
+      "loss": 0.5998,
+      "step": 728
+    },
+    {
+      "epoch": 0.7714285714285715,
+      "grad_norm": 0.27441250036308384,
+      "learning_rate": 4.129753038024304e-05,
+      "loss": 0.5271,
+      "step": 729
+    },
+    {
+      "epoch": 0.7724867724867724,
+      "grad_norm": 0.2522029487623296,
+      "learning_rate": 4.127793022344179e-05,
+      "loss": 0.5339,
+      "step": 730
+    },
+    {
+      "epoch": 0.7735449735449735,
+      "grad_norm": 0.25217726436063065,
+      "learning_rate": 4.1258330066640534e-05,
+      "loss": 0.578,
+      "step": 731
+    },
+    {
+      "epoch": 0.7746031746031746,
+      "grad_norm": 0.25344645901788887,
+      "learning_rate": 4.123872990983928e-05,
+      "loss": 0.5418,
+      "step": 732
+    },
+    {
+      "epoch": 0.7756613756613756,
+      "grad_norm": 0.2396746178638489,
+      "learning_rate": 4.1219129753038025e-05,
+      "loss": 0.5244,
+      "step": 733
+    },
+    {
+      "epoch": 0.7767195767195767,
+      "grad_norm": 0.31881877046038376,
+      "learning_rate": 4.1199529596236775e-05,
+      "loss": 0.5785,
+      "step": 734
+    },
+    {
+      "epoch": 0.7777777777777778,
+      "grad_norm": 0.23060282015738198,
+      "learning_rate": 4.117992943943552e-05,
+      "loss": 0.4795,
+      "step": 735
+    },
+    {
+      "epoch": 0.7788359788359789,
+      "grad_norm": 0.26057461199701626,
+      "learning_rate": 4.1160329282634266e-05,
+      "loss": 0.4967,
+      "step": 736
+    },
+    {
+      "epoch": 0.7798941798941799,
+      "grad_norm": 0.24273599035257443,
+      "learning_rate": 4.114072912583301e-05,
+      "loss": 0.5449,
+      "step": 737
+    },
+    {
+      "epoch": 0.780952380952381,
+      "grad_norm": 0.23318919061727453,
+      "learning_rate": 4.112112896903175e-05,
+      "loss": 0.4872,
+      "step": 738
+    },
+    {
+      "epoch": 0.782010582010582,
+      "grad_norm": 0.28999779048722424,
+      "learning_rate": 4.11015288122305e-05,
+      "loss": 0.6129,
+      "step": 739
+    },
+    {
+      "epoch": 0.783068783068783,
+      "grad_norm": 0.25705614998142,
+      "learning_rate": 4.108192865542925e-05,
+      "loss": 0.5755,
+      "step": 740
+    },
+    {
+      "epoch": 0.7841269841269841,
+      "grad_norm": 0.24516780262995658,
+      "learning_rate": 4.1062328498627986e-05,
+      "loss": 0.5313,
+      "step": 741
+    },
+    {
+      "epoch": 0.7851851851851852,
+      "grad_norm": 0.2632105354730925,
+      "learning_rate": 4.1042728341826735e-05,
+      "loss": 0.4978,
+      "step": 742
+    },
+    {
+      "epoch": 0.7862433862433862,
+      "grad_norm": 0.2950535479592732,
+      "learning_rate": 4.1023128185025484e-05,
+      "loss": 0.5477,
+      "step": 743
+    },
+    {
+      "epoch": 0.7873015873015873,
+      "grad_norm": 0.2890918764180236,
+      "learning_rate": 4.100352802822423e-05,
+      "loss": 0.6027,
+      "step": 744
+    },
+    {
+      "epoch": 0.7883597883597884,
+      "grad_norm": 0.24905516173631076,
+      "learning_rate": 4.098392787142297e-05,
+      "loss": 0.4916,
+      "step": 745
+    },
+    {
+      "epoch": 0.7894179894179895,
+      "grad_norm": 0.30918231275089586,
+      "learning_rate": 4.096432771462172e-05,
+      "loss": 0.5521,
+      "step": 746
+    },
+    {
+      "epoch": 0.7904761904761904,
+      "grad_norm": 0.254678981001304,
+      "learning_rate": 4.094472755782047e-05,
+      "loss": 0.5693,
+      "step": 747
+    },
+    {
+      "epoch": 0.7915343915343915,
+      "grad_norm": 0.24323586920042417,
+      "learning_rate": 4.092512740101921e-05,
+      "loss": 0.5497,
+      "step": 748
+    },
+    {
+      "epoch": 0.7925925925925926,
+      "grad_norm": 0.2888285005504316,
+      "learning_rate": 4.090552724421795e-05,
+      "loss": 0.4983,
+      "step": 749
+    },
+    {
+      "epoch": 0.7936507936507936,
+      "grad_norm": 0.259386381669663,
+      "learning_rate": 4.08859270874167e-05,
+      "loss": 0.6195,
+      "step": 750
+    },
+    {
+      "epoch": 0.7947089947089947,
+      "grad_norm": 0.23036461293567234,
+      "learning_rate": 4.0866326930615444e-05,
+      "loss": 0.5492,
+      "step": 751
+    },
+    {
+      "epoch": 0.7957671957671958,
+      "grad_norm": 0.2678855648948368,
+      "learning_rate": 4.0846726773814194e-05,
+      "loss": 0.5072,
+      "step": 752
+    },
+    {
+      "epoch": 0.7968253968253968,
+      "grad_norm": 0.2887180222360751,
+      "learning_rate": 4.082712661701294e-05,
+      "loss": 0.5246,
+      "step": 753
+    },
+    {
+      "epoch": 0.7978835978835979,
+      "grad_norm": 0.2568292309960633,
+      "learning_rate": 4.080752646021168e-05,
+      "loss": 0.5184,
+      "step": 754
+    },
+    {
+      "epoch": 0.798941798941799,
+      "grad_norm": 0.24711251400823,
+      "learning_rate": 4.078792630341043e-05,
+      "loss": 0.5341,
+      "step": 755
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.31036770266770836,
+      "learning_rate": 4.076832614660918e-05,
+      "loss": 0.6273,
+      "step": 756
+    },
+    {
+      "epoch": 0.801058201058201,
+      "grad_norm": 0.2571757220271915,
+      "learning_rate": 4.074872598980792e-05,
+      "loss": 0.5696,
+      "step": 757
+    },
+    {
+      "epoch": 0.8021164021164021,
+      "grad_norm": 0.26980125079686357,
+      "learning_rate": 4.072912583300666e-05,
+      "loss": 0.5467,
+      "step": 758
+    },
+    {
+      "epoch": 0.8031746031746032,
+      "grad_norm": 0.24951832038119673,
+      "learning_rate": 4.070952567620541e-05,
+      "loss": 0.5083,
+      "step": 759
+    },
+    {
+      "epoch": 0.8042328042328042,
+      "grad_norm": 0.25244068781335705,
+      "learning_rate": 4.068992551940416e-05,
+      "loss": 0.4868,
+      "step": 760
+    },
+    {
+      "epoch": 0.8052910052910053,
+      "grad_norm": 0.24075958410510967,
+      "learning_rate": 4.06703253626029e-05,
+      "loss": 0.5132,
+      "step": 761
+    },
+    {
+      "epoch": 0.8063492063492064,
+      "grad_norm": 0.299818761325055,
+      "learning_rate": 4.0650725205801646e-05,
+      "loss": 0.5085,
+      "step": 762
+    },
+    {
+      "epoch": 0.8074074074074075,
+      "grad_norm": 0.2556890383489905,
+      "learning_rate": 4.0631125049000395e-05,
+      "loss": 0.6116,
+      "step": 763
+    },
+    {
+      "epoch": 0.8084656084656084,
+      "grad_norm": 0.25781514459024585,
+      "learning_rate": 4.061152489219914e-05,
+      "loss": 0.5138,
+      "step": 764
+    },
+    {
+      "epoch": 0.8095238095238095,
+      "grad_norm": 0.4812654512833488,
+      "learning_rate": 4.059192473539789e-05,
+      "loss": 0.5117,
+      "step": 765
+    },
+    {
+      "epoch": 0.8105820105820106,
+      "grad_norm": 0.43995941221107904,
+      "learning_rate": 4.057232457859663e-05,
+      "loss": 0.523,
+      "step": 766
+    },
+    {
+      "epoch": 0.8116402116402116,
+      "grad_norm": 0.2570086400263589,
+      "learning_rate": 4.055272442179537e-05,
+      "loss": 0.4922,
+      "step": 767
+    },
+    {
+      "epoch": 0.8126984126984127,
+      "grad_norm": 0.27355262696697646,
+      "learning_rate": 4.053312426499412e-05,
+      "loss": 0.5214,
+      "step": 768
+    },
+    {
+      "epoch": 0.8137566137566138,
+      "grad_norm": 0.30929404940417793,
+      "learning_rate": 4.051352410819287e-05,
+      "loss": 0.5595,
+      "step": 769
+    },
+    {
+      "epoch": 0.8148148148148148,
+      "grad_norm": 0.25651422442360594,
+      "learning_rate": 4.049392395139161e-05,
+      "loss": 0.5977,
+      "step": 770
+    },
+    {
+      "epoch": 0.8158730158730159,
+      "grad_norm": 0.3254465197758939,
+      "learning_rate": 4.0474323794590355e-05,
+      "loss": 0.5534,
+      "step": 771
+    },
+    {
+      "epoch": 0.816931216931217,
+      "grad_norm": 0.2859692925402336,
+      "learning_rate": 4.0454723637789104e-05,
+      "loss": 0.56,
+      "step": 772
+    },
+    {
+      "epoch": 0.817989417989418,
+      "grad_norm": 0.24530950245641758,
+      "learning_rate": 4.0435123480987854e-05,
+      "loss": 0.4727,
+      "step": 773
+    },
+    {
+      "epoch": 0.819047619047619,
+      "grad_norm": 0.2852863920952454,
+      "learning_rate": 4.0415523324186596e-05,
+      "loss": 0.5166,
+      "step": 774
+    },
+    {
+      "epoch": 0.8201058201058201,
+      "grad_norm": 0.28797276683526324,
+      "learning_rate": 4.039592316738534e-05,
+      "loss": 0.6108,
+      "step": 775
+    },
+    {
+      "epoch": 0.8211640211640212,
+      "grad_norm": 0.24315570228929462,
+      "learning_rate": 4.037632301058409e-05,
+      "loss": 0.5383,
+      "step": 776
+    },
+    {
+      "epoch": 0.8222222222222222,
+      "grad_norm": 0.27806729681229697,
+      "learning_rate": 4.035672285378283e-05,
+      "loss": 0.5445,
+      "step": 777
+    },
+    {
+      "epoch": 0.8232804232804233,
+      "grad_norm": 0.27598442145764895,
+      "learning_rate": 4.033712269698158e-05,
+      "loss": 0.4656,
+      "step": 778
+    },
+    {
+      "epoch": 0.8243386243386244,
+      "grad_norm": 0.22788260383702866,
+      "learning_rate": 4.031752254018032e-05,
+      "loss": 0.4989,
+      "step": 779
+    },
+    {
+      "epoch": 0.8253968253968254,
+      "grad_norm": 0.2799668545860428,
+      "learning_rate": 4.0297922383379065e-05,
+      "loss": 0.6928,
+      "step": 780
+    },
+    {
+      "epoch": 0.8264550264550264,
+      "grad_norm": 0.2393669331758126,
+      "learning_rate": 4.0278322226577814e-05,
+      "loss": 0.4514,
+      "step": 781
+    },
+    {
+      "epoch": 0.8275132275132275,
+      "grad_norm": 0.2625212858096316,
+      "learning_rate": 4.025872206977656e-05,
+      "loss": 0.5826,
+      "step": 782
+    },
+    {
+      "epoch": 0.8285714285714286,
+      "grad_norm": 0.28090132618519864,
+      "learning_rate": 4.0239121912975306e-05,
+      "loss": 0.5763,
+      "step": 783
+    },
+    {
+      "epoch": 0.8296296296296296,
+      "grad_norm": 0.254545648971496,
+      "learning_rate": 4.021952175617405e-05,
+      "loss": 0.5603,
+      "step": 784
+    },
+    {
+      "epoch": 0.8306878306878307,
+      "grad_norm": 0.3026212789118765,
+      "learning_rate": 4.01999215993728e-05,
+      "loss": 0.5572,
+      "step": 785
+    },
+    {
+      "epoch": 0.8317460317460318,
+      "grad_norm": 0.2583084756689895,
+      "learning_rate": 4.018032144257155e-05,
+      "loss": 0.5687,
+      "step": 786
+    },
+    {
+      "epoch": 0.8328042328042328,
+      "grad_norm": 0.2912070717602156,
+      "learning_rate": 4.016072128577029e-05,
+      "loss": 0.4976,
+      "step": 787
+    },
+    {
+      "epoch": 0.8338624338624339,
+      "grad_norm": 0.3091276282835055,
+      "learning_rate": 4.014112112896903e-05,
+      "loss": 0.602,
+      "step": 788
+    },
+    {
+      "epoch": 0.834920634920635,
+      "grad_norm": 0.2614685006059475,
+      "learning_rate": 4.012152097216778e-05,
+      "loss": 0.6047,
+      "step": 789
+    },
+    {
+      "epoch": 0.8359788359788359,
+      "grad_norm": 0.2976678306406433,
+      "learning_rate": 4.010192081536652e-05,
+      "loss": 0.5572,
+      "step": 790
+    },
+    {
+      "epoch": 0.837037037037037,
+      "grad_norm": 0.26695480665010896,
+      "learning_rate": 4.008232065856527e-05,
+      "loss": 0.5368,
+      "step": 791
+    },
+    {
+      "epoch": 0.8380952380952381,
+      "grad_norm": 0.27224551418357,
+      "learning_rate": 4.0062720501764015e-05,
+      "loss": 0.5715,
+      "step": 792
+    },
+    {
+      "epoch": 0.8391534391534392,
+      "grad_norm": 0.24476160512891385,
+      "learning_rate": 4.004312034496276e-05,
+      "loss": 0.5734,
+      "step": 793
+    },
+    {
+      "epoch": 0.8402116402116402,
+      "grad_norm": 0.26511024491224705,
+      "learning_rate": 4.002352018816151e-05,
+      "loss": 0.5795,
+      "step": 794
+    },
+    {
+      "epoch": 0.8412698412698413,
+      "grad_norm": 0.34773604650933276,
+      "learning_rate": 4.0003920031360256e-05,
+      "loss": 0.5005,
+      "step": 795
+    },
+    {
+      "epoch": 0.8423280423280424,
+      "grad_norm": 0.2615336683961906,
+      "learning_rate": 3.9984319874559e-05,
+      "loss": 0.5364,
+      "step": 796
+    },
+    {
+      "epoch": 0.8433862433862434,
+      "grad_norm": 0.2739470851171991,
+      "learning_rate": 3.996471971775774e-05,
+      "loss": 0.4933,
+      "step": 797
+    },
+    {
+      "epoch": 0.8444444444444444,
+      "grad_norm": 0.3482221259725657,
+      "learning_rate": 3.994511956095649e-05,
+      "loss": 0.6087,
+      "step": 798
+    },
+    {
+      "epoch": 0.8455026455026455,
+      "grad_norm": 7.054547078978019,
+      "learning_rate": 3.992551940415524e-05,
+      "loss": 0.5531,
+      "step": 799
+    },
+    {
+      "epoch": 0.8465608465608465,
+      "grad_norm": 0.3025752493200541,
+      "learning_rate": 3.990591924735398e-05,
+      "loss": 0.5327,
+      "step": 800
+    },
+    {
+      "epoch": 0.8476190476190476,
+      "grad_norm": 0.2515881041243724,
+      "learning_rate": 3.9886319090552725e-05,
+      "loss": 0.5089,
+      "step": 801
+    },
+    {
+      "epoch": 0.8486772486772487,
+      "grad_norm": 0.29447997518392294,
+      "learning_rate": 3.9866718933751474e-05,
+      "loss": 0.5635,
+      "step": 802
+    },
+    {
+      "epoch": 0.8497354497354498,
+      "grad_norm": 1.2056063054042254,
+      "learning_rate": 3.9847118776950216e-05,
+      "loss": 0.568,
+      "step": 803
+    },
+    {
+      "epoch": 0.8507936507936508,
+      "grad_norm": 0.24867885592706862,
+      "learning_rate": 3.9827518620148966e-05,
+      "loss": 0.5172,
+      "step": 804
+    },
+    {
+      "epoch": 0.8518518518518519,
+      "grad_norm": 0.2833308927773852,
+      "learning_rate": 3.980791846334771e-05,
+      "loss": 0.5552,
+      "step": 805
+    },
+    {
+      "epoch": 0.852910052910053,
+      "grad_norm": 0.43186449303900065,
+      "learning_rate": 3.978831830654645e-05,
+      "loss": 0.4626,
+      "step": 806
+    },
+    {
+      "epoch": 0.8539682539682539,
+      "grad_norm": 0.28455060749504574,
+      "learning_rate": 3.97687181497452e-05,
+      "loss": 0.5228,
+      "step": 807
+    },
+    {
+      "epoch": 0.855026455026455,
+      "grad_norm": 0.29552626608893695,
+      "learning_rate": 3.974911799294395e-05,
+      "loss": 0.5452,
+      "step": 808
+    },
+    {
+      "epoch": 0.8560846560846561,
+      "grad_norm": 0.24084287912602692,
+      "learning_rate": 3.972951783614269e-05,
+      "loss": 0.54,
+      "step": 809
+    },
+    {
+      "epoch": 0.8571428571428571,
+      "grad_norm": 0.28708908508185177,
+      "learning_rate": 3.9709917679341434e-05,
+      "loss": 0.524,
+      "step": 810
+    },
+    {
+      "epoch": 0.8582010582010582,
+      "grad_norm": 0.26965168845733906,
+      "learning_rate": 3.969031752254018e-05,
+      "loss": 0.5084,
+      "step": 811
+    },
+    {
+      "epoch": 0.8592592592592593,
+      "grad_norm": 0.27200841727390407,
+      "learning_rate": 3.967071736573893e-05,
+      "loss": 0.6633,
+      "step": 812
+    },
+    {
+      "epoch": 0.8603174603174604,
+      "grad_norm": 0.4577420835451772,
+      "learning_rate": 3.965111720893767e-05,
+      "loss": 0.5984,
+      "step": 813
+    },
+    {
+      "epoch": 0.8613756613756614,
+      "grad_norm": 0.23226446549107316,
+      "learning_rate": 3.963151705213642e-05,
+      "loss": 0.5383,
+      "step": 814
+    },
+    {
+      "epoch": 0.8624338624338624,
+      "grad_norm": 0.24732625840438446,
+      "learning_rate": 3.961191689533517e-05,
+      "loss": 0.4663,
+      "step": 815
+    },
+    {
+      "epoch": 0.8634920634920635,
+      "grad_norm": 0.2442585038196715,
+      "learning_rate": 3.959231673853391e-05,
+      "loss": 0.4915,
+      "step": 816
+    },
+    {
+      "epoch": 0.8645502645502645,
+      "grad_norm": 0.24133270544480462,
+      "learning_rate": 3.957271658173265e-05,
+      "loss": 0.5099,
+      "step": 817
+    },
+    {
+      "epoch": 0.8656084656084656,
+      "grad_norm": 0.2375341297040655,
+      "learning_rate": 3.95531164249314e-05,
+      "loss": 0.4853,
+      "step": 818
+    },
+    {
+      "epoch": 0.8666666666666667,
+      "grad_norm": 0.2374457163459558,
+      "learning_rate": 3.953351626813015e-05,
+      "loss": 0.497,
+      "step": 819
+    },
+    {
+      "epoch": 0.8677248677248677,
+      "grad_norm": 0.2944923584520418,
+      "learning_rate": 3.951391611132889e-05,
+      "loss": 0.5841,
+      "step": 820
+    },
+    {
+      "epoch": 0.8687830687830688,
+      "grad_norm": 0.3103954396715373,
+      "learning_rate": 3.9494315954527635e-05,
+      "loss": 0.6308,
+      "step": 821
+    },
+    {
+      "epoch": 0.8698412698412699,
+      "grad_norm": 0.3044635700570842,
+      "learning_rate": 3.9474715797726385e-05,
+      "loss": 0.6198,
+      "step": 822
+    },
+    {
+      "epoch": 0.870899470899471,
+      "grad_norm": 0.25175807643491943,
+      "learning_rate": 3.945511564092513e-05,
+      "loss": 0.5508,
+      "step": 823
+    },
+    {
+      "epoch": 0.8719576719576719,
+      "grad_norm": 0.22937033545757074,
+      "learning_rate": 3.9435515484123876e-05,
+      "loss": 0.5033,
+      "step": 824
+    },
+    {
+      "epoch": 0.873015873015873,
+      "grad_norm": 0.2414211430197171,
+      "learning_rate": 3.9415915327322626e-05,
+      "loss": 0.5281,
+      "step": 825
+    },
+    {
+      "epoch": 0.8740740740740741,
+      "grad_norm": 0.36303292523703895,
+      "learning_rate": 3.939631517052136e-05,
+      "loss": 0.4688,
+      "step": 826
+    },
+    {
+      "epoch": 0.8751322751322751,
+      "grad_norm": 0.2840607840231142,
+      "learning_rate": 3.937671501372011e-05,
+      "loss": 0.5765,
+      "step": 827
+    },
+    {
+      "epoch": 0.8761904761904762,
+      "grad_norm": 0.26470574352059545,
+      "learning_rate": 3.935711485691886e-05,
+      "loss": 0.5827,
+      "step": 828
+    },
+    {
+      "epoch": 0.8772486772486773,
+      "grad_norm": 0.23448435580139299,
+      "learning_rate": 3.93375147001176e-05,
+      "loss": 0.4977,
+      "step": 829
+    },
+    {
+      "epoch": 0.8783068783068783,
+      "grad_norm": 0.2635026772188355,
+      "learning_rate": 3.9317914543316345e-05,
+      "loss": 0.521,
+      "step": 830
+    },
+    {
+      "epoch": 0.8793650793650793,
+      "grad_norm": 0.26987056951962557,
+      "learning_rate": 3.9298314386515094e-05,
+      "loss": 0.5562,
+      "step": 831
+    },
+    {
+      "epoch": 0.8804232804232804,
+      "grad_norm": 0.2365896896674794,
+      "learning_rate": 3.927871422971384e-05,
+      "loss": 0.5206,
+      "step": 832
+    },
+    {
+      "epoch": 0.8814814814814815,
+      "grad_norm": 0.2617025654901415,
+      "learning_rate": 3.9259114072912586e-05,
+      "loss": 0.53,
+      "step": 833
+    },
+    {
+      "epoch": 0.8825396825396825,
+      "grad_norm": 0.24319463532557173,
+      "learning_rate": 3.923951391611133e-05,
+      "loss": 0.5784,
+      "step": 834
+    },
+    {
+      "epoch": 0.8835978835978836,
+      "grad_norm": 0.30018809738193136,
+      "learning_rate": 3.921991375931008e-05,
+      "loss": 0.5594,
+      "step": 835
+    },
+    {
+      "epoch": 0.8846560846560847,
+      "grad_norm": 0.24564159380840794,
+      "learning_rate": 3.920031360250882e-05,
+      "loss": 0.5678,
+      "step": 836
+    },
+    {
+      "epoch": 0.8857142857142857,
+      "grad_norm": 0.24469707503949964,
+      "learning_rate": 3.918071344570757e-05,
+      "loss": 0.6056,
+      "step": 837
+    },
+    {
+      "epoch": 0.8867724867724868,
+      "grad_norm": 0.28670136066959284,
+      "learning_rate": 3.916111328890631e-05,
+      "loss": 0.503,
+      "step": 838
+    },
+    {
+      "epoch": 0.8878306878306879,
+      "grad_norm": 0.25036374517538473,
+      "learning_rate": 3.9141513132105054e-05,
+      "loss": 0.536,
+      "step": 839
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.22936770624952652,
+      "learning_rate": 3.9121912975303804e-05,
+      "loss": 0.4632,
+      "step": 840
+    },
+    {
+      "epoch": 0.8899470899470899,
+      "grad_norm": 0.23888772118569762,
+      "learning_rate": 3.910231281850255e-05,
+      "loss": 0.539,
+      "step": 841
+    },
+    {
+      "epoch": 0.891005291005291,
+      "grad_norm": 0.26618130929626477,
+      "learning_rate": 3.9082712661701295e-05,
+      "loss": 0.5135,
+      "step": 842
+    },
+    {
+      "epoch": 0.8920634920634921,
+      "grad_norm": 0.23048542050631554,
+      "learning_rate": 3.906311250490004e-05,
+      "loss": 0.4616,
+      "step": 843
+    },
+    {
+      "epoch": 0.8931216931216931,
+      "grad_norm": 0.23846574548002752,
+      "learning_rate": 3.904351234809879e-05,
+      "loss": 0.56,
+      "step": 844
+    },
+    {
+      "epoch": 0.8941798941798942,
+      "grad_norm": 0.25578679656745784,
+      "learning_rate": 3.9023912191297536e-05,
+      "loss": 0.5171,
+      "step": 845
+    },
+    {
+      "epoch": 0.8952380952380953,
+      "grad_norm": 0.2604178853945741,
+      "learning_rate": 3.900431203449628e-05,
+      "loss": 0.4736,
+      "step": 846
+    },
+    {
+      "epoch": 0.8962962962962963,
+      "grad_norm": 0.2543159342822874,
+      "learning_rate": 3.898471187769502e-05,
+      "loss": 0.553,
+      "step": 847
+    },
+    {
+      "epoch": 0.8973544973544973,
+      "grad_norm": 0.25668454175903266,
+      "learning_rate": 3.896511172089377e-05,
+      "loss": 0.5986,
+      "step": 848
+    },
+    {
+      "epoch": 0.8984126984126984,
+      "grad_norm": 0.27128711537893746,
+      "learning_rate": 3.894551156409251e-05,
+      "loss": 0.6013,
+      "step": 849
+    },
+    {
+      "epoch": 0.8994708994708994,
+      "grad_norm": 0.25661383335622456,
+      "learning_rate": 3.892591140729126e-05,
+      "loss": 0.6018,
+      "step": 850
+    },
+    {
+      "epoch": 0.9005291005291005,
+      "grad_norm": 0.26144285059562283,
+      "learning_rate": 3.8906311250490005e-05,
+      "loss": 0.5544,
+      "step": 851
+    },
+    {
+      "epoch": 0.9015873015873016,
+      "grad_norm": 0.24851759763750786,
+      "learning_rate": 3.888671109368875e-05,
+      "loss": 0.5417,
+      "step": 852
+    },
+    {
+      "epoch": 0.9026455026455027,
+      "grad_norm": 0.22366950486260603,
+      "learning_rate": 3.8867110936887497e-05,
+      "loss": 0.5216,
+      "step": 853
+    },
+    {
+      "epoch": 0.9037037037037037,
+      "grad_norm": 0.25522407995231283,
+      "learning_rate": 3.8847510780086246e-05,
+      "loss": 0.5511,
+      "step": 854
+    },
+    {
+      "epoch": 0.9047619047619048,
+      "grad_norm": 0.2459244748277179,
+      "learning_rate": 3.882791062328499e-05,
+      "loss": 0.4738,
+      "step": 855
+    },
+    {
+      "epoch": 0.9058201058201059,
+      "grad_norm": 0.23773875962661492,
+      "learning_rate": 3.880831046648373e-05,
+      "loss": 0.5866,
+      "step": 856
+    },
+    {
+      "epoch": 0.9068783068783068,
+      "grad_norm": 0.2580145368233666,
+      "learning_rate": 3.878871030968248e-05,
+      "loss": 0.5448,
+      "step": 857
+    },
+    {
+      "epoch": 0.9079365079365079,
+      "grad_norm": 0.24118543676000287,
+      "learning_rate": 3.876911015288123e-05,
+      "loss": 0.5223,
+      "step": 858
+    },
+    {
+      "epoch": 0.908994708994709,
+      "grad_norm": 0.24514536124293135,
+      "learning_rate": 3.874950999607997e-05,
+      "loss": 0.5641,
+      "step": 859
+    },
+    {
+      "epoch": 0.91005291005291,
+      "grad_norm": 0.23032586872284455,
+      "learning_rate": 3.8729909839278714e-05,
+      "loss": 0.5285,
+      "step": 860
+    },
+    {
+      "epoch": 0.9111111111111111,
+      "grad_norm": 0.24435066352876703,
+      "learning_rate": 3.8710309682477464e-05,
+      "loss": 0.5423,
+      "step": 861
+    },
+    {
+      "epoch": 0.9121693121693122,
+      "grad_norm": 0.2622407130624908,
+      "learning_rate": 3.8690709525676206e-05,
+      "loss": 0.5741,
+      "step": 862
+    },
+    {
+      "epoch": 0.9132275132275133,
+      "grad_norm": 1.3636871490670779,
+      "learning_rate": 3.8671109368874955e-05,
+      "loss": 0.7745,
+      "step": 863
+    },
+    {
+      "epoch": 0.9142857142857143,
+      "grad_norm": 0.3780723682991901,
+      "learning_rate": 3.86515092120737e-05,
+      "loss": 0.5085,
+      "step": 864
+    },
+    {
+      "epoch": 0.9153439153439153,
+      "grad_norm": 0.24726218347601356,
+      "learning_rate": 3.863190905527244e-05,
+      "loss": 0.5712,
+      "step": 865
+    },
+    {
+      "epoch": 0.9164021164021164,
+      "grad_norm": 0.27126893307687105,
+      "learning_rate": 3.861230889847119e-05,
+      "loss": 0.5255,
+      "step": 866
+    },
+    {
+      "epoch": 0.9174603174603174,
+      "grad_norm": 0.2426515970790078,
+      "learning_rate": 3.859270874166994e-05,
+      "loss": 0.5768,
+      "step": 867
+    },
+    {
+      "epoch": 0.9185185185185185,
+      "grad_norm": 0.28594907335057773,
+      "learning_rate": 3.857310858486868e-05,
+      "loss": 0.5528,
+      "step": 868
+    },
+    {
+      "epoch": 0.9195767195767196,
+      "grad_norm": 0.23606697714730085,
+      "learning_rate": 3.8553508428067424e-05,
+      "loss": 0.5133,
+      "step": 869
+    },
+    {
+      "epoch": 0.9206349206349206,
+      "grad_norm": 0.27926385748395915,
+      "learning_rate": 3.853390827126617e-05,
+      "loss": 0.5796,
+      "step": 870
+    },
+    {
+      "epoch": 0.9216931216931217,
+      "grad_norm": 0.24343714813051942,
+      "learning_rate": 3.851430811446492e-05,
+      "loss": 0.5295,
+      "step": 871
+    },
+    {
+      "epoch": 0.9227513227513228,
+      "grad_norm": 0.26244227600390824,
+      "learning_rate": 3.8494707957663665e-05,
+      "loss": 0.4905,
+      "step": 872
+    },
+    {
+      "epoch": 0.9238095238095239,
+      "grad_norm": 0.2765749086021136,
+      "learning_rate": 3.847510780086241e-05,
+      "loss": 0.5227,
+      "step": 873
+    },
+    {
+      "epoch": 0.9248677248677248,
+      "grad_norm": 0.22924047622529786,
+      "learning_rate": 3.8455507644061157e-05,
+      "loss": 0.5447,
+      "step": 874
+    },
+    {
+      "epoch": 0.9259259259259259,
+      "grad_norm": 0.26004927012669626,
+      "learning_rate": 3.84359074872599e-05,
+      "loss": 0.5876,
+      "step": 875
+    },
+    {
+      "epoch": 0.926984126984127,
+      "grad_norm": 0.23257781793426832,
+      "learning_rate": 3.841630733045865e-05,
+      "loss": 0.4462,
+      "step": 876
+    },
+    {
+      "epoch": 0.928042328042328,
+      "grad_norm": 0.23530057335980378,
+      "learning_rate": 3.839670717365739e-05,
+      "loss": 0.5099,
+      "step": 877
+    },
+    {
+      "epoch": 0.9291005291005291,
+      "grad_norm": 0.22869829817586848,
+      "learning_rate": 3.837710701685613e-05,
+      "loss": 0.4933,
+      "step": 878
+    },
+    {
+      "epoch": 0.9301587301587302,
+      "grad_norm": 0.2538072480043844,
+      "learning_rate": 3.835750686005488e-05,
+      "loss": 0.4838,
+      "step": 879
+    },
+    {
+      "epoch": 0.9312169312169312,
+      "grad_norm": 0.23947855637218263,
+      "learning_rate": 3.833790670325363e-05,
+      "loss": 0.5569,
+      "step": 880
+    },
+    {
+      "epoch": 0.9322751322751323,
+      "grad_norm": 0.2379463855293634,
+      "learning_rate": 3.8318306546452374e-05,
+      "loss": 0.5169,
+      "step": 881
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "grad_norm": 0.2706377388516219,
+      "learning_rate": 3.829870638965112e-05,
+      "loss": 0.4809,
+      "step": 882
+    },
+    {
+      "epoch": 0.9343915343915344,
+      "grad_norm": 0.2599921459464922,
+      "learning_rate": 3.8279106232849866e-05,
+      "loss": 0.5511,
+      "step": 883
+    },
+    {
+      "epoch": 0.9354497354497354,
+      "grad_norm": 0.23700018733026876,
+      "learning_rate": 3.8259506076048615e-05,
+      "loss": 0.5508,
+      "step": 884
+    },
+    {
+      "epoch": 0.9365079365079365,
+      "grad_norm": 0.27148173561209227,
+      "learning_rate": 3.823990591924735e-05,
+      "loss": 0.5598,
+      "step": 885
+    },
+    {
+      "epoch": 0.9375661375661376,
+      "grad_norm": 0.23284787985918257,
+      "learning_rate": 3.82203057624461e-05,
+      "loss": 0.4889,
+      "step": 886
+    },
+    {
+      "epoch": 0.9386243386243386,
+      "grad_norm": 0.2289097690533918,
+      "learning_rate": 3.820070560564485e-05,
+      "loss": 0.484,
+      "step": 887
+    },
+    {
+      "epoch": 0.9396825396825397,
+      "grad_norm": 0.29024067491700634,
+      "learning_rate": 3.818110544884359e-05,
+      "loss": 0.527,
+      "step": 888
+    },
+    {
+      "epoch": 0.9407407407407408,
+      "grad_norm": 0.2573481910146356,
+      "learning_rate": 3.8161505292042334e-05,
+      "loss": 0.4921,
+      "step": 889
+    },
+    {
+      "epoch": 0.9417989417989417,
+      "grad_norm": 0.2287016832994264,
+      "learning_rate": 3.8141905135241084e-05,
+      "loss": 0.4975,
+      "step": 890
+    },
+    {
+      "epoch": 0.9428571428571428,
+      "grad_norm": 0.2793182907010499,
+      "learning_rate": 3.8122304978439826e-05,
+      "loss": 0.5901,
+      "step": 891
+    },
+    {
+      "epoch": 0.9439153439153439,
+      "grad_norm": 0.2815450954484471,
+      "learning_rate": 3.8102704821638575e-05,
+      "loss": 0.4907,
+      "step": 892
+    },
+    {
+      "epoch": 0.944973544973545,
+      "grad_norm": 0.24825039265577378,
+      "learning_rate": 3.808310466483732e-05,
+      "loss": 0.5845,
+      "step": 893
+    },
+    {
+      "epoch": 0.946031746031746,
+      "grad_norm": 0.2684156933456286,
+      "learning_rate": 3.806350450803607e-05,
+      "loss": 0.4972,
+      "step": 894
+    },
+    {
+      "epoch": 0.9470899470899471,
+      "grad_norm": 0.2777947126936837,
+      "learning_rate": 3.804390435123481e-05,
+      "loss": 0.5588,
+      "step": 895
+    },
+    {
+      "epoch": 0.9481481481481482,
+      "grad_norm": 0.2479091360698014,
+      "learning_rate": 3.802430419443356e-05,
+      "loss": 0.5882,
+      "step": 896
+    },
+    {
+      "epoch": 0.9492063492063492,
+      "grad_norm": 0.2657834372378511,
+      "learning_rate": 3.800470403763231e-05,
+      "loss": 0.5495,
+      "step": 897
+    },
+    {
+      "epoch": 0.9502645502645503,
+      "grad_norm": 0.9282022073586812,
+      "learning_rate": 3.7985103880831044e-05,
+      "loss": 0.5746,
+      "step": 898
+    },
+    {
+      "epoch": 0.9513227513227513,
+      "grad_norm": 0.25172528574506164,
+      "learning_rate": 3.796550372402979e-05,
+      "loss": 0.4896,
+      "step": 899
+    },
+    {
+      "epoch": 0.9523809523809523,
+      "grad_norm": 0.2555445883398386,
+      "learning_rate": 3.794590356722854e-05,
+      "loss": 0.5736,
+      "step": 900
+    },
+    {
+      "epoch": 0.9534391534391534,
+      "grad_norm": 0.249355354605511,
+      "learning_rate": 3.7926303410427285e-05,
+      "loss": 0.4817,
+      "step": 901
+    },
+    {
+      "epoch": 0.9544973544973545,
+      "grad_norm": 0.2149812919434468,
+      "learning_rate": 3.790670325362603e-05,
+      "loss": 0.4524,
+      "step": 902
+    },
+    {
+      "epoch": 0.9555555555555556,
+      "grad_norm": 0.2703911118438873,
+      "learning_rate": 3.788710309682478e-05,
+      "loss": 0.5443,
+      "step": 903
+    },
+    {
+      "epoch": 0.9566137566137566,
+      "grad_norm": 0.223977288588384,
+      "learning_rate": 3.786750294002352e-05,
+      "loss": 0.4925,
+      "step": 904
+    },
+    {
+      "epoch": 0.9576719576719577,
+      "grad_norm": 0.27011460419403965,
+      "learning_rate": 3.784790278322227e-05,
+      "loss": 0.5812,
+      "step": 905
+    },
+    {
+      "epoch": 0.9587301587301588,
+      "grad_norm": 0.23957948838890472,
+      "learning_rate": 3.782830262642101e-05,
+      "loss": 0.5007,
+      "step": 906
+    },
+    {
+      "epoch": 0.9597883597883597,
+      "grad_norm": 0.2264533105409485,
+      "learning_rate": 3.780870246961976e-05,
+      "loss": 0.5141,
+      "step": 907
+    },
+    {
+      "epoch": 0.9608465608465608,
+      "grad_norm": 0.24934710629027704,
+      "learning_rate": 3.77891023128185e-05,
+      "loss": 0.5271,
+      "step": 908
+    },
+    {
+      "epoch": 0.9619047619047619,
+      "grad_norm": 0.24400597517616687,
+      "learning_rate": 3.776950215601725e-05,
+      "loss": 0.5524,
+      "step": 909
+    },
+    {
+      "epoch": 0.9629629629629629,
+      "grad_norm": 0.24087812537974612,
+      "learning_rate": 3.7749901999215994e-05,
+      "loss": 0.4046,
+      "step": 910
+    },
+    {
+      "epoch": 0.964021164021164,
+      "grad_norm": 2.6978887892991747,
+      "learning_rate": 3.773030184241474e-05,
+      "loss": 0.5807,
+      "step": 911
+    },
+    {
+      "epoch": 0.9650793650793651,
+      "grad_norm": 0.31338696022947576,
+      "learning_rate": 3.7710701685613486e-05,
+      "loss": 0.5033,
+      "step": 912
+    },
+    {
+      "epoch": 0.9661375661375662,
+      "grad_norm": 0.3061821128532338,
+      "learning_rate": 3.7691101528812235e-05,
+      "loss": 0.5131,
+      "step": 913
+    },
+    {
+      "epoch": 0.9671957671957672,
+      "grad_norm": 0.23119947689356746,
+      "learning_rate": 3.767150137201098e-05,
+      "loss": 0.4257,
+      "step": 914
+    },
+    {
+      "epoch": 0.9682539682539683,
+      "grad_norm": 0.27958747828152924,
+      "learning_rate": 3.765190121520972e-05,
+      "loss": 0.543,
+      "step": 915
+    },
+    {
+      "epoch": 0.9693121693121693,
+      "grad_norm": 0.30495291790476814,
+      "learning_rate": 3.763230105840847e-05,
+      "loss": 0.5112,
+      "step": 916
+    },
+    {
+      "epoch": 0.9703703703703703,
+      "grad_norm": 0.2815945843930539,
+      "learning_rate": 3.761270090160722e-05,
+      "loss": 0.6096,
+      "step": 917
+    },
+    {
+      "epoch": 0.9714285714285714,
+      "grad_norm": 0.27333650578935825,
+      "learning_rate": 3.759310074480596e-05,
+      "loss": 0.5376,
+      "step": 918
+    },
+    {
+      "epoch": 0.9724867724867725,
+      "grad_norm": 0.2982346639473295,
+      "learning_rate": 3.7573500588004704e-05,
+      "loss": 0.452,
+      "step": 919
+    },
+    {
+      "epoch": 0.9735449735449735,
+      "grad_norm": 0.2677922848009343,
+      "learning_rate": 3.755390043120345e-05,
+      "loss": 0.4959,
+      "step": 920
+    },
+    {
+      "epoch": 0.9746031746031746,
+      "grad_norm": 0.2598899505130424,
+      "learning_rate": 3.7534300274402196e-05,
+      "loss": 0.5954,
+      "step": 921
+    },
+    {
+      "epoch": 0.9756613756613757,
+      "grad_norm": 0.29688775573892967,
+      "learning_rate": 3.7514700117600945e-05,
+      "loss": 0.4943,
+      "step": 922
+    },
+    {
+      "epoch": 0.9767195767195768,
+      "grad_norm": 0.21278741966910175,
+      "learning_rate": 3.749509996079969e-05,
+      "loss": 0.469,
+      "step": 923
+    },
+    {
+      "epoch": 0.9777777777777777,
+      "grad_norm": 0.27458087883654253,
+      "learning_rate": 3.747549980399843e-05,
+      "loss": 0.5852,
+      "step": 924
+    },
+    {
+      "epoch": 0.9788359788359788,
+      "grad_norm": 0.26092414227041094,
+      "learning_rate": 3.745589964719718e-05,
+      "loss": 0.5247,
+      "step": 925
+    },
+    {
+      "epoch": 0.9798941798941799,
+      "grad_norm": 0.22726587133642012,
+      "learning_rate": 3.743629949039593e-05,
+      "loss": 0.5845,
+      "step": 926
+    },
+    {
+      "epoch": 0.9809523809523809,
+      "grad_norm": 0.24328946071436178,
+      "learning_rate": 3.741669933359467e-05,
+      "loss": 0.5947,
+      "step": 927
+    },
+    {
+      "epoch": 0.982010582010582,
+      "grad_norm": 0.25227625801156794,
+      "learning_rate": 3.7397099176793413e-05,
+      "loss": 0.5245,
+      "step": 928
+    },
+    {
+      "epoch": 0.9830687830687831,
+      "grad_norm": 1.8750127781404309,
+      "learning_rate": 3.737749901999216e-05,
+      "loss": 0.5496,
+      "step": 929
+    },
+    {
+      "epoch": 0.9841269841269841,
+      "grad_norm": 0.2690191369601627,
+      "learning_rate": 3.735789886319091e-05,
+      "loss": 0.5706,
+      "step": 930
+    },
+    {
+      "epoch": 0.9851851851851852,
+      "grad_norm": 0.23072633191225864,
+      "learning_rate": 3.7338298706389654e-05,
+      "loss": 0.5585,
+      "step": 931
+    },
+    {
+      "epoch": 0.9862433862433863,
+      "grad_norm": 0.24321911538830931,
+      "learning_rate": 3.73186985495884e-05,
+      "loss": 0.5314,
+      "step": 932
+    },
+    {
+      "epoch": 0.9873015873015873,
+      "grad_norm": 0.2265652217349377,
+      "learning_rate": 3.7299098392787146e-05,
+      "loss": 0.5157,
+      "step": 933
+    },
+    {
+      "epoch": 0.9883597883597883,
+      "grad_norm": 0.2798879913889636,
+      "learning_rate": 3.727949823598589e-05,
+      "loss": 0.527,
+      "step": 934
+    },
+    {
+      "epoch": 0.9894179894179894,
+      "grad_norm": 0.3392043127421207,
+      "learning_rate": 3.725989807918464e-05,
+      "loss": 0.5649,
+      "step": 935
+    },
+    {
+      "epoch": 0.9904761904761905,
+      "grad_norm": 0.46994870334965927,
+      "learning_rate": 3.724029792238338e-05,
+      "loss": 0.5019,
+      "step": 936
+    },
+    {
+      "epoch": 0.9915343915343915,
+      "grad_norm": 0.31670821542256694,
+      "learning_rate": 3.722069776558212e-05,
+      "loss": 0.4984,
+      "step": 937
+    },
+    {
+      "epoch": 0.9925925925925926,
+      "grad_norm": 0.30042583096668124,
+      "learning_rate": 3.720109760878087e-05,
+      "loss": 0.443,
+      "step": 938
+    },
+    {
+      "epoch": 0.9936507936507937,
+      "grad_norm": 0.26554660858783213,
+      "learning_rate": 3.718149745197962e-05,
+      "loss": 0.5143,
+      "step": 939
+    },
+    {
+      "epoch": 0.9947089947089947,
+      "grad_norm": 0.31047007588956205,
+      "learning_rate": 3.7161897295178364e-05,
+      "loss": 0.5113,
+      "step": 940
+    },
+    {
+      "epoch": 0.9957671957671957,
+      "grad_norm": 0.32328394275671823,
+      "learning_rate": 3.7142297138377106e-05,
+      "loss": 0.5627,
+      "step": 941
+    },
+    {
+      "epoch": 0.9968253968253968,
+      "grad_norm": 0.30050781187053605,
+      "learning_rate": 3.7122696981575856e-05,
+      "loss": 0.4492,
+      "step": 942
+    },
+    {
+      "epoch": 0.9978835978835979,
+      "grad_norm": 0.2883856032205587,
+      "learning_rate": 3.7103096824774605e-05,
+      "loss": 0.573,
+      "step": 943
+    },
+    {
+      "epoch": 0.9989417989417989,
+      "grad_norm": 0.3095113640917161,
+      "learning_rate": 3.708349666797335e-05,
+      "loss": 0.5471,
+      "step": 944
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.2740280000061703,
+      "learning_rate": 3.706389651117209e-05,
+      "loss": 0.5048,
+      "step": 945
+    },
+    {
+      "epoch": 1.001058201058201,
+      "grad_norm": 0.3430281338192382,
+      "learning_rate": 3.704429635437084e-05,
+      "loss": 0.484,
+      "step": 946
+    },
+    {
+      "epoch": 1.0021164021164022,
+      "grad_norm": 0.5765855281602044,
+      "learning_rate": 3.702469619756958e-05,
+      "loss": 0.421,
+      "step": 947
+    },
+    {
+      "epoch": 1.0031746031746032,
+      "grad_norm": 0.32549789453659606,
+      "learning_rate": 3.700509604076833e-05,
+      "loss": 0.5094,
+      "step": 948
+    },
+    {
+      "epoch": 1.0042328042328041,
+      "grad_norm": 0.27167110880361595,
+      "learning_rate": 3.698549588396707e-05,
+      "loss": 0.4292,
+      "step": 949
+    },
+    {
+      "epoch": 1.0052910052910053,
+      "grad_norm": 0.2634410312734115,
+      "learning_rate": 3.6965895727165816e-05,
+      "loss": 0.4056,
+      "step": 950
+    },
+    {
+      "epoch": 1.0063492063492063,
+      "grad_norm": 0.290932789089174,
+      "learning_rate": 3.6946295570364565e-05,
+      "loss": 0.4966,
+      "step": 951
+    },
+    {
+      "epoch": 1.0074074074074073,
+      "grad_norm": 0.277649198156178,
+      "learning_rate": 3.6926695413563314e-05,
+      "loss": 0.5468,
+      "step": 952
+    },
+    {
+      "epoch": 1.0084656084656085,
+      "grad_norm": 0.2871264585009383,
+      "learning_rate": 3.690709525676205e-05,
+      "loss": 0.4619,
+      "step": 953
+    },
+    {
+      "epoch": 1.0095238095238095,
+      "grad_norm": 0.26922960005310304,
+      "learning_rate": 3.68874950999608e-05,
+      "loss": 0.4126,
+      "step": 954
+    },
+    {
+      "epoch": 1.0105820105820107,
+      "grad_norm": 0.2630654741035761,
+      "learning_rate": 3.686789494315955e-05,
+      "loss": 0.4967,
+      "step": 955
+    },
+    {
+      "epoch": 1.0116402116402117,
+      "grad_norm": 0.30760949861741693,
+      "learning_rate": 3.68482947863583e-05,
+      "loss": 0.3988,
+      "step": 956
+    },
+    {
+      "epoch": 1.0126984126984127,
+      "grad_norm": 0.30191598263981867,
+      "learning_rate": 3.6828694629557034e-05,
+      "loss": 0.4891,
+      "step": 957
+    },
+    {
+      "epoch": 1.0137566137566139,
+      "grad_norm": 0.24918234554246604,
+      "learning_rate": 3.680909447275578e-05,
+      "loss": 0.4265,
+      "step": 958
+    },
+    {
+      "epoch": 1.0148148148148148,
+      "grad_norm": 0.25444420109426397,
+      "learning_rate": 3.678949431595453e-05,
+      "loss": 0.4755,
+      "step": 959
+    },
+    {
+      "epoch": 1.0158730158730158,
+      "grad_norm": 0.2704029549763871,
+      "learning_rate": 3.6769894159153275e-05,
+      "loss": 0.4622,
+      "step": 960
+    },
+    {
+      "epoch": 1.016931216931217,
+      "grad_norm": 0.25814705258519083,
+      "learning_rate": 3.675029400235202e-05,
+      "loss": 0.45,
+      "step": 961
+    },
+    {
+      "epoch": 1.017989417989418,
+      "grad_norm": 0.24396615670006316,
+      "learning_rate": 3.6730693845550766e-05,
+      "loss": 0.4621,
+      "step": 962
+    },
+    {
+      "epoch": 1.019047619047619,
+      "grad_norm": 0.24613409465997452,
+      "learning_rate": 3.671109368874951e-05,
+      "loss": 0.3997,
+      "step": 963
+    },
+    {
+      "epoch": 1.0201058201058202,
+      "grad_norm": 0.2446865998835932,
+      "learning_rate": 3.669149353194826e-05,
+      "loss": 0.4312,
+      "step": 964
+    },
+    {
+      "epoch": 1.0211640211640212,
+      "grad_norm": 0.24064127384639672,
+      "learning_rate": 3.6671893375147e-05,
+      "loss": 0.4879,
+      "step": 965
+    },
+    {
+      "epoch": 1.0222222222222221,
+      "grad_norm": 0.24643135413567363,
+      "learning_rate": 3.665229321834575e-05,
+      "loss": 0.4811,
+      "step": 966
+    },
+    {
+      "epoch": 1.0232804232804233,
+      "grad_norm": 0.3002457243261394,
+      "learning_rate": 3.663269306154449e-05,
+      "loss": 0.5209,
+      "step": 967
+    },
+    {
+      "epoch": 1.0243386243386243,
+      "grad_norm": 0.22426207621628652,
+      "learning_rate": 3.661309290474324e-05,
+      "loss": 0.4123,
+      "step": 968
+    },
+    {
+      "epoch": 1.0253968253968253,
+      "grad_norm": 0.24667559321374757,
+      "learning_rate": 3.659349274794199e-05,
+      "loss": 0.462,
+      "step": 969
+    },
+    {
+      "epoch": 1.0264550264550265,
+      "grad_norm": 0.21521332687361075,
+      "learning_rate": 3.6573892591140727e-05,
+      "loss": 0.3874,
+      "step": 970
+    },
+    {
+      "epoch": 1.0275132275132275,
+      "grad_norm": 0.21371142578879992,
+      "learning_rate": 3.6554292434339476e-05,
+      "loss": 0.4216,
+      "step": 971
+    },
+    {
+      "epoch": 1.0285714285714285,
+      "grad_norm": 0.2457752256589608,
+      "learning_rate": 3.6534692277538225e-05,
+      "loss": 0.4751,
+      "step": 972
+    },
+    {
+      "epoch": 1.0296296296296297,
+      "grad_norm": 0.23036724262890812,
+      "learning_rate": 3.651509212073697e-05,
+      "loss": 0.4582,
+      "step": 973
+    },
+    {
+      "epoch": 1.0306878306878307,
+      "grad_norm": 0.2398481443361909,
+      "learning_rate": 3.649549196393571e-05,
+      "loss": 0.4908,
+      "step": 974
+    },
+    {
+      "epoch": 1.0317460317460316,
+      "grad_norm": 0.2191458862749879,
+      "learning_rate": 3.647589180713446e-05,
+      "loss": 0.3675,
+      "step": 975
+    },
+    {
+      "epoch": 1.0328042328042328,
+      "grad_norm": 0.24624710147713,
+      "learning_rate": 3.64562916503332e-05,
+      "loss": 0.4168,
+      "step": 976
+    },
+    {
+      "epoch": 1.0338624338624338,
+      "grad_norm": 0.2505135808681229,
+      "learning_rate": 3.643669149353195e-05,
+      "loss": 0.4317,
+      "step": 977
+    },
+    {
+      "epoch": 1.034920634920635,
+      "grad_norm": 0.23347770709457338,
+      "learning_rate": 3.6417091336730694e-05,
+      "loss": 0.4385,
+      "step": 978
+    },
+    {
+      "epoch": 1.035978835978836,
+      "grad_norm": 0.228128885877278,
+      "learning_rate": 3.639749117992944e-05,
+      "loss": 0.4258,
+      "step": 979
+    },
+    {
+      "epoch": 1.037037037037037,
+      "grad_norm": 0.2629471686081972,
+      "learning_rate": 3.6377891023128185e-05,
+      "loss": 0.4682,
+      "step": 980
+    },
+    {
+      "epoch": 1.0380952380952382,
+      "grad_norm": 0.2649161639864007,
+      "learning_rate": 3.6358290866326935e-05,
+      "loss": 0.4061,
+      "step": 981
+    },
+    {
+      "epoch": 1.0391534391534392,
+      "grad_norm": 0.23269995847151623,
+      "learning_rate": 3.633869070952568e-05,
+      "loss": 0.3948,
+      "step": 982
+    },
+    {
+      "epoch": 1.0402116402116401,
+      "grad_norm": 0.3069159899042451,
+      "learning_rate": 3.631909055272442e-05,
+      "loss": 0.4964,
+      "step": 983
+    },
+    {
+      "epoch": 1.0412698412698413,
+      "grad_norm": 0.28329260505792503,
+      "learning_rate": 3.629949039592317e-05,
+      "loss": 0.446,
+      "step": 984
+    },
+    {
+      "epoch": 1.0423280423280423,
+      "grad_norm": 0.22085634036978702,
+      "learning_rate": 3.627989023912192e-05,
+      "loss": 0.397,
+      "step": 985
+    },
+    {
+      "epoch": 1.0433862433862433,
+      "grad_norm": 0.2533209615376046,
+      "learning_rate": 3.626029008232066e-05,
+      "loss": 0.4478,
+      "step": 986
+    },
+    {
+      "epoch": 1.0444444444444445,
+      "grad_norm": 0.24209341289293657,
+      "learning_rate": 3.62406899255194e-05,
+      "loss": 0.4514,
+      "step": 987
+    },
+    {
+      "epoch": 1.0455026455026455,
+      "grad_norm": 0.22467546448533063,
+      "learning_rate": 3.622108976871815e-05,
+      "loss": 0.4335,
+      "step": 988
+    },
+    {
+      "epoch": 1.0465608465608465,
+      "grad_norm": 0.21049472574558023,
+      "learning_rate": 3.6201489611916895e-05,
+      "loss": 0.408,
+      "step": 989
+    },
+    {
+      "epoch": 1.0476190476190477,
+      "grad_norm": 0.27894693629024636,
+      "learning_rate": 3.6181889455115644e-05,
+      "loss": 0.4497,
+      "step": 990
+    },
+    {
+      "epoch": 1.0486772486772487,
+      "grad_norm": 0.2569069796077995,
+      "learning_rate": 3.6162289298314387e-05,
+      "loss": 0.4601,
+      "step": 991
+    },
+    {
+      "epoch": 1.0497354497354496,
+      "grad_norm": 0.2284829852995991,
+      "learning_rate": 3.6142689141513136e-05,
+      "loss": 0.3953,
+      "step": 992
+    },
+    {
+      "epoch": 1.0507936507936508,
+      "grad_norm": 0.29209053034281596,
+      "learning_rate": 3.612308898471188e-05,
+      "loss": 0.5659,
+      "step": 993
+    },
+    {
+      "epoch": 1.0518518518518518,
+      "grad_norm": 0.3674972112779229,
+      "learning_rate": 3.610348882791063e-05,
+      "loss": 0.5056,
+      "step": 994
+    },
+    {
+      "epoch": 1.052910052910053,
+      "grad_norm": 0.24428585787867374,
+      "learning_rate": 3.608388867110937e-05,
+      "loss": 0.4746,
+      "step": 995
+    },
+    {
+      "epoch": 1.053968253968254,
+      "grad_norm": 0.24244488023201471,
+      "learning_rate": 3.606428851430811e-05,
+      "loss": 0.4863,
+      "step": 996
+    },
+    {
+      "epoch": 1.055026455026455,
+      "grad_norm": 0.2261470911889456,
+      "learning_rate": 3.604468835750686e-05,
+      "loss": 0.449,
+      "step": 997
+    },
+    {
+      "epoch": 1.0560846560846562,
+      "grad_norm": 0.23100103297173594,
+      "learning_rate": 3.602508820070561e-05,
+      "loss": 0.4325,
+      "step": 998
+    },
+    {
+      "epoch": 1.0571428571428572,
+      "grad_norm": 0.2383765185611898,
+      "learning_rate": 3.6005488043904354e-05,
+      "loss": 0.4625,
+      "step": 999
+    },
+    {
+      "epoch": 1.0582010582010581,
+      "grad_norm": 0.22794426514581065,
+      "learning_rate": 3.5985887887103096e-05,
+      "loss": 0.4055,
+      "step": 1000
+    },
+    {
+      "epoch": 1.0592592592592593,
+      "grad_norm": 0.24878598922435188,
+      "learning_rate": 3.5966287730301845e-05,
+      "loss": 0.4176,
+      "step": 1001
+    },
+    {
+      "epoch": 1.0603174603174603,
+      "grad_norm": 0.24946112914172103,
+      "learning_rate": 3.594668757350059e-05,
+      "loss": 0.4949,
+      "step": 1002
+    },
+    {
+      "epoch": 1.0613756613756613,
+      "grad_norm": 0.24129845809349035,
+      "learning_rate": 3.592708741669934e-05,
+      "loss": 0.3912,
+      "step": 1003
+    },
+    {
+      "epoch": 1.0624338624338625,
+      "grad_norm": 0.24281750665894428,
+      "learning_rate": 3.590748725989808e-05,
+      "loss": 0.4568,
+      "step": 1004
+    },
+    {
+      "epoch": 1.0634920634920635,
+      "grad_norm": 0.20760278598384954,
+      "learning_rate": 3.588788710309683e-05,
+      "loss": 0.3948,
+      "step": 1005
+    },
+    {
+      "epoch": 1.0645502645502645,
+      "grad_norm": 0.23431721866007746,
+      "learning_rate": 3.586828694629557e-05,
+      "loss": 0.424,
+      "step": 1006
+    },
+    {
+      "epoch": 1.0656084656084657,
+      "grad_norm": 0.2471701897811236,
+      "learning_rate": 3.584868678949432e-05,
+      "loss": 0.4256,
+      "step": 1007
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "grad_norm": 0.2302012676934357,
+      "learning_rate": 3.582908663269306e-05,
+      "loss": 0.4277,
+      "step": 1008
+    },
+    {
+      "epoch": 1.0677248677248676,
+      "grad_norm": 0.22452613295126128,
+      "learning_rate": 3.5809486475891806e-05,
+      "loss": 0.4465,
+      "step": 1009
+    },
+    {
+      "epoch": 1.0687830687830688,
+      "grad_norm": 0.2257507725040729,
+      "learning_rate": 3.5789886319090555e-05,
+      "loss": 0.4698,
+      "step": 1010
+    },
+    {
+      "epoch": 1.0698412698412698,
+      "grad_norm": 0.24453737754436786,
+      "learning_rate": 3.5770286162289304e-05,
+      "loss": 0.4469,
+      "step": 1011
+    },
+    {
+      "epoch": 1.0708994708994708,
+      "grad_norm": 0.24956877063093863,
+      "learning_rate": 3.5750686005488047e-05,
+      "loss": 0.4626,
+      "step": 1012
+    },
+    {
+      "epoch": 1.071957671957672,
+      "grad_norm": 0.2600831587111751,
+      "learning_rate": 3.573108584868679e-05,
+      "loss": 0.4537,
+      "step": 1013
+    },
+    {
+      "epoch": 1.073015873015873,
+      "grad_norm": 4.587346913530301,
+      "learning_rate": 3.571148569188554e-05,
+      "loss": 0.558,
+      "step": 1014
+    },
+    {
+      "epoch": 1.074074074074074,
+      "grad_norm": 0.29359742422528,
+      "learning_rate": 3.569188553508429e-05,
+      "loss": 0.4682,
+      "step": 1015
+    },
+    {
+      "epoch": 1.0751322751322752,
+      "grad_norm": 0.31833732588739005,
+      "learning_rate": 3.567228537828303e-05,
+      "loss": 0.4743,
+      "step": 1016
+    },
+    {
+      "epoch": 1.0761904761904761,
+      "grad_norm": 0.2553867628775377,
+      "learning_rate": 3.565268522148177e-05,
+      "loss": 0.551,
+      "step": 1017
+    },
+    {
+      "epoch": 1.0772486772486773,
+      "grad_norm": 0.24879754575790247,
+      "learning_rate": 3.563308506468052e-05,
+      "loss": 0.4524,
+      "step": 1018
+    },
+    {
+      "epoch": 1.0783068783068783,
+      "grad_norm": 0.28766655345957765,
+      "learning_rate": 3.5613484907879264e-05,
+      "loss": 0.4384,
+      "step": 1019
+    },
+    {
+      "epoch": 1.0793650793650793,
+      "grad_norm": 0.41507355236002946,
+      "learning_rate": 3.5593884751078014e-05,
+      "loss": 0.4206,
+      "step": 1020
+    },
+    {
+      "epoch": 1.0804232804232805,
+      "grad_norm": 0.23507148230554162,
+      "learning_rate": 3.5574284594276756e-05,
+      "loss": 0.4353,
+      "step": 1021
+    },
+    {
+      "epoch": 1.0814814814814815,
+      "grad_norm": 0.2622046524529905,
+      "learning_rate": 3.55546844374755e-05,
+      "loss": 0.487,
+      "step": 1022
+    },
+    {
+      "epoch": 1.0825396825396825,
+      "grad_norm": 0.2963869501406697,
+      "learning_rate": 3.553508428067425e-05,
+      "loss": 0.5234,
+      "step": 1023
+    },
+    {
+      "epoch": 1.0835978835978837,
+      "grad_norm": 0.22883220209770097,
+      "learning_rate": 3.5515484123873e-05,
+      "loss": 0.4493,
+      "step": 1024
+    },
+    {
+      "epoch": 1.0846560846560847,
+      "grad_norm": 0.2569611944727975,
+      "learning_rate": 3.549588396707173e-05,
+      "loss": 0.466,
+      "step": 1025
+    },
+    {
+      "epoch": 1.0857142857142856,
+      "grad_norm": 0.254023601263868,
+      "learning_rate": 3.547628381027048e-05,
+      "loss": 0.459,
+      "step": 1026
+    },
+    {
+      "epoch": 1.0867724867724868,
+      "grad_norm": 0.23407024600575801,
+      "learning_rate": 3.545668365346923e-05,
+      "loss": 0.45,
+      "step": 1027
+    },
+    {
+      "epoch": 1.0878306878306878,
+      "grad_norm": 0.2599527517255244,
+      "learning_rate": 3.543708349666798e-05,
+      "loss": 0.4115,
+      "step": 1028
+    },
+    {
+      "epoch": 1.0888888888888888,
+      "grad_norm": 0.22311611567918718,
+      "learning_rate": 3.5417483339866716e-05,
+      "loss": 0.3979,
+      "step": 1029
+    },
+    {
+      "epoch": 1.08994708994709,
+      "grad_norm": 0.2299712856828212,
+      "learning_rate": 3.5397883183065466e-05,
+      "loss": 0.3694,
+      "step": 1030
+    },
+    {
+      "epoch": 1.091005291005291,
+      "grad_norm": 0.21532417571199142,
+      "learning_rate": 3.5378283026264215e-05,
+      "loss": 0.3551,
+      "step": 1031
+    },
+    {
+      "epoch": 1.0920634920634922,
+      "grad_norm": 0.23843932815464314,
+      "learning_rate": 3.535868286946296e-05,
+      "loss": 0.4513,
+      "step": 1032
+    },
+    {
+      "epoch": 1.0931216931216932,
+      "grad_norm": 0.21531850573097505,
+      "learning_rate": 3.53390827126617e-05,
+      "loss": 0.3931,
+      "step": 1033
+    },
+    {
+      "epoch": 1.0941798941798941,
+      "grad_norm": 0.3068198759456981,
+      "learning_rate": 3.531948255586045e-05,
+      "loss": 0.4891,
+      "step": 1034
+    },
+    {
+      "epoch": 1.0952380952380953,
+      "grad_norm": 0.23836601091918505,
+      "learning_rate": 3.529988239905919e-05,
+      "loss": 0.439,
+      "step": 1035
+    },
+    {
+      "epoch": 1.0962962962962963,
+      "grad_norm": 0.2437498779338309,
+      "learning_rate": 3.528028224225794e-05,
+      "loss": 0.4149,
+      "step": 1036
+    },
+    {
+      "epoch": 1.0973544973544973,
+      "grad_norm": 0.27351894671972754,
+      "learning_rate": 3.526068208545668e-05,
+      "loss": 0.4416,
+      "step": 1037
+    },
+    {
+      "epoch": 1.0984126984126985,
+      "grad_norm": 0.27770230078648367,
+      "learning_rate": 3.5241081928655426e-05,
+      "loss": 0.4884,
+      "step": 1038
+    },
+    {
+      "epoch": 1.0994708994708995,
+      "grad_norm": 0.25718315589569163,
+      "learning_rate": 3.5221481771854175e-05,
+      "loss": 0.4578,
+      "step": 1039
+    },
+    {
+      "epoch": 1.1005291005291005,
+      "grad_norm": 0.2594326299087,
+      "learning_rate": 3.5201881615052924e-05,
+      "loss": 0.4864,
+      "step": 1040
+    },
+    {
+      "epoch": 1.1015873015873017,
+      "grad_norm": 0.2472630187462841,
+      "learning_rate": 3.5182281458251674e-05,
+      "loss": 0.4451,
+      "step": 1041
+    },
+    {
+      "epoch": 1.1026455026455027,
+      "grad_norm": 0.25781560522935376,
+      "learning_rate": 3.516268130145041e-05,
+      "loss": 0.4182,
+      "step": 1042
+    },
+    {
+      "epoch": 1.1037037037037036,
+      "grad_norm": 0.2535547232578919,
+      "learning_rate": 3.514308114464916e-05,
+      "loss": 0.4749,
+      "step": 1043
+    },
+    {
+      "epoch": 1.1047619047619048,
+      "grad_norm": 0.254021416442326,
+      "learning_rate": 3.512348098784791e-05,
+      "loss": 0.471,
+      "step": 1044
+    },
+    {
+      "epoch": 1.1058201058201058,
+      "grad_norm": 0.28102597311344235,
+      "learning_rate": 3.510388083104665e-05,
+      "loss": 0.4615,
+      "step": 1045
+    },
+    {
+      "epoch": 1.1068783068783068,
+      "grad_norm": 0.23113213821636622,
+      "learning_rate": 3.508428067424539e-05,
+      "loss": 0.4147,
+      "step": 1046
+    },
+    {
+      "epoch": 1.107936507936508,
+      "grad_norm": 0.2474923745981431,
+      "learning_rate": 3.506468051744414e-05,
+      "loss": 0.4437,
+      "step": 1047
+    },
+    {
+      "epoch": 1.108994708994709,
+      "grad_norm": 0.24830934850829475,
+      "learning_rate": 3.5045080360642884e-05,
+      "loss": 0.4376,
+      "step": 1048
+    },
+    {
+      "epoch": 1.11005291005291,
+      "grad_norm": 0.47806726597633753,
+      "learning_rate": 3.5025480203841634e-05,
+      "loss": 0.4324,
+      "step": 1049
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 0.32025751473392294,
+      "learning_rate": 3.5005880047040376e-05,
+      "loss": 0.5246,
+      "step": 1050
+    },
+    {
+      "epoch": 1.1121693121693121,
+      "grad_norm": 0.28097198860260764,
+      "learning_rate": 3.498627989023912e-05,
+      "loss": 0.4598,
+      "step": 1051
+    },
+    {
+      "epoch": 1.1132275132275131,
+      "grad_norm": 0.254947622562914,
+      "learning_rate": 3.496667973343787e-05,
+      "loss": 0.4222,
+      "step": 1052
+    },
+    {
+      "epoch": 1.1142857142857143,
+      "grad_norm": 0.3349593604089576,
+      "learning_rate": 3.494707957663662e-05,
+      "loss": 0.4155,
+      "step": 1053
+    },
+    {
+      "epoch": 1.1153439153439153,
+      "grad_norm": 0.26874038889591473,
+      "learning_rate": 3.492747941983536e-05,
+      "loss": 0.4375,
+      "step": 1054
+    },
+    {
+      "epoch": 1.1164021164021163,
+      "grad_norm": 0.26633236590448656,
+      "learning_rate": 3.49078792630341e-05,
+      "loss": 0.4267,
+      "step": 1055
+    },
+    {
+      "epoch": 1.1174603174603175,
+      "grad_norm": 0.2884717153428804,
+      "learning_rate": 3.488827910623285e-05,
+      "loss": 0.4257,
+      "step": 1056
+    },
+    {
+      "epoch": 1.1185185185185185,
+      "grad_norm": 0.4452379521735504,
+      "learning_rate": 3.48686789494316e-05,
+      "loss": 0.4332,
+      "step": 1057
+    },
+    {
+      "epoch": 1.1195767195767197,
+      "grad_norm": 0.7702289596361585,
+      "learning_rate": 3.484907879263034e-05,
+      "loss": 0.4349,
+      "step": 1058
+    },
+    {
+      "epoch": 1.1206349206349207,
+      "grad_norm": 0.3656397372057208,
+      "learning_rate": 3.4829478635829086e-05,
+      "loss": 0.4238,
+      "step": 1059
+    },
+    {
+      "epoch": 1.1216931216931216,
+      "grad_norm": 0.4866179217310575,
+      "learning_rate": 3.4809878479027835e-05,
+      "loss": 0.4216,
+      "step": 1060
+    },
+    {
+      "epoch": 1.1227513227513228,
+      "grad_norm": 0.33474959074863353,
+      "learning_rate": 3.479027832222658e-05,
+      "loss": 0.4418,
+      "step": 1061
+    },
+    {
+      "epoch": 1.1238095238095238,
+      "grad_norm": 0.27607814779984247,
+      "learning_rate": 3.477067816542533e-05,
+      "loss": 0.4364,
+      "step": 1062
+    },
+    {
+      "epoch": 1.1248677248677248,
+      "grad_norm": 0.3407943035819788,
+      "learning_rate": 3.475107800862407e-05,
+      "loss": 0.4665,
+      "step": 1063
+    },
+    {
+      "epoch": 1.125925925925926,
+      "grad_norm": 0.33884649086749913,
+      "learning_rate": 3.473147785182282e-05,
+      "loss": 0.4543,
+      "step": 1064
+    },
+    {
+      "epoch": 1.126984126984127,
+      "grad_norm": 0.29045801993635223,
+      "learning_rate": 3.471187769502156e-05,
+      "loss": 0.4612,
+      "step": 1065
+    },
+    {
+      "epoch": 1.128042328042328,
+      "grad_norm": 0.3424125515301145,
+      "learning_rate": 3.469227753822031e-05,
+      "loss": 0.5199,
+      "step": 1066
+    },
+    {
+      "epoch": 1.1291005291005292,
+      "grad_norm": 0.29631324600571024,
+      "learning_rate": 3.467267738141905e-05,
+      "loss": 0.4505,
+      "step": 1067
+    },
+    {
+      "epoch": 1.1301587301587301,
+      "grad_norm": 0.2254979652739058,
+      "learning_rate": 3.4653077224617795e-05,
+      "loss": 0.4902,
+      "step": 1068
+    },
+    {
+      "epoch": 1.1312169312169311,
+      "grad_norm": 0.29453491815047816,
+      "learning_rate": 3.4633477067816544e-05,
+      "loss": 0.4537,
+      "step": 1069
+    },
+    {
+      "epoch": 1.1322751322751323,
+      "grad_norm": 0.24120272460563255,
+      "learning_rate": 3.4613876911015294e-05,
+      "loss": 0.4156,
+      "step": 1070
+    },
+    {
+      "epoch": 1.1333333333333333,
+      "grad_norm": 0.2417667303754181,
+      "learning_rate": 3.4594276754214036e-05,
+      "loss": 0.464,
+      "step": 1071
+    },
+    {
+      "epoch": 1.1343915343915345,
+      "grad_norm": 0.24479296740003978,
+      "learning_rate": 3.457467659741278e-05,
+      "loss": 0.425,
+      "step": 1072
+    },
+    {
+      "epoch": 1.1354497354497355,
+      "grad_norm": 0.2606469824823141,
+      "learning_rate": 3.455507644061153e-05,
+      "loss": 0.4171,
+      "step": 1073
+    },
+    {
+      "epoch": 1.1365079365079365,
+      "grad_norm": 1.160342854664124,
+      "learning_rate": 3.453547628381027e-05,
+      "loss": 0.4907,
+      "step": 1074
+    },
+    {
+      "epoch": 1.1375661375661377,
+      "grad_norm": 0.27475204211746457,
+      "learning_rate": 3.451587612700902e-05,
+      "loss": 0.4412,
+      "step": 1075
+    },
+    {
+      "epoch": 1.1386243386243386,
+      "grad_norm": 0.27882185123707176,
+      "learning_rate": 3.449627597020776e-05,
+      "loss": 0.4304,
+      "step": 1076
+    },
+    {
+      "epoch": 1.1396825396825396,
+      "grad_norm": 0.24769703151282585,
+      "learning_rate": 3.447667581340651e-05,
+      "loss": 0.4258,
+      "step": 1077
+    },
+    {
+      "epoch": 1.1407407407407408,
+      "grad_norm": 0.2624061999879434,
+      "learning_rate": 3.4457075656605254e-05,
+      "loss": 0.4202,
+      "step": 1078
+    },
+    {
+      "epoch": 1.1417989417989418,
+      "grad_norm": 0.23516866564878172,
+      "learning_rate": 3.4437475499804e-05,
+      "loss": 0.4201,
+      "step": 1079
+    },
+    {
+      "epoch": 1.1428571428571428,
+      "grad_norm": 0.2591076428810598,
+      "learning_rate": 3.4417875343002746e-05,
+      "loss": 0.4179,
+      "step": 1080
+    },
+    {
+      "epoch": 1.143915343915344,
+      "grad_norm": 0.245392393657956,
+      "learning_rate": 3.439827518620149e-05,
+      "loss": 0.3815,
+      "step": 1081
+    },
+    {
+      "epoch": 1.144973544973545,
+      "grad_norm": 0.23935170354064206,
+      "learning_rate": 3.437867502940024e-05,
+      "loss": 0.4652,
+      "step": 1082
+    },
+    {
+      "epoch": 1.146031746031746,
+      "grad_norm": 0.24157446804740304,
+      "learning_rate": 3.435907487259899e-05,
+      "loss": 0.4123,
+      "step": 1083
+    },
+    {
+      "epoch": 1.1470899470899472,
+      "grad_norm": 0.2997924271290018,
+      "learning_rate": 3.433947471579773e-05,
+      "loss": 0.4711,
+      "step": 1084
+    },
+    {
+      "epoch": 1.1481481481481481,
+      "grad_norm": 0.2513216731368017,
+      "learning_rate": 3.431987455899647e-05,
+      "loss": 0.4065,
+      "step": 1085
+    },
+    {
+      "epoch": 1.1492063492063491,
+      "grad_norm": 0.2522352354653566,
+      "learning_rate": 3.430027440219522e-05,
+      "loss": 0.4234,
+      "step": 1086
+    },
+    {
+      "epoch": 1.1502645502645503,
+      "grad_norm": 0.29099731437781645,
+      "learning_rate": 3.4280674245393963e-05,
+      "loss": 0.4617,
+      "step": 1087
+    },
+    {
+      "epoch": 1.1513227513227513,
+      "grad_norm": 0.2940806275764955,
+      "learning_rate": 3.426107408859271e-05,
+      "loss": 0.4231,
+      "step": 1088
+    },
+    {
+      "epoch": 1.1523809523809523,
+      "grad_norm": 0.27282227795834785,
+      "learning_rate": 3.4241473931791455e-05,
+      "loss": 0.4884,
+      "step": 1089
+    },
+    {
+      "epoch": 1.1534391534391535,
+      "grad_norm": 0.25357226775494324,
+      "learning_rate": 3.4221873774990204e-05,
+      "loss": 0.4817,
+      "step": 1090
+    },
+    {
+      "epoch": 1.1544973544973545,
+      "grad_norm": 0.29654179010898835,
+      "learning_rate": 3.420227361818895e-05,
+      "loss": 0.4057,
+      "step": 1091
+    },
+    {
+      "epoch": 1.1555555555555554,
+      "grad_norm": 0.27272067824064034,
+      "learning_rate": 3.4182673461387696e-05,
+      "loss": 0.4649,
+      "step": 1092
+    },
+    {
+      "epoch": 1.1566137566137566,
+      "grad_norm": 0.25231760876479825,
+      "learning_rate": 3.416307330458644e-05,
+      "loss": 0.4467,
+      "step": 1093
+    },
+    {
+      "epoch": 1.1576719576719576,
+      "grad_norm": 1.4439868836946126,
+      "learning_rate": 3.414347314778518e-05,
+      "loss": 0.5031,
+      "step": 1094
+    },
+    {
+      "epoch": 1.1587301587301586,
+      "grad_norm": 0.2669246549318654,
+      "learning_rate": 3.412387299098393e-05,
+      "loss": 0.4242,
+      "step": 1095
+    },
+    {
+      "epoch": 1.1597883597883598,
+      "grad_norm": 0.4725447349327677,
+      "learning_rate": 3.410427283418268e-05,
+      "loss": 0.4037,
+      "step": 1096
+    },
+    {
+      "epoch": 1.1608465608465608,
+      "grad_norm": 0.24176894320740916,
+      "learning_rate": 3.4084672677381415e-05,
+      "loss": 0.4461,
+      "step": 1097
+    },
+    {
+      "epoch": 1.161904761904762,
+      "grad_norm": 0.2571630439924496,
+      "learning_rate": 3.4065072520580165e-05,
+      "loss": 0.4435,
+      "step": 1098
+    },
+    {
+      "epoch": 1.162962962962963,
+      "grad_norm": 0.25405311821960447,
+      "learning_rate": 3.4045472363778914e-05,
+      "loss": 0.4008,
+      "step": 1099
+    },
+    {
+      "epoch": 1.164021164021164,
+      "grad_norm": 0.2360510636639726,
+      "learning_rate": 3.4025872206977656e-05,
+      "loss": 0.4418,
+      "step": 1100
+    },
+    {
+      "epoch": 1.1650793650793652,
+      "grad_norm": 0.24349573097824267,
+      "learning_rate": 3.40062720501764e-05,
+      "loss": 0.4242,
+      "step": 1101
+    },
+    {
+      "epoch": 1.1661375661375661,
+      "grad_norm": 0.22205185004409997,
+      "learning_rate": 3.398667189337515e-05,
+      "loss": 0.4319,
+      "step": 1102
+    },
+    {
+      "epoch": 1.1671957671957671,
+      "grad_norm": 0.25476677264568914,
+      "learning_rate": 3.39670717365739e-05,
+      "loss": 0.4214,
+      "step": 1103
+    },
+    {
+      "epoch": 1.1682539682539683,
+      "grad_norm": 0.24140105278129537,
+      "learning_rate": 3.394747157977264e-05,
+      "loss": 0.4331,
+      "step": 1104
+    },
+    {
+      "epoch": 1.1693121693121693,
+      "grad_norm": 0.23526559056529114,
+      "learning_rate": 3.392787142297138e-05,
+      "loss": 0.4854,
+      "step": 1105
+    },
+    {
+      "epoch": 1.1703703703703703,
+      "grad_norm": 3.2639568571243096,
+      "learning_rate": 3.390827126617013e-05,
+      "loss": 0.444,
+      "step": 1106
+    },
+    {
+      "epoch": 1.1714285714285715,
+      "grad_norm": 0.29307877776027735,
+      "learning_rate": 3.3888671109368874e-05,
+      "loss": 0.491,
+      "step": 1107
+    },
+    {
+      "epoch": 1.1724867724867725,
+      "grad_norm": 0.25173250808327663,
+      "learning_rate": 3.3869070952567623e-05,
+      "loss": 0.4304,
+      "step": 1108
+    },
+    {
+      "epoch": 1.1735449735449737,
+      "grad_norm": 0.23834729259967638,
+      "learning_rate": 3.3849470795766366e-05,
+      "loss": 0.4242,
+      "step": 1109
+    },
+    {
+      "epoch": 1.1746031746031746,
+      "grad_norm": 0.2820981143799431,
+      "learning_rate": 3.382987063896511e-05,
+      "loss": 0.4531,
+      "step": 1110
+    },
+    {
+      "epoch": 1.1756613756613756,
+      "grad_norm": 0.2684834031217499,
+      "learning_rate": 3.381027048216386e-05,
+      "loss": 0.4423,
+      "step": 1111
+    },
+    {
+      "epoch": 1.1767195767195768,
+      "grad_norm": 0.22060471509608026,
+      "learning_rate": 3.379067032536261e-05,
+      "loss": 0.4567,
+      "step": 1112
+    },
+    {
+      "epoch": 1.1777777777777778,
+      "grad_norm": 0.2938768585311419,
+      "learning_rate": 3.3771070168561356e-05,
+      "loss": 0.5653,
+      "step": 1113
+    },
+    {
+      "epoch": 1.1788359788359788,
+      "grad_norm": 0.2185303188886883,
+      "learning_rate": 3.375147001176009e-05,
+      "loss": 0.4376,
+      "step": 1114
+    },
+    {
+      "epoch": 1.17989417989418,
+      "grad_norm": 0.31115860640653836,
+      "learning_rate": 3.373186985495884e-05,
+      "loss": 0.477,
+      "step": 1115
+    },
+    {
+      "epoch": 1.180952380952381,
+      "grad_norm": 0.2546303170293118,
+      "learning_rate": 3.371226969815759e-05,
+      "loss": 0.4421,
+      "step": 1116
+    },
+    {
+      "epoch": 1.182010582010582,
+      "grad_norm": 0.24401462771526597,
+      "learning_rate": 3.369266954135633e-05,
+      "loss": 0.4158,
+      "step": 1117
+    },
+    {
+      "epoch": 1.1830687830687832,
+      "grad_norm": 0.2371965772098529,
+      "learning_rate": 3.3673069384555075e-05,
+      "loss": 0.4148,
+      "step": 1118
+    },
+    {
+      "epoch": 1.1841269841269841,
+      "grad_norm": 0.25311128441059516,
+      "learning_rate": 3.3653469227753825e-05,
+      "loss": 0.3996,
+      "step": 1119
+    },
+    {
+      "epoch": 1.1851851851851851,
+      "grad_norm": 0.25406846795302634,
+      "learning_rate": 3.363386907095257e-05,
+      "loss": 0.4802,
+      "step": 1120
+    },
+    {
+      "epoch": 1.1862433862433863,
+      "grad_norm": 0.28969576250269813,
+      "learning_rate": 3.3614268914151316e-05,
+      "loss": 0.4434,
+      "step": 1121
+    },
+    {
+      "epoch": 1.1873015873015873,
+      "grad_norm": 0.25113888422786323,
+      "learning_rate": 3.359466875735006e-05,
+      "loss": 0.4693,
+      "step": 1122
+    },
+    {
+      "epoch": 1.1883597883597883,
+      "grad_norm": 0.26242305511278513,
+      "learning_rate": 3.35750686005488e-05,
+      "loss": 0.4408,
+      "step": 1123
+    },
+    {
+      "epoch": 1.1894179894179895,
+      "grad_norm": 0.24851482302114736,
+      "learning_rate": 3.355546844374755e-05,
+      "loss": 0.4855,
+      "step": 1124
+    },
+    {
+      "epoch": 1.1904761904761905,
+      "grad_norm": 0.21264747963193062,
+      "learning_rate": 3.35358682869463e-05,
+      "loss": 0.403,
+      "step": 1125
+    },
+    {
+      "epoch": 1.1915343915343914,
+      "grad_norm": 0.22264008089350337,
+      "learning_rate": 3.351626813014504e-05,
+      "loss": 0.4466,
+      "step": 1126
+    },
+    {
+      "epoch": 1.1925925925925926,
+      "grad_norm": 0.27737325502984783,
+      "learning_rate": 3.3496667973343785e-05,
+      "loss": 0.4601,
+      "step": 1127
+    },
+    {
+      "epoch": 1.1936507936507936,
+      "grad_norm": 3.368075981212778,
+      "learning_rate": 3.3477067816542534e-05,
+      "loss": 0.4871,
+      "step": 1128
+    },
+    {
+      "epoch": 1.1947089947089946,
+      "grad_norm": 0.3148955129252137,
+      "learning_rate": 3.345746765974128e-05,
+      "loss": 0.4603,
+      "step": 1129
+    },
+    {
+      "epoch": 1.1957671957671958,
+      "grad_norm": 0.2533234823076518,
+      "learning_rate": 3.3437867502940026e-05,
+      "loss": 0.4263,
+      "step": 1130
+    },
+    {
+      "epoch": 1.1968253968253968,
+      "grad_norm": 0.25463206395686244,
+      "learning_rate": 3.341826734613877e-05,
+      "loss": 0.452,
+      "step": 1131
+    },
+    {
+      "epoch": 1.1978835978835978,
+      "grad_norm": 0.27879057651843764,
+      "learning_rate": 3.339866718933752e-05,
+      "loss": 0.4882,
+      "step": 1132
+    },
+    {
+      "epoch": 1.198941798941799,
+      "grad_norm": 0.26539022543940605,
+      "learning_rate": 3.337906703253626e-05,
+      "loss": 0.4372,
+      "step": 1133
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.279151992029459,
+      "learning_rate": 3.335946687573501e-05,
+      "loss": 0.4505,
+      "step": 1134
+    },
+    {
+      "epoch": 1.201058201058201,
+      "grad_norm": 0.3036270580949587,
+      "learning_rate": 3.333986671893375e-05,
+      "loss": 0.4974,
+      "step": 1135
+    },
+    {
+      "epoch": 1.2021164021164021,
+      "grad_norm": 0.27528174866789473,
+      "learning_rate": 3.3320266562132494e-05,
+      "loss": 0.4994,
+      "step": 1136
+    },
+    {
+      "epoch": 1.2031746031746031,
+      "grad_norm": 0.29856950983454983,
+      "learning_rate": 3.3300666405331244e-05,
+      "loss": 0.4534,
+      "step": 1137
+    },
+    {
+      "epoch": 1.2042328042328043,
+      "grad_norm": 0.27683008135091847,
+      "learning_rate": 3.328106624852999e-05,
+      "loss": 0.4379,
+      "step": 1138
+    },
+    {
+      "epoch": 1.2052910052910053,
+      "grad_norm": 0.26301228869162596,
+      "learning_rate": 3.3261466091728735e-05,
+      "loss": 0.4369,
+      "step": 1139
+    },
+    {
+      "epoch": 1.2063492063492063,
+      "grad_norm": 0.4106472014462593,
+      "learning_rate": 3.324186593492748e-05,
+      "loss": 0.4226,
+      "step": 1140
+    },
+    {
+      "epoch": 1.2074074074074075,
+      "grad_norm": 0.2843670121313921,
+      "learning_rate": 3.322226577812623e-05,
+      "loss": 0.4895,
+      "step": 1141
+    },
+    {
+      "epoch": 1.2084656084656085,
+      "grad_norm": 0.31274125765332483,
+      "learning_rate": 3.3202665621324976e-05,
+      "loss": 0.5111,
+      "step": 1142
+    },
+    {
+      "epoch": 1.2095238095238094,
+      "grad_norm": 0.34931031172201393,
+      "learning_rate": 3.318306546452372e-05,
+      "loss": 0.4436,
+      "step": 1143
+    },
+    {
+      "epoch": 1.2105820105820106,
+      "grad_norm": 0.23718925178525013,
+      "learning_rate": 3.316346530772246e-05,
+      "loss": 0.3974,
+      "step": 1144
+    },
+    {
+      "epoch": 1.2116402116402116,
+      "grad_norm": 0.2668299284915015,
+      "learning_rate": 3.314386515092121e-05,
+      "loss": 0.4508,
+      "step": 1145
+    },
+    {
+      "epoch": 1.2126984126984126,
+      "grad_norm": 0.2801312984615186,
+      "learning_rate": 3.312426499411995e-05,
+      "loss": 0.4694,
+      "step": 1146
+    },
+    {
+      "epoch": 1.2137566137566138,
+      "grad_norm": 0.24312099943011134,
+      "learning_rate": 3.31046648373187e-05,
+      "loss": 0.4546,
+      "step": 1147
+    },
+    {
+      "epoch": 1.2148148148148148,
+      "grad_norm": 0.25424837283402324,
+      "learning_rate": 3.3085064680517445e-05,
+      "loss": 0.4747,
+      "step": 1148
+    },
+    {
+      "epoch": 1.215873015873016,
+      "grad_norm": 0.2674867566361891,
+      "learning_rate": 3.306546452371619e-05,
+      "loss": 0.4574,
+      "step": 1149
+    },
+    {
+      "epoch": 1.216931216931217,
+      "grad_norm": 0.2627651556600422,
+      "learning_rate": 3.3045864366914937e-05,
+      "loss": 0.4378,
+      "step": 1150
+    },
+    {
+      "epoch": 1.217989417989418,
+      "grad_norm": 0.23716263801602822,
+      "learning_rate": 3.3026264210113686e-05,
+      "loss": 0.4161,
+      "step": 1151
+    },
+    {
+      "epoch": 1.2190476190476192,
+      "grad_norm": 0.3138567642984657,
+      "learning_rate": 3.300666405331243e-05,
+      "loss": 0.4383,
+      "step": 1152
+    },
+    {
+      "epoch": 1.2201058201058201,
+      "grad_norm": 0.276967866343535,
+      "learning_rate": 3.298706389651117e-05,
+      "loss": 0.4737,
+      "step": 1153
+    },
+    {
+      "epoch": 1.2211640211640211,
+      "grad_norm": 0.2607801018967495,
+      "learning_rate": 3.296746373970992e-05,
+      "loss": 0.4962,
+      "step": 1154
+    },
+    {
+      "epoch": 1.2222222222222223,
+      "grad_norm": 0.24637415714449645,
+      "learning_rate": 3.294786358290867e-05,
+      "loss": 0.4259,
+      "step": 1155
+    },
+    {
+      "epoch": 1.2232804232804233,
+      "grad_norm": 0.25635788195924125,
+      "learning_rate": 3.292826342610741e-05,
+      "loss": 0.4851,
+      "step": 1156
+    },
+    {
+      "epoch": 1.2243386243386243,
+      "grad_norm": 21.708465964907735,
+      "learning_rate": 3.2908663269306154e-05,
+      "loss": 0.5324,
+      "step": 1157
+    },
+    {
+      "epoch": 1.2253968253968255,
+      "grad_norm": 0.2480738276020431,
+      "learning_rate": 3.2889063112504904e-05,
+      "loss": 0.4119,
+      "step": 1158
+    },
+    {
+      "epoch": 1.2264550264550265,
+      "grad_norm": 0.2781677513733888,
+      "learning_rate": 3.2869462955703646e-05,
+      "loss": 0.5124,
+      "step": 1159
+    },
+    {
+      "epoch": 1.2275132275132274,
+      "grad_norm": 0.20926116289505092,
+      "learning_rate": 3.2849862798902395e-05,
+      "loss": 0.3981,
+      "step": 1160
+    },
+    {
+      "epoch": 1.2285714285714286,
+      "grad_norm": 0.23499136854110111,
+      "learning_rate": 3.283026264210114e-05,
+      "loss": 0.4563,
+      "step": 1161
+    },
+    {
+      "epoch": 1.2296296296296296,
+      "grad_norm": 0.28418800988255766,
+      "learning_rate": 3.281066248529989e-05,
+      "loss": 0.4683,
+      "step": 1162
+    },
+    {
+      "epoch": 1.2306878306878306,
+      "grad_norm": 0.5732237050386909,
+      "learning_rate": 3.279106232849863e-05,
+      "loss": 0.454,
+      "step": 1163
+    },
+    {
+      "epoch": 1.2317460317460318,
+      "grad_norm": 0.29349759306686635,
+      "learning_rate": 3.277146217169738e-05,
+      "loss": 0.3962,
+      "step": 1164
+    },
+    {
+      "epoch": 1.2328042328042328,
+      "grad_norm": 0.3073242580524234,
+      "learning_rate": 3.275186201489612e-05,
+      "loss": 0.4743,
+      "step": 1165
+    },
+    {
+      "epoch": 1.2338624338624338,
+      "grad_norm": 0.2695714862968926,
+      "learning_rate": 3.2732261858094864e-05,
+      "loss": 0.4684,
+      "step": 1166
+    },
+    {
+      "epoch": 1.234920634920635,
+      "grad_norm": 0.24526233061405303,
+      "learning_rate": 3.271266170129361e-05,
+      "loss": 0.4098,
+      "step": 1167
+    },
+    {
+      "epoch": 1.235978835978836,
+      "grad_norm": 0.282480189507544,
+      "learning_rate": 3.269306154449236e-05,
+      "loss": 0.4649,
+      "step": 1168
+    },
+    {
+      "epoch": 1.237037037037037,
+      "grad_norm": 0.28035139011097865,
+      "learning_rate": 3.26734613876911e-05,
+      "loss": 0.5027,
+      "step": 1169
+    },
+    {
+      "epoch": 1.2380952380952381,
+      "grad_norm": 0.2291595625165518,
+      "learning_rate": 3.265386123088985e-05,
+      "loss": 0.3896,
+      "step": 1170
+    },
+    {
+      "epoch": 1.2391534391534391,
+      "grad_norm": 0.2566644949612776,
+      "learning_rate": 3.2634261074088597e-05,
+      "loss": 0.4192,
+      "step": 1171
+    },
+    {
+      "epoch": 1.24021164021164,
+      "grad_norm": 0.24470481325359464,
+      "learning_rate": 3.261466091728734e-05,
+      "loss": 0.4963,
+      "step": 1172
+    },
+    {
+      "epoch": 1.2412698412698413,
+      "grad_norm": 0.24563820953102872,
+      "learning_rate": 3.259506076048608e-05,
+      "loss": 0.4184,
+      "step": 1173
+    },
+    {
+      "epoch": 1.2423280423280423,
+      "grad_norm": 0.2347354332412375,
+      "learning_rate": 3.257546060368483e-05,
+      "loss": 0.4514,
+      "step": 1174
+    },
+    {
+      "epoch": 1.2433862433862433,
+      "grad_norm": 2.958020037172032,
+      "learning_rate": 3.255586044688358e-05,
+      "loss": 0.5267,
+      "step": 1175
+    },
+    {
+      "epoch": 1.2444444444444445,
+      "grad_norm": 0.2843310735518099,
+      "learning_rate": 3.253626029008232e-05,
+      "loss": 0.4906,
+      "step": 1176
+    },
+    {
+      "epoch": 1.2455026455026454,
+      "grad_norm": 0.2542336692718839,
+      "learning_rate": 3.2516660133281065e-05,
+      "loss": 0.4443,
+      "step": 1177
+    },
+    {
+      "epoch": 1.2465608465608466,
+      "grad_norm": 0.22674151401535145,
+      "learning_rate": 3.2497059976479814e-05,
+      "loss": 0.4132,
+      "step": 1178
+    },
+    {
+      "epoch": 1.2476190476190476,
+      "grad_norm": 0.24724783616086354,
+      "learning_rate": 3.247745981967856e-05,
+      "loss": 0.4808,
+      "step": 1179
+    },
+    {
+      "epoch": 1.2486772486772486,
+      "grad_norm": 0.24184869409515652,
+      "learning_rate": 3.2457859662877306e-05,
+      "loss": 0.4494,
+      "step": 1180
+    },
+    {
+      "epoch": 1.2497354497354498,
+      "grad_norm": 0.23220783704881004,
+      "learning_rate": 3.243825950607605e-05,
+      "loss": 0.4236,
+      "step": 1181
+    },
+    {
+      "epoch": 1.2507936507936508,
+      "grad_norm": 0.21779940633439762,
+      "learning_rate": 3.241865934927479e-05,
+      "loss": 0.3891,
+      "step": 1182
+    },
+    {
+      "epoch": 1.2518518518518518,
+      "grad_norm": 0.22367557629440174,
+      "learning_rate": 3.239905919247354e-05,
+      "loss": 0.4615,
+      "step": 1183
+    },
+    {
+      "epoch": 1.252910052910053,
+      "grad_norm": 0.24819272296895625,
+      "learning_rate": 3.237945903567229e-05,
+      "loss": 0.4737,
+      "step": 1184
+    },
+    {
+      "epoch": 1.253968253968254,
+      "grad_norm": 0.21940462041140096,
+      "learning_rate": 3.235985887887103e-05,
+      "loss": 0.4358,
+      "step": 1185
+    },
+    {
+      "epoch": 1.2550264550264552,
+      "grad_norm": 0.23333157969435228,
+      "learning_rate": 3.2340258722069775e-05,
+      "loss": 0.4305,
+      "step": 1186
+    },
+    {
+      "epoch": 1.2560846560846561,
+      "grad_norm": 0.2878829627468074,
+      "learning_rate": 3.2320658565268524e-05,
+      "loss": 0.429,
+      "step": 1187
+    },
+    {
+      "epoch": 1.2571428571428571,
+      "grad_norm": 0.24749463547096354,
+      "learning_rate": 3.230105840846727e-05,
+      "loss": 0.4438,
+      "step": 1188
+    },
+    {
+      "epoch": 1.2582010582010583,
+      "grad_norm": 0.228540686736987,
+      "learning_rate": 3.2281458251666016e-05,
+      "loss": 0.4468,
+      "step": 1189
+    },
+    {
+      "epoch": 1.2592592592592593,
+      "grad_norm": 0.2531138794775126,
+      "learning_rate": 3.226185809486476e-05,
+      "loss": 0.4759,
+      "step": 1190
+    },
+    {
+      "epoch": 1.2603174603174603,
+      "grad_norm": 0.24125935658343106,
+      "learning_rate": 3.224225793806351e-05,
+      "loss": 0.4406,
+      "step": 1191
+    },
+    {
+      "epoch": 1.2613756613756615,
+      "grad_norm": 0.25935447680403384,
+      "learning_rate": 3.222265778126225e-05,
+      "loss": 0.4795,
+      "step": 1192
+    },
+    {
+      "epoch": 1.2624338624338625,
+      "grad_norm": 0.23939043891539874,
+      "learning_rate": 3.2203057624461e-05,
+      "loss": 0.4322,
+      "step": 1193
+    },
+    {
+      "epoch": 1.2634920634920634,
+      "grad_norm": 0.23783030731256977,
+      "learning_rate": 3.218345746765974e-05,
+      "loss": 0.4159,
+      "step": 1194
+    },
+    {
+      "epoch": 1.2645502645502646,
+      "grad_norm": 0.2801793027048429,
+      "learning_rate": 3.2163857310858484e-05,
+      "loss": 0.5096,
+      "step": 1195
+    },
+    {
+      "epoch": 1.2656084656084656,
+      "grad_norm": 0.25672962049215936,
+      "learning_rate": 3.214425715405723e-05,
+      "loss": 0.4475,
+      "step": 1196
+    },
+    {
+      "epoch": 1.2666666666666666,
+      "grad_norm": 2.98727547896751,
+      "learning_rate": 3.212465699725598e-05,
+      "loss": 0.5605,
+      "step": 1197
+    },
+    {
+      "epoch": 1.2677248677248678,
+      "grad_norm": 0.31486994948060426,
+      "learning_rate": 3.2105056840454725e-05,
+      "loss": 0.5239,
+      "step": 1198
+    },
+    {
+      "epoch": 1.2687830687830688,
+      "grad_norm": 0.2651388574204949,
+      "learning_rate": 3.208545668365347e-05,
+      "loss": 0.4854,
+      "step": 1199
+    },
+    {
+      "epoch": 1.2698412698412698,
+      "grad_norm": 0.23407941410701735,
+      "learning_rate": 3.206585652685222e-05,
+      "loss": 0.4403,
+      "step": 1200
+    },
+    {
+      "epoch": 1.270899470899471,
+      "grad_norm": 0.25844549625625296,
+      "learning_rate": 3.2046256370050966e-05,
+      "loss": 0.4974,
+      "step": 1201
+    },
+    {
+      "epoch": 1.271957671957672,
+      "grad_norm": 0.357313584767075,
+      "learning_rate": 3.202665621324971e-05,
+      "loss": 0.48,
+      "step": 1202
+    },
+    {
+      "epoch": 1.273015873015873,
+      "grad_norm": 0.2498660944411242,
+      "learning_rate": 3.200705605644845e-05,
+      "loss": 0.4524,
+      "step": 1203
+    },
+    {
+      "epoch": 1.2740740740740741,
+      "grad_norm": 0.2296750952608947,
+      "learning_rate": 3.19874558996472e-05,
+      "loss": 0.4473,
+      "step": 1204
+    },
+    {
+      "epoch": 1.2751322751322751,
+      "grad_norm": 0.23977764063922036,
+      "learning_rate": 3.196785574284594e-05,
+      "loss": 0.4306,
+      "step": 1205
+    },
+    {
+      "epoch": 1.276190476190476,
+      "grad_norm": 0.25984180898265835,
+      "learning_rate": 3.194825558604469e-05,
+      "loss": 0.4463,
+      "step": 1206
+    },
+    {
+      "epoch": 1.2772486772486773,
+      "grad_norm": 0.2205818526717538,
+      "learning_rate": 3.1928655429243434e-05,
+      "loss": 0.4162,
+      "step": 1207
+    },
+    {
+      "epoch": 1.2783068783068783,
+      "grad_norm": 0.20685381181973775,
+      "learning_rate": 3.190905527244218e-05,
+      "loss": 0.4067,
+      "step": 1208
+    },
+    {
+      "epoch": 1.2793650793650793,
+      "grad_norm": 0.2622315576719718,
+      "learning_rate": 3.1889455115640926e-05,
+      "loss": 0.4535,
+      "step": 1209
+    },
+    {
+      "epoch": 1.2804232804232805,
+      "grad_norm": 0.27562631291061185,
+      "learning_rate": 3.1869854958839675e-05,
+      "loss": 0.5411,
+      "step": 1210
+    },
+    {
+      "epoch": 1.2814814814814814,
+      "grad_norm": 0.2436787808592762,
+      "learning_rate": 3.185025480203842e-05,
+      "loss": 0.4231,
+      "step": 1211
+    },
+    {
+      "epoch": 1.2825396825396824,
+      "grad_norm": 0.26452961096327726,
+      "learning_rate": 3.183065464523716e-05,
+      "loss": 0.4682,
+      "step": 1212
+    },
+    {
+      "epoch": 1.2835978835978836,
+      "grad_norm": 0.2561065853683004,
+      "learning_rate": 3.181105448843591e-05,
+      "loss": 0.4672,
+      "step": 1213
+    },
+    {
+      "epoch": 1.2846560846560846,
+      "grad_norm": 0.23228316782055128,
+      "learning_rate": 3.179145433163466e-05,
+      "loss": 0.4774,
+      "step": 1214
+    },
+    {
+      "epoch": 1.2857142857142856,
+      "grad_norm": 2.661626366284025,
+      "learning_rate": 3.17718541748334e-05,
+      "loss": 0.492,
+      "step": 1215
+    },
+    {
+      "epoch": 1.2867724867724868,
+      "grad_norm": 0.2550139339573797,
+      "learning_rate": 3.1752254018032144e-05,
+      "loss": 0.393,
+      "step": 1216
+    },
+    {
+      "epoch": 1.2878306878306878,
+      "grad_norm": 0.2982626987372917,
+      "learning_rate": 3.173265386123089e-05,
+      "loss": 0.4337,
+      "step": 1217
+    },
+    {
+      "epoch": 1.2888888888888888,
+      "grad_norm": 0.28057911093701576,
+      "learning_rate": 3.1713053704429636e-05,
+      "loss": 0.4806,
+      "step": 1218
+    },
+    {
+      "epoch": 1.28994708994709,
+      "grad_norm": 0.4561185589253618,
+      "learning_rate": 3.1693453547628385e-05,
+      "loss": 0.4365,
+      "step": 1219
+    },
+    {
+      "epoch": 1.291005291005291,
+      "grad_norm": 0.29263696706005893,
+      "learning_rate": 3.167385339082713e-05,
+      "loss": 0.4656,
+      "step": 1220
+    },
+    {
+      "epoch": 1.2920634920634921,
+      "grad_norm": 0.27093691099606854,
+      "learning_rate": 3.165425323402587e-05,
+      "loss": 0.4962,
+      "step": 1221
+    },
+    {
+      "epoch": 1.2931216931216931,
+      "grad_norm": 0.28064868068280735,
+      "learning_rate": 3.163465307722462e-05,
+      "loss": 0.4068,
+      "step": 1222
+    },
+    {
+      "epoch": 1.294179894179894,
+      "grad_norm": 0.3148596581856991,
+      "learning_rate": 3.161505292042337e-05,
+      "loss": 0.4815,
+      "step": 1223
+    },
+    {
+      "epoch": 1.2952380952380953,
+      "grad_norm": 0.7434992980537427,
+      "learning_rate": 3.159545276362211e-05,
+      "loss": 0.5237,
+      "step": 1224
+    },
+    {
+      "epoch": 1.2962962962962963,
+      "grad_norm": 0.27981974942403515,
+      "learning_rate": 3.1575852606820853e-05,
+      "loss": 0.4476,
+      "step": 1225
+    },
+    {
+      "epoch": 1.2973544973544975,
+      "grad_norm": 0.30917000871173816,
+      "learning_rate": 3.15562524500196e-05,
+      "loss": 0.433,
+      "step": 1226
+    },
+    {
+      "epoch": 1.2984126984126985,
+      "grad_norm": 0.2756768577076242,
+      "learning_rate": 3.153665229321835e-05,
+      "loss": 0.4463,
+      "step": 1227
+    },
+    {
+      "epoch": 1.2994708994708994,
+      "grad_norm": 0.29298489584722304,
+      "learning_rate": 3.1517052136417094e-05,
+      "loss": 0.4787,
+      "step": 1228
+    },
+    {
+      "epoch": 1.3005291005291006,
+      "grad_norm": 0.23814302832789452,
+      "learning_rate": 3.149745197961584e-05,
+      "loss": 0.4115,
+      "step": 1229
+    },
+    {
+      "epoch": 1.3015873015873016,
+      "grad_norm": 0.3031186340550912,
+      "learning_rate": 3.1477851822814586e-05,
+      "loss": 0.4138,
+      "step": 1230
+    },
+    {
+      "epoch": 1.3026455026455026,
+      "grad_norm": 0.28701409757098467,
+      "learning_rate": 3.145825166601333e-05,
+      "loss": 0.4445,
+      "step": 1231
+    },
+    {
+      "epoch": 1.3037037037037038,
+      "grad_norm": 0.24322652680335896,
+      "learning_rate": 3.143865150921208e-05,
+      "loss": 0.4955,
+      "step": 1232
+    },
+    {
+      "epoch": 1.3047619047619048,
+      "grad_norm": 0.2672959117899414,
+      "learning_rate": 3.141905135241082e-05,
+      "loss": 0.495,
+      "step": 1233
+    },
+    {
+      "epoch": 1.3058201058201058,
+      "grad_norm": 0.25188935994394157,
+      "learning_rate": 3.139945119560956e-05,
+      "loss": 0.4609,
+      "step": 1234
+    },
+    {
+      "epoch": 1.306878306878307,
+      "grad_norm": 0.24955613360702394,
+      "learning_rate": 3.137985103880831e-05,
+      "loss": 0.4466,
+      "step": 1235
+    },
+    {
+      "epoch": 1.307936507936508,
+      "grad_norm": 0.23518682524284248,
+      "learning_rate": 3.136025088200706e-05,
+      "loss": 0.3943,
+      "step": 1236
+    },
+    {
+      "epoch": 1.308994708994709,
+      "grad_norm": 0.23720969598849823,
+      "learning_rate": 3.1340650725205804e-05,
+      "loss": 0.5164,
+      "step": 1237
+    },
+    {
+      "epoch": 1.3100529100529101,
+      "grad_norm": 0.21537376471764513,
+      "learning_rate": 3.1321050568404546e-05,
+      "loss": 0.4002,
+      "step": 1238
+    },
+    {
+      "epoch": 1.3111111111111111,
+      "grad_norm": 1.6860157051017535,
+      "learning_rate": 3.1301450411603296e-05,
+      "loss": 0.4859,
+      "step": 1239
+    },
+    {
+      "epoch": 1.312169312169312,
+      "grad_norm": 0.2651741235611922,
+      "learning_rate": 3.1281850254802045e-05,
+      "loss": 0.421,
+      "step": 1240
+    },
+    {
+      "epoch": 1.3132275132275133,
+      "grad_norm": 0.23836910046795176,
+      "learning_rate": 3.126225009800078e-05,
+      "loss": 0.4484,
+      "step": 1241
+    },
+    {
+      "epoch": 1.3142857142857143,
+      "grad_norm": 0.34511017850329395,
+      "learning_rate": 3.124264994119953e-05,
+      "loss": 0.4587,
+      "step": 1242
+    },
+    {
+      "epoch": 1.3153439153439153,
+      "grad_norm": 0.3025522362899887,
+      "learning_rate": 3.122304978439828e-05,
+      "loss": 0.4275,
+      "step": 1243
+    },
+    {
+      "epoch": 1.3164021164021165,
+      "grad_norm": 0.247152508905285,
+      "learning_rate": 3.120344962759702e-05,
+      "loss": 0.4214,
+      "step": 1244
+    },
+    {
+      "epoch": 1.3174603174603174,
+      "grad_norm": 0.25398227298952686,
+      "learning_rate": 3.1183849470795764e-05,
+      "loss": 0.4581,
+      "step": 1245
+    },
+    {
+      "epoch": 1.3185185185185184,
+      "grad_norm": 0.22633308872087934,
+      "learning_rate": 3.1164249313994513e-05,
+      "loss": 0.4584,
+      "step": 1246
+    },
+    {
+      "epoch": 1.3195767195767196,
+      "grad_norm": 0.24210217710444076,
+      "learning_rate": 3.1144649157193256e-05,
+      "loss": 0.4206,
+      "step": 1247
+    },
+    {
+      "epoch": 1.3206349206349206,
+      "grad_norm": 0.23335804933192775,
+      "learning_rate": 3.1125049000392005e-05,
+      "loss": 0.4248,
+      "step": 1248
+    },
+    {
+      "epoch": 1.3216931216931216,
+      "grad_norm": 0.24164931346146493,
+      "learning_rate": 3.110544884359075e-05,
+      "loss": 0.4013,
+      "step": 1249
+    },
+    {
+      "epoch": 1.3227513227513228,
+      "grad_norm": 1.2954500627805814,
+      "learning_rate": 3.10858486867895e-05,
+      "loss": 0.5315,
+      "step": 1250
+    },
+    {
+      "epoch": 1.3238095238095238,
+      "grad_norm": 0.23445766500023593,
+      "learning_rate": 3.106624852998824e-05,
+      "loss": 0.4013,
+      "step": 1251
+    },
+    {
+      "epoch": 1.3248677248677247,
+      "grad_norm": 0.24783454283131148,
+      "learning_rate": 3.104664837318699e-05,
+      "loss": 0.4648,
+      "step": 1252
+    },
+    {
+      "epoch": 1.325925925925926,
+      "grad_norm": 0.2659873598285454,
+      "learning_rate": 3.102704821638574e-05,
+      "loss": 0.4992,
+      "step": 1253
+    },
+    {
+      "epoch": 1.326984126984127,
+      "grad_norm": 0.6888495280810113,
+      "learning_rate": 3.1007448059584474e-05,
+      "loss": 0.4971,
+      "step": 1254
+    },
+    {
+      "epoch": 1.328042328042328,
+      "grad_norm": 0.24722561577784258,
+      "learning_rate": 3.098784790278322e-05,
+      "loss": 0.4246,
+      "step": 1255
+    },
+    {
+      "epoch": 1.3291005291005291,
+      "grad_norm": 0.27176108829651746,
+      "learning_rate": 3.096824774598197e-05,
+      "loss": 0.493,
+      "step": 1256
+    },
+    {
+      "epoch": 1.33015873015873,
+      "grad_norm": 0.2922219870244475,
+      "learning_rate": 3.0948647589180715e-05,
+      "loss": 0.4643,
+      "step": 1257
+    },
+    {
+      "epoch": 1.3312169312169313,
+      "grad_norm": 0.2495354668993547,
+      "learning_rate": 3.092904743237946e-05,
+      "loss": 0.4807,
+      "step": 1258
+    },
+    {
+      "epoch": 1.3322751322751323,
+      "grad_norm": 0.25416175375930267,
+      "learning_rate": 3.0909447275578206e-05,
+      "loss": 0.4559,
+      "step": 1259
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.2505309654320362,
+      "learning_rate": 3.088984711877695e-05,
+      "loss": 0.4175,
+      "step": 1260
+    },
+    {
+      "epoch": 1.3343915343915345,
+      "grad_norm": 0.22168988059391945,
+      "learning_rate": 3.08702469619757e-05,
+      "loss": 0.3638,
+      "step": 1261
+    },
+    {
+      "epoch": 1.3354497354497354,
+      "grad_norm": 0.2088949121467012,
+      "learning_rate": 3.085064680517444e-05,
+      "loss": 0.3575,
+      "step": 1262
+    },
+    {
+      "epoch": 1.3365079365079366,
+      "grad_norm": 0.26660498435295343,
+      "learning_rate": 3.083104664837319e-05,
+      "loss": 0.4729,
+      "step": 1263
+    },
+    {
+      "epoch": 1.3375661375661376,
+      "grad_norm": 0.24658883170058563,
+      "learning_rate": 3.081144649157193e-05,
+      "loss": 0.4545,
+      "step": 1264
+    },
+    {
+      "epoch": 1.3386243386243386,
+      "grad_norm": 0.2236745034029776,
+      "learning_rate": 3.079184633477068e-05,
+      "loss": 0.4503,
+      "step": 1265
+    },
+    {
+      "epoch": 1.3396825396825398,
+      "grad_norm": 0.24135305732779005,
+      "learning_rate": 3.0772246177969424e-05,
+      "loss": 0.4874,
+      "step": 1266
+    },
+    {
+      "epoch": 1.3407407407407408,
+      "grad_norm": 0.3256043618189639,
+      "learning_rate": 3.075264602116817e-05,
+      "loss": 0.5231,
+      "step": 1267
+    },
+    {
+      "epoch": 1.3417989417989418,
+      "grad_norm": 0.23737346236972562,
+      "learning_rate": 3.0733045864366916e-05,
+      "loss": 0.4316,
+      "step": 1268
+    },
+    {
+      "epoch": 1.342857142857143,
+      "grad_norm": 0.23544854426751335,
+      "learning_rate": 3.0713445707565665e-05,
+      "loss": 0.5169,
+      "step": 1269
+    },
+    {
+      "epoch": 1.343915343915344,
+      "grad_norm": 1.659064568731437,
+      "learning_rate": 3.069384555076441e-05,
+      "loss": 0.4348,
+      "step": 1270
+    },
+    {
+      "epoch": 1.344973544973545,
+      "grad_norm": 0.2534937144399376,
+      "learning_rate": 3.067424539396315e-05,
+      "loss": 0.4652,
+      "step": 1271
+    },
+    {
+      "epoch": 1.3460317460317461,
+      "grad_norm": 0.23526593221138056,
+      "learning_rate": 3.06546452371619e-05,
+      "loss": 0.4616,
+      "step": 1272
+    },
+    {
+      "epoch": 1.3470899470899471,
+      "grad_norm": 0.23742609091689737,
+      "learning_rate": 3.063504508036065e-05,
+      "loss": 0.4369,
+      "step": 1273
+    },
+    {
+      "epoch": 1.348148148148148,
+      "grad_norm": 0.2222447404589058,
+      "learning_rate": 3.061544492355939e-05,
+      "loss": 0.4124,
+      "step": 1274
+    },
+    {
+      "epoch": 1.3492063492063493,
+      "grad_norm": 0.29184250980945264,
+      "learning_rate": 3.0595844766758134e-05,
+      "loss": 0.4454,
+      "step": 1275
+    },
+    {
+      "epoch": 1.3502645502645503,
+      "grad_norm": 0.2349625400452902,
+      "learning_rate": 3.057624460995688e-05,
+      "loss": 0.4706,
+      "step": 1276
+    },
+    {
+      "epoch": 1.3513227513227513,
+      "grad_norm": 0.21431852615283079,
+      "learning_rate": 3.0556644453155625e-05,
+      "loss": 0.4258,
+      "step": 1277
+    },
+    {
+      "epoch": 1.3523809523809525,
+      "grad_norm": 0.21372092688801175,
+      "learning_rate": 3.0537044296354375e-05,
+      "loss": 0.4199,
+      "step": 1278
+    },
+    {
+      "epoch": 1.3534391534391534,
+      "grad_norm": 0.2579456014466511,
+      "learning_rate": 3.051744413955312e-05,
+      "loss": 0.4978,
+      "step": 1279
+    },
+    {
+      "epoch": 1.3544973544973544,
+      "grad_norm": 0.2507941348475173,
+      "learning_rate": 3.0497843982751863e-05,
+      "loss": 0.4665,
+      "step": 1280
+    },
+    {
+      "epoch": 1.3555555555555556,
+      "grad_norm": 0.232129817590424,
+      "learning_rate": 3.047824382595061e-05,
+      "loss": 0.456,
+      "step": 1281
+    },
+    {
+      "epoch": 1.3566137566137566,
+      "grad_norm": 0.2514526079081348,
+      "learning_rate": 3.0458643669149355e-05,
+      "loss": 0.4404,
+      "step": 1282
+    },
+    {
+      "epoch": 1.3576719576719576,
+      "grad_norm": 0.2572184939136445,
+      "learning_rate": 3.0439043512348097e-05,
+      "loss": 0.5318,
+      "step": 1283
+    },
+    {
+      "epoch": 1.3587301587301588,
+      "grad_norm": 0.4039307927131717,
+      "learning_rate": 3.0419443355546846e-05,
+      "loss": 0.3901,
+      "step": 1284
+    },
+    {
+      "epoch": 1.3597883597883598,
+      "grad_norm": 0.24715025836977536,
+      "learning_rate": 3.0399843198745592e-05,
+      "loss": 0.4698,
+      "step": 1285
+    },
+    {
+      "epoch": 1.3608465608465607,
+      "grad_norm": 0.23085610616983096,
+      "learning_rate": 3.0380243041944338e-05,
+      "loss": 0.4939,
+      "step": 1286
+    },
+    {
+      "epoch": 1.361904761904762,
+      "grad_norm": 0.2364958065579408,
+      "learning_rate": 3.036064288514308e-05,
+      "loss": 0.4244,
+      "step": 1287
+    },
+    {
+      "epoch": 1.362962962962963,
+      "grad_norm": 0.25662478506333875,
+      "learning_rate": 3.034104272834183e-05,
+      "loss": 0.4692,
+      "step": 1288
+    },
+    {
+      "epoch": 1.364021164021164,
+      "grad_norm": 0.23895684567679154,
+      "learning_rate": 3.0321442571540576e-05,
+      "loss": 0.4296,
+      "step": 1289
+    },
+    {
+      "epoch": 1.3650793650793651,
+      "grad_norm": 0.2440970628813272,
+      "learning_rate": 3.030184241473932e-05,
+      "loss": 0.4288,
+      "step": 1290
+    },
+    {
+      "epoch": 1.366137566137566,
+      "grad_norm": 0.22375993243823888,
+      "learning_rate": 3.0282242257938064e-05,
+      "loss": 0.4409,
+      "step": 1291
+    },
+    {
+      "epoch": 1.367195767195767,
+      "grad_norm": 0.2213878463198946,
+      "learning_rate": 3.0262642101136813e-05,
+      "loss": 0.4122,
+      "step": 1292
+    },
+    {
+      "epoch": 1.3682539682539683,
+      "grad_norm": 0.23752017382355897,
+      "learning_rate": 3.0243041944335553e-05,
+      "loss": 0.4666,
+      "step": 1293
+    },
+    {
+      "epoch": 1.3693121693121693,
+      "grad_norm": 0.25742210373427815,
+      "learning_rate": 3.0223441787534302e-05,
+      "loss": 0.4956,
+      "step": 1294
+    },
+    {
+      "epoch": 1.3703703703703702,
+      "grad_norm": 0.234897224069936,
+      "learning_rate": 3.0203841630733048e-05,
+      "loss": 0.4467,
+      "step": 1295
+    },
+    {
+      "epoch": 1.3714285714285714,
+      "grad_norm": 0.23263694871443225,
+      "learning_rate": 3.018424147393179e-05,
+      "loss": 0.4795,
+      "step": 1296
+    },
+    {
+      "epoch": 1.3724867724867724,
+      "grad_norm": 0.2712687227877605,
+      "learning_rate": 3.0164641317130536e-05,
+      "loss": 0.5024,
+      "step": 1297
+    },
+    {
+      "epoch": 1.3735449735449736,
+      "grad_norm": 0.22372235767515633,
+      "learning_rate": 3.0145041160329285e-05,
+      "loss": 0.48,
+      "step": 1298
+    },
+    {
+      "epoch": 1.3746031746031746,
+      "grad_norm": 0.2406938475308407,
+      "learning_rate": 3.012544100352803e-05,
+      "loss": 0.4314,
+      "step": 1299
+    },
+    {
+      "epoch": 1.3756613756613756,
+      "grad_norm": 0.21389054293693965,
+      "learning_rate": 3.0105840846726774e-05,
+      "loss": 0.4229,
+      "step": 1300
+    },
+    {
+      "epoch": 1.3767195767195768,
+      "grad_norm": 0.23149026826112992,
+      "learning_rate": 3.008624068992552e-05,
+      "loss": 0.4707,
+      "step": 1301
+    },
+    {
+      "epoch": 1.3777777777777778,
+      "grad_norm": 0.2691143078356963,
+      "learning_rate": 3.006664053312427e-05,
+      "loss": 0.4282,
+      "step": 1302
+    },
+    {
+      "epoch": 1.378835978835979,
+      "grad_norm": 0.20755788867950342,
+      "learning_rate": 3.004704037632301e-05,
+      "loss": 0.3924,
+      "step": 1303
+    },
+    {
+      "epoch": 1.37989417989418,
+      "grad_norm": 0.23052600055188638,
+      "learning_rate": 3.0027440219521757e-05,
+      "loss": 0.4129,
+      "step": 1304
+    },
+    {
+      "epoch": 1.380952380952381,
+      "grad_norm": 0.22668631216324492,
+      "learning_rate": 3.0007840062720503e-05,
+      "loss": 0.4708,
+      "step": 1305
+    },
+    {
+      "epoch": 1.3820105820105821,
+      "grad_norm": 0.2660877251492311,
+      "learning_rate": 2.9988239905919246e-05,
+      "loss": 0.5037,
+      "step": 1306
+    },
+    {
+      "epoch": 1.3830687830687831,
+      "grad_norm": 0.22823512577697444,
+      "learning_rate": 2.9968639749117995e-05,
+      "loss": 0.4464,
+      "step": 1307
+    },
+    {
+      "epoch": 1.384126984126984,
+      "grad_norm": 0.24777055372668572,
+      "learning_rate": 2.994903959231674e-05,
+      "loss": 0.5314,
+      "step": 1308
+    },
+    {
+      "epoch": 1.3851851851851853,
+      "grad_norm": 0.23660719349812223,
+      "learning_rate": 2.9929439435515483e-05,
+      "loss": 0.4372,
+      "step": 1309
+    },
+    {
+      "epoch": 1.3862433862433863,
+      "grad_norm": 0.25379316726262235,
+      "learning_rate": 2.990983927871423e-05,
+      "loss": 0.4353,
+      "step": 1310
+    },
+    {
+      "epoch": 1.3873015873015873,
+      "grad_norm": 0.25004667495021365,
+      "learning_rate": 2.989023912191298e-05,
+      "loss": 0.4608,
+      "step": 1311
+    },
+    {
+      "epoch": 1.3883597883597885,
+      "grad_norm": 0.22311146023995984,
+      "learning_rate": 2.9870638965111724e-05,
+      "loss": 0.4398,
+      "step": 1312
+    },
+    {
+      "epoch": 1.3894179894179894,
+      "grad_norm": 0.2557529215124851,
+      "learning_rate": 2.9851038808310467e-05,
+      "loss": 0.5005,
+      "step": 1313
+    },
+    {
+      "epoch": 1.3904761904761904,
+      "grad_norm": 0.219279598189162,
+      "learning_rate": 2.9831438651509213e-05,
+      "loss": 0.4129,
+      "step": 1314
+    },
+    {
+      "epoch": 1.3915343915343916,
+      "grad_norm": 0.2269485021477995,
+      "learning_rate": 2.9811838494707962e-05,
+      "loss": 0.4145,
+      "step": 1315
+    },
+    {
+      "epoch": 1.3925925925925926,
+      "grad_norm": 0.2331786311014036,
+      "learning_rate": 2.9792238337906704e-05,
+      "loss": 0.42,
+      "step": 1316
+    },
+    {
+      "epoch": 1.3936507936507936,
+      "grad_norm": 0.31888841203624635,
+      "learning_rate": 2.977263818110545e-05,
+      "loss": 0.4383,
+      "step": 1317
+    },
+    {
+      "epoch": 1.3947089947089948,
+      "grad_norm": 0.212841055960245,
+      "learning_rate": 2.9753038024304196e-05,
+      "loss": 0.4061,
+      "step": 1318
+    },
+    {
+      "epoch": 1.3957671957671958,
+      "grad_norm": 0.24247705953739102,
+      "learning_rate": 2.973343786750294e-05,
+      "loss": 0.4658,
+      "step": 1319
+    },
+    {
+      "epoch": 1.3968253968253967,
+      "grad_norm": 0.30703930058231754,
+      "learning_rate": 2.9713837710701688e-05,
+      "loss": 0.3601,
+      "step": 1320
+    },
+    {
+      "epoch": 1.397883597883598,
+      "grad_norm": 0.23472988733850123,
+      "learning_rate": 2.9694237553900434e-05,
+      "loss": 0.4945,
+      "step": 1321
+    },
+    {
+      "epoch": 1.398941798941799,
+      "grad_norm": 0.30870145935773974,
+      "learning_rate": 2.967463739709918e-05,
+      "loss": 0.4775,
+      "step": 1322
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.22196743152262746,
+      "learning_rate": 2.9655037240297922e-05,
+      "loss": 0.457,
+      "step": 1323
+    },
+    {
+      "epoch": 1.4010582010582011,
+      "grad_norm": 0.23880933160360443,
+      "learning_rate": 2.963543708349667e-05,
+      "loss": 0.4267,
+      "step": 1324
+    },
+    {
+      "epoch": 1.402116402116402,
+      "grad_norm": 0.20702662843283434,
+      "learning_rate": 2.9615836926695417e-05,
+      "loss": 0.441,
+      "step": 1325
+    },
+    {
+      "epoch": 1.403174603174603,
+      "grad_norm": 0.2167350518937471,
+      "learning_rate": 2.959623676989416e-05,
+      "loss": 0.4408,
+      "step": 1326
+    },
+    {
+      "epoch": 1.4042328042328043,
+      "grad_norm": 0.24089833902821953,
+      "learning_rate": 2.9576636613092906e-05,
+      "loss": 0.4968,
+      "step": 1327
+    },
+    {
+      "epoch": 1.4052910052910053,
+      "grad_norm": 0.2241238478376523,
+      "learning_rate": 2.9557036456291655e-05,
+      "loss": 0.4517,
+      "step": 1328
+    },
+    {
+      "epoch": 1.4063492063492062,
+      "grad_norm": 0.22856799607317055,
+      "learning_rate": 2.9537436299490394e-05,
+      "loss": 0.4579,
+      "step": 1329
+    },
+    {
+      "epoch": 1.4074074074074074,
+      "grad_norm": 0.2268187851367138,
+      "learning_rate": 2.9517836142689143e-05,
+      "loss": 0.4649,
+      "step": 1330
+    },
+    {
+      "epoch": 1.4084656084656084,
+      "grad_norm": 0.22233549147880394,
+      "learning_rate": 2.949823598588789e-05,
+      "loss": 0.4465,
+      "step": 1331
+    },
+    {
+      "epoch": 1.4095238095238094,
+      "grad_norm": 0.2236855488492212,
+      "learning_rate": 2.947863582908663e-05,
+      "loss": 0.4228,
+      "step": 1332
+    },
+    {
+      "epoch": 1.4105820105820106,
+      "grad_norm": 0.23801323711217828,
+      "learning_rate": 2.9459035672285377e-05,
+      "loss": 0.4252,
+      "step": 1333
+    },
+    {
+      "epoch": 1.4116402116402116,
+      "grad_norm": 0.24798530897647444,
+      "learning_rate": 2.9439435515484127e-05,
+      "loss": 0.5064,
+      "step": 1334
+    },
+    {
+      "epoch": 1.4126984126984126,
+      "grad_norm": 0.23941337695722817,
+      "learning_rate": 2.9419835358682873e-05,
+      "loss": 0.4679,
+      "step": 1335
+    },
+    {
+      "epoch": 1.4137566137566138,
+      "grad_norm": 0.2623716098058726,
+      "learning_rate": 2.9400235201881615e-05,
+      "loss": 0.4473,
+      "step": 1336
+    },
+    {
+      "epoch": 1.4148148148148147,
+      "grad_norm": 0.21199389424437273,
+      "learning_rate": 2.938063504508036e-05,
+      "loss": 0.4225,
+      "step": 1337
+    },
+    {
+      "epoch": 1.415873015873016,
+      "grad_norm": 0.21541116818374437,
+      "learning_rate": 2.936103488827911e-05,
+      "loss": 0.4108,
+      "step": 1338
+    },
+    {
+      "epoch": 1.416931216931217,
+      "grad_norm": 0.24853955951034445,
+      "learning_rate": 2.9341434731477853e-05,
+      "loss": 0.447,
+      "step": 1339
+    },
+    {
+      "epoch": 1.417989417989418,
+      "grad_norm": 0.2215945031478815,
+      "learning_rate": 2.93218345746766e-05,
+      "loss": 0.4761,
+      "step": 1340
+    },
+    {
+      "epoch": 1.4190476190476191,
+      "grad_norm": 0.2133339228199975,
+      "learning_rate": 2.9302234417875344e-05,
+      "loss": 0.4524,
+      "step": 1341
+    },
+    {
+      "epoch": 1.42010582010582,
+      "grad_norm": 0.24231327015852916,
+      "learning_rate": 2.9282634261074087e-05,
+      "loss": 0.4629,
+      "step": 1342
+    },
+    {
+      "epoch": 1.4211640211640213,
+      "grad_norm": 0.21685366161636507,
+      "learning_rate": 2.9263034104272836e-05,
+      "loss": 0.4234,
+      "step": 1343
+    },
+    {
+      "epoch": 1.4222222222222223,
+      "grad_norm": 0.22253006527800145,
+      "learning_rate": 2.9243433947471582e-05,
+      "loss": 0.4323,
+      "step": 1344
+    },
+    {
+      "epoch": 1.4232804232804233,
+      "grad_norm": 0.22176071929381708,
+      "learning_rate": 2.9223833790670325e-05,
+      "loss": 0.4427,
+      "step": 1345
+    },
+    {
+      "epoch": 1.4243386243386245,
+      "grad_norm": 0.24232602640124465,
+      "learning_rate": 2.920423363386907e-05,
+      "loss": 0.3828,
+      "step": 1346
+    },
+    {
+      "epoch": 1.4253968253968254,
+      "grad_norm": 0.22581122208104695,
+      "learning_rate": 2.918463347706782e-05,
+      "loss": 0.4805,
+      "step": 1347
+    },
+    {
+      "epoch": 1.4264550264550264,
+      "grad_norm": 0.20026012333803486,
+      "learning_rate": 2.9165033320266566e-05,
+      "loss": 0.3919,
+      "step": 1348
+    },
+    {
+      "epoch": 1.4275132275132276,
+      "grad_norm": 0.22333787876129188,
+      "learning_rate": 2.9145433163465308e-05,
+      "loss": 0.4131,
+      "step": 1349
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 0.2687484229021416,
+      "learning_rate": 2.9125833006664054e-05,
+      "loss": 0.5264,
+      "step": 1350
+    },
+    {
+      "epoch": 1.4296296296296296,
+      "grad_norm": 0.2103685914143013,
+      "learning_rate": 2.9106232849862803e-05,
+      "loss": 0.4576,
+      "step": 1351
+    },
+    {
+      "epoch": 1.4306878306878308,
+      "grad_norm": 0.21771460457762667,
+      "learning_rate": 2.9086632693061546e-05,
+      "loss": 0.4361,
+      "step": 1352
+    },
+    {
+      "epoch": 1.4317460317460318,
+      "grad_norm": 0.20303241796561644,
+      "learning_rate": 2.906703253626029e-05,
+      "loss": 0.3971,
+      "step": 1353
+    },
+    {
+      "epoch": 1.4328042328042327,
+      "grad_norm": 0.20143577073807975,
+      "learning_rate": 2.9047432379459037e-05,
+      "loss": 0.3797,
+      "step": 1354
+    },
+    {
+      "epoch": 1.433862433862434,
+      "grad_norm": 0.2329009387387878,
+      "learning_rate": 2.902783222265778e-05,
+      "loss": 0.506,
+      "step": 1355
+    },
+    {
+      "epoch": 1.434920634920635,
+      "grad_norm": 0.2702723280262594,
+      "learning_rate": 2.900823206585653e-05,
+      "loss": 0.3775,
+      "step": 1356
+    },
+    {
+      "epoch": 1.435978835978836,
+      "grad_norm": 0.2834676972961095,
+      "learning_rate": 2.8988631909055275e-05,
+      "loss": 0.462,
+      "step": 1357
+    },
+    {
+      "epoch": 1.4370370370370371,
+      "grad_norm": 0.23065496781142314,
+      "learning_rate": 2.8969031752254017e-05,
+      "loss": 0.4994,
+      "step": 1358
+    },
+    {
+      "epoch": 1.438095238095238,
+      "grad_norm": 0.22768658403107855,
+      "learning_rate": 2.8949431595452763e-05,
+      "loss": 0.4218,
+      "step": 1359
+    },
+    {
+      "epoch": 1.439153439153439,
+      "grad_norm": 0.2680457798964453,
+      "learning_rate": 2.8929831438651513e-05,
+      "loss": 0.4527,
+      "step": 1360
+    },
+    {
+      "epoch": 1.4402116402116403,
+      "grad_norm": 0.2275408510504903,
+      "learning_rate": 2.891023128185026e-05,
+      "loss": 0.4718,
+      "step": 1361
+    },
+    {
+      "epoch": 1.4412698412698413,
+      "grad_norm": 0.23860010786360963,
+      "learning_rate": 2.8890631125049e-05,
+      "loss": 0.4656,
+      "step": 1362
+    },
+    {
+      "epoch": 1.4423280423280422,
+      "grad_norm": 0.5143958846694693,
+      "learning_rate": 2.8871030968247747e-05,
+      "loss": 0.4567,
+      "step": 1363
+    },
+    {
+      "epoch": 1.4433862433862434,
+      "grad_norm": 0.23216914200826835,
+      "learning_rate": 2.8851430811446496e-05,
+      "loss": 0.4873,
+      "step": 1364
+    },
+    {
+      "epoch": 1.4444444444444444,
+      "grad_norm": 0.25213100827470797,
+      "learning_rate": 2.8831830654645235e-05,
+      "loss": 0.4681,
+      "step": 1365
+    },
+    {
+      "epoch": 1.4455026455026454,
+      "grad_norm": 0.22446515285108198,
+      "learning_rate": 2.8812230497843984e-05,
+      "loss": 0.3967,
+      "step": 1366
+    },
+    {
+      "epoch": 1.4465608465608466,
+      "grad_norm": 0.22388321159591337,
+      "learning_rate": 2.879263034104273e-05,
+      "loss": 0.409,
+      "step": 1367
+    },
+    {
+      "epoch": 1.4476190476190476,
+      "grad_norm": 0.22054153685138833,
+      "learning_rate": 2.8773030184241473e-05,
+      "loss": 0.4393,
+      "step": 1368
+    },
+    {
+      "epoch": 1.4486772486772486,
+      "grad_norm": 0.23961761676424165,
+      "learning_rate": 2.875343002744022e-05,
+      "loss": 0.416,
+      "step": 1369
+    },
+    {
+      "epoch": 1.4497354497354498,
+      "grad_norm": 0.24967858263135762,
+      "learning_rate": 2.8733829870638968e-05,
+      "loss": 0.4897,
+      "step": 1370
+    },
+    {
+      "epoch": 1.4507936507936507,
+      "grad_norm": 0.23131162096590446,
+      "learning_rate": 2.8714229713837714e-05,
+      "loss": 0.4876,
+      "step": 1371
+    },
+    {
+      "epoch": 1.4518518518518517,
+      "grad_norm": 0.9779701550723804,
+      "learning_rate": 2.8694629557036456e-05,
+      "loss": 0.4304,
+      "step": 1372
+    },
+    {
+      "epoch": 1.452910052910053,
+      "grad_norm": 0.25630422299181627,
+      "learning_rate": 2.8675029400235202e-05,
+      "loss": 0.4348,
+      "step": 1373
+    },
+    {
+      "epoch": 1.453968253968254,
+      "grad_norm": 0.25784690293324475,
+      "learning_rate": 2.865542924343395e-05,
+      "loss": 0.4629,
+      "step": 1374
+    },
+    {
+      "epoch": 1.455026455026455,
+      "grad_norm": 0.27761793043496447,
+      "learning_rate": 2.8635829086632694e-05,
+      "loss": 0.4485,
+      "step": 1375
+    },
+    {
+      "epoch": 1.456084656084656,
+      "grad_norm": 0.26282171909723734,
+      "learning_rate": 2.861622892983144e-05,
+      "loss": 0.4981,
+      "step": 1376
+    },
+    {
+      "epoch": 1.457142857142857,
+      "grad_norm": 0.7366784020256089,
+      "learning_rate": 2.8596628773030186e-05,
+      "loss": 0.4384,
+      "step": 1377
+    },
+    {
+      "epoch": 1.4582010582010583,
+      "grad_norm": 0.23451220676925627,
+      "learning_rate": 2.8577028616228928e-05,
+      "loss": 0.4342,
+      "step": 1378
+    },
+    {
+      "epoch": 1.4592592592592593,
+      "grad_norm": 0.2232768861934982,
+      "learning_rate": 2.8557428459427677e-05,
+      "loss": 0.4122,
+      "step": 1379
+    },
+    {
+      "epoch": 1.4603174603174602,
+      "grad_norm": 0.2330751792602911,
+      "learning_rate": 2.8537828302626423e-05,
+      "loss": 0.4716,
+      "step": 1380
+    },
+    {
+      "epoch": 1.4613756613756614,
+      "grad_norm": 0.2363839837673862,
+      "learning_rate": 2.8518228145825166e-05,
+      "loss": 0.4455,
+      "step": 1381
+    },
+    {
+      "epoch": 1.4624338624338624,
+      "grad_norm": 0.22199158535950497,
+      "learning_rate": 2.8498627989023912e-05,
+      "loss": 0.4017,
+      "step": 1382
+    },
+    {
+      "epoch": 1.4634920634920636,
+      "grad_norm": 0.2140913960403465,
+      "learning_rate": 2.847902783222266e-05,
+      "loss": 0.4171,
+      "step": 1383
+    },
+    {
+      "epoch": 1.4645502645502646,
+      "grad_norm": 0.21942455553450194,
+      "learning_rate": 2.8459427675421407e-05,
+      "loss": 0.4345,
+      "step": 1384
+    },
+    {
+      "epoch": 1.4656084656084656,
+      "grad_norm": 0.2589886529373693,
+      "learning_rate": 2.843982751862015e-05,
+      "loss": 0.4751,
+      "step": 1385
+    },
+    {
+      "epoch": 1.4666666666666668,
+      "grad_norm": 0.2161806830799717,
+      "learning_rate": 2.8420227361818895e-05,
+      "loss": 0.449,
+      "step": 1386
+    },
+    {
+      "epoch": 1.4677248677248678,
+      "grad_norm": 0.2155045931577355,
+      "learning_rate": 2.8400627205017644e-05,
+      "loss": 0.3958,
+      "step": 1387
+    },
+    {
+      "epoch": 1.4687830687830687,
+      "grad_norm": 34.14419147235214,
+      "learning_rate": 2.8381027048216387e-05,
+      "loss": 0.931,
+      "step": 1388
+    },
+    {
+      "epoch": 1.46984126984127,
+      "grad_norm": 0.2349647452936981,
+      "learning_rate": 2.8361426891415133e-05,
+      "loss": 0.4205,
+      "step": 1389
+    },
+    {
+      "epoch": 1.470899470899471,
+      "grad_norm": 0.4986074723963525,
+      "learning_rate": 2.834182673461388e-05,
+      "loss": 0.4416,
+      "step": 1390
+    },
+    {
+      "epoch": 1.471957671957672,
+      "grad_norm": 0.23543109220326247,
+      "learning_rate": 2.832222657781262e-05,
+      "loss": 0.4605,
+      "step": 1391
+    },
+    {
+      "epoch": 1.4730158730158731,
+      "grad_norm": 0.2041343822972172,
+      "learning_rate": 2.830262642101137e-05,
+      "loss": 0.4152,
+      "step": 1392
+    },
+    {
+      "epoch": 1.474074074074074,
+      "grad_norm": 0.22858488741114016,
+      "learning_rate": 2.8283026264210116e-05,
+      "loss": 0.4099,
+      "step": 1393
+    },
+    {
+      "epoch": 1.475132275132275,
+      "grad_norm": 0.2583668529158323,
+      "learning_rate": 2.826342610740886e-05,
+      "loss": 0.4569,
+      "step": 1394
+    },
+    {
+      "epoch": 1.4761904761904763,
+      "grad_norm": 0.2110916980288495,
+      "learning_rate": 2.8243825950607605e-05,
+      "loss": 0.4281,
+      "step": 1395
+    },
+    {
+      "epoch": 1.4772486772486773,
+      "grad_norm": 0.2335971618684065,
+      "learning_rate": 2.8224225793806354e-05,
+      "loss": 0.4363,
+      "step": 1396
+    },
+    {
+      "epoch": 1.4783068783068782,
+      "grad_norm": 0.24852243373430788,
+      "learning_rate": 2.82046256370051e-05,
+      "loss": 0.4723,
+      "step": 1397
+    },
+    {
+      "epoch": 1.4793650793650794,
+      "grad_norm": 1.7834274229889742,
+      "learning_rate": 2.8185025480203842e-05,
+      "loss": 0.4434,
+      "step": 1398
+    },
+    {
+      "epoch": 1.4804232804232804,
+      "grad_norm": 0.24286885099131655,
+      "learning_rate": 2.8165425323402588e-05,
+      "loss": 0.4564,
+      "step": 1399
+    },
+    {
+      "epoch": 1.4814814814814814,
+      "grad_norm": 0.2609544012499301,
+      "learning_rate": 2.8145825166601337e-05,
+      "loss": 0.4881,
+      "step": 1400
+    },
+    {
+      "epoch": 1.4825396825396826,
+      "grad_norm": 0.2078688963633995,
+      "learning_rate": 2.8126225009800077e-05,
+      "loss": 0.4035,
+      "step": 1401
+    },
+    {
+      "epoch": 1.4835978835978836,
+      "grad_norm": 0.2183047037176754,
+      "learning_rate": 2.8106624852998826e-05,
+      "loss": 0.4606,
+      "step": 1402
+    },
+    {
+      "epoch": 1.4846560846560846,
+      "grad_norm": 0.24125277886676127,
+      "learning_rate": 2.808702469619757e-05,
+      "loss": 0.3907,
+      "step": 1403
+    },
+    {
+      "epoch": 1.4857142857142858,
+      "grad_norm": 0.21563203906268566,
+      "learning_rate": 2.8067424539396314e-05,
+      "loss": 0.3794,
+      "step": 1404
+    },
+    {
+      "epoch": 1.4867724867724867,
+      "grad_norm": 0.23564668760181676,
+      "learning_rate": 2.804782438259506e-05,
+      "loss": 0.3883,
+      "step": 1405
+    },
+    {
+      "epoch": 1.4878306878306877,
+      "grad_norm": 0.22700381452133556,
+      "learning_rate": 2.802822422579381e-05,
+      "loss": 0.4516,
+      "step": 1406
+    },
+    {
+      "epoch": 1.488888888888889,
+      "grad_norm": 0.21734610674555102,
+      "learning_rate": 2.8008624068992552e-05,
+      "loss": 0.4437,
+      "step": 1407
+    },
+    {
+      "epoch": 1.48994708994709,
+      "grad_norm": 0.23992886346228195,
+      "learning_rate": 2.7989023912191298e-05,
+      "loss": 0.4216,
+      "step": 1408
+    },
+    {
+      "epoch": 1.491005291005291,
+      "grad_norm": 0.2628261185467104,
+      "learning_rate": 2.7969423755390044e-05,
+      "loss": 0.5153,
+      "step": 1409
+    },
+    {
+      "epoch": 1.492063492063492,
+      "grad_norm": 0.2224824662272724,
+      "learning_rate": 2.7949823598588793e-05,
+      "loss": 0.4091,
+      "step": 1410
+    },
+    {
+      "epoch": 1.493121693121693,
+      "grad_norm": 0.23493481404403857,
+      "learning_rate": 2.7930223441787535e-05,
+      "loss": 0.4866,
+      "step": 1411
+    },
+    {
+      "epoch": 1.494179894179894,
+      "grad_norm": 0.2833535564343193,
+      "learning_rate": 2.791062328498628e-05,
+      "loss": 0.5036,
+      "step": 1412
+    },
+    {
+      "epoch": 1.4952380952380953,
+      "grad_norm": 0.2436710459859125,
+      "learning_rate": 2.7891023128185027e-05,
+      "loss": 0.4242,
+      "step": 1413
+    },
+    {
+      "epoch": 1.4962962962962962,
+      "grad_norm": 0.2137011724525208,
+      "learning_rate": 2.787142297138377e-05,
+      "loss": 0.4408,
+      "step": 1414
+    },
+    {
+      "epoch": 1.4973544973544972,
+      "grad_norm": 0.21652484554883622,
+      "learning_rate": 2.785182281458252e-05,
+      "loss": 0.3977,
+      "step": 1415
+    },
+    {
+      "epoch": 1.4984126984126984,
+      "grad_norm": 0.22029526293438814,
+      "learning_rate": 2.7832222657781265e-05,
+      "loss": 0.4157,
+      "step": 1416
+    },
+    {
+      "epoch": 1.4994708994708994,
+      "grad_norm": 1.4501431445796187,
+      "learning_rate": 2.7812622500980007e-05,
+      "loss": 0.5586,
+      "step": 1417
+    },
+    {
+      "epoch": 1.5005291005291004,
+      "grad_norm": 0.22081871178721485,
+      "learning_rate": 2.7793022344178753e-05,
+      "loss": 0.388,
+      "step": 1418
+    },
+    {
+      "epoch": 1.5015873015873016,
+      "grad_norm": 0.23512336150560495,
+      "learning_rate": 2.7773422187377502e-05,
+      "loss": 0.5023,
+      "step": 1419
+    },
+    {
+      "epoch": 1.5026455026455028,
+      "grad_norm": 0.20454204650582183,
+      "learning_rate": 2.7753822030576248e-05,
+      "loss": 0.4074,
+      "step": 1420
+    },
+    {
+      "epoch": 1.5037037037037035,
+      "grad_norm": 0.23959488063488388,
+      "learning_rate": 2.773422187377499e-05,
+      "loss": 0.408,
+      "step": 1421
+    },
+    {
+      "epoch": 1.5047619047619047,
+      "grad_norm": 0.21797011158561788,
+      "learning_rate": 2.7714621716973737e-05,
+      "loss": 0.4108,
+      "step": 1422
+    },
+    {
+      "epoch": 1.505820105820106,
+      "grad_norm": 0.2199227438643587,
+      "learning_rate": 2.7695021560172486e-05,
+      "loss": 0.4537,
+      "step": 1423
+    },
+    {
+      "epoch": 1.506878306878307,
+      "grad_norm": 0.2197835414815512,
+      "learning_rate": 2.7675421403371228e-05,
+      "loss": 0.4258,
+      "step": 1424
+    },
+    {
+      "epoch": 1.507936507936508,
+      "grad_norm": 0.2527966111155505,
+      "learning_rate": 2.7655821246569974e-05,
+      "loss": 0.5248,
+      "step": 1425
+    },
+    {
+      "epoch": 1.508994708994709,
+      "grad_norm": 0.2261829434017835,
+      "learning_rate": 2.763622108976872e-05,
+      "loss": 0.4856,
+      "step": 1426
+    },
+    {
+      "epoch": 1.51005291005291,
+      "grad_norm": 0.22385869251344861,
+      "learning_rate": 2.7616620932967463e-05,
+      "loss": 0.4547,
+      "step": 1427
+    },
+    {
+      "epoch": 1.511111111111111,
+      "grad_norm": 0.23763869903387455,
+      "learning_rate": 2.7597020776166212e-05,
+      "loss": 0.4457,
+      "step": 1428
+    },
+    {
+      "epoch": 1.5121693121693123,
+      "grad_norm": 0.24360801662833237,
+      "learning_rate": 2.7577420619364958e-05,
+      "loss": 0.453,
+      "step": 1429
+    },
+    {
+      "epoch": 1.5132275132275133,
+      "grad_norm": 0.21514915559651115,
+      "learning_rate": 2.75578204625637e-05,
+      "loss": 0.4227,
+      "step": 1430
+    },
+    {
+      "epoch": 1.5142857142857142,
+      "grad_norm": 0.25074559433170956,
+      "learning_rate": 2.7538220305762446e-05,
+      "loss": 0.4814,
+      "step": 1431
+    },
+    {
+      "epoch": 1.5153439153439154,
+      "grad_norm": 0.2130270827244537,
+      "learning_rate": 2.7518620148961195e-05,
+      "loss": 0.4009,
+      "step": 1432
+    },
+    {
+      "epoch": 1.5164021164021164,
+      "grad_norm": 0.2152015038995373,
+      "learning_rate": 2.749901999215994e-05,
+      "loss": 0.3733,
+      "step": 1433
+    },
+    {
+      "epoch": 1.5174603174603174,
+      "grad_norm": 0.24155192005071274,
+      "learning_rate": 2.7479419835358684e-05,
+      "loss": 0.3632,
+      "step": 1434
+    },
+    {
+      "epoch": 1.5185185185185186,
+      "grad_norm": 0.2503955177301381,
+      "learning_rate": 2.745981967855743e-05,
+      "loss": 0.3957,
+      "step": 1435
+    },
+    {
+      "epoch": 1.5195767195767196,
+      "grad_norm": 0.21447061332958312,
+      "learning_rate": 2.744021952175618e-05,
+      "loss": 0.4501,
+      "step": 1436
+    },
+    {
+      "epoch": 1.5206349206349206,
+      "grad_norm": 0.2142050063037971,
+      "learning_rate": 2.7420619364954918e-05,
+      "loss": 0.4045,
+      "step": 1437
+    },
+    {
+      "epoch": 1.5216931216931218,
+      "grad_norm": 0.23469888425670518,
+      "learning_rate": 2.7401019208153667e-05,
+      "loss": 0.4069,
+      "step": 1438
+    },
+    {
+      "epoch": 1.5227513227513227,
+      "grad_norm": 0.2373104219367434,
+      "learning_rate": 2.7381419051352413e-05,
+      "loss": 0.4439,
+      "step": 1439
+    },
+    {
+      "epoch": 1.5238095238095237,
+      "grad_norm": 0.22931990044472228,
+      "learning_rate": 2.7361818894551155e-05,
+      "loss": 0.4105,
+      "step": 1440
+    },
+    {
+      "epoch": 1.524867724867725,
+      "grad_norm": 0.24468342878432897,
+      "learning_rate": 2.73422187377499e-05,
+      "loss": 0.4859,
+      "step": 1441
+    },
+    {
+      "epoch": 1.525925925925926,
+      "grad_norm": 0.23888600508604607,
+      "learning_rate": 2.732261858094865e-05,
+      "loss": 0.4692,
+      "step": 1442
+    },
+    {
+      "epoch": 1.5269841269841269,
+      "grad_norm": 0.2379674355949921,
+      "learning_rate": 2.7303018424147393e-05,
+      "loss": 0.4565,
+      "step": 1443
+    },
+    {
+      "epoch": 1.528042328042328,
+      "grad_norm": 0.21540467607096592,
+      "learning_rate": 2.728341826734614e-05,
+      "loss": 0.4183,
+      "step": 1444
+    },
+    {
+      "epoch": 1.529100529100529,
+      "grad_norm": 0.2185049084742189,
+      "learning_rate": 2.7263818110544885e-05,
+      "loss": 0.4343,
+      "step": 1445
+    },
+    {
+      "epoch": 1.53015873015873,
+      "grad_norm": 0.3364663153817796,
+      "learning_rate": 2.7244217953743634e-05,
+      "loss": 0.4324,
+      "step": 1446
+    },
+    {
+      "epoch": 1.5312169312169313,
+      "grad_norm": 0.2053796079743368,
+      "learning_rate": 2.7224617796942377e-05,
+      "loss": 0.3719,
+      "step": 1447
+    },
+    {
+      "epoch": 1.5322751322751322,
+      "grad_norm": 0.22693217543362984,
+      "learning_rate": 2.7205017640141122e-05,
+      "loss": 0.4356,
+      "step": 1448
+    },
+    {
+      "epoch": 1.5333333333333332,
+      "grad_norm": 0.27177436391000964,
+      "learning_rate": 2.718541748333987e-05,
+      "loss": 0.5063,
+      "step": 1449
+    },
+    {
+      "epoch": 1.5343915343915344,
+      "grad_norm": 0.2272754512480649,
+      "learning_rate": 2.716581732653861e-05,
+      "loss": 0.4487,
+      "step": 1450
+    },
+    {
+      "epoch": 1.5354497354497354,
+      "grad_norm": 0.2410315063196938,
+      "learning_rate": 2.714621716973736e-05,
+      "loss": 0.4438,
+      "step": 1451
+    },
+    {
+      "epoch": 1.5365079365079364,
+      "grad_norm": 0.25742696962818656,
+      "learning_rate": 2.7126617012936106e-05,
+      "loss": 0.4149,
+      "step": 1452
+    },
+    {
+      "epoch": 1.5375661375661376,
+      "grad_norm": 0.21908883154703712,
+      "learning_rate": 2.710701685613485e-05,
+      "loss": 0.4035,
+      "step": 1453
+    },
+    {
+      "epoch": 1.5386243386243388,
+      "grad_norm": 0.2720610233232513,
+      "learning_rate": 2.7087416699333594e-05,
+      "loss": 0.4956,
+      "step": 1454
+    },
+    {
+      "epoch": 1.5396825396825395,
+      "grad_norm": 0.25478178075039626,
+      "learning_rate": 2.7067816542532344e-05,
+      "loss": 0.4651,
+      "step": 1455
+    },
+    {
+      "epoch": 1.5407407407407407,
+      "grad_norm": 0.24946076321329155,
+      "learning_rate": 2.7048216385731083e-05,
+      "loss": 0.4157,
+      "step": 1456
+    },
+    {
+      "epoch": 1.541798941798942,
+      "grad_norm": 0.22111397690870904,
+      "learning_rate": 2.7028616228929832e-05,
+      "loss": 0.4587,
+      "step": 1457
+    },
+    {
+      "epoch": 1.5428571428571427,
+      "grad_norm": 0.2324140499446646,
+      "learning_rate": 2.7009016072128578e-05,
+      "loss": 0.405,
+      "step": 1458
+    },
+    {
+      "epoch": 1.543915343915344,
+      "grad_norm": 0.22732415750667306,
+      "learning_rate": 2.6989415915327327e-05,
+      "loss": 0.4055,
+      "step": 1459
+    },
+    {
+      "epoch": 1.544973544973545,
+      "grad_norm": 0.24012853054594216,
+      "learning_rate": 2.696981575852607e-05,
+      "loss": 0.4123,
+      "step": 1460
+    },
+    {
+      "epoch": 1.5460317460317459,
+      "grad_norm": 0.23424926428598633,
+      "learning_rate": 2.6950215601724815e-05,
+      "loss": 0.4752,
+      "step": 1461
+    },
+    {
+      "epoch": 1.547089947089947,
+      "grad_norm": 0.21660322408001612,
+      "learning_rate": 2.693061544492356e-05,
+      "loss": 0.4178,
+      "step": 1462
+    },
+    {
+      "epoch": 1.5481481481481483,
+      "grad_norm": 0.24724861696989267,
+      "learning_rate": 2.6911015288122304e-05,
+      "loss": 0.4814,
+      "step": 1463
+    },
+    {
+      "epoch": 1.5492063492063493,
+      "grad_norm": 2.4202839601143395,
+      "learning_rate": 2.6891415131321053e-05,
+      "loss": 0.4914,
+      "step": 1464
+    },
+    {
+      "epoch": 1.5502645502645502,
+      "grad_norm": 0.2730981950672889,
+      "learning_rate": 2.68718149745198e-05,
+      "loss": 0.4757,
+      "step": 1465
+    },
+    {
+      "epoch": 1.5513227513227514,
+      "grad_norm": 0.25371740302385765,
+      "learning_rate": 2.685221481771854e-05,
+      "loss": 0.4303,
+      "step": 1466
+    },
+    {
+      "epoch": 1.5523809523809524,
+      "grad_norm": 0.23514106327423817,
+      "learning_rate": 2.6832614660917287e-05,
+      "loss": 0.4854,
+      "step": 1467
+    },
+    {
+      "epoch": 1.5534391534391534,
+      "grad_norm": 0.2708610524635199,
+      "learning_rate": 2.6813014504116037e-05,
+      "loss": 0.4422,
+      "step": 1468
+    },
+    {
+      "epoch": 1.5544973544973546,
+      "grad_norm": 0.30542783481347546,
+      "learning_rate": 2.6793414347314782e-05,
+      "loss": 0.528,
+      "step": 1469
+    },
+    {
+      "epoch": 1.5555555555555556,
+      "grad_norm": 0.2966857021941013,
+      "learning_rate": 2.6773814190513525e-05,
+      "loss": 0.5021,
+      "step": 1470
+    },
+    {
+      "epoch": 1.5566137566137566,
+      "grad_norm": 0.2860444368837978,
+      "learning_rate": 2.675421403371227e-05,
+      "loss": 0.4251,
+      "step": 1471
+    },
+    {
+      "epoch": 1.5576719576719578,
+      "grad_norm": 0.28665886400785867,
+      "learning_rate": 2.673461387691102e-05,
+      "loss": 0.4435,
+      "step": 1472
+    },
+    {
+      "epoch": 1.5587301587301587,
+      "grad_norm": 0.2500775008406684,
+      "learning_rate": 2.671501372010976e-05,
+      "loss": 0.4658,
+      "step": 1473
+    },
+    {
+      "epoch": 1.5597883597883597,
+      "grad_norm": 0.22253440194972182,
+      "learning_rate": 2.669541356330851e-05,
+      "loss": 0.4646,
+      "step": 1474
+    },
+    {
+      "epoch": 1.560846560846561,
+      "grad_norm": 0.2146340288566786,
+      "learning_rate": 2.6675813406507254e-05,
+      "loss": 0.3825,
+      "step": 1475
+    },
+    {
+      "epoch": 1.561904761904762,
+      "grad_norm": 0.25064823580585577,
+      "learning_rate": 2.6656213249705997e-05,
+      "loss": 0.4584,
+      "step": 1476
+    },
+    {
+      "epoch": 1.5629629629629629,
+      "grad_norm": 0.2134463383511302,
+      "learning_rate": 2.6636613092904743e-05,
+      "loss": 0.3789,
+      "step": 1477
+    },
+    {
+      "epoch": 1.564021164021164,
+      "grad_norm": 0.2151371289165454,
+      "learning_rate": 2.6617012936103492e-05,
+      "loss": 0.4082,
+      "step": 1478
+    },
+    {
+      "epoch": 1.565079365079365,
+      "grad_norm": 0.211315815255196,
+      "learning_rate": 2.6597412779302234e-05,
+      "loss": 0.4444,
+      "step": 1479
+    },
+    {
+      "epoch": 1.566137566137566,
+      "grad_norm": 0.2638512019499202,
+      "learning_rate": 2.657781262250098e-05,
+      "loss": 0.4507,
+      "step": 1480
+    },
+    {
+      "epoch": 1.5671957671957673,
+      "grad_norm": 0.2249085194794952,
+      "learning_rate": 2.6558212465699726e-05,
+      "loss": 0.4262,
+      "step": 1481
+    },
+    {
+      "epoch": 1.5682539682539682,
+      "grad_norm": 0.22540466561989844,
+      "learning_rate": 2.6538612308898475e-05,
+      "loss": 0.4295,
+      "step": 1482
+    },
+    {
+      "epoch": 1.5693121693121692,
+      "grad_norm": 0.30665171807723735,
+      "learning_rate": 2.6519012152097218e-05,
+      "loss": 0.4453,
+      "step": 1483
+    },
+    {
+      "epoch": 1.5703703703703704,
+      "grad_norm": 0.22641322915874576,
+      "learning_rate": 2.6499411995295964e-05,
+      "loss": 0.4591,
+      "step": 1484
+    },
+    {
+      "epoch": 1.5714285714285714,
+      "grad_norm": 0.23344135403479024,
+      "learning_rate": 2.647981183849471e-05,
+      "loss": 0.4637,
+      "step": 1485
+    },
+    {
+      "epoch": 1.5724867724867724,
+      "grad_norm": 0.22588214050919211,
+      "learning_rate": 2.6460211681693452e-05,
+      "loss": 0.4211,
+      "step": 1486
+    },
+    {
+      "epoch": 1.5735449735449736,
+      "grad_norm": 0.21399920852193247,
+      "learning_rate": 2.64406115248922e-05,
+      "loss": 0.411,
+      "step": 1487
+    },
+    {
+      "epoch": 1.5746031746031746,
+      "grad_norm": 0.23919553348000056,
+      "learning_rate": 2.6421011368090947e-05,
+      "loss": 0.4556,
+      "step": 1488
+    },
+    {
+      "epoch": 1.5756613756613755,
+      "grad_norm": 0.2540160136398132,
+      "learning_rate": 2.640141121128969e-05,
+      "loss": 0.4276,
+      "step": 1489
+    },
+    {
+      "epoch": 1.5767195767195767,
+      "grad_norm": 0.21268692149914978,
+      "learning_rate": 2.6381811054488436e-05,
+      "loss": 0.4493,
+      "step": 1490
+    },
+    {
+      "epoch": 1.5777777777777777,
+      "grad_norm": 0.2504004750600532,
+      "learning_rate": 2.6362210897687185e-05,
+      "loss": 0.4475,
+      "step": 1491
+    },
+    {
+      "epoch": 1.5788359788359787,
+      "grad_norm": 0.23627610401779106,
+      "learning_rate": 2.6342610740885927e-05,
+      "loss": 0.4387,
+      "step": 1492
+    },
+    {
+      "epoch": 1.57989417989418,
+      "grad_norm": 0.24142802272346445,
+      "learning_rate": 2.6323010584084673e-05,
+      "loss": 0.4786,
+      "step": 1493
+    },
+    {
+      "epoch": 1.580952380952381,
+      "grad_norm": 0.2175227164819747,
+      "learning_rate": 2.630341042728342e-05,
+      "loss": 0.3726,
+      "step": 1494
+    },
+    {
+      "epoch": 1.5820105820105819,
+      "grad_norm": 0.2753522156824328,
+      "learning_rate": 2.628381027048217e-05,
+      "loss": 0.476,
+      "step": 1495
+    },
+    {
+      "epoch": 1.583068783068783,
+      "grad_norm": 0.23691656483032675,
+      "learning_rate": 2.626421011368091e-05,
+      "loss": 0.4573,
+      "step": 1496
+    },
+    {
+      "epoch": 1.5841269841269843,
+      "grad_norm": 0.2710431505452728,
+      "learning_rate": 2.6244609956879657e-05,
+      "loss": 0.4574,
+      "step": 1497
+    },
+    {
+      "epoch": 1.585185185185185,
+      "grad_norm": 0.23025660214601665,
+      "learning_rate": 2.6225009800078403e-05,
+      "loss": 0.4509,
+      "step": 1498
+    },
+    {
+      "epoch": 1.5862433862433862,
+      "grad_norm": 0.23275216490060904,
+      "learning_rate": 2.6205409643277145e-05,
+      "loss": 0.4509,
+      "step": 1499
+    },
+    {
+      "epoch": 1.5873015873015874,
+      "grad_norm": 0.2515703737282682,
+      "learning_rate": 2.6185809486475894e-05,
+      "loss": 0.418,
+      "step": 1500
+    },
+    {
+      "epoch": 1.5883597883597882,
+      "grad_norm": 0.22937595837034908,
+      "learning_rate": 2.616620932967464e-05,
+      "loss": 0.4749,
+      "step": 1501
+    },
+    {
+      "epoch": 1.5894179894179894,
+      "grad_norm": 0.22662452164402827,
+      "learning_rate": 2.6146609172873383e-05,
+      "loss": 0.431,
+      "step": 1502
+    },
+    {
+      "epoch": 1.5904761904761906,
+      "grad_norm": 0.24520042229444927,
+      "learning_rate": 2.612700901607213e-05,
+      "loss": 0.4334,
+      "step": 1503
+    },
+    {
+      "epoch": 1.5915343915343916,
+      "grad_norm": 0.21389765221210016,
+      "learning_rate": 2.6107408859270878e-05,
+      "loss": 0.3985,
+      "step": 1504
+    },
+    {
+      "epoch": 1.5925925925925926,
+      "grad_norm": 0.2520836403046001,
+      "learning_rate": 2.6087808702469617e-05,
+      "loss": 0.4952,
+      "step": 1505
+    },
+    {
+      "epoch": 1.5936507936507938,
+      "grad_norm": 0.2590281741764871,
+      "learning_rate": 2.6068208545668366e-05,
+      "loss": 0.4153,
+      "step": 1506
+    },
+    {
+      "epoch": 1.5947089947089947,
+      "grad_norm": 0.24382084508627283,
+      "learning_rate": 2.6048608388867112e-05,
+      "loss": 0.428,
+      "step": 1507
+    },
+    {
+      "epoch": 1.5957671957671957,
+      "grad_norm": 0.2738433714268017,
+      "learning_rate": 2.602900823206586e-05,
+      "loss": 0.5297,
+      "step": 1508
+    },
+    {
+      "epoch": 1.596825396825397,
+      "grad_norm": 0.24412194302796253,
+      "learning_rate": 2.60094080752646e-05,
+      "loss": 0.4653,
+      "step": 1509
+    },
+    {
+      "epoch": 1.597883597883598,
+      "grad_norm": 0.24762704362472204,
+      "learning_rate": 2.598980791846335e-05,
+      "loss": 0.4746,
+      "step": 1510
+    },
+    {
+      "epoch": 1.5989417989417989,
+      "grad_norm": 0.21515171581042802,
+      "learning_rate": 2.5970207761662096e-05,
+      "loss": 0.4151,
+      "step": 1511
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.2500799916715579,
+      "learning_rate": 2.5950607604860838e-05,
+      "loss": 0.417,
+      "step": 1512
+    },
+    {
+      "epoch": 1.601058201058201,
+      "grad_norm": 0.2381688667695587,
+      "learning_rate": 2.5931007448059584e-05,
+      "loss": 0.4448,
+      "step": 1513
+    },
+    {
+      "epoch": 1.602116402116402,
+      "grad_norm": 0.24191214514605877,
+      "learning_rate": 2.5911407291258333e-05,
+      "loss": 0.4683,
+      "step": 1514
+    },
+    {
+      "epoch": 1.6031746031746033,
+      "grad_norm": 0.23784184704388528,
+      "learning_rate": 2.5891807134457076e-05,
+      "loss": 0.4265,
+      "step": 1515
+    },
+    {
+      "epoch": 1.6042328042328042,
+      "grad_norm": 0.2734834110756681,
+      "learning_rate": 2.587220697765582e-05,
+      "loss": 0.439,
+      "step": 1516
+    },
+    {
+      "epoch": 1.6052910052910052,
+      "grad_norm": 0.25396515727811014,
+      "learning_rate": 2.5852606820854568e-05,
+      "loss": 0.461,
+      "step": 1517
+    },
+    {
+      "epoch": 1.6063492063492064,
+      "grad_norm": 0.242809898846282,
+      "learning_rate": 2.5833006664053317e-05,
+      "loss": 0.4573,
+      "step": 1518
+    },
+    {
+      "epoch": 1.6074074074074074,
+      "grad_norm": 0.24330748961980134,
+      "learning_rate": 2.581340650725206e-05,
+      "loss": 0.4954,
+      "step": 1519
+    },
+    {
+      "epoch": 1.6084656084656084,
+      "grad_norm": 0.27485852795834315,
+      "learning_rate": 2.5793806350450805e-05,
+      "loss": 0.428,
+      "step": 1520
+    },
+    {
+      "epoch": 1.6095238095238096,
+      "grad_norm": 0.2445502116760993,
+      "learning_rate": 2.577420619364955e-05,
+      "loss": 0.4789,
+      "step": 1521
+    },
+    {
+      "epoch": 1.6105820105820106,
+      "grad_norm": 0.20220337369742722,
+      "learning_rate": 2.5754606036848293e-05,
+      "loss": 0.4299,
+      "step": 1522
+    },
+    {
+      "epoch": 1.6116402116402115,
+      "grad_norm": 0.2587743856048611,
+      "learning_rate": 2.5735005880047043e-05,
+      "loss": 0.4562,
+      "step": 1523
+    },
+    {
+      "epoch": 1.6126984126984127,
+      "grad_norm": 0.24738675168343857,
+      "learning_rate": 2.571540572324579e-05,
+      "loss": 0.4662,
+      "step": 1524
+    },
+    {
+      "epoch": 1.6137566137566137,
+      "grad_norm": 0.24385074926826564,
+      "learning_rate": 2.569580556644453e-05,
+      "loss": 0.5195,
+      "step": 1525
+    },
+    {
+      "epoch": 1.6148148148148147,
+      "grad_norm": 0.23106098933576363,
+      "learning_rate": 2.5676205409643277e-05,
+      "loss": 0.4307,
+      "step": 1526
+    },
+    {
+      "epoch": 1.615873015873016,
+      "grad_norm": 0.2717986108900208,
+      "learning_rate": 2.5656605252842026e-05,
+      "loss": 0.4243,
+      "step": 1527
+    },
+    {
+      "epoch": 1.6169312169312169,
+      "grad_norm": 0.23631217436069807,
+      "learning_rate": 2.563700509604077e-05,
+      "loss": 0.4001,
+      "step": 1528
+    },
+    {
+      "epoch": 1.6179894179894179,
+      "grad_norm": 0.21316396332645726,
+      "learning_rate": 2.5617404939239515e-05,
+      "loss": 0.4231,
+      "step": 1529
+    },
+    {
+      "epoch": 1.619047619047619,
+      "grad_norm": 0.2529199744670375,
+      "learning_rate": 2.559780478243826e-05,
+      "loss": 0.4383,
+      "step": 1530
+    },
+    {
+      "epoch": 1.6201058201058203,
+      "grad_norm": 0.25696462723939195,
+      "learning_rate": 2.557820462563701e-05,
+      "loss": 0.4237,
+      "step": 1531
+    },
+    {
+      "epoch": 1.621164021164021,
+      "grad_norm": 0.21751147659743908,
+      "learning_rate": 2.5558604468835752e-05,
+      "loss": 0.4341,
+      "step": 1532
+    },
+    {
+      "epoch": 1.6222222222222222,
+      "grad_norm": 0.23705868496582927,
+      "learning_rate": 2.5539004312034498e-05,
+      "loss": 0.4627,
+      "step": 1533
+    },
+    {
+      "epoch": 1.6232804232804234,
+      "grad_norm": 0.228268665473131,
+      "learning_rate": 2.5519404155233244e-05,
+      "loss": 0.4752,
+      "step": 1534
+    },
+    {
+      "epoch": 1.6243386243386242,
+      "grad_norm": 0.22482394544205297,
+      "learning_rate": 2.5499803998431986e-05,
+      "loss": 0.3858,
+      "step": 1535
+    },
+    {
+      "epoch": 1.6253968253968254,
+      "grad_norm": 0.22061168915586668,
+      "learning_rate": 2.5480203841630736e-05,
+      "loss": 0.4521,
+      "step": 1536
+    },
+    {
+      "epoch": 1.6264550264550266,
+      "grad_norm": 0.2187772213768095,
+      "learning_rate": 2.546060368482948e-05,
+      "loss": 0.4239,
+      "step": 1537
+    },
+    {
+      "epoch": 1.6275132275132274,
+      "grad_norm": 0.24727489561841617,
+      "learning_rate": 2.5441003528028224e-05,
+      "loss": 0.4374,
+      "step": 1538
+    },
+    {
+      "epoch": 1.6285714285714286,
+      "grad_norm": 0.22113247609707468,
+      "learning_rate": 2.542140337122697e-05,
+      "loss": 0.4134,
+      "step": 1539
+    },
+    {
+      "epoch": 1.6296296296296298,
+      "grad_norm": 0.19784291165488221,
+      "learning_rate": 2.540180321442572e-05,
+      "loss": 0.4032,
+      "step": 1540
+    },
+    {
+      "epoch": 1.6306878306878307,
+      "grad_norm": 0.2619577804116681,
+      "learning_rate": 2.538220305762446e-05,
+      "loss": 0.4637,
+      "step": 1541
+    },
+    {
+      "epoch": 1.6317460317460317,
+      "grad_norm": 0.22538233425794543,
+      "learning_rate": 2.5362602900823208e-05,
+      "loss": 0.4441,
+      "step": 1542
+    },
+    {
+      "epoch": 1.632804232804233,
+      "grad_norm": 0.24391278793008334,
+      "learning_rate": 2.5343002744021953e-05,
+      "loss": 0.4614,
+      "step": 1543
+    },
+    {
+      "epoch": 1.633862433862434,
+      "grad_norm": 0.20585304887187394,
+      "learning_rate": 2.5323402587220703e-05,
+      "loss": 0.4181,
+      "step": 1544
+    },
+    {
+      "epoch": 1.6349206349206349,
+      "grad_norm": 0.22818882395320167,
+      "learning_rate": 2.5303802430419442e-05,
+      "loss": 0.4149,
+      "step": 1545
+    },
+    {
+      "epoch": 1.635978835978836,
+      "grad_norm": 0.2350090079830974,
+      "learning_rate": 2.528420227361819e-05,
+      "loss": 0.4735,
+      "step": 1546
+    },
+    {
+      "epoch": 1.637037037037037,
+      "grad_norm": 0.22082978235289127,
+      "learning_rate": 2.5264602116816937e-05,
+      "loss": 0.4501,
+      "step": 1547
+    },
+    {
+      "epoch": 1.638095238095238,
+      "grad_norm": 0.21032695498106663,
+      "learning_rate": 2.524500196001568e-05,
+      "loss": 0.4295,
+      "step": 1548
+    },
+    {
+      "epoch": 1.6391534391534393,
+      "grad_norm": 0.21324437007116495,
+      "learning_rate": 2.5225401803214425e-05,
+      "loss": 0.3839,
+      "step": 1549
+    },
+    {
+      "epoch": 1.6402116402116402,
+      "grad_norm": 0.2389234288277766,
+      "learning_rate": 2.5205801646413175e-05,
+      "loss": 0.4673,
+      "step": 1550
+    },
+    {
+      "epoch": 1.6412698412698412,
+      "grad_norm": 0.20583896765782994,
+      "learning_rate": 2.5186201489611917e-05,
+      "loss": 0.4092,
+      "step": 1551
+    },
+    {
+      "epoch": 1.6423280423280424,
+      "grad_norm": 0.22499957708636312,
+      "learning_rate": 2.5166601332810663e-05,
+      "loss": 0.4721,
+      "step": 1552
+    },
+    {
+      "epoch": 1.6433862433862434,
+      "grad_norm": 0.8465391318181628,
+      "learning_rate": 2.514700117600941e-05,
+      "loss": 0.4525,
+      "step": 1553
+    },
+    {
+      "epoch": 1.6444444444444444,
+      "grad_norm": 0.2196056190105969,
+      "learning_rate": 2.512740101920815e-05,
+      "loss": 0.3989,
+      "step": 1554
+    },
+    {
+      "epoch": 1.6455026455026456,
+      "grad_norm": 0.2638083915277794,
+      "learning_rate": 2.51078008624069e-05,
+      "loss": 0.4352,
+      "step": 1555
+    },
+    {
+      "epoch": 1.6465608465608466,
+      "grad_norm": 0.25435791435225086,
+      "learning_rate": 2.5088200705605646e-05,
+      "loss": 0.4399,
+      "step": 1556
+    },
+    {
+      "epoch": 1.6476190476190475,
+      "grad_norm": 0.23162932695663496,
+      "learning_rate": 2.5068600548804392e-05,
+      "loss": 0.4378,
+      "step": 1557
+    },
+    {
+      "epoch": 1.6486772486772487,
+      "grad_norm": 0.24325491659040138,
+      "learning_rate": 2.5049000392003135e-05,
+      "loss": 0.4726,
+      "step": 1558
+    },
+    {
+      "epoch": 1.6497354497354497,
+      "grad_norm": 0.2692936406441785,
+      "learning_rate": 2.5029400235201884e-05,
+      "loss": 0.503,
+      "step": 1559
+    },
+    {
+      "epoch": 1.6507936507936507,
+      "grad_norm": 0.23044637601493173,
+      "learning_rate": 2.500980007840063e-05,
+      "loss": 0.4153,
+      "step": 1560
+    },
+    {
+      "epoch": 1.651851851851852,
+      "grad_norm": 0.19978472969279348,
+      "learning_rate": 2.4990199921599376e-05,
+      "loss": 0.4189,
+      "step": 1561
+    },
+    {
+      "epoch": 1.6529100529100529,
+      "grad_norm": 2.0130289348317545,
+      "learning_rate": 2.497059976479812e-05,
+      "loss": 0.5285,
+      "step": 1562
+    },
+    {
+      "epoch": 1.6539682539682539,
+      "grad_norm": 0.6382100207364493,
+      "learning_rate": 2.4950999607996864e-05,
+      "loss": 0.498,
+      "step": 1563
+    },
+    {
+      "epoch": 1.655026455026455,
+      "grad_norm": 0.2873980573844794,
+      "learning_rate": 2.493139945119561e-05,
+      "loss": 0.4345,
+      "step": 1564
+    },
+    {
+      "epoch": 1.656084656084656,
+      "grad_norm": 0.2238132320590369,
+      "learning_rate": 2.4911799294394356e-05,
+      "loss": 0.4107,
+      "step": 1565
+    },
+    {
+      "epoch": 1.657142857142857,
+      "grad_norm": 0.22847079158253494,
+      "learning_rate": 2.4892199137593102e-05,
+      "loss": 0.4377,
+      "step": 1566
+    },
+    {
+      "epoch": 1.6582010582010582,
+      "grad_norm": 0.27256045565172915,
+      "learning_rate": 2.4872598980791848e-05,
+      "loss": 0.4313,
+      "step": 1567
+    },
+    {
+      "epoch": 1.6592592592592592,
+      "grad_norm": 0.2586293011878613,
+      "learning_rate": 2.4852998823990594e-05,
+      "loss": 0.4437,
+      "step": 1568
+    },
+    {
+      "epoch": 1.6603174603174602,
+      "grad_norm": 0.22800723424295755,
+      "learning_rate": 2.483339866718934e-05,
+      "loss": 0.4273,
+      "step": 1569
+    },
+    {
+      "epoch": 1.6613756613756614,
+      "grad_norm": 0.24485218448636267,
+      "learning_rate": 2.4813798510388085e-05,
+      "loss": 0.4393,
+      "step": 1570
+    },
+    {
+      "epoch": 1.6624338624338626,
+      "grad_norm": 0.2484620989572041,
+      "learning_rate": 2.479419835358683e-05,
+      "loss": 0.3975,
+      "step": 1571
+    },
+    {
+      "epoch": 1.6634920634920634,
+      "grad_norm": 0.2124965525509564,
+      "learning_rate": 2.4774598196785577e-05,
+      "loss": 0.4508,
+      "step": 1572
+    },
+    {
+      "epoch": 1.6645502645502646,
+      "grad_norm": 0.24457740342964363,
+      "learning_rate": 2.475499803998432e-05,
+      "loss": 0.4299,
+      "step": 1573
+    },
+    {
+      "epoch": 1.6656084656084658,
+      "grad_norm": 0.25001350248928084,
+      "learning_rate": 2.473539788318307e-05,
+      "loss": 0.4924,
+      "step": 1574
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.2307802816899225,
+      "learning_rate": 2.471579772638181e-05,
+      "loss": 0.4466,
+      "step": 1575
+    },
+    {
+      "epoch": 1.6677248677248677,
+      "grad_norm": 0.22210528313961542,
+      "learning_rate": 2.469619756958056e-05,
+      "loss": 0.404,
+      "step": 1576
+    },
+    {
+      "epoch": 1.668783068783069,
+      "grad_norm": 0.21443376857710403,
+      "learning_rate": 2.4676597412779303e-05,
+      "loss": 0.4274,
+      "step": 1577
+    },
+    {
+      "epoch": 1.6698412698412697,
+      "grad_norm": 0.22624353684944185,
+      "learning_rate": 2.465699725597805e-05,
+      "loss": 0.436,
+      "step": 1578
+    },
+    {
+      "epoch": 1.6708994708994709,
+      "grad_norm": 0.23099050262288628,
+      "learning_rate": 2.4637397099176795e-05,
+      "loss": 0.4582,
+      "step": 1579
+    },
+    {
+      "epoch": 1.671957671957672,
+      "grad_norm": 0.20700052537048644,
+      "learning_rate": 2.461779694237554e-05,
+      "loss": 0.4105,
+      "step": 1580
+    },
+    {
+      "epoch": 1.673015873015873,
+      "grad_norm": 0.21408619017989325,
+      "learning_rate": 2.4598196785574283e-05,
+      "loss": 0.4435,
+      "step": 1581
+    },
+    {
+      "epoch": 1.674074074074074,
+      "grad_norm": 0.22932373910076814,
+      "learning_rate": 2.4578596628773032e-05,
+      "loss": 0.4423,
+      "step": 1582
+    },
+    {
+      "epoch": 1.6751322751322753,
+      "grad_norm": 0.2701302455070178,
+      "learning_rate": 2.4558996471971775e-05,
+      "loss": 0.491,
+      "step": 1583
+    },
+    {
+      "epoch": 1.6761904761904762,
+      "grad_norm": 0.22544337265587341,
+      "learning_rate": 2.4539396315170524e-05,
+      "loss": 0.4743,
+      "step": 1584
+    },
+    {
+      "epoch": 1.6772486772486772,
+      "grad_norm": 0.21709705988111216,
+      "learning_rate": 2.4519796158369267e-05,
+      "loss": 0.4065,
+      "step": 1585
+    },
+    {
+      "epoch": 1.6783068783068784,
+      "grad_norm": 0.24996850315949026,
+      "learning_rate": 2.4500196001568013e-05,
+      "loss": 0.4375,
+      "step": 1586
+    },
+    {
+      "epoch": 1.6793650793650794,
+      "grad_norm": 0.9469921058616022,
+      "learning_rate": 2.448059584476676e-05,
+      "loss": 0.4191,
+      "step": 1587
+    },
+    {
+      "epoch": 1.6804232804232804,
+      "grad_norm": 0.22085668242729445,
+      "learning_rate": 2.4460995687965504e-05,
+      "loss": 0.48,
+      "step": 1588
+    },
+    {
+      "epoch": 1.6814814814814816,
+      "grad_norm": 0.23494830806105532,
+      "learning_rate": 2.444139553116425e-05,
+      "loss": 0.4781,
+      "step": 1589
+    },
+    {
+      "epoch": 1.6825396825396826,
+      "grad_norm": 0.26542532348970826,
+      "learning_rate": 2.4421795374362996e-05,
+      "loss": 0.4089,
+      "step": 1590
+    },
+    {
+      "epoch": 1.6835978835978835,
+      "grad_norm": 0.25254211616363226,
+      "learning_rate": 2.4402195217561742e-05,
+      "loss": 0.4629,
+      "step": 1591
+    },
+    {
+      "epoch": 1.6846560846560847,
+      "grad_norm": 0.2273568902013295,
+      "learning_rate": 2.4382595060760488e-05,
+      "loss": 0.4592,
+      "step": 1592
+    },
+    {
+      "epoch": 1.6857142857142857,
+      "grad_norm": 0.2594410680121094,
+      "learning_rate": 2.4362994903959234e-05,
+      "loss": 0.439,
+      "step": 1593
+    },
+    {
+      "epoch": 1.6867724867724867,
+      "grad_norm": 0.2274722145506129,
+      "learning_rate": 2.4343394747157976e-05,
+      "loss": 0.4246,
+      "step": 1594
+    },
+    {
+      "epoch": 1.687830687830688,
+      "grad_norm": 0.2400296192105033,
+      "learning_rate": 2.4323794590356725e-05,
+      "loss": 0.4985,
+      "step": 1595
+    },
+    {
+      "epoch": 1.6888888888888889,
+      "grad_norm": 0.2335755377011187,
+      "learning_rate": 2.4304194433555468e-05,
+      "loss": 0.5176,
+      "step": 1596
+    },
+    {
+      "epoch": 1.6899470899470899,
+      "grad_norm": 0.22541799468462043,
+      "learning_rate": 2.4284594276754217e-05,
+      "loss": 0.391,
+      "step": 1597
+    },
+    {
+      "epoch": 1.691005291005291,
+      "grad_norm": 0.25348107176315676,
+      "learning_rate": 2.426499411995296e-05,
+      "loss": 0.4718,
+      "step": 1598
+    },
+    {
+      "epoch": 1.692063492063492,
+      "grad_norm": 0.2134398598463444,
+      "learning_rate": 2.4245393963151706e-05,
+      "loss": 0.4563,
+      "step": 1599
+    },
+    {
+      "epoch": 1.693121693121693,
+      "grad_norm": 0.2799436267035962,
+      "learning_rate": 2.422579380635045e-05,
+      "loss": 0.4554,
+      "step": 1600
+    },
+    {
+      "epoch": 1.6941798941798942,
+      "grad_norm": 0.2239050910441869,
+      "learning_rate": 2.4206193649549197e-05,
+      "loss": 0.4303,
+      "step": 1601
+    },
+    {
+      "epoch": 1.6952380952380952,
+      "grad_norm": 5.342628537787954,
+      "learning_rate": 2.4186593492747943e-05,
+      "loss": 0.4961,
+      "step": 1602
+    },
+    {
+      "epoch": 1.6962962962962962,
+      "grad_norm": 1.9189218328198483,
+      "learning_rate": 2.416699333594669e-05,
+      "loss": 0.5465,
+      "step": 1603
+    },
+    {
+      "epoch": 1.6973544973544974,
+      "grad_norm": 0.22908151431119422,
+      "learning_rate": 2.4147393179145435e-05,
+      "loss": 0.4628,
+      "step": 1604
+    },
+    {
+      "epoch": 1.6984126984126984,
+      "grad_norm": 0.20485996984971452,
+      "learning_rate": 2.412779302234418e-05,
+      "loss": 0.4013,
+      "step": 1605
+    },
+    {
+      "epoch": 1.6994708994708994,
+      "grad_norm": 0.21384044603948377,
+      "learning_rate": 2.4108192865542927e-05,
+      "loss": 0.4111,
+      "step": 1606
+    },
+    {
+      "epoch": 1.7005291005291006,
+      "grad_norm": 0.20944814894771038,
+      "learning_rate": 2.4088592708741673e-05,
+      "loss": 0.4487,
+      "step": 1607
+    },
+    {
+      "epoch": 1.7015873015873015,
+      "grad_norm": 0.21567540070670468,
+      "learning_rate": 2.406899255194042e-05,
+      "loss": 0.4157,
+      "step": 1608
+    },
+    {
+      "epoch": 1.7026455026455025,
+      "grad_norm": 0.21951670349669852,
+      "learning_rate": 2.404939239513916e-05,
+      "loss": 0.3811,
+      "step": 1609
+    },
+    {
+      "epoch": 1.7037037037037037,
+      "grad_norm": 0.21975151815133592,
+      "learning_rate": 2.402979223833791e-05,
+      "loss": 0.4788,
+      "step": 1610
+    },
+    {
+      "epoch": 1.704761904761905,
+      "grad_norm": 0.20831580759962212,
+      "learning_rate": 2.4010192081536653e-05,
+      "loss": 0.4237,
+      "step": 1611
+    },
+    {
+      "epoch": 1.7058201058201057,
+      "grad_norm": 0.20479233675329178,
+      "learning_rate": 2.39905919247354e-05,
+      "loss": 0.4125,
+      "step": 1612
+    },
+    {
+      "epoch": 1.7068783068783069,
+      "grad_norm": 0.21043974195059237,
+      "learning_rate": 2.3970991767934144e-05,
+      "loss": 0.4114,
+      "step": 1613
+    },
+    {
+      "epoch": 1.707936507936508,
+      "grad_norm": 0.19363060033043297,
+      "learning_rate": 2.395139161113289e-05,
+      "loss": 0.3958,
+      "step": 1614
+    },
+    {
+      "epoch": 1.7089947089947088,
+      "grad_norm": 0.2203472309140231,
+      "learning_rate": 2.3931791454331636e-05,
+      "loss": 0.4136,
+      "step": 1615
+    },
+    {
+      "epoch": 1.71005291005291,
+      "grad_norm": 0.27731224351626677,
+      "learning_rate": 2.3912191297530382e-05,
+      "loss": 0.428,
+      "step": 1616
+    },
+    {
+      "epoch": 1.7111111111111112,
+      "grad_norm": 0.2165312271941672,
+      "learning_rate": 2.3892591140729124e-05,
+      "loss": 0.4455,
+      "step": 1617
+    },
+    {
+      "epoch": 1.712169312169312,
+      "grad_norm": 0.21755495769093408,
+      "learning_rate": 2.3872990983927874e-05,
+      "loss": 0.4031,
+      "step": 1618
+    },
+    {
+      "epoch": 1.7132275132275132,
+      "grad_norm": 0.24222900168163267,
+      "learning_rate": 2.3853390827126616e-05,
+      "loss": 0.4786,
+      "step": 1619
+    },
+    {
+      "epoch": 1.7142857142857144,
+      "grad_norm": 0.23418054019764165,
+      "learning_rate": 2.3833790670325365e-05,
+      "loss": 0.4608,
+      "step": 1620
+    },
+    {
+      "epoch": 1.7153439153439154,
+      "grad_norm": 0.21694499784225765,
+      "learning_rate": 2.3814190513524108e-05,
+      "loss": 0.4283,
+      "step": 1621
+    },
+    {
+      "epoch": 1.7164021164021164,
+      "grad_norm": 0.2546334351383069,
+      "learning_rate": 2.3794590356722854e-05,
+      "loss": 0.4606,
+      "step": 1622
+    },
+    {
+      "epoch": 1.7174603174603176,
+      "grad_norm": 0.2319584310490375,
+      "learning_rate": 2.37749901999216e-05,
+      "loss": 0.451,
+      "step": 1623
+    },
+    {
+      "epoch": 1.7185185185185186,
+      "grad_norm": 0.24477601621399064,
+      "learning_rate": 2.3755390043120346e-05,
+      "loss": 0.5059,
+      "step": 1624
+    },
+    {
+      "epoch": 1.7195767195767195,
+      "grad_norm": 0.36838675048518105,
+      "learning_rate": 2.373578988631909e-05,
+      "loss": 0.4285,
+      "step": 1625
+    },
+    {
+      "epoch": 1.7206349206349207,
+      "grad_norm": 0.2466732852047628,
+      "learning_rate": 2.3716189729517837e-05,
+      "loss": 0.4437,
+      "step": 1626
+    },
+    {
+      "epoch": 1.7216931216931217,
+      "grad_norm": 0.2135957001631125,
+      "learning_rate": 2.3696589572716583e-05,
+      "loss": 0.421,
+      "step": 1627
+    },
+    {
+      "epoch": 1.7227513227513227,
+      "grad_norm": 0.24026169452843113,
+      "learning_rate": 2.367698941591533e-05,
+      "loss": 0.4532,
+      "step": 1628
+    },
+    {
+      "epoch": 1.723809523809524,
+      "grad_norm": 0.22254374311366967,
+      "learning_rate": 2.3657389259114075e-05,
+      "loss": 0.4547,
+      "step": 1629
+    },
+    {
+      "epoch": 1.7248677248677249,
+      "grad_norm": 0.2379541965307219,
+      "learning_rate": 2.3637789102312817e-05,
+      "loss": 0.4735,
+      "step": 1630
+    },
+    {
+      "epoch": 1.7259259259259259,
+      "grad_norm": 0.21785973896658944,
+      "learning_rate": 2.3618188945511567e-05,
+      "loss": 0.4757,
+      "step": 1631
+    },
+    {
+      "epoch": 1.726984126984127,
+      "grad_norm": 0.20413785133119947,
+      "learning_rate": 2.359858878871031e-05,
+      "loss": 0.4213,
+      "step": 1632
+    },
+    {
+      "epoch": 1.728042328042328,
+      "grad_norm": 0.25684940141007967,
+      "learning_rate": 2.357898863190906e-05,
+      "loss": 0.4839,
+      "step": 1633
+    },
+    {
+      "epoch": 1.729100529100529,
+      "grad_norm": 0.22575583816002107,
+      "learning_rate": 2.35593884751078e-05,
+      "loss": 0.4319,
+      "step": 1634
+    },
+    {
+      "epoch": 1.7301587301587302,
+      "grad_norm": 0.2102700377358579,
+      "learning_rate": 2.3539788318306547e-05,
+      "loss": 0.4465,
+      "step": 1635
+    },
+    {
+      "epoch": 1.7312169312169312,
+      "grad_norm": 0.23871894363562643,
+      "learning_rate": 2.3520188161505293e-05,
+      "loss": 0.4692,
+      "step": 1636
+    },
+    {
+      "epoch": 1.7322751322751322,
+      "grad_norm": 0.2257335899512697,
+      "learning_rate": 2.350058800470404e-05,
+      "loss": 0.4308,
+      "step": 1637
+    },
+    {
+      "epoch": 1.7333333333333334,
+      "grad_norm": 0.21199670925667166,
+      "learning_rate": 2.3480987847902784e-05,
+      "loss": 0.4399,
+      "step": 1638
+    },
+    {
+      "epoch": 1.7343915343915344,
+      "grad_norm": 0.2637943802055583,
+      "learning_rate": 2.346138769110153e-05,
+      "loss": 0.469,
+      "step": 1639
+    },
+    {
+      "epoch": 1.7354497354497354,
+      "grad_norm": 0.2091292729089544,
+      "learning_rate": 2.3441787534300276e-05,
+      "loss": 0.4079,
+      "step": 1640
+    },
+    {
+      "epoch": 1.7365079365079366,
+      "grad_norm": 0.21242035801784248,
+      "learning_rate": 2.3422187377499022e-05,
+      "loss": 0.433,
+      "step": 1641
+    },
+    {
+      "epoch": 1.7375661375661375,
+      "grad_norm": 0.21411367696110664,
+      "learning_rate": 2.3402587220697768e-05,
+      "loss": 0.445,
+      "step": 1642
+    },
+    {
+      "epoch": 1.7386243386243385,
+      "grad_norm": 0.22666171986236078,
+      "learning_rate": 2.338298706389651e-05,
+      "loss": 0.4352,
+      "step": 1643
+    },
+    {
+      "epoch": 1.7396825396825397,
+      "grad_norm": 0.2154746362675361,
+      "learning_rate": 2.336338690709526e-05,
+      "loss": 0.4225,
+      "step": 1644
+    },
+    {
+      "epoch": 1.7407407407407407,
+      "grad_norm": 0.20835776969269476,
+      "learning_rate": 2.3343786750294002e-05,
+      "loss": 0.423,
+      "step": 1645
+    },
+    {
+      "epoch": 1.7417989417989417,
+      "grad_norm": 0.22888559147228213,
+      "learning_rate": 2.332418659349275e-05,
+      "loss": 0.4047,
+      "step": 1646
+    },
+    {
+      "epoch": 1.7428571428571429,
+      "grad_norm": 0.2211280651344821,
+      "learning_rate": 2.3304586436691494e-05,
+      "loss": 0.3904,
+      "step": 1647
+    },
+    {
+      "epoch": 1.7439153439153439,
+      "grad_norm": 0.23013001218859871,
+      "learning_rate": 2.328498627989024e-05,
+      "loss": 0.4983,
+      "step": 1648
+    },
+    {
+      "epoch": 1.7449735449735448,
+      "grad_norm": 0.2202927426992636,
+      "learning_rate": 2.3265386123088986e-05,
+      "loss": 0.4676,
+      "step": 1649
+    },
+    {
+      "epoch": 1.746031746031746,
+      "grad_norm": 0.22953761733384298,
+      "learning_rate": 2.324578596628773e-05,
+      "loss": 0.45,
+      "step": 1650
+    },
+    {
+      "epoch": 1.7470899470899472,
+      "grad_norm": 0.22282806910302103,
+      "learning_rate": 2.3226185809486477e-05,
+      "loss": 0.4218,
+      "step": 1651
+    },
+    {
+      "epoch": 1.748148148148148,
+      "grad_norm": 0.23634054688203343,
+      "learning_rate": 2.3206585652685223e-05,
+      "loss": 0.464,
+      "step": 1652
+    },
+    {
+      "epoch": 1.7492063492063492,
+      "grad_norm": 0.22465725068012996,
+      "learning_rate": 2.3186985495883966e-05,
+      "loss": 0.448,
+      "step": 1653
+    },
+    {
+      "epoch": 1.7502645502645504,
+      "grad_norm": 0.23369312668049108,
+      "learning_rate": 2.3167385339082715e-05,
+      "loss": 0.4475,
+      "step": 1654
+    },
+    {
+      "epoch": 1.7513227513227512,
+      "grad_norm": 0.21668131882841285,
+      "learning_rate": 2.3147785182281458e-05,
+      "loss": 0.4246,
+      "step": 1655
+    },
+    {
+      "epoch": 1.7523809523809524,
+      "grad_norm": 0.22432886200596663,
+      "learning_rate": 2.3128185025480207e-05,
+      "loss": 0.4726,
+      "step": 1656
+    },
+    {
+      "epoch": 1.7534391534391536,
+      "grad_norm": 0.23217067772618433,
+      "learning_rate": 2.310858486867895e-05,
+      "loss": 0.4217,
+      "step": 1657
+    },
+    {
+      "epoch": 1.7544973544973543,
+      "grad_norm": 0.23998057651431876,
+      "learning_rate": 2.3088984711877695e-05,
+      "loss": 0.4478,
+      "step": 1658
+    },
+    {
+      "epoch": 1.7555555555555555,
+      "grad_norm": 0.20789843322357746,
+      "learning_rate": 2.306938455507644e-05,
+      "loss": 0.4093,
+      "step": 1659
+    },
+    {
+      "epoch": 1.7566137566137567,
+      "grad_norm": 0.18321420188469675,
+      "learning_rate": 2.3049784398275187e-05,
+      "loss": 0.3687,
+      "step": 1660
+    },
+    {
+      "epoch": 1.7576719576719577,
+      "grad_norm": 0.24493094649795308,
+      "learning_rate": 2.3030184241473933e-05,
+      "loss": 0.4494,
+      "step": 1661
+    },
+    {
+      "epoch": 1.7587301587301587,
+      "grad_norm": 0.2091468538339814,
+      "learning_rate": 2.301058408467268e-05,
+      "loss": 0.4199,
+      "step": 1662
+    },
+    {
+      "epoch": 1.75978835978836,
+      "grad_norm": 0.2071184685848974,
+      "learning_rate": 2.2990983927871425e-05,
+      "loss": 0.4324,
+      "step": 1663
+    },
+    {
+      "epoch": 1.7608465608465609,
+      "grad_norm": 0.24913461114215388,
+      "learning_rate": 2.297138377107017e-05,
+      "loss": 0.467,
+      "step": 1664
+    },
+    {
+      "epoch": 1.7619047619047619,
+      "grad_norm": 0.2391544344366436,
+      "learning_rate": 2.2951783614268916e-05,
+      "loss": 0.4312,
+      "step": 1665
+    },
+    {
+      "epoch": 1.762962962962963,
+      "grad_norm": 0.22154350725206487,
+      "learning_rate": 2.293218345746766e-05,
+      "loss": 0.4597,
+      "step": 1666
+    },
+    {
+      "epoch": 1.764021164021164,
+      "grad_norm": 0.24501309931712809,
+      "learning_rate": 2.2912583300666408e-05,
+      "loss": 0.459,
+      "step": 1667
+    },
+    {
+      "epoch": 1.765079365079365,
+      "grad_norm": 0.25526129904002803,
+      "learning_rate": 2.289298314386515e-05,
+      "loss": 0.4835,
+      "step": 1668
+    },
+    {
+      "epoch": 1.7661375661375662,
+      "grad_norm": 0.21646297720766336,
+      "learning_rate": 2.28733829870639e-05,
+      "loss": 0.4335,
+      "step": 1669
+    },
+    {
+      "epoch": 1.7671957671957672,
+      "grad_norm": 0.22733884215028632,
+      "learning_rate": 2.2853782830262642e-05,
+      "loss": 0.4552,
+      "step": 1670
+    },
+    {
+      "epoch": 1.7682539682539682,
+      "grad_norm": 0.2075067401096603,
+      "learning_rate": 2.2834182673461388e-05,
+      "loss": 0.4138,
+      "step": 1671
+    },
+    {
+      "epoch": 1.7693121693121694,
+      "grad_norm": 0.2257968183277266,
+      "learning_rate": 2.2814582516660134e-05,
+      "loss": 0.5253,
+      "step": 1672
+    },
+    {
+      "epoch": 1.7703703703703704,
+      "grad_norm": 0.21533345478212504,
+      "learning_rate": 2.279498235985888e-05,
+      "loss": 0.4942,
+      "step": 1673
+    },
+    {
+      "epoch": 1.7714285714285714,
+      "grad_norm": 0.22258782531542667,
+      "learning_rate": 2.2775382203057626e-05,
+      "loss": 0.4511,
+      "step": 1674
+    },
+    {
+      "epoch": 1.7724867724867726,
+      "grad_norm": 0.2323804687576553,
+      "learning_rate": 2.275578204625637e-05,
+      "loss": 0.439,
+      "step": 1675
+    },
+    {
+      "epoch": 1.7735449735449735,
+      "grad_norm": 0.22142223429635433,
+      "learning_rate": 2.2736181889455118e-05,
+      "loss": 0.4404,
+      "step": 1676
+    },
+    {
+      "epoch": 1.7746031746031745,
+      "grad_norm": 0.24472248169409241,
+      "learning_rate": 2.2716581732653863e-05,
+      "loss": 0.4276,
+      "step": 1677
+    },
+    {
+      "epoch": 1.7756613756613757,
+      "grad_norm": 0.49844277498031275,
+      "learning_rate": 2.269698157585261e-05,
+      "loss": 0.3956,
+      "step": 1678
+    },
+    {
+      "epoch": 1.7767195767195767,
+      "grad_norm": 0.228348531852297,
+      "learning_rate": 2.2677381419051352e-05,
+      "loss": 0.4511,
+      "step": 1679
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "grad_norm": 0.2886498501182977,
+      "learning_rate": 2.26577812622501e-05,
+      "loss": 0.4456,
+      "step": 1680
+    },
+    {
+      "epoch": 1.7788359788359789,
+      "grad_norm": 4.492214718550815,
+      "learning_rate": 2.2638181105448844e-05,
+      "loss": 0.5011,
+      "step": 1681
+    },
+    {
+      "epoch": 1.7798941798941799,
+      "grad_norm": 0.24204760220136187,
+      "learning_rate": 2.2618580948647593e-05,
+      "loss": 0.4279,
+      "step": 1682
+    },
+    {
+      "epoch": 1.7809523809523808,
+      "grad_norm": 0.23756281310500446,
+      "learning_rate": 2.2598980791846335e-05,
+      "loss": 0.4534,
+      "step": 1683
+    },
+    {
+      "epoch": 1.782010582010582,
+      "grad_norm": 0.21980783401447584,
+      "learning_rate": 2.257938063504508e-05,
+      "loss": 0.4446,
+      "step": 1684
+    },
+    {
+      "epoch": 1.783068783068783,
+      "grad_norm": 0.2087115413343064,
+      "learning_rate": 2.2559780478243827e-05,
+      "loss": 0.3605,
+      "step": 1685
+    },
+    {
+      "epoch": 1.784126984126984,
+      "grad_norm": 0.2543465932738077,
+      "learning_rate": 2.2540180321442573e-05,
+      "loss": 0.4842,
+      "step": 1686
+    },
+    {
+      "epoch": 1.7851851851851852,
+      "grad_norm": 0.22912923330057322,
+      "learning_rate": 2.252058016464132e-05,
+      "loss": 0.486,
+      "step": 1687
+    },
+    {
+      "epoch": 1.7862433862433862,
+      "grad_norm": 0.21781329156771057,
+      "learning_rate": 2.2500980007840065e-05,
+      "loss": 0.4683,
+      "step": 1688
+    },
+    {
+      "epoch": 1.7873015873015872,
+      "grad_norm": 0.21952765550999895,
+      "learning_rate": 2.2481379851038807e-05,
+      "loss": 0.449,
+      "step": 1689
+    },
+    {
+      "epoch": 1.7883597883597884,
+      "grad_norm": 0.2419599965780607,
+      "learning_rate": 2.2461779694237556e-05,
+      "loss": 0.5227,
+      "step": 1690
+    },
+    {
+      "epoch": 1.7894179894179896,
+      "grad_norm": 0.23359674854150045,
+      "learning_rate": 2.24421795374363e-05,
+      "loss": 0.4485,
+      "step": 1691
+    },
+    {
+      "epoch": 1.7904761904761903,
+      "grad_norm": 0.21983049948624178,
+      "learning_rate": 2.2422579380635045e-05,
+      "loss": 0.4553,
+      "step": 1692
+    },
+    {
+      "epoch": 1.7915343915343915,
+      "grad_norm": 0.22263523405847083,
+      "learning_rate": 2.240297922383379e-05,
+      "loss": 0.4627,
+      "step": 1693
+    },
+    {
+      "epoch": 1.7925925925925927,
+      "grad_norm": 0.21875294678933027,
+      "learning_rate": 2.2383379067032536e-05,
+      "loss": 0.452,
+      "step": 1694
+    },
+    {
+      "epoch": 1.7936507936507935,
+      "grad_norm": 0.2190892904324516,
+      "learning_rate": 2.2363778910231282e-05,
+      "loss": 0.4176,
+      "step": 1695
+    },
+    {
+      "epoch": 1.7947089947089947,
+      "grad_norm": 0.20100867962420854,
+      "learning_rate": 2.2344178753430028e-05,
+      "loss": 0.4424,
+      "step": 1696
+    },
+    {
+      "epoch": 1.795767195767196,
+      "grad_norm": 0.22957554331685712,
+      "learning_rate": 2.2324578596628774e-05,
+      "loss": 0.471,
+      "step": 1697
+    },
+    {
+      "epoch": 1.7968253968253967,
+      "grad_norm": 0.22879161164574066,
+      "learning_rate": 2.230497843982752e-05,
+      "loss": 0.497,
+      "step": 1698
+    },
+    {
+      "epoch": 1.7978835978835979,
+      "grad_norm": 0.19578829621547492,
+      "learning_rate": 2.2285378283026266e-05,
+      "loss": 0.3854,
+      "step": 1699
+    },
+    {
+      "epoch": 1.798941798941799,
+      "grad_norm": 0.21006387291689507,
+      "learning_rate": 2.2265778126225012e-05,
+      "loss": 0.4098,
+      "step": 1700
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.23977296269604864,
+      "learning_rate": 2.2246177969423758e-05,
+      "loss": 0.4905,
+      "step": 1701
+    },
+    {
+      "epoch": 1.801058201058201,
+      "grad_norm": 0.2105622462139618,
+      "learning_rate": 2.22265778126225e-05,
+      "loss": 0.4283,
+      "step": 1702
+    },
+    {
+      "epoch": 1.8021164021164022,
+      "grad_norm": 0.22056822514598742,
+      "learning_rate": 2.220697765582125e-05,
+      "loss": 0.4205,
+      "step": 1703
+    },
+    {
+      "epoch": 1.8031746031746032,
+      "grad_norm": 0.20582405401017348,
+      "learning_rate": 2.2187377499019992e-05,
+      "loss": 0.4471,
+      "step": 1704
+    },
+    {
+      "epoch": 1.8042328042328042,
+      "grad_norm": 0.24182764269625914,
+      "learning_rate": 2.216777734221874e-05,
+      "loss": 0.4553,
+      "step": 1705
+    },
+    {
+      "epoch": 1.8052910052910054,
+      "grad_norm": 0.20679090700136382,
+      "learning_rate": 2.2148177185417484e-05,
+      "loss": 0.451,
+      "step": 1706
+    },
+    {
+      "epoch": 1.8063492063492064,
+      "grad_norm": 0.19092027693402705,
+      "learning_rate": 2.212857702861623e-05,
+      "loss": 0.3698,
+      "step": 1707
+    },
+    {
+      "epoch": 1.8074074074074074,
+      "grad_norm": 0.23205314819347692,
+      "learning_rate": 2.2108976871814975e-05,
+      "loss": 0.4511,
+      "step": 1708
+    },
+    {
+      "epoch": 1.8084656084656086,
+      "grad_norm": 0.21244419197626072,
+      "learning_rate": 2.208937671501372e-05,
+      "loss": 0.426,
+      "step": 1709
+    },
+    {
+      "epoch": 1.8095238095238095,
+      "grad_norm": 0.22435395018849072,
+      "learning_rate": 2.2069776558212467e-05,
+      "loss": 0.4626,
+      "step": 1710
+    },
+    {
+      "epoch": 1.8105820105820105,
+      "grad_norm": 0.24022212831540976,
+      "learning_rate": 2.2050176401411213e-05,
+      "loss": 0.4383,
+      "step": 1711
+    },
+    {
+      "epoch": 1.8116402116402117,
+      "grad_norm": 0.20705066488075852,
+      "learning_rate": 2.203057624460996e-05,
+      "loss": 0.4122,
+      "step": 1712
+    },
+    {
+      "epoch": 1.8126984126984127,
+      "grad_norm": 0.22592755674231862,
+      "learning_rate": 2.2010976087808705e-05,
+      "loss": 0.4842,
+      "step": 1713
+    },
+    {
+      "epoch": 1.8137566137566137,
+      "grad_norm": 0.21271788549637888,
+      "learning_rate": 2.199137593100745e-05,
+      "loss": 0.4389,
+      "step": 1714
+    },
+    {
+      "epoch": 1.8148148148148149,
+      "grad_norm": 0.23889518702932439,
+      "learning_rate": 2.1971775774206193e-05,
+      "loss": 0.4717,
+      "step": 1715
+    },
+    {
+      "epoch": 1.8158730158730159,
+      "grad_norm": 0.2069283317504962,
+      "learning_rate": 2.1952175617404942e-05,
+      "loss": 0.4386,
+      "step": 1716
+    },
+    {
+      "epoch": 1.8169312169312168,
+      "grad_norm": 0.20579833750263932,
+      "learning_rate": 2.1932575460603685e-05,
+      "loss": 0.4147,
+      "step": 1717
+    },
+    {
+      "epoch": 1.817989417989418,
+      "grad_norm": 0.2249576933304194,
+      "learning_rate": 2.1912975303802434e-05,
+      "loss": 0.4128,
+      "step": 1718
+    },
+    {
+      "epoch": 1.819047619047619,
+      "grad_norm": 0.24258567681504786,
+      "learning_rate": 2.1893375147001177e-05,
+      "loss": 0.496,
+      "step": 1719
+    },
+    {
+      "epoch": 1.82010582010582,
+      "grad_norm": 2.4445652502189414,
+      "learning_rate": 2.1873774990199922e-05,
+      "loss": 0.5013,
+      "step": 1720
+    },
+    {
+      "epoch": 1.8211640211640212,
+      "grad_norm": 0.22937999540510728,
+      "learning_rate": 2.185417483339867e-05,
+      "loss": 0.458,
+      "step": 1721
+    },
+    {
+      "epoch": 1.8222222222222222,
+      "grad_norm": 0.22188766496336873,
+      "learning_rate": 2.1834574676597414e-05,
+      "loss": 0.4118,
+      "step": 1722
+    },
+    {
+      "epoch": 1.8232804232804232,
+      "grad_norm": 0.20985186497122293,
+      "learning_rate": 2.1814974519796157e-05,
+      "loss": 0.4202,
+      "step": 1723
+    },
+    {
+      "epoch": 1.8243386243386244,
+      "grad_norm": 5.620824290236189,
+      "learning_rate": 2.1795374362994906e-05,
+      "loss": 0.6836,
+      "step": 1724
+    },
+    {
+      "epoch": 1.8253968253968254,
+      "grad_norm": 0.2732860325354266,
+      "learning_rate": 2.177577420619365e-05,
+      "loss": 0.5006,
+      "step": 1725
+    },
+    {
+      "epoch": 1.8264550264550263,
+      "grad_norm": 0.2518168942236389,
+      "learning_rate": 2.1756174049392398e-05,
+      "loss": 0.4803,
+      "step": 1726
+    },
+    {
+      "epoch": 1.8275132275132275,
+      "grad_norm": 0.30831471115695636,
+      "learning_rate": 2.173657389259114e-05,
+      "loss": 0.4059,
+      "step": 1727
+    },
+    {
+      "epoch": 1.8285714285714287,
+      "grad_norm": 0.2518548837185105,
+      "learning_rate": 2.1716973735789886e-05,
+      "loss": 0.4423,
+      "step": 1728
+    },
+    {
+      "epoch": 1.8296296296296295,
+      "grad_norm": 0.23578369543941782,
+      "learning_rate": 2.1697373578988632e-05,
+      "loss": 0.4302,
+      "step": 1729
+    },
+    {
+      "epoch": 1.8306878306878307,
+      "grad_norm": 0.2417044417728648,
+      "learning_rate": 2.1677773422187378e-05,
+      "loss": 0.4478,
+      "step": 1730
+    },
+    {
+      "epoch": 1.831746031746032,
+      "grad_norm": 0.21978275736190653,
+      "learning_rate": 2.1658173265386124e-05,
+      "loss": 0.4316,
+      "step": 1731
+    },
+    {
+      "epoch": 1.8328042328042327,
+      "grad_norm": 0.22025067994263706,
+      "learning_rate": 2.163857310858487e-05,
+      "loss": 0.4013,
+      "step": 1732
+    },
+    {
+      "epoch": 1.8338624338624339,
+      "grad_norm": 0.23323951970705392,
+      "learning_rate": 2.1618972951783615e-05,
+      "loss": 0.4603,
+      "step": 1733
+    },
+    {
+      "epoch": 1.834920634920635,
+      "grad_norm": 0.2120845002669665,
+      "learning_rate": 2.159937279498236e-05,
+      "loss": 0.4597,
+      "step": 1734
+    },
+    {
+      "epoch": 1.8359788359788358,
+      "grad_norm": 0.22367193469028424,
+      "learning_rate": 2.1579772638181107e-05,
+      "loss": 0.4438,
+      "step": 1735
+    },
+    {
+      "epoch": 1.837037037037037,
+      "grad_norm": 0.21769442259199298,
+      "learning_rate": 2.1560172481379853e-05,
+      "loss": 0.4325,
+      "step": 1736
+    },
+    {
+      "epoch": 1.8380952380952382,
+      "grad_norm": 0.2248745754640526,
+      "learning_rate": 2.15405723245786e-05,
+      "loss": 0.432,
+      "step": 1737
+    },
+    {
+      "epoch": 1.8391534391534392,
+      "grad_norm": 0.21085112873884912,
+      "learning_rate": 2.152097216777734e-05,
+      "loss": 0.4293,
+      "step": 1738
+    },
+    {
+      "epoch": 1.8402116402116402,
+      "grad_norm": 0.20872522105623528,
+      "learning_rate": 2.150137201097609e-05,
+      "loss": 0.4546,
+      "step": 1739
+    },
+    {
+      "epoch": 1.8412698412698414,
+      "grad_norm": 0.2285263541285715,
+      "learning_rate": 2.1481771854174833e-05,
+      "loss": 0.4051,
+      "step": 1740
+    },
+    {
+      "epoch": 1.8423280423280424,
+      "grad_norm": 0.22784803646681845,
+      "learning_rate": 2.146217169737358e-05,
+      "loss": 0.446,
+      "step": 1741
+    },
+    {
+      "epoch": 1.8433862433862434,
+      "grad_norm": 0.26036748975604307,
+      "learning_rate": 2.1442571540572325e-05,
+      "loss": 0.4154,
+      "step": 1742
+    },
+    {
+      "epoch": 1.8444444444444446,
+      "grad_norm": 0.2387042021248444,
+      "learning_rate": 2.142297138377107e-05,
+      "loss": 0.4309,
+      "step": 1743
+    },
+    {
+      "epoch": 1.8455026455026455,
+      "grad_norm": 0.2182840280137568,
+      "learning_rate": 2.1403371226969817e-05,
+      "loss": 0.4367,
+      "step": 1744
+    },
+    {
+      "epoch": 1.8465608465608465,
+      "grad_norm": 0.22907755338440494,
+      "learning_rate": 2.1383771070168563e-05,
+      "loss": 0.4255,
+      "step": 1745
+    },
+    {
+      "epoch": 1.8476190476190477,
+      "grad_norm": 0.20993653159537567,
+      "learning_rate": 2.136417091336731e-05,
+      "loss": 0.4389,
+      "step": 1746
+    },
+    {
+      "epoch": 1.8486772486772487,
+      "grad_norm": 0.21338544578060106,
+      "learning_rate": 2.1344570756566054e-05,
+      "loss": 0.461,
+      "step": 1747
+    },
+    {
+      "epoch": 1.8497354497354497,
+      "grad_norm": 0.21341559351141076,
+      "learning_rate": 2.13249705997648e-05,
+      "loss": 0.4839,
+      "step": 1748
+    },
+    {
+      "epoch": 1.8507936507936509,
+      "grad_norm": 0.24079068460679487,
+      "learning_rate": 2.1305370442963546e-05,
+      "loss": 0.4823,
+      "step": 1749
+    },
+    {
+      "epoch": 1.8518518518518519,
+      "grad_norm": 0.20444170286570748,
+      "learning_rate": 2.1285770286162292e-05,
+      "loss": 0.4461,
+      "step": 1750
+    },
+    {
+      "epoch": 1.8529100529100528,
+      "grad_norm": 0.2387410951821536,
+      "learning_rate": 2.1266170129361034e-05,
+      "loss": 0.5042,
+      "step": 1751
+    },
+    {
+      "epoch": 1.853968253968254,
+      "grad_norm": 0.19937829720047517,
+      "learning_rate": 2.1246569972559784e-05,
+      "loss": 0.3733,
+      "step": 1752
+    },
+    {
+      "epoch": 1.855026455026455,
+      "grad_norm": 0.20349986087488134,
+      "learning_rate": 2.1226969815758526e-05,
+      "loss": 0.4192,
+      "step": 1753
+    },
+    {
+      "epoch": 1.856084656084656,
+      "grad_norm": 0.19942778332550556,
+      "learning_rate": 2.1207369658957275e-05,
+      "loss": 0.4123,
+      "step": 1754
+    },
+    {
+      "epoch": 1.8571428571428572,
+      "grad_norm": 0.2395297045424899,
+      "learning_rate": 2.1187769502156018e-05,
+      "loss": 0.4858,
+      "step": 1755
+    },
+    {
+      "epoch": 1.8582010582010582,
+      "grad_norm": 0.20249034937090846,
+      "learning_rate": 2.1168169345354764e-05,
+      "loss": 0.4741,
+      "step": 1756
+    },
+    {
+      "epoch": 1.8592592592592592,
+      "grad_norm": 0.22048497676461043,
+      "learning_rate": 2.114856918855351e-05,
+      "loss": 0.45,
+      "step": 1757
+    },
+    {
+      "epoch": 1.8603174603174604,
+      "grad_norm": 0.22049360912083787,
+      "learning_rate": 2.1128969031752256e-05,
+      "loss": 0.5033,
+      "step": 1758
+    },
+    {
+      "epoch": 1.8613756613756614,
+      "grad_norm": 0.22074914003644552,
+      "learning_rate": 2.1109368874950998e-05,
+      "loss": 0.4399,
+      "step": 1759
+    },
+    {
+      "epoch": 1.8624338624338623,
+      "grad_norm": 0.2160312595040178,
+      "learning_rate": 2.1089768718149747e-05,
+      "loss": 0.4702,
+      "step": 1760
+    },
+    {
+      "epoch": 1.8634920634920635,
+      "grad_norm": 0.206865371111694,
+      "learning_rate": 2.107016856134849e-05,
+      "loss": 0.4652,
+      "step": 1761
+    },
+    {
+      "epoch": 1.8645502645502645,
+      "grad_norm": 0.2091592072021074,
+      "learning_rate": 2.105056840454724e-05,
+      "loss": 0.4182,
+      "step": 1762
+    },
+    {
+      "epoch": 1.8656084656084655,
+      "grad_norm": 0.23339044723826416,
+      "learning_rate": 2.103096824774598e-05,
+      "loss": 0.5277,
+      "step": 1763
+    },
+    {
+      "epoch": 1.8666666666666667,
+      "grad_norm": 0.4460094384942643,
+      "learning_rate": 2.1011368090944727e-05,
+      "loss": 0.4579,
+      "step": 1764
+    },
+    {
+      "epoch": 1.8677248677248677,
+      "grad_norm": 0.2100939143704434,
+      "learning_rate": 2.0991767934143473e-05,
+      "loss": 0.4655,
+      "step": 1765
+    },
+    {
+      "epoch": 1.8687830687830687,
+      "grad_norm": 0.23434130765094305,
+      "learning_rate": 2.097216777734222e-05,
+      "loss": 0.5209,
+      "step": 1766
+    },
+    {
+      "epoch": 1.8698412698412699,
+      "grad_norm": 0.2434629762626319,
+      "learning_rate": 2.0952567620540965e-05,
+      "loss": 0.4495,
+      "step": 1767
+    },
+    {
+      "epoch": 1.870899470899471,
+      "grad_norm": 0.21781894326153597,
+      "learning_rate": 2.093296746373971e-05,
+      "loss": 0.4462,
+      "step": 1768
+    },
+    {
+      "epoch": 1.8719576719576718,
+      "grad_norm": 0.20899498020213622,
+      "learning_rate": 2.0913367306938457e-05,
+      "loss": 0.4457,
+      "step": 1769
+    },
+    {
+      "epoch": 1.873015873015873,
+      "grad_norm": 0.2167051331331649,
+      "learning_rate": 2.0893767150137203e-05,
+      "loss": 0.4375,
+      "step": 1770
+    },
+    {
+      "epoch": 1.8740740740740742,
+      "grad_norm": 0.2240331588887899,
+      "learning_rate": 2.087416699333595e-05,
+      "loss": 0.5185,
+      "step": 1771
+    },
+    {
+      "epoch": 1.875132275132275,
+      "grad_norm": 0.22090206678208632,
+      "learning_rate": 2.085456683653469e-05,
+      "loss": 0.4808,
+      "step": 1772
+    },
+    {
+      "epoch": 1.8761904761904762,
+      "grad_norm": 0.22040696733634577,
+      "learning_rate": 2.083496667973344e-05,
+      "loss": 0.456,
+      "step": 1773
+    },
+    {
+      "epoch": 1.8772486772486774,
+      "grad_norm": 0.2079955864538246,
+      "learning_rate": 2.0815366522932183e-05,
+      "loss": 0.4143,
+      "step": 1774
+    },
+    {
+      "epoch": 1.8783068783068781,
+      "grad_norm": 0.22169752369395546,
+      "learning_rate": 2.0795766366130932e-05,
+      "loss": 0.4858,
+      "step": 1775
+    },
+    {
+      "epoch": 1.8793650793650793,
+      "grad_norm": 0.2402070315956562,
+      "learning_rate": 2.0776166209329674e-05,
+      "loss": 0.4504,
+      "step": 1776
+    },
+    {
+      "epoch": 1.8804232804232806,
+      "grad_norm": 0.23435927645511864,
+      "learning_rate": 2.075656605252842e-05,
+      "loss": 0.4593,
+      "step": 1777
+    },
+    {
+      "epoch": 1.8814814814814815,
+      "grad_norm": 0.20818590974484366,
+      "learning_rate": 2.0736965895727166e-05,
+      "loss": 0.4381,
+      "step": 1778
+    },
+    {
+      "epoch": 1.8825396825396825,
+      "grad_norm": 0.2130354907179979,
+      "learning_rate": 2.0717365738925912e-05,
+      "loss": 0.4297,
+      "step": 1779
+    },
+    {
+      "epoch": 1.8835978835978837,
+      "grad_norm": 0.21660044821589397,
+      "learning_rate": 2.0697765582124658e-05,
+      "loss": 0.4697,
+      "step": 1780
+    },
+    {
+      "epoch": 1.8846560846560847,
+      "grad_norm": 0.20309183557121804,
+      "learning_rate": 2.0678165425323404e-05,
+      "loss": 0.4025,
+      "step": 1781
+    },
+    {
+      "epoch": 1.8857142857142857,
+      "grad_norm": 0.21087610186232195,
+      "learning_rate": 2.065856526852215e-05,
+      "loss": 0.4546,
+      "step": 1782
+    },
+    {
+      "epoch": 1.8867724867724869,
+      "grad_norm": 0.22732457303926343,
+      "learning_rate": 2.0638965111720896e-05,
+      "loss": 0.4668,
+      "step": 1783
+    },
+    {
+      "epoch": 1.8878306878306879,
+      "grad_norm": 0.22180452040864543,
+      "learning_rate": 2.061936495491964e-05,
+      "loss": 0.4584,
+      "step": 1784
+    },
+    {
+      "epoch": 1.8888888888888888,
+      "grad_norm": 0.21261990698971484,
+      "learning_rate": 2.0599764798118387e-05,
+      "loss": 0.4406,
+      "step": 1785
+    },
+    {
+      "epoch": 1.88994708994709,
+      "grad_norm": 0.21071003692457904,
+      "learning_rate": 2.0580164641317133e-05,
+      "loss": 0.4143,
+      "step": 1786
+    },
+    {
+      "epoch": 1.891005291005291,
+      "grad_norm": 0.2204518775713819,
+      "learning_rate": 2.0560564484515876e-05,
+      "loss": 0.4135,
+      "step": 1787
+    },
+    {
+      "epoch": 1.892063492063492,
+      "grad_norm": 0.2078948195319659,
+      "learning_rate": 2.0540964327714625e-05,
+      "loss": 0.3921,
+      "step": 1788
+    },
+    {
+      "epoch": 1.8931216931216932,
+      "grad_norm": 0.2196279086951217,
+      "learning_rate": 2.0521364170913367e-05,
+      "loss": 0.3873,
+      "step": 1789
+    },
+    {
+      "epoch": 1.8941798941798942,
+      "grad_norm": 0.24347249045633615,
+      "learning_rate": 2.0501764014112113e-05,
+      "loss": 0.4223,
+      "step": 1790
+    },
+    {
+      "epoch": 1.8952380952380952,
+      "grad_norm": 0.21502367020868784,
+      "learning_rate": 2.048216385731086e-05,
+      "loss": 0.4241,
+      "step": 1791
+    },
+    {
+      "epoch": 1.8962962962962964,
+      "grad_norm": 0.2223665065572327,
+      "learning_rate": 2.0462563700509605e-05,
+      "loss": 0.4585,
+      "step": 1792
+    },
+    {
+      "epoch": 1.8973544973544973,
+      "grad_norm": 0.2210410067693124,
+      "learning_rate": 2.044296354370835e-05,
+      "loss": 0.4468,
+      "step": 1793
+    },
+    {
+      "epoch": 1.8984126984126983,
+      "grad_norm": 0.21557122963124056,
+      "learning_rate": 2.0423363386907097e-05,
+      "loss": 0.458,
+      "step": 1794
+    },
+    {
+      "epoch": 1.8994708994708995,
+      "grad_norm": 0.24519100116436493,
+      "learning_rate": 2.040376323010584e-05,
+      "loss": 0.4466,
+      "step": 1795
+    },
+    {
+      "epoch": 1.9005291005291005,
+      "grad_norm": 0.24801432016192543,
+      "learning_rate": 2.038416307330459e-05,
+      "loss": 0.4839,
+      "step": 1796
+    },
+    {
+      "epoch": 1.9015873015873015,
+      "grad_norm": 0.20049577359290413,
+      "learning_rate": 2.036456291650333e-05,
+      "loss": 0.4127,
+      "step": 1797
+    },
+    {
+      "epoch": 1.9026455026455027,
+      "grad_norm": 0.23520752147502938,
+      "learning_rate": 2.034496275970208e-05,
+      "loss": 0.448,
+      "step": 1798
+    },
+    {
+      "epoch": 1.9037037037037037,
+      "grad_norm": 0.24490985101151913,
+      "learning_rate": 2.0325362602900823e-05,
+      "loss": 0.431,
+      "step": 1799
+    },
+    {
+      "epoch": 1.9047619047619047,
+      "grad_norm": 0.2664715652435904,
+      "learning_rate": 2.030576244609957e-05,
+      "loss": 0.4766,
+      "step": 1800
+    },
+    {
+      "epoch": 1.9058201058201059,
+      "grad_norm": 0.2485082923716217,
+      "learning_rate": 2.0286162289298315e-05,
+      "loss": 0.573,
+      "step": 1801
+    },
+    {
+      "epoch": 1.9068783068783068,
+      "grad_norm": 0.24251809787369136,
+      "learning_rate": 2.026656213249706e-05,
+      "loss": 0.4245,
+      "step": 1802
+    },
+    {
+      "epoch": 1.9079365079365078,
+      "grad_norm": 0.27387491257447555,
+      "learning_rate": 2.0246961975695806e-05,
+      "loss": 0.4171,
+      "step": 1803
+    },
+    {
+      "epoch": 1.908994708994709,
+      "grad_norm": 0.23047285152697722,
+      "learning_rate": 2.0227361818894552e-05,
+      "loss": 0.493,
+      "step": 1804
+    },
+    {
+      "epoch": 1.91005291005291,
+      "grad_norm": 7.990646854444911,
+      "learning_rate": 2.0207761662093298e-05,
+      "loss": 0.635,
+      "step": 1805
+    },
+    {
+      "epoch": 1.911111111111111,
+      "grad_norm": 0.26163599966519524,
+      "learning_rate": 2.0188161505292044e-05,
+      "loss": 0.4351,
+      "step": 1806
+    },
+    {
+      "epoch": 1.9121693121693122,
+      "grad_norm": 0.25130856546308006,
+      "learning_rate": 2.016856134849079e-05,
+      "loss": 0.4399,
+      "step": 1807
+    },
+    {
+      "epoch": 1.9132275132275134,
+      "grad_norm": 0.20476985393616431,
+      "learning_rate": 2.0148961191689532e-05,
+      "loss": 0.4158,
+      "step": 1808
+    },
+    {
+      "epoch": 1.9142857142857141,
+      "grad_norm": 0.22396937696533817,
+      "learning_rate": 2.012936103488828e-05,
+      "loss": 0.4953,
+      "step": 1809
+    },
+    {
+      "epoch": 1.9153439153439153,
+      "grad_norm": 0.2100685359464842,
+      "learning_rate": 2.0109760878087024e-05,
+      "loss": 0.4604,
+      "step": 1810
+    },
+    {
+      "epoch": 1.9164021164021166,
+      "grad_norm": 0.2347068667100795,
+      "learning_rate": 2.0090160721285773e-05,
+      "loss": 0.4656,
+      "step": 1811
+    },
+    {
+      "epoch": 1.9174603174603173,
+      "grad_norm": 0.2134340622258882,
+      "learning_rate": 2.0070560564484516e-05,
+      "loss": 0.4155,
+      "step": 1812
+    },
+    {
+      "epoch": 1.9185185185185185,
+      "grad_norm": 0.20915306386821994,
+      "learning_rate": 2.005096040768326e-05,
+      "loss": 0.4005,
+      "step": 1813
+    },
+    {
+      "epoch": 1.9195767195767197,
+      "grad_norm": 0.2261359347456008,
+      "learning_rate": 2.0031360250882008e-05,
+      "loss": 0.4248,
+      "step": 1814
+    },
+    {
+      "epoch": 1.9206349206349205,
+      "grad_norm": 0.21813479888223342,
+      "learning_rate": 2.0011760094080753e-05,
+      "loss": 0.4638,
+      "step": 1815
+    },
+    {
+      "epoch": 1.9216931216931217,
+      "grad_norm": 0.2101057981170796,
+      "learning_rate": 1.99921599372795e-05,
+      "loss": 0.4028,
+      "step": 1816
+    },
+    {
+      "epoch": 1.9227513227513229,
+      "grad_norm": 0.22511459770933143,
+      "learning_rate": 1.9972559780478245e-05,
+      "loss": 0.5111,
+      "step": 1817
+    },
+    {
+      "epoch": 1.9238095238095239,
+      "grad_norm": 0.23388911916732105,
+      "learning_rate": 1.995295962367699e-05,
+      "loss": 0.4683,
+      "step": 1818
+    },
+    {
+      "epoch": 1.9248677248677248,
+      "grad_norm": 0.22435740200537527,
+      "learning_rate": 1.9933359466875737e-05,
+      "loss": 0.438,
+      "step": 1819
+    },
+    {
+      "epoch": 1.925925925925926,
+      "grad_norm": 0.21611950819751236,
+      "learning_rate": 1.9913759310074483e-05,
+      "loss": 0.378,
+      "step": 1820
+    },
+    {
+      "epoch": 1.926984126984127,
+      "grad_norm": 0.20291845845078743,
+      "learning_rate": 1.9894159153273225e-05,
+      "loss": 0.4277,
+      "step": 1821
+    },
+    {
+      "epoch": 1.928042328042328,
+      "grad_norm": 0.22847172320277195,
+      "learning_rate": 1.9874558996471975e-05,
+      "loss": 0.4674,
+      "step": 1822
+    },
+    {
+      "epoch": 1.9291005291005292,
+      "grad_norm": 0.23053558937849467,
+      "learning_rate": 1.9854958839670717e-05,
+      "loss": 0.4015,
+      "step": 1823
+    },
+    {
+      "epoch": 1.9301587301587302,
+      "grad_norm": 0.23708550014377422,
+      "learning_rate": 1.9835358682869466e-05,
+      "loss": 0.4684,
+      "step": 1824
+    },
+    {
+      "epoch": 1.9312169312169312,
+      "grad_norm": 0.316817125246121,
+      "learning_rate": 1.981575852606821e-05,
+      "loss": 0.4263,
+      "step": 1825
+    },
+    {
+      "epoch": 1.9322751322751324,
+      "grad_norm": 0.2254829079471954,
+      "learning_rate": 1.9796158369266955e-05,
+      "loss": 0.4851,
+      "step": 1826
+    },
+    {
+      "epoch": 1.9333333333333333,
+      "grad_norm": 0.2316587670641986,
+      "learning_rate": 1.97765582124657e-05,
+      "loss": 0.4229,
+      "step": 1827
+    },
+    {
+      "epoch": 1.9343915343915343,
+      "grad_norm": 0.21297157984215134,
+      "learning_rate": 1.9756958055664446e-05,
+      "loss": 0.4415,
+      "step": 1828
+    },
+    {
+      "epoch": 1.9354497354497355,
+      "grad_norm": 0.2612957037749709,
+      "learning_rate": 1.9737357898863192e-05,
+      "loss": 0.4589,
+      "step": 1829
+    },
+    {
+      "epoch": 1.9365079365079365,
+      "grad_norm": 0.2129663181734509,
+      "learning_rate": 1.9717757742061938e-05,
+      "loss": 0.4511,
+      "step": 1830
+    },
+    {
+      "epoch": 1.9375661375661375,
+      "grad_norm": 0.23503195721500833,
+      "learning_rate": 1.969815758526068e-05,
+      "loss": 0.4617,
+      "step": 1831
+    },
+    {
+      "epoch": 1.9386243386243387,
+      "grad_norm": 0.2287134737206407,
+      "learning_rate": 1.967855742845943e-05,
+      "loss": 0.471,
+      "step": 1832
+    },
+    {
+      "epoch": 1.9396825396825397,
+      "grad_norm": 0.21262604276669486,
+      "learning_rate": 1.9658957271658172e-05,
+      "loss": 0.47,
+      "step": 1833
+    },
+    {
+      "epoch": 1.9407407407407407,
+      "grad_norm": 0.220344769052406,
+      "learning_rate": 1.963935711485692e-05,
+      "loss": 0.4505,
+      "step": 1834
+    },
+    {
+      "epoch": 1.9417989417989419,
+      "grad_norm": 0.220152899947779,
+      "learning_rate": 1.9619756958055664e-05,
+      "loss": 0.4263,
+      "step": 1835
+    },
+    {
+      "epoch": 1.9428571428571428,
+      "grad_norm": 0.20366436651004108,
+      "learning_rate": 1.960015680125441e-05,
+      "loss": 0.3771,
+      "step": 1836
+    },
+    {
+      "epoch": 1.9439153439153438,
+      "grad_norm": 0.2200894429401924,
+      "learning_rate": 1.9580556644453156e-05,
+      "loss": 0.4203,
+      "step": 1837
+    },
+    {
+      "epoch": 1.944973544973545,
+      "grad_norm": 0.21590610062905347,
+      "learning_rate": 1.9560956487651902e-05,
+      "loss": 0.4558,
+      "step": 1838
+    },
+    {
+      "epoch": 1.946031746031746,
+      "grad_norm": 0.22356270505912842,
+      "learning_rate": 1.9541356330850648e-05,
+      "loss": 0.4233,
+      "step": 1839
+    },
+    {
+      "epoch": 1.947089947089947,
+      "grad_norm": 0.20737807332328057,
+      "learning_rate": 1.9521756174049394e-05,
+      "loss": 0.3993,
+      "step": 1840
+    },
+    {
+      "epoch": 1.9481481481481482,
+      "grad_norm": 0.21342097386972908,
+      "learning_rate": 1.950215601724814e-05,
+      "loss": 0.4127,
+      "step": 1841
+    },
+    {
+      "epoch": 1.9492063492063492,
+      "grad_norm": 0.2031800844818702,
+      "learning_rate": 1.9482555860446885e-05,
+      "loss": 0.383,
+      "step": 1842
+    },
+    {
+      "epoch": 1.9502645502645501,
+      "grad_norm": 0.22744285466424116,
+      "learning_rate": 1.946295570364563e-05,
+      "loss": 0.4748,
+      "step": 1843
+    },
+    {
+      "epoch": 1.9513227513227513,
+      "grad_norm": 0.23141003857099457,
+      "learning_rate": 1.9443355546844374e-05,
+      "loss": 0.5051,
+      "step": 1844
+    },
+    {
+      "epoch": 1.9523809523809523,
+      "grad_norm": 0.2145215316356645,
+      "learning_rate": 1.9423755390043123e-05,
+      "loss": 0.4225,
+      "step": 1845
+    },
+    {
+      "epoch": 1.9534391534391533,
+      "grad_norm": 0.2026905601029007,
+      "learning_rate": 1.9404155233241865e-05,
+      "loss": 0.4007,
+      "step": 1846
+    },
+    {
+      "epoch": 1.9544973544973545,
+      "grad_norm": 0.22999532724274202,
+      "learning_rate": 1.9384555076440615e-05,
+      "loss": 0.4617,
+      "step": 1847
+    },
+    {
+      "epoch": 1.9555555555555557,
+      "grad_norm": 0.2082995676826002,
+      "learning_rate": 1.9364954919639357e-05,
+      "loss": 0.4278,
+      "step": 1848
+    },
+    {
+      "epoch": 1.9566137566137565,
+      "grad_norm": 0.22475386311924242,
+      "learning_rate": 1.9345354762838103e-05,
+      "loss": 0.4241,
+      "step": 1849
+    },
+    {
+      "epoch": 1.9576719576719577,
+      "grad_norm": 0.22479359963728507,
+      "learning_rate": 1.932575460603685e-05,
+      "loss": 0.4338,
+      "step": 1850
+    },
+    {
+      "epoch": 1.9587301587301589,
+      "grad_norm": 0.20367965304182695,
+      "learning_rate": 1.9306154449235595e-05,
+      "loss": 0.3818,
+      "step": 1851
+    },
+    {
+      "epoch": 1.9597883597883596,
+      "grad_norm": 0.21527243201486918,
+      "learning_rate": 1.928655429243434e-05,
+      "loss": 0.3774,
+      "step": 1852
+    },
+    {
+      "epoch": 1.9608465608465608,
+      "grad_norm": 0.2249500078225003,
+      "learning_rate": 1.9266954135633087e-05,
+      "loss": 0.4967,
+      "step": 1853
+    },
+    {
+      "epoch": 1.961904761904762,
+      "grad_norm": 0.21710410127581037,
+      "learning_rate": 1.9247353978831832e-05,
+      "loss": 0.456,
+      "step": 1854
+    },
+    {
+      "epoch": 1.9629629629629628,
+      "grad_norm": 0.21285160925599858,
+      "learning_rate": 1.9227753822030578e-05,
+      "loss": 0.4345,
+      "step": 1855
+    },
+    {
+      "epoch": 1.964021164021164,
+      "grad_norm": 0.19869936479793823,
+      "learning_rate": 1.9208153665229324e-05,
+      "loss": 0.4267,
+      "step": 1856
+    },
+    {
+      "epoch": 1.9650793650793652,
+      "grad_norm": 0.20189396892728878,
+      "learning_rate": 1.9188553508428067e-05,
+      "loss": 0.4246,
+      "step": 1857
+    },
+    {
+      "epoch": 1.9661375661375662,
+      "grad_norm": 0.21026152529101638,
+      "learning_rate": 1.9168953351626816e-05,
+      "loss": 0.4074,
+      "step": 1858
+    },
+    {
+      "epoch": 1.9671957671957672,
+      "grad_norm": 0.2143954764423591,
+      "learning_rate": 1.914935319482556e-05,
+      "loss": 0.4739,
+      "step": 1859
+    },
+    {
+      "epoch": 1.9682539682539684,
+      "grad_norm": 0.19350444830703373,
+      "learning_rate": 1.9129753038024308e-05,
+      "loss": 0.3873,
+      "step": 1860
+    },
+    {
+      "epoch": 1.9693121693121693,
+      "grad_norm": 0.19855752381553382,
+      "learning_rate": 1.911015288122305e-05,
+      "loss": 0.4127,
+      "step": 1861
+    },
+    {
+      "epoch": 1.9703703703703703,
+      "grad_norm": 0.21567120927747255,
+      "learning_rate": 1.9090552724421796e-05,
+      "loss": 0.4648,
+      "step": 1862
+    },
+    {
+      "epoch": 1.9714285714285715,
+      "grad_norm": 0.21153824863129872,
+      "learning_rate": 1.9070952567620542e-05,
+      "loss": 0.4796,
+      "step": 1863
+    },
+    {
+      "epoch": 1.9724867724867725,
+      "grad_norm": 0.19320130815422823,
+      "learning_rate": 1.9051352410819288e-05,
+      "loss": 0.4275,
+      "step": 1864
+    },
+    {
+      "epoch": 1.9735449735449735,
+      "grad_norm": 0.19887236706758227,
+      "learning_rate": 1.9031752254018034e-05,
+      "loss": 0.4148,
+      "step": 1865
+    },
+    {
+      "epoch": 1.9746031746031747,
+      "grad_norm": 0.1932406080084438,
+      "learning_rate": 1.901215209721678e-05,
+      "loss": 0.4067,
+      "step": 1866
+    },
+    {
+      "epoch": 1.9756613756613757,
+      "grad_norm": 0.21807163033247312,
+      "learning_rate": 1.8992551940415522e-05,
+      "loss": 0.4882,
+      "step": 1867
+    },
+    {
+      "epoch": 1.9767195767195767,
+      "grad_norm": 0.20618559739939532,
+      "learning_rate": 1.897295178361427e-05,
+      "loss": 0.4276,
+      "step": 1868
+    },
+    {
+      "epoch": 1.9777777777777779,
+      "grad_norm": 0.19090448828455922,
+      "learning_rate": 1.8953351626813014e-05,
+      "loss": 0.4043,
+      "step": 1869
+    },
+    {
+      "epoch": 1.9788359788359788,
+      "grad_norm": 0.1947353451525142,
+      "learning_rate": 1.893375147001176e-05,
+      "loss": 0.4039,
+      "step": 1870
+    },
+    {
+      "epoch": 1.9798941798941798,
+      "grad_norm": 0.22563886214800088,
+      "learning_rate": 1.8914151313210505e-05,
+      "loss": 0.4928,
+      "step": 1871
+    },
+    {
+      "epoch": 1.980952380952381,
+      "grad_norm": 0.24654476642054673,
+      "learning_rate": 1.889455115640925e-05,
+      "loss": 0.5016,
+      "step": 1872
+    },
+    {
+      "epoch": 1.982010582010582,
+      "grad_norm": 0.21770134373827096,
+      "learning_rate": 1.8874950999607997e-05,
+      "loss": 0.4636,
+      "step": 1873
+    },
+    {
+      "epoch": 1.983068783068783,
+      "grad_norm": 0.21078921223109962,
+      "learning_rate": 1.8855350842806743e-05,
+      "loss": 0.4529,
+      "step": 1874
+    },
+    {
+      "epoch": 1.9841269841269842,
+      "grad_norm": 0.2042699274002301,
+      "learning_rate": 1.883575068600549e-05,
+      "loss": 0.4443,
+      "step": 1875
+    },
+    {
+      "epoch": 1.9851851851851852,
+      "grad_norm": 0.20988400394066842,
+      "learning_rate": 1.8816150529204235e-05,
+      "loss": 0.4441,
+      "step": 1876
+    },
+    {
+      "epoch": 1.9862433862433861,
+      "grad_norm": 0.231811845633082,
+      "learning_rate": 1.879655037240298e-05,
+      "loss": 0.4538,
+      "step": 1877
+    },
+    {
+      "epoch": 1.9873015873015873,
+      "grad_norm": 0.19558466017276782,
+      "learning_rate": 1.8776950215601727e-05,
+      "loss": 0.4031,
+      "step": 1878
+    },
+    {
+      "epoch": 1.9883597883597883,
+      "grad_norm": 0.20830804273827055,
+      "learning_rate": 1.8757350058800472e-05,
+      "loss": 0.454,
+      "step": 1879
+    },
+    {
+      "epoch": 1.9894179894179893,
+      "grad_norm": 0.2316902360278231,
+      "learning_rate": 1.8737749901999215e-05,
+      "loss": 0.4099,
+      "step": 1880
+    },
+    {
+      "epoch": 1.9904761904761905,
+      "grad_norm": 0.6294053779165485,
+      "learning_rate": 1.8718149745197964e-05,
+      "loss": 0.5239,
+      "step": 1881
+    },
+    {
+      "epoch": 1.9915343915343915,
+      "grad_norm": 0.1914593214711506,
+      "learning_rate": 1.8698549588396707e-05,
+      "loss": 0.3855,
+      "step": 1882
+    },
+    {
+      "epoch": 1.9925925925925925,
+      "grad_norm": 0.21642178461292894,
+      "learning_rate": 1.8678949431595456e-05,
+      "loss": 0.467,
+      "step": 1883
+    },
+    {
+      "epoch": 1.9936507936507937,
+      "grad_norm": 0.23743786345884751,
+      "learning_rate": 1.86593492747942e-05,
+      "loss": 0.4487,
+      "step": 1884
+    },
+    {
+      "epoch": 1.9947089947089947,
+      "grad_norm": 0.23198695415553083,
+      "learning_rate": 1.8639749117992944e-05,
+      "loss": 0.5057,
+      "step": 1885
+    },
+    {
+      "epoch": 1.9957671957671956,
+      "grad_norm": 0.3471270912766888,
+      "learning_rate": 1.862014896119169e-05,
+      "loss": 0.3937,
+      "step": 1886
+    },
+    {
+      "epoch": 1.9968253968253968,
+      "grad_norm": 0.22576252798533672,
+      "learning_rate": 1.8600548804390436e-05,
+      "loss": 0.4228,
+      "step": 1887
+    },
+    {
+      "epoch": 1.997883597883598,
+      "grad_norm": 0.21560388020282445,
+      "learning_rate": 1.8580948647589182e-05,
+      "loss": 0.4147,
+      "step": 1888
+    },
+    {
+      "epoch": 1.9989417989417988,
+      "grad_norm": 0.22358325111497787,
+      "learning_rate": 1.8561348490787928e-05,
+      "loss": 0.4384,
+      "step": 1889
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.22079447072147176,
+      "learning_rate": 1.8541748333986674e-05,
+      "loss": 0.4229,
+      "step": 1890
+    },
+    {
+      "epoch": 2.001058201058201,
+      "grad_norm": 0.298496877542824,
+      "learning_rate": 1.852214817718542e-05,
+      "loss": 0.3341,
+      "step": 1891
+    },
+    {
+      "epoch": 2.002116402116402,
+      "grad_norm": 0.2769200047722302,
+      "learning_rate": 1.8502548020384165e-05,
+      "loss": 0.3833,
+      "step": 1892
+    },
+    {
+      "epoch": 2.003174603174603,
+      "grad_norm": 0.29171539792667067,
+      "learning_rate": 1.8482947863582908e-05,
+      "loss": 0.3858,
+      "step": 1893
+    },
+    {
+      "epoch": 2.0042328042328044,
+      "grad_norm": 0.2789134735368325,
+      "learning_rate": 1.8463347706781657e-05,
+      "loss": 0.3568,
+      "step": 1894
+    },
+    {
+      "epoch": 2.005291005291005,
+      "grad_norm": 0.34633898150173376,
+      "learning_rate": 1.84437475499804e-05,
+      "loss": 0.3465,
+      "step": 1895
+    },
+    {
+      "epoch": 2.0063492063492063,
+      "grad_norm": 0.2793278281768595,
+      "learning_rate": 1.842414739317915e-05,
+      "loss": 0.4017,
+      "step": 1896
+    },
+    {
+      "epoch": 2.0074074074074075,
+      "grad_norm": 0.3059555073494133,
+      "learning_rate": 1.840454723637789e-05,
+      "loss": 0.3392,
+      "step": 1897
+    },
+    {
+      "epoch": 2.0084656084656083,
+      "grad_norm": 0.26671660329309566,
+      "learning_rate": 1.8384947079576637e-05,
+      "loss": 0.3768,
+      "step": 1898
+    },
+    {
+      "epoch": 2.0095238095238095,
+      "grad_norm": 0.276118400635007,
+      "learning_rate": 1.8365346922775383e-05,
+      "loss": 0.3967,
+      "step": 1899
+    },
+    {
+      "epoch": 2.0105820105820107,
+      "grad_norm": 0.23862871261186194,
+      "learning_rate": 1.834574676597413e-05,
+      "loss": 0.3411,
+      "step": 1900
+    },
+    {
+      "epoch": 2.0116402116402115,
+      "grad_norm": 0.2602752683930678,
+      "learning_rate": 1.8326146609172875e-05,
+      "loss": 0.3598,
+      "step": 1901
+    },
+    {
+      "epoch": 2.0126984126984127,
+      "grad_norm": 0.24165262446248684,
+      "learning_rate": 1.830654645237162e-05,
+      "loss": 0.3637,
+      "step": 1902
+    },
+    {
+      "epoch": 2.013756613756614,
+      "grad_norm": 0.2211835840328273,
+      "learning_rate": 1.8286946295570363e-05,
+      "loss": 0.3475,
+      "step": 1903
+    },
+    {
+      "epoch": 2.0148148148148146,
+      "grad_norm": 0.2536735392338854,
+      "learning_rate": 1.8267346138769113e-05,
+      "loss": 0.371,
+      "step": 1904
+    },
+    {
+      "epoch": 2.015873015873016,
+      "grad_norm": 0.23225659690495537,
+      "learning_rate": 1.8247745981967855e-05,
+      "loss": 0.3231,
+      "step": 1905
+    },
+    {
+      "epoch": 2.016931216931217,
+      "grad_norm": 0.2363270132100584,
+      "learning_rate": 1.82281458251666e-05,
+      "loss": 0.4065,
+      "step": 1906
+    },
+    {
+      "epoch": 2.0179894179894178,
+      "grad_norm": 0.21698478882185174,
+      "learning_rate": 1.8208545668365347e-05,
+      "loss": 0.372,
+      "step": 1907
+    },
+    {
+      "epoch": 2.019047619047619,
+      "grad_norm": 0.23211712957073785,
+      "learning_rate": 1.8188945511564093e-05,
+      "loss": 0.3939,
+      "step": 1908
+    },
+    {
+      "epoch": 2.02010582010582,
+      "grad_norm": 0.21859722986589825,
+      "learning_rate": 1.816934535476284e-05,
+      "loss": 0.3096,
+      "step": 1909
+    },
+    {
+      "epoch": 2.0211640211640214,
+      "grad_norm": 0.22198129532334243,
+      "learning_rate": 1.8149745197961584e-05,
+      "loss": 0.3603,
+      "step": 1910
+    },
+    {
+      "epoch": 2.022222222222222,
+      "grad_norm": 0.20161388915274267,
+      "learning_rate": 1.813014504116033e-05,
+      "loss": 0.3316,
+      "step": 1911
+    },
+    {
+      "epoch": 2.0232804232804233,
+      "grad_norm": 0.2169686284719689,
+      "learning_rate": 1.8110544884359076e-05,
+      "loss": 0.3631,
+      "step": 1912
+    },
+    {
+      "epoch": 2.0243386243386245,
+      "grad_norm": 0.20992722910943554,
+      "learning_rate": 1.8090944727557822e-05,
+      "loss": 0.354,
+      "step": 1913
+    },
+    {
+      "epoch": 2.0253968253968253,
+      "grad_norm": 0.21467603286816797,
+      "learning_rate": 1.8071344570756568e-05,
+      "loss": 0.3362,
+      "step": 1914
+    },
+    {
+      "epoch": 2.0264550264550265,
+      "grad_norm": 0.20786206596359347,
+      "learning_rate": 1.8051744413955314e-05,
+      "loss": 0.3244,
+      "step": 1915
+    },
+    {
+      "epoch": 2.0275132275132277,
+      "grad_norm": 0.21372256780280968,
+      "learning_rate": 1.8032144257154056e-05,
+      "loss": 0.371,
+      "step": 1916
+    },
+    {
+      "epoch": 2.0285714285714285,
+      "grad_norm": 0.21738963933410124,
+      "learning_rate": 1.8012544100352806e-05,
+      "loss": 0.3022,
+      "step": 1917
+    },
+    {
+      "epoch": 2.0296296296296297,
+      "grad_norm": 0.20703116847143097,
+      "learning_rate": 1.7992943943551548e-05,
+      "loss": 0.3333,
+      "step": 1918
+    },
+    {
+      "epoch": 2.030687830687831,
+      "grad_norm": 0.22313107632965132,
+      "learning_rate": 1.7973343786750294e-05,
+      "loss": 0.3615,
+      "step": 1919
+    },
+    {
+      "epoch": 2.0317460317460316,
+      "grad_norm": 0.20832627765877879,
+      "learning_rate": 1.795374362994904e-05,
+      "loss": 0.3038,
+      "step": 1920
+    },
+    {
+      "epoch": 2.032804232804233,
+      "grad_norm": 0.21323384648245985,
+      "learning_rate": 1.7934143473147786e-05,
+      "loss": 0.3509,
+      "step": 1921
+    },
+    {
+      "epoch": 2.033862433862434,
+      "grad_norm": 0.22323179664908244,
+      "learning_rate": 1.791454331634653e-05,
+      "loss": 0.3586,
+      "step": 1922
+    },
+    {
+      "epoch": 2.034920634920635,
+      "grad_norm": 0.2223456687807534,
+      "learning_rate": 1.7894943159545277e-05,
+      "loss": 0.3692,
+      "step": 1923
+    },
+    {
+      "epoch": 2.035978835978836,
+      "grad_norm": 0.20164634068506698,
+      "learning_rate": 1.7875343002744023e-05,
+      "loss": 0.3694,
+      "step": 1924
+    },
+    {
+      "epoch": 2.037037037037037,
+      "grad_norm": 0.20323586163027332,
+      "learning_rate": 1.785574284594277e-05,
+      "loss": 0.3405,
+      "step": 1925
+    },
+    {
+      "epoch": 2.038095238095238,
+      "grad_norm": 0.19413426270312215,
+      "learning_rate": 1.7836142689141515e-05,
+      "loss": 0.3272,
+      "step": 1926
+    },
+    {
+      "epoch": 2.039153439153439,
+      "grad_norm": 0.21682781764578435,
+      "learning_rate": 1.781654253234026e-05,
+      "loss": 0.3543,
+      "step": 1927
+    },
+    {
+      "epoch": 2.0402116402116404,
+      "grad_norm": 0.20226324872136348,
+      "learning_rate": 1.7796942375539007e-05,
+      "loss": 0.3249,
+      "step": 1928
+    },
+    {
+      "epoch": 2.041269841269841,
+      "grad_norm": 0.22193251820682366,
+      "learning_rate": 1.777734221873775e-05,
+      "loss": 0.3416,
+      "step": 1929
+    },
+    {
+      "epoch": 2.0423280423280423,
+      "grad_norm": 0.1835440881260506,
+      "learning_rate": 1.77577420619365e-05,
+      "loss": 0.3253,
+      "step": 1930
+    },
+    {
+      "epoch": 2.0433862433862435,
+      "grad_norm": 0.23125563586478098,
+      "learning_rate": 1.773814190513524e-05,
+      "loss": 0.3172,
+      "step": 1931
+    },
+    {
+      "epoch": 2.0444444444444443,
+      "grad_norm": 0.2192023103480394,
+      "learning_rate": 1.771854174833399e-05,
+      "loss": 0.3675,
+      "step": 1932
+    },
+    {
+      "epoch": 2.0455026455026455,
+      "grad_norm": 0.19764009006450245,
+      "learning_rate": 1.7698941591532733e-05,
+      "loss": 0.3108,
+      "step": 1933
+    },
+    {
+      "epoch": 2.0465608465608467,
+      "grad_norm": 0.21277420085835824,
+      "learning_rate": 1.767934143473148e-05,
+      "loss": 0.3796,
+      "step": 1934
+    },
+    {
+      "epoch": 2.0476190476190474,
+      "grad_norm": 0.2221209417138681,
+      "learning_rate": 1.7659741277930225e-05,
+      "loss": 0.3444,
+      "step": 1935
+    },
+    {
+      "epoch": 2.0486772486772487,
+      "grad_norm": 0.208598589107057,
+      "learning_rate": 1.764014112112897e-05,
+      "loss": 0.3416,
+      "step": 1936
+    },
+    {
+      "epoch": 2.04973544973545,
+      "grad_norm": 0.19795154801112497,
+      "learning_rate": 1.7620540964327713e-05,
+      "loss": 0.295,
+      "step": 1937
+    },
+    {
+      "epoch": 2.0507936507936506,
+      "grad_norm": 0.21861746828164197,
+      "learning_rate": 1.7600940807526462e-05,
+      "loss": 0.3672,
+      "step": 1938
+    },
+    {
+      "epoch": 2.051851851851852,
+      "grad_norm": 0.20848142512577938,
+      "learning_rate": 1.7581340650725205e-05,
+      "loss": 0.3503,
+      "step": 1939
+    },
+    {
+      "epoch": 2.052910052910053,
+      "grad_norm": 0.20406572548942745,
+      "learning_rate": 1.7561740493923954e-05,
+      "loss": 0.3594,
+      "step": 1940
+    },
+    {
+      "epoch": 2.0539682539682538,
+      "grad_norm": 0.22043618465867792,
+      "learning_rate": 1.7542140337122696e-05,
+      "loss": 0.3388,
+      "step": 1941
+    },
+    {
+      "epoch": 2.055026455026455,
+      "grad_norm": 0.20857631557038028,
+      "learning_rate": 1.7522540180321442e-05,
+      "loss": 0.3665,
+      "step": 1942
+    },
+    {
+      "epoch": 2.056084656084656,
+      "grad_norm": 0.21727483413067578,
+      "learning_rate": 1.7502940023520188e-05,
+      "loss": 0.4127,
+      "step": 1943
+    },
+    {
+      "epoch": 2.057142857142857,
+      "grad_norm": 0.23262861070941543,
+      "learning_rate": 1.7483339866718934e-05,
+      "loss": 0.408,
+      "step": 1944
+    },
+    {
+      "epoch": 2.058201058201058,
+      "grad_norm": 0.19879438803020888,
+      "learning_rate": 1.746373970991768e-05,
+      "loss": 0.3254,
+      "step": 1945
+    },
+    {
+      "epoch": 2.0592592592592593,
+      "grad_norm": 0.21005560252157807,
+      "learning_rate": 1.7444139553116426e-05,
+      "loss": 0.3522,
+      "step": 1946
+    },
+    {
+      "epoch": 2.06031746031746,
+      "grad_norm": 0.22966010387646338,
+      "learning_rate": 1.742453939631517e-05,
+      "loss": 0.3749,
+      "step": 1947
+    },
+    {
+      "epoch": 2.0613756613756613,
+      "grad_norm": 0.19379588380401883,
+      "learning_rate": 1.7404939239513917e-05,
+      "loss": 0.3103,
+      "step": 1948
+    },
+    {
+      "epoch": 2.0624338624338625,
+      "grad_norm": 0.22978352813918101,
+      "learning_rate": 1.7385339082712663e-05,
+      "loss": 0.3673,
+      "step": 1949
+    },
+    {
+      "epoch": 2.0634920634920633,
+      "grad_norm": 0.2212514519753634,
+      "learning_rate": 1.736573892591141e-05,
+      "loss": 0.3597,
+      "step": 1950
+    },
+    {
+      "epoch": 2.0645502645502645,
+      "grad_norm": 0.20997192399458178,
+      "learning_rate": 1.7346138769110155e-05,
+      "loss": 0.3736,
+      "step": 1951
+    },
+    {
+      "epoch": 2.0656084656084657,
+      "grad_norm": 0.20665636238190924,
+      "learning_rate": 1.7326538612308898e-05,
+      "loss": 0.3523,
+      "step": 1952
+    },
+    {
+      "epoch": 2.066666666666667,
+      "grad_norm": 0.20758995846695377,
+      "learning_rate": 1.7306938455507647e-05,
+      "loss": 0.3748,
+      "step": 1953
+    },
+    {
+      "epoch": 2.0677248677248676,
+      "grad_norm": 0.22564678363361657,
+      "learning_rate": 1.728733829870639e-05,
+      "loss": 0.336,
+      "step": 1954
+    },
+    {
+      "epoch": 2.068783068783069,
+      "grad_norm": 0.2516870186148532,
+      "learning_rate": 1.7267738141905135e-05,
+      "loss": 0.4016,
+      "step": 1955
+    },
+    {
+      "epoch": 2.06984126984127,
+      "grad_norm": 0.20252710114468478,
+      "learning_rate": 1.724813798510388e-05,
+      "loss": 0.3567,
+      "step": 1956
+    },
+    {
+      "epoch": 2.070899470899471,
+      "grad_norm": 0.2282832684870184,
+      "learning_rate": 1.7228537828302627e-05,
+      "loss": 0.3136,
+      "step": 1957
+    },
+    {
+      "epoch": 2.071957671957672,
+      "grad_norm": 0.2797927938402852,
+      "learning_rate": 1.7208937671501373e-05,
+      "loss": 0.3932,
+      "step": 1958
+    },
+    {
+      "epoch": 2.073015873015873,
+      "grad_norm": 0.20461018617775642,
+      "learning_rate": 1.718933751470012e-05,
+      "loss": 0.3397,
+      "step": 1959
+    },
+    {
+      "epoch": 2.074074074074074,
+      "grad_norm": 0.21544569146507456,
+      "learning_rate": 1.7169737357898865e-05,
+      "loss": 0.3519,
+      "step": 1960
+    },
+    {
+      "epoch": 2.075132275132275,
+      "grad_norm": 0.22183783680712937,
+      "learning_rate": 1.715013720109761e-05,
+      "loss": 0.3515,
+      "step": 1961
+    },
+    {
+      "epoch": 2.0761904761904764,
+      "grad_norm": 0.22146564807753902,
+      "learning_rate": 1.7130537044296356e-05,
+      "loss": 0.3354,
+      "step": 1962
+    },
+    {
+      "epoch": 2.077248677248677,
+      "grad_norm": 0.19857023572678106,
+      "learning_rate": 1.7110936887495102e-05,
+      "loss": 0.3213,
+      "step": 1963
+    },
+    {
+      "epoch": 2.0783068783068783,
+      "grad_norm": 0.217266771807875,
+      "learning_rate": 1.7091336730693848e-05,
+      "loss": 0.3532,
+      "step": 1964
+    },
+    {
+      "epoch": 2.0793650793650795,
+      "grad_norm": 0.20005495691125305,
+      "learning_rate": 1.707173657389259e-05,
+      "loss": 0.3409,
+      "step": 1965
+    },
+    {
+      "epoch": 2.0804232804232803,
+      "grad_norm": 0.21330150110247303,
+      "learning_rate": 1.705213641709134e-05,
+      "loss": 0.3741,
+      "step": 1966
+    },
+    {
+      "epoch": 2.0814814814814815,
+      "grad_norm": 0.2278780973049878,
+      "learning_rate": 1.7032536260290082e-05,
+      "loss": 0.3468,
+      "step": 1967
+    },
+    {
+      "epoch": 2.0825396825396827,
+      "grad_norm": 0.2109353272677484,
+      "learning_rate": 1.7012936103488828e-05,
+      "loss": 0.3471,
+      "step": 1968
+    },
+    {
+      "epoch": 2.0835978835978834,
+      "grad_norm": 0.19895351419178237,
+      "learning_rate": 1.6993335946687574e-05,
+      "loss": 0.3492,
+      "step": 1969
+    },
+    {
+      "epoch": 2.0846560846560847,
+      "grad_norm": 0.207632263472837,
+      "learning_rate": 1.697373578988632e-05,
+      "loss": 0.3399,
+      "step": 1970
+    },
+    {
+      "epoch": 2.085714285714286,
+      "grad_norm": 0.2163337881929324,
+      "learning_rate": 1.6954135633085066e-05,
+      "loss": 0.3255,
+      "step": 1971
+    },
+    {
+      "epoch": 2.0867724867724866,
+      "grad_norm": 0.23040692314017927,
+      "learning_rate": 1.6934535476283812e-05,
+      "loss": 0.3606,
+      "step": 1972
+    },
+    {
+      "epoch": 2.087830687830688,
+      "grad_norm": 0.21989016356512248,
+      "learning_rate": 1.6914935319482554e-05,
+      "loss": 0.3647,
+      "step": 1973
+    },
+    {
+      "epoch": 2.088888888888889,
+      "grad_norm": 0.19613496455842805,
+      "learning_rate": 1.6895335162681303e-05,
+      "loss": 0.3203,
+      "step": 1974
+    },
+    {
+      "epoch": 2.0899470899470898,
+      "grad_norm": 0.22567442720889747,
+      "learning_rate": 1.6875735005880046e-05,
+      "loss": 0.3365,
+      "step": 1975
+    },
+    {
+      "epoch": 2.091005291005291,
+      "grad_norm": 0.22947949949294724,
+      "learning_rate": 1.6856134849078795e-05,
+      "loss": 0.3584,
+      "step": 1976
+    },
+    {
+      "epoch": 2.092063492063492,
+      "grad_norm": 0.23041866230411318,
+      "learning_rate": 1.6836534692277538e-05,
+      "loss": 0.3731,
+      "step": 1977
+    },
+    {
+      "epoch": 2.093121693121693,
+      "grad_norm": 0.21985790184342138,
+      "learning_rate": 1.6816934535476284e-05,
+      "loss": 0.3709,
+      "step": 1978
+    },
+    {
+      "epoch": 2.094179894179894,
+      "grad_norm": 0.19774950421675602,
+      "learning_rate": 1.679733437867503e-05,
+      "loss": 0.3409,
+      "step": 1979
+    },
+    {
+      "epoch": 2.0952380952380953,
+      "grad_norm": 0.2269563084196834,
+      "learning_rate": 1.6777734221873775e-05,
+      "loss": 0.3242,
+      "step": 1980
+    },
+    {
+      "epoch": 2.096296296296296,
+      "grad_norm": 0.23498422047692288,
+      "learning_rate": 1.675813406507252e-05,
+      "loss": 0.3583,
+      "step": 1981
+    },
+    {
+      "epoch": 2.0973544973544973,
+      "grad_norm": 0.20236703887981325,
+      "learning_rate": 1.6738533908271267e-05,
+      "loss": 0.3557,
+      "step": 1982
+    },
+    {
+      "epoch": 2.0984126984126985,
+      "grad_norm": 0.20883374092288098,
+      "learning_rate": 1.6718933751470013e-05,
+      "loss": 0.3285,
+      "step": 1983
+    },
+    {
+      "epoch": 2.0994708994708993,
+      "grad_norm": 0.2366397155580664,
+      "learning_rate": 1.669933359466876e-05,
+      "loss": 0.3539,
+      "step": 1984
+    },
+    {
+      "epoch": 2.1005291005291005,
+      "grad_norm": 0.2178711255251422,
+      "learning_rate": 1.6679733437867505e-05,
+      "loss": 0.329,
+      "step": 1985
+    },
+    {
+      "epoch": 2.1015873015873017,
+      "grad_norm": 0.21131026834720637,
+      "learning_rate": 1.6660133281066247e-05,
+      "loss": 0.3361,
+      "step": 1986
+    },
+    {
+      "epoch": 2.102645502645503,
+      "grad_norm": 0.2055578261425351,
+      "learning_rate": 1.6640533124264996e-05,
+      "loss": 0.3032,
+      "step": 1987
+    },
+    {
+      "epoch": 2.1037037037037036,
+      "grad_norm": 0.21812782402014477,
+      "learning_rate": 1.662093296746374e-05,
+      "loss": 0.296,
+      "step": 1988
+    },
+    {
+      "epoch": 2.104761904761905,
+      "grad_norm": 0.21287530507237762,
+      "learning_rate": 1.6601332810662488e-05,
+      "loss": 0.351,
+      "step": 1989
+    },
+    {
+      "epoch": 2.105820105820106,
+      "grad_norm": 0.20950973871134726,
+      "learning_rate": 1.658173265386123e-05,
+      "loss": 0.3575,
+      "step": 1990
+    },
+    {
+      "epoch": 2.106878306878307,
+      "grad_norm": 0.21396955753725935,
+      "learning_rate": 1.6562132497059977e-05,
+      "loss": 0.3424,
+      "step": 1991
+    },
+    {
+      "epoch": 2.107936507936508,
+      "grad_norm": 0.22616878855251082,
+      "learning_rate": 1.6542532340258722e-05,
+      "loss": 0.3662,
+      "step": 1992
+    },
+    {
+      "epoch": 2.108994708994709,
+      "grad_norm": 0.20008820458876664,
+      "learning_rate": 1.6522932183457468e-05,
+      "loss": 0.3108,
+      "step": 1993
+    },
+    {
+      "epoch": 2.11005291005291,
+      "grad_norm": 4.143863598055369,
+      "learning_rate": 1.6503332026656214e-05,
+      "loss": 0.4833,
+      "step": 1994
+    },
+    {
+      "epoch": 2.111111111111111,
+      "grad_norm": 0.2292235575997881,
+      "learning_rate": 1.648373186985496e-05,
+      "loss": 0.3343,
+      "step": 1995
+    },
+    {
+      "epoch": 2.1121693121693124,
+      "grad_norm": 0.23419107839319792,
+      "learning_rate": 1.6464131713053706e-05,
+      "loss": 0.3347,
+      "step": 1996
+    },
+    {
+      "epoch": 2.113227513227513,
+      "grad_norm": 0.22940581819547085,
+      "learning_rate": 1.6444531556252452e-05,
+      "loss": 0.3951,
+      "step": 1997
+    },
+    {
+      "epoch": 2.1142857142857143,
+      "grad_norm": 0.23526414274011145,
+      "learning_rate": 1.6424931399451198e-05,
+      "loss": 0.3424,
+      "step": 1998
+    },
+    {
+      "epoch": 2.1153439153439155,
+      "grad_norm": 0.22735375376942824,
+      "learning_rate": 1.6405331242649944e-05,
+      "loss": 0.3649,
+      "step": 1999
+    },
+    {
+      "epoch": 2.1164021164021163,
+      "grad_norm": 0.21944975202575878,
+      "learning_rate": 1.638573108584869e-05,
+      "loss": 0.3376,
+      "step": 2000
+    },
+    {
+      "epoch": 2.1174603174603175,
+      "grad_norm": 0.5029143453384859,
+      "learning_rate": 1.6366130929047432e-05,
+      "loss": 0.3368,
+      "step": 2001
+    },
+    {
+      "epoch": 2.1185185185185187,
+      "grad_norm": 0.2504647540949935,
+      "learning_rate": 1.634653077224618e-05,
+      "loss": 0.365,
+      "step": 2002
+    },
+    {
+      "epoch": 2.1195767195767194,
+      "grad_norm": 0.2468901550955016,
+      "learning_rate": 1.6326930615444924e-05,
+      "loss": 0.3448,
+      "step": 2003
+    },
+    {
+      "epoch": 2.1206349206349207,
+      "grad_norm": 0.2197505442465071,
+      "learning_rate": 1.630733045864367e-05,
+      "loss": 0.361,
+      "step": 2004
+    },
+    {
+      "epoch": 2.121693121693122,
+      "grad_norm": 0.25391417814496214,
+      "learning_rate": 1.6287730301842415e-05,
+      "loss": 0.3581,
+      "step": 2005
+    },
+    {
+      "epoch": 2.1227513227513226,
+      "grad_norm": 0.25883936021880843,
+      "learning_rate": 1.626813014504116e-05,
+      "loss": 0.3505,
+      "step": 2006
+    },
+    {
+      "epoch": 2.123809523809524,
+      "grad_norm": 0.2520258145131729,
+      "learning_rate": 1.6248529988239907e-05,
+      "loss": 0.3398,
+      "step": 2007
+    },
+    {
+      "epoch": 2.124867724867725,
+      "grad_norm": 0.2203661316211868,
+      "learning_rate": 1.6228929831438653e-05,
+      "loss": 0.3627,
+      "step": 2008
+    },
+    {
+      "epoch": 2.1259259259259258,
+      "grad_norm": 0.23435328040308137,
+      "learning_rate": 1.6209329674637396e-05,
+      "loss": 0.3389,
+      "step": 2009
+    },
+    {
+      "epoch": 2.126984126984127,
+      "grad_norm": 0.27743951042442594,
+      "learning_rate": 1.6189729517836145e-05,
+      "loss": 0.3761,
+      "step": 2010
+    },
+    {
+      "epoch": 2.128042328042328,
+      "grad_norm": 0.20244205174436367,
+      "learning_rate": 1.6170129361034887e-05,
+      "loss": 0.3152,
+      "step": 2011
+    },
+    {
+      "epoch": 2.129100529100529,
+      "grad_norm": 0.20894323608250834,
+      "learning_rate": 1.6150529204233637e-05,
+      "loss": 0.3854,
+      "step": 2012
+    },
+    {
+      "epoch": 2.13015873015873,
+      "grad_norm": 0.24428589140845183,
+      "learning_rate": 1.613092904743238e-05,
+      "loss": 0.3876,
+      "step": 2013
+    },
+    {
+      "epoch": 2.1312169312169313,
+      "grad_norm": 0.24322774885752654,
+      "learning_rate": 1.6111328890631125e-05,
+      "loss": 0.3581,
+      "step": 2014
+    },
+    {
+      "epoch": 2.132275132275132,
+      "grad_norm": 0.20361725365232156,
+      "learning_rate": 1.609172873382987e-05,
+      "loss": 0.288,
+      "step": 2015
+    },
+    {
+      "epoch": 2.1333333333333333,
+      "grad_norm": 0.23886166021463448,
+      "learning_rate": 1.6072128577028617e-05,
+      "loss": 0.3867,
+      "step": 2016
+    },
+    {
+      "epoch": 2.1343915343915345,
+      "grad_norm": 0.20682128386588214,
+      "learning_rate": 1.6052528420227363e-05,
+      "loss": 0.3289,
+      "step": 2017
+    },
+    {
+      "epoch": 2.1354497354497353,
+      "grad_norm": 0.22555396551236676,
+      "learning_rate": 1.603292826342611e-05,
+      "loss": 0.336,
+      "step": 2018
+    },
+    {
+      "epoch": 2.1365079365079365,
+      "grad_norm": 0.22141671607674948,
+      "learning_rate": 1.6013328106624854e-05,
+      "loss": 0.3584,
+      "step": 2019
+    },
+    {
+      "epoch": 2.1375661375661377,
+      "grad_norm": 0.21896453846877967,
+      "learning_rate": 1.59937279498236e-05,
+      "loss": 0.3527,
+      "step": 2020
+    },
+    {
+      "epoch": 2.1386243386243384,
+      "grad_norm": 0.22228976521690386,
+      "learning_rate": 1.5974127793022346e-05,
+      "loss": 0.3923,
+      "step": 2021
+    },
+    {
+      "epoch": 2.1396825396825396,
+      "grad_norm": 0.22066786689884066,
+      "learning_rate": 1.595452763622109e-05,
+      "loss": 0.3625,
+      "step": 2022
+    },
+    {
+      "epoch": 2.140740740740741,
+      "grad_norm": 0.22383398867863769,
+      "learning_rate": 1.5934927479419838e-05,
+      "loss": 0.3741,
+      "step": 2023
+    },
+    {
+      "epoch": 2.1417989417989416,
+      "grad_norm": 0.21926212085215122,
+      "learning_rate": 1.591532732261858e-05,
+      "loss": 0.305,
+      "step": 2024
+    },
+    {
+      "epoch": 2.142857142857143,
+      "grad_norm": 0.21328408231111298,
+      "learning_rate": 1.589572716581733e-05,
+      "loss": 0.3539,
+      "step": 2025
+    },
+    {
+      "epoch": 2.143915343915344,
+      "grad_norm": 0.20310210789488775,
+      "learning_rate": 1.5876127009016072e-05,
+      "loss": 0.3406,
+      "step": 2026
+    },
+    {
+      "epoch": 2.1449735449735448,
+      "grad_norm": 0.20985690382441458,
+      "learning_rate": 1.5856526852214818e-05,
+      "loss": 0.3298,
+      "step": 2027
+    },
+    {
+      "epoch": 2.146031746031746,
+      "grad_norm": 0.21426037653049163,
+      "learning_rate": 1.5836926695413564e-05,
+      "loss": 0.3076,
+      "step": 2028
+    },
+    {
+      "epoch": 2.147089947089947,
+      "grad_norm": 0.21677917505372724,
+      "learning_rate": 1.581732653861231e-05,
+      "loss": 0.3824,
+      "step": 2029
+    },
+    {
+      "epoch": 2.148148148148148,
+      "grad_norm": 0.2064869233207918,
+      "learning_rate": 1.5797726381811055e-05,
+      "loss": 0.3436,
+      "step": 2030
+    },
+    {
+      "epoch": 2.149206349206349,
+      "grad_norm": 0.21971899343914275,
+      "learning_rate": 1.57781262250098e-05,
+      "loss": 0.369,
+      "step": 2031
+    },
+    {
+      "epoch": 2.1502645502645503,
+      "grad_norm": 0.3221845152874034,
+      "learning_rate": 1.5758526068208547e-05,
+      "loss": 0.3425,
+      "step": 2032
+    },
+    {
+      "epoch": 2.1513227513227515,
+      "grad_norm": 0.19874490771824377,
+      "learning_rate": 1.5738925911407293e-05,
+      "loss": 0.3184,
+      "step": 2033
+    },
+    {
+      "epoch": 2.1523809523809523,
+      "grad_norm": 10.905047393025432,
+      "learning_rate": 1.571932575460604e-05,
+      "loss": 1.0908,
+      "step": 2034
+    },
+    {
+      "epoch": 2.1534391534391535,
+      "grad_norm": 0.21735676542498997,
+      "learning_rate": 1.569972559780478e-05,
+      "loss": 0.3473,
+      "step": 2035
+    },
+    {
+      "epoch": 2.1544973544973547,
+      "grad_norm": 0.216545692844694,
+      "learning_rate": 1.568012544100353e-05,
+      "loss": 0.3582,
+      "step": 2036
+    },
+    {
+      "epoch": 2.1555555555555554,
+      "grad_norm": 0.21730583612765947,
+      "learning_rate": 1.5660525284202273e-05,
+      "loss": 0.3126,
+      "step": 2037
+    },
+    {
+      "epoch": 2.1566137566137566,
+      "grad_norm": 0.2004424058198479,
+      "learning_rate": 1.5640925127401022e-05,
+      "loss": 0.34,
+      "step": 2038
+    },
+    {
+      "epoch": 2.157671957671958,
+      "grad_norm": 0.2118306305932416,
+      "learning_rate": 1.5621324970599765e-05,
+      "loss": 0.307,
+      "step": 2039
+    },
+    {
+      "epoch": 2.1587301587301586,
+      "grad_norm": 0.22105484159810915,
+      "learning_rate": 1.560172481379851e-05,
+      "loss": 0.3747,
+      "step": 2040
+    },
+    {
+      "epoch": 2.15978835978836,
+      "grad_norm": 0.20960788510999773,
+      "learning_rate": 1.5582124656997257e-05,
+      "loss": 0.3417,
+      "step": 2041
+    },
+    {
+      "epoch": 2.160846560846561,
+      "grad_norm": 0.21884953306007163,
+      "learning_rate": 1.5562524500196003e-05,
+      "loss": 0.3512,
+      "step": 2042
+    },
+    {
+      "epoch": 2.1619047619047618,
+      "grad_norm": 0.23243491580846684,
+      "learning_rate": 1.554292434339475e-05,
+      "loss": 0.3841,
+      "step": 2043
+    },
+    {
+      "epoch": 2.162962962962963,
+      "grad_norm": 0.23140426219629298,
+      "learning_rate": 1.5523324186593494e-05,
+      "loss": 0.405,
+      "step": 2044
+    },
+    {
+      "epoch": 2.164021164021164,
+      "grad_norm": 0.20298786257942408,
+      "learning_rate": 1.5503724029792237e-05,
+      "loss": 0.3551,
+      "step": 2045
+    },
+    {
+      "epoch": 2.165079365079365,
+      "grad_norm": 0.21348196250544565,
+      "learning_rate": 1.5484123872990986e-05,
+      "loss": 0.3378,
+      "step": 2046
+    },
+    {
+      "epoch": 2.166137566137566,
+      "grad_norm": 0.2439977018458661,
+      "learning_rate": 1.546452371618973e-05,
+      "loss": 0.3773,
+      "step": 2047
+    },
+    {
+      "epoch": 2.1671957671957673,
+      "grad_norm": 0.19895306399691473,
+      "learning_rate": 1.5444923559388474e-05,
+      "loss": 0.3209,
+      "step": 2048
+    },
+    {
+      "epoch": 2.168253968253968,
+      "grad_norm": 0.21695598379114167,
+      "learning_rate": 1.542532340258722e-05,
+      "loss": 0.3675,
+      "step": 2049
+    },
+    {
+      "epoch": 2.1693121693121693,
+      "grad_norm": 0.1794392725994556,
+      "learning_rate": 1.5405723245785966e-05,
+      "loss": 0.2931,
+      "step": 2050
+    },
+    {
+      "epoch": 2.1703703703703705,
+      "grad_norm": 0.2038700225689888,
+      "learning_rate": 1.5386123088984712e-05,
+      "loss": 0.3375,
+      "step": 2051
+    },
+    {
+      "epoch": 2.1714285714285713,
+      "grad_norm": 0.22597709606412955,
+      "learning_rate": 1.5366522932183458e-05,
+      "loss": 0.3576,
+      "step": 2052
+    },
+    {
+      "epoch": 2.1724867724867725,
+      "grad_norm": 0.2077246480006038,
+      "learning_rate": 1.5346922775382204e-05,
+      "loss": 0.3344,
+      "step": 2053
+    },
+    {
+      "epoch": 2.1735449735449737,
+      "grad_norm": 0.20000941756704047,
+      "learning_rate": 1.532732261858095e-05,
+      "loss": 0.3382,
+      "step": 2054
+    },
+    {
+      "epoch": 2.1746031746031744,
+      "grad_norm": 0.19713016775489853,
+      "learning_rate": 1.5307722461779696e-05,
+      "loss": 0.3333,
+      "step": 2055
+    },
+    {
+      "epoch": 2.1756613756613756,
+      "grad_norm": 0.19246485509834052,
+      "learning_rate": 1.528812230497844e-05,
+      "loss": 0.305,
+      "step": 2056
+    },
+    {
+      "epoch": 2.176719576719577,
+      "grad_norm": 0.20627954988108957,
+      "learning_rate": 1.5268522148177187e-05,
+      "loss": 0.3102,
+      "step": 2057
+    },
+    {
+      "epoch": 2.1777777777777776,
+      "grad_norm": 0.21828606128129674,
+      "learning_rate": 1.5248921991375931e-05,
+      "loss": 0.38,
+      "step": 2058
+    },
+    {
+      "epoch": 2.178835978835979,
+      "grad_norm": 0.20648370010623027,
+      "learning_rate": 1.5229321834574677e-05,
+      "loss": 0.3356,
+      "step": 2059
+    },
+    {
+      "epoch": 2.17989417989418,
+      "grad_norm": 0.21130639938010654,
+      "learning_rate": 1.5209721677773423e-05,
+      "loss": 0.3638,
+      "step": 2060
+    },
+    {
+      "epoch": 2.1809523809523808,
+      "grad_norm": 0.2115670503852573,
+      "learning_rate": 1.5190121520972169e-05,
+      "loss": 0.3432,
+      "step": 2061
+    },
+    {
+      "epoch": 2.182010582010582,
+      "grad_norm": 0.1908436166806636,
+      "learning_rate": 1.5170521364170915e-05,
+      "loss": 0.3041,
+      "step": 2062
+    },
+    {
+      "epoch": 2.183068783068783,
+      "grad_norm": 0.2020564072365534,
+      "learning_rate": 1.515092120736966e-05,
+      "loss": 0.336,
+      "step": 2063
+    },
+    {
+      "epoch": 2.1841269841269844,
+      "grad_norm": 0.2161693720721347,
+      "learning_rate": 1.5131321050568407e-05,
+      "loss": 0.3545,
+      "step": 2064
+    },
+    {
+      "epoch": 2.185185185185185,
+      "grad_norm": 0.22473602950532073,
+      "learning_rate": 1.5111720893767151e-05,
+      "loss": 0.342,
+      "step": 2065
+    },
+    {
+      "epoch": 2.1862433862433863,
+      "grad_norm": 0.2150375377284277,
+      "learning_rate": 1.5092120736965895e-05,
+      "loss": 0.3833,
+      "step": 2066
+    },
+    {
+      "epoch": 2.1873015873015875,
+      "grad_norm": 0.21870865973475231,
+      "learning_rate": 1.5072520580164643e-05,
+      "loss": 0.3534,
+      "step": 2067
+    },
+    {
+      "epoch": 2.1883597883597883,
+      "grad_norm": 0.21446734495632633,
+      "learning_rate": 1.5052920423363387e-05,
+      "loss": 0.3606,
+      "step": 2068
+    },
+    {
+      "epoch": 2.1894179894179895,
+      "grad_norm": 0.22888453880818366,
+      "learning_rate": 1.5033320266562134e-05,
+      "loss": 0.3907,
+      "step": 2069
+    },
+    {
+      "epoch": 2.1904761904761907,
+      "grad_norm": 0.21937778969235877,
+      "learning_rate": 1.5013720109760879e-05,
+      "loss": 0.3293,
+      "step": 2070
+    },
+    {
+      "epoch": 2.1915343915343914,
+      "grad_norm": 0.20058436964581325,
+      "learning_rate": 1.4994119952959623e-05,
+      "loss": 0.3137,
+      "step": 2071
+    },
+    {
+      "epoch": 2.1925925925925926,
+      "grad_norm": 0.23325324698372354,
+      "learning_rate": 1.497451979615837e-05,
+      "loss": 0.3512,
+      "step": 2072
+    },
+    {
+      "epoch": 2.193650793650794,
+      "grad_norm": 0.21063385449656516,
+      "learning_rate": 1.4954919639357115e-05,
+      "loss": 0.3487,
+      "step": 2073
+    },
+    {
+      "epoch": 2.1947089947089946,
+      "grad_norm": 0.22182976034778512,
+      "learning_rate": 1.4935319482555862e-05,
+      "loss": 0.3783,
+      "step": 2074
+    },
+    {
+      "epoch": 2.195767195767196,
+      "grad_norm": 0.20256863181044094,
+      "learning_rate": 1.4915719325754606e-05,
+      "loss": 0.3231,
+      "step": 2075
+    },
+    {
+      "epoch": 2.196825396825397,
+      "grad_norm": 0.23952493244674197,
+      "learning_rate": 1.4896119168953352e-05,
+      "loss": 0.3828,
+      "step": 2076
+    },
+    {
+      "epoch": 2.1978835978835978,
+      "grad_norm": 0.2161109491007153,
+      "learning_rate": 1.4876519012152098e-05,
+      "loss": 0.3379,
+      "step": 2077
+    },
+    {
+      "epoch": 2.198941798941799,
+      "grad_norm": 0.21208428593234505,
+      "learning_rate": 1.4856918855350844e-05,
+      "loss": 0.3386,
+      "step": 2078
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.20469783702775474,
+      "learning_rate": 1.483731869854959e-05,
+      "loss": 0.3285,
+      "step": 2079
+    },
+    {
+      "epoch": 2.201058201058201,
+      "grad_norm": 0.20106413317496333,
+      "learning_rate": 1.4817718541748336e-05,
+      "loss": 0.3191,
+      "step": 2080
+    },
+    {
+      "epoch": 2.202116402116402,
+      "grad_norm": 0.20886186796806508,
+      "learning_rate": 1.479811838494708e-05,
+      "loss": 0.3638,
+      "step": 2081
+    },
+    {
+      "epoch": 2.2031746031746033,
+      "grad_norm": 0.2024596948339384,
+      "learning_rate": 1.4778518228145827e-05,
+      "loss": 0.3497,
+      "step": 2082
+    },
+    {
+      "epoch": 2.204232804232804,
+      "grad_norm": 0.19966325145216773,
+      "learning_rate": 1.4758918071344572e-05,
+      "loss": 0.3598,
+      "step": 2083
+    },
+    {
+      "epoch": 2.2052910052910053,
+      "grad_norm": 0.19469952860860135,
+      "learning_rate": 1.4739317914543316e-05,
+      "loss": 0.3035,
+      "step": 2084
+    },
+    {
+      "epoch": 2.2063492063492065,
+      "grad_norm": 0.21170064193002702,
+      "learning_rate": 1.4719717757742063e-05,
+      "loss": 0.304,
+      "step": 2085
+    },
+    {
+      "epoch": 2.2074074074074073,
+      "grad_norm": 0.20805361694025837,
+      "learning_rate": 1.4700117600940808e-05,
+      "loss": 0.3511,
+      "step": 2086
+    },
+    {
+      "epoch": 2.2084656084656085,
+      "grad_norm": 0.2055245437744949,
+      "learning_rate": 1.4680517444139555e-05,
+      "loss": 0.3544,
+      "step": 2087
+    },
+    {
+      "epoch": 2.2095238095238097,
+      "grad_norm": 0.201237038494924,
+      "learning_rate": 1.46609172873383e-05,
+      "loss": 0.3253,
+      "step": 2088
+    },
+    {
+      "epoch": 2.2105820105820104,
+      "grad_norm": 0.22354747868539224,
+      "learning_rate": 1.4641317130537043e-05,
+      "loss": 0.3795,
+      "step": 2089
+    },
+    {
+      "epoch": 2.2116402116402116,
+      "grad_norm": 0.22190512018803507,
+      "learning_rate": 1.4621716973735791e-05,
+      "loss": 0.3626,
+      "step": 2090
+    },
+    {
+      "epoch": 2.212698412698413,
+      "grad_norm": 0.3024061933530375,
+      "learning_rate": 1.4602116816934535e-05,
+      "loss": 0.3426,
+      "step": 2091
+    },
+    {
+      "epoch": 2.2137566137566136,
+      "grad_norm": 0.22362434974122583,
+      "learning_rate": 1.4582516660133283e-05,
+      "loss": 0.3568,
+      "step": 2092
+    },
+    {
+      "epoch": 2.214814814814815,
+      "grad_norm": 0.2120000539713426,
+      "learning_rate": 1.4562916503332027e-05,
+      "loss": 0.3436,
+      "step": 2093
+    },
+    {
+      "epoch": 2.215873015873016,
+      "grad_norm": 0.2273394265772914,
+      "learning_rate": 1.4543316346530773e-05,
+      "loss": 0.347,
+      "step": 2094
+    },
+    {
+      "epoch": 2.2169312169312168,
+      "grad_norm": 0.20044541135515623,
+      "learning_rate": 1.4523716189729519e-05,
+      "loss": 0.3415,
+      "step": 2095
+    },
+    {
+      "epoch": 2.217989417989418,
+      "grad_norm": 0.20384591070530153,
+      "learning_rate": 1.4504116032928265e-05,
+      "loss": 0.3278,
+      "step": 2096
+    },
+    {
+      "epoch": 2.219047619047619,
+      "grad_norm": 0.25810597441442096,
+      "learning_rate": 1.4484515876127009e-05,
+      "loss": 0.3534,
+      "step": 2097
+    },
+    {
+      "epoch": 2.22010582010582,
+      "grad_norm": 0.22795352736979752,
+      "learning_rate": 1.4464915719325756e-05,
+      "loss": 0.3455,
+      "step": 2098
+    },
+    {
+      "epoch": 2.221164021164021,
+      "grad_norm": 0.21049254714677118,
+      "learning_rate": 1.44453155625245e-05,
+      "loss": 0.3808,
+      "step": 2099
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 0.23067771453190306,
+      "learning_rate": 1.4425715405723248e-05,
+      "loss": 0.3285,
+      "step": 2100
+    },
+    {
+      "epoch": 2.223280423280423,
+      "grad_norm": 0.2200823890048532,
+      "learning_rate": 1.4406115248921992e-05,
+      "loss": 0.3497,
+      "step": 2101
+    },
+    {
+      "epoch": 2.2243386243386243,
+      "grad_norm": 3.4221269378218153,
+      "learning_rate": 1.4386515092120736e-05,
+      "loss": 0.4657,
+      "step": 2102
+    },
+    {
+      "epoch": 2.2253968253968255,
+      "grad_norm": 0.23261505421987935,
+      "learning_rate": 1.4366914935319484e-05,
+      "loss": 0.3376,
+      "step": 2103
+    },
+    {
+      "epoch": 2.2264550264550262,
+      "grad_norm": 0.20182994591620118,
+      "learning_rate": 1.4347314778518228e-05,
+      "loss": 0.3526,
+      "step": 2104
+    },
+    {
+      "epoch": 2.2275132275132274,
+      "grad_norm": 0.21082285581411647,
+      "learning_rate": 1.4327714621716976e-05,
+      "loss": 0.3552,
+      "step": 2105
+    },
+    {
+      "epoch": 2.2285714285714286,
+      "grad_norm": 0.22032489334949037,
+      "learning_rate": 1.430811446491572e-05,
+      "loss": 0.3756,
+      "step": 2106
+    },
+    {
+      "epoch": 2.2296296296296294,
+      "grad_norm": 0.2159520496931439,
+      "learning_rate": 1.4288514308114464e-05,
+      "loss": 0.3592,
+      "step": 2107
+    },
+    {
+      "epoch": 2.2306878306878306,
+      "grad_norm": 0.20689680272282845,
+      "learning_rate": 1.4268914151313212e-05,
+      "loss": 0.3491,
+      "step": 2108
+    },
+    {
+      "epoch": 2.231746031746032,
+      "grad_norm": 0.19169011186080528,
+      "learning_rate": 1.4249313994511956e-05,
+      "loss": 0.3355,
+      "step": 2109
+    },
+    {
+      "epoch": 2.2328042328042326,
+      "grad_norm": 0.19803274438625507,
+      "learning_rate": 1.4229713837710703e-05,
+      "loss": 0.3211,
+      "step": 2110
+    },
+    {
+      "epoch": 2.2338624338624338,
+      "grad_norm": 0.2169935695452419,
+      "learning_rate": 1.4210113680909448e-05,
+      "loss": 0.3789,
+      "step": 2111
+    },
+    {
+      "epoch": 2.234920634920635,
+      "grad_norm": 0.20431499248719218,
+      "learning_rate": 1.4190513524108193e-05,
+      "loss": 0.3162,
+      "step": 2112
+    },
+    {
+      "epoch": 2.235978835978836,
+      "grad_norm": 0.2079250549519578,
+      "learning_rate": 1.417091336730694e-05,
+      "loss": 0.3469,
+      "step": 2113
+    },
+    {
+      "epoch": 2.237037037037037,
+      "grad_norm": 0.22941584714468002,
+      "learning_rate": 1.4151313210505685e-05,
+      "loss": 0.405,
+      "step": 2114
+    },
+    {
+      "epoch": 2.238095238095238,
+      "grad_norm": 0.19630828002495715,
+      "learning_rate": 1.413171305370443e-05,
+      "loss": 0.3515,
+      "step": 2115
+    },
+    {
+      "epoch": 2.2391534391534393,
+      "grad_norm": 0.210404649350759,
+      "learning_rate": 1.4112112896903177e-05,
+      "loss": 0.3376,
+      "step": 2116
+    },
+    {
+      "epoch": 2.24021164021164,
+      "grad_norm": 0.2376328983905229,
+      "learning_rate": 1.4092512740101921e-05,
+      "loss": 0.388,
+      "step": 2117
+    },
+    {
+      "epoch": 2.2412698412698413,
+      "grad_norm": 0.21383169199679727,
+      "learning_rate": 1.4072912583300669e-05,
+      "loss": 0.3199,
+      "step": 2118
+    },
+    {
+      "epoch": 2.2423280423280425,
+      "grad_norm": 0.20202389531154502,
+      "learning_rate": 1.4053312426499413e-05,
+      "loss": 0.3684,
+      "step": 2119
+    },
+    {
+      "epoch": 2.2433862433862433,
+      "grad_norm": 0.32250759227392084,
+      "learning_rate": 1.4033712269698157e-05,
+      "loss": 0.3633,
+      "step": 2120
+    },
+    {
+      "epoch": 2.2444444444444445,
+      "grad_norm": 0.23290731003562973,
+      "learning_rate": 1.4014112112896905e-05,
+      "loss": 0.4027,
+      "step": 2121
+    },
+    {
+      "epoch": 2.2455026455026457,
+      "grad_norm": 0.2116123949601418,
+      "learning_rate": 1.3994511956095649e-05,
+      "loss": 0.3819,
+      "step": 2122
+    },
+    {
+      "epoch": 2.2465608465608464,
+      "grad_norm": 0.21423684536621376,
+      "learning_rate": 1.3974911799294396e-05,
+      "loss": 0.3684,
+      "step": 2123
+    },
+    {
+      "epoch": 2.2476190476190476,
+      "grad_norm": 0.2052592452761068,
+      "learning_rate": 1.395531164249314e-05,
+      "loss": 0.3344,
+      "step": 2124
+    },
+    {
+      "epoch": 2.248677248677249,
+      "grad_norm": 0.20079823755779902,
+      "learning_rate": 1.3935711485691885e-05,
+      "loss": 0.3618,
+      "step": 2125
+    },
+    {
+      "epoch": 2.2497354497354496,
+      "grad_norm": 1.0894959314523671,
+      "learning_rate": 1.3916111328890632e-05,
+      "loss": 0.3475,
+      "step": 2126
+    },
+    {
+      "epoch": 2.250793650793651,
+      "grad_norm": 0.2099112875651436,
+      "learning_rate": 1.3896511172089377e-05,
+      "loss": 0.3252,
+      "step": 2127
+    },
+    {
+      "epoch": 2.251851851851852,
+      "grad_norm": 0.20136693529793254,
+      "learning_rate": 1.3876911015288124e-05,
+      "loss": 0.3378,
+      "step": 2128
+    },
+    {
+      "epoch": 2.2529100529100528,
+      "grad_norm": 0.2105920879605314,
+      "learning_rate": 1.3857310858486868e-05,
+      "loss": 0.3455,
+      "step": 2129
+    },
+    {
+      "epoch": 2.253968253968254,
+      "grad_norm": 0.21728105108056908,
+      "learning_rate": 1.3837710701685614e-05,
+      "loss": 0.3555,
+      "step": 2130
+    },
+    {
+      "epoch": 2.255026455026455,
+      "grad_norm": 0.19483145128816542,
+      "learning_rate": 1.381811054488436e-05,
+      "loss": 0.3163,
+      "step": 2131
+    },
+    {
+      "epoch": 2.256084656084656,
+      "grad_norm": 0.18761723281777778,
+      "learning_rate": 1.3798510388083106e-05,
+      "loss": 0.3057,
+      "step": 2132
+    },
+    {
+      "epoch": 2.257142857142857,
+      "grad_norm": 0.2031218040229401,
+      "learning_rate": 1.377891023128185e-05,
+      "loss": 0.3118,
+      "step": 2133
+    },
+    {
+      "epoch": 2.2582010582010583,
+      "grad_norm": 0.19461075746598536,
+      "learning_rate": 1.3759310074480598e-05,
+      "loss": 0.3307,
+      "step": 2134
+    },
+    {
+      "epoch": 2.259259259259259,
+      "grad_norm": 0.23054743512215764,
+      "learning_rate": 1.3739709917679342e-05,
+      "loss": 0.4172,
+      "step": 2135
+    },
+    {
+      "epoch": 2.2603174603174603,
+      "grad_norm": 0.21187621751878205,
+      "learning_rate": 1.372010976087809e-05,
+      "loss": 0.3666,
+      "step": 2136
+    },
+    {
+      "epoch": 2.2613756613756615,
+      "grad_norm": 0.22640129012026444,
+      "learning_rate": 1.3700509604076834e-05,
+      "loss": 0.3875,
+      "step": 2137
+    },
+    {
+      "epoch": 2.2624338624338622,
+      "grad_norm": 0.19823862095773512,
+      "learning_rate": 1.3680909447275578e-05,
+      "loss": 0.2947,
+      "step": 2138
+    },
+    {
+      "epoch": 2.2634920634920634,
+      "grad_norm": 0.21825276904693997,
+      "learning_rate": 1.3661309290474325e-05,
+      "loss": 0.3833,
+      "step": 2139
+    },
+    {
+      "epoch": 2.2645502645502646,
+      "grad_norm": 0.21409936187628198,
+      "learning_rate": 1.364170913367307e-05,
+      "loss": 0.3485,
+      "step": 2140
+    },
+    {
+      "epoch": 2.265608465608466,
+      "grad_norm": 0.19696631739592757,
+      "learning_rate": 1.3622108976871817e-05,
+      "loss": 0.325,
+      "step": 2141
+    },
+    {
+      "epoch": 2.2666666666666666,
+      "grad_norm": 0.18972546473558874,
+      "learning_rate": 1.3602508820070561e-05,
+      "loss": 0.3315,
+      "step": 2142
+    },
+    {
+      "epoch": 2.267724867724868,
+      "grad_norm": 0.20576822255139696,
+      "learning_rate": 1.3582908663269305e-05,
+      "loss": 0.3432,
+      "step": 2143
+    },
+    {
+      "epoch": 2.268783068783069,
+      "grad_norm": 1.5364918959173623,
+      "learning_rate": 1.3563308506468053e-05,
+      "loss": 0.3966,
+      "step": 2144
+    },
+    {
+      "epoch": 2.2698412698412698,
+      "grad_norm": 0.21478874797886527,
+      "learning_rate": 1.3543708349666797e-05,
+      "loss": 0.3491,
+      "step": 2145
+    },
+    {
+      "epoch": 2.270899470899471,
+      "grad_norm": 0.21276572642119787,
+      "learning_rate": 1.3524108192865541e-05,
+      "loss": 0.3825,
+      "step": 2146
+    },
+    {
+      "epoch": 2.271957671957672,
+      "grad_norm": 0.1955401249488189,
+      "learning_rate": 1.3504508036064289e-05,
+      "loss": 0.3066,
+      "step": 2147
+    },
+    {
+      "epoch": 2.273015873015873,
+      "grad_norm": 0.19895969650589623,
+      "learning_rate": 1.3484907879263035e-05,
+      "loss": 0.3479,
+      "step": 2148
+    },
+    {
+      "epoch": 2.274074074074074,
+      "grad_norm": 0.21571935832308445,
+      "learning_rate": 1.346530772246178e-05,
+      "loss": 0.3688,
+      "step": 2149
+    },
+    {
+      "epoch": 2.2751322751322753,
+      "grad_norm": 0.21735528938360044,
+      "learning_rate": 1.3445707565660527e-05,
+      "loss": 0.3332,
+      "step": 2150
+    },
+    {
+      "epoch": 2.276190476190476,
+      "grad_norm": 0.19334954892944037,
+      "learning_rate": 1.342610740885927e-05,
+      "loss": 0.313,
+      "step": 2151
+    },
+    {
+      "epoch": 2.2772486772486773,
+      "grad_norm": 0.20885026871074683,
+      "learning_rate": 1.3406507252058018e-05,
+      "loss": 0.3296,
+      "step": 2152
+    },
+    {
+      "epoch": 2.2783068783068785,
+      "grad_norm": 0.1976772642773375,
+      "learning_rate": 1.3386907095256762e-05,
+      "loss": 0.3354,
+      "step": 2153
+    },
+    {
+      "epoch": 2.2793650793650793,
+      "grad_norm": 0.21922541090556855,
+      "learning_rate": 1.336730693845551e-05,
+      "loss": 0.3702,
+      "step": 2154
+    },
+    {
+      "epoch": 2.2804232804232805,
+      "grad_norm": 0.22298245670974845,
+      "learning_rate": 1.3347706781654254e-05,
+      "loss": 0.3339,
+      "step": 2155
+    },
+    {
+      "epoch": 2.2814814814814817,
+      "grad_norm": 0.21558822449093906,
+      "learning_rate": 1.3328106624852998e-05,
+      "loss": 0.3699,
+      "step": 2156
+    },
+    {
+      "epoch": 2.2825396825396824,
+      "grad_norm": 0.20618573290980355,
+      "learning_rate": 1.3308506468051746e-05,
+      "loss": 0.3411,
+      "step": 2157
+    },
+    {
+      "epoch": 2.2835978835978836,
+      "grad_norm": 0.22181612653063038,
+      "learning_rate": 1.328890631125049e-05,
+      "loss": 0.3479,
+      "step": 2158
+    },
+    {
+      "epoch": 2.284656084656085,
+      "grad_norm": 0.2033737787576765,
+      "learning_rate": 1.3269306154449238e-05,
+      "loss": 0.309,
+      "step": 2159
+    },
+    {
+      "epoch": 2.2857142857142856,
+      "grad_norm": 0.2304280611065484,
+      "learning_rate": 1.3249705997647982e-05,
+      "loss": 0.3606,
+      "step": 2160
+    },
+    {
+      "epoch": 2.286772486772487,
+      "grad_norm": 0.21105846127089964,
+      "learning_rate": 1.3230105840846726e-05,
+      "loss": 0.3343,
+      "step": 2161
+    },
+    {
+      "epoch": 2.287830687830688,
+      "grad_norm": 0.21136860640342398,
+      "learning_rate": 1.3210505684045474e-05,
+      "loss": 0.3407,
+      "step": 2162
+    },
+    {
+      "epoch": 2.2888888888888888,
+      "grad_norm": 0.21010473995437073,
+      "learning_rate": 1.3190905527244218e-05,
+      "loss": 0.3338,
+      "step": 2163
+    },
+    {
+      "epoch": 2.28994708994709,
+      "grad_norm": 0.21899937987852755,
+      "learning_rate": 1.3171305370442964e-05,
+      "loss": 0.3486,
+      "step": 2164
+    },
+    {
+      "epoch": 2.291005291005291,
+      "grad_norm": 0.1975690727274802,
+      "learning_rate": 1.315170521364171e-05,
+      "loss": 0.3115,
+      "step": 2165
+    },
+    {
+      "epoch": 2.292063492063492,
+      "grad_norm": 0.19198092778024467,
+      "learning_rate": 1.3132105056840455e-05,
+      "loss": 0.3275,
+      "step": 2166
+    },
+    {
+      "epoch": 2.293121693121693,
+      "grad_norm": 0.19304872180177207,
+      "learning_rate": 1.3112504900039201e-05,
+      "loss": 0.3058,
+      "step": 2167
+    },
+    {
+      "epoch": 2.2941798941798943,
+      "grad_norm": 0.22184189861680062,
+      "learning_rate": 1.3092904743237947e-05,
+      "loss": 0.3557,
+      "step": 2168
+    },
+    {
+      "epoch": 2.295238095238095,
+      "grad_norm": 0.2044727357753798,
+      "learning_rate": 1.3073304586436691e-05,
+      "loss": 0.3775,
+      "step": 2169
+    },
+    {
+      "epoch": 2.2962962962962963,
+      "grad_norm": 0.2109940769725802,
+      "learning_rate": 1.3053704429635439e-05,
+      "loss": 0.3606,
+      "step": 2170
+    },
+    {
+      "epoch": 2.2973544973544975,
+      "grad_norm": 0.20603745974637752,
+      "learning_rate": 1.3034104272834183e-05,
+      "loss": 0.3568,
+      "step": 2171
+    },
+    {
+      "epoch": 2.2984126984126982,
+      "grad_norm": 0.20723859027326352,
+      "learning_rate": 1.301450411603293e-05,
+      "loss": 0.3598,
+      "step": 2172
+    },
+    {
+      "epoch": 2.2994708994708994,
+      "grad_norm": 0.22014955361649102,
+      "learning_rate": 1.2994903959231675e-05,
+      "loss": 0.3645,
+      "step": 2173
+    },
+    {
+      "epoch": 2.3005291005291006,
+      "grad_norm": 0.21240912935645942,
+      "learning_rate": 1.2975303802430419e-05,
+      "loss": 0.3227,
+      "step": 2174
+    },
+    {
+      "epoch": 2.3015873015873014,
+      "grad_norm": 0.1984055417276385,
+      "learning_rate": 1.2955703645629167e-05,
+      "loss": 0.3403,
+      "step": 2175
+    },
+    {
+      "epoch": 2.3026455026455026,
+      "grad_norm": 0.20615107635995517,
+      "learning_rate": 1.293610348882791e-05,
+      "loss": 0.357,
+      "step": 2176
+    },
+    {
+      "epoch": 2.303703703703704,
+      "grad_norm": 0.2883475072855802,
+      "learning_rate": 1.2916503332026658e-05,
+      "loss": 0.328,
+      "step": 2177
+    },
+    {
+      "epoch": 2.3047619047619046,
+      "grad_norm": 0.21617403311285718,
+      "learning_rate": 1.2896903175225403e-05,
+      "loss": 0.3207,
+      "step": 2178
+    },
+    {
+      "epoch": 2.3058201058201058,
+      "grad_norm": 0.22261165487391843,
+      "learning_rate": 1.2877303018424147e-05,
+      "loss": 0.3905,
+      "step": 2179
+    },
+    {
+      "epoch": 2.306878306878307,
+      "grad_norm": 0.19746052299888087,
+      "learning_rate": 1.2857702861622894e-05,
+      "loss": 0.3344,
+      "step": 2180
+    },
+    {
+      "epoch": 2.3079365079365077,
+      "grad_norm": 0.20619793904799188,
+      "learning_rate": 1.2838102704821638e-05,
+      "loss": 0.3187,
+      "step": 2181
+    },
+    {
+      "epoch": 2.308994708994709,
+      "grad_norm": 0.21727096923527425,
+      "learning_rate": 1.2818502548020384e-05,
+      "loss": 0.393,
+      "step": 2182
+    },
+    {
+      "epoch": 2.31005291005291,
+      "grad_norm": 0.22496945697900078,
+      "learning_rate": 1.279890239121913e-05,
+      "loss": 0.4016,
+      "step": 2183
+    },
+    {
+      "epoch": 2.311111111111111,
+      "grad_norm": 0.20878049645676627,
+      "learning_rate": 1.2779302234417876e-05,
+      "loss": 0.3308,
+      "step": 2184
+    },
+    {
+      "epoch": 2.312169312169312,
+      "grad_norm": 0.19876898454737266,
+      "learning_rate": 1.2759702077616622e-05,
+      "loss": 0.3472,
+      "step": 2185
+    },
+    {
+      "epoch": 2.3132275132275133,
+      "grad_norm": 0.22470198675510053,
+      "learning_rate": 1.2740101920815368e-05,
+      "loss": 0.3723,
+      "step": 2186
+    },
+    {
+      "epoch": 2.314285714285714,
+      "grad_norm": 0.2149186208399833,
+      "learning_rate": 1.2720501764014112e-05,
+      "loss": 0.3609,
+      "step": 2187
+    },
+    {
+      "epoch": 2.3153439153439153,
+      "grad_norm": 0.21505615763181798,
+      "learning_rate": 1.270090160721286e-05,
+      "loss": 0.3775,
+      "step": 2188
+    },
+    {
+      "epoch": 2.3164021164021165,
+      "grad_norm": 0.21734973914888298,
+      "learning_rate": 1.2681301450411604e-05,
+      "loss": 0.3578,
+      "step": 2189
+    },
+    {
+      "epoch": 2.317460317460317,
+      "grad_norm": 0.19738584826580072,
+      "learning_rate": 1.2661701293610351e-05,
+      "loss": 0.3114,
+      "step": 2190
+    },
+    {
+      "epoch": 2.3185185185185184,
+      "grad_norm": 0.2059651241318454,
+      "learning_rate": 1.2642101136809096e-05,
+      "loss": 0.3503,
+      "step": 2191
+    },
+    {
+      "epoch": 2.3195767195767196,
+      "grad_norm": 0.21191185476329366,
+      "learning_rate": 1.262250098000784e-05,
+      "loss": 0.3277,
+      "step": 2192
+    },
+    {
+      "epoch": 2.320634920634921,
+      "grad_norm": 0.21822223673790192,
+      "learning_rate": 1.2602900823206587e-05,
+      "loss": 0.3402,
+      "step": 2193
+    },
+    {
+      "epoch": 2.3216931216931216,
+      "grad_norm": 0.22119523492818854,
+      "learning_rate": 1.2583300666405331e-05,
+      "loss": 0.3931,
+      "step": 2194
+    },
+    {
+      "epoch": 2.322751322751323,
+      "grad_norm": 0.18136041095386,
+      "learning_rate": 1.2563700509604076e-05,
+      "loss": 0.3065,
+      "step": 2195
+    },
+    {
+      "epoch": 2.323809523809524,
+      "grad_norm": 0.28983751655981344,
+      "learning_rate": 1.2544100352802823e-05,
+      "loss": 0.366,
+      "step": 2196
+    },
+    {
+      "epoch": 2.3248677248677247,
+      "grad_norm": 0.2191979489235065,
+      "learning_rate": 1.2524500196001567e-05,
+      "loss": 0.3444,
+      "step": 2197
+    },
+    {
+      "epoch": 2.325925925925926,
+      "grad_norm": 0.200384296867351,
+      "learning_rate": 1.2504900039200315e-05,
+      "loss": 0.3277,
+      "step": 2198
+    },
+    {
+      "epoch": 2.326984126984127,
+      "grad_norm": 0.20207450606748373,
+      "learning_rate": 1.248529988239906e-05,
+      "loss": 0.3619,
+      "step": 2199
+    },
+    {
+      "epoch": 2.328042328042328,
+      "grad_norm": 0.20817037763377247,
+      "learning_rate": 1.2465699725597805e-05,
+      "loss": 0.349,
+      "step": 2200
+    },
+    {
+      "epoch": 2.329100529100529,
+      "grad_norm": 0.20772850458509592,
+      "learning_rate": 1.2446099568796551e-05,
+      "loss": 0.3289,
+      "step": 2201
+    },
+    {
+      "epoch": 2.3301587301587303,
+      "grad_norm": 1.2036917950362793,
+      "learning_rate": 1.2426499411995297e-05,
+      "loss": 0.4814,
+      "step": 2202
+    },
+    {
+      "epoch": 2.331216931216931,
+      "grad_norm": 0.21881788351835138,
+      "learning_rate": 1.2406899255194043e-05,
+      "loss": 0.3625,
+      "step": 2203
+    },
+    {
+      "epoch": 2.3322751322751323,
+      "grad_norm": 0.20154036096455355,
+      "learning_rate": 1.2387299098392789e-05,
+      "loss": 0.3686,
+      "step": 2204
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 0.20121002730380697,
+      "learning_rate": 1.2367698941591534e-05,
+      "loss": 0.3603,
+      "step": 2205
+    },
+    {
+      "epoch": 2.3343915343915342,
+      "grad_norm": 0.2197435442897782,
+      "learning_rate": 1.234809878479028e-05,
+      "loss": 0.349,
+      "step": 2206
+    },
+    {
+      "epoch": 2.3354497354497354,
+      "grad_norm": 0.21748291890475052,
+      "learning_rate": 1.2328498627989024e-05,
+      "loss": 0.4108,
+      "step": 2207
+    },
+    {
+      "epoch": 2.3365079365079366,
+      "grad_norm": 0.18641463759702262,
+      "learning_rate": 1.230889847118777e-05,
+      "loss": 0.31,
+      "step": 2208
+    },
+    {
+      "epoch": 2.3375661375661374,
+      "grad_norm": 0.19503644892500818,
+      "learning_rate": 1.2289298314386516e-05,
+      "loss": 0.3595,
+      "step": 2209
+    },
+    {
+      "epoch": 2.3386243386243386,
+      "grad_norm": 0.20574304094175103,
+      "learning_rate": 1.2269698157585262e-05,
+      "loss": 0.3666,
+      "step": 2210
+    },
+    {
+      "epoch": 2.33968253968254,
+      "grad_norm": 0.20069061348310832,
+      "learning_rate": 1.2250098000784006e-05,
+      "loss": 0.3123,
+      "step": 2211
+    },
+    {
+      "epoch": 2.3407407407407406,
+      "grad_norm": 0.20015298378799415,
+      "learning_rate": 1.2230497843982752e-05,
+      "loss": 0.3545,
+      "step": 2212
+    },
+    {
+      "epoch": 2.3417989417989418,
+      "grad_norm": 0.19565316573476668,
+      "learning_rate": 1.2210897687181498e-05,
+      "loss": 0.3481,
+      "step": 2213
+    },
+    {
+      "epoch": 2.342857142857143,
+      "grad_norm": 0.213652830993624,
+      "learning_rate": 1.2191297530380244e-05,
+      "loss": 0.3492,
+      "step": 2214
+    },
+    {
+      "epoch": 2.3439153439153437,
+      "grad_norm": 0.211527905920233,
+      "learning_rate": 1.2171697373578988e-05,
+      "loss": 0.372,
+      "step": 2215
+    },
+    {
+      "epoch": 2.344973544973545,
+      "grad_norm": 0.20693192379565375,
+      "learning_rate": 1.2152097216777734e-05,
+      "loss": 0.3772,
+      "step": 2216
+    },
+    {
+      "epoch": 2.346031746031746,
+      "grad_norm": 0.20506549919992692,
+      "learning_rate": 1.213249705997648e-05,
+      "loss": 0.3506,
+      "step": 2217
+    },
+    {
+      "epoch": 2.3470899470899473,
+      "grad_norm": 0.29640440277261043,
+      "learning_rate": 1.2112896903175226e-05,
+      "loss": 0.4008,
+      "step": 2218
+    },
+    {
+      "epoch": 2.348148148148148,
+      "grad_norm": 0.21563211935876653,
+      "learning_rate": 1.2093296746373972e-05,
+      "loss": 0.366,
+      "step": 2219
+    },
+    {
+      "epoch": 2.3492063492063493,
+      "grad_norm": 0.20236613701883793,
+      "learning_rate": 1.2073696589572717e-05,
+      "loss": 0.3492,
+      "step": 2220
+    },
+    {
+      "epoch": 2.3502645502645505,
+      "grad_norm": 0.1966092544544384,
+      "learning_rate": 1.2054096432771463e-05,
+      "loss": 0.3317,
+      "step": 2221
+    },
+    {
+      "epoch": 2.3513227513227513,
+      "grad_norm": 0.21784607574068482,
+      "learning_rate": 1.203449627597021e-05,
+      "loss": 0.3847,
+      "step": 2222
+    },
+    {
+      "epoch": 2.3523809523809525,
+      "grad_norm": 0.21084416216589955,
+      "learning_rate": 1.2014896119168955e-05,
+      "loss": 0.3637,
+      "step": 2223
+    },
+    {
+      "epoch": 2.3534391534391537,
+      "grad_norm": 0.1991773340459715,
+      "learning_rate": 1.19952959623677e-05,
+      "loss": 0.3303,
+      "step": 2224
+    },
+    {
+      "epoch": 2.3544973544973544,
+      "grad_norm": 0.1918709626897971,
+      "learning_rate": 1.1975695805566445e-05,
+      "loss": 0.3025,
+      "step": 2225
+    },
+    {
+      "epoch": 2.3555555555555556,
+      "grad_norm": 0.19838360255526297,
+      "learning_rate": 1.1956095648765191e-05,
+      "loss": 0.3105,
+      "step": 2226
+    },
+    {
+      "epoch": 2.356613756613757,
+      "grad_norm": 0.20910709785260412,
+      "learning_rate": 1.1936495491963937e-05,
+      "loss": 0.3434,
+      "step": 2227
+    },
+    {
+      "epoch": 2.3576719576719576,
+      "grad_norm": 0.20813311118331637,
+      "learning_rate": 1.1916895335162683e-05,
+      "loss": 0.3388,
+      "step": 2228
+    },
+    {
+      "epoch": 2.358730158730159,
+      "grad_norm": 0.21726113463115548,
+      "learning_rate": 1.1897295178361427e-05,
+      "loss": 0.3551,
+      "step": 2229
+    },
+    {
+      "epoch": 2.35978835978836,
+      "grad_norm": 0.21011962247860588,
+      "learning_rate": 1.1877695021560173e-05,
+      "loss": 0.3509,
+      "step": 2230
+    },
+    {
+      "epoch": 2.3608465608465607,
+      "grad_norm": 0.1953218231695064,
+      "learning_rate": 1.1858094864758919e-05,
+      "loss": 0.319,
+      "step": 2231
+    },
+    {
+      "epoch": 2.361904761904762,
+      "grad_norm": 0.20596213011515244,
+      "learning_rate": 1.1838494707957665e-05,
+      "loss": 0.3517,
+      "step": 2232
+    },
+    {
+      "epoch": 2.362962962962963,
+      "grad_norm": 0.2126925193532187,
+      "learning_rate": 1.1818894551156409e-05,
+      "loss": 0.3767,
+      "step": 2233
+    },
+    {
+      "epoch": 2.364021164021164,
+      "grad_norm": 0.19743855767120733,
+      "learning_rate": 1.1799294394355155e-05,
+      "loss": 0.3302,
+      "step": 2234
+    },
+    {
+      "epoch": 2.365079365079365,
+      "grad_norm": 0.1867269633062068,
+      "learning_rate": 1.17796942375539e-05,
+      "loss": 0.3275,
+      "step": 2235
+    },
+    {
+      "epoch": 2.3661375661375663,
+      "grad_norm": 0.19798324638566683,
+      "learning_rate": 1.1760094080752646e-05,
+      "loss": 0.3358,
+      "step": 2236
+    },
+    {
+      "epoch": 2.367195767195767,
+      "grad_norm": 0.21204816572937232,
+      "learning_rate": 1.1740493923951392e-05,
+      "loss": 0.3557,
+      "step": 2237
+    },
+    {
+      "epoch": 2.3682539682539683,
+      "grad_norm": 0.20561285406488616,
+      "learning_rate": 1.1720893767150138e-05,
+      "loss": 0.3859,
+      "step": 2238
+    },
+    {
+      "epoch": 2.3693121693121695,
+      "grad_norm": 0.1844307143572627,
+      "learning_rate": 1.1701293610348884e-05,
+      "loss": 0.3093,
+      "step": 2239
+    },
+    {
+      "epoch": 2.3703703703703702,
+      "grad_norm": 0.18285268707845753,
+      "learning_rate": 1.168169345354763e-05,
+      "loss": 0.2879,
+      "step": 2240
+    },
+    {
+      "epoch": 2.3714285714285714,
+      "grad_norm": 0.18891596600216767,
+      "learning_rate": 1.1662093296746376e-05,
+      "loss": 0.3164,
+      "step": 2241
+    },
+    {
+      "epoch": 2.3724867724867726,
+      "grad_norm": 0.21443463118106196,
+      "learning_rate": 1.164249313994512e-05,
+      "loss": 0.3724,
+      "step": 2242
+    },
+    {
+      "epoch": 2.3735449735449734,
+      "grad_norm": 0.20092191567682116,
+      "learning_rate": 1.1622892983143866e-05,
+      "loss": 0.336,
+      "step": 2243
+    },
+    {
+      "epoch": 2.3746031746031746,
+      "grad_norm": 0.21661040764716843,
+      "learning_rate": 1.1603292826342612e-05,
+      "loss": 0.3457,
+      "step": 2244
+    },
+    {
+      "epoch": 2.375661375661376,
+      "grad_norm": 0.190041744765612,
+      "learning_rate": 1.1583692669541358e-05,
+      "loss": 0.3251,
+      "step": 2245
+    },
+    {
+      "epoch": 2.3767195767195766,
+      "grad_norm": 0.1886509935215561,
+      "learning_rate": 1.1564092512740103e-05,
+      "loss": 0.3268,
+      "step": 2246
+    },
+    {
+      "epoch": 2.3777777777777778,
+      "grad_norm": 0.19137796325706583,
+      "learning_rate": 1.1544492355938848e-05,
+      "loss": 0.3137,
+      "step": 2247
+    },
+    {
+      "epoch": 2.378835978835979,
+      "grad_norm": 0.19633048102603665,
+      "learning_rate": 1.1524892199137593e-05,
+      "loss": 0.3428,
+      "step": 2248
+    },
+    {
+      "epoch": 2.3798941798941797,
+      "grad_norm": 0.20298501174663242,
+      "learning_rate": 1.150529204233634e-05,
+      "loss": 0.3336,
+      "step": 2249
+    },
+    {
+      "epoch": 2.380952380952381,
+      "grad_norm": 0.20915086314748638,
+      "learning_rate": 1.1485691885535085e-05,
+      "loss": 0.3538,
+      "step": 2250
+    },
+    {
+      "epoch": 2.382010582010582,
+      "grad_norm": 0.1959406139546007,
+      "learning_rate": 1.146609172873383e-05,
+      "loss": 0.3204,
+      "step": 2251
+    },
+    {
+      "epoch": 2.383068783068783,
+      "grad_norm": 0.1980356952193389,
+      "learning_rate": 1.1446491571932575e-05,
+      "loss": 0.3369,
+      "step": 2252
+    },
+    {
+      "epoch": 2.384126984126984,
+      "grad_norm": 0.21603527021327792,
+      "learning_rate": 1.1426891415131321e-05,
+      "loss": 0.4106,
+      "step": 2253
+    },
+    {
+      "epoch": 2.3851851851851853,
+      "grad_norm": 0.19272692236278238,
+      "learning_rate": 1.1407291258330067e-05,
+      "loss": 0.3393,
+      "step": 2254
+    },
+    {
+      "epoch": 2.386243386243386,
+      "grad_norm": 0.1956248707684031,
+      "learning_rate": 1.1387691101528813e-05,
+      "loss": 0.3411,
+      "step": 2255
+    },
+    {
+      "epoch": 2.3873015873015873,
+      "grad_norm": 0.19014685494467712,
+      "learning_rate": 1.1368090944727559e-05,
+      "loss": 0.3185,
+      "step": 2256
+    },
+    {
+      "epoch": 2.3883597883597885,
+      "grad_norm": 0.20447615014158566,
+      "learning_rate": 1.1348490787926305e-05,
+      "loss": 0.3513,
+      "step": 2257
+    },
+    {
+      "epoch": 2.389417989417989,
+      "grad_norm": 0.21079536892016623,
+      "learning_rate": 1.132889063112505e-05,
+      "loss": 0.3377,
+      "step": 2258
+    },
+    {
+      "epoch": 2.3904761904761904,
+      "grad_norm": 0.20403855917854744,
+      "learning_rate": 1.1309290474323796e-05,
+      "loss": 0.3728,
+      "step": 2259
+    },
+    {
+      "epoch": 2.3915343915343916,
+      "grad_norm": 0.19160044266304876,
+      "learning_rate": 1.128969031752254e-05,
+      "loss": 0.3113,
+      "step": 2260
+    },
+    {
+      "epoch": 2.3925925925925924,
+      "grad_norm": 0.2081785958094779,
+      "learning_rate": 1.1270090160721286e-05,
+      "loss": 0.3618,
+      "step": 2261
+    },
+    {
+      "epoch": 2.3936507936507936,
+      "grad_norm": 0.2080027481507071,
+      "learning_rate": 1.1250490003920032e-05,
+      "loss": 0.3484,
+      "step": 2262
+    },
+    {
+      "epoch": 2.394708994708995,
+      "grad_norm": 0.2048194108111222,
+      "learning_rate": 1.1230889847118778e-05,
+      "loss": 0.3429,
+      "step": 2263
+    },
+    {
+      "epoch": 2.3957671957671955,
+      "grad_norm": 0.2103500941268802,
+      "learning_rate": 1.1211289690317522e-05,
+      "loss": 0.3665,
+      "step": 2264
+    },
+    {
+      "epoch": 2.3968253968253967,
+      "grad_norm": 0.21252569969121238,
+      "learning_rate": 1.1191689533516268e-05,
+      "loss": 0.3346,
+      "step": 2265
+    },
+    {
+      "epoch": 2.397883597883598,
+      "grad_norm": 0.19587149638160067,
+      "learning_rate": 1.1172089376715014e-05,
+      "loss": 0.3352,
+      "step": 2266
+    },
+    {
+      "epoch": 2.3989417989417987,
+      "grad_norm": 0.21361691084135886,
+      "learning_rate": 1.115248921991376e-05,
+      "loss": 0.3695,
+      "step": 2267
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.1860233174405519,
+      "learning_rate": 1.1132889063112506e-05,
+      "loss": 0.3049,
+      "step": 2268
+    },
+    {
+      "epoch": 2.401058201058201,
+      "grad_norm": 0.21531066517258854,
+      "learning_rate": 1.111328890631125e-05,
+      "loss": 0.3919,
+      "step": 2269
+    },
+    {
+      "epoch": 2.402116402116402,
+      "grad_norm": 0.20973338346864226,
+      "learning_rate": 1.1093688749509996e-05,
+      "loss": 0.309,
+      "step": 2270
+    },
+    {
+      "epoch": 2.403174603174603,
+      "grad_norm": 0.19942109842684344,
+      "learning_rate": 1.1074088592708742e-05,
+      "loss": 0.3438,
+      "step": 2271
+    },
+    {
+      "epoch": 2.4042328042328043,
+      "grad_norm": 0.20332330726245493,
+      "learning_rate": 1.1054488435907488e-05,
+      "loss": 0.3716,
+      "step": 2272
+    },
+    {
+      "epoch": 2.4052910052910055,
+      "grad_norm": 0.2109034509123887,
+      "learning_rate": 1.1034888279106234e-05,
+      "loss": 0.3815,
+      "step": 2273
+    },
+    {
+      "epoch": 2.4063492063492062,
+      "grad_norm": 0.20967094697117655,
+      "learning_rate": 1.101528812230498e-05,
+      "loss": 0.3791,
+      "step": 2274
+    },
+    {
+      "epoch": 2.4074074074074074,
+      "grad_norm": 0.20137962520759609,
+      "learning_rate": 1.0995687965503725e-05,
+      "loss": 0.352,
+      "step": 2275
+    },
+    {
+      "epoch": 2.4084656084656086,
+      "grad_norm": 0.1959407595415704,
+      "learning_rate": 1.0976087808702471e-05,
+      "loss": 0.3128,
+      "step": 2276
+    },
+    {
+      "epoch": 2.4095238095238094,
+      "grad_norm": 0.2025181599605024,
+      "learning_rate": 1.0956487651901217e-05,
+      "loss": 0.3643,
+      "step": 2277
+    },
+    {
+      "epoch": 2.4105820105820106,
+      "grad_norm": 0.230544012868408,
+      "learning_rate": 1.0936887495099961e-05,
+      "loss": 0.3726,
+      "step": 2278
+    },
+    {
+      "epoch": 2.411640211640212,
+      "grad_norm": 0.1990066422140862,
+      "learning_rate": 1.0917287338298707e-05,
+      "loss": 0.3638,
+      "step": 2279
+    },
+    {
+      "epoch": 2.4126984126984126,
+      "grad_norm": 0.18998917405534244,
+      "learning_rate": 1.0897687181497453e-05,
+      "loss": 0.3274,
+      "step": 2280
+    },
+    {
+      "epoch": 2.4137566137566138,
+      "grad_norm": 0.20400621001125108,
+      "learning_rate": 1.0878087024696199e-05,
+      "loss": 0.3977,
+      "step": 2281
+    },
+    {
+      "epoch": 2.414814814814815,
+      "grad_norm": 0.19362884565369348,
+      "learning_rate": 1.0858486867894943e-05,
+      "loss": 0.2982,
+      "step": 2282
+    },
+    {
+      "epoch": 2.4158730158730157,
+      "grad_norm": 0.19304320410771325,
+      "learning_rate": 1.0838886711093689e-05,
+      "loss": 0.3363,
+      "step": 2283
+    },
+    {
+      "epoch": 2.416931216931217,
+      "grad_norm": 0.18681427914983176,
+      "learning_rate": 1.0819286554292435e-05,
+      "loss": 0.2841,
+      "step": 2284
+    },
+    {
+      "epoch": 2.417989417989418,
+      "grad_norm": 0.20184125426045912,
+      "learning_rate": 1.079968639749118e-05,
+      "loss": 0.3428,
+      "step": 2285
+    },
+    {
+      "epoch": 2.419047619047619,
+      "grad_norm": 0.2037706142983145,
+      "learning_rate": 1.0780086240689927e-05,
+      "loss": 0.3649,
+      "step": 2286
+    },
+    {
+      "epoch": 2.42010582010582,
+      "grad_norm": 0.19817399493675728,
+      "learning_rate": 1.076048608388867e-05,
+      "loss": 0.3506,
+      "step": 2287
+    },
+    {
+      "epoch": 2.4211640211640213,
+      "grad_norm": 0.20169589584258404,
+      "learning_rate": 1.0740885927087417e-05,
+      "loss": 0.357,
+      "step": 2288
+    },
+    {
+      "epoch": 2.422222222222222,
+      "grad_norm": 0.20942728369583866,
+      "learning_rate": 1.0721285770286162e-05,
+      "loss": 0.3732,
+      "step": 2289
+    },
+    {
+      "epoch": 2.4232804232804233,
+      "grad_norm": 0.19265517759072814,
+      "learning_rate": 1.0701685613484908e-05,
+      "loss": 0.3176,
+      "step": 2290
+    },
+    {
+      "epoch": 2.4243386243386245,
+      "grad_norm": 0.1982086084106544,
+      "learning_rate": 1.0682085456683654e-05,
+      "loss": 0.3467,
+      "step": 2291
+    },
+    {
+      "epoch": 2.425396825396825,
+      "grad_norm": 0.19447098693465864,
+      "learning_rate": 1.06624852998824e-05,
+      "loss": 0.3299,
+      "step": 2292
+    },
+    {
+      "epoch": 2.4264550264550264,
+      "grad_norm": 0.20888212480507384,
+      "learning_rate": 1.0642885143081146e-05,
+      "loss": 0.3459,
+      "step": 2293
+    },
+    {
+      "epoch": 2.4275132275132276,
+      "grad_norm": 0.208144303971245,
+      "learning_rate": 1.0623284986279892e-05,
+      "loss": 0.367,
+      "step": 2294
+    },
+    {
+      "epoch": 2.4285714285714284,
+      "grad_norm": 0.20876100051287758,
+      "learning_rate": 1.0603684829478638e-05,
+      "loss": 0.3745,
+      "step": 2295
+    },
+    {
+      "epoch": 2.4296296296296296,
+      "grad_norm": 0.20791252154146914,
+      "learning_rate": 1.0584084672677382e-05,
+      "loss": 0.3491,
+      "step": 2296
+    },
+    {
+      "epoch": 2.430687830687831,
+      "grad_norm": 0.19777559356235025,
+      "learning_rate": 1.0564484515876128e-05,
+      "loss": 0.3374,
+      "step": 2297
+    },
+    {
+      "epoch": 2.431746031746032,
+      "grad_norm": 0.19110723755738349,
+      "learning_rate": 1.0544884359074874e-05,
+      "loss": 0.3271,
+      "step": 2298
+    },
+    {
+      "epoch": 2.4328042328042327,
+      "grad_norm": 0.2008378278505105,
+      "learning_rate": 1.052528420227362e-05,
+      "loss": 0.3387,
+      "step": 2299
+    },
+    {
+      "epoch": 2.433862433862434,
+      "grad_norm": 0.21027780382458788,
+      "learning_rate": 1.0505684045472364e-05,
+      "loss": 0.3179,
+      "step": 2300
+    },
+    {
+      "epoch": 2.434920634920635,
+      "grad_norm": 0.2026493429086862,
+      "learning_rate": 1.048608388867111e-05,
+      "loss": 0.3345,
+      "step": 2301
+    },
+    {
+      "epoch": 2.435978835978836,
+      "grad_norm": 0.20274404356172807,
+      "learning_rate": 1.0466483731869855e-05,
+      "loss": 0.3801,
+      "step": 2302
+    },
+    {
+      "epoch": 2.437037037037037,
+      "grad_norm": 0.19273549407390878,
+      "learning_rate": 1.0446883575068601e-05,
+      "loss": 0.3467,
+      "step": 2303
+    },
+    {
+      "epoch": 2.4380952380952383,
+      "grad_norm": 0.1818577084303832,
+      "learning_rate": 1.0427283418267345e-05,
+      "loss": 0.3202,
+      "step": 2304
+    },
+    {
+      "epoch": 2.439153439153439,
+      "grad_norm": 0.19560068167078504,
+      "learning_rate": 1.0407683261466091e-05,
+      "loss": 0.3273,
+      "step": 2305
+    },
+    {
+      "epoch": 2.4402116402116403,
+      "grad_norm": 0.20370517607190086,
+      "learning_rate": 1.0388083104664837e-05,
+      "loss": 0.3403,
+      "step": 2306
+    },
+    {
+      "epoch": 2.4412698412698415,
+      "grad_norm": 0.20114205886933845,
+      "learning_rate": 1.0368482947863583e-05,
+      "loss": 0.3522,
+      "step": 2307
+    },
+    {
+      "epoch": 2.4423280423280422,
+      "grad_norm": 0.20882464854711777,
+      "learning_rate": 1.0348882791062329e-05,
+      "loss": 0.3666,
+      "step": 2308
+    },
+    {
+      "epoch": 2.4433862433862434,
+      "grad_norm": 0.20438756339590045,
+      "learning_rate": 1.0329282634261075e-05,
+      "loss": 0.3435,
+      "step": 2309
+    },
+    {
+      "epoch": 2.4444444444444446,
+      "grad_norm": 0.22064605217007735,
+      "learning_rate": 1.030968247745982e-05,
+      "loss": 0.4142,
+      "step": 2310
+    },
+    {
+      "epoch": 2.4455026455026454,
+      "grad_norm": 0.1817014882330108,
+      "learning_rate": 1.0290082320658567e-05,
+      "loss": 0.2751,
+      "step": 2311
+    },
+    {
+      "epoch": 2.4465608465608466,
+      "grad_norm": 0.19417363691429626,
+      "learning_rate": 1.0270482163857312e-05,
+      "loss": 0.347,
+      "step": 2312
+    },
+    {
+      "epoch": 2.447619047619048,
+      "grad_norm": 0.19984236991627735,
+      "learning_rate": 1.0250882007056057e-05,
+      "loss": 0.358,
+      "step": 2313
+    },
+    {
+      "epoch": 2.4486772486772486,
+      "grad_norm": 0.20561224905848144,
+      "learning_rate": 1.0231281850254803e-05,
+      "loss": 0.3748,
+      "step": 2314
+    },
+    {
+      "epoch": 2.4497354497354498,
+      "grad_norm": 0.2030555882275226,
+      "learning_rate": 1.0211681693453548e-05,
+      "loss": 0.3333,
+      "step": 2315
+    },
+    {
+      "epoch": 2.450793650793651,
+      "grad_norm": 0.18955391416822132,
+      "learning_rate": 1.0192081536652294e-05,
+      "loss": 0.3067,
+      "step": 2316
+    },
+    {
+      "epoch": 2.4518518518518517,
+      "grad_norm": 0.214593330571037,
+      "learning_rate": 1.017248137985104e-05,
+      "loss": 0.4056,
+      "step": 2317
+    },
+    {
+      "epoch": 2.452910052910053,
+      "grad_norm": 0.20084964435456928,
+      "learning_rate": 1.0152881223049784e-05,
+      "loss": 0.3302,
+      "step": 2318
+    },
+    {
+      "epoch": 2.453968253968254,
+      "grad_norm": 0.19398682980431026,
+      "learning_rate": 1.013328106624853e-05,
+      "loss": 0.3213,
+      "step": 2319
+    },
+    {
+      "epoch": 2.455026455026455,
+      "grad_norm": 0.22095190181637556,
+      "learning_rate": 1.0113680909447276e-05,
+      "loss": 0.4104,
+      "step": 2320
+    },
+    {
+      "epoch": 2.456084656084656,
+      "grad_norm": 0.21142265935421076,
+      "learning_rate": 1.0094080752646022e-05,
+      "loss": 0.3691,
+      "step": 2321
+    },
+    {
+      "epoch": 2.4571428571428573,
+      "grad_norm": 0.20774600162711074,
+      "learning_rate": 1.0074480595844766e-05,
+      "loss": 0.3576,
+      "step": 2322
+    },
+    {
+      "epoch": 2.458201058201058,
+      "grad_norm": 0.19470750734709114,
+      "learning_rate": 1.0054880439043512e-05,
+      "loss": 0.3391,
+      "step": 2323
+    },
+    {
+      "epoch": 2.4592592592592593,
+      "grad_norm": 0.20179308051014488,
+      "learning_rate": 1.0035280282242258e-05,
+      "loss": 0.3753,
+      "step": 2324
+    },
+    {
+      "epoch": 2.4603174603174605,
+      "grad_norm": 0.1964229149495673,
+      "learning_rate": 1.0015680125441004e-05,
+      "loss": 0.323,
+      "step": 2325
+    },
+    {
+      "epoch": 2.461375661375661,
+      "grad_norm": 0.20063757362174908,
+      "learning_rate": 9.99607996863975e-06,
+      "loss": 0.3787,
+      "step": 2326
+    },
+    {
+      "epoch": 2.4624338624338624,
+      "grad_norm": 0.1930747913002795,
+      "learning_rate": 9.976479811838496e-06,
+      "loss": 0.319,
+      "step": 2327
+    },
+    {
+      "epoch": 2.4634920634920636,
+      "grad_norm": 0.21711063943337874,
+      "learning_rate": 9.956879655037241e-06,
+      "loss": 0.3542,
+      "step": 2328
+    },
+    {
+      "epoch": 2.4645502645502644,
+      "grad_norm": 0.1903790970553673,
+      "learning_rate": 9.937279498235987e-06,
+      "loss": 0.3267,
+      "step": 2329
+    },
+    {
+      "epoch": 2.4656084656084656,
+      "grad_norm": 0.3974441833827003,
+      "learning_rate": 9.917679341434733e-06,
+      "loss": 0.3863,
+      "step": 2330
+    },
+    {
+      "epoch": 2.466666666666667,
+      "grad_norm": 0.21646430718812706,
+      "learning_rate": 9.898079184633477e-06,
+      "loss": 0.3876,
+      "step": 2331
+    },
+    {
+      "epoch": 2.4677248677248675,
+      "grad_norm": 0.22008541249629765,
+      "learning_rate": 9.878479027832223e-06,
+      "loss": 0.328,
+      "step": 2332
+    },
+    {
+      "epoch": 2.4687830687830687,
+      "grad_norm": 0.19270815027961807,
+      "learning_rate": 9.858878871030969e-06,
+      "loss": 0.3422,
+      "step": 2333
+    },
+    {
+      "epoch": 2.46984126984127,
+      "grad_norm": 0.2002872486600827,
+      "learning_rate": 9.839278714229715e-06,
+      "loss": 0.3509,
+      "step": 2334
+    },
+    {
+      "epoch": 2.4708994708994707,
+      "grad_norm": 0.216151009251505,
+      "learning_rate": 9.81967855742846e-06,
+      "loss": 0.3872,
+      "step": 2335
+    },
+    {
+      "epoch": 2.471957671957672,
+      "grad_norm": 0.21555323390899783,
+      "learning_rate": 9.800078400627205e-06,
+      "loss": 0.3542,
+      "step": 2336
+    },
+    {
+      "epoch": 2.473015873015873,
+      "grad_norm": 0.18992518981588918,
+      "learning_rate": 9.780478243825951e-06,
+      "loss": 0.3061,
+      "step": 2337
+    },
+    {
+      "epoch": 2.474074074074074,
+      "grad_norm": 0.20207716846236165,
+      "learning_rate": 9.760878087024697e-06,
+      "loss": 0.3492,
+      "step": 2338
+    },
+    {
+      "epoch": 2.475132275132275,
+      "grad_norm": 0.3238919978391247,
+      "learning_rate": 9.741277930223443e-06,
+      "loss": 0.3811,
+      "step": 2339
+    },
+    {
+      "epoch": 2.4761904761904763,
+      "grad_norm": 0.20016631554309963,
+      "learning_rate": 9.721677773422187e-06,
+      "loss": 0.3375,
+      "step": 2340
+    },
+    {
+      "epoch": 2.477248677248677,
+      "grad_norm": 0.2099603304873576,
+      "learning_rate": 9.702077616620933e-06,
+      "loss": 0.3601,
+      "step": 2341
+    },
+    {
+      "epoch": 2.4783068783068782,
+      "grad_norm": 0.2163092177109193,
+      "learning_rate": 9.682477459819679e-06,
+      "loss": 0.3929,
+      "step": 2342
+    },
+    {
+      "epoch": 2.4793650793650794,
+      "grad_norm": 0.195942799842103,
+      "learning_rate": 9.662877303018424e-06,
+      "loss": 0.3259,
+      "step": 2343
+    },
+    {
+      "epoch": 2.48042328042328,
+      "grad_norm": 0.20028542223736842,
+      "learning_rate": 9.64327714621717e-06,
+      "loss": 0.3195,
+      "step": 2344
+    },
+    {
+      "epoch": 2.4814814814814814,
+      "grad_norm": 0.2034772057737831,
+      "learning_rate": 9.623676989415916e-06,
+      "loss": 0.3481,
+      "step": 2345
+    },
+    {
+      "epoch": 2.4825396825396826,
+      "grad_norm": 0.19210503275161703,
+      "learning_rate": 9.604076832614662e-06,
+      "loss": 0.3252,
+      "step": 2346
+    },
+    {
+      "epoch": 2.4835978835978834,
+      "grad_norm": 0.2096974621839747,
+      "learning_rate": 9.584476675813408e-06,
+      "loss": 0.3513,
+      "step": 2347
+    },
+    {
+      "epoch": 2.4846560846560846,
+      "grad_norm": 0.20429072459845538,
+      "learning_rate": 9.564876519012154e-06,
+      "loss": 0.3549,
+      "step": 2348
+    },
+    {
+      "epoch": 2.4857142857142858,
+      "grad_norm": 0.18861327099569755,
+      "learning_rate": 9.545276362210898e-06,
+      "loss": 0.3381,
+      "step": 2349
+    },
+    {
+      "epoch": 2.4867724867724865,
+      "grad_norm": 0.19260260318882602,
+      "learning_rate": 9.525676205409644e-06,
+      "loss": 0.3248,
+      "step": 2350
+    },
+    {
+      "epoch": 2.4878306878306877,
+      "grad_norm": 0.2121251443373568,
+      "learning_rate": 9.50607604860839e-06,
+      "loss": 0.3893,
+      "step": 2351
+    },
+    {
+      "epoch": 2.488888888888889,
+      "grad_norm": 0.2153555638932981,
+      "learning_rate": 9.486475891807136e-06,
+      "loss": 0.4048,
+      "step": 2352
+    },
+    {
+      "epoch": 2.48994708994709,
+      "grad_norm": 0.21789073463498923,
+      "learning_rate": 9.46687573500588e-06,
+      "loss": 0.3747,
+      "step": 2353
+    },
+    {
+      "epoch": 2.491005291005291,
+      "grad_norm": 0.20505225693714865,
+      "learning_rate": 9.447275578204626e-06,
+      "loss": 0.3438,
+      "step": 2354
+    },
+    {
+      "epoch": 2.492063492063492,
+      "grad_norm": 0.19519515384302688,
+      "learning_rate": 9.427675421403372e-06,
+      "loss": 0.324,
+      "step": 2355
+    },
+    {
+      "epoch": 2.4931216931216933,
+      "grad_norm": 0.19903623860168693,
+      "learning_rate": 9.408075264602117e-06,
+      "loss": 0.3503,
+      "step": 2356
+    },
+    {
+      "epoch": 2.494179894179894,
+      "grad_norm": 0.2075371111288816,
+      "learning_rate": 9.388475107800863e-06,
+      "loss": 0.3645,
+      "step": 2357
+    },
+    {
+      "epoch": 2.4952380952380953,
+      "grad_norm": 0.20375379109475755,
+      "learning_rate": 9.368874950999607e-06,
+      "loss": 0.322,
+      "step": 2358
+    },
+    {
+      "epoch": 2.4962962962962965,
+      "grad_norm": 0.22033682656514744,
+      "learning_rate": 9.349274794198353e-06,
+      "loss": 0.3896,
+      "step": 2359
+    },
+    {
+      "epoch": 2.497354497354497,
+      "grad_norm": 0.19640384262879929,
+      "learning_rate": 9.3296746373971e-06,
+      "loss": 0.312,
+      "step": 2360
+    },
+    {
+      "epoch": 2.4984126984126984,
+      "grad_norm": 0.2081466305517114,
+      "learning_rate": 9.310074480595845e-06,
+      "loss": 0.3352,
+      "step": 2361
+    },
+    {
+      "epoch": 2.4994708994708996,
+      "grad_norm": 0.1999811658493017,
+      "learning_rate": 9.290474323794591e-06,
+      "loss": 0.34,
+      "step": 2362
+    },
+    {
+      "epoch": 2.5005291005291004,
+      "grad_norm": 0.20242709998710268,
+      "learning_rate": 9.270874166993337e-06,
+      "loss": 0.324,
+      "step": 2363
+    },
+    {
+      "epoch": 2.5015873015873016,
+      "grad_norm": 0.1890758571530466,
+      "learning_rate": 9.251274010192083e-06,
+      "loss": 0.3205,
+      "step": 2364
+    },
+    {
+      "epoch": 2.502645502645503,
+      "grad_norm": 0.2018935760610972,
+      "learning_rate": 9.231673853390829e-06,
+      "loss": 0.3512,
+      "step": 2365
+    },
+    {
+      "epoch": 2.5037037037037035,
+      "grad_norm": 0.1905973585991993,
+      "learning_rate": 9.212073696589574e-06,
+      "loss": 0.3184,
+      "step": 2366
+    },
+    {
+      "epoch": 2.5047619047619047,
+      "grad_norm": 0.20119197145153586,
+      "learning_rate": 9.192473539788319e-06,
+      "loss": 0.3715,
+      "step": 2367
+    },
+    {
+      "epoch": 2.505820105820106,
+      "grad_norm": 0.22913278918931926,
+      "learning_rate": 9.172873382987065e-06,
+      "loss": 0.3569,
+      "step": 2368
+    },
+    {
+      "epoch": 2.506878306878307,
+      "grad_norm": 0.21451859754659375,
+      "learning_rate": 9.15327322618581e-06,
+      "loss": 0.3475,
+      "step": 2369
+    },
+    {
+      "epoch": 2.507936507936508,
+      "grad_norm": 0.18032408092733698,
+      "learning_rate": 9.133673069384556e-06,
+      "loss": 0.2882,
+      "step": 2370
+    },
+    {
+      "epoch": 2.508994708994709,
+      "grad_norm": 0.18785546074799372,
+      "learning_rate": 9.1140729125833e-06,
+      "loss": 0.321,
+      "step": 2371
+    },
+    {
+      "epoch": 2.5100529100529103,
+      "grad_norm": 0.2081663751474157,
+      "learning_rate": 9.094472755782046e-06,
+      "loss": 0.37,
+      "step": 2372
+    },
+    {
+      "epoch": 2.511111111111111,
+      "grad_norm": 0.19467624754830543,
+      "learning_rate": 9.074872598980792e-06,
+      "loss": 0.3151,
+      "step": 2373
+    },
+    {
+      "epoch": 2.5121693121693123,
+      "grad_norm": 0.21111318671913534,
+      "learning_rate": 9.055272442179538e-06,
+      "loss": 0.3559,
+      "step": 2374
+    },
+    {
+      "epoch": 2.5132275132275135,
+      "grad_norm": 0.20158857860626356,
+      "learning_rate": 9.035672285378284e-06,
+      "loss": 0.3413,
+      "step": 2375
+    },
+    {
+      "epoch": 2.5142857142857142,
+      "grad_norm": 0.19967004938402297,
+      "learning_rate": 9.016072128577028e-06,
+      "loss": 0.324,
+      "step": 2376
+    },
+    {
+      "epoch": 2.5153439153439154,
+      "grad_norm": 0.20043339966519105,
+      "learning_rate": 8.996471971775774e-06,
+      "loss": 0.3575,
+      "step": 2377
+    },
+    {
+      "epoch": 2.5164021164021166,
+      "grad_norm": 0.2090345217187549,
+      "learning_rate": 8.97687181497452e-06,
+      "loss": 0.3647,
+      "step": 2378
+    },
+    {
+      "epoch": 2.5174603174603174,
+      "grad_norm": 0.19573505746191802,
+      "learning_rate": 8.957271658173266e-06,
+      "loss": 0.3559,
+      "step": 2379
+    },
+    {
+      "epoch": 2.5185185185185186,
+      "grad_norm": 0.2108974750518579,
+      "learning_rate": 8.937671501372012e-06,
+      "loss": 0.3635,
+      "step": 2380
+    },
+    {
+      "epoch": 2.51957671957672,
+      "grad_norm": 0.200163482912999,
+      "learning_rate": 8.918071344570758e-06,
+      "loss": 0.3469,
+      "step": 2381
+    },
+    {
+      "epoch": 2.5206349206349206,
+      "grad_norm": 0.20864956888060593,
+      "learning_rate": 8.898471187769503e-06,
+      "loss": 0.3657,
+      "step": 2382
+    },
+    {
+      "epoch": 2.5216931216931218,
+      "grad_norm": 0.19842095672056884,
+      "learning_rate": 8.87887103096825e-06,
+      "loss": 0.3738,
+      "step": 2383
+    },
+    {
+      "epoch": 2.522751322751323,
+      "grad_norm": 0.2135741099116242,
+      "learning_rate": 8.859270874166995e-06,
+      "loss": 0.4217,
+      "step": 2384
+    },
+    {
+      "epoch": 2.5238095238095237,
+      "grad_norm": 0.1869862686126683,
+      "learning_rate": 8.83967071736574e-06,
+      "loss": 0.3287,
+      "step": 2385
+    },
+    {
+      "epoch": 2.524867724867725,
+      "grad_norm": 0.2079464545823666,
+      "learning_rate": 8.820070560564485e-06,
+      "loss": 0.3214,
+      "step": 2386
+    },
+    {
+      "epoch": 2.525925925925926,
+      "grad_norm": 0.2196057878276039,
+      "learning_rate": 8.800470403763231e-06,
+      "loss": 0.3939,
+      "step": 2387
+    },
+    {
+      "epoch": 2.526984126984127,
+      "grad_norm": 0.20564907375985877,
+      "learning_rate": 8.780870246961977e-06,
+      "loss": 0.3698,
+      "step": 2388
+    },
+    {
+      "epoch": 2.528042328042328,
+      "grad_norm": 0.18394751990532987,
+      "learning_rate": 8.761270090160721e-06,
+      "loss": 0.293,
+      "step": 2389
+    },
+    {
+      "epoch": 2.5291005291005293,
+      "grad_norm": 0.21363483672053832,
+      "learning_rate": 8.741669933359467e-06,
+      "loss": 0.303,
+      "step": 2390
+    },
+    {
+      "epoch": 2.53015873015873,
+      "grad_norm": 0.20780527897074721,
+      "learning_rate": 8.722069776558213e-06,
+      "loss": 0.3877,
+      "step": 2391
+    },
+    {
+      "epoch": 2.5312169312169313,
+      "grad_norm": 0.19783156319626172,
+      "learning_rate": 8.702469619756959e-06,
+      "loss": 0.3512,
+      "step": 2392
+    },
+    {
+      "epoch": 2.5322751322751325,
+      "grad_norm": 0.1892822612166525,
+      "learning_rate": 8.682869462955705e-06,
+      "loss": 0.3275,
+      "step": 2393
+    },
+    {
+      "epoch": 2.533333333333333,
+      "grad_norm": 0.19170942956582057,
+      "learning_rate": 8.663269306154449e-06,
+      "loss": 0.3128,
+      "step": 2394
+    },
+    {
+      "epoch": 2.5343915343915344,
+      "grad_norm": 0.21434651400329688,
+      "learning_rate": 8.643669149353195e-06,
+      "loss": 0.371,
+      "step": 2395
+    },
+    {
+      "epoch": 2.5354497354497356,
+      "grad_norm": 0.20111551143204334,
+      "learning_rate": 8.62406899255194e-06,
+      "loss": 0.3564,
+      "step": 2396
+    },
+    {
+      "epoch": 2.5365079365079364,
+      "grad_norm": 0.18854536443130432,
+      "learning_rate": 8.604468835750686e-06,
+      "loss": 0.3282,
+      "step": 2397
+    },
+    {
+      "epoch": 2.5375661375661376,
+      "grad_norm": 0.2024488750022803,
+      "learning_rate": 8.584868678949432e-06,
+      "loss": 0.3738,
+      "step": 2398
+    },
+    {
+      "epoch": 2.538624338624339,
+      "grad_norm": 0.20611891154190004,
+      "learning_rate": 8.565268522148178e-06,
+      "loss": 0.3811,
+      "step": 2399
+    },
+    {
+      "epoch": 2.5396825396825395,
+      "grad_norm": 0.20495838632725324,
+      "learning_rate": 8.545668365346924e-06,
+      "loss": 0.3613,
+      "step": 2400
+    },
+    {
+      "epoch": 2.5407407407407407,
+      "grad_norm": 0.19072363427225575,
+      "learning_rate": 8.52606820854567e-06,
+      "loss": 0.3369,
+      "step": 2401
+    },
+    {
+      "epoch": 2.541798941798942,
+      "grad_norm": 0.19789802048128188,
+      "learning_rate": 8.506468051744414e-06,
+      "loss": 0.3239,
+      "step": 2402
+    },
+    {
+      "epoch": 2.5428571428571427,
+      "grad_norm": 0.21881605201200133,
+      "learning_rate": 8.48686789494316e-06,
+      "loss": 0.3218,
+      "step": 2403
+    },
+    {
+      "epoch": 2.543915343915344,
+      "grad_norm": 0.20069563327958265,
+      "learning_rate": 8.467267738141906e-06,
+      "loss": 0.3424,
+      "step": 2404
+    },
+    {
+      "epoch": 2.544973544973545,
+      "grad_norm": 0.2075442824382541,
+      "learning_rate": 8.447667581340652e-06,
+      "loss": 0.3579,
+      "step": 2405
+    },
+    {
+      "epoch": 2.546031746031746,
+      "grad_norm": 0.20209877990797678,
+      "learning_rate": 8.428067424539398e-06,
+      "loss": 0.3658,
+      "step": 2406
+    },
+    {
+      "epoch": 2.547089947089947,
+      "grad_norm": 0.20347999261766514,
+      "learning_rate": 8.408467267738142e-06,
+      "loss": 0.3624,
+      "step": 2407
+    },
+    {
+      "epoch": 2.5481481481481483,
+      "grad_norm": 0.19601569897380608,
+      "learning_rate": 8.388867110936888e-06,
+      "loss": 0.3342,
+      "step": 2408
+    },
+    {
+      "epoch": 2.549206349206349,
+      "grad_norm": 0.20238610460423004,
+      "learning_rate": 8.369266954135634e-06,
+      "loss": 0.3509,
+      "step": 2409
+    },
+    {
+      "epoch": 2.5502645502645502,
+      "grad_norm": 0.20063742492135442,
+      "learning_rate": 8.34966679733438e-06,
+      "loss": 0.3398,
+      "step": 2410
+    },
+    {
+      "epoch": 2.5513227513227514,
+      "grad_norm": 0.20819913291416195,
+      "learning_rate": 8.330066640533124e-06,
+      "loss": 0.3712,
+      "step": 2411
+    },
+    {
+      "epoch": 2.552380952380952,
+      "grad_norm": 0.20770506244650877,
+      "learning_rate": 8.31046648373187e-06,
+      "loss": 0.3702,
+      "step": 2412
+    },
+    {
+      "epoch": 2.5534391534391534,
+      "grad_norm": 0.1908698216977231,
+      "learning_rate": 8.290866326930615e-06,
+      "loss": 0.3442,
+      "step": 2413
+    },
+    {
+      "epoch": 2.5544973544973546,
+      "grad_norm": 0.22309789081285533,
+      "learning_rate": 8.271266170129361e-06,
+      "loss": 0.3393,
+      "step": 2414
+    },
+    {
+      "epoch": 2.5555555555555554,
+      "grad_norm": 0.2047928716854341,
+      "learning_rate": 8.251666013328107e-06,
+      "loss": 0.3602,
+      "step": 2415
+    },
+    {
+      "epoch": 2.5566137566137566,
+      "grad_norm": 0.19373128203836848,
+      "learning_rate": 8.232065856526853e-06,
+      "loss": 0.3192,
+      "step": 2416
+    },
+    {
+      "epoch": 2.5576719576719578,
+      "grad_norm": 0.2243631469556603,
+      "learning_rate": 8.212465699725599e-06,
+      "loss": 0.3471,
+      "step": 2417
+    },
+    {
+      "epoch": 2.5587301587301585,
+      "grad_norm": 0.2030278112277129,
+      "learning_rate": 8.192865542924345e-06,
+      "loss": 0.2942,
+      "step": 2418
+    },
+    {
+      "epoch": 2.5597883597883597,
+      "grad_norm": 0.21497459789916237,
+      "learning_rate": 8.17326538612309e-06,
+      "loss": 0.4136,
+      "step": 2419
+    },
+    {
+      "epoch": 2.560846560846561,
+      "grad_norm": 0.22645848971121996,
+      "learning_rate": 8.153665229321835e-06,
+      "loss": 0.3705,
+      "step": 2420
+    },
+    {
+      "epoch": 2.5619047619047617,
+      "grad_norm": 0.18927547831969765,
+      "learning_rate": 8.13406507252058e-06,
+      "loss": 0.287,
+      "step": 2421
+    },
+    {
+      "epoch": 2.562962962962963,
+      "grad_norm": 0.19829367281019175,
+      "learning_rate": 8.114464915719327e-06,
+      "loss": 0.3342,
+      "step": 2422
+    },
+    {
+      "epoch": 2.564021164021164,
+      "grad_norm": 0.21043939582778504,
+      "learning_rate": 8.094864758918072e-06,
+      "loss": 0.371,
+      "step": 2423
+    },
+    {
+      "epoch": 2.565079365079365,
+      "grad_norm": 0.21176652701168425,
+      "learning_rate": 8.075264602116818e-06,
+      "loss": 0.361,
+      "step": 2424
+    },
+    {
+      "epoch": 2.566137566137566,
+      "grad_norm": 0.19858482433563368,
+      "learning_rate": 8.055664445315562e-06,
+      "loss": 0.3435,
+      "step": 2425
+    },
+    {
+      "epoch": 2.5671957671957673,
+      "grad_norm": 0.1930389532486009,
+      "learning_rate": 8.036064288514308e-06,
+      "loss": 0.3116,
+      "step": 2426
+    },
+    {
+      "epoch": 2.568253968253968,
+      "grad_norm": 0.2017032704364643,
+      "learning_rate": 8.016464131713054e-06,
+      "loss": 0.3526,
+      "step": 2427
+    },
+    {
+      "epoch": 2.569312169312169,
+      "grad_norm": 0.2100015873501434,
+      "learning_rate": 7.9968639749118e-06,
+      "loss": 0.3647,
+      "step": 2428
+    },
+    {
+      "epoch": 2.5703703703703704,
+      "grad_norm": 0.20235907257311597,
+      "learning_rate": 7.977263818110544e-06,
+      "loss": 0.3405,
+      "step": 2429
+    },
+    {
+      "epoch": 2.571428571428571,
+      "grad_norm": 0.19998918287012984,
+      "learning_rate": 7.95766366130929e-06,
+      "loss": 0.376,
+      "step": 2430
+    },
+    {
+      "epoch": 2.5724867724867724,
+      "grad_norm": 0.19691447139385335,
+      "learning_rate": 7.938063504508036e-06,
+      "loss": 0.3494,
+      "step": 2431
+    },
+    {
+      "epoch": 2.5735449735449736,
+      "grad_norm": 0.2135251585389222,
+      "learning_rate": 7.918463347706782e-06,
+      "loss": 0.3532,
+      "step": 2432
+    },
+    {
+      "epoch": 2.5746031746031743,
+      "grad_norm": 0.201995066867421,
+      "learning_rate": 7.898863190905528e-06,
+      "loss": 0.3443,
+      "step": 2433
+    },
+    {
+      "epoch": 2.5756613756613755,
+      "grad_norm": 0.18780759974783323,
+      "learning_rate": 7.879263034104274e-06,
+      "loss": 0.3067,
+      "step": 2434
+    },
+    {
+      "epoch": 2.5767195767195767,
+      "grad_norm": 0.19161770345353862,
+      "learning_rate": 7.85966287730302e-06,
+      "loss": 0.3165,
+      "step": 2435
+    },
+    {
+      "epoch": 2.5777777777777775,
+      "grad_norm": 2.4738622305406635,
+      "learning_rate": 7.840062720501765e-06,
+      "loss": 0.4718,
+      "step": 2436
+    },
+    {
+      "epoch": 2.5788359788359787,
+      "grad_norm": 0.2163467306235148,
+      "learning_rate": 7.820462563700511e-06,
+      "loss": 0.3272,
+      "step": 2437
+    },
+    {
+      "epoch": 2.57989417989418,
+      "grad_norm": 0.20305099508166402,
+      "learning_rate": 7.800862406899255e-06,
+      "loss": 0.3354,
+      "step": 2438
+    },
+    {
+      "epoch": 2.580952380952381,
+      "grad_norm": 0.2251581241069781,
+      "learning_rate": 7.781262250098001e-06,
+      "loss": 0.3525,
+      "step": 2439
+    },
+    {
+      "epoch": 2.582010582010582,
+      "grad_norm": 0.20829738004575854,
+      "learning_rate": 7.761662093296747e-06,
+      "loss": 0.357,
+      "step": 2440
+    },
+    {
+      "epoch": 2.583068783068783,
+      "grad_norm": 0.20453941398613518,
+      "learning_rate": 7.742061936495493e-06,
+      "loss": 0.3655,
+      "step": 2441
+    },
+    {
+      "epoch": 2.5841269841269843,
+      "grad_norm": 0.2084221126285046,
+      "learning_rate": 7.722461779694237e-06,
+      "loss": 0.3545,
+      "step": 2442
+    },
+    {
+      "epoch": 2.585185185185185,
+      "grad_norm": 0.20422846872501063,
+      "learning_rate": 7.702861622892983e-06,
+      "loss": 0.3505,
+      "step": 2443
+    },
+    {
+      "epoch": 2.5862433862433862,
+      "grad_norm": 0.23322575042300528,
+      "learning_rate": 7.683261466091729e-06,
+      "loss": 0.4038,
+      "step": 2444
+    },
+    {
+      "epoch": 2.5873015873015874,
+      "grad_norm": 0.2149540741964415,
+      "learning_rate": 7.663661309290475e-06,
+      "loss": 0.365,
+      "step": 2445
+    },
+    {
+      "epoch": 2.588359788359788,
+      "grad_norm": 0.23658255878401824,
+      "learning_rate": 7.64406115248922e-06,
+      "loss": 0.3352,
+      "step": 2446
+    },
+    {
+      "epoch": 2.5894179894179894,
+      "grad_norm": 0.193188372100089,
+      "learning_rate": 7.624460995687966e-06,
+      "loss": 0.3533,
+      "step": 2447
+    },
+    {
+      "epoch": 2.5904761904761906,
+      "grad_norm": 0.2156461994282791,
+      "learning_rate": 7.604860838886712e-06,
+      "loss": 0.3092,
+      "step": 2448
+    },
+    {
+      "epoch": 2.591534391534392,
+      "grad_norm": 0.21459254756654236,
+      "learning_rate": 7.5852606820854575e-06,
+      "loss": 0.371,
+      "step": 2449
+    },
+    {
+      "epoch": 2.5925925925925926,
+      "grad_norm": 0.20067238928848508,
+      "learning_rate": 7.565660525284203e-06,
+      "loss": 0.3334,
+      "step": 2450
+    },
+    {
+      "epoch": 2.5936507936507938,
+      "grad_norm": 0.20761937713286907,
+      "learning_rate": 7.5460603684829476e-06,
+      "loss": 0.348,
+      "step": 2451
+    },
+    {
+      "epoch": 2.594708994708995,
+      "grad_norm": 0.19344185405933195,
+      "learning_rate": 7.5264602116816934e-06,
+      "loss": 0.3343,
+      "step": 2452
+    },
+    {
+      "epoch": 2.5957671957671957,
+      "grad_norm": 0.20457097110759648,
+      "learning_rate": 7.506860054880439e-06,
+      "loss": 0.3588,
+      "step": 2453
+    },
+    {
+      "epoch": 2.596825396825397,
+      "grad_norm": 0.19256191638601902,
+      "learning_rate": 7.487259898079185e-06,
+      "loss": 0.3044,
+      "step": 2454
+    },
+    {
+      "epoch": 2.597883597883598,
+      "grad_norm": 0.19715791865716,
+      "learning_rate": 7.467659741277931e-06,
+      "loss": 0.3438,
+      "step": 2455
+    },
+    {
+      "epoch": 2.598941798941799,
+      "grad_norm": 0.20066467071473018,
+      "learning_rate": 7.448059584476676e-06,
+      "loss": 0.3708,
+      "step": 2456
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.2061072533569407,
+      "learning_rate": 7.428459427675422e-06,
+      "loss": 0.365,
+      "step": 2457
+    },
+    {
+      "epoch": 2.6010582010582013,
+      "grad_norm": 0.190635839627958,
+      "learning_rate": 7.408859270874168e-06,
+      "loss": 0.3185,
+      "step": 2458
+    },
+    {
+      "epoch": 2.602116402116402,
+      "grad_norm": 0.21002157908314345,
+      "learning_rate": 7.389259114072914e-06,
+      "loss": 0.3322,
+      "step": 2459
+    },
+    {
+      "epoch": 2.6031746031746033,
+      "grad_norm": 0.18713946271024856,
+      "learning_rate": 7.369658957271658e-06,
+      "loss": 0.3126,
+      "step": 2460
+    },
+    {
+      "epoch": 2.6042328042328045,
+      "grad_norm": 0.2132276955579535,
+      "learning_rate": 7.350058800470404e-06,
+      "loss": 0.3714,
+      "step": 2461
+    },
+    {
+      "epoch": 2.605291005291005,
+      "grad_norm": 0.1976682334299022,
+      "learning_rate": 7.33045864366915e-06,
+      "loss": 0.3446,
+      "step": 2462
+    },
+    {
+      "epoch": 2.6063492063492064,
+      "grad_norm": 0.19144499037515725,
+      "learning_rate": 7.3108584868678955e-06,
+      "loss": 0.334,
+      "step": 2463
+    },
+    {
+      "epoch": 2.6074074074074076,
+      "grad_norm": 0.20970237193504582,
+      "learning_rate": 7.291258330066641e-06,
+      "loss": 0.3803,
+      "step": 2464
+    },
+    {
+      "epoch": 2.6084656084656084,
+      "grad_norm": 0.19369116393515454,
+      "learning_rate": 7.271658173265386e-06,
+      "loss": 0.2944,
+      "step": 2465
+    },
+    {
+      "epoch": 2.6095238095238096,
+      "grad_norm": 0.20464263215117728,
+      "learning_rate": 7.252058016464132e-06,
+      "loss": 0.3118,
+      "step": 2466
+    },
+    {
+      "epoch": 2.610582010582011,
+      "grad_norm": 0.19514819457072263,
+      "learning_rate": 7.232457859662878e-06,
+      "loss": 0.3373,
+      "step": 2467
+    },
+    {
+      "epoch": 2.6116402116402115,
+      "grad_norm": 0.2054577341252547,
+      "learning_rate": 7.212857702861624e-06,
+      "loss": 0.3312,
+      "step": 2468
+    },
+    {
+      "epoch": 2.6126984126984127,
+      "grad_norm": 0.218003406758833,
+      "learning_rate": 7.193257546060368e-06,
+      "loss": 0.3636,
+      "step": 2469
+    },
+    {
+      "epoch": 2.613756613756614,
+      "grad_norm": 0.44213851229624174,
+      "learning_rate": 7.173657389259114e-06,
+      "loss": 0.3976,
+      "step": 2470
+    },
+    {
+      "epoch": 2.6148148148148147,
+      "grad_norm": 0.2009534741403274,
+      "learning_rate": 7.15405723245786e-06,
+      "loss": 0.3442,
+      "step": 2471
+    },
+    {
+      "epoch": 2.615873015873016,
+      "grad_norm": 0.20912608516036513,
+      "learning_rate": 7.134457075656606e-06,
+      "loss": 0.3854,
+      "step": 2472
+    },
+    {
+      "epoch": 2.616931216931217,
+      "grad_norm": 0.20078773548602796,
+      "learning_rate": 7.114856918855352e-06,
+      "loss": 0.3793,
+      "step": 2473
+    },
+    {
+      "epoch": 2.617989417989418,
+      "grad_norm": 0.21374608990215083,
+      "learning_rate": 7.095256762054097e-06,
+      "loss": 0.3883,
+      "step": 2474
+    },
+    {
+      "epoch": 2.619047619047619,
+      "grad_norm": 0.20326986212714018,
+      "learning_rate": 7.075656605252843e-06,
+      "loss": 0.3634,
+      "step": 2475
+    },
+    {
+      "epoch": 2.6201058201058203,
+      "grad_norm": 0.184483550265106,
+      "learning_rate": 7.0560564484515885e-06,
+      "loss": 0.3251,
+      "step": 2476
+    },
+    {
+      "epoch": 2.621164021164021,
+      "grad_norm": 0.20518627904667391,
+      "learning_rate": 7.036456291650334e-06,
+      "loss": 0.3383,
+      "step": 2477
+    },
+    {
+      "epoch": 2.6222222222222222,
+      "grad_norm": 0.2213975218224379,
+      "learning_rate": 7.0168561348490785e-06,
+      "loss": 0.3805,
+      "step": 2478
+    },
+    {
+      "epoch": 2.6232804232804234,
+      "grad_norm": 0.20669880976906882,
+      "learning_rate": 6.997255978047824e-06,
+      "loss": 0.3293,
+      "step": 2479
+    },
+    {
+      "epoch": 2.624338624338624,
+      "grad_norm": 0.20421330013118466,
+      "learning_rate": 6.97765582124657e-06,
+      "loss": 0.3308,
+      "step": 2480
+    },
+    {
+      "epoch": 2.6253968253968254,
+      "grad_norm": 0.23282007125318938,
+      "learning_rate": 6.958055664445316e-06,
+      "loss": 0.3844,
+      "step": 2481
+    },
+    {
+      "epoch": 2.6264550264550266,
+      "grad_norm": 0.20766235706851663,
+      "learning_rate": 6.938455507644062e-06,
+      "loss": 0.3729,
+      "step": 2482
+    },
+    {
+      "epoch": 2.6275132275132274,
+      "grad_norm": 0.20961331760975993,
+      "learning_rate": 6.918855350842807e-06,
+      "loss": 0.3733,
+      "step": 2483
+    },
+    {
+      "epoch": 2.6285714285714286,
+      "grad_norm": 0.20224319032663668,
+      "learning_rate": 6.899255194041553e-06,
+      "loss": 0.3737,
+      "step": 2484
+    },
+    {
+      "epoch": 2.6296296296296298,
+      "grad_norm": 0.17940459522018293,
+      "learning_rate": 6.879655037240299e-06,
+      "loss": 0.3172,
+      "step": 2485
+    },
+    {
+      "epoch": 2.6306878306878305,
+      "grad_norm": 0.18842546291187015,
+      "learning_rate": 6.860054880439045e-06,
+      "loss": 0.3237,
+      "step": 2486
+    },
+    {
+      "epoch": 2.6317460317460317,
+      "grad_norm": 0.20586867202259435,
+      "learning_rate": 6.840454723637789e-06,
+      "loss": 0.3341,
+      "step": 2487
+    },
+    {
+      "epoch": 2.632804232804233,
+      "grad_norm": 0.19471736282988128,
+      "learning_rate": 6.820854566836535e-06,
+      "loss": 0.3179,
+      "step": 2488
+    },
+    {
+      "epoch": 2.6338624338624337,
+      "grad_norm": 0.20284155346377497,
+      "learning_rate": 6.801254410035281e-06,
+      "loss": 0.3665,
+      "step": 2489
+    },
+    {
+      "epoch": 2.634920634920635,
+      "grad_norm": 0.18879946930455782,
+      "learning_rate": 6.7816542532340265e-06,
+      "loss": 0.3259,
+      "step": 2490
+    },
+    {
+      "epoch": 2.635978835978836,
+      "grad_norm": 0.19398530466899666,
+      "learning_rate": 6.762054096432771e-06,
+      "loss": 0.3179,
+      "step": 2491
+    },
+    {
+      "epoch": 2.637037037037037,
+      "grad_norm": 0.19234318749673635,
+      "learning_rate": 6.742453939631517e-06,
+      "loss": 0.3446,
+      "step": 2492
+    },
+    {
+      "epoch": 2.638095238095238,
+      "grad_norm": 0.21354926873262906,
+      "learning_rate": 6.722853782830263e-06,
+      "loss": 0.3525,
+      "step": 2493
+    },
+    {
+      "epoch": 2.6391534391534393,
+      "grad_norm": 0.20334093193943195,
+      "learning_rate": 6.703253626029009e-06,
+      "loss": 0.36,
+      "step": 2494
+    },
+    {
+      "epoch": 2.64021164021164,
+      "grad_norm": 0.2035547977347391,
+      "learning_rate": 6.683653469227755e-06,
+      "loss": 0.3542,
+      "step": 2495
+    },
+    {
+      "epoch": 2.641269841269841,
+      "grad_norm": 0.19627021534313335,
+      "learning_rate": 6.664053312426499e-06,
+      "loss": 0.334,
+      "step": 2496
+    },
+    {
+      "epoch": 2.6423280423280424,
+      "grad_norm": 0.2086234488088156,
+      "learning_rate": 6.644453155625245e-06,
+      "loss": 0.3619,
+      "step": 2497
+    },
+    {
+      "epoch": 2.643386243386243,
+      "grad_norm": 0.2201861596648691,
+      "learning_rate": 6.624852998823991e-06,
+      "loss": 0.389,
+      "step": 2498
+    },
+    {
+      "epoch": 2.6444444444444444,
+      "grad_norm": 0.20846191600277475,
+      "learning_rate": 6.605252842022737e-06,
+      "loss": 0.3779,
+      "step": 2499
+    },
+    {
+      "epoch": 2.6455026455026456,
+      "grad_norm": 0.20734012321675435,
+      "learning_rate": 6.585652685221482e-06,
+      "loss": 0.3508,
+      "step": 2500
+    },
+    {
+      "epoch": 2.6465608465608463,
+      "grad_norm": 0.19345863392721405,
+      "learning_rate": 6.566052528420228e-06,
+      "loss": 0.3295,
+      "step": 2501
+    },
+    {
+      "epoch": 2.6476190476190475,
+      "grad_norm": 0.20354235487098435,
+      "learning_rate": 6.546452371618974e-06,
+      "loss": 0.376,
+      "step": 2502
+    },
+    {
+      "epoch": 2.6486772486772487,
+      "grad_norm": 0.19936131307592084,
+      "learning_rate": 6.5268522148177195e-06,
+      "loss": 0.3372,
+      "step": 2503
+    },
+    {
+      "epoch": 2.6497354497354495,
+      "grad_norm": 0.19833040616276873,
+      "learning_rate": 6.507252058016465e-06,
+      "loss": 0.3249,
+      "step": 2504
+    },
+    {
+      "epoch": 2.6507936507936507,
+      "grad_norm": 0.22382325734125558,
+      "learning_rate": 6.4876519012152095e-06,
+      "loss": 0.3898,
+      "step": 2505
+    },
+    {
+      "epoch": 2.651851851851852,
+      "grad_norm": 0.2086391866302971,
+      "learning_rate": 6.468051744413955e-06,
+      "loss": 0.3822,
+      "step": 2506
+    },
+    {
+      "epoch": 2.6529100529100527,
+      "grad_norm": 0.19118652570324526,
+      "learning_rate": 6.448451587612701e-06,
+      "loss": 0.3232,
+      "step": 2507
+    },
+    {
+      "epoch": 2.653968253968254,
+      "grad_norm": 0.1992916481134372,
+      "learning_rate": 6.428851430811447e-06,
+      "loss": 0.3596,
+      "step": 2508
+    },
+    {
+      "epoch": 2.655026455026455,
+      "grad_norm": 0.20052752929381737,
+      "learning_rate": 6.409251274010192e-06,
+      "loss": 0.3474,
+      "step": 2509
+    },
+    {
+      "epoch": 2.656084656084656,
+      "grad_norm": 0.1892696179940764,
+      "learning_rate": 6.389651117208938e-06,
+      "loss": 0.3491,
+      "step": 2510
+    },
+    {
+      "epoch": 2.657142857142857,
+      "grad_norm": 0.18821467544895704,
+      "learning_rate": 6.370050960407684e-06,
+      "loss": 0.3406,
+      "step": 2511
+    },
+    {
+      "epoch": 2.6582010582010582,
+      "grad_norm": 0.2146316899839704,
+      "learning_rate": 6.35045080360643e-06,
+      "loss": 0.3268,
+      "step": 2512
+    },
+    {
+      "epoch": 2.659259259259259,
+      "grad_norm": 0.1924428845095873,
+      "learning_rate": 6.330850646805176e-06,
+      "loss": 0.3278,
+      "step": 2513
+    },
+    {
+      "epoch": 2.66031746031746,
+      "grad_norm": 0.20147304345647865,
+      "learning_rate": 6.31125049000392e-06,
+      "loss": 0.3452,
+      "step": 2514
+    },
+    {
+      "epoch": 2.6613756613756614,
+      "grad_norm": 0.18452957282061885,
+      "learning_rate": 6.291650333202666e-06,
+      "loss": 0.3281,
+      "step": 2515
+    },
+    {
+      "epoch": 2.6624338624338626,
+      "grad_norm": 0.1857986422958092,
+      "learning_rate": 6.272050176401412e-06,
+      "loss": 0.3101,
+      "step": 2516
+    },
+    {
+      "epoch": 2.6634920634920634,
+      "grad_norm": 0.21536712844728778,
+      "learning_rate": 6.2524500196001575e-06,
+      "loss": 0.3835,
+      "step": 2517
+    },
+    {
+      "epoch": 2.6645502645502646,
+      "grad_norm": 0.19196672833526432,
+      "learning_rate": 6.2328498627989025e-06,
+      "loss": 0.331,
+      "step": 2518
+    },
+    {
+      "epoch": 2.6656084656084658,
+      "grad_norm": 0.20346805393197445,
+      "learning_rate": 6.213249705997648e-06,
+      "loss": 0.3718,
+      "step": 2519
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.19865890016978405,
+      "learning_rate": 6.193649549196394e-06,
+      "loss": 0.3476,
+      "step": 2520
+    },
+    {
+      "epoch": 2.6677248677248677,
+      "grad_norm": 0.1868360318396176,
+      "learning_rate": 6.17404939239514e-06,
+      "loss": 0.3461,
+      "step": 2521
+    },
+    {
+      "epoch": 2.668783068783069,
+      "grad_norm": 0.18283098776282525,
+      "learning_rate": 6.154449235593885e-06,
+      "loss": 0.3185,
+      "step": 2522
+    },
+    {
+      "epoch": 2.6698412698412697,
+      "grad_norm": 0.19611612634805634,
+      "learning_rate": 6.134849078792631e-06,
+      "loss": 0.3348,
+      "step": 2523
+    },
+    {
+      "epoch": 2.670899470899471,
+      "grad_norm": 0.19050047780185728,
+      "learning_rate": 6.115248921991376e-06,
+      "loss": 0.3277,
+      "step": 2524
+    },
+    {
+      "epoch": 2.671957671957672,
+      "grad_norm": 0.2084510520678467,
+      "learning_rate": 6.095648765190122e-06,
+      "loss": 0.3225,
+      "step": 2525
+    },
+    {
+      "epoch": 2.6730158730158733,
+      "grad_norm": 0.1922454758562699,
+      "learning_rate": 6.076048608388867e-06,
+      "loss": 0.3602,
+      "step": 2526
+    },
+    {
+      "epoch": 2.674074074074074,
+      "grad_norm": 0.19098409079480982,
+      "learning_rate": 6.056448451587613e-06,
+      "loss": 0.3501,
+      "step": 2527
+    },
+    {
+      "epoch": 2.6751322751322753,
+      "grad_norm": 0.19240132190670328,
+      "learning_rate": 6.036848294786359e-06,
+      "loss": 0.3455,
+      "step": 2528
+    },
+    {
+      "epoch": 2.6761904761904765,
+      "grad_norm": 0.18234297191193383,
+      "learning_rate": 6.017248137985105e-06,
+      "loss": 0.3248,
+      "step": 2529
+    },
+    {
+      "epoch": 2.677248677248677,
+      "grad_norm": 0.18076554021932248,
+      "learning_rate": 5.99764798118385e-06,
+      "loss": 0.3065,
+      "step": 2530
+    },
+    {
+      "epoch": 2.6783068783068784,
+      "grad_norm": 0.1929979054056131,
+      "learning_rate": 5.9780478243825955e-06,
+      "loss": 0.3604,
+      "step": 2531
+    },
+    {
+      "epoch": 2.6793650793650796,
+      "grad_norm": 0.20221424529505985,
+      "learning_rate": 5.958447667581341e-06,
+      "loss": 0.3857,
+      "step": 2532
+    },
+    {
+      "epoch": 2.6804232804232804,
+      "grad_norm": 0.20020493436720654,
+      "learning_rate": 5.938847510780086e-06,
+      "loss": 0.3353,
+      "step": 2533
+    },
+    {
+      "epoch": 2.6814814814814816,
+      "grad_norm": 0.1979878253122416,
+      "learning_rate": 5.919247353978832e-06,
+      "loss": 0.3524,
+      "step": 2534
+    },
+    {
+      "epoch": 2.682539682539683,
+      "grad_norm": 0.1878994433039551,
+      "learning_rate": 5.899647197177577e-06,
+      "loss": 0.3234,
+      "step": 2535
+    },
+    {
+      "epoch": 2.6835978835978835,
+      "grad_norm": 0.20303922395745672,
+      "learning_rate": 5.880047040376323e-06,
+      "loss": 0.3472,
+      "step": 2536
+    },
+    {
+      "epoch": 2.6846560846560847,
+      "grad_norm": 0.1964831514888397,
+      "learning_rate": 5.860446883575069e-06,
+      "loss": 0.3512,
+      "step": 2537
+    },
+    {
+      "epoch": 2.685714285714286,
+      "grad_norm": 0.20050158418913078,
+      "learning_rate": 5.840846726773815e-06,
+      "loss": 0.368,
+      "step": 2538
+    },
+    {
+      "epoch": 2.6867724867724867,
+      "grad_norm": 0.19809108859289698,
+      "learning_rate": 5.82124656997256e-06,
+      "loss": 0.3642,
+      "step": 2539
+    },
+    {
+      "epoch": 2.687830687830688,
+      "grad_norm": 0.19542084982885585,
+      "learning_rate": 5.801646413171306e-06,
+      "loss": 0.3505,
+      "step": 2540
+    },
+    {
+      "epoch": 2.688888888888889,
+      "grad_norm": 0.25026144659287575,
+      "learning_rate": 5.782046256370052e-06,
+      "loss": 0.3564,
+      "step": 2541
+    },
+    {
+      "epoch": 2.68994708994709,
+      "grad_norm": 0.19368060067900453,
+      "learning_rate": 5.762446099568797e-06,
+      "loss": 0.3444,
+      "step": 2542
+    },
+    {
+      "epoch": 2.691005291005291,
+      "grad_norm": 0.2015994007513886,
+      "learning_rate": 5.742845942767543e-06,
+      "loss": 0.3331,
+      "step": 2543
+    },
+    {
+      "epoch": 2.6920634920634923,
+      "grad_norm": 0.20202195926051253,
+      "learning_rate": 5.723245785966288e-06,
+      "loss": 0.3452,
+      "step": 2544
+    },
+    {
+      "epoch": 2.693121693121693,
+      "grad_norm": 0.18149608421071714,
+      "learning_rate": 5.7036456291650335e-06,
+      "loss": 0.2989,
+      "step": 2545
+    },
+    {
+      "epoch": 2.6941798941798942,
+      "grad_norm": 0.21151146883978197,
+      "learning_rate": 5.684045472363779e-06,
+      "loss": 0.3899,
+      "step": 2546
+    },
+    {
+      "epoch": 2.6952380952380954,
+      "grad_norm": 0.19106657712419947,
+      "learning_rate": 5.664445315562525e-06,
+      "loss": 0.3133,
+      "step": 2547
+    },
+    {
+      "epoch": 2.696296296296296,
+      "grad_norm": 0.204486097348059,
+      "learning_rate": 5.64484515876127e-06,
+      "loss": 0.3243,
+      "step": 2548
+    },
+    {
+      "epoch": 2.6973544973544974,
+      "grad_norm": 0.2120307771160055,
+      "learning_rate": 5.625245001960016e-06,
+      "loss": 0.3616,
+      "step": 2549
+    },
+    {
+      "epoch": 2.6984126984126986,
+      "grad_norm": 0.17095776864486206,
+      "learning_rate": 5.605644845158761e-06,
+      "loss": 0.2899,
+      "step": 2550
+    },
+    {
+      "epoch": 2.6994708994708994,
+      "grad_norm": 0.21843454968847806,
+      "learning_rate": 5.586044688357507e-06,
+      "loss": 0.3463,
+      "step": 2551
+    },
+    {
+      "epoch": 2.7005291005291006,
+      "grad_norm": 0.19118514569525785,
+      "learning_rate": 5.566444531556253e-06,
+      "loss": 0.3216,
+      "step": 2552
+    },
+    {
+      "epoch": 2.7015873015873018,
+      "grad_norm": 0.25238654920190934,
+      "learning_rate": 5.546844374754998e-06,
+      "loss": 0.3795,
+      "step": 2553
+    },
+    {
+      "epoch": 2.7026455026455025,
+      "grad_norm": 0.20643432168257772,
+      "learning_rate": 5.527244217953744e-06,
+      "loss": 0.366,
+      "step": 2554
+    },
+    {
+      "epoch": 2.7037037037037037,
+      "grad_norm": 0.1951728012706216,
+      "learning_rate": 5.50764406115249e-06,
+      "loss": 0.3211,
+      "step": 2555
+    },
+    {
+      "epoch": 2.704761904761905,
+      "grad_norm": 0.19406462105051225,
+      "learning_rate": 5.488043904351236e-06,
+      "loss": 0.3226,
+      "step": 2556
+    },
+    {
+      "epoch": 2.7058201058201057,
+      "grad_norm": 0.19140790364896687,
+      "learning_rate": 5.468443747549981e-06,
+      "loss": 0.32,
+      "step": 2557
+    },
+    {
+      "epoch": 2.706878306878307,
+      "grad_norm": 0.19071360052522665,
+      "learning_rate": 5.4488435907487265e-06,
+      "loss": 0.3619,
+      "step": 2558
+    },
+    {
+      "epoch": 2.707936507936508,
+      "grad_norm": 0.2066029798654874,
+      "learning_rate": 5.4292434339474715e-06,
+      "loss": 0.37,
+      "step": 2559
+    },
+    {
+      "epoch": 2.708994708994709,
+      "grad_norm": 0.2000206802025288,
+      "learning_rate": 5.409643277146217e-06,
+      "loss": 0.3739,
+      "step": 2560
+    },
+    {
+      "epoch": 2.71005291005291,
+      "grad_norm": 0.1992451517241142,
+      "learning_rate": 5.390043120344963e-06,
+      "loss": 0.2976,
+      "step": 2561
+    },
+    {
+      "epoch": 2.7111111111111112,
+      "grad_norm": 0.21041064516580602,
+      "learning_rate": 5.370442963543708e-06,
+      "loss": 0.3372,
+      "step": 2562
+    },
+    {
+      "epoch": 2.712169312169312,
+      "grad_norm": 0.1965382228229818,
+      "learning_rate": 5.350842806742454e-06,
+      "loss": 0.3181,
+      "step": 2563
+    },
+    {
+      "epoch": 2.713227513227513,
+      "grad_norm": 0.19553975072971144,
+      "learning_rate": 5.3312426499412e-06,
+      "loss": 0.3334,
+      "step": 2564
+    },
+    {
+      "epoch": 2.7142857142857144,
+      "grad_norm": 0.182293586725453,
+      "learning_rate": 5.311642493139946e-06,
+      "loss": 0.312,
+      "step": 2565
+    },
+    {
+      "epoch": 2.715343915343915,
+      "grad_norm": 0.22281231749006977,
+      "learning_rate": 5.292042336338691e-06,
+      "loss": 0.4265,
+      "step": 2566
+    },
+    {
+      "epoch": 2.7164021164021164,
+      "grad_norm": 0.19852030519721917,
+      "learning_rate": 5.272442179537437e-06,
+      "loss": 0.3736,
+      "step": 2567
+    },
+    {
+      "epoch": 2.7174603174603176,
+      "grad_norm": 0.20060353048459667,
+      "learning_rate": 5.252842022736182e-06,
+      "loss": 0.3608,
+      "step": 2568
+    },
+    {
+      "epoch": 2.7185185185185183,
+      "grad_norm": 0.2043124213401307,
+      "learning_rate": 5.233241865934928e-06,
+      "loss": 0.3695,
+      "step": 2569
+    },
+    {
+      "epoch": 2.7195767195767195,
+      "grad_norm": 0.21279444075120416,
+      "learning_rate": 5.213641709133673e-06,
+      "loss": 0.3461,
+      "step": 2570
+    },
+    {
+      "epoch": 2.7206349206349207,
+      "grad_norm": 0.1962353653907673,
+      "learning_rate": 5.194041552332419e-06,
+      "loss": 0.3479,
+      "step": 2571
+    },
+    {
+      "epoch": 2.7216931216931215,
+      "grad_norm": 0.19136776040919806,
+      "learning_rate": 5.1744413955311645e-06,
+      "loss": 0.3381,
+      "step": 2572
+    },
+    {
+      "epoch": 2.7227513227513227,
+      "grad_norm": 0.19168537594384008,
+      "learning_rate": 5.15484123872991e-06,
+      "loss": 0.332,
+      "step": 2573
+    },
+    {
+      "epoch": 2.723809523809524,
+      "grad_norm": 0.18822366734415621,
+      "learning_rate": 5.135241081928656e-06,
+      "loss": 0.3287,
+      "step": 2574
+    },
+    {
+      "epoch": 2.7248677248677247,
+      "grad_norm": 0.2077200076431067,
+      "learning_rate": 5.115640925127401e-06,
+      "loss": 0.3431,
+      "step": 2575
+    },
+    {
+      "epoch": 2.725925925925926,
+      "grad_norm": 0.19269703180103093,
+      "learning_rate": 5.096040768326147e-06,
+      "loss": 0.3086,
+      "step": 2576
+    },
+    {
+      "epoch": 2.726984126984127,
+      "grad_norm": 0.18860716399222538,
+      "learning_rate": 5.076440611524892e-06,
+      "loss": 0.3154,
+      "step": 2577
+    },
+    {
+      "epoch": 2.728042328042328,
+      "grad_norm": 0.1958899798494481,
+      "learning_rate": 5.056840454723638e-06,
+      "loss": 0.3111,
+      "step": 2578
+    },
+    {
+      "epoch": 2.729100529100529,
+      "grad_norm": 0.1820101679127598,
+      "learning_rate": 5.037240297922383e-06,
+      "loss": 0.3145,
+      "step": 2579
+    },
+    {
+      "epoch": 2.7301587301587302,
+      "grad_norm": 0.1983802921074873,
+      "learning_rate": 5.017640141121129e-06,
+      "loss": 0.3595,
+      "step": 2580
+    },
+    {
+      "epoch": 2.731216931216931,
+      "grad_norm": 0.19970794198008765,
+      "learning_rate": 4.998039984319875e-06,
+      "loss": 0.3572,
+      "step": 2581
+    },
+    {
+      "epoch": 2.732275132275132,
+      "grad_norm": 0.1893406404078852,
+      "learning_rate": 4.978439827518621e-06,
+      "loss": 0.3426,
+      "step": 2582
+    },
+    {
+      "epoch": 2.7333333333333334,
+      "grad_norm": 0.20404466163446328,
+      "learning_rate": 4.9588396707173666e-06,
+      "loss": 0.3658,
+      "step": 2583
+    },
+    {
+      "epoch": 2.734391534391534,
+      "grad_norm": 0.18412114768156596,
+      "learning_rate": 4.939239513916112e-06,
+      "loss": 0.3046,
+      "step": 2584
+    },
+    {
+      "epoch": 2.7354497354497354,
+      "grad_norm": 0.2075384994168872,
+      "learning_rate": 4.9196393571148575e-06,
+      "loss": 0.338,
+      "step": 2585
+    },
+    {
+      "epoch": 2.7365079365079366,
+      "grad_norm": 0.22412000158196252,
+      "learning_rate": 4.9000392003136025e-06,
+      "loss": 0.4155,
+      "step": 2586
+    },
+    {
+      "epoch": 2.7375661375661373,
+      "grad_norm": 0.20401992100764668,
+      "learning_rate": 4.880439043512348e-06,
+      "loss": 0.3865,
+      "step": 2587
+    },
+    {
+      "epoch": 2.7386243386243385,
+      "grad_norm": 0.18604920498095107,
+      "learning_rate": 4.860838886711093e-06,
+      "loss": 0.3134,
+      "step": 2588
+    },
+    {
+      "epoch": 2.7396825396825397,
+      "grad_norm": 0.21195152122045988,
+      "learning_rate": 4.841238729909839e-06,
+      "loss": 0.3833,
+      "step": 2589
+    },
+    {
+      "epoch": 2.7407407407407405,
+      "grad_norm": 0.19793521415832854,
+      "learning_rate": 4.821638573108585e-06,
+      "loss": 0.3654,
+      "step": 2590
+    },
+    {
+      "epoch": 2.7417989417989417,
+      "grad_norm": 0.19239288309471586,
+      "learning_rate": 4.802038416307331e-06,
+      "loss": 0.3311,
+      "step": 2591
+    },
+    {
+      "epoch": 2.742857142857143,
+      "grad_norm": 0.18478342433128606,
+      "learning_rate": 4.782438259506077e-06,
+      "loss": 0.3196,
+      "step": 2592
+    },
+    {
+      "epoch": 2.7439153439153436,
+      "grad_norm": 0.19737300368776284,
+      "learning_rate": 4.762838102704822e-06,
+      "loss": 0.3399,
+      "step": 2593
+    },
+    {
+      "epoch": 2.744973544973545,
+      "grad_norm": 0.2044197261835309,
+      "learning_rate": 4.743237945903568e-06,
+      "loss": 0.3874,
+      "step": 2594
+    },
+    {
+      "epoch": 2.746031746031746,
+      "grad_norm": 0.2160010979562428,
+      "learning_rate": 4.723637789102313e-06,
+      "loss": 0.385,
+      "step": 2595
+    },
+    {
+      "epoch": 2.7470899470899472,
+      "grad_norm": 0.18127732926658155,
+      "learning_rate": 4.704037632301059e-06,
+      "loss": 0.3192,
+      "step": 2596
+    },
+    {
+      "epoch": 2.748148148148148,
+      "grad_norm": 0.18852738977755046,
+      "learning_rate": 4.684437475499804e-06,
+      "loss": 0.3439,
+      "step": 2597
+    },
+    {
+      "epoch": 2.749206349206349,
+      "grad_norm": 0.17640615561096873,
+      "learning_rate": 4.66483731869855e-06,
+      "loss": 0.2992,
+      "step": 2598
+    },
+    {
+      "epoch": 2.7502645502645504,
+      "grad_norm": 0.20969394677698452,
+      "learning_rate": 4.6452371618972955e-06,
+      "loss": 0.4039,
+      "step": 2599
+    },
+    {
+      "epoch": 2.751322751322751,
+      "grad_norm": 0.1938761616618117,
+      "learning_rate": 4.625637005096041e-06,
+      "loss": 0.3456,
+      "step": 2600
+    },
+    {
+      "epoch": 2.7523809523809524,
+      "grad_norm": 0.20065804261742967,
+      "learning_rate": 4.606036848294787e-06,
+      "loss": 0.379,
+      "step": 2601
+    },
+    {
+      "epoch": 2.7534391534391536,
+      "grad_norm": 0.18022036699122285,
+      "learning_rate": 4.586436691493532e-06,
+      "loss": 0.2838,
+      "step": 2602
+    },
+    {
+      "epoch": 2.7544973544973543,
+      "grad_norm": 0.20021693200514237,
+      "learning_rate": 4.566836534692278e-06,
+      "loss": 0.3744,
+      "step": 2603
+    },
+    {
+      "epoch": 2.7555555555555555,
+      "grad_norm": 0.19141878650651184,
+      "learning_rate": 4.547236377891023e-06,
+      "loss": 0.2937,
+      "step": 2604
+    },
+    {
+      "epoch": 2.7566137566137567,
+      "grad_norm": 0.2529365210436403,
+      "learning_rate": 4.527636221089769e-06,
+      "loss": 0.3638,
+      "step": 2605
+    },
+    {
+      "epoch": 2.757671957671958,
+      "grad_norm": 0.1854468401388474,
+      "learning_rate": 4.508036064288514e-06,
+      "loss": 0.3123,
+      "step": 2606
+    },
+    {
+      "epoch": 2.7587301587301587,
+      "grad_norm": 0.20565208226029344,
+      "learning_rate": 4.48843590748726e-06,
+      "loss": 0.3706,
+      "step": 2607
+    },
+    {
+      "epoch": 2.75978835978836,
+      "grad_norm": 0.19044667922854414,
+      "learning_rate": 4.468835750686006e-06,
+      "loss": 0.3196,
+      "step": 2608
+    },
+    {
+      "epoch": 2.760846560846561,
+      "grad_norm": 0.19238853879054066,
+      "learning_rate": 4.449235593884752e-06,
+      "loss": 0.3148,
+      "step": 2609
+    },
+    {
+      "epoch": 2.761904761904762,
+      "grad_norm": 0.1956487628537959,
+      "learning_rate": 4.4296354370834976e-06,
+      "loss": 0.3521,
+      "step": 2610
+    },
+    {
+      "epoch": 2.762962962962963,
+      "grad_norm": 0.20209286152850098,
+      "learning_rate": 4.410035280282243e-06,
+      "loss": 0.3717,
+      "step": 2611
+    },
+    {
+      "epoch": 2.7640211640211643,
+      "grad_norm": 0.19340479563705504,
+      "learning_rate": 4.3904351234809885e-06,
+      "loss": 0.3598,
+      "step": 2612
+    },
+    {
+      "epoch": 2.765079365079365,
+      "grad_norm": 0.19526534766192782,
+      "learning_rate": 4.3708349666797335e-06,
+      "loss": 0.3709,
+      "step": 2613
+    },
+    {
+      "epoch": 2.7661375661375662,
+      "grad_norm": 0.1876434539839283,
+      "learning_rate": 4.351234809878479e-06,
+      "loss": 0.3356,
+      "step": 2614
+    },
+    {
+      "epoch": 2.7671957671957674,
+      "grad_norm": 0.18913882028697904,
+      "learning_rate": 4.331634653077224e-06,
+      "loss": 0.3331,
+      "step": 2615
+    },
+    {
+      "epoch": 2.768253968253968,
+      "grad_norm": 0.18531612005736103,
+      "learning_rate": 4.31203449627597e-06,
+      "loss": 0.3121,
+      "step": 2616
+    },
+    {
+      "epoch": 2.7693121693121694,
+      "grad_norm": 0.18210121950128233,
+      "learning_rate": 4.292434339474716e-06,
+      "loss": 0.3217,
+      "step": 2617
+    },
+    {
+      "epoch": 2.7703703703703706,
+      "grad_norm": 0.17631791372733657,
+      "learning_rate": 4.272834182673462e-06,
+      "loss": 0.3145,
+      "step": 2618
+    },
+    {
+      "epoch": 2.7714285714285714,
+      "grad_norm": 0.2052048552010276,
+      "learning_rate": 4.253234025872207e-06,
+      "loss": 0.3445,
+      "step": 2619
+    },
+    {
+      "epoch": 2.7724867724867726,
+      "grad_norm": 0.20391385871973938,
+      "learning_rate": 4.233633869070953e-06,
+      "loss": 0.3821,
+      "step": 2620
+    },
+    {
+      "epoch": 2.7735449735449738,
+      "grad_norm": 0.19379016138412378,
+      "learning_rate": 4.214033712269699e-06,
+      "loss": 0.3808,
+      "step": 2621
+    },
+    {
+      "epoch": 2.7746031746031745,
+      "grad_norm": 0.18659539827425273,
+      "learning_rate": 4.194433555468444e-06,
+      "loss": 0.3072,
+      "step": 2622
+    },
+    {
+      "epoch": 2.7756613756613757,
+      "grad_norm": 0.1894212818459714,
+      "learning_rate": 4.17483339866719e-06,
+      "loss": 0.3268,
+      "step": 2623
+    },
+    {
+      "epoch": 2.776719576719577,
+      "grad_norm": 0.190978704730525,
+      "learning_rate": 4.155233241865935e-06,
+      "loss": 0.3382,
+      "step": 2624
+    },
+    {
+      "epoch": 2.7777777777777777,
+      "grad_norm": 0.20109476739966778,
+      "learning_rate": 4.135633085064681e-06,
+      "loss": 0.3827,
+      "step": 2625
+    },
+    {
+      "epoch": 2.778835978835979,
+      "grad_norm": 0.19363672859817238,
+      "learning_rate": 4.1160329282634265e-06,
+      "loss": 0.3587,
+      "step": 2626
+    },
+    {
+      "epoch": 2.77989417989418,
+      "grad_norm": 0.2010695366446098,
+      "learning_rate": 4.096432771462172e-06,
+      "loss": 0.3657,
+      "step": 2627
+    },
+    {
+      "epoch": 2.780952380952381,
+      "grad_norm": 0.1910110293287592,
+      "learning_rate": 4.076832614660917e-06,
+      "loss": 0.3155,
+      "step": 2628
+    },
+    {
+      "epoch": 2.782010582010582,
+      "grad_norm": 0.19681762572714664,
+      "learning_rate": 4.057232457859663e-06,
+      "loss": 0.3781,
+      "step": 2629
+    },
+    {
+      "epoch": 2.7830687830687832,
+      "grad_norm": 0.2038735269817466,
+      "learning_rate": 4.037632301058409e-06,
+      "loss": 0.3807,
+      "step": 2630
+    },
+    {
+      "epoch": 2.784126984126984,
+      "grad_norm": 0.19777937024940856,
+      "learning_rate": 4.018032144257154e-06,
+      "loss": 0.3569,
+      "step": 2631
+    },
+    {
+      "epoch": 2.785185185185185,
+      "grad_norm": 0.1995201890239187,
+      "learning_rate": 3.9984319874559e-06,
+      "loss": 0.3701,
+      "step": 2632
+    },
+    {
+      "epoch": 2.7862433862433864,
+      "grad_norm": 0.20317171069691647,
+      "learning_rate": 3.978831830654645e-06,
+      "loss": 0.3569,
+      "step": 2633
+    },
+    {
+      "epoch": 2.787301587301587,
+      "grad_norm": 0.19068734602892104,
+      "learning_rate": 3.959231673853391e-06,
+      "loss": 0.3366,
+      "step": 2634
+    },
+    {
+      "epoch": 2.7883597883597884,
+      "grad_norm": 0.1828626835621772,
+      "learning_rate": 3.939631517052137e-06,
+      "loss": 0.3116,
+      "step": 2635
+    },
+    {
+      "epoch": 2.7894179894179896,
+      "grad_norm": 0.19548905750250636,
+      "learning_rate": 3.920031360250883e-06,
+      "loss": 0.3472,
+      "step": 2636
+    },
+    {
+      "epoch": 2.7904761904761903,
+      "grad_norm": 0.19023591210044766,
+      "learning_rate": 3.900431203449628e-06,
+      "loss": 0.3046,
+      "step": 2637
+    },
+    {
+      "epoch": 2.7915343915343915,
+      "grad_norm": 0.24034345295992002,
+      "learning_rate": 3.880831046648374e-06,
+      "loss": 0.402,
+      "step": 2638
+    },
+    {
+      "epoch": 2.7925925925925927,
+      "grad_norm": 0.1934131985141875,
+      "learning_rate": 3.861230889847119e-06,
+      "loss": 0.3431,
+      "step": 2639
+    },
+    {
+      "epoch": 2.7936507936507935,
+      "grad_norm": 0.19510076751913827,
+      "learning_rate": 3.8416307330458645e-06,
+      "loss": 0.3473,
+      "step": 2640
+    },
+    {
+      "epoch": 2.7947089947089947,
+      "grad_norm": 0.20150653464491855,
+      "learning_rate": 3.82203057624461e-06,
+      "loss": 0.3688,
+      "step": 2641
+    },
+    {
+      "epoch": 2.795767195767196,
+      "grad_norm": 0.18941533891094486,
+      "learning_rate": 3.802430419443356e-06,
+      "loss": 0.3482,
+      "step": 2642
+    },
+    {
+      "epoch": 2.7968253968253967,
+      "grad_norm": 0.20133005930916215,
+      "learning_rate": 3.7828302626421017e-06,
+      "loss": 0.3549,
+      "step": 2643
+    },
+    {
+      "epoch": 2.797883597883598,
+      "grad_norm": 0.20525961667412126,
+      "learning_rate": 3.7632301058408467e-06,
+      "loss": 0.3616,
+      "step": 2644
+    },
+    {
+      "epoch": 2.798941798941799,
+      "grad_norm": 0.19156388319779546,
+      "learning_rate": 3.7436299490395926e-06,
+      "loss": 0.3417,
+      "step": 2645
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.2011256721429364,
+      "learning_rate": 3.724029792238338e-06,
+      "loss": 0.3601,
+      "step": 2646
+    },
+    {
+      "epoch": 2.801058201058201,
+      "grad_norm": 0.19619574219995886,
+      "learning_rate": 3.704429635437084e-06,
+      "loss": 0.3485,
+      "step": 2647
+    },
+    {
+      "epoch": 2.8021164021164022,
+      "grad_norm": 0.19499163232482158,
+      "learning_rate": 3.684829478635829e-06,
+      "loss": 0.3725,
+      "step": 2648
+    },
+    {
+      "epoch": 2.803174603174603,
+      "grad_norm": 0.19294474196059933,
+      "learning_rate": 3.665229321834575e-06,
+      "loss": 0.3555,
+      "step": 2649
+    },
+    {
+      "epoch": 2.804232804232804,
+      "grad_norm": 0.2032516666422444,
+      "learning_rate": 3.6456291650333207e-06,
+      "loss": 0.3661,
+      "step": 2650
+    },
+    {
+      "epoch": 2.8052910052910054,
+      "grad_norm": 0.2020734831226216,
+      "learning_rate": 3.626029008232066e-06,
+      "loss": 0.3556,
+      "step": 2651
+    },
+    {
+      "epoch": 2.806349206349206,
+      "grad_norm": 0.1916619404390491,
+      "learning_rate": 3.606428851430812e-06,
+      "loss": 0.3329,
+      "step": 2652
+    },
+    {
+      "epoch": 2.8074074074074074,
+      "grad_norm": 0.1952962673766634,
+      "learning_rate": 3.586828694629557e-06,
+      "loss": 0.3554,
+      "step": 2653
+    },
+    {
+      "epoch": 2.8084656084656086,
+      "grad_norm": 0.1962712621825182,
+      "learning_rate": 3.567228537828303e-06,
+      "loss": 0.3574,
+      "step": 2654
+    },
+    {
+      "epoch": 2.8095238095238093,
+      "grad_norm": 0.1844963945964023,
+      "learning_rate": 3.5476283810270484e-06,
+      "loss": 0.3231,
+      "step": 2655
+    },
+    {
+      "epoch": 2.8105820105820105,
+      "grad_norm": 0.18348101787536167,
+      "learning_rate": 3.5280282242257942e-06,
+      "loss": 0.3159,
+      "step": 2656
+    },
+    {
+      "epoch": 2.8116402116402117,
+      "grad_norm": 0.1849348467642649,
+      "learning_rate": 3.5084280674245393e-06,
+      "loss": 0.3231,
+      "step": 2657
+    },
+    {
+      "epoch": 2.8126984126984125,
+      "grad_norm": 0.18599313201602283,
+      "learning_rate": 3.488827910623285e-06,
+      "loss": 0.3381,
+      "step": 2658
+    },
+    {
+      "epoch": 2.8137566137566137,
+      "grad_norm": 0.18221416638152477,
+      "learning_rate": 3.469227753822031e-06,
+      "loss": 0.3136,
+      "step": 2659
+    },
+    {
+      "epoch": 2.814814814814815,
+      "grad_norm": 0.18255639179737698,
+      "learning_rate": 3.4496275970207765e-06,
+      "loss": 0.322,
+      "step": 2660
+    },
+    {
+      "epoch": 2.8158730158730156,
+      "grad_norm": 0.2095837906111075,
+      "learning_rate": 3.4300274402195223e-06,
+      "loss": 0.3377,
+      "step": 2661
+    },
+    {
+      "epoch": 2.816931216931217,
+      "grad_norm": 0.18665268042560523,
+      "learning_rate": 3.4104272834182674e-06,
+      "loss": 0.3431,
+      "step": 2662
+    },
+    {
+      "epoch": 2.817989417989418,
+      "grad_norm": 0.19002223272665303,
+      "learning_rate": 3.3908271266170132e-06,
+      "loss": 0.3342,
+      "step": 2663
+    },
+    {
+      "epoch": 2.819047619047619,
+      "grad_norm": 0.18705897801580285,
+      "learning_rate": 3.3712269698157587e-06,
+      "loss": 0.3181,
+      "step": 2664
+    },
+    {
+      "epoch": 2.82010582010582,
+      "grad_norm": 0.20862688975705004,
+      "learning_rate": 3.3516268130145046e-06,
+      "loss": 0.3668,
+      "step": 2665
+    },
+    {
+      "epoch": 2.821164021164021,
+      "grad_norm": 0.19225010946678575,
+      "learning_rate": 3.3320266562132496e-06,
+      "loss": 0.3359,
+      "step": 2666
+    },
+    {
+      "epoch": 2.822222222222222,
+      "grad_norm": 0.20161732029347787,
+      "learning_rate": 3.3124264994119955e-06,
+      "loss": 0.3367,
+      "step": 2667
+    },
+    {
+      "epoch": 2.823280423280423,
+      "grad_norm": 0.18828704734970103,
+      "learning_rate": 3.292826342610741e-06,
+      "loss": 0.3094,
+      "step": 2668
+    },
+    {
+      "epoch": 2.8243386243386244,
+      "grad_norm": 0.18924625950342489,
+      "learning_rate": 3.273226185809487e-06,
+      "loss": 0.3434,
+      "step": 2669
+    },
+    {
+      "epoch": 2.825396825396825,
+      "grad_norm": 0.18417340787076836,
+      "learning_rate": 3.2536260290082327e-06,
+      "loss": 0.3303,
+      "step": 2670
+    },
+    {
+      "epoch": 2.8264550264550263,
+      "grad_norm": 0.20379631500237214,
+      "learning_rate": 3.2340258722069777e-06,
+      "loss": 0.3635,
+      "step": 2671
+    },
+    {
+      "epoch": 2.8275132275132275,
+      "grad_norm": 0.19838887354654922,
+      "learning_rate": 3.2144257154057236e-06,
+      "loss": 0.3639,
+      "step": 2672
+    },
+    {
+      "epoch": 2.8285714285714287,
+      "grad_norm": 0.19794810913649097,
+      "learning_rate": 3.194825558604469e-06,
+      "loss": 0.3438,
+      "step": 2673
+    },
+    {
+      "epoch": 2.8296296296296295,
+      "grad_norm": 0.185510630825901,
+      "learning_rate": 3.175225401803215e-06,
+      "loss": 0.3467,
+      "step": 2674
+    },
+    {
+      "epoch": 2.8306878306878307,
+      "grad_norm": 0.192065774599022,
+      "learning_rate": 3.15562524500196e-06,
+      "loss": 0.3375,
+      "step": 2675
+    },
+    {
+      "epoch": 2.831746031746032,
+      "grad_norm": 0.19435756312714894,
+      "learning_rate": 3.136025088200706e-06,
+      "loss": 0.3493,
+      "step": 2676
+    },
+    {
+      "epoch": 2.8328042328042327,
+      "grad_norm": 0.2009853526245991,
+      "learning_rate": 3.1164249313994513e-06,
+      "loss": 0.3707,
+      "step": 2677
+    },
+    {
+      "epoch": 2.833862433862434,
+      "grad_norm": 0.1744316971987279,
+      "learning_rate": 3.096824774598197e-06,
+      "loss": 0.2686,
+      "step": 2678
+    },
+    {
+      "epoch": 2.834920634920635,
+      "grad_norm": 0.19758049236544187,
+      "learning_rate": 3.0772246177969426e-06,
+      "loss": 0.3461,
+      "step": 2679
+    },
+    {
+      "epoch": 2.835978835978836,
+      "grad_norm": 0.19599350952580494,
+      "learning_rate": 3.057624460995688e-06,
+      "loss": 0.3551,
+      "step": 2680
+    },
+    {
+      "epoch": 2.837037037037037,
+      "grad_norm": 0.18930967657748698,
+      "learning_rate": 3.0380243041944335e-06,
+      "loss": 0.3696,
+      "step": 2681
+    },
+    {
+      "epoch": 2.8380952380952382,
+      "grad_norm": 0.20202839534587627,
+      "learning_rate": 3.0184241473931794e-06,
+      "loss": 0.3899,
+      "step": 2682
+    },
+    {
+      "epoch": 2.8391534391534394,
+      "grad_norm": 0.1889193752419766,
+      "learning_rate": 2.998823990591925e-06,
+      "loss": 0.3159,
+      "step": 2683
+    },
+    {
+      "epoch": 2.84021164021164,
+      "grad_norm": 0.1911231246544652,
+      "learning_rate": 2.9792238337906707e-06,
+      "loss": 0.3272,
+      "step": 2684
+    },
+    {
+      "epoch": 2.8412698412698414,
+      "grad_norm": 0.20056244525036598,
+      "learning_rate": 2.959623676989416e-06,
+      "loss": 0.3516,
+      "step": 2685
+    },
+    {
+      "epoch": 2.8423280423280426,
+      "grad_norm": 0.19474804822394667,
+      "learning_rate": 2.9400235201881616e-06,
+      "loss": 0.3619,
+      "step": 2686
+    },
+    {
+      "epoch": 2.8433862433862434,
+      "grad_norm": 0.1922225375564539,
+      "learning_rate": 2.9204233633869075e-06,
+      "loss": 0.3583,
+      "step": 2687
+    },
+    {
+      "epoch": 2.8444444444444446,
+      "grad_norm": 0.2114409910111819,
+      "learning_rate": 2.900823206585653e-06,
+      "loss": 0.3864,
+      "step": 2688
+    },
+    {
+      "epoch": 2.8455026455026458,
+      "grad_norm": 0.18768776804906878,
+      "learning_rate": 2.8812230497843984e-06,
+      "loss": 0.3316,
+      "step": 2689
+    },
+    {
+      "epoch": 2.8465608465608465,
+      "grad_norm": 0.19630860306480566,
+      "learning_rate": 2.861622892983144e-06,
+      "loss": 0.3418,
+      "step": 2690
+    },
+    {
+      "epoch": 2.8476190476190477,
+      "grad_norm": 0.18278700681759866,
+      "learning_rate": 2.8420227361818897e-06,
+      "loss": 0.3305,
+      "step": 2691
+    },
+    {
+      "epoch": 2.848677248677249,
+      "grad_norm": 0.1828410519755495,
+      "learning_rate": 2.822422579380635e-06,
+      "loss": 0.3093,
+      "step": 2692
+    },
+    {
+      "epoch": 2.8497354497354497,
+      "grad_norm": 0.19191081583592245,
+      "learning_rate": 2.8028224225793806e-06,
+      "loss": 0.3178,
+      "step": 2693
+    },
+    {
+      "epoch": 2.850793650793651,
+      "grad_norm": 0.19137029288992774,
+      "learning_rate": 2.7832222657781265e-06,
+      "loss": 0.3494,
+      "step": 2694
+    },
+    {
+      "epoch": 2.851851851851852,
+      "grad_norm": 0.18943110594155188,
+      "learning_rate": 2.763622108976872e-06,
+      "loss": 0.346,
+      "step": 2695
+    },
+    {
+      "epoch": 2.852910052910053,
+      "grad_norm": 0.20066459093560637,
+      "learning_rate": 2.744021952175618e-06,
+      "loss": 0.3665,
+      "step": 2696
+    },
+    {
+      "epoch": 2.853968253968254,
+      "grad_norm": 0.20097973469627337,
+      "learning_rate": 2.7244217953743632e-06,
+      "loss": 0.3853,
+      "step": 2697
+    },
+    {
+      "epoch": 2.8550264550264552,
+      "grad_norm": 0.19008839219881032,
+      "learning_rate": 2.7048216385731087e-06,
+      "loss": 0.346,
+      "step": 2698
+    },
+    {
+      "epoch": 2.856084656084656,
+      "grad_norm": 0.17411703036920584,
+      "learning_rate": 2.685221481771854e-06,
+      "loss": 0.3064,
+      "step": 2699
+    },
+    {
+      "epoch": 2.857142857142857,
+      "grad_norm": 0.19049701322996757,
+      "learning_rate": 2.6656213249706e-06,
+      "loss": 0.3472,
+      "step": 2700
+    },
+    {
+      "epoch": 2.8582010582010584,
+      "grad_norm": 0.17636675029171775,
+      "learning_rate": 2.6460211681693455e-06,
+      "loss": 0.3013,
+      "step": 2701
+    },
+    {
+      "epoch": 2.859259259259259,
+      "grad_norm": 0.18713094779664108,
+      "learning_rate": 2.626421011368091e-06,
+      "loss": 0.332,
+      "step": 2702
+    },
+    {
+      "epoch": 2.8603174603174604,
+      "grad_norm": 0.1840880400812757,
+      "learning_rate": 2.6068208545668364e-06,
+      "loss": 0.3173,
+      "step": 2703
+    },
+    {
+      "epoch": 2.8613756613756616,
+      "grad_norm": 0.18365151688406114,
+      "learning_rate": 2.5872206977655822e-06,
+      "loss": 0.3351,
+      "step": 2704
+    },
+    {
+      "epoch": 2.8624338624338623,
+      "grad_norm": 0.19835551815444752,
+      "learning_rate": 2.567620540964328e-06,
+      "loss": 0.3591,
+      "step": 2705
+    },
+    {
+      "epoch": 2.8634920634920635,
+      "grad_norm": 0.1821324213120818,
+      "learning_rate": 2.5480203841630736e-06,
+      "loss": 0.3202,
+      "step": 2706
+    },
+    {
+      "epoch": 2.8645502645502647,
+      "grad_norm": 0.18706640843666805,
+      "learning_rate": 2.528420227361819e-06,
+      "loss": 0.3202,
+      "step": 2707
+    },
+    {
+      "epoch": 2.8656084656084655,
+      "grad_norm": 0.1890910955252759,
+      "learning_rate": 2.5088200705605645e-06,
+      "loss": 0.3466,
+      "step": 2708
+    },
+    {
+      "epoch": 2.8666666666666667,
+      "grad_norm": 0.19715376592862416,
+      "learning_rate": 2.4892199137593104e-06,
+      "loss": 0.3587,
+      "step": 2709
+    },
+    {
+      "epoch": 2.867724867724868,
+      "grad_norm": 0.1929758625937477,
+      "learning_rate": 2.469619756958056e-06,
+      "loss": 0.3364,
+      "step": 2710
+    },
+    {
+      "epoch": 2.8687830687830687,
+      "grad_norm": 0.19274028108990232,
+      "learning_rate": 2.4500196001568013e-06,
+      "loss": 0.3349,
+      "step": 2711
+    },
+    {
+      "epoch": 2.86984126984127,
+      "grad_norm": 0.18700906272661122,
+      "learning_rate": 2.4304194433555467e-06,
+      "loss": 0.3221,
+      "step": 2712
+    },
+    {
+      "epoch": 2.870899470899471,
+      "grad_norm": 0.1815493059168304,
+      "learning_rate": 2.4108192865542926e-06,
+      "loss": 0.3283,
+      "step": 2713
+    },
+    {
+      "epoch": 2.871957671957672,
+      "grad_norm": 0.19775169552678987,
+      "learning_rate": 2.3912191297530385e-06,
+      "loss": 0.3233,
+      "step": 2714
+    },
+    {
+      "epoch": 2.873015873015873,
+      "grad_norm": 0.2659234152993213,
+      "learning_rate": 2.371618972951784e-06,
+      "loss": 0.3772,
+      "step": 2715
+    },
+    {
+      "epoch": 2.8740740740740742,
+      "grad_norm": 0.2176181455659064,
+      "learning_rate": 2.3520188161505294e-06,
+      "loss": 0.3911,
+      "step": 2716
+    },
+    {
+      "epoch": 2.875132275132275,
+      "grad_norm": 0.17472137810214414,
+      "learning_rate": 2.332418659349275e-06,
+      "loss": 0.2846,
+      "step": 2717
+    },
+    {
+      "epoch": 2.876190476190476,
+      "grad_norm": 0.19723914254783917,
+      "learning_rate": 2.3128185025480207e-06,
+      "loss": 0.3504,
+      "step": 2718
+    },
+    {
+      "epoch": 2.8772486772486774,
+      "grad_norm": 0.18663011406982238,
+      "learning_rate": 2.293218345746766e-06,
+      "loss": 0.3262,
+      "step": 2719
+    },
+    {
+      "epoch": 2.878306878306878,
+      "grad_norm": 0.19137413050629865,
+      "learning_rate": 2.2736181889455116e-06,
+      "loss": 0.3648,
+      "step": 2720
+    },
+    {
+      "epoch": 2.8793650793650793,
+      "grad_norm": 0.19471470387541595,
+      "learning_rate": 2.254018032144257e-06,
+      "loss": 0.3447,
+      "step": 2721
+    },
+    {
+      "epoch": 2.8804232804232806,
+      "grad_norm": 0.18568099963810475,
+      "learning_rate": 2.234417875343003e-06,
+      "loss": 0.3104,
+      "step": 2722
+    },
+    {
+      "epoch": 2.8814814814814813,
+      "grad_norm": 0.17907029988352577,
+      "learning_rate": 2.2148177185417488e-06,
+      "loss": 0.3088,
+      "step": 2723
+    },
+    {
+      "epoch": 2.8825396825396825,
+      "grad_norm": 0.18879588257488686,
+      "learning_rate": 2.1952175617404942e-06,
+      "loss": 0.362,
+      "step": 2724
+    },
+    {
+      "epoch": 2.8835978835978837,
+      "grad_norm": 0.17652030997033405,
+      "learning_rate": 2.1756174049392397e-06,
+      "loss": 0.3012,
+      "step": 2725
+    },
+    {
+      "epoch": 2.8846560846560845,
+      "grad_norm": 0.1763918711495294,
+      "learning_rate": 2.156017248137985e-06,
+      "loss": 0.2995,
+      "step": 2726
+    },
+    {
+      "epoch": 2.8857142857142857,
+      "grad_norm": 0.2041375473388221,
+      "learning_rate": 2.136417091336731e-06,
+      "loss": 0.3903,
+      "step": 2727
+    },
+    {
+      "epoch": 2.886772486772487,
+      "grad_norm": 0.1799297516998756,
+      "learning_rate": 2.1168169345354765e-06,
+      "loss": 0.3152,
+      "step": 2728
+    },
+    {
+      "epoch": 2.8878306878306876,
+      "grad_norm": 0.1995448891297022,
+      "learning_rate": 2.097216777734222e-06,
+      "loss": 0.3903,
+      "step": 2729
+    },
+    {
+      "epoch": 2.888888888888889,
+      "grad_norm": 0.19215679672923563,
+      "learning_rate": 2.0776166209329674e-06,
+      "loss": 0.3628,
+      "step": 2730
+    },
+    {
+      "epoch": 2.88994708994709,
+      "grad_norm": 0.18610530187878963,
+      "learning_rate": 2.0580164641317132e-06,
+      "loss": 0.3251,
+      "step": 2731
+    },
+    {
+      "epoch": 2.891005291005291,
+      "grad_norm": 0.17806080573632393,
+      "learning_rate": 2.0384163073304587e-06,
+      "loss": 0.3056,
+      "step": 2732
+    },
+    {
+      "epoch": 2.892063492063492,
+      "grad_norm": 0.18795661792601714,
+      "learning_rate": 2.0188161505292046e-06,
+      "loss": 0.3396,
+      "step": 2733
+    },
+    {
+      "epoch": 2.893121693121693,
+      "grad_norm": 0.18063664445735603,
+      "learning_rate": 1.99921599372795e-06,
+      "loss": 0.2997,
+      "step": 2734
+    },
+    {
+      "epoch": 2.894179894179894,
+      "grad_norm": 0.18511841091215842,
+      "learning_rate": 1.9796158369266955e-06,
+      "loss": 0.3388,
+      "step": 2735
+    },
+    {
+      "epoch": 2.895238095238095,
+      "grad_norm": 0.2003527156439393,
+      "learning_rate": 1.9600156801254413e-06,
+      "loss": 0.3554,
+      "step": 2736
+    },
+    {
+      "epoch": 2.8962962962962964,
+      "grad_norm": 0.19842421312034014,
+      "learning_rate": 1.940415523324187e-06,
+      "loss": 0.3761,
+      "step": 2737
+    },
+    {
+      "epoch": 2.897354497354497,
+      "grad_norm": 0.2742847809137376,
+      "learning_rate": 1.9208153665229322e-06,
+      "loss": 0.374,
+      "step": 2738
+    },
+    {
+      "epoch": 2.8984126984126983,
+      "grad_norm": 0.19759325790955387,
+      "learning_rate": 1.901215209721678e-06,
+      "loss": 0.3458,
+      "step": 2739
+    },
+    {
+      "epoch": 2.8994708994708995,
+      "grad_norm": 0.1949299864756493,
+      "learning_rate": 1.8816150529204234e-06,
+      "loss": 0.3851,
+      "step": 2740
+    },
+    {
+      "epoch": 2.9005291005291003,
+      "grad_norm": 0.18716277246541932,
+      "learning_rate": 1.862014896119169e-06,
+      "loss": 0.3562,
+      "step": 2741
+    },
+    {
+      "epoch": 2.9015873015873015,
+      "grad_norm": 0.18868051576949735,
+      "learning_rate": 1.8424147393179145e-06,
+      "loss": 0.3559,
+      "step": 2742
+    },
+    {
+      "epoch": 2.9026455026455027,
+      "grad_norm": 0.1992670631408877,
+      "learning_rate": 1.8228145825166603e-06,
+      "loss": 0.3953,
+      "step": 2743
+    },
+    {
+      "epoch": 2.9037037037037035,
+      "grad_norm": 0.190011431296961,
+      "learning_rate": 1.803214425715406e-06,
+      "loss": 0.3517,
+      "step": 2744
+    },
+    {
+      "epoch": 2.9047619047619047,
+      "grad_norm": 0.1898685284634927,
+      "learning_rate": 1.7836142689141515e-06,
+      "loss": 0.324,
+      "step": 2745
+    },
+    {
+      "epoch": 2.905820105820106,
+      "grad_norm": 0.20798301264228833,
+      "learning_rate": 1.7640141121128971e-06,
+      "loss": 0.4022,
+      "step": 2746
+    },
+    {
+      "epoch": 2.9068783068783066,
+      "grad_norm": 0.1884618131179482,
+      "learning_rate": 1.7444139553116426e-06,
+      "loss": 0.3162,
+      "step": 2747
+    },
+    {
+      "epoch": 2.907936507936508,
+      "grad_norm": 0.2062849190831743,
+      "learning_rate": 1.7248137985103882e-06,
+      "loss": 0.3713,
+      "step": 2748
+    },
+    {
+      "epoch": 2.908994708994709,
+      "grad_norm": 0.21572426764959662,
+      "learning_rate": 1.7052136417091337e-06,
+      "loss": 0.3902,
+      "step": 2749
+    },
+    {
+      "epoch": 2.91005291005291,
+      "grad_norm": 0.29553314531980623,
+      "learning_rate": 1.6856134849078794e-06,
+      "loss": 0.3407,
+      "step": 2750
+    },
+    {
+      "epoch": 2.911111111111111,
+      "grad_norm": 0.20393510044120217,
+      "learning_rate": 1.6660133281066248e-06,
+      "loss": 0.4114,
+      "step": 2751
+    },
+    {
+      "epoch": 2.912169312169312,
+      "grad_norm": 0.16933792741507034,
+      "learning_rate": 1.6464131713053705e-06,
+      "loss": 0.3066,
+      "step": 2752
+    },
+    {
+      "epoch": 2.9132275132275134,
+      "grad_norm": 0.1996783103654981,
+      "learning_rate": 1.6268130145041163e-06,
+      "loss": 0.3649,
+      "step": 2753
+    },
+    {
+      "epoch": 2.914285714285714,
+      "grad_norm": 0.1967458515577118,
+      "learning_rate": 1.6072128577028618e-06,
+      "loss": 0.3568,
+      "step": 2754
+    },
+    {
+      "epoch": 2.9153439153439153,
+      "grad_norm": 0.18448247900585343,
+      "learning_rate": 1.5876127009016075e-06,
+      "loss": 0.326,
+      "step": 2755
+    },
+    {
+      "epoch": 2.9164021164021166,
+      "grad_norm": 0.19461118793102933,
+      "learning_rate": 1.568012544100353e-06,
+      "loss": 0.3213,
+      "step": 2756
+    },
+    {
+      "epoch": 2.9174603174603173,
+      "grad_norm": 0.1899035298993076,
+      "learning_rate": 1.5484123872990986e-06,
+      "loss": 0.3387,
+      "step": 2757
+    },
+    {
+      "epoch": 2.9185185185185185,
+      "grad_norm": 0.1973718574120133,
+      "learning_rate": 1.528812230497844e-06,
+      "loss": 0.3617,
+      "step": 2758
+    },
+    {
+      "epoch": 2.9195767195767197,
+      "grad_norm": 0.29546988416171227,
+      "learning_rate": 1.5092120736965897e-06,
+      "loss": 0.338,
+      "step": 2759
+    },
+    {
+      "epoch": 2.9206349206349205,
+      "grad_norm": 0.18641514359807654,
+      "learning_rate": 1.4896119168953353e-06,
+      "loss": 0.3353,
+      "step": 2760
+    },
+    {
+      "epoch": 2.9216931216931217,
+      "grad_norm": 0.1986244733946501,
+      "learning_rate": 1.4700117600940808e-06,
+      "loss": 0.3652,
+      "step": 2761
+    },
+    {
+      "epoch": 2.922751322751323,
+      "grad_norm": 0.1840275855494628,
+      "learning_rate": 1.4504116032928265e-06,
+      "loss": 0.34,
+      "step": 2762
+    },
+    {
+      "epoch": 2.923809523809524,
+      "grad_norm": 0.1955588620037554,
+      "learning_rate": 1.430811446491572e-06,
+      "loss": 0.3592,
+      "step": 2763
+    },
+    {
+      "epoch": 2.924867724867725,
+      "grad_norm": 0.19691689936476783,
+      "learning_rate": 1.4112112896903176e-06,
+      "loss": 0.377,
+      "step": 2764
+    },
+    {
+      "epoch": 2.925925925925926,
+      "grad_norm": 0.18009646768548268,
+      "learning_rate": 1.3916111328890632e-06,
+      "loss": 0.3205,
+      "step": 2765
+    },
+    {
+      "epoch": 2.9269841269841272,
+      "grad_norm": 0.1898019017223925,
+      "learning_rate": 1.372010976087809e-06,
+      "loss": 0.3549,
+      "step": 2766
+    },
+    {
+      "epoch": 2.928042328042328,
+      "grad_norm": 0.19704920070535306,
+      "learning_rate": 1.3524108192865543e-06,
+      "loss": 0.3775,
+      "step": 2767
+    },
+    {
+      "epoch": 2.929100529100529,
+      "grad_norm": 0.18705440894929945,
+      "learning_rate": 1.3328106624853e-06,
+      "loss": 0.3265,
+      "step": 2768
+    },
+    {
+      "epoch": 2.9301587301587304,
+      "grad_norm": 0.16473519945175205,
+      "learning_rate": 1.3132105056840455e-06,
+      "loss": 0.2888,
+      "step": 2769
+    },
+    {
+      "epoch": 2.931216931216931,
+      "grad_norm": 0.19655934519795515,
+      "learning_rate": 1.2936103488827911e-06,
+      "loss": 0.3502,
+      "step": 2770
+    },
+    {
+      "epoch": 2.9322751322751324,
+      "grad_norm": 0.19677250020354922,
+      "learning_rate": 1.2740101920815368e-06,
+      "loss": 0.377,
+      "step": 2771
+    },
+    {
+      "epoch": 2.9333333333333336,
+      "grad_norm": 0.1937023595690031,
+      "learning_rate": 1.2544100352802822e-06,
+      "loss": 0.3376,
+      "step": 2772
+    },
+    {
+      "epoch": 2.9343915343915343,
+      "grad_norm": 0.1884504777247503,
+      "learning_rate": 1.234809878479028e-06,
+      "loss": 0.3534,
+      "step": 2773
+    },
+    {
+      "epoch": 2.9354497354497355,
+      "grad_norm": 0.1807776339927901,
+      "learning_rate": 1.2152097216777734e-06,
+      "loss": 0.3001,
+      "step": 2774
+    },
+    {
+      "epoch": 2.9365079365079367,
+      "grad_norm": 0.19494893044097508,
+      "learning_rate": 1.1956095648765192e-06,
+      "loss": 0.3601,
+      "step": 2775
+    },
+    {
+      "epoch": 2.9375661375661375,
+      "grad_norm": 0.18173859837169326,
+      "learning_rate": 1.1760094080752647e-06,
+      "loss": 0.3351,
+      "step": 2776
+    },
+    {
+      "epoch": 2.9386243386243387,
+      "grad_norm": 0.19155272417454372,
+      "learning_rate": 1.1564092512740103e-06,
+      "loss": 0.3246,
+      "step": 2777
+    },
+    {
+      "epoch": 2.93968253968254,
+      "grad_norm": 0.18097278381295773,
+      "learning_rate": 1.1368090944727558e-06,
+      "loss": 0.3325,
+      "step": 2778
+    },
+    {
+      "epoch": 2.9407407407407407,
+      "grad_norm": 0.18795150930344018,
+      "learning_rate": 1.1172089376715015e-06,
+      "loss": 0.3289,
+      "step": 2779
+    },
+    {
+      "epoch": 2.941798941798942,
+      "grad_norm": 0.19915958751004778,
+      "learning_rate": 1.0976087808702471e-06,
+      "loss": 0.3626,
+      "step": 2780
+    },
+    {
+      "epoch": 2.942857142857143,
+      "grad_norm": 0.21232105482687197,
+      "learning_rate": 1.0780086240689926e-06,
+      "loss": 0.4117,
+      "step": 2781
+    },
+    {
+      "epoch": 2.943915343915344,
+      "grad_norm": 0.1864031377940111,
+      "learning_rate": 1.0584084672677382e-06,
+      "loss": 0.3617,
+      "step": 2782
+    },
+    {
+      "epoch": 2.944973544973545,
+      "grad_norm": 0.20266505522326464,
+      "learning_rate": 1.0388083104664837e-06,
+      "loss": 0.4006,
+      "step": 2783
+    },
+    {
+      "epoch": 2.9460317460317462,
+      "grad_norm": 0.17805463329627502,
+      "learning_rate": 1.0192081536652293e-06,
+      "loss": 0.3185,
+      "step": 2784
+    },
+    {
+      "epoch": 2.947089947089947,
+      "grad_norm": 0.1865921965755665,
+      "learning_rate": 9.99607996863975e-07,
+      "loss": 0.3616,
+      "step": 2785
+    },
+    {
+      "epoch": 2.948148148148148,
+      "grad_norm": 0.19565848130960645,
+      "learning_rate": 9.800078400627207e-07,
+      "loss": 0.3565,
+      "step": 2786
+    },
+    {
+      "epoch": 2.9492063492063494,
+      "grad_norm": 0.19030143596622032,
+      "learning_rate": 9.604076832614661e-07,
+      "loss": 0.3379,
+      "step": 2787
+    },
+    {
+      "epoch": 2.95026455026455,
+      "grad_norm": 0.2078056379962277,
+      "learning_rate": 9.408075264602117e-07,
+      "loss": 0.3939,
+      "step": 2788
+    },
+    {
+      "epoch": 2.9513227513227513,
+      "grad_norm": 0.18813468584278845,
+      "learning_rate": 9.212073696589572e-07,
+      "loss": 0.3346,
+      "step": 2789
+    },
+    {
+      "epoch": 2.9523809523809526,
+      "grad_norm": 0.17678708921482536,
+      "learning_rate": 9.01607212857703e-07,
+      "loss": 0.3052,
+      "step": 2790
+    },
+    {
+      "epoch": 2.9534391534391533,
+      "grad_norm": 0.18575208703680052,
+      "learning_rate": 8.820070560564486e-07,
+      "loss": 0.323,
+      "step": 2791
+    },
+    {
+      "epoch": 2.9544973544973545,
+      "grad_norm": 0.18700053280637205,
+      "learning_rate": 8.624068992551941e-07,
+      "loss": 0.3312,
+      "step": 2792
+    },
+    {
+      "epoch": 2.9555555555555557,
+      "grad_norm": 0.1947341580545555,
+      "learning_rate": 8.428067424539397e-07,
+      "loss": 0.3652,
+      "step": 2793
+    },
+    {
+      "epoch": 2.9566137566137565,
+      "grad_norm": 0.18336157333209027,
+      "learning_rate": 8.232065856526852e-07,
+      "loss": 0.299,
+      "step": 2794
+    },
+    {
+      "epoch": 2.9576719576719577,
+      "grad_norm": 0.1989438129617965,
+      "learning_rate": 8.036064288514309e-07,
+      "loss": 0.3758,
+      "step": 2795
+    },
+    {
+      "epoch": 2.958730158730159,
+      "grad_norm": 0.1795010261425146,
+      "learning_rate": 7.840062720501765e-07,
+      "loss": 0.3095,
+      "step": 2796
+    },
+    {
+      "epoch": 2.9597883597883596,
+      "grad_norm": 0.1878025427152609,
+      "learning_rate": 7.64406115248922e-07,
+      "loss": 0.343,
+      "step": 2797
+    },
+    {
+      "epoch": 2.960846560846561,
+      "grad_norm": 0.1860910101745494,
+      "learning_rate": 7.448059584476677e-07,
+      "loss": 0.3434,
+      "step": 2798
+    },
+    {
+      "epoch": 2.961904761904762,
+      "grad_norm": 0.2080637929191924,
+      "learning_rate": 7.252058016464132e-07,
+      "loss": 0.3734,
+      "step": 2799
+    },
+    {
+      "epoch": 2.962962962962963,
+      "grad_norm": 0.19340393517672708,
+      "learning_rate": 7.056056448451588e-07,
+      "loss": 0.3631,
+      "step": 2800
+    },
+    {
+      "epoch": 2.964021164021164,
+      "grad_norm": 0.20271654196591155,
+      "learning_rate": 6.860054880439044e-07,
+      "loss": 0.3958,
+      "step": 2801
+    },
+    {
+      "epoch": 2.965079365079365,
+      "grad_norm": 0.19696852611219293,
+      "learning_rate": 6.6640533124265e-07,
+      "loss": 0.3719,
+      "step": 2802
+    },
+    {
+      "epoch": 2.966137566137566,
+      "grad_norm": 0.1884693790216164,
+      "learning_rate": 6.468051744413956e-07,
+      "loss": 0.3243,
+      "step": 2803
+    },
+    {
+      "epoch": 2.967195767195767,
+      "grad_norm": 0.1875853128029582,
+      "learning_rate": 6.272050176401411e-07,
+      "loss": 0.3405,
+      "step": 2804
+    },
+    {
+      "epoch": 2.9682539682539684,
+      "grad_norm": 0.19330130887895539,
+      "learning_rate": 6.076048608388867e-07,
+      "loss": 0.3742,
+      "step": 2805
+    },
+    {
+      "epoch": 2.969312169312169,
+      "grad_norm": 0.2007548359643975,
+      "learning_rate": 5.880047040376323e-07,
+      "loss": 0.3831,
+      "step": 2806
+    },
+    {
+      "epoch": 2.9703703703703703,
+      "grad_norm": 0.18320658631267467,
+      "learning_rate": 5.684045472363779e-07,
+      "loss": 0.3455,
+      "step": 2807
+    },
+    {
+      "epoch": 2.9714285714285715,
+      "grad_norm": 0.19352908453997061,
+      "learning_rate": 5.488043904351236e-07,
+      "loss": 0.3337,
+      "step": 2808
+    },
+    {
+      "epoch": 2.9724867724867723,
+      "grad_norm": 0.18917045289367823,
+      "learning_rate": 5.292042336338691e-07,
+      "loss": 0.3668,
+      "step": 2809
+    },
+    {
+      "epoch": 2.9735449735449735,
+      "grad_norm": 0.28331421422089914,
+      "learning_rate": 5.096040768326147e-07,
+      "loss": 0.3877,
+      "step": 2810
+    },
+    {
+      "epoch": 2.9746031746031747,
+      "grad_norm": 0.18730859428046542,
+      "learning_rate": 4.900039200313603e-07,
+      "loss": 0.3235,
+      "step": 2811
+    },
+    {
+      "epoch": 2.9756613756613755,
+      "grad_norm": 0.2061919680406255,
+      "learning_rate": 4.7040376323010584e-07,
+      "loss": 0.3877,
+      "step": 2812
+    },
+    {
+      "epoch": 2.9767195767195767,
+      "grad_norm": 0.17991475264599735,
+      "learning_rate": 4.508036064288515e-07,
+      "loss": 0.3117,
+      "step": 2813
+    },
+    {
+      "epoch": 2.977777777777778,
+      "grad_norm": 0.18979406930672754,
+      "learning_rate": 4.3120344962759706e-07,
+      "loss": 0.3523,
+      "step": 2814
+    },
+    {
+      "epoch": 2.9788359788359786,
+      "grad_norm": 0.20105316878907992,
+      "learning_rate": 4.116032928263426e-07,
+      "loss": 0.3547,
+      "step": 2815
+    },
+    {
+      "epoch": 2.97989417989418,
+      "grad_norm": 0.18428131124919112,
+      "learning_rate": 3.920031360250882e-07,
+      "loss": 0.323,
+      "step": 2816
+    },
+    {
+      "epoch": 2.980952380952381,
+      "grad_norm": 0.1812934812970745,
+      "learning_rate": 3.7240297922383384e-07,
+      "loss": 0.3181,
+      "step": 2817
+    },
+    {
+      "epoch": 2.982010582010582,
+      "grad_norm": 0.1863114001092859,
+      "learning_rate": 3.528028224225794e-07,
+      "loss": 0.3177,
+      "step": 2818
+    },
+    {
+      "epoch": 2.983068783068783,
+      "grad_norm": 0.18605121181705242,
+      "learning_rate": 3.33202665621325e-07,
+      "loss": 0.336,
+      "step": 2819
+    },
+    {
+      "epoch": 2.984126984126984,
+      "grad_norm": 0.20781903726047113,
+      "learning_rate": 3.1360250882007056e-07,
+      "loss": 0.4122,
+      "step": 2820
+    },
+    {
+      "epoch": 2.985185185185185,
+      "grad_norm": 0.2016418889379911,
+      "learning_rate": 2.9400235201881617e-07,
+      "loss": 0.3835,
+      "step": 2821
+    },
+    {
+      "epoch": 2.986243386243386,
+      "grad_norm": 0.17435892542662546,
+      "learning_rate": 2.744021952175618e-07,
+      "loss": 0.302,
+      "step": 2822
+    },
+    {
+      "epoch": 2.9873015873015873,
+      "grad_norm": 0.18917626577977661,
+      "learning_rate": 2.5480203841630734e-07,
+      "loss": 0.3654,
+      "step": 2823
+    },
+    {
+      "epoch": 2.988359788359788,
+      "grad_norm": 0.18622957441576268,
+      "learning_rate": 2.3520188161505292e-07,
+      "loss": 0.3388,
+      "step": 2824
+    },
+    {
+      "epoch": 2.9894179894179893,
+      "grad_norm": 0.1859174632852059,
+      "learning_rate": 2.1560172481379853e-07,
+      "loss": 0.3393,
+      "step": 2825
+    },
+    {
+      "epoch": 2.9904761904761905,
+      "grad_norm": 0.19065097631350006,
+      "learning_rate": 1.960015680125441e-07,
+      "loss": 0.3422,
+      "step": 2826
+    },
+    {
+      "epoch": 2.9915343915343913,
+      "grad_norm": 0.18272454868166982,
+      "learning_rate": 1.764014112112897e-07,
+      "loss": 0.3127,
+      "step": 2827
+    },
+    {
+      "epoch": 2.9925925925925925,
+      "grad_norm": 0.18725014042815322,
+      "learning_rate": 1.5680125441003528e-07,
+      "loss": 0.3275,
+      "step": 2828
+    },
+    {
+      "epoch": 2.9936507936507937,
+      "grad_norm": 0.17561038959172634,
+      "learning_rate": 1.372010976087809e-07,
+      "loss": 0.3066,
+      "step": 2829
+    },
+    {
+      "epoch": 2.9947089947089944,
+      "grad_norm": 0.18415272195204574,
+      "learning_rate": 1.1760094080752646e-07,
+      "loss": 0.3328,
+      "step": 2830
+    },
+    {
+      "epoch": 2.9957671957671956,
+      "grad_norm": 0.17845429273021535,
+      "learning_rate": 9.800078400627206e-08,
+      "loss": 0.2933,
+      "step": 2831
+    },
+    {
+      "epoch": 2.996825396825397,
+      "grad_norm": 0.1784414113949217,
+      "learning_rate": 7.840062720501764e-08,
+      "loss": 0.3129,
+      "step": 2832
+    },
+    {
+      "epoch": 2.997883597883598,
+      "grad_norm": 0.18326094077219415,
+      "learning_rate": 5.880047040376323e-08,
+      "loss": 0.339,
+      "step": 2833
+    },
+    {
+      "epoch": 2.998941798941799,
+      "grad_norm": 0.181202143145444,
+      "learning_rate": 3.920031360250882e-08,
+      "loss": 0.2975,
+      "step": 2834
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.1791412063210063,
+      "learning_rate": 1.960015680125441e-08,
+      "loss": 0.3422,
+      "step": 2835
+    },
+    {
+      "epoch": 3.0,
+      "step": 2835,
+      "total_flos": 3.152810456905679e+19,
+      "train_loss": 0.5287227819941451,
+      "train_runtime": 91700.2784,
+      "train_samples_per_second": 0.495,
+      "train_steps_per_second": 0.031
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2835,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "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": 3.152810456905679e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}