diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4794 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 680,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0014716703458425313,
+      "grad_norm": 4.058454990386963,
+      "learning_rate": 0.0,
+      "loss": 1.0378,
+      "step": 1
+    },
+    {
+      "epoch": 0.0029433406916850625,
+      "grad_norm": 4.680859565734863,
+      "learning_rate": 2.9411764705882356e-07,
+      "loss": 0.9258,
+      "step": 2
+    },
+    {
+      "epoch": 0.004415011037527594,
+      "grad_norm": 3.905587673187256,
+      "learning_rate": 5.882352941176471e-07,
+      "loss": 0.872,
+      "step": 3
+    },
+    {
+      "epoch": 0.005886681383370125,
+      "grad_norm": 4.3216776847839355,
+      "learning_rate": 8.823529411764707e-07,
+      "loss": 0.934,
+      "step": 4
+    },
+    {
+      "epoch": 0.007358351729212656,
+      "grad_norm": 4.033775329589844,
+      "learning_rate": 1.1764705882352942e-06,
+      "loss": 0.8904,
+      "step": 5
+    },
+    {
+      "epoch": 0.008830022075055188,
+      "grad_norm": 3.678755283355713,
+      "learning_rate": 1.4705882352941177e-06,
+      "loss": 0.9328,
+      "step": 6
+    },
+    {
+      "epoch": 0.010301692420897719,
+      "grad_norm": 3.711585521697998,
+      "learning_rate": 1.7647058823529414e-06,
+      "loss": 0.8471,
+      "step": 7
+    },
+    {
+      "epoch": 0.01177336276674025,
+      "grad_norm": 4.169286251068115,
+      "learning_rate": 2.058823529411765e-06,
+      "loss": 1.0045,
+      "step": 8
+    },
+    {
+      "epoch": 0.013245033112582781,
+      "grad_norm": 3.5065314769744873,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 0.8635,
+      "step": 9
+    },
+    {
+      "epoch": 0.014716703458425313,
+      "grad_norm": 3.752119302749634,
+      "learning_rate": 2.647058823529412e-06,
+      "loss": 0.9662,
+      "step": 10
+    },
+    {
+      "epoch": 0.016188373804267846,
+      "grad_norm": 3.2777113914489746,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 0.7967,
+      "step": 11
+    },
+    {
+      "epoch": 0.017660044150110375,
+      "grad_norm": 3.3518145084381104,
+      "learning_rate": 3.2352941176470594e-06,
+      "loss": 0.8465,
+      "step": 12
+    },
+    {
+      "epoch": 0.01913171449595291,
+      "grad_norm": 3.0732693672180176,
+      "learning_rate": 3.529411764705883e-06,
+      "loss": 0.8289,
+      "step": 13
+    },
+    {
+      "epoch": 0.020603384841795438,
+      "grad_norm": 2.9257853031158447,
+      "learning_rate": 3.8235294117647055e-06,
+      "loss": 0.8942,
+      "step": 14
+    },
+    {
+      "epoch": 0.02207505518763797,
+      "grad_norm": 2.98813533782959,
+      "learning_rate": 4.11764705882353e-06,
+      "loss": 0.9784,
+      "step": 15
+    },
+    {
+      "epoch": 0.0235467255334805,
+      "grad_norm": 3.698316812515259,
+      "learning_rate": 4.411764705882353e-06,
+      "loss": 1.0652,
+      "step": 16
+    },
+    {
+      "epoch": 0.025018395879323033,
+      "grad_norm": 3.1177361011505127,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 0.8983,
+      "step": 17
+    },
+    {
+      "epoch": 0.026490066225165563,
+      "grad_norm": 2.7175192832946777,
+      "learning_rate": 5e-06,
+      "loss": 0.8029,
+      "step": 18
+    },
+    {
+      "epoch": 0.027961736571008096,
+      "grad_norm": 2.897536039352417,
+      "learning_rate": 5.294117647058824e-06,
+      "loss": 0.8776,
+      "step": 19
+    },
+    {
+      "epoch": 0.029433406916850625,
+      "grad_norm": 2.9727461338043213,
+      "learning_rate": 5.588235294117647e-06,
+      "loss": 0.7791,
+      "step": 20
+    },
+    {
+      "epoch": 0.03090507726269316,
+      "grad_norm": 2.6092498302459717,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 0.7451,
+      "step": 21
+    },
+    {
+      "epoch": 0.03237674760853569,
+      "grad_norm": 2.913469076156616,
+      "learning_rate": 6.176470588235295e-06,
+      "loss": 0.9554,
+      "step": 22
+    },
+    {
+      "epoch": 0.03384841795437822,
+      "grad_norm": 2.8593599796295166,
+      "learning_rate": 6.470588235294119e-06,
+      "loss": 0.8882,
+      "step": 23
+    },
+    {
+      "epoch": 0.03532008830022075,
+      "grad_norm": 2.838365077972412,
+      "learning_rate": 6.764705882352942e-06,
+      "loss": 0.8554,
+      "step": 24
+    },
+    {
+      "epoch": 0.03679175864606328,
+      "grad_norm": 2.7695982456207275,
+      "learning_rate": 7.058823529411766e-06,
+      "loss": 0.8275,
+      "step": 25
+    },
+    {
+      "epoch": 0.03826342899190582,
+      "grad_norm": 2.9007530212402344,
+      "learning_rate": 7.352941176470589e-06,
+      "loss": 0.908,
+      "step": 26
+    },
+    {
+      "epoch": 0.039735099337748346,
+      "grad_norm": 2.676220178604126,
+      "learning_rate": 7.647058823529411e-06,
+      "loss": 0.9996,
+      "step": 27
+    },
+    {
+      "epoch": 0.041206769683590876,
+      "grad_norm": 2.886916160583496,
+      "learning_rate": 7.941176470588236e-06,
+      "loss": 0.8241,
+      "step": 28
+    },
+    {
+      "epoch": 0.042678440029433405,
+      "grad_norm": 2.9885916709899902,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 0.8173,
+      "step": 29
+    },
+    {
+      "epoch": 0.04415011037527594,
+      "grad_norm": 2.797389030456543,
+      "learning_rate": 8.529411764705883e-06,
+      "loss": 0.7265,
+      "step": 30
+    },
+    {
+      "epoch": 0.04562178072111847,
+      "grad_norm": 3.0593364238739014,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 0.8292,
+      "step": 31
+    },
+    {
+      "epoch": 0.047093451066961,
+      "grad_norm": 2.8509674072265625,
+      "learning_rate": 9.11764705882353e-06,
+      "loss": 0.7311,
+      "step": 32
+    },
+    {
+      "epoch": 0.04856512141280353,
+      "grad_norm": 3.2090744972229004,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 0.8574,
+      "step": 33
+    },
+    {
+      "epoch": 0.05003679175864607,
+      "grad_norm": 2.5766093730926514,
+      "learning_rate": 9.705882352941177e-06,
+      "loss": 0.8513,
+      "step": 34
+    },
+    {
+      "epoch": 0.051508462104488596,
+      "grad_norm": 3.0090904235839844,
+      "learning_rate": 1e-05,
+      "loss": 0.8444,
+      "step": 35
+    },
+    {
+      "epoch": 0.052980132450331126,
+      "grad_norm": 2.665348529815674,
+      "learning_rate": 9.999940874631278e-06,
+      "loss": 0.6617,
+      "step": 36
+    },
+    {
+      "epoch": 0.054451802796173655,
+      "grad_norm": 2.9271090030670166,
+      "learning_rate": 9.999763499923432e-06,
+      "loss": 0.808,
+      "step": 37
+    },
+    {
+      "epoch": 0.05592347314201619,
+      "grad_norm": 3.011321544647217,
+      "learning_rate": 9.999467880071402e-06,
+      "loss": 0.725,
+      "step": 38
+    },
+    {
+      "epoch": 0.05739514348785872,
+      "grad_norm": 3.2543551921844482,
+      "learning_rate": 9.999054022066643e-06,
+      "loss": 0.9299,
+      "step": 39
+    },
+    {
+      "epoch": 0.05886681383370125,
+      "grad_norm": 2.519449234008789,
+      "learning_rate": 9.998521935696953e-06,
+      "loss": 0.7564,
+      "step": 40
+    },
+    {
+      "epoch": 0.06033848417954378,
+      "grad_norm": 2.8531711101531982,
+      "learning_rate": 9.997871633546257e-06,
+      "loss": 0.87,
+      "step": 41
+    },
+    {
+      "epoch": 0.06181015452538632,
+      "grad_norm": 3.165675640106201,
+      "learning_rate": 9.997103130994295e-06,
+      "loss": 0.9265,
+      "step": 42
+    },
+    {
+      "epoch": 0.06328182487122884,
+      "grad_norm": 3.373549222946167,
+      "learning_rate": 9.996216446216267e-06,
+      "loss": 0.9618,
+      "step": 43
+    },
+    {
+      "epoch": 0.06475349521707138,
+      "grad_norm": 2.9908242225646973,
+      "learning_rate": 9.995211600182397e-06,
+      "loss": 0.8371,
+      "step": 44
+    },
+    {
+      "epoch": 0.06622516556291391,
+      "grad_norm": 2.482671022415161,
+      "learning_rate": 9.994088616657445e-06,
+      "loss": 0.7564,
+      "step": 45
+    },
+    {
+      "epoch": 0.06769683590875644,
+      "grad_norm": 3.005289316177368,
+      "learning_rate": 9.992847522200132e-06,
+      "loss": 0.9316,
+      "step": 46
+    },
+    {
+      "epoch": 0.06916850625459897,
+      "grad_norm": 2.875523805618286,
+      "learning_rate": 9.99148834616253e-06,
+      "loss": 0.7165,
+      "step": 47
+    },
+    {
+      "epoch": 0.0706401766004415,
+      "grad_norm": 2.9502875804901123,
+      "learning_rate": 9.990011120689352e-06,
+      "loss": 1.0304,
+      "step": 48
+    },
+    {
+      "epoch": 0.07211184694628403,
+      "grad_norm": 2.928637981414795,
+      "learning_rate": 9.988415880717195e-06,
+      "loss": 0.7055,
+      "step": 49
+    },
+    {
+      "epoch": 0.07358351729212656,
+      "grad_norm": 2.8092260360717773,
+      "learning_rate": 9.986702663973722e-06,
+      "loss": 0.7569,
+      "step": 50
+    },
+    {
+      "epoch": 0.07505518763796909,
+      "grad_norm": 2.9287312030792236,
+      "learning_rate": 9.98487151097676e-06,
+      "loss": 0.7917,
+      "step": 51
+    },
+    {
+      "epoch": 0.07652685798381163,
+      "grad_norm": 3.0313379764556885,
+      "learning_rate": 9.98292246503335e-06,
+      "loss": 0.88,
+      "step": 52
+    },
+    {
+      "epoch": 0.07799852832965416,
+      "grad_norm": 3.0787267684936523,
+      "learning_rate": 9.980855572238715e-06,
+      "loss": 0.7935,
+      "step": 53
+    },
+    {
+      "epoch": 0.07947019867549669,
+      "grad_norm": 2.687150716781616,
+      "learning_rate": 9.978670881475173e-06,
+      "loss": 0.7144,
+      "step": 54
+    },
+    {
+      "epoch": 0.08094186902133922,
+      "grad_norm": 2.613748073577881,
+      "learning_rate": 9.976368444410985e-06,
+      "loss": 0.7784,
+      "step": 55
+    },
+    {
+      "epoch": 0.08241353936718175,
+      "grad_norm": 2.7247557640075684,
+      "learning_rate": 9.973948315499126e-06,
+      "loss": 0.878,
+      "step": 56
+    },
+    {
+      "epoch": 0.08388520971302428,
+      "grad_norm": 2.8546860218048096,
+      "learning_rate": 9.971410551976001e-06,
+      "loss": 1.0515,
+      "step": 57
+    },
+    {
+      "epoch": 0.08535688005886681,
+      "grad_norm": 3.1416189670562744,
+      "learning_rate": 9.968755213860094e-06,
+      "loss": 0.913,
+      "step": 58
+    },
+    {
+      "epoch": 0.08682855040470934,
+      "grad_norm": 3.013488292694092,
+      "learning_rate": 9.96598236395054e-06,
+      "loss": 0.9151,
+      "step": 59
+    },
+    {
+      "epoch": 0.08830022075055188,
+      "grad_norm": 2.837449073791504,
+      "learning_rate": 9.963092067825651e-06,
+      "loss": 0.8935,
+      "step": 60
+    },
+    {
+      "epoch": 0.08977189109639441,
+      "grad_norm": 2.8535547256469727,
+      "learning_rate": 9.960084393841355e-06,
+      "loss": 0.8951,
+      "step": 61
+    },
+    {
+      "epoch": 0.09124356144223694,
+      "grad_norm": 2.8073694705963135,
+      "learning_rate": 9.956959413129586e-06,
+      "loss": 0.9974,
+      "step": 62
+    },
+    {
+      "epoch": 0.09271523178807947,
+      "grad_norm": 2.9127821922302246,
+      "learning_rate": 9.953717199596598e-06,
+      "loss": 0.9056,
+      "step": 63
+    },
+    {
+      "epoch": 0.094186902133922,
+      "grad_norm": 2.8197975158691406,
+      "learning_rate": 9.95035782992122e-06,
+      "loss": 0.7322,
+      "step": 64
+    },
+    {
+      "epoch": 0.09565857247976453,
+      "grad_norm": 2.5801289081573486,
+      "learning_rate": 9.94688138355304e-06,
+      "loss": 0.7477,
+      "step": 65
+    },
+    {
+      "epoch": 0.09713024282560706,
+      "grad_norm": 2.6256613731384277,
+      "learning_rate": 9.943287942710527e-06,
+      "loss": 0.7042,
+      "step": 66
+    },
+    {
+      "epoch": 0.09860191317144959,
+      "grad_norm": 3.3652524948120117,
+      "learning_rate": 9.939577592379088e-06,
+      "loss": 0.888,
+      "step": 67
+    },
+    {
+      "epoch": 0.10007358351729213,
+      "grad_norm": 3.101560354232788,
+      "learning_rate": 9.935750420309055e-06,
+      "loss": 0.8977,
+      "step": 68
+    },
+    {
+      "epoch": 0.10154525386313466,
+      "grad_norm": 2.5174732208251953,
+      "learning_rate": 9.931806517013612e-06,
+      "loss": 0.6729,
+      "step": 69
+    },
+    {
+      "epoch": 0.10301692420897719,
+      "grad_norm": 2.921755313873291,
+      "learning_rate": 9.927745975766654e-06,
+      "loss": 0.7846,
+      "step": 70
+    },
+    {
+      "epoch": 0.10448859455481972,
+      "grad_norm": 2.7532265186309814,
+      "learning_rate": 9.923568892600579e-06,
+      "loss": 0.9492,
+      "step": 71
+    },
+    {
+      "epoch": 0.10596026490066225,
+      "grad_norm": 2.8316383361816406,
+      "learning_rate": 9.919275366304021e-06,
+      "loss": 0.689,
+      "step": 72
+    },
+    {
+      "epoch": 0.10743193524650478,
+      "grad_norm": 2.870181083679199,
+      "learning_rate": 9.91486549841951e-06,
+      "loss": 0.6709,
+      "step": 73
+    },
+    {
+      "epoch": 0.10890360559234731,
+      "grad_norm": 2.7353837490081787,
+      "learning_rate": 9.91033939324107e-06,
+      "loss": 0.7755,
+      "step": 74
+    },
+    {
+      "epoch": 0.11037527593818984,
+      "grad_norm": 3.06223726272583,
+      "learning_rate": 9.905697157811761e-06,
+      "loss": 0.8336,
+      "step": 75
+    },
+    {
+      "epoch": 0.11184694628403238,
+      "grad_norm": 3.273441791534424,
+      "learning_rate": 9.90093890192113e-06,
+      "loss": 0.9732,
+      "step": 76
+    },
+    {
+      "epoch": 0.11331861662987491,
+      "grad_norm": 3.0267012119293213,
+      "learning_rate": 9.896064738102635e-06,
+      "loss": 0.8882,
+      "step": 77
+    },
+    {
+      "epoch": 0.11479028697571744,
+      "grad_norm": 2.8442041873931885,
+      "learning_rate": 9.891074781630967e-06,
+      "loss": 0.6856,
+      "step": 78
+    },
+    {
+      "epoch": 0.11626195732155997,
+      "grad_norm": 2.679326057434082,
+      "learning_rate": 9.885969150519332e-06,
+      "loss": 0.8473,
+      "step": 79
+    },
+    {
+      "epoch": 0.1177336276674025,
+      "grad_norm": 3.2508740425109863,
+      "learning_rate": 9.88074796551666e-06,
+      "loss": 0.8845,
+      "step": 80
+    },
+    {
+      "epoch": 0.11920529801324503,
+      "grad_norm": 3.125309944152832,
+      "learning_rate": 9.875411350104745e-06,
+      "loss": 0.8611,
+      "step": 81
+    },
+    {
+      "epoch": 0.12067696835908756,
+      "grad_norm": 2.9621620178222656,
+      "learning_rate": 9.869959430495329e-06,
+      "loss": 0.8959,
+      "step": 82
+    },
+    {
+      "epoch": 0.12214863870493009,
+      "grad_norm": 2.8529794216156006,
+      "learning_rate": 9.864392335627118e-06,
+      "loss": 0.7757,
+      "step": 83
+    },
+    {
+      "epoch": 0.12362030905077263,
+      "grad_norm": 2.8269686698913574,
+      "learning_rate": 9.858710197162722e-06,
+      "loss": 0.9969,
+      "step": 84
+    },
+    {
+      "epoch": 0.12509197939661515,
+      "grad_norm": 2.6117818355560303,
+      "learning_rate": 9.852913149485556e-06,
+      "loss": 0.7997,
+      "step": 85
+    },
+    {
+      "epoch": 0.12656364974245768,
+      "grad_norm": 2.948822259902954,
+      "learning_rate": 9.847001329696653e-06,
+      "loss": 0.9769,
+      "step": 86
+    },
+    {
+      "epoch": 0.1280353200883002,
+      "grad_norm": 3.2538185119628906,
+      "learning_rate": 9.840974877611423e-06,
+      "loss": 0.8283,
+      "step": 87
+    },
+    {
+      "epoch": 0.12950699043414277,
+      "grad_norm": 2.8442130088806152,
+      "learning_rate": 9.834833935756345e-06,
+      "loss": 0.8773,
+      "step": 88
+    },
+    {
+      "epoch": 0.1309786607799853,
+      "grad_norm": 2.388075828552246,
+      "learning_rate": 9.8285786493656e-06,
+      "loss": 0.7164,
+      "step": 89
+    },
+    {
+      "epoch": 0.13245033112582782,
+      "grad_norm": 2.6131460666656494,
+      "learning_rate": 9.822209166377635e-06,
+      "loss": 0.7686,
+      "step": 90
+    },
+    {
+      "epoch": 0.13392200147167035,
+      "grad_norm": 2.897113800048828,
+      "learning_rate": 9.815725637431663e-06,
+      "loss": 0.8232,
+      "step": 91
+    },
+    {
+      "epoch": 0.13539367181751288,
+      "grad_norm": 2.9289119243621826,
+      "learning_rate": 9.809128215864096e-06,
+      "loss": 0.8337,
+      "step": 92
+    },
+    {
+      "epoch": 0.1368653421633554,
+      "grad_norm": 2.462143898010254,
+      "learning_rate": 9.80241705770493e-06,
+      "loss": 0.7856,
+      "step": 93
+    },
+    {
+      "epoch": 0.13833701250919794,
+      "grad_norm": 2.8170995712280273,
+      "learning_rate": 9.795592321674046e-06,
+      "loss": 0.8723,
+      "step": 94
+    },
+    {
+      "epoch": 0.13980868285504047,
+      "grad_norm": 2.745225667953491,
+      "learning_rate": 9.788654169177454e-06,
+      "loss": 0.8504,
+      "step": 95
+    },
+    {
+      "epoch": 0.141280353200883,
+      "grad_norm": 2.677945613861084,
+      "learning_rate": 9.781602764303488e-06,
+      "loss": 0.7281,
+      "step": 96
+    },
+    {
+      "epoch": 0.14275202354672553,
+      "grad_norm": 2.890650987625122,
+      "learning_rate": 9.77443827381891e-06,
+      "loss": 0.7535,
+      "step": 97
+    },
+    {
+      "epoch": 0.14422369389256806,
+      "grad_norm": 3.1169748306274414,
+      "learning_rate": 9.76716086716498e-06,
+      "loss": 0.8581,
+      "step": 98
+    },
+    {
+      "epoch": 0.1456953642384106,
+      "grad_norm": 2.939659357070923,
+      "learning_rate": 9.759770716453436e-06,
+      "loss": 0.7432,
+      "step": 99
+    },
+    {
+      "epoch": 0.14716703458425312,
+      "grad_norm": 2.7667946815490723,
+      "learning_rate": 9.752267996462435e-06,
+      "loss": 1.0398,
+      "step": 100
+    },
+    {
+      "epoch": 0.14863870493009565,
+      "grad_norm": 2.3578202724456787,
+      "learning_rate": 9.744652884632406e-06,
+      "loss": 0.7488,
+      "step": 101
+    },
+    {
+      "epoch": 0.15011037527593818,
+      "grad_norm": 2.921765089035034,
+      "learning_rate": 9.736925561061871e-06,
+      "loss": 0.8559,
+      "step": 102
+    },
+    {
+      "epoch": 0.1515820456217807,
+      "grad_norm": 3.2434988021850586,
+      "learning_rate": 9.729086208503174e-06,
+      "loss": 0.8243,
+      "step": 103
+    },
+    {
+      "epoch": 0.15305371596762327,
+      "grad_norm": 2.7646985054016113,
+      "learning_rate": 9.721135012358156e-06,
+      "loss": 0.7898,
+      "step": 104
+    },
+    {
+      "epoch": 0.1545253863134658,
+      "grad_norm": 2.821303606033325,
+      "learning_rate": 9.713072160673778e-06,
+      "loss": 0.7296,
+      "step": 105
+    },
+    {
+      "epoch": 0.15599705665930833,
+      "grad_norm": 2.873602867126465,
+      "learning_rate": 9.704897844137674e-06,
+      "loss": 1.0217,
+      "step": 106
+    },
+    {
+      "epoch": 0.15746872700515085,
+      "grad_norm": 2.6251001358032227,
+      "learning_rate": 9.696612256073634e-06,
+      "loss": 0.7348,
+      "step": 107
+    },
+    {
+      "epoch": 0.15894039735099338,
+      "grad_norm": 3.296396255493164,
+      "learning_rate": 9.68821559243704e-06,
+      "loss": 0.9856,
+      "step": 108
+    },
+    {
+      "epoch": 0.1604120676968359,
+      "grad_norm": 2.97357439994812,
+      "learning_rate": 9.679708051810222e-06,
+      "loss": 0.8755,
+      "step": 109
+    },
+    {
+      "epoch": 0.16188373804267844,
+      "grad_norm": 2.6917476654052734,
+      "learning_rate": 9.671089835397772e-06,
+      "loss": 0.8255,
+      "step": 110
+    },
+    {
+      "epoch": 0.16335540838852097,
+      "grad_norm": 2.70802903175354,
+      "learning_rate": 9.66236114702178e-06,
+      "loss": 0.7903,
+      "step": 111
+    },
+    {
+      "epoch": 0.1648270787343635,
+      "grad_norm": 3.122664451599121,
+      "learning_rate": 9.653522193117014e-06,
+      "loss": 1.193,
+      "step": 112
+    },
+    {
+      "epoch": 0.16629874908020603,
+      "grad_norm": 2.777764081954956,
+      "learning_rate": 9.644573182726035e-06,
+      "loss": 0.8138,
+      "step": 113
+    },
+    {
+      "epoch": 0.16777041942604856,
+      "grad_norm": 2.6206493377685547,
+      "learning_rate": 9.63551432749426e-06,
+      "loss": 0.7676,
+      "step": 114
+    },
+    {
+      "epoch": 0.1692420897718911,
+      "grad_norm": 2.8496499061584473,
+      "learning_rate": 9.626345841664953e-06,
+      "loss": 0.7495,
+      "step": 115
+    },
+    {
+      "epoch": 0.17071376011773362,
+      "grad_norm": 2.744779586791992,
+      "learning_rate": 9.617067942074155e-06,
+      "loss": 0.8713,
+      "step": 116
+    },
+    {
+      "epoch": 0.17218543046357615,
+      "grad_norm": 2.4520657062530518,
+      "learning_rate": 9.607680848145557e-06,
+      "loss": 0.6581,
+      "step": 117
+    },
+    {
+      "epoch": 0.17365710080941868,
+      "grad_norm": 3.1236605644226074,
+      "learning_rate": 9.59818478188532e-06,
+      "loss": 0.8834,
+      "step": 118
+    },
+    {
+      "epoch": 0.1751287711552612,
+      "grad_norm": 3.0055110454559326,
+      "learning_rate": 9.588579967876806e-06,
+      "loss": 0.8016,
+      "step": 119
+    },
+    {
+      "epoch": 0.17660044150110377,
+      "grad_norm": 2.935249090194702,
+      "learning_rate": 9.578866633275289e-06,
+      "loss": 0.8539,
+      "step": 120
+    },
+    {
+      "epoch": 0.1780721118469463,
+      "grad_norm": 2.7148215770721436,
+      "learning_rate": 9.569045007802558e-06,
+      "loss": 0.7939,
+      "step": 121
+    },
+    {
+      "epoch": 0.17954378219278883,
+      "grad_norm": 3.0759003162384033,
+      "learning_rate": 9.55911532374151e-06,
+      "loss": 0.734,
+      "step": 122
+    },
+    {
+      "epoch": 0.18101545253863136,
+      "grad_norm": 2.664663553237915,
+      "learning_rate": 9.549077815930636e-06,
+      "loss": 0.9403,
+      "step": 123
+    },
+    {
+      "epoch": 0.18248712288447388,
+      "grad_norm": 2.898247241973877,
+      "learning_rate": 9.538932721758474e-06,
+      "loss": 0.7705,
+      "step": 124
+    },
+    {
+      "epoch": 0.18395879323031641,
+      "grad_norm": 2.6949071884155273,
+      "learning_rate": 9.528680281157999e-06,
+      "loss": 0.8279,
+      "step": 125
+    },
+    {
+      "epoch": 0.18543046357615894,
+      "grad_norm": 2.5309793949127197,
+      "learning_rate": 9.518320736600943e-06,
+      "loss": 0.7256,
+      "step": 126
+    },
+    {
+      "epoch": 0.18690213392200147,
+      "grad_norm": 2.75760817527771,
+      "learning_rate": 9.507854333092064e-06,
+      "loss": 0.7586,
+      "step": 127
+    },
+    {
+      "epoch": 0.188373804267844,
+      "grad_norm": 2.6482796669006348,
+      "learning_rate": 9.497281318163347e-06,
+      "loss": 0.7408,
+      "step": 128
+    },
+    {
+      "epoch": 0.18984547461368653,
+      "grad_norm": 3.0589354038238525,
+      "learning_rate": 9.486601941868155e-06,
+      "loss": 0.9713,
+      "step": 129
+    },
+    {
+      "epoch": 0.19131714495952906,
+      "grad_norm": 2.661060333251953,
+      "learning_rate": 9.475816456775313e-06,
+      "loss": 0.9774,
+      "step": 130
+    },
+    {
+      "epoch": 0.1927888153053716,
+      "grad_norm": 2.9241509437561035,
+      "learning_rate": 9.464925117963133e-06,
+      "loss": 0.9132,
+      "step": 131
+    },
+    {
+      "epoch": 0.19426048565121412,
+      "grad_norm": 2.5484488010406494,
+      "learning_rate": 9.453928183013385e-06,
+      "loss": 0.7035,
+      "step": 132
+    },
+    {
+      "epoch": 0.19573215599705665,
+      "grad_norm": 2.797853946685791,
+      "learning_rate": 9.442825912005203e-06,
+      "loss": 0.8403,
+      "step": 133
+    },
+    {
+      "epoch": 0.19720382634289918,
+      "grad_norm": 2.6880745887756348,
+      "learning_rate": 9.431618567508933e-06,
+      "loss": 0.7808,
+      "step": 134
+    },
+    {
+      "epoch": 0.1986754966887417,
+      "grad_norm": 3.096302032470703,
+      "learning_rate": 9.420306414579925e-06,
+      "loss": 0.7849,
+      "step": 135
+    },
+    {
+      "epoch": 0.20014716703458427,
+      "grad_norm": 2.483643054962158,
+      "learning_rate": 9.408889720752265e-06,
+      "loss": 0.6982,
+      "step": 136
+    },
+    {
+      "epoch": 0.2016188373804268,
+      "grad_norm": 2.664942741394043,
+      "learning_rate": 9.397368756032445e-06,
+      "loss": 0.8164,
+      "step": 137
+    },
+    {
+      "epoch": 0.20309050772626933,
+      "grad_norm": 2.9081645011901855,
+      "learning_rate": 9.385743792892983e-06,
+      "loss": 0.8065,
+      "step": 138
+    },
+    {
+      "epoch": 0.20456217807211186,
+      "grad_norm": 2.6566150188446045,
+      "learning_rate": 9.374015106265968e-06,
+      "loss": 0.6634,
+      "step": 139
+    },
+    {
+      "epoch": 0.20603384841795438,
+      "grad_norm": 2.695697069168091,
+      "learning_rate": 9.362182973536568e-06,
+      "loss": 0.7934,
+      "step": 140
+    },
+    {
+      "epoch": 0.20750551876379691,
+      "grad_norm": 2.575732469558716,
+      "learning_rate": 9.35024767453647e-06,
+      "loss": 0.8003,
+      "step": 141
+    },
+    {
+      "epoch": 0.20897718910963944,
+      "grad_norm": 2.4942550659179688,
+      "learning_rate": 9.338209491537257e-06,
+      "loss": 0.7969,
+      "step": 142
+    },
+    {
+      "epoch": 0.21044885945548197,
+      "grad_norm": 2.8870980739593506,
+      "learning_rate": 9.326068709243727e-06,
+      "loss": 1.0334,
+      "step": 143
+    },
+    {
+      "epoch": 0.2119205298013245,
+      "grad_norm": 2.619992971420288,
+      "learning_rate": 9.313825614787178e-06,
+      "loss": 0.7597,
+      "step": 144
+    },
+    {
+      "epoch": 0.21339220014716703,
+      "grad_norm": 2.5091888904571533,
+      "learning_rate": 9.301480497718594e-06,
+      "loss": 0.6728,
+      "step": 145
+    },
+    {
+      "epoch": 0.21486387049300956,
+      "grad_norm": 2.789379358291626,
+      "learning_rate": 9.289033650001817e-06,
+      "loss": 0.8862,
+      "step": 146
+    },
+    {
+      "epoch": 0.2163355408388521,
+      "grad_norm": 2.4175891876220703,
+      "learning_rate": 9.276485366006634e-06,
+      "loss": 0.7874,
+      "step": 147
+    },
+    {
+      "epoch": 0.21780721118469462,
+      "grad_norm": 2.8280205726623535,
+      "learning_rate": 9.263835942501807e-06,
+      "loss": 0.7887,
+      "step": 148
+    },
+    {
+      "epoch": 0.21927888153053715,
+      "grad_norm": 3.125808000564575,
+      "learning_rate": 9.251085678648072e-06,
+      "loss": 0.9522,
+      "step": 149
+    },
+    {
+      "epoch": 0.22075055187637968,
+      "grad_norm": 2.627514362335205,
+      "learning_rate": 9.238234875991048e-06,
+      "loss": 0.9322,
+      "step": 150
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 2.8124518394470215,
+      "learning_rate": 9.225283838454111e-06,
+      "loss": 0.7869,
+      "step": 151
+    },
+    {
+      "epoch": 0.22369389256806477,
+      "grad_norm": 3.086568593978882,
+      "learning_rate": 9.21223287233121e-06,
+      "loss": 0.9827,
+      "step": 152
+    },
+    {
+      "epoch": 0.2251655629139073,
+      "grad_norm": 2.854715347290039,
+      "learning_rate": 9.199082286279622e-06,
+      "loss": 0.7919,
+      "step": 153
+    },
+    {
+      "epoch": 0.22663723325974983,
+      "grad_norm": 2.5665128231048584,
+      "learning_rate": 9.185832391312644e-06,
+      "loss": 0.7764,
+      "step": 154
+    },
+    {
+      "epoch": 0.22810890360559236,
+      "grad_norm": 2.915379285812378,
+      "learning_rate": 9.172483500792246e-06,
+      "loss": 0.8804,
+      "step": 155
+    },
+    {
+      "epoch": 0.22958057395143489,
+      "grad_norm": 2.723245620727539,
+      "learning_rate": 9.159035930421658e-06,
+      "loss": 0.769,
+      "step": 156
+    },
+    {
+      "epoch": 0.23105224429727741,
+      "grad_norm": 3.021594285964966,
+      "learning_rate": 9.145489998237902e-06,
+      "loss": 0.8767,
+      "step": 157
+    },
+    {
+      "epoch": 0.23252391464311994,
+      "grad_norm": 2.540314197540283,
+      "learning_rate": 9.131846024604275e-06,
+      "loss": 0.7199,
+      "step": 158
+    },
+    {
+      "epoch": 0.23399558498896247,
+      "grad_norm": 2.591949462890625,
+      "learning_rate": 9.11810433220276e-06,
+      "loss": 0.8694,
+      "step": 159
+    },
+    {
+      "epoch": 0.235467255334805,
+      "grad_norm": 2.764286518096924,
+      "learning_rate": 9.104265246026414e-06,
+      "loss": 0.9038,
+      "step": 160
+    },
+    {
+      "epoch": 0.23693892568064753,
+      "grad_norm": 2.5182137489318848,
+      "learning_rate": 9.090329093371667e-06,
+      "loss": 0.7015,
+      "step": 161
+    },
+    {
+      "epoch": 0.23841059602649006,
+      "grad_norm": 2.872434616088867,
+      "learning_rate": 9.07629620383058e-06,
+      "loss": 0.8608,
+      "step": 162
+    },
+    {
+      "epoch": 0.2398822663723326,
+      "grad_norm": 2.6351943016052246,
+      "learning_rate": 9.062166909283062e-06,
+      "loss": 0.8356,
+      "step": 163
+    },
+    {
+      "epoch": 0.24135393671817512,
+      "grad_norm": 2.9412364959716797,
+      "learning_rate": 9.047941543889014e-06,
+      "loss": 0.9061,
+      "step": 164
+    },
+    {
+      "epoch": 0.24282560706401765,
+      "grad_norm": 2.7398324012756348,
+      "learning_rate": 9.033620444080427e-06,
+      "loss": 0.8046,
+      "step": 165
+    },
+    {
+      "epoch": 0.24429727740986018,
+      "grad_norm": 2.4704911708831787,
+      "learning_rate": 9.019203948553422e-06,
+      "loss": 0.6613,
+      "step": 166
+    },
+    {
+      "epoch": 0.2457689477557027,
+      "grad_norm": 2.9570794105529785,
+      "learning_rate": 9.004692398260243e-06,
+      "loss": 0.9917,
+      "step": 167
+    },
+    {
+      "epoch": 0.24724061810154527,
+      "grad_norm": 2.62516713142395,
+      "learning_rate": 8.990086136401199e-06,
+      "loss": 0.8395,
+      "step": 168
+    },
+    {
+      "epoch": 0.2487122884473878,
+      "grad_norm": 2.8056349754333496,
+      "learning_rate": 8.975385508416532e-06,
+      "loss": 0.7698,
+      "step": 169
+    },
+    {
+      "epoch": 0.2501839587932303,
+      "grad_norm": 2.83166241645813,
+      "learning_rate": 8.960590861978265e-06,
+      "loss": 0.6347,
+      "step": 170
+    },
+    {
+      "epoch": 0.25165562913907286,
+      "grad_norm": 2.616415500640869,
+      "learning_rate": 8.94570254698197e-06,
+      "loss": 0.8078,
+      "step": 171
+    },
+    {
+      "epoch": 0.25312729948491536,
+      "grad_norm": 2.677455186843872,
+      "learning_rate": 8.930720915538487e-06,
+      "loss": 0.8911,
+      "step": 172
+    },
+    {
+      "epoch": 0.2545989698307579,
+      "grad_norm": 3.1193928718566895,
+      "learning_rate": 8.915646321965615e-06,
+      "loss": 0.977,
+      "step": 173
+    },
+    {
+      "epoch": 0.2560706401766004,
+      "grad_norm": 5.6962971687316895,
+      "learning_rate": 8.900479122779712e-06,
+      "loss": 0.7694,
+      "step": 174
+    },
+    {
+      "epoch": 0.257542310522443,
+      "grad_norm": 3.3536412715911865,
+      "learning_rate": 8.885219676687277e-06,
+      "loss": 1.0121,
+      "step": 175
+    },
+    {
+      "epoch": 0.25901398086828553,
+      "grad_norm": 2.662569999694824,
+      "learning_rate": 8.86986834457646e-06,
+      "loss": 0.8054,
+      "step": 176
+    },
+    {
+      "epoch": 0.26048565121412803,
+      "grad_norm": 2.8296239376068115,
+      "learning_rate": 8.85442548950853e-06,
+      "loss": 0.832,
+      "step": 177
+    },
+    {
+      "epoch": 0.2619573215599706,
+      "grad_norm": 2.915475606918335,
+      "learning_rate": 8.838891476709289e-06,
+      "loss": 0.7431,
+      "step": 178
+    },
+    {
+      "epoch": 0.2634289919058131,
+      "grad_norm": 3.094879150390625,
+      "learning_rate": 8.823266673560426e-06,
+      "loss": 0.7247,
+      "step": 179
+    },
+    {
+      "epoch": 0.26490066225165565,
+      "grad_norm": 2.888617753982544,
+      "learning_rate": 8.807551449590846e-06,
+      "loss": 0.932,
+      "step": 180
+    },
+    {
+      "epoch": 0.26637233259749815,
+      "grad_norm": 2.9324824810028076,
+      "learning_rate": 8.791746176467908e-06,
+      "loss": 0.8315,
+      "step": 181
+    },
+    {
+      "epoch": 0.2678440029433407,
+      "grad_norm": 2.9685041904449463,
+      "learning_rate": 8.775851227988655e-06,
+      "loss": 0.84,
+      "step": 182
+    },
+    {
+      "epoch": 0.2693156732891832,
+      "grad_norm": 2.951185703277588,
+      "learning_rate": 8.759866980070963e-06,
+      "loss": 0.7088,
+      "step": 183
+    },
+    {
+      "epoch": 0.27078734363502577,
+      "grad_norm": 2.66342830657959,
+      "learning_rate": 8.743793810744655e-06,
+      "loss": 0.8356,
+      "step": 184
+    },
+    {
+      "epoch": 0.27225901398086827,
+      "grad_norm": 2.9835357666015625,
+      "learning_rate": 8.72763210014255e-06,
+      "loss": 0.8564,
+      "step": 185
+    },
+    {
+      "epoch": 0.2737306843267108,
+      "grad_norm": 2.647819757461548,
+      "learning_rate": 8.711382230491494e-06,
+      "loss": 0.7725,
+      "step": 186
+    },
+    {
+      "epoch": 0.27520235467255333,
+      "grad_norm": 2.6995792388916016,
+      "learning_rate": 8.695044586103297e-06,
+      "loss": 0.6293,
+      "step": 187
+    },
+    {
+      "epoch": 0.2766740250183959,
+      "grad_norm": 3.1014583110809326,
+      "learning_rate": 8.67861955336566e-06,
+      "loss": 0.805,
+      "step": 188
+    },
+    {
+      "epoch": 0.2781456953642384,
+      "grad_norm": 3.188678741455078,
+      "learning_rate": 8.662107520733027e-06,
+      "loss": 0.7317,
+      "step": 189
+    },
+    {
+      "epoch": 0.27961736571008095,
+      "grad_norm": 2.689643621444702,
+      "learning_rate": 8.64550887871741e-06,
+      "loss": 0.6776,
+      "step": 190
+    },
+    {
+      "epoch": 0.28108903605592345,
+      "grad_norm": 2.912175416946411,
+      "learning_rate": 8.628824019879137e-06,
+      "loss": 0.8686,
+      "step": 191
+    },
+    {
+      "epoch": 0.282560706401766,
+      "grad_norm": 2.829907178878784,
+      "learning_rate": 8.612053338817582e-06,
+      "loss": 0.7567,
+      "step": 192
+    },
+    {
+      "epoch": 0.28403237674760856,
+      "grad_norm": 3.0386757850646973,
+      "learning_rate": 8.595197232161824e-06,
+      "loss": 0.8411,
+      "step": 193
+    },
+    {
+      "epoch": 0.28550404709345106,
+      "grad_norm": 2.5806057453155518,
+      "learning_rate": 8.578256098561276e-06,
+      "loss": 0.7135,
+      "step": 194
+    },
+    {
+      "epoch": 0.2869757174392936,
+      "grad_norm": 2.757744789123535,
+      "learning_rate": 8.56123033867624e-06,
+      "loss": 0.7537,
+      "step": 195
+    },
+    {
+      "epoch": 0.2884473877851361,
+      "grad_norm": 2.6642186641693115,
+      "learning_rate": 8.544120355168451e-06,
+      "loss": 1.0471,
+      "step": 196
+    },
+    {
+      "epoch": 0.2899190581309787,
+      "grad_norm": 2.6267518997192383,
+      "learning_rate": 8.526926552691545e-06,
+      "loss": 0.777,
+      "step": 197
+    },
+    {
+      "epoch": 0.2913907284768212,
+      "grad_norm": 2.747035264968872,
+      "learning_rate": 8.509649337881483e-06,
+      "loss": 0.727,
+      "step": 198
+    },
+    {
+      "epoch": 0.29286239882266374,
+      "grad_norm": 2.958754539489746,
+      "learning_rate": 8.492289119346944e-06,
+      "loss": 0.7843,
+      "step": 199
+    },
+    {
+      "epoch": 0.29433406916850624,
+      "grad_norm": 2.807094097137451,
+      "learning_rate": 8.47484630765966e-06,
+      "loss": 0.9576,
+      "step": 200
+    },
+    {
+      "epoch": 0.2958057395143488,
+      "grad_norm": 2.713832139968872,
+      "learning_rate": 8.457321315344695e-06,
+      "loss": 0.9781,
+      "step": 201
+    },
+    {
+      "epoch": 0.2972774098601913,
+      "grad_norm": 3.011497974395752,
+      "learning_rate": 8.439714556870705e-06,
+      "loss": 0.8523,
+      "step": 202
+    },
+    {
+      "epoch": 0.29874908020603386,
+      "grad_norm": 3.302286386489868,
+      "learning_rate": 8.422026448640124e-06,
+      "loss": 0.8937,
+      "step": 203
+    },
+    {
+      "epoch": 0.30022075055187636,
+      "grad_norm": 2.5852441787719727,
+      "learning_rate": 8.404257408979322e-06,
+      "loss": 0.7258,
+      "step": 204
+    },
+    {
+      "epoch": 0.3016924208977189,
+      "grad_norm": 2.5731096267700195,
+      "learning_rate": 8.386407858128707e-06,
+      "loss": 0.7649,
+      "step": 205
+    },
+    {
+      "epoch": 0.3031640912435614,
+      "grad_norm": 2.4777395725250244,
+      "learning_rate": 8.368478218232787e-06,
+      "loss": 0.7454,
+      "step": 206
+    },
+    {
+      "epoch": 0.304635761589404,
+      "grad_norm": 2.754692554473877,
+      "learning_rate": 8.350468913330192e-06,
+      "loss": 0.8684,
+      "step": 207
+    },
+    {
+      "epoch": 0.30610743193524653,
+      "grad_norm": 2.4948291778564453,
+      "learning_rate": 8.33238036934364e-06,
+      "loss": 0.7159,
+      "step": 208
+    },
+    {
+      "epoch": 0.30757910228108903,
+      "grad_norm": 2.582134962081909,
+      "learning_rate": 8.31421301406986e-06,
+      "loss": 0.6708,
+      "step": 209
+    },
+    {
+      "epoch": 0.3090507726269316,
+      "grad_norm": 3.2686588764190674,
+      "learning_rate": 8.29596727716949e-06,
+      "loss": 0.9558,
+      "step": 210
+    },
+    {
+      "epoch": 0.3105224429727741,
+      "grad_norm": 2.770517110824585,
+      "learning_rate": 8.277643590156893e-06,
+      "loss": 0.8734,
+      "step": 211
+    },
+    {
+      "epoch": 0.31199411331861665,
+      "grad_norm": 2.6695854663848877,
+      "learning_rate": 8.259242386389975e-06,
+      "loss": 0.8677,
+      "step": 212
+    },
+    {
+      "epoch": 0.31346578366445915,
+      "grad_norm": 2.725999116897583,
+      "learning_rate": 8.240764101059913e-06,
+      "loss": 0.6776,
+      "step": 213
+    },
+    {
+      "epoch": 0.3149374540103017,
+      "grad_norm": 2.592602014541626,
+      "learning_rate": 8.222209171180883e-06,
+      "loss": 0.826,
+      "step": 214
+    },
+    {
+      "epoch": 0.3164091243561442,
+      "grad_norm": 2.528812885284424,
+      "learning_rate": 8.203578035579716e-06,
+      "loss": 0.7403,
+      "step": 215
+    },
+    {
+      "epoch": 0.31788079470198677,
+      "grad_norm": 2.584472894668579,
+      "learning_rate": 8.184871134885512e-06,
+      "loss": 0.6792,
+      "step": 216
+    },
+    {
+      "epoch": 0.31935246504782927,
+      "grad_norm": 2.6827452182769775,
+      "learning_rate": 8.166088911519236e-06,
+      "loss": 0.8447,
+      "step": 217
+    },
+    {
+      "epoch": 0.3208241353936718,
+      "grad_norm": 2.8367629051208496,
+      "learning_rate": 8.147231809683236e-06,
+      "loss": 0.7953,
+      "step": 218
+    },
+    {
+      "epoch": 0.32229580573951433,
+      "grad_norm": 2.950148820877075,
+      "learning_rate": 8.128300275350756e-06,
+      "loss": 0.8487,
+      "step": 219
+    },
+    {
+      "epoch": 0.3237674760853569,
+      "grad_norm": 2.701542854309082,
+      "learning_rate": 8.109294756255375e-06,
+      "loss": 0.7345,
+      "step": 220
+    },
+    {
+      "epoch": 0.3252391464311994,
+      "grad_norm": 2.7395551204681396,
+      "learning_rate": 8.090215701880418e-06,
+      "loss": 0.864,
+      "step": 221
+    },
+    {
+      "epoch": 0.32671081677704195,
+      "grad_norm": 2.761073350906372,
+      "learning_rate": 8.071063563448341e-06,
+      "loss": 0.7823,
+      "step": 222
+    },
+    {
+      "epoch": 0.32818248712288445,
+      "grad_norm": 2.613734006881714,
+      "learning_rate": 8.051838793910038e-06,
+      "loss": 0.6439,
+      "step": 223
+    },
+    {
+      "epoch": 0.329654157468727,
+      "grad_norm": 2.8382253646850586,
+      "learning_rate": 8.032541847934145e-06,
+      "loss": 0.7498,
+      "step": 224
+    },
+    {
+      "epoch": 0.33112582781456956,
+      "grad_norm": 2.5751657485961914,
+      "learning_rate": 8.013173181896283e-06,
+      "loss": 0.7132,
+      "step": 225
+    },
+    {
+      "epoch": 0.33259749816041206,
+      "grad_norm": 2.6825568675994873,
+      "learning_rate": 7.993733253868256e-06,
+      "loss": 0.8806,
+      "step": 226
+    },
+    {
+      "epoch": 0.3340691685062546,
+      "grad_norm": 2.8396565914154053,
+      "learning_rate": 7.974222523607236e-06,
+      "loss": 0.7735,
+      "step": 227
+    },
+    {
+      "epoch": 0.3355408388520971,
+      "grad_norm": 2.496772289276123,
+      "learning_rate": 7.954641452544864e-06,
+      "loss": 0.6571,
+      "step": 228
+    },
+    {
+      "epoch": 0.3370125091979397,
+      "grad_norm": 2.668145179748535,
+      "learning_rate": 7.934990503776363e-06,
+      "loss": 0.8631,
+      "step": 229
+    },
+    {
+      "epoch": 0.3384841795437822,
+      "grad_norm": 2.763549327850342,
+      "learning_rate": 7.915270142049566e-06,
+      "loss": 0.7821,
+      "step": 230
+    },
+    {
+      "epoch": 0.33995584988962474,
+      "grad_norm": 2.9257652759552,
+      "learning_rate": 7.895480833753942e-06,
+      "loss": 0.8285,
+      "step": 231
+    },
+    {
+      "epoch": 0.34142752023546724,
+      "grad_norm": 2.4478650093078613,
+      "learning_rate": 7.875623046909544e-06,
+      "loss": 0.7354,
+      "step": 232
+    },
+    {
+      "epoch": 0.3428991905813098,
+      "grad_norm": 2.6642887592315674,
+      "learning_rate": 7.855697251155967e-06,
+      "loss": 0.644,
+      "step": 233
+    },
+    {
+      "epoch": 0.3443708609271523,
+      "grad_norm": 2.843369960784912,
+      "learning_rate": 7.835703917741213e-06,
+      "loss": 0.9386,
+      "step": 234
+    },
+    {
+      "epoch": 0.34584253127299486,
+      "grad_norm": 3.1602954864501953,
+      "learning_rate": 7.815643519510571e-06,
+      "loss": 0.8077,
+      "step": 235
+    },
+    {
+      "epoch": 0.34731420161883736,
+      "grad_norm": 2.789182186126709,
+      "learning_rate": 7.795516530895414e-06,
+      "loss": 0.8464,
+      "step": 236
+    },
+    {
+      "epoch": 0.3487858719646799,
+      "grad_norm": 2.463135004043579,
+      "learning_rate": 7.775323427901993e-06,
+      "loss": 0.7189,
+      "step": 237
+    },
+    {
+      "epoch": 0.3502575423105224,
+      "grad_norm": 2.626633882522583,
+      "learning_rate": 7.75506468810017e-06,
+      "loss": 0.7559,
+      "step": 238
+    },
+    {
+      "epoch": 0.351729212656365,
+      "grad_norm": 2.4405670166015625,
+      "learning_rate": 7.734740790612137e-06,
+      "loss": 0.8049,
+      "step": 239
+    },
+    {
+      "epoch": 0.35320088300220753,
+      "grad_norm": 2.6157946586608887,
+      "learning_rate": 7.714352216101055e-06,
+      "loss": 0.7195,
+      "step": 240
+    },
+    {
+      "epoch": 0.35467255334805003,
+      "grad_norm": 2.6621780395507812,
+      "learning_rate": 7.693899446759727e-06,
+      "loss": 0.8505,
+      "step": 241
+    },
+    {
+      "epoch": 0.3561442236938926,
+      "grad_norm": 3.248483657836914,
+      "learning_rate": 7.673382966299163e-06,
+      "loss": 0.8385,
+      "step": 242
+    },
+    {
+      "epoch": 0.3576158940397351,
+      "grad_norm": 2.657336950302124,
+      "learning_rate": 7.65280325993715e-06,
+      "loss": 0.7362,
+      "step": 243
+    },
+    {
+      "epoch": 0.35908756438557765,
+      "grad_norm": 2.793397903442383,
+      "learning_rate": 7.63216081438678e-06,
+      "loss": 0.8892,
+      "step": 244
+    },
+    {
+      "epoch": 0.36055923473142015,
+      "grad_norm": 2.8806533813476562,
+      "learning_rate": 7.611456117844934e-06,
+      "loss": 0.7511,
+      "step": 245
+    },
+    {
+      "epoch": 0.3620309050772627,
+      "grad_norm": 2.9854753017425537,
+      "learning_rate": 7.59068965998074e-06,
+      "loss": 0.8891,
+      "step": 246
+    },
+    {
+      "epoch": 0.3635025754231052,
+      "grad_norm": 3.109421491622925,
+      "learning_rate": 7.569861931923989e-06,
+      "loss": 0.8743,
+      "step": 247
+    },
+    {
+      "epoch": 0.36497424576894777,
+      "grad_norm": 2.67236065864563,
+      "learning_rate": 7.548973426253521e-06,
+      "loss": 0.6766,
+      "step": 248
+    },
+    {
+      "epoch": 0.36644591611479027,
+      "grad_norm": 2.873359441757202,
+      "learning_rate": 7.528024636985575e-06,
+      "loss": 0.9396,
+      "step": 249
+    },
+    {
+      "epoch": 0.36791758646063283,
+      "grad_norm": 2.40339994430542,
+      "learning_rate": 7.507016059562107e-06,
+      "loss": 0.7551,
+      "step": 250
+    },
+    {
+      "epoch": 0.36938925680647533,
+      "grad_norm": 2.6877894401550293,
+      "learning_rate": 7.485948190839076e-06,
+      "loss": 0.7781,
+      "step": 251
+    },
+    {
+      "epoch": 0.3708609271523179,
+      "grad_norm": 2.939539670944214,
+      "learning_rate": 7.464821529074678e-06,
+      "loss": 0.7794,
+      "step": 252
+    },
+    {
+      "epoch": 0.3723325974981604,
+      "grad_norm": 2.9500153064727783,
+      "learning_rate": 7.443636573917585e-06,
+      "loss": 0.8368,
+      "step": 253
+    },
+    {
+      "epoch": 0.37380426784400295,
+      "grad_norm": 2.593169689178467,
+      "learning_rate": 7.4223938263951075e-06,
+      "loss": 0.806,
+      "step": 254
+    },
+    {
+      "epoch": 0.37527593818984545,
+      "grad_norm": 2.776139736175537,
+      "learning_rate": 7.40109378890136e-06,
+      "loss": 0.9303,
+      "step": 255
+    },
+    {
+      "epoch": 0.376747608535688,
+      "grad_norm": 2.4220099449157715,
+      "learning_rate": 7.379736965185369e-06,
+      "loss": 0.6452,
+      "step": 256
+    },
+    {
+      "epoch": 0.37821927888153056,
+      "grad_norm": 2.896829605102539,
+      "learning_rate": 7.358323860339165e-06,
+      "loss": 0.9025,
+      "step": 257
+    },
+    {
+      "epoch": 0.37969094922737306,
+      "grad_norm": 2.9651646614074707,
+      "learning_rate": 7.336854980785839e-06,
+      "loss": 0.7699,
+      "step": 258
+    },
+    {
+      "epoch": 0.3811626195732156,
+      "grad_norm": 2.693281650543213,
+      "learning_rate": 7.315330834267553e-06,
+      "loss": 0.7204,
+      "step": 259
+    },
+    {
+      "epoch": 0.3826342899190581,
+      "grad_norm": 2.8250482082366943,
+      "learning_rate": 7.293751929833553e-06,
+      "loss": 0.8387,
+      "step": 260
+    },
+    {
+      "epoch": 0.3841059602649007,
+      "grad_norm": 2.729510545730591,
+      "learning_rate": 7.272118777828109e-06,
+      "loss": 0.789,
+      "step": 261
+    },
+    {
+      "epoch": 0.3855776306107432,
+      "grad_norm": 3.190199613571167,
+      "learning_rate": 7.250431889878455e-06,
+      "loss": 0.735,
+      "step": 262
+    },
+    {
+      "epoch": 0.38704930095658574,
+      "grad_norm": 2.878964900970459,
+      "learning_rate": 7.2286917788826926e-06,
+      "loss": 0.6908,
+      "step": 263
+    },
+    {
+      "epoch": 0.38852097130242824,
+      "grad_norm": 2.516129970550537,
+      "learning_rate": 7.20689895899765e-06,
+      "loss": 0.6908,
+      "step": 264
+    },
+    {
+      "epoch": 0.3899926416482708,
+      "grad_norm": 2.970674991607666,
+      "learning_rate": 7.185053945626734e-06,
+      "loss": 0.8865,
+      "step": 265
+    },
+    {
+      "epoch": 0.3914643119941133,
+      "grad_norm": 2.7611653804779053,
+      "learning_rate": 7.163157255407732e-06,
+      "loss": 0.6888,
+      "step": 266
+    },
+    {
+      "epoch": 0.39293598233995586,
+      "grad_norm": 2.6418163776397705,
+      "learning_rate": 7.1412094062005985e-06,
+      "loss": 0.7193,
+      "step": 267
+    },
+    {
+      "epoch": 0.39440765268579836,
+      "grad_norm": 2.6448776721954346,
+      "learning_rate": 7.119210917075201e-06,
+      "loss": 0.6769,
+      "step": 268
+    },
+    {
+      "epoch": 0.3958793230316409,
+      "grad_norm": 3.3905608654022217,
+      "learning_rate": 7.097162308299055e-06,
+      "loss": 0.922,
+      "step": 269
+    },
+    {
+      "epoch": 0.3973509933774834,
+      "grad_norm": 2.664294481277466,
+      "learning_rate": 7.07506410132501e-06,
+      "loss": 0.7169,
+      "step": 270
+    },
+    {
+      "epoch": 0.398822663723326,
+      "grad_norm": 3.0975229740142822,
+      "learning_rate": 7.052916818778918e-06,
+      "loss": 0.8299,
+      "step": 271
+    },
+    {
+      "epoch": 0.40029433406916853,
+      "grad_norm": 2.8812222480773926,
+      "learning_rate": 7.030720984447279e-06,
+      "loss": 0.9557,
+      "step": 272
+    },
+    {
+      "epoch": 0.40176600441501104,
+      "grad_norm": 2.807002544403076,
+      "learning_rate": 7.008477123264849e-06,
+      "loss": 0.7065,
+      "step": 273
+    },
+    {
+      "epoch": 0.4032376747608536,
+      "grad_norm": 3.06130313873291,
+      "learning_rate": 6.986185761302224e-06,
+      "loss": 0.8652,
+      "step": 274
+    },
+    {
+      "epoch": 0.4047093451066961,
+      "grad_norm": 3.1718475818634033,
+      "learning_rate": 6.9638474257534025e-06,
+      "loss": 0.8435,
+      "step": 275
+    },
+    {
+      "epoch": 0.40618101545253865,
+      "grad_norm": 2.6698691844940186,
+      "learning_rate": 6.941462644923318e-06,
+      "loss": 0.7995,
+      "step": 276
+    },
+    {
+      "epoch": 0.40765268579838115,
+      "grad_norm": 2.624782085418701,
+      "learning_rate": 6.919031948215335e-06,
+      "loss": 0.7063,
+      "step": 277
+    },
+    {
+      "epoch": 0.4091243561442237,
+      "grad_norm": 3.034499406814575,
+      "learning_rate": 6.89655586611874e-06,
+      "loss": 0.8636,
+      "step": 278
+    },
+    {
+      "epoch": 0.4105960264900662,
+      "grad_norm": 2.70481538772583,
+      "learning_rate": 6.874034930196191e-06,
+      "loss": 0.8662,
+      "step": 279
+    },
+    {
+      "epoch": 0.41206769683590877,
+      "grad_norm": 2.350039005279541,
+      "learning_rate": 6.851469673071143e-06,
+      "loss": 0.6598,
+      "step": 280
+    },
+    {
+      "epoch": 0.41353936718175127,
+      "grad_norm": 2.4506044387817383,
+      "learning_rate": 6.8288606284152535e-06,
+      "loss": 0.6876,
+      "step": 281
+    },
+    {
+      "epoch": 0.41501103752759383,
+      "grad_norm": 2.7478151321411133,
+      "learning_rate": 6.806208330935766e-06,
+      "loss": 0.6849,
+      "step": 282
+    },
+    {
+      "epoch": 0.41648270787343633,
+      "grad_norm": 2.9902756214141846,
+      "learning_rate": 6.783513316362855e-06,
+      "loss": 0.8701,
+      "step": 283
+    },
+    {
+      "epoch": 0.4179543782192789,
+      "grad_norm": 3.0882210731506348,
+      "learning_rate": 6.760776121436963e-06,
+      "loss": 0.9867,
+      "step": 284
+    },
+    {
+      "epoch": 0.4194260485651214,
+      "grad_norm": 2.815107822418213,
+      "learning_rate": 6.737997283896104e-06,
+      "loss": 0.8299,
+      "step": 285
+    },
+    {
+      "epoch": 0.42089771891096395,
+      "grad_norm": 2.409449815750122,
+      "learning_rate": 6.715177342463145e-06,
+      "loss": 0.6086,
+      "step": 286
+    },
+    {
+      "epoch": 0.42236938925680645,
+      "grad_norm": 3.0505027770996094,
+      "learning_rate": 6.692316836833066e-06,
+      "loss": 0.8025,
+      "step": 287
+    },
+    {
+      "epoch": 0.423841059602649,
+      "grad_norm": 2.561631679534912,
+      "learning_rate": 6.6694163076601995e-06,
+      "loss": 0.7739,
+      "step": 288
+    },
+    {
+      "epoch": 0.42531272994849156,
+      "grad_norm": 2.783630847930908,
+      "learning_rate": 6.646476296545434e-06,
+      "loss": 0.73,
+      "step": 289
+    },
+    {
+      "epoch": 0.42678440029433407,
+      "grad_norm": 2.918511390686035,
+      "learning_rate": 6.6234973460234184e-06,
+      "loss": 0.8336,
+      "step": 290
+    },
+    {
+      "epoch": 0.4282560706401766,
+      "grad_norm": 2.899632692337036,
+      "learning_rate": 6.600479999549721e-06,
+      "loss": 0.847,
+      "step": 291
+    },
+    {
+      "epoch": 0.4297277409860191,
+      "grad_norm": 2.849313974380493,
+      "learning_rate": 6.57742480148798e-06,
+      "loss": 0.675,
+      "step": 292
+    },
+    {
+      "epoch": 0.4311994113318617,
+      "grad_norm": 2.822368860244751,
+      "learning_rate": 6.554332297097032e-06,
+      "loss": 0.8094,
+      "step": 293
+    },
+    {
+      "epoch": 0.4326710816777042,
+      "grad_norm": 2.5822582244873047,
+      "learning_rate": 6.53120303251801e-06,
+      "loss": 0.7853,
+      "step": 294
+    },
+    {
+      "epoch": 0.43414275202354674,
+      "grad_norm": 2.8273942470550537,
+      "learning_rate": 6.5080375547614325e-06,
+      "loss": 1.074,
+      "step": 295
+    },
+    {
+      "epoch": 0.43561442236938924,
+      "grad_norm": 2.904768705368042,
+      "learning_rate": 6.4848364116942665e-06,
+      "loss": 0.7764,
+      "step": 296
+    },
+    {
+      "epoch": 0.4370860927152318,
+      "grad_norm": 2.6361513137817383,
+      "learning_rate": 6.461600152026966e-06,
+      "loss": 0.7447,
+      "step": 297
+    },
+    {
+      "epoch": 0.4385577630610743,
+      "grad_norm": 2.717353105545044,
+      "learning_rate": 6.4383293253004996e-06,
+      "loss": 0.6936,
+      "step": 298
+    },
+    {
+      "epoch": 0.44002943340691686,
+      "grad_norm": 2.722168207168579,
+      "learning_rate": 6.415024481873352e-06,
+      "loss": 0.7681,
+      "step": 299
+    },
+    {
+      "epoch": 0.44150110375275936,
+      "grad_norm": 2.6191341876983643,
+      "learning_rate": 6.391686172908507e-06,
+      "loss": 0.7627,
+      "step": 300
+    },
+    {
+      "epoch": 0.4429727740986019,
+      "grad_norm": 2.391493082046509,
+      "learning_rate": 6.368314950360416e-06,
+      "loss": 0.6849,
+      "step": 301
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 2.654991865158081,
+      "learning_rate": 6.344911366961935e-06,
+      "loss": 0.8075,
+      "step": 302
+    },
+    {
+      "epoch": 0.445916114790287,
+      "grad_norm": 2.5275535583496094,
+      "learning_rate": 6.321475976211267e-06,
+      "loss": 0.7684,
+      "step": 303
+    },
+    {
+      "epoch": 0.44738778513612953,
+      "grad_norm": 2.607572078704834,
+      "learning_rate": 6.298009332358857e-06,
+      "loss": 0.7934,
+      "step": 304
+    },
+    {
+      "epoch": 0.44885945548197204,
+      "grad_norm": 2.6352803707122803,
+      "learning_rate": 6.274511990394294e-06,
+      "loss": 0.8401,
+      "step": 305
+    },
+    {
+      "epoch": 0.4503311258278146,
+      "grad_norm": 2.862342357635498,
+      "learning_rate": 6.250984506033183e-06,
+      "loss": 0.8266,
+      "step": 306
+    },
+    {
+      "epoch": 0.4518027961736571,
+      "grad_norm": 2.898751735687256,
+      "learning_rate": 6.227427435703997e-06,
+      "loss": 0.78,
+      "step": 307
+    },
+    {
+      "epoch": 0.45327446651949965,
+      "grad_norm": 3.0323238372802734,
+      "learning_rate": 6.203841336534924e-06,
+      "loss": 0.7944,
+      "step": 308
+    },
+    {
+      "epoch": 0.45474613686534215,
+      "grad_norm": 2.3359568119049072,
+      "learning_rate": 6.180226766340688e-06,
+      "loss": 0.7331,
+      "step": 309
+    },
+    {
+      "epoch": 0.4562178072111847,
+      "grad_norm": 2.904798746109009,
+      "learning_rate": 6.156584283609359e-06,
+      "loss": 0.7173,
+      "step": 310
+    },
+    {
+      "epoch": 0.4576894775570272,
+      "grad_norm": 3.093244791030884,
+      "learning_rate": 6.132914447489137e-06,
+      "loss": 0.8657,
+      "step": 311
+    },
+    {
+      "epoch": 0.45916114790286977,
+      "grad_norm": 3.100961685180664,
+      "learning_rate": 6.109217817775139e-06,
+      "loss": 1.0079,
+      "step": 312
+    },
+    {
+      "epoch": 0.4606328182487123,
+      "grad_norm": 2.5757603645324707,
+      "learning_rate": 6.085494954896156e-06,
+      "loss": 0.9887,
+      "step": 313
+    },
+    {
+      "epoch": 0.46210448859455483,
+      "grad_norm": 2.6319212913513184,
+      "learning_rate": 6.061746419901389e-06,
+      "loss": 0.5864,
+      "step": 314
+    },
+    {
+      "epoch": 0.46357615894039733,
+      "grad_norm": 3.017703056335449,
+      "learning_rate": 6.037972774447194e-06,
+      "loss": 0.828,
+      "step": 315
+    },
+    {
+      "epoch": 0.4650478292862399,
+      "grad_norm": 2.824852466583252,
+      "learning_rate": 6.014174580783794e-06,
+      "loss": 0.9,
+      "step": 316
+    },
+    {
+      "epoch": 0.4665194996320824,
+      "grad_norm": 2.583486795425415,
+      "learning_rate": 5.990352401741981e-06,
+      "loss": 0.8048,
+      "step": 317
+    },
+    {
+      "epoch": 0.46799116997792495,
+      "grad_norm": 3.027487277984619,
+      "learning_rate": 5.966506800719798e-06,
+      "loss": 0.938,
+      "step": 318
+    },
+    {
+      "epoch": 0.46946284032376745,
+      "grad_norm": 2.3798038959503174,
+      "learning_rate": 5.94263834166923e-06,
+      "loss": 0.7072,
+      "step": 319
+    },
+    {
+      "epoch": 0.47093451066961,
+      "grad_norm": 3.0898847579956055,
+      "learning_rate": 5.918747589082853e-06,
+      "loss": 0.8603,
+      "step": 320
+    },
+    {
+      "epoch": 0.47240618101545256,
+      "grad_norm": 3.080111026763916,
+      "learning_rate": 5.8948351079804875e-06,
+      "loss": 0.7156,
+      "step": 321
+    },
+    {
+      "epoch": 0.47387785136129507,
+      "grad_norm": 3.2610669136047363,
+      "learning_rate": 5.8709014638958406e-06,
+      "loss": 1.0839,
+      "step": 322
+    },
+    {
+      "epoch": 0.4753495217071376,
+      "grad_norm": 2.785217046737671,
+      "learning_rate": 5.846947222863123e-06,
+      "loss": 0.7786,
+      "step": 323
+    },
+    {
+      "epoch": 0.4768211920529801,
+      "grad_norm": 2.5710747241973877,
+      "learning_rate": 5.82297295140367e-06,
+      "loss": 0.7557,
+      "step": 324
+    },
+    {
+      "epoch": 0.4782928623988227,
+      "grad_norm": 2.669019937515259,
+      "learning_rate": 5.798979216512536e-06,
+      "loss": 0.7343,
+      "step": 325
+    },
+    {
+      "epoch": 0.4797645327446652,
+      "grad_norm": 2.868227005004883,
+      "learning_rate": 5.774966585645092e-06,
+      "loss": 0.7073,
+      "step": 326
+    },
+    {
+      "epoch": 0.48123620309050774,
+      "grad_norm": 2.7177627086639404,
+      "learning_rate": 5.750935626703598e-06,
+      "loss": 0.671,
+      "step": 327
+    },
+    {
+      "epoch": 0.48270787343635024,
+      "grad_norm": 2.884152412414551,
+      "learning_rate": 5.726886908023777e-06,
+      "loss": 0.7695,
+      "step": 328
+    },
+    {
+      "epoch": 0.4841795437821928,
+      "grad_norm": 2.8466315269470215,
+      "learning_rate": 5.702820998361374e-06,
+      "loss": 0.7373,
+      "step": 329
+    },
+    {
+      "epoch": 0.4856512141280353,
+      "grad_norm": 2.944166660308838,
+      "learning_rate": 5.678738466878699e-06,
+      "loss": 0.701,
+      "step": 330
+    },
+    {
+      "epoch": 0.48712288447387786,
+      "grad_norm": 2.806544303894043,
+      "learning_rate": 5.6546398831311774e-06,
+      "loss": 0.9049,
+      "step": 331
+    },
+    {
+      "epoch": 0.48859455481972036,
+      "grad_norm": 2.7961156368255615,
+      "learning_rate": 5.6305258170538676e-06,
+      "loss": 0.6243,
+      "step": 332
+    },
+    {
+      "epoch": 0.4900662251655629,
+      "grad_norm": 2.8264875411987305,
+      "learning_rate": 5.606396838947988e-06,
+      "loss": 0.8022,
+      "step": 333
+    },
+    {
+      "epoch": 0.4915378955114054,
+      "grad_norm": 2.3879430294036865,
+      "learning_rate": 5.582253519467432e-06,
+      "loss": 0.7052,
+      "step": 334
+    },
+    {
+      "epoch": 0.493009565857248,
+      "grad_norm": 2.72257137298584,
+      "learning_rate": 5.558096429605263e-06,
+      "loss": 0.7944,
+      "step": 335
+    },
+    {
+      "epoch": 0.49448123620309054,
+      "grad_norm": 2.5829052925109863,
+      "learning_rate": 5.533926140680222e-06,
+      "loss": 0.6631,
+      "step": 336
+    },
+    {
+      "epoch": 0.49595290654893304,
+      "grad_norm": 2.6868889331817627,
+      "learning_rate": 5.509743224323203e-06,
+      "loss": 0.7841,
+      "step": 337
+    },
+    {
+      "epoch": 0.4974245768947756,
+      "grad_norm": 2.834749698638916,
+      "learning_rate": 5.485548252463749e-06,
+      "loss": 0.7468,
+      "step": 338
+    },
+    {
+      "epoch": 0.4988962472406181,
+      "grad_norm": 2.4310719966888428,
+      "learning_rate": 5.46134179731651e-06,
+      "loss": 0.6794,
+      "step": 339
+    },
+    {
+      "epoch": 0.5003679175864606,
+      "grad_norm": 2.9393675327301025,
+      "learning_rate": 5.437124431367723e-06,
+      "loss": 0.7932,
+      "step": 340
+    },
+    {
+      "epoch": 0.5018395879323032,
+      "grad_norm": 2.730503797531128,
+      "learning_rate": 5.412896727361663e-06,
+      "loss": 0.8225,
+      "step": 341
+    },
+    {
+      "epoch": 0.5033112582781457,
+      "grad_norm": 3.03702449798584,
+      "learning_rate": 5.388659258287103e-06,
+      "loss": 0.7743,
+      "step": 342
+    },
+    {
+      "epoch": 0.5047829286239882,
+      "grad_norm": 2.5906665325164795,
+      "learning_rate": 5.36441259736376e-06,
+      "loss": 0.6833,
+      "step": 343
+    },
+    {
+      "epoch": 0.5062545989698307,
+      "grad_norm": 2.9820640087127686,
+      "learning_rate": 5.340157318028743e-06,
+      "loss": 0.9421,
+      "step": 344
+    },
+    {
+      "epoch": 0.5077262693156733,
+      "grad_norm": 2.8592886924743652,
+      "learning_rate": 5.3158939939229855e-06,
+      "loss": 0.7296,
+      "step": 345
+    },
+    {
+      "epoch": 0.5091979396615158,
+      "grad_norm": 2.825484275817871,
+      "learning_rate": 5.29162319887768e-06,
+      "loss": 0.8671,
+      "step": 346
+    },
+    {
+      "epoch": 0.5106696100073583,
+      "grad_norm": 2.661414384841919,
+      "learning_rate": 5.26734550690071e-06,
+      "loss": 0.8022,
+      "step": 347
+    },
+    {
+      "epoch": 0.5121412803532008,
+      "grad_norm": 2.814080238342285,
+      "learning_rate": 5.243061492163073e-06,
+      "loss": 0.8754,
+      "step": 348
+    },
+    {
+      "epoch": 0.5136129506990434,
+      "grad_norm": 2.7929255962371826,
+      "learning_rate": 5.218771728985296e-06,
+      "loss": 0.6977,
+      "step": 349
+    },
+    {
+      "epoch": 0.515084621044886,
+      "grad_norm": 3.3023571968078613,
+      "learning_rate": 5.1944767918238625e-06,
+      "loss": 0.9073,
+      "step": 350
+    },
+    {
+      "epoch": 0.5165562913907285,
+      "grad_norm": 2.8338370323181152,
+      "learning_rate": 5.170177255257618e-06,
+      "loss": 0.8119,
+      "step": 351
+    },
+    {
+      "epoch": 0.5180279617365711,
+      "grad_norm": 2.5478556156158447,
+      "learning_rate": 5.145873693974188e-06,
+      "loss": 0.7282,
+      "step": 352
+    },
+    {
+      "epoch": 0.5194996320824136,
+      "grad_norm": 2.494419813156128,
+      "learning_rate": 5.12156668275638e-06,
+      "loss": 0.7521,
+      "step": 353
+    },
+    {
+      "epoch": 0.5209713024282561,
+      "grad_norm": 2.797395706176758,
+      "learning_rate": 5.097256796468598e-06,
+      "loss": 0.7097,
+      "step": 354
+    },
+    {
+      "epoch": 0.5224429727740986,
+      "grad_norm": 2.939812421798706,
+      "learning_rate": 5.0729446100432326e-06,
+      "loss": 0.7828,
+      "step": 355
+    },
+    {
+      "epoch": 0.5239146431199412,
+      "grad_norm": 2.9113619327545166,
+      "learning_rate": 5.048630698467081e-06,
+      "loss": 0.844,
+      "step": 356
+    },
+    {
+      "epoch": 0.5253863134657837,
+      "grad_norm": 2.550365924835205,
+      "learning_rate": 5.024315636767738e-06,
+      "loss": 0.7853,
+      "step": 357
+    },
+    {
+      "epoch": 0.5268579838116262,
+      "grad_norm": 2.595287799835205,
+      "learning_rate": 5e-06,
+      "loss": 0.6457,
+      "step": 358
+    },
+    {
+      "epoch": 0.5283296541574687,
+      "grad_norm": 2.7226860523223877,
+      "learning_rate": 4.975684363232263e-06,
+      "loss": 0.734,
+      "step": 359
+    },
+    {
+      "epoch": 0.5298013245033113,
+      "grad_norm": 2.8578407764434814,
+      "learning_rate": 4.951369301532919e-06,
+      "loss": 0.8679,
+      "step": 360
+    },
+    {
+      "epoch": 0.5312729948491538,
+      "grad_norm": 2.848055124282837,
+      "learning_rate": 4.927055389956768e-06,
+      "loss": 0.7845,
+      "step": 361
+    },
+    {
+      "epoch": 0.5327446651949963,
+      "grad_norm": 2.7717485427856445,
+      "learning_rate": 4.9027432035314045e-06,
+      "loss": 0.8777,
+      "step": 362
+    },
+    {
+      "epoch": 0.5342163355408388,
+      "grad_norm": 2.854522466659546,
+      "learning_rate": 4.878433317243621e-06,
+      "loss": 0.8327,
+      "step": 363
+    },
+    {
+      "epoch": 0.5356880058866814,
+      "grad_norm": 3.19992733001709,
+      "learning_rate": 4.854126306025813e-06,
+      "loss": 0.7984,
+      "step": 364
+    },
+    {
+      "epoch": 0.5371596762325239,
+      "grad_norm": 2.4490439891815186,
+      "learning_rate": 4.829822744742383e-06,
+      "loss": 0.7675,
+      "step": 365
+    },
+    {
+      "epoch": 0.5386313465783664,
+      "grad_norm": 2.8721706867218018,
+      "learning_rate": 4.805523208176139e-06,
+      "loss": 0.7735,
+      "step": 366
+    },
+    {
+      "epoch": 0.5401030169242089,
+      "grad_norm": 2.759693145751953,
+      "learning_rate": 4.781228271014704e-06,
+      "loss": 0.8394,
+      "step": 367
+    },
+    {
+      "epoch": 0.5415746872700515,
+      "grad_norm": 2.8939993381500244,
+      "learning_rate": 4.756938507836929e-06,
+      "loss": 0.7914,
+      "step": 368
+    },
+    {
+      "epoch": 0.543046357615894,
+      "grad_norm": 2.637204647064209,
+      "learning_rate": 4.7326544930992905e-06,
+      "loss": 0.7225,
+      "step": 369
+    },
+    {
+      "epoch": 0.5445180279617365,
+      "grad_norm": 2.537356376647949,
+      "learning_rate": 4.708376801122321e-06,
+      "loss": 0.7063,
+      "step": 370
+    },
+    {
+      "epoch": 0.5459896983075792,
+      "grad_norm": 2.9717514514923096,
+      "learning_rate": 4.684106006077015e-06,
+      "loss": 1.0508,
+      "step": 371
+    },
+    {
+      "epoch": 0.5474613686534217,
+      "grad_norm": 2.7512850761413574,
+      "learning_rate": 4.659842681971258e-06,
+      "loss": 0.9937,
+      "step": 372
+    },
+    {
+      "epoch": 0.5489330389992642,
+      "grad_norm": 2.5516979694366455,
+      "learning_rate": 4.635587402636241e-06,
+      "loss": 0.7076,
+      "step": 373
+    },
+    {
+      "epoch": 0.5504047093451067,
+      "grad_norm": 2.5767934322357178,
+      "learning_rate": 4.611340741712901e-06,
+      "loss": 0.6927,
+      "step": 374
+    },
+    {
+      "epoch": 0.5518763796909493,
+      "grad_norm": 2.8047547340393066,
+      "learning_rate": 4.587103272638339e-06,
+      "loss": 0.7365,
+      "step": 375
+    },
+    {
+      "epoch": 0.5533480500367918,
+      "grad_norm": 2.8013343811035156,
+      "learning_rate": 4.562875568632278e-06,
+      "loss": 0.8549,
+      "step": 376
+    },
+    {
+      "epoch": 0.5548197203826343,
+      "grad_norm": 2.835747718811035,
+      "learning_rate": 4.53865820268349e-06,
+      "loss": 0.8297,
+      "step": 377
+    },
+    {
+      "epoch": 0.5562913907284768,
+      "grad_norm": 2.6400527954101562,
+      "learning_rate": 4.514451747536252e-06,
+      "loss": 0.8164,
+      "step": 378
+    },
+    {
+      "epoch": 0.5577630610743194,
+      "grad_norm": 2.340898036956787,
+      "learning_rate": 4.4902567756767976e-06,
+      "loss": 0.7111,
+      "step": 379
+    },
+    {
+      "epoch": 0.5592347314201619,
+      "grad_norm": 2.1810595989227295,
+      "learning_rate": 4.466073859319781e-06,
+      "loss": 0.6738,
+      "step": 380
+    },
+    {
+      "epoch": 0.5607064017660044,
+      "grad_norm": 2.5656542778015137,
+      "learning_rate": 4.441903570394739e-06,
+      "loss": 0.8892,
+      "step": 381
+    },
+    {
+      "epoch": 0.5621780721118469,
+      "grad_norm": 2.4357686042785645,
+      "learning_rate": 4.4177464805325695e-06,
+      "loss": 0.6686,
+      "step": 382
+    },
+    {
+      "epoch": 0.5636497424576895,
+      "grad_norm": 2.548100471496582,
+      "learning_rate": 4.3936031610520126e-06,
+      "loss": 0.6647,
+      "step": 383
+    },
+    {
+      "epoch": 0.565121412803532,
+      "grad_norm": 2.8117403984069824,
+      "learning_rate": 4.369474182946133e-06,
+      "loss": 0.8539,
+      "step": 384
+    },
+    {
+      "epoch": 0.5665930831493745,
+      "grad_norm": 2.70920467376709,
+      "learning_rate": 4.3453601168688225e-06,
+      "loss": 0.7258,
+      "step": 385
+    },
+    {
+      "epoch": 0.5680647534952171,
+      "grad_norm": 2.609267234802246,
+      "learning_rate": 4.321261533121303e-06,
+      "loss": 0.9332,
+      "step": 386
+    },
+    {
+      "epoch": 0.5695364238410596,
+      "grad_norm": 2.978053331375122,
+      "learning_rate": 4.297179001638629e-06,
+      "loss": 0.8062,
+      "step": 387
+    },
+    {
+      "epoch": 0.5710080941869021,
+      "grad_norm": 2.7160444259643555,
+      "learning_rate": 4.273113091976226e-06,
+      "loss": 0.7262,
+      "step": 388
+    },
+    {
+      "epoch": 0.5724797645327446,
+      "grad_norm": 2.662433624267578,
+      "learning_rate": 4.249064373296403e-06,
+      "loss": 1.0035,
+      "step": 389
+    },
+    {
+      "epoch": 0.5739514348785872,
+      "grad_norm": 2.56451416015625,
+      "learning_rate": 4.225033414354909e-06,
+      "loss": 0.7963,
+      "step": 390
+    },
+    {
+      "epoch": 0.5754231052244297,
+      "grad_norm": 2.2902348041534424,
+      "learning_rate": 4.201020783487465e-06,
+      "loss": 0.7863,
+      "step": 391
+    },
+    {
+      "epoch": 0.5768947755702722,
+      "grad_norm": 2.604950428009033,
+      "learning_rate": 4.17702704859633e-06,
+      "loss": 0.7413,
+      "step": 392
+    },
+    {
+      "epoch": 0.5783664459161147,
+      "grad_norm": 2.8261375427246094,
+      "learning_rate": 4.153052777136879e-06,
+      "loss": 0.8013,
+      "step": 393
+    },
+    {
+      "epoch": 0.5798381162619574,
+      "grad_norm": 2.6362226009368896,
+      "learning_rate": 4.129098536104161e-06,
+      "loss": 0.8815,
+      "step": 394
+    },
+    {
+      "epoch": 0.5813097866077999,
+      "grad_norm": 2.562701463699341,
+      "learning_rate": 4.105164892019514e-06,
+      "loss": 0.7645,
+      "step": 395
+    },
+    {
+      "epoch": 0.5827814569536424,
+      "grad_norm": 2.8253800868988037,
+      "learning_rate": 4.081252410917148e-06,
+      "loss": 0.8393,
+      "step": 396
+    },
+    {
+      "epoch": 0.5842531272994849,
+      "grad_norm": 3.0047738552093506,
+      "learning_rate": 4.0573616583307705e-06,
+      "loss": 0.8258,
+      "step": 397
+    },
+    {
+      "epoch": 0.5857247976453275,
+      "grad_norm": 2.6175127029418945,
+      "learning_rate": 4.033493199280203e-06,
+      "loss": 0.8083,
+      "step": 398
+    },
+    {
+      "epoch": 0.58719646799117,
+      "grad_norm": 2.4778695106506348,
+      "learning_rate": 4.009647598258022e-06,
+      "loss": 0.7379,
+      "step": 399
+    },
+    {
+      "epoch": 0.5886681383370125,
+      "grad_norm": 2.7437469959259033,
+      "learning_rate": 3.985825419216207e-06,
+      "loss": 0.8363,
+      "step": 400
+    },
+    {
+      "epoch": 0.5901398086828551,
+      "grad_norm": 2.574824571609497,
+      "learning_rate": 3.962027225552807e-06,
+      "loss": 0.6493,
+      "step": 401
+    },
+    {
+      "epoch": 0.5916114790286976,
+      "grad_norm": 2.9498939514160156,
+      "learning_rate": 3.938253580098613e-06,
+      "loss": 0.8471,
+      "step": 402
+    },
+    {
+      "epoch": 0.5930831493745401,
+      "grad_norm": 2.7222161293029785,
+      "learning_rate": 3.914505045103845e-06,
+      "loss": 0.8087,
+      "step": 403
+    },
+    {
+      "epoch": 0.5945548197203826,
+      "grad_norm": 2.566035747528076,
+      "learning_rate": 3.89078218222486e-06,
+      "loss": 0.585,
+      "step": 404
+    },
+    {
+      "epoch": 0.5960264900662252,
+      "grad_norm": 2.4814834594726562,
+      "learning_rate": 3.867085552510865e-06,
+      "loss": 0.7181,
+      "step": 405
+    },
+    {
+      "epoch": 0.5974981604120677,
+      "grad_norm": 2.8687238693237305,
+      "learning_rate": 3.843415716390644e-06,
+      "loss": 0.6498,
+      "step": 406
+    },
+    {
+      "epoch": 0.5989698307579102,
+      "grad_norm": 2.9591500759124756,
+      "learning_rate": 3.819773233659314e-06,
+      "loss": 0.8105,
+      "step": 407
+    },
+    {
+      "epoch": 0.6004415011037527,
+      "grad_norm": 2.970616579055786,
+      "learning_rate": 3.7961586634650773e-06,
+      "loss": 0.8648,
+      "step": 408
+    },
+    {
+      "epoch": 0.6019131714495953,
+      "grad_norm": 2.613100290298462,
+      "learning_rate": 3.7725725642960047e-06,
+      "loss": 0.6968,
+      "step": 409
+    },
+    {
+      "epoch": 0.6033848417954378,
+      "grad_norm": 2.374007225036621,
+      "learning_rate": 3.7490154939668176e-06,
+      "loss": 0.6306,
+      "step": 410
+    },
+    {
+      "epoch": 0.6048565121412803,
+      "grad_norm": 2.7874433994293213,
+      "learning_rate": 3.725488009605708e-06,
+      "loss": 0.7445,
+      "step": 411
+    },
+    {
+      "epoch": 0.6063281824871228,
+      "grad_norm": 3.110804319381714,
+      "learning_rate": 3.701990667641145e-06,
+      "loss": 0.8443,
+      "step": 412
+    },
+    {
+      "epoch": 0.6077998528329654,
+      "grad_norm": 2.5718374252319336,
+      "learning_rate": 3.6785240237887355e-06,
+      "loss": 0.6842,
+      "step": 413
+    },
+    {
+      "epoch": 0.609271523178808,
+      "grad_norm": 3.108994722366333,
+      "learning_rate": 3.655088633038067e-06,
+      "loss": 0.6848,
+      "step": 414
+    },
+    {
+      "epoch": 0.6107431935246505,
+      "grad_norm": 2.580518960952759,
+      "learning_rate": 3.6316850496395863e-06,
+      "loss": 0.7368,
+      "step": 415
+    },
+    {
+      "epoch": 0.6122148638704931,
+      "grad_norm": 2.642230749130249,
+      "learning_rate": 3.6083138270914932e-06,
+      "loss": 0.7295,
+      "step": 416
+    },
+    {
+      "epoch": 0.6136865342163356,
+      "grad_norm": 3.21877121925354,
+      "learning_rate": 3.584975518126648e-06,
+      "loss": 0.8949,
+      "step": 417
+    },
+    {
+      "epoch": 0.6151582045621781,
+      "grad_norm": 2.6950385570526123,
+      "learning_rate": 3.561670674699503e-06,
+      "loss": 0.942,
+      "step": 418
+    },
+    {
+      "epoch": 0.6166298749080206,
+      "grad_norm": 3.3686113357543945,
+      "learning_rate": 3.5383998479730357e-06,
+      "loss": 0.9905,
+      "step": 419
+    },
+    {
+      "epoch": 0.6181015452538632,
+      "grad_norm": 2.3857452869415283,
+      "learning_rate": 3.5151635883057356e-06,
+      "loss": 0.7793,
+      "step": 420
+    },
+    {
+      "epoch": 0.6195732155997057,
+      "grad_norm": 2.945770740509033,
+      "learning_rate": 3.491962445238569e-06,
+      "loss": 0.8128,
+      "step": 421
+    },
+    {
+      "epoch": 0.6210448859455482,
+      "grad_norm": 2.611788511276245,
+      "learning_rate": 3.4687969674819912e-06,
+      "loss": 0.7271,
+      "step": 422
+    },
+    {
+      "epoch": 0.6225165562913907,
+      "grad_norm": 2.4041404724121094,
+      "learning_rate": 3.4456677029029687e-06,
+      "loss": 0.7543,
+      "step": 423
+    },
+    {
+      "epoch": 0.6239882266372333,
+      "grad_norm": 2.8908700942993164,
+      "learning_rate": 3.4225751985120213e-06,
+      "loss": 0.7535,
+      "step": 424
+    },
+    {
+      "epoch": 0.6254598969830758,
+      "grad_norm": 2.6423447132110596,
+      "learning_rate": 3.3995200004502814e-06,
+      "loss": 0.7395,
+      "step": 425
+    },
+    {
+      "epoch": 0.6269315673289183,
+      "grad_norm": 2.6107118129730225,
+      "learning_rate": 3.3765026539765832e-06,
+      "loss": 0.9585,
+      "step": 426
+    },
+    {
+      "epoch": 0.6284032376747608,
+      "grad_norm": 2.6921772956848145,
+      "learning_rate": 3.3535237034545677e-06,
+      "loss": 0.867,
+      "step": 427
+    },
+    {
+      "epoch": 0.6298749080206034,
+      "grad_norm": 2.6669552326202393,
+      "learning_rate": 3.3305836923398026e-06,
+      "loss": 0.785,
+      "step": 428
+    },
+    {
+      "epoch": 0.6313465783664459,
+      "grad_norm": 2.6771135330200195,
+      "learning_rate": 3.307683163166934e-06,
+      "loss": 0.7238,
+      "step": 429
+    },
+    {
+      "epoch": 0.6328182487122884,
+      "grad_norm": 2.9073171615600586,
+      "learning_rate": 3.2848226575368565e-06,
+      "loss": 0.8712,
+      "step": 430
+    },
+    {
+      "epoch": 0.6342899190581309,
+      "grad_norm": 2.675462484359741,
+      "learning_rate": 3.2620027161038975e-06,
+      "loss": 0.8208,
+      "step": 431
+    },
+    {
+      "epoch": 0.6357615894039735,
+      "grad_norm": 2.6564865112304688,
+      "learning_rate": 3.2392238785630387e-06,
+      "loss": 0.8071,
+      "step": 432
+    },
+    {
+      "epoch": 0.637233259749816,
+      "grad_norm": 3.0897388458251953,
+      "learning_rate": 3.216486683637146e-06,
+      "loss": 0.9577,
+      "step": 433
+    },
+    {
+      "epoch": 0.6387049300956585,
+      "grad_norm": 2.678072452545166,
+      "learning_rate": 3.1937916690642356e-06,
+      "loss": 0.7587,
+      "step": 434
+    },
+    {
+      "epoch": 0.6401766004415012,
+      "grad_norm": 2.514816999435425,
+      "learning_rate": 3.1711393715847477e-06,
+      "loss": 0.7648,
+      "step": 435
+    },
+    {
+      "epoch": 0.6416482707873437,
+      "grad_norm": 2.598339319229126,
+      "learning_rate": 3.1485303269288603e-06,
+      "loss": 0.9055,
+      "step": 436
+    },
+    {
+      "epoch": 0.6431199411331862,
+      "grad_norm": 2.5613248348236084,
+      "learning_rate": 3.1259650698038106e-06,
+      "loss": 0.8907,
+      "step": 437
+    },
+    {
+      "epoch": 0.6445916114790287,
+      "grad_norm": 2.8210527896881104,
+      "learning_rate": 3.103444133881261e-06,
+      "loss": 1.0472,
+      "step": 438
+    },
+    {
+      "epoch": 0.6460632818248713,
+      "grad_norm": 2.8982555866241455,
+      "learning_rate": 3.0809680517846664e-06,
+      "loss": 0.7641,
+      "step": 439
+    },
+    {
+      "epoch": 0.6475349521707138,
+      "grad_norm": 2.70125150680542,
+      "learning_rate": 3.058537355076683e-06,
+      "loss": 0.7527,
+      "step": 440
+    },
+    {
+      "epoch": 0.6490066225165563,
+      "grad_norm": 2.6196935176849365,
+      "learning_rate": 3.0361525742465975e-06,
+      "loss": 0.7548,
+      "step": 441
+    },
+    {
+      "epoch": 0.6504782928623988,
+      "grad_norm": 2.6116650104522705,
+      "learning_rate": 3.0138142386977786e-06,
+      "loss": 0.866,
+      "step": 442
+    },
+    {
+      "epoch": 0.6519499632082414,
+      "grad_norm": 2.4863133430480957,
+      "learning_rate": 2.991522876735154e-06,
+      "loss": 0.8648,
+      "step": 443
+    },
+    {
+      "epoch": 0.6534216335540839,
+      "grad_norm": 2.4774370193481445,
+      "learning_rate": 2.9692790155527228e-06,
+      "loss": 0.7423,
+      "step": 444
+    },
+    {
+      "epoch": 0.6548933038999264,
+      "grad_norm": 2.4120702743530273,
+      "learning_rate": 2.9470831812210836e-06,
+      "loss": 0.6506,
+      "step": 445
+    },
+    {
+      "epoch": 0.6563649742457689,
+      "grad_norm": 2.8675732612609863,
+      "learning_rate": 2.9249358986749922e-06,
+      "loss": 0.6814,
+      "step": 446
+    },
+    {
+      "epoch": 0.6578366445916115,
+      "grad_norm": 2.7161166667938232,
+      "learning_rate": 2.9028376917009448e-06,
+      "loss": 0.7128,
+      "step": 447
+    },
+    {
+      "epoch": 0.659308314937454,
+      "grad_norm": 2.4354002475738525,
+      "learning_rate": 2.880789082924798e-06,
+      "loss": 0.7436,
+      "step": 448
+    },
+    {
+      "epoch": 0.6607799852832965,
+      "grad_norm": 2.698469400405884,
+      "learning_rate": 2.858790593799405e-06,
+      "loss": 0.8305,
+      "step": 449
+    },
+    {
+      "epoch": 0.6622516556291391,
+      "grad_norm": 2.746195077896118,
+      "learning_rate": 2.8368427445922697e-06,
+      "loss": 0.8079,
+      "step": 450
+    },
+    {
+      "epoch": 0.6637233259749816,
+      "grad_norm": 2.750650644302368,
+      "learning_rate": 2.8149460543732666e-06,
+      "loss": 0.7356,
+      "step": 451
+    },
+    {
+      "epoch": 0.6651949963208241,
+      "grad_norm": 2.8698570728302,
+      "learning_rate": 2.7931010410023516e-06,
+      "loss": 0.7823,
+      "step": 452
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 2.5994505882263184,
+      "learning_rate": 2.771308221117309e-06,
+      "loss": 0.8341,
+      "step": 453
+    },
+    {
+      "epoch": 0.6681383370125092,
+      "grad_norm": 2.789361000061035,
+      "learning_rate": 2.749568110121545e-06,
+      "loss": 0.7507,
+      "step": 454
+    },
+    {
+      "epoch": 0.6696100073583517,
+      "grad_norm": 2.711209774017334,
+      "learning_rate": 2.7278812221718927e-06,
+      "loss": 0.7451,
+      "step": 455
+    },
+    {
+      "epoch": 0.6710816777041942,
+      "grad_norm": 2.647339105606079,
+      "learning_rate": 2.706248070166449e-06,
+      "loss": 0.6768,
+      "step": 456
+    },
+    {
+      "epoch": 0.6725533480500367,
+      "grad_norm": 2.946976661682129,
+      "learning_rate": 2.6846691657324473e-06,
+      "loss": 0.8795,
+      "step": 457
+    },
+    {
+      "epoch": 0.6740250183958794,
+      "grad_norm": 2.607889175415039,
+      "learning_rate": 2.663145019214163e-06,
+      "loss": 0.7953,
+      "step": 458
+    },
+    {
+      "epoch": 0.6754966887417219,
+      "grad_norm": 2.4289133548736572,
+      "learning_rate": 2.6416761396608365e-06,
+      "loss": 0.7475,
+      "step": 459
+    },
+    {
+      "epoch": 0.6769683590875644,
+      "grad_norm": 2.715745449066162,
+      "learning_rate": 2.6202630348146323e-06,
+      "loss": 0.6455,
+      "step": 460
+    },
+    {
+      "epoch": 0.6784400294334069,
+      "grad_norm": 2.5252177715301514,
+      "learning_rate": 2.598906211098643e-06,
+      "loss": 0.7237,
+      "step": 461
+    },
+    {
+      "epoch": 0.6799116997792495,
+      "grad_norm": 2.5386343002319336,
+      "learning_rate": 2.577606173604894e-06,
+      "loss": 0.793,
+      "step": 462
+    },
+    {
+      "epoch": 0.681383370125092,
+      "grad_norm": 2.5121896266937256,
+      "learning_rate": 2.556363426082418e-06,
+      "loss": 0.7519,
+      "step": 463
+    },
+    {
+      "epoch": 0.6828550404709345,
+      "grad_norm": 2.628500461578369,
+      "learning_rate": 2.535178470925323e-06,
+      "loss": 0.8999,
+      "step": 464
+    },
+    {
+      "epoch": 0.6843267108167771,
+      "grad_norm": 2.4954497814178467,
+      "learning_rate": 2.5140518091609254e-06,
+      "loss": 0.7667,
+      "step": 465
+    },
+    {
+      "epoch": 0.6857983811626196,
+      "grad_norm": 2.9571664333343506,
+      "learning_rate": 2.4929839404378935e-06,
+      "loss": 0.9206,
+      "step": 466
+    },
+    {
+      "epoch": 0.6872700515084621,
+      "grad_norm": 2.7304599285125732,
+      "learning_rate": 2.4719753630144283e-06,
+      "loss": 0.8066,
+      "step": 467
+    },
+    {
+      "epoch": 0.6887417218543046,
+      "grad_norm": 2.7657480239868164,
+      "learning_rate": 2.451026573746482e-06,
+      "loss": 0.7235,
+      "step": 468
+    },
+    {
+      "epoch": 0.6902133922001472,
+      "grad_norm": 2.851828098297119,
+      "learning_rate": 2.430138068076013e-06,
+      "loss": 0.7381,
+      "step": 469
+    },
+    {
+      "epoch": 0.6916850625459897,
+      "grad_norm": 2.8263611793518066,
+      "learning_rate": 2.4093103400192626e-06,
+      "loss": 0.9067,
+      "step": 470
+    },
+    {
+      "epoch": 0.6931567328918322,
+      "grad_norm": 2.27524995803833,
+      "learning_rate": 2.388543882155067e-06,
+      "loss": 0.5643,
+      "step": 471
+    },
+    {
+      "epoch": 0.6946284032376747,
+      "grad_norm": 2.7731189727783203,
+      "learning_rate": 2.3678391856132203e-06,
+      "loss": 0.7023,
+      "step": 472
+    },
+    {
+      "epoch": 0.6961000735835173,
+      "grad_norm": 2.6988525390625,
+      "learning_rate": 2.3471967400628513e-06,
+      "loss": 0.7669,
+      "step": 473
+    },
+    {
+      "epoch": 0.6975717439293598,
+      "grad_norm": 2.60081148147583,
+      "learning_rate": 2.32661703370084e-06,
+      "loss": 0.8822,
+      "step": 474
+    },
+    {
+      "epoch": 0.6990434142752023,
+      "grad_norm": 2.901594638824463,
+      "learning_rate": 2.306100553240274e-06,
+      "loss": 0.7002,
+      "step": 475
+    },
+    {
+      "epoch": 0.7005150846210448,
+      "grad_norm": 2.74725079536438,
+      "learning_rate": 2.2856477838989455e-06,
+      "loss": 0.7208,
+      "step": 476
+    },
+    {
+      "epoch": 0.7019867549668874,
+      "grad_norm": 2.5290257930755615,
+      "learning_rate": 2.265259209387867e-06,
+      "loss": 0.9018,
+      "step": 477
+    },
+    {
+      "epoch": 0.70345842531273,
+      "grad_norm": 3.207190752029419,
+      "learning_rate": 2.244935311899829e-06,
+      "loss": 0.9887,
+      "step": 478
+    },
+    {
+      "epoch": 0.7049300956585725,
+      "grad_norm": 3.089996099472046,
+      "learning_rate": 2.2246765720980074e-06,
+      "loss": 0.8796,
+      "step": 479
+    },
+    {
+      "epoch": 0.7064017660044151,
+      "grad_norm": 2.8044493198394775,
+      "learning_rate": 2.2044834691045873e-06,
+      "loss": 0.8091,
+      "step": 480
+    },
+    {
+      "epoch": 0.7078734363502576,
+      "grad_norm": 2.7987630367279053,
+      "learning_rate": 2.1843564804894316e-06,
+      "loss": 0.792,
+      "step": 481
+    },
+    {
+      "epoch": 0.7093451066961001,
+      "grad_norm": 3.069977045059204,
+      "learning_rate": 2.1642960822587878e-06,
+      "loss": 0.7951,
+      "step": 482
+    },
+    {
+      "epoch": 0.7108167770419426,
+      "grad_norm": 3.010993003845215,
+      "learning_rate": 2.1443027488440338e-06,
+      "loss": 0.8252,
+      "step": 483
+    },
+    {
+      "epoch": 0.7122884473877852,
+      "grad_norm": 2.7961068153381348,
+      "learning_rate": 2.1243769530904564e-06,
+      "loss": 0.89,
+      "step": 484
+    },
+    {
+      "epoch": 0.7137601177336277,
+      "grad_norm": 2.8578946590423584,
+      "learning_rate": 2.104519166246059e-06,
+      "loss": 0.8419,
+      "step": 485
+    },
+    {
+      "epoch": 0.7152317880794702,
+      "grad_norm": 2.810152769088745,
+      "learning_rate": 2.0847298579504344e-06,
+      "loss": 0.7479,
+      "step": 486
+    },
+    {
+      "epoch": 0.7167034584253127,
+      "grad_norm": 2.567078113555908,
+      "learning_rate": 2.065009496223638e-06,
+      "loss": 0.664,
+      "step": 487
+    },
+    {
+      "epoch": 0.7181751287711553,
+      "grad_norm": 2.8592417240142822,
+      "learning_rate": 2.045358547455138e-06,
+      "loss": 0.8539,
+      "step": 488
+    },
+    {
+      "epoch": 0.7196467991169978,
+      "grad_norm": 2.859276533126831,
+      "learning_rate": 2.0257774763927656e-06,
+      "loss": 0.8118,
+      "step": 489
+    },
+    {
+      "epoch": 0.7211184694628403,
+      "grad_norm": 2.916217088699341,
+      "learning_rate": 2.0062667461317425e-06,
+      "loss": 0.7133,
+      "step": 490
+    },
+    {
+      "epoch": 0.7225901398086828,
+      "grad_norm": 2.890087127685547,
+      "learning_rate": 1.9868268181037186e-06,
+      "loss": 0.8926,
+      "step": 491
+    },
+    {
+      "epoch": 0.7240618101545254,
+      "grad_norm": 2.791355848312378,
+      "learning_rate": 1.967458152065857e-06,
+      "loss": 0.9062,
+      "step": 492
+    },
+    {
+      "epoch": 0.7255334805003679,
+      "grad_norm": 2.536635398864746,
+      "learning_rate": 1.9481612060899646e-06,
+      "loss": 0.7114,
+      "step": 493
+    },
+    {
+      "epoch": 0.7270051508462104,
+      "grad_norm": 2.708343505859375,
+      "learning_rate": 1.928936436551661e-06,
+      "loss": 0.8861,
+      "step": 494
+    },
+    {
+      "epoch": 0.7284768211920529,
+      "grad_norm": 2.842942953109741,
+      "learning_rate": 1.9097842981195836e-06,
+      "loss": 0.7926,
+      "step": 495
+    },
+    {
+      "epoch": 0.7299484915378955,
+      "grad_norm": 2.975794792175293,
+      "learning_rate": 1.8907052437446271e-06,
+      "loss": 0.801,
+      "step": 496
+    },
+    {
+      "epoch": 0.731420161883738,
+      "grad_norm": 2.9826395511627197,
+      "learning_rate": 1.871699724649244e-06,
+      "loss": 0.6895,
+      "step": 497
+    },
+    {
+      "epoch": 0.7328918322295805,
+      "grad_norm": 2.5878913402557373,
+      "learning_rate": 1.8527681903167644e-06,
+      "loss": 0.6327,
+      "step": 498
+    },
+    {
+      "epoch": 0.7343635025754232,
+      "grad_norm": 2.8014075756073,
+      "learning_rate": 1.8339110884807671e-06,
+      "loss": 0.7839,
+      "step": 499
+    },
+    {
+      "epoch": 0.7358351729212657,
+      "grad_norm": 2.62280535697937,
+      "learning_rate": 1.8151288651144894e-06,
+      "loss": 0.6315,
+      "step": 500
+    },
+    {
+      "epoch": 0.7373068432671082,
+      "grad_norm": 2.3114869594573975,
+      "learning_rate": 1.7964219644202852e-06,
+      "loss": 0.8394,
+      "step": 501
+    },
+    {
+      "epoch": 0.7387785136129507,
+      "grad_norm": 2.415231466293335,
+      "learning_rate": 1.7777908288191176e-06,
+      "loss": 0.724,
+      "step": 502
+    },
+    {
+      "epoch": 0.7402501839587933,
+      "grad_norm": 2.5043222904205322,
+      "learning_rate": 1.7592358989400882e-06,
+      "loss": 0.8248,
+      "step": 503
+    },
+    {
+      "epoch": 0.7417218543046358,
+      "grad_norm": 2.547647476196289,
+      "learning_rate": 1.7407576136100278e-06,
+      "loss": 0.7288,
+      "step": 504
+    },
+    {
+      "epoch": 0.7431935246504783,
+      "grad_norm": 2.986689329147339,
+      "learning_rate": 1.7223564098431067e-06,
+      "loss": 0.8131,
+      "step": 505
+    },
+    {
+      "epoch": 0.7446651949963208,
+      "grad_norm": 2.9111227989196777,
+      "learning_rate": 1.704032722830512e-06,
+      "loss": 0.7931,
+      "step": 506
+    },
+    {
+      "epoch": 0.7461368653421634,
+      "grad_norm": 2.6043741703033447,
+      "learning_rate": 1.6857869859301401e-06,
+      "loss": 0.68,
+      "step": 507
+    },
+    {
+      "epoch": 0.7476085356880059,
+      "grad_norm": 2.713120460510254,
+      "learning_rate": 1.6676196306563614e-06,
+      "loss": 0.7447,
+      "step": 508
+    },
+    {
+      "epoch": 0.7490802060338484,
+      "grad_norm": 2.875211000442505,
+      "learning_rate": 1.6495310866698095e-06,
+      "loss": 0.8778,
+      "step": 509
+    },
+    {
+      "epoch": 0.7505518763796909,
+      "grad_norm": 2.747056722640991,
+      "learning_rate": 1.6315217817672142e-06,
+      "loss": 0.8072,
+      "step": 510
+    },
+    {
+      "epoch": 0.7520235467255335,
+      "grad_norm": 2.6293632984161377,
+      "learning_rate": 1.6135921418712959e-06,
+      "loss": 0.8272,
+      "step": 511
+    },
+    {
+      "epoch": 0.753495217071376,
+      "grad_norm": 2.7001330852508545,
+      "learning_rate": 1.5957425910206787e-06,
+      "loss": 0.9759,
+      "step": 512
+    },
+    {
+      "epoch": 0.7549668874172185,
+      "grad_norm": 2.5239274501800537,
+      "learning_rate": 1.577973551359877e-06,
+      "loss": 0.717,
+      "step": 513
+    },
+    {
+      "epoch": 0.7564385577630611,
+      "grad_norm": 2.681542158126831,
+      "learning_rate": 1.5602854431292963e-06,
+      "loss": 0.6227,
+      "step": 514
+    },
+    {
+      "epoch": 0.7579102281089036,
+      "grad_norm": 2.667667865753174,
+      "learning_rate": 1.542678684655306e-06,
+      "loss": 0.8233,
+      "step": 515
+    },
+    {
+      "epoch": 0.7593818984547461,
+      "grad_norm": 2.79180908203125,
+      "learning_rate": 1.5251536923403427e-06,
+      "loss": 0.7726,
+      "step": 516
+    },
+    {
+      "epoch": 0.7608535688005886,
+      "grad_norm": 3.0569746494293213,
+      "learning_rate": 1.5077108806530582e-06,
+      "loss": 0.7847,
+      "step": 517
+    },
+    {
+      "epoch": 0.7623252391464312,
+      "grad_norm": 2.5163207054138184,
+      "learning_rate": 1.4903506621185193e-06,
+      "loss": 0.7946,
+      "step": 518
+    },
+    {
+      "epoch": 0.7637969094922737,
+      "grad_norm": 2.5696611404418945,
+      "learning_rate": 1.4730734473084568e-06,
+      "loss": 0.7233,
+      "step": 519
+    },
+    {
+      "epoch": 0.7652685798381162,
+      "grad_norm": 2.863888740539551,
+      "learning_rate": 1.4558796448315504e-06,
+      "loss": 0.7795,
+      "step": 520
+    },
+    {
+      "epoch": 0.7667402501839587,
+      "grad_norm": 2.5073134899139404,
+      "learning_rate": 1.4387696613237612e-06,
+      "loss": 0.782,
+      "step": 521
+    },
+    {
+      "epoch": 0.7682119205298014,
+      "grad_norm": 3.094791889190674,
+      "learning_rate": 1.4217439014387251e-06,
+      "loss": 0.936,
+      "step": 522
+    },
+    {
+      "epoch": 0.7696835908756439,
+      "grad_norm": 2.4089252948760986,
+      "learning_rate": 1.404802767838176e-06,
+      "loss": 0.7706,
+      "step": 523
+    },
+    {
+      "epoch": 0.7711552612214864,
+      "grad_norm": 2.9336740970611572,
+      "learning_rate": 1.38794666118242e-06,
+      "loss": 0.8929,
+      "step": 524
+    },
+    {
+      "epoch": 0.7726269315673289,
+      "grad_norm": 2.5258219242095947,
+      "learning_rate": 1.371175980120864e-06,
+      "loss": 0.763,
+      "step": 525
+    },
+    {
+      "epoch": 0.7740986019131715,
+      "grad_norm": 2.525669574737549,
+      "learning_rate": 1.3544911212825907e-06,
+      "loss": 0.8683,
+      "step": 526
+    },
+    {
+      "epoch": 0.775570272259014,
+      "grad_norm": 2.4933722019195557,
+      "learning_rate": 1.337892479266974e-06,
+      "loss": 0.7019,
+      "step": 527
+    },
+    {
+      "epoch": 0.7770419426048565,
+      "grad_norm": 2.8783955574035645,
+      "learning_rate": 1.321380446634342e-06,
+      "loss": 0.914,
+      "step": 528
+    },
+    {
+      "epoch": 0.7785136129506991,
+      "grad_norm": 3.0792112350463867,
+      "learning_rate": 1.3049554138967052e-06,
+      "loss": 0.768,
+      "step": 529
+    },
+    {
+      "epoch": 0.7799852832965416,
+      "grad_norm": 2.811859130859375,
+      "learning_rate": 1.2886177695085078e-06,
+      "loss": 0.782,
+      "step": 530
+    },
+    {
+      "epoch": 0.7814569536423841,
+      "grad_norm": 2.7325949668884277,
+      "learning_rate": 1.2723678998574512e-06,
+      "loss": 0.8395,
+      "step": 531
+    },
+    {
+      "epoch": 0.7829286239882266,
+      "grad_norm": 2.7767133712768555,
+      "learning_rate": 1.2562061892553472e-06,
+      "loss": 0.766,
+      "step": 532
+    },
+    {
+      "epoch": 0.7844002943340692,
+      "grad_norm": 2.8727259635925293,
+      "learning_rate": 1.2401330199290368e-06,
+      "loss": 0.7734,
+      "step": 533
+    },
+    {
+      "epoch": 0.7858719646799117,
+      "grad_norm": 2.7274365425109863,
+      "learning_rate": 1.224148772011346e-06,
+      "loss": 0.8786,
+      "step": 534
+    },
+    {
+      "epoch": 0.7873436350257542,
+      "grad_norm": 2.7339842319488525,
+      "learning_rate": 1.2082538235320928e-06,
+      "loss": 0.7648,
+      "step": 535
+    },
+    {
+      "epoch": 0.7888153053715967,
+      "grad_norm": 2.509180784225464,
+      "learning_rate": 1.1924485504091565e-06,
+      "loss": 0.7314,
+      "step": 536
+    },
+    {
+      "epoch": 0.7902869757174393,
+      "grad_norm": 3.0106115341186523,
+      "learning_rate": 1.1767333264395735e-06,
+      "loss": 0.7766,
+      "step": 537
+    },
+    {
+      "epoch": 0.7917586460632818,
+      "grad_norm": 2.565518617630005,
+      "learning_rate": 1.1611085232907132e-06,
+      "loss": 0.683,
+      "step": 538
+    },
+    {
+      "epoch": 0.7932303164091243,
+      "grad_norm": 2.6307907104492188,
+      "learning_rate": 1.14557451049147e-06,
+      "loss": 0.833,
+      "step": 539
+    },
+    {
+      "epoch": 0.7947019867549668,
+      "grad_norm": 2.797319173812866,
+      "learning_rate": 1.1301316554235397e-06,
+      "loss": 0.7519,
+      "step": 540
+    },
+    {
+      "epoch": 0.7961736571008095,
+      "grad_norm": 2.6741998195648193,
+      "learning_rate": 1.1147803233127241e-06,
+      "loss": 0.7264,
+      "step": 541
+    },
+    {
+      "epoch": 0.797645327446652,
+      "grad_norm": 2.9562933444976807,
+      "learning_rate": 1.0995208772202898e-06,
+      "loss": 0.9802,
+      "step": 542
+    },
+    {
+      "epoch": 0.7991169977924945,
+      "grad_norm": 2.957308053970337,
+      "learning_rate": 1.0843536780343866e-06,
+      "loss": 0.904,
+      "step": 543
+    },
+    {
+      "epoch": 0.8005886681383371,
+      "grad_norm": 3.1149260997772217,
+      "learning_rate": 1.0692790844615131e-06,
+      "loss": 0.8182,
+      "step": 544
+    },
+    {
+      "epoch": 0.8020603384841796,
+      "grad_norm": 3.102614641189575,
+      "learning_rate": 1.0542974530180327e-06,
+      "loss": 0.7731,
+      "step": 545
+    },
+    {
+      "epoch": 0.8035320088300221,
+      "grad_norm": 2.8476767539978027,
+      "learning_rate": 1.0394091380217354e-06,
+      "loss": 0.7407,
+      "step": 546
+    },
+    {
+      "epoch": 0.8050036791758646,
+      "grad_norm": 2.722993850708008,
+      "learning_rate": 1.0246144915834683e-06,
+      "loss": 0.6995,
+      "step": 547
+    },
+    {
+      "epoch": 0.8064753495217072,
+      "grad_norm": 2.6003148555755615,
+      "learning_rate": 1.0099138635988026e-06,
+      "loss": 0.7062,
+      "step": 548
+    },
+    {
+      "epoch": 0.8079470198675497,
+      "grad_norm": 2.6799726486206055,
+      "learning_rate": 9.953076017397579e-07,
+      "loss": 0.6479,
+      "step": 549
+    },
+    {
+      "epoch": 0.8094186902133922,
+      "grad_norm": 2.7827186584472656,
+      "learning_rate": 9.807960514465792e-07,
+      "loss": 0.7404,
+      "step": 550
+    },
+    {
+      "epoch": 0.8108903605592347,
+      "grad_norm": 2.6582939624786377,
+      "learning_rate": 9.663795559195733e-07,
+      "loss": 0.7495,
+      "step": 551
+    },
+    {
+      "epoch": 0.8123620309050773,
+      "grad_norm": 2.6799185276031494,
+      "learning_rate": 9.520584561109863e-07,
+      "loss": 0.8731,
+      "step": 552
+    },
+    {
+      "epoch": 0.8138337012509198,
+      "grad_norm": 2.924217462539673,
+      "learning_rate": 9.378330907169387e-07,
+      "loss": 0.6908,
+      "step": 553
+    },
+    {
+      "epoch": 0.8153053715967623,
+      "grad_norm": 2.823230028152466,
+      "learning_rate": 9.237037961694223e-07,
+      "loss": 0.7609,
+      "step": 554
+    },
+    {
+      "epoch": 0.8167770419426048,
+      "grad_norm": 2.8368513584136963,
+      "learning_rate": 9.096709066283355e-07,
+      "loss": 0.8756,
+      "step": 555
+    },
+    {
+      "epoch": 0.8182487122884474,
+      "grad_norm": 2.3523733615875244,
+      "learning_rate": 8.957347539735872e-07,
+      "loss": 0.6146,
+      "step": 556
+    },
+    {
+      "epoch": 0.8197203826342899,
+      "grad_norm": 2.600253105163574,
+      "learning_rate": 8.818956677972407e-07,
+      "loss": 0.6998,
+      "step": 557
+    },
+    {
+      "epoch": 0.8211920529801324,
+      "grad_norm": 2.78033447265625,
+      "learning_rate": 8.681539753957268e-07,
+      "loss": 0.8612,
+      "step": 558
+    },
+    {
+      "epoch": 0.8226637233259749,
+      "grad_norm": 2.650313138961792,
+      "learning_rate": 8.545100017620988e-07,
+      "loss": 0.7443,
+      "step": 559
+    },
+    {
+      "epoch": 0.8241353936718175,
+      "grad_norm": 2.694070816040039,
+      "learning_rate": 8.409640695783444e-07,
+      "loss": 0.7601,
+      "step": 560
+    },
+    {
+      "epoch": 0.82560706401766,
+      "grad_norm": 3.236161708831787,
+      "learning_rate": 8.275164992077555e-07,
+      "loss": 0.8981,
+      "step": 561
+    },
+    {
+      "epoch": 0.8270787343635025,
+      "grad_norm": 2.4480512142181396,
+      "learning_rate": 8.141676086873574e-07,
+      "loss": 0.6256,
+      "step": 562
+    },
+    {
+      "epoch": 0.8285504047093452,
+      "grad_norm": 2.66650128364563,
+      "learning_rate": 8.009177137203794e-07,
+      "loss": 0.6718,
+      "step": 563
+    },
+    {
+      "epoch": 0.8300220750551877,
+      "grad_norm": 2.679651975631714,
+      "learning_rate": 7.877671276687899e-07,
+      "loss": 0.67,
+      "step": 564
+    },
+    {
+      "epoch": 0.8314937454010302,
+      "grad_norm": 2.603134870529175,
+      "learning_rate": 7.747161615458903e-07,
+      "loss": 0.6928,
+      "step": 565
+    },
+    {
+      "epoch": 0.8329654157468727,
+      "grad_norm": 2.7582461833953857,
+      "learning_rate": 7.617651240089546e-07,
+      "loss": 0.8008,
+      "step": 566
+    },
+    {
+      "epoch": 0.8344370860927153,
+      "grad_norm": 2.8830525875091553,
+      "learning_rate": 7.489143213519301e-07,
+      "loss": 0.8107,
+      "step": 567
+    },
+    {
+      "epoch": 0.8359087564385578,
+      "grad_norm": 2.789246082305908,
+      "learning_rate": 7.361640574981938e-07,
+      "loss": 0.7996,
+      "step": 568
+    },
+    {
+      "epoch": 0.8373804267844003,
+      "grad_norm": 3.0257482528686523,
+      "learning_rate": 7.235146339933674e-07,
+      "loss": 0.7135,
+      "step": 569
+    },
+    {
+      "epoch": 0.8388520971302428,
+      "grad_norm": 2.429319381713867,
+      "learning_rate": 7.109663499981834e-07,
+      "loss": 0.8229,
+      "step": 570
+    },
+    {
+      "epoch": 0.8403237674760854,
+      "grad_norm": 2.646169424057007,
+      "learning_rate": 6.985195022814068e-07,
+      "loss": 0.7295,
+      "step": 571
+    },
+    {
+      "epoch": 0.8417954378219279,
+      "grad_norm": 2.7810020446777344,
+      "learning_rate": 6.861743852128233e-07,
+      "loss": 0.8911,
+      "step": 572
+    },
+    {
+      "epoch": 0.8432671081677704,
+      "grad_norm": 2.963247537612915,
+      "learning_rate": 6.739312907562734e-07,
+      "loss": 0.9285,
+      "step": 573
+    },
+    {
+      "epoch": 0.8447387785136129,
+      "grad_norm": 2.7100367546081543,
+      "learning_rate": 6.617905084627452e-07,
+      "loss": 0.7045,
+      "step": 574
+    },
+    {
+      "epoch": 0.8462104488594555,
+      "grad_norm": 2.6360723972320557,
+      "learning_rate": 6.497523254635296e-07,
+      "loss": 0.923,
+      "step": 575
+    },
+    {
+      "epoch": 0.847682119205298,
+      "grad_norm": 2.7827036380767822,
+      "learning_rate": 6.37817026463432e-07,
+      "loss": 0.7223,
+      "step": 576
+    },
+    {
+      "epoch": 0.8491537895511405,
+      "grad_norm": 3.0738601684570312,
+      "learning_rate": 6.25984893734034e-07,
+      "loss": 0.952,
+      "step": 577
+    },
+    {
+      "epoch": 0.8506254598969831,
+      "grad_norm": 2.955704689025879,
+      "learning_rate": 6.142562071070179e-07,
+      "loss": 0.9231,
+      "step": 578
+    },
+    {
+      "epoch": 0.8520971302428256,
+      "grad_norm": 2.607506036758423,
+      "learning_rate": 6.026312439675553e-07,
+      "loss": 0.8037,
+      "step": 579
+    },
+    {
+      "epoch": 0.8535688005886681,
+      "grad_norm": 2.722355842590332,
+      "learning_rate": 5.911102792477358e-07,
+      "loss": 0.8769,
+      "step": 580
+    },
+    {
+      "epoch": 0.8550404709345106,
+      "grad_norm": 2.6345765590667725,
+      "learning_rate": 5.796935854200764e-07,
+      "loss": 0.6607,
+      "step": 581
+    },
+    {
+      "epoch": 0.8565121412803532,
+      "grad_norm": 2.4528110027313232,
+      "learning_rate": 5.683814324910685e-07,
+      "loss": 0.7011,
+      "step": 582
+    },
+    {
+      "epoch": 0.8579838116261957,
+      "grad_norm": 2.753122329711914,
+      "learning_rate": 5.571740879947979e-07,
+      "loss": 0.7738,
+      "step": 583
+    },
+    {
+      "epoch": 0.8594554819720382,
+      "grad_norm": 2.6649703979492188,
+      "learning_rate": 5.460718169866163e-07,
+      "loss": 0.7858,
+      "step": 584
+    },
+    {
+      "epoch": 0.8609271523178808,
+      "grad_norm": 2.739748477935791,
+      "learning_rate": 5.350748820368689e-07,
+      "loss": 0.7256,
+      "step": 585
+    },
+    {
+      "epoch": 0.8623988226637234,
+      "grad_norm": 2.549894094467163,
+      "learning_rate": 5.241835432246888e-07,
+      "loss": 0.6335,
+      "step": 586
+    },
+    {
+      "epoch": 0.8638704930095659,
+      "grad_norm": 2.778587579727173,
+      "learning_rate": 5.133980581318459e-07,
+      "loss": 0.8065,
+      "step": 587
+    },
+    {
+      "epoch": 0.8653421633554084,
+      "grad_norm": 2.5618858337402344,
+      "learning_rate": 5.027186818366542e-07,
+      "loss": 0.7347,
+      "step": 588
+    },
+    {
+      "epoch": 0.8668138337012509,
+      "grad_norm": 2.583333730697632,
+      "learning_rate": 4.921456669079366e-07,
+      "loss": 0.8103,
+      "step": 589
+    },
+    {
+      "epoch": 0.8682855040470935,
+      "grad_norm": 2.6635944843292236,
+      "learning_rate": 4.81679263399057e-07,
+      "loss": 0.8243,
+      "step": 590
+    },
+    {
+      "epoch": 0.869757174392936,
+      "grad_norm": 3.3781163692474365,
+      "learning_rate": 4.713197188420027e-07,
+      "loss": 0.8512,
+      "step": 591
+    },
+    {
+      "epoch": 0.8712288447387785,
+      "grad_norm": 2.3862738609313965,
+      "learning_rate": 4.6106727824152764e-07,
+      "loss": 0.6945,
+      "step": 592
+    },
+    {
+      "epoch": 0.8727005150846211,
+      "grad_norm": 2.710103750228882,
+      "learning_rate": 4.509221840693656e-07,
+      "loss": 0.7842,
+      "step": 593
+    },
+    {
+      "epoch": 0.8741721854304636,
+      "grad_norm": 3.1467630863189697,
+      "learning_rate": 4.408846762584901e-07,
+      "loss": 0.8328,
+      "step": 594
+    },
+    {
+      "epoch": 0.8756438557763061,
+      "grad_norm": 2.763000965118408,
+      "learning_rate": 4.309549921974421e-07,
+      "loss": 0.8947,
+      "step": 595
+    },
+    {
+      "epoch": 0.8771155261221486,
+      "grad_norm": 3.0412211418151855,
+      "learning_rate": 4.211333667247125e-07,
+      "loss": 0.8716,
+      "step": 596
+    },
+    {
+      "epoch": 0.8785871964679912,
+      "grad_norm": 2.719144344329834,
+      "learning_rate": 4.114200321231937e-07,
+      "loss": 0.7778,
+      "step": 597
+    },
+    {
+      "epoch": 0.8800588668138337,
+      "grad_norm": 2.2873966693878174,
+      "learning_rate": 4.018152181146823e-07,
+      "loss": 0.6861,
+      "step": 598
+    },
+    {
+      "epoch": 0.8815305371596762,
+      "grad_norm": 2.654785633087158,
+      "learning_rate": 3.9231915185444337e-07,
+      "loss": 0.8151,
+      "step": 599
+    },
+    {
+      "epoch": 0.8830022075055187,
+      "grad_norm": 2.84719181060791,
+      "learning_rate": 3.8293205792584666e-07,
+      "loss": 0.9433,
+      "step": 600
+    },
+    {
+      "epoch": 0.8844738778513613,
+      "grad_norm": 2.5994880199432373,
+      "learning_rate": 3.736541583350473e-07,
+      "loss": 0.9252,
+      "step": 601
+    },
+    {
+      "epoch": 0.8859455481972038,
+      "grad_norm": 2.488598108291626,
+      "learning_rate": 3.6448567250574053e-07,
+      "loss": 0.6793,
+      "step": 602
+    },
+    {
+      "epoch": 0.8874172185430463,
+      "grad_norm": 2.968190908432007,
+      "learning_rate": 3.5542681727396613e-07,
+      "loss": 0.908,
+      "step": 603
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 2.297053813934326,
+      "learning_rate": 3.464778068829883e-07,
+      "loss": 0.6731,
+      "step": 604
+    },
+    {
+      "epoch": 0.8903605592347315,
+      "grad_norm": 2.4779140949249268,
+      "learning_rate": 3.3763885297822153e-07,
+      "loss": 0.6418,
+      "step": 605
+    },
+    {
+      "epoch": 0.891832229580574,
+      "grad_norm": 2.8664562702178955,
+      "learning_rate": 3.289101646022297e-07,
+      "loss": 0.7856,
+      "step": 606
+    },
+    {
+      "epoch": 0.8933038999264165,
+      "grad_norm": 2.8044114112854004,
+      "learning_rate": 3.2029194818977984e-07,
+      "loss": 0.9089,
+      "step": 607
+    },
+    {
+      "epoch": 0.8947755702722591,
+      "grad_norm": 2.656986951828003,
+      "learning_rate": 3.117844075629617e-07,
+      "loss": 0.8366,
+      "step": 608
+    },
+    {
+      "epoch": 0.8962472406181016,
+      "grad_norm": 2.7499563694000244,
+      "learning_rate": 3.033877439263666e-07,
+      "loss": 0.7709,
+      "step": 609
+    },
+    {
+      "epoch": 0.8977189109639441,
+      "grad_norm": 2.770153760910034,
+      "learning_rate": 2.9510215586232737e-07,
+      "loss": 0.8263,
+      "step": 610
+    },
+    {
+      "epoch": 0.8991905813097866,
+      "grad_norm": 2.6473796367645264,
+      "learning_rate": 2.869278393262226e-07,
+      "loss": 0.7757,
+      "step": 611
+    },
+    {
+      "epoch": 0.9006622516556292,
+      "grad_norm": 2.7503674030303955,
+      "learning_rate": 2.7886498764184587e-07,
+      "loss": 0.7354,
+      "step": 612
+    },
+    {
+      "epoch": 0.9021339220014717,
+      "grad_norm": 2.681307792663574,
+      "learning_rate": 2.7091379149682683e-07,
+      "loss": 0.8607,
+      "step": 613
+    },
+    {
+      "epoch": 0.9036055923473142,
+      "grad_norm": 2.6204802989959717,
+      "learning_rate": 2.6307443893812847e-07,
+      "loss": 0.8717,
+      "step": 614
+    },
+    {
+      "epoch": 0.9050772626931567,
+      "grad_norm": 2.9451098442077637,
+      "learning_rate": 2.55347115367594e-07,
+      "loss": 0.895,
+      "step": 615
+    },
+    {
+      "epoch": 0.9065489330389993,
+      "grad_norm": 2.728437662124634,
+      "learning_rate": 2.47732003537568e-07,
+      "loss": 0.8632,
+      "step": 616
+    },
+    {
+      "epoch": 0.9080206033848418,
+      "grad_norm": 2.671208381652832,
+      "learning_rate": 2.402292835465647e-07,
+      "loss": 0.7702,
+      "step": 617
+    },
+    {
+      "epoch": 0.9094922737306843,
+      "grad_norm": 2.5721795558929443,
+      "learning_rate": 2.3283913283502047e-07,
+      "loss": 0.8497,
+      "step": 618
+    },
+    {
+      "epoch": 0.9109639440765268,
+      "grad_norm": 2.4093384742736816,
+      "learning_rate": 2.2556172618108996e-07,
+      "loss": 0.8805,
+      "step": 619
+    },
+    {
+      "epoch": 0.9124356144223694,
+      "grad_norm": 3.0314748287200928,
+      "learning_rate": 2.183972356965125e-07,
+      "loss": 0.8581,
+      "step": 620
+    },
+    {
+      "epoch": 0.9139072847682119,
+      "grad_norm": 2.5950942039489746,
+      "learning_rate": 2.113458308225458e-07,
+      "loss": 0.6963,
+      "step": 621
+    },
+    {
+      "epoch": 0.9153789551140544,
+      "grad_norm": 2.590554714202881,
+      "learning_rate": 2.0440767832595576e-07,
+      "loss": 0.6748,
+      "step": 622
+    },
+    {
+      "epoch": 0.9168506254598969,
+      "grad_norm": 2.457000970840454,
+      "learning_rate": 1.9758294229507092e-07,
+      "loss": 0.761,
+      "step": 623
+    },
+    {
+      "epoch": 0.9183222958057395,
+      "grad_norm": 2.3966705799102783,
+      "learning_rate": 1.908717841359048e-07,
+      "loss": 0.7255,
+      "step": 624
+    },
+    {
+      "epoch": 0.919793966151582,
+      "grad_norm": 2.606142520904541,
+      "learning_rate": 1.8427436256833853e-07,
+      "loss": 0.7811,
+      "step": 625
+    },
+    {
+      "epoch": 0.9212656364974245,
+      "grad_norm": 2.725306749343872,
+      "learning_rate": 1.777908336223655e-07,
+      "loss": 0.7978,
+      "step": 626
+    },
+    {
+      "epoch": 0.9227373068432672,
+      "grad_norm": 2.595306634902954,
+      "learning_rate": 1.7142135063440034e-07,
+      "loss": 0.7847,
+      "step": 627
+    },
+    {
+      "epoch": 0.9242089771891097,
+      "grad_norm": 2.832662582397461,
+      "learning_rate": 1.6516606424365644e-07,
+      "loss": 0.6639,
+      "step": 628
+    },
+    {
+      "epoch": 0.9256806475349522,
+      "grad_norm": 2.824404239654541,
+      "learning_rate": 1.590251223885786e-07,
+      "loss": 0.6387,
+      "step": 629
+    },
+    {
+      "epoch": 0.9271523178807947,
+      "grad_norm": 2.551408052444458,
+      "learning_rate": 1.5299867030334815e-07,
+      "loss": 0.7776,
+      "step": 630
+    },
+    {
+      "epoch": 0.9286239882266373,
+      "grad_norm": 2.958336591720581,
+      "learning_rate": 1.4708685051444515e-07,
+      "loss": 0.9973,
+      "step": 631
+    },
+    {
+      "epoch": 0.9300956585724798,
+      "grad_norm": 2.609131336212158,
+      "learning_rate": 1.4128980283727946e-07,
+      "loss": 0.9308,
+      "step": 632
+    },
+    {
+      "epoch": 0.9315673289183223,
+      "grad_norm": 2.8172359466552734,
+      "learning_rate": 1.3560766437288432e-07,
+      "loss": 0.7915,
+      "step": 633
+    },
+    {
+      "epoch": 0.9330389992641648,
+      "grad_norm": 2.5778167247772217,
+      "learning_rate": 1.3004056950467135e-07,
+      "loss": 0.8021,
+      "step": 634
+    },
+    {
+      "epoch": 0.9345106696100074,
+      "grad_norm": 3.02774977684021,
+      "learning_rate": 1.24588649895257e-07,
+      "loss": 0.857,
+      "step": 635
+    },
+    {
+      "epoch": 0.9359823399558499,
+      "grad_norm": 2.6817007064819336,
+      "learning_rate": 1.19252034483342e-07,
+      "loss": 0.7217,
+      "step": 636
+    },
+    {
+      "epoch": 0.9374540103016924,
+      "grad_norm": 3.296025037765503,
+      "learning_rate": 1.1403084948067023e-07,
+      "loss": 0.9432,
+      "step": 637
+    },
+    {
+      "epoch": 0.9389256806475349,
+      "grad_norm": 2.963472604751587,
+      "learning_rate": 1.089252183690348e-07,
+      "loss": 0.7671,
+      "step": 638
+    },
+    {
+      "epoch": 0.9403973509933775,
+      "grad_norm": 2.72013258934021,
+      "learning_rate": 1.0393526189736602e-07,
+      "loss": 0.725,
+      "step": 639
+    },
+    {
+      "epoch": 0.94186902133922,
+      "grad_norm": 2.7649495601654053,
+      "learning_rate": 9.906109807887032e-08,
+      "loss": 0.8975,
+      "step": 640
+    },
+    {
+      "epoch": 0.9433406916850625,
+      "grad_norm": 2.6894352436065674,
+      "learning_rate": 9.430284218824026e-08,
+      "loss": 0.832,
+      "step": 641
+    },
+    {
+      "epoch": 0.9448123620309051,
+      "grad_norm": 2.740673780441284,
+      "learning_rate": 8.966060675892951e-08,
+      "loss": 1.0232,
+      "step": 642
+    },
+    {
+      "epoch": 0.9462840323767476,
+      "grad_norm": 2.584526777267456,
+      "learning_rate": 8.513450158049109e-08,
+      "loss": 0.7357,
+      "step": 643
+    },
+    {
+      "epoch": 0.9477557027225901,
+      "grad_norm": 2.6124203205108643,
+      "learning_rate": 8.072463369597994e-08,
+      "loss": 0.7387,
+      "step": 644
+    },
+    {
+      "epoch": 0.9492273730684326,
+      "grad_norm": 2.9415154457092285,
+      "learning_rate": 7.643110739942172e-08,
+      "loss": 0.7183,
+      "step": 645
+    },
+    {
+      "epoch": 0.9506990434142752,
+      "grad_norm": 2.719373941421509,
+      "learning_rate": 7.225402423334694e-08,
+      "loss": 0.69,
+      "step": 646
+    },
+    {
+      "epoch": 0.9521707137601177,
+      "grad_norm": 2.724714517593384,
+      "learning_rate": 6.819348298638839e-08,
+      "loss": 0.9065,
+      "step": 647
+    },
+    {
+      "epoch": 0.9536423841059603,
+      "grad_norm": 2.491793155670166,
+      "learning_rate": 6.424957969094536e-08,
+      "loss": 0.6751,
+      "step": 648
+    },
+    {
+      "epoch": 0.9551140544518028,
+      "grad_norm": 2.8342292308807373,
+      "learning_rate": 6.0422407620912e-08,
+      "loss": 0.8441,
+      "step": 649
+    },
+    {
+      "epoch": 0.9565857247976454,
+      "grad_norm": 2.551548719406128,
+      "learning_rate": 5.6712057289473047e-08,
+      "loss": 0.7815,
+      "step": 650
+    },
+    {
+      "epoch": 0.9580573951434879,
+      "grad_norm": 2.459239959716797,
+      "learning_rate": 5.3118616446960484e-08,
+      "loss": 0.7014,
+      "step": 651
+    },
+    {
+      "epoch": 0.9595290654893304,
+      "grad_norm": 2.655364751815796,
+      "learning_rate": 4.9642170078780804e-08,
+      "loss": 0.7887,
+      "step": 652
+    },
+    {
+      "epoch": 0.9610007358351729,
+      "grad_norm": 3.058354616165161,
+      "learning_rate": 4.628280040340272e-08,
+      "loss": 0.8193,
+      "step": 653
+    },
+    {
+      "epoch": 0.9624724061810155,
+      "grad_norm": 2.8031606674194336,
+      "learning_rate": 4.3040586870415346e-08,
+      "loss": 0.8415,
+      "step": 654
+    },
+    {
+      "epoch": 0.963944076526858,
+      "grad_norm": 2.788810968399048,
+      "learning_rate": 3.991560615864587e-08,
+      "loss": 0.6981,
+      "step": 655
+    },
+    {
+      "epoch": 0.9654157468727005,
+      "grad_norm": 2.4134199619293213,
+      "learning_rate": 3.690793217434985e-08,
+      "loss": 0.6854,
+      "step": 656
+    },
+    {
+      "epoch": 0.9668874172185431,
+      "grad_norm": 2.4256997108459473,
+      "learning_rate": 3.40176360494604e-08,
+      "loss": 0.8412,
+      "step": 657
+    },
+    {
+      "epoch": 0.9683590875643856,
+      "grad_norm": 2.961256742477417,
+      "learning_rate": 3.1244786139907334e-08,
+      "loss": 0.8928,
+      "step": 658
+    },
+    {
+      "epoch": 0.9698307579102281,
+      "grad_norm": 3.0347156524658203,
+      "learning_rate": 2.858944802399899e-08,
+      "loss": 0.862,
+      "step": 659
+    },
+    {
+      "epoch": 0.9713024282560706,
+      "grad_norm": 2.7489945888519287,
+      "learning_rate": 2.605168450087514e-08,
+      "loss": 0.8402,
+      "step": 660
+    },
+    {
+      "epoch": 0.9727740986019132,
+      "grad_norm": 2.5865862369537354,
+      "learning_rate": 2.363155558901542e-08,
+      "loss": 0.7846,
+      "step": 661
+    },
+    {
+      "epoch": 0.9742457689477557,
+      "grad_norm": 2.7627742290496826,
+      "learning_rate": 2.1329118524827662e-08,
+      "loss": 0.8333,
+      "step": 662
+    },
+    {
+      "epoch": 0.9757174392935982,
+      "grad_norm": 2.5887227058410645,
+      "learning_rate": 1.914442776128622e-08,
+      "loss": 0.8554,
+      "step": 663
+    },
+    {
+      "epoch": 0.9771891096394407,
+      "grad_norm": 2.511735439300537,
+      "learning_rate": 1.7077534966650767e-08,
+      "loss": 0.705,
+      "step": 664
+    },
+    {
+      "epoch": 0.9786607799852833,
+      "grad_norm": 2.7016990184783936,
+      "learning_rate": 1.51284890232406e-08,
+      "loss": 0.7558,
+      "step": 665
+    },
+    {
+      "epoch": 0.9801324503311258,
+      "grad_norm": 2.836970329284668,
+      "learning_rate": 1.3297336026280027e-08,
+      "loss": 0.8053,
+      "step": 666
+    },
+    {
+      "epoch": 0.9816041206769683,
+      "grad_norm": 2.5258963108062744,
+      "learning_rate": 1.158411928280645e-08,
+      "loss": 0.5653,
+      "step": 667
+    },
+    {
+      "epoch": 0.9830757910228108,
+      "grad_norm": 2.4458508491516113,
+      "learning_rate": 9.988879310649513e-09,
+      "loss": 0.6538,
+      "step": 668
+    },
+    {
+      "epoch": 0.9845474613686535,
+      "grad_norm": 2.5343708992004395,
+      "learning_rate": 8.511653837470212e-09,
+      "loss": 0.703,
+      "step": 669
+    },
+    {
+      "epoch": 0.986019131714496,
+      "grad_norm": 2.5260910987854004,
+      "learning_rate": 7.152477799867718e-09,
+      "loss": 0.67,
+      "step": 670
+    },
+    {
+      "epoch": 0.9874908020603385,
+      "grad_norm": 2.7152626514434814,
+      "learning_rate": 5.911383342556143e-09,
+      "loss": 0.9093,
+      "step": 671
+    },
+    {
+      "epoch": 0.9889624724061811,
+      "grad_norm": 2.4940526485443115,
+      "learning_rate": 4.788399817602929e-09,
+      "loss": 0.6923,
+      "step": 672
+    },
+    {
+      "epoch": 0.9904341427520236,
+      "grad_norm": 2.8715314865112305,
+      "learning_rate": 3.783553783733851e-09,
+      "loss": 0.7952,
+      "step": 673
+    },
+    {
+      "epoch": 0.9919058130978661,
+      "grad_norm": 2.9330999851226807,
+      "learning_rate": 2.896869005705183e-09,
+      "loss": 0.9401,
+      "step": 674
+    },
+    {
+      "epoch": 0.9933774834437086,
+      "grad_norm": 2.560947895050049,
+      "learning_rate": 2.128366453743591e-09,
+      "loss": 0.7578,
+      "step": 675
+    },
+    {
+      "epoch": 0.9948491537895512,
+      "grad_norm": 2.549210786819458,
+      "learning_rate": 1.4780643030476439e-09,
+      "loss": 0.8439,
+      "step": 676
+    },
+    {
+      "epoch": 0.9963208241353937,
+      "grad_norm": 2.915217876434326,
+      "learning_rate": 9.459779333587104e-10,
+      "loss": 0.7632,
+      "step": 677
+    },
+    {
+      "epoch": 0.9977924944812362,
+      "grad_norm": 2.8781983852386475,
+      "learning_rate": 5.321199285979184e-10,
+      "loss": 0.8137,
+      "step": 678
+    },
+    {
+      "epoch": 0.9992641648270787,
+      "grad_norm": 2.7309377193450928,
+      "learning_rate": 2.3650007656805804e-10,
+      "loss": 0.6736,
+      "step": 679
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.366690158843994,
+      "learning_rate": 5.912536872321184e-11,
+      "loss": 0.6984,
+      "step": 680
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 680,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "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": 3.879511026381619e+16,
+  "train_batch_size": 6,
+  "trial_name": null,
+  "trial_params": null
+}