diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5850 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 831,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0012033694344163659,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.9999928539313353e-05,
+      "loss": 1.9787,
+      "step": 1
+    },
+    {
+      "epoch": 0.0024067388688327317,
+      "grad_norm": 8.375,
+      "learning_rate": 1.9999714158274743e-05,
+      "loss": 3.438,
+      "step": 2
+    },
+    {
+      "epoch": 0.0036101083032490976,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.9999356859948126e-05,
+      "loss": 1.7121,
+      "step": 3
+    },
+    {
+      "epoch": 0.0048134777376654635,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.9998856649440058e-05,
+      "loss": 3.1317,
+      "step": 4
+    },
+    {
+      "epoch": 0.006016847172081829,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.9998213533899625e-05,
+      "loss": 2.1194,
+      "step": 5
+    },
+    {
+      "epoch": 0.007220216606498195,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.9997427522518315e-05,
+      "loss": 2.833,
+      "step": 6
+    },
+    {
+      "epoch": 0.00842358604091456,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.9996498626529914e-05,
+      "loss": 2.1864,
+      "step": 7
+    },
+    {
+      "epoch": 0.009626955475330927,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.999542685921033e-05,
+      "loss": 2.7296,
+      "step": 8
+    },
+    {
+      "epoch": 0.010830324909747292,
+      "grad_norm": 9.4375,
+      "learning_rate": 1.9994212235877407e-05,
+      "loss": 1.9642,
+      "step": 9
+    },
+    {
+      "epoch": 0.012033694344163659,
+      "grad_norm": 6.46875,
+      "learning_rate": 1.9992854773890714e-05,
+      "loss": 2.6336,
+      "step": 10
+    },
+    {
+      "epoch": 0.013237063778580024,
+      "grad_norm": 7.875,
+      "learning_rate": 1.9991354492651283e-05,
+      "loss": 1.7843,
+      "step": 11
+    },
+    {
+      "epoch": 0.01444043321299639,
+      "grad_norm": 4.8125,
+      "learning_rate": 1.9989711413601332e-05,
+      "loss": 2.5641,
+      "step": 12
+    },
+    {
+      "epoch": 0.015643802647412757,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.998792556022398e-05,
+      "loss": 2.2421,
+      "step": 13
+    },
+    {
+      "epoch": 0.01684717208182912,
+      "grad_norm": 3.984375,
+      "learning_rate": 1.9985996958042887e-05,
+      "loss": 2.4817,
+      "step": 14
+    },
+    {
+      "epoch": 0.018050541516245487,
+      "grad_norm": 15.875,
+      "learning_rate": 1.9983925634621894e-05,
+      "loss": 2.2535,
+      "step": 15
+    },
+    {
+      "epoch": 0.019253910950661854,
+      "grad_norm": 3.5625,
+      "learning_rate": 1.9981711619564654e-05,
+      "loss": 2.3963,
+      "step": 16
+    },
+    {
+      "epoch": 0.02045728038507822,
+      "grad_norm": 10.75,
+      "learning_rate": 1.997935494451416e-05,
+      "loss": 2.03,
+      "step": 17
+    },
+    {
+      "epoch": 0.021660649819494584,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.997685564315234e-05,
+      "loss": 2.344,
+      "step": 18
+    },
+    {
+      "epoch": 0.02286401925391095,
+      "grad_norm": 12.25,
+      "learning_rate": 1.9974213751199556e-05,
+      "loss": 1.928,
+      "step": 19
+    },
+    {
+      "epoch": 0.024067388688327317,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.9971429306414087e-05,
+      "loss": 2.3099,
+      "step": 20
+    },
+    {
+      "epoch": 0.02527075812274368,
+      "grad_norm": 7.84375,
+      "learning_rate": 1.99685023485916e-05,
+      "loss": 1.7257,
+      "step": 21
+    },
+    {
+      "epoch": 0.026474127557160047,
+      "grad_norm": 2.671875,
+      "learning_rate": 1.9965432919564583e-05,
+      "loss": 2.2477,
+      "step": 22
+    },
+    {
+      "epoch": 0.027677496991576414,
+      "grad_norm": 8.625,
+      "learning_rate": 1.9962221063201734e-05,
+      "loss": 1.8483,
+      "step": 23
+    },
+    {
+      "epoch": 0.02888086642599278,
+      "grad_norm": 3.21875,
+      "learning_rate": 1.995886682540734e-05,
+      "loss": 2.1731,
+      "step": 24
+    },
+    {
+      "epoch": 0.030084235860409144,
+      "grad_norm": 7.625,
+      "learning_rate": 1.9955370254120635e-05,
+      "loss": 1.5969,
+      "step": 25
+    },
+    {
+      "epoch": 0.031287605294825514,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.9951731399315095e-05,
+      "loss": 2.2097,
+      "step": 26
+    },
+    {
+      "epoch": 0.032490974729241874,
+      "grad_norm": 9.0,
+      "learning_rate": 1.994795031299773e-05,
+      "loss": 1.4314,
+      "step": 27
+    },
+    {
+      "epoch": 0.03369434416365824,
+      "grad_norm": 2.59375,
+      "learning_rate": 1.9944027049208347e-05,
+      "loss": 2.14,
+      "step": 28
+    },
+    {
+      "epoch": 0.03489771359807461,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.993996166401877e-05,
+      "loss": 1.7424,
+      "step": 29
+    },
+    {
+      "epoch": 0.036101083032490974,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.993575421553204e-05,
+      "loss": 2.1018,
+      "step": 30
+    },
+    {
+      "epoch": 0.03730445246690734,
+      "grad_norm": 12.0,
+      "learning_rate": 1.9931404763881598e-05,
+      "loss": 1.9668,
+      "step": 31
+    },
+    {
+      "epoch": 0.03850782190132371,
+      "grad_norm": 2.25,
+      "learning_rate": 1.9926913371230393e-05,
+      "loss": 2.1388,
+      "step": 32
+    },
+    {
+      "epoch": 0.039711191335740074,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.992228010177003e-05,
+      "loss": 2.1242,
+      "step": 33
+    },
+    {
+      "epoch": 0.04091456077015644,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.9917505021719833e-05,
+      "loss": 2.0531,
+      "step": 34
+    },
+    {
+      "epoch": 0.0421179302045728,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.99125881993259e-05,
+      "loss": 2.0355,
+      "step": 35
+    },
+    {
+      "epoch": 0.04332129963898917,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.990752970486014e-05,
+      "loss": 2.1009,
+      "step": 36
+    },
+    {
+      "epoch": 0.044524669073405534,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.990232961061924e-05,
+      "loss": 1.6248,
+      "step": 37
+    },
+    {
+      "epoch": 0.0457280385078219,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.989698799092366e-05,
+      "loss": 2.0782,
+      "step": 38
+    },
+    {
+      "epoch": 0.04693140794223827,
+      "grad_norm": 9.625,
+      "learning_rate": 1.9891504922116572e-05,
+      "loss": 1.6215,
+      "step": 39
+    },
+    {
+      "epoch": 0.048134777376654635,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.988588048256274e-05,
+      "loss": 2.0563,
+      "step": 40
+    },
+    {
+      "epoch": 0.049338146811071,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.9880114752647434e-05,
+      "loss": 2.0851,
+      "step": 41
+    },
+    {
+      "epoch": 0.05054151624548736,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.9874207814775252e-05,
+      "loss": 2.0189,
+      "step": 42
+    },
+    {
+      "epoch": 0.05174488567990373,
+      "grad_norm": 9.375,
+      "learning_rate": 1.9868159753368964e-05,
+      "loss": 1.8673,
+      "step": 43
+    },
+    {
+      "epoch": 0.052948255114320095,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.9861970654868292e-05,
+      "loss": 2.0445,
+      "step": 44
+    },
+    {
+      "epoch": 0.05415162454873646,
+      "grad_norm": 8.625,
+      "learning_rate": 1.9855640607728684e-05,
+      "loss": 2.1062,
+      "step": 45
+    },
+    {
+      "epoch": 0.05535499398315283,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.9849169702420044e-05,
+      "loss": 2.0244,
+      "step": 46
+    },
+    {
+      "epoch": 0.056558363417569195,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.9842558031425434e-05,
+      "loss": 1.9904,
+      "step": 47
+    },
+    {
+      "epoch": 0.05776173285198556,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.983580568923977e-05,
+      "loss": 2.0685,
+      "step": 48
+    },
+    {
+      "epoch": 0.05896510228640193,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.982891277236845e-05,
+      "loss": 1.6353,
+      "step": 49
+    },
+    {
+      "epoch": 0.06016847172081829,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.9821879379325985e-05,
+      "loss": 1.9923,
+      "step": 50
+    },
+    {
+      "epoch": 0.061371841155234655,
+      "grad_norm": 8.125,
+      "learning_rate": 1.9814705610634602e-05,
+      "loss": 1.9369,
+      "step": 51
+    },
+    {
+      "epoch": 0.06257521058965103,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.9807391568822785e-05,
+      "loss": 1.9696,
+      "step": 52
+    },
+    {
+      "epoch": 0.06377858002406739,
+      "grad_norm": 9.5,
+      "learning_rate": 1.9799937358423826e-05,
+      "loss": 2.1856,
+      "step": 53
+    },
+    {
+      "epoch": 0.06498194945848375,
+      "grad_norm": 2.375,
+      "learning_rate": 1.9792343085974316e-05,
+      "loss": 2.0433,
+      "step": 54
+    },
+    {
+      "epoch": 0.06618531889290012,
+      "grad_norm": 9.25,
+      "learning_rate": 1.9784608860012652e-05,
+      "loss": 2.0544,
+      "step": 55
+    },
+    {
+      "epoch": 0.06738868832731648,
+      "grad_norm": 1.875,
+      "learning_rate": 1.9776734791077442e-05,
+      "loss": 1.9472,
+      "step": 56
+    },
+    {
+      "epoch": 0.06859205776173286,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.976872099170597e-05,
+      "loss": 1.8057,
+      "step": 57
+    },
+    {
+      "epoch": 0.06979542719614922,
+      "grad_norm": 2.125,
+      "learning_rate": 1.976056757643255e-05,
+      "loss": 1.9963,
+      "step": 58
+    },
+    {
+      "epoch": 0.07099879663056559,
+      "grad_norm": 11.375,
+      "learning_rate": 1.9752274661786916e-05,
+      "loss": 1.8518,
+      "step": 59
+    },
+    {
+      "epoch": 0.07220216606498195,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.9743842366292544e-05,
+      "loss": 1.9508,
+      "step": 60
+    },
+    {
+      "epoch": 0.07340553549939831,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.9735270810464958e-05,
+      "loss": 1.9033,
+      "step": 61
+    },
+    {
+      "epoch": 0.07460890493381468,
+      "grad_norm": 1.625,
+      "learning_rate": 1.9726560116810006e-05,
+      "loss": 1.9837,
+      "step": 62
+    },
+    {
+      "epoch": 0.07581227436823104,
+      "grad_norm": 15.0,
+      "learning_rate": 1.971771040982213e-05,
+      "loss": 1.8616,
+      "step": 63
+    },
+    {
+      "epoch": 0.07701564380264742,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.9708721815982543e-05,
+      "loss": 1.9847,
+      "step": 64
+    },
+    {
+      "epoch": 0.07821901323706378,
+      "grad_norm": 7.59375,
+      "learning_rate": 1.9699594463757475e-05,
+      "loss": 1.7556,
+      "step": 65
+    },
+    {
+      "epoch": 0.07942238267148015,
+      "grad_norm": 1.6796875,
+      "learning_rate": 1.9690328483596287e-05,
+      "loss": 1.9431,
+      "step": 66
+    },
+    {
+      "epoch": 0.08062575210589651,
+      "grad_norm": 8.25,
+      "learning_rate": 1.968092400792965e-05,
+      "loss": 2.1827,
+      "step": 67
+    },
+    {
+      "epoch": 0.08182912154031288,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.9671381171167616e-05,
+      "loss": 1.9171,
+      "step": 68
+    },
+    {
+      "epoch": 0.08303249097472924,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.9661700109697718e-05,
+      "loss": 1.8986,
+      "step": 69
+    },
+    {
+      "epoch": 0.0842358604091456,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.9651880961883025e-05,
+      "loss": 2.0004,
+      "step": 70
+    },
+    {
+      "epoch": 0.08543922984356198,
+      "grad_norm": 8.875,
+      "learning_rate": 1.964192386806013e-05,
+      "loss": 1.4482,
+      "step": 71
+    },
+    {
+      "epoch": 0.08664259927797834,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.9631828970537196e-05,
+      "loss": 1.918,
+      "step": 72
+    },
+    {
+      "epoch": 0.08784596871239471,
+      "grad_norm": 11.375,
+      "learning_rate": 1.9621596413591885e-05,
+      "loss": 1.8718,
+      "step": 73
+    },
+    {
+      "epoch": 0.08904933814681107,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.96112263434693e-05,
+      "loss": 1.9367,
+      "step": 74
+    },
+    {
+      "epoch": 0.09025270758122744,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.960071890837991e-05,
+      "loss": 1.8661,
+      "step": 75
+    },
+    {
+      "epoch": 0.0914560770156438,
+      "grad_norm": 1.75,
+      "learning_rate": 1.9590074258497423e-05,
+      "loss": 1.9062,
+      "step": 76
+    },
+    {
+      "epoch": 0.09265944645006016,
+      "grad_norm": 8.625,
+      "learning_rate": 1.957929254595664e-05,
+      "loss": 2.0265,
+      "step": 77
+    },
+    {
+      "epoch": 0.09386281588447654,
+      "grad_norm": 2.125,
+      "learning_rate": 1.9568373924851267e-05,
+      "loss": 1.9558,
+      "step": 78
+    },
+    {
+      "epoch": 0.0950661853188929,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.9557318551231745e-05,
+      "loss": 1.7762,
+      "step": 79
+    },
+    {
+      "epoch": 0.09626955475330927,
+      "grad_norm": 1.875,
+      "learning_rate": 1.9546126583102983e-05,
+      "loss": 1.8972,
+      "step": 80
+    },
+    {
+      "epoch": 0.09747292418772563,
+      "grad_norm": 9.625,
+      "learning_rate": 1.953479818042214e-05,
+      "loss": 2.105,
+      "step": 81
+    },
+    {
+      "epoch": 0.098676293622142,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.952333350509629e-05,
+      "loss": 1.9388,
+      "step": 82
+    },
+    {
+      "epoch": 0.09987966305655836,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.9511732720980156e-05,
+      "loss": 1.7736,
+      "step": 83
+    },
+    {
+      "epoch": 0.10108303249097472,
+      "grad_norm": 2.375,
+      "learning_rate": 1.949999599387373e-05,
+      "loss": 1.8761,
+      "step": 84
+    },
+    {
+      "epoch": 0.1022864019253911,
+      "grad_norm": 7.71875,
+      "learning_rate": 1.9488123491519935e-05,
+      "loss": 1.8785,
+      "step": 85
+    },
+    {
+      "epoch": 0.10348977135980746,
+      "grad_norm": 2.125,
+      "learning_rate": 1.94761153836022e-05,
+      "loss": 1.9391,
+      "step": 86
+    },
+    {
+      "epoch": 0.10469314079422383,
+      "grad_norm": 7.875,
+      "learning_rate": 1.9463971841742057e-05,
+      "loss": 1.3668,
+      "step": 87
+    },
+    {
+      "epoch": 0.10589651022864019,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.9451693039496665e-05,
+      "loss": 1.9245,
+      "step": 88
+    },
+    {
+      "epoch": 0.10709987966305656,
+      "grad_norm": 8.5,
+      "learning_rate": 1.9439279152356363e-05,
+      "loss": 2.1398,
+      "step": 89
+    },
+    {
+      "epoch": 0.10830324909747292,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.9426730357742123e-05,
+      "loss": 1.9277,
+      "step": 90
+    },
+    {
+      "epoch": 0.1095066185318893,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.9414046835003043e-05,
+      "loss": 1.6924,
+      "step": 91
+    },
+    {
+      "epoch": 0.11070998796630566,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.9401228765413774e-05,
+      "loss": 1.8802,
+      "step": 92
+    },
+    {
+      "epoch": 0.11191335740072202,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.938827633217193e-05,
+      "loss": 2.0444,
+      "step": 93
+    },
+    {
+      "epoch": 0.11311672683513839,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.9375189720395454e-05,
+      "loss": 1.8791,
+      "step": 94
+    },
+    {
+      "epoch": 0.11432009626955475,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.936196911712001e-05,
+      "loss": 1.5531,
+      "step": 95
+    },
+    {
+      "epoch": 0.11552346570397112,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.934861471129627e-05,
+      "loss": 1.886,
+      "step": 96
+    },
+    {
+      "epoch": 0.11672683513838748,
+      "grad_norm": 12.375,
+      "learning_rate": 1.9335126693787237e-05,
+      "loss": 1.464,
+      "step": 97
+    },
+    {
+      "epoch": 0.11793020457280386,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.9321505257365508e-05,
+      "loss": 1.8344,
+      "step": 98
+    },
+    {
+      "epoch": 0.11913357400722022,
+      "grad_norm": 9.25,
+      "learning_rate": 1.930775059671053e-05,
+      "loss": 2.2084,
+      "step": 99
+    },
+    {
+      "epoch": 0.12033694344163658,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.9293862908405795e-05,
+      "loss": 1.8806,
+      "step": 100
+    },
+    {
+      "epoch": 0.12154031287605295,
+      "grad_norm": 8.125,
+      "learning_rate": 1.927984239093605e-05,
+      "loss": 1.5898,
+      "step": 101
+    },
+    {
+      "epoch": 0.12274368231046931,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.926568924468446e-05,
+      "loss": 1.8556,
+      "step": 102
+    },
+    {
+      "epoch": 0.12394705174488568,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.9251403671929738e-05,
+      "loss": 1.7307,
+      "step": 103
+    },
+    {
+      "epoch": 0.12515042117930206,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.9236985876843243e-05,
+      "loss": 1.8625,
+      "step": 104
+    },
+    {
+      "epoch": 0.1263537906137184,
+      "grad_norm": 7.875,
+      "learning_rate": 1.922243606548609e-05,
+      "loss": 2.0208,
+      "step": 105
+    },
+    {
+      "epoch": 0.12755716004813478,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.9207754445806176e-05,
+      "loss": 1.9865,
+      "step": 106
+    },
+    {
+      "epoch": 0.12876052948255115,
+      "grad_norm": 10.0,
+      "learning_rate": 1.9192941227635232e-05,
+      "loss": 1.6236,
+      "step": 107
+    },
+    {
+      "epoch": 0.1299638989169675,
+      "grad_norm": 3.140625,
+      "learning_rate": 1.91779966226858e-05,
+      "loss": 1.9087,
+      "step": 108
+    },
+    {
+      "epoch": 0.13116726835138387,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.9162920844548227e-05,
+      "loss": 1.8531,
+      "step": 109
+    },
+    {
+      "epoch": 0.13237063778580024,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.914771410868761e-05,
+      "loss": 1.8767,
+      "step": 110
+    },
+    {
+      "epoch": 0.13357400722021662,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.91323766324407e-05,
+      "loss": 1.867,
+      "step": 111
+    },
+    {
+      "epoch": 0.13477737665463296,
+      "grad_norm": 2.46875,
+      "learning_rate": 1.9116908635012813e-05,
+      "loss": 1.8482,
+      "step": 112
+    },
+    {
+      "epoch": 0.13598074608904934,
+      "grad_norm": 7.53125,
+      "learning_rate": 1.91013103374747e-05,
+      "loss": 1.7155,
+      "step": 113
+    },
+    {
+      "epoch": 0.1371841155234657,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.9085581962759366e-05,
+      "loss": 1.9164,
+      "step": 114
+    },
+    {
+      "epoch": 0.13838748495788206,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.9069723735658903e-05,
+      "loss": 1.7375,
+      "step": 115
+    },
+    {
+      "epoch": 0.13959085439229843,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.905373588282127e-05,
+      "loss": 1.9173,
+      "step": 116
+    },
+    {
+      "epoch": 0.1407942238267148,
+      "grad_norm": 11.875,
+      "learning_rate": 1.903761863274706e-05,
+      "loss": 2.1522,
+      "step": 117
+    },
+    {
+      "epoch": 0.14199759326113118,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.9021372215786218e-05,
+      "loss": 1.9097,
+      "step": 118
+    },
+    {
+      "epoch": 0.14320096269554752,
+      "grad_norm": 11.875,
+      "learning_rate": 1.9004996864134767e-05,
+      "loss": 1.7723,
+      "step": 119
+    },
+    {
+      "epoch": 0.1444043321299639,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.8988492811831485e-05,
+      "loss": 1.8475,
+      "step": 120
+    },
+    {
+      "epoch": 0.14560770156438027,
+      "grad_norm": 8.125,
+      "learning_rate": 1.8971860294754554e-05,
+      "loss": 1.8492,
+      "step": 121
+    },
+    {
+      "epoch": 0.14681107099879662,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.8955099550618194e-05,
+      "loss": 1.8399,
+      "step": 122
+    },
+    {
+      "epoch": 0.148014440433213,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.8938210818969257e-05,
+      "loss": 2.1739,
+      "step": 123
+    },
+    {
+      "epoch": 0.14921780986762936,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.8921194341183815e-05,
+      "loss": 1.8505,
+      "step": 124
+    },
+    {
+      "epoch": 0.15042117930204574,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.8904050360463708e-05,
+      "loss": 1.9786,
+      "step": 125
+    },
+    {
+      "epoch": 0.15162454873646208,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.8886779121833065e-05,
+      "loss": 1.8304,
+      "step": 126
+    },
+    {
+      "epoch": 0.15282791817087846,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.886938087213479e-05,
+      "loss": 1.6926,
+      "step": 127
+    },
+    {
+      "epoch": 0.15403128760529483,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.885185586002707e-05,
+      "loss": 1.8204,
+      "step": 128
+    },
+    {
+      "epoch": 0.1552346570397112,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.8834204335979777e-05,
+      "loss": 1.8364,
+      "step": 129
+    },
+    {
+      "epoch": 0.15643802647412755,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.8816426552270922e-05,
+      "loss": 1.8379,
+      "step": 130
+    },
+    {
+      "epoch": 0.15764139590854392,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.8798522762983026e-05,
+      "loss": 1.5707,
+      "step": 131
+    },
+    {
+      "epoch": 0.1588447653429603,
+      "grad_norm": 1.703125,
+      "learning_rate": 1.8780493223999508e-05,
+      "loss": 1.8673,
+      "step": 132
+    },
+    {
+      "epoch": 0.16004813477737664,
+      "grad_norm": 8.625,
+      "learning_rate": 1.8762338193001013e-05,
+      "loss": 2.0149,
+      "step": 133
+    },
+    {
+      "epoch": 0.16125150421179302,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.8744057929461736e-05,
+      "loss": 1.8588,
+      "step": 134
+    },
+    {
+      "epoch": 0.1624548736462094,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.8725652694645714e-05,
+      "loss": 1.6913,
+      "step": 135
+    },
+    {
+      "epoch": 0.16365824308062576,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.8707122751603098e-05,
+      "loss": 1.8392,
+      "step": 136
+    },
+    {
+      "epoch": 0.1648616125150421,
+      "grad_norm": 10.0,
+      "learning_rate": 1.868846836516637e-05,
+      "loss": 1.7601,
+      "step": 137
+    },
+    {
+      "epoch": 0.16606498194945848,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.8669689801946585e-05,
+      "loss": 1.7836,
+      "step": 138
+    },
+    {
+      "epoch": 0.16726835138387486,
+      "grad_norm": 12.0,
+      "learning_rate": 1.8650787330329546e-05,
+      "loss": 1.9955,
+      "step": 139
+    },
+    {
+      "epoch": 0.1684717208182912,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.863176122047198e-05,
+      "loss": 1.8541,
+      "step": 140
+    },
+    {
+      "epoch": 0.16967509025270758,
+      "grad_norm": 15.75,
+      "learning_rate": 1.861261174429765e-05,
+      "loss": 1.9102,
+      "step": 141
+    },
+    {
+      "epoch": 0.17087845968712395,
+      "grad_norm": 2.734375,
+      "learning_rate": 1.8593339175493514e-05,
+      "loss": 1.9752,
+      "step": 142
+    },
+    {
+      "epoch": 0.17208182912154033,
+      "grad_norm": 8.0,
+      "learning_rate": 1.8573943789505762e-05,
+      "loss": 1.7416,
+      "step": 143
+    },
+    {
+      "epoch": 0.17328519855595667,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.8554425863535915e-05,
+      "loss": 1.8621,
+      "step": 144
+    },
+    {
+      "epoch": 0.17448856799037304,
+      "grad_norm": 9.875,
+      "learning_rate": 1.8534785676536856e-05,
+      "loss": 1.4806,
+      "step": 145
+    },
+    {
+      "epoch": 0.17569193742478942,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.851502350920883e-05,
+      "loss": 1.8894,
+      "step": 146
+    },
+    {
+      "epoch": 0.17689530685920576,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.849513964399545e-05,
+      "loss": 1.5029,
+      "step": 147
+    },
+    {
+      "epoch": 0.17809867629362214,
+      "grad_norm": 2.125,
+      "learning_rate": 1.8475134365079642e-05,
+      "loss": 1.8094,
+      "step": 148
+    },
+    {
+      "epoch": 0.1793020457280385,
+      "grad_norm": 9.5,
+      "learning_rate": 1.8455007958379604e-05,
+      "loss": 1.325,
+      "step": 149
+    },
+    {
+      "epoch": 0.18050541516245489,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.8434760711544707e-05,
+      "loss": 1.8688,
+      "step": 150
+    },
+    {
+      "epoch": 0.18170878459687123,
+      "grad_norm": 8.25,
+      "learning_rate": 1.8414392913951382e-05,
+      "loss": 1.7627,
+      "step": 151
+    },
+    {
+      "epoch": 0.1829121540312876,
+      "grad_norm": 2.0,
+      "learning_rate": 1.8393904856698987e-05,
+      "loss": 1.843,
+      "step": 152
+    },
+    {
+      "epoch": 0.18411552346570398,
+      "grad_norm": 9.9375,
+      "learning_rate": 1.8373296832605647e-05,
+      "loss": 1.5985,
+      "step": 153
+    },
+    {
+      "epoch": 0.18531889290012032,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.835256913620408e-05,
+      "loss": 1.8304,
+      "step": 154
+    },
+    {
+      "epoch": 0.1865222623345367,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.8331722063737365e-05,
+      "loss": 1.7801,
+      "step": 155
+    },
+    {
+      "epoch": 0.18772563176895307,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.8310755913154726e-05,
+      "loss": 1.8439,
+      "step": 156
+    },
+    {
+      "epoch": 0.18892900120336945,
+      "grad_norm": 8.125,
+      "learning_rate": 1.8289670984107263e-05,
+      "loss": 1.5793,
+      "step": 157
+    },
+    {
+      "epoch": 0.1901323706377858,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.826846757794368e-05,
+      "loss": 1.8254,
+      "step": 158
+    },
+    {
+      "epoch": 0.19133574007220217,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.8247145997705977e-05,
+      "loss": 1.5562,
+      "step": 159
+    },
+    {
+      "epoch": 0.19253910950661854,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.8225706548125094e-05,
+      "loss": 1.8317,
+      "step": 160
+    },
+    {
+      "epoch": 0.19374247894103488,
+      "grad_norm": 8.125,
+      "learning_rate": 1.8204149535616596e-05,
+      "loss": 1.7808,
+      "step": 161
+    },
+    {
+      "epoch": 0.19494584837545126,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.8182475268276265e-05,
+      "loss": 1.8565,
+      "step": 162
+    },
+    {
+      "epoch": 0.19614921780986763,
+      "grad_norm": 9.0,
+      "learning_rate": 1.8160684055875704e-05,
+      "loss": 1.6122,
+      "step": 163
+    },
+    {
+      "epoch": 0.197352587244284,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.813877620985792e-05,
+      "loss": 1.8802,
+      "step": 164
+    },
+    {
+      "epoch": 0.19855595667870035,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.8116752043332848e-05,
+      "loss": 1.8362,
+      "step": 165
+    },
+    {
+      "epoch": 0.19975932611311673,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.8094611871072906e-05,
+      "loss": 1.9092,
+      "step": 166
+    },
+    {
+      "epoch": 0.2009626955475331,
+      "grad_norm": 10.25,
+      "learning_rate": 1.8072356009508473e-05,
+      "loss": 2.1806,
+      "step": 167
+    },
+    {
+      "epoch": 0.20216606498194944,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.8049984776723383e-05,
+      "loss": 1.8313,
+      "step": 168
+    },
+    {
+      "epoch": 0.20336943441636582,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.8027498492450367e-05,
+      "loss": 1.8958,
+      "step": 169
+    },
+    {
+      "epoch": 0.2045728038507822,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.8004897478066482e-05,
+      "loss": 1.8194,
+      "step": 170
+    },
+    {
+      "epoch": 0.20577617328519857,
+      "grad_norm": 8.875,
+      "learning_rate": 1.7982182056588536e-05,
+      "loss": 1.4826,
+      "step": 171
+    },
+    {
+      "epoch": 0.2069795427196149,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.795935255266845e-05,
+      "loss": 1.8559,
+      "step": 172
+    },
+    {
+      "epoch": 0.20818291215403129,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.7936409292588627e-05,
+      "loss": 1.6988,
+      "step": 173
+    },
+    {
+      "epoch": 0.20938628158844766,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.791335260425729e-05,
+      "loss": 1.8318,
+      "step": 174
+    },
+    {
+      "epoch": 0.21058965102286403,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.7890182817203806e-05,
+      "loss": 1.7352,
+      "step": 175
+    },
+    {
+      "epoch": 0.21179302045728038,
+      "grad_norm": 2.65625,
+      "learning_rate": 1.786690026257394e-05,
+      "loss": 1.8497,
+      "step": 176
+    },
+    {
+      "epoch": 0.21299638989169675,
+      "grad_norm": 6.6875,
+      "learning_rate": 1.7843505273125164e-05,
+      "loss": 1.4856,
+      "step": 177
+    },
+    {
+      "epoch": 0.21419975932611313,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.7819998183221883e-05,
+      "loss": 1.8217,
+      "step": 178
+    },
+    {
+      "epoch": 0.21540312876052947,
+      "grad_norm": 9.375,
+      "learning_rate": 1.7796379328830652e-05,
+      "loss": 1.9262,
+      "step": 179
+    },
+    {
+      "epoch": 0.21660649819494585,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7772649047515384e-05,
+      "loss": 1.8533,
+      "step": 180
+    },
+    {
+      "epoch": 0.21780986762936222,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.7748807678432514e-05,
+      "loss": 1.869,
+      "step": 181
+    },
+    {
+      "epoch": 0.2190132370637786,
+      "grad_norm": 3.3125,
+      "learning_rate": 1.7724855562326167e-05,
+      "loss": 1.7709,
+      "step": 182
+    },
+    {
+      "epoch": 0.22021660649819494,
+      "grad_norm": 6.625,
+      "learning_rate": 1.7700793041523272e-05,
+      "loss": 1.3658,
+      "step": 183
+    },
+    {
+      "epoch": 0.2214199759326113,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.7676620459928683e-05,
+      "loss": 1.9054,
+      "step": 184
+    },
+    {
+      "epoch": 0.2226233453670277,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.7652338163020257e-05,
+      "loss": 1.9153,
+      "step": 185
+    },
+    {
+      "epoch": 0.22382671480144403,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7627946497843917e-05,
+      "loss": 1.9365,
+      "step": 186
+    },
+    {
+      "epoch": 0.2250300842358604,
+      "grad_norm": 7.15625,
+      "learning_rate": 1.7603445813008685e-05,
+      "loss": 1.5009,
+      "step": 187
+    },
+    {
+      "epoch": 0.22623345367027678,
+      "grad_norm": 2.765625,
+      "learning_rate": 1.7578836458681718e-05,
+      "loss": 1.862,
+      "step": 188
+    },
+    {
+      "epoch": 0.22743682310469315,
+      "grad_norm": 11.0,
+      "learning_rate": 1.755411878658329e-05,
+      "loss": 1.7548,
+      "step": 189
+    },
+    {
+      "epoch": 0.2286401925391095,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.7529293149981758e-05,
+      "loss": 1.8371,
+      "step": 190
+    },
+    {
+      "epoch": 0.22984356197352587,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.7504359903688537e-05,
+      "loss": 1.9448,
+      "step": 191
+    },
+    {
+      "epoch": 0.23104693140794225,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.7479319404053004e-05,
+      "loss": 1.8082,
+      "step": 192
+    },
+    {
+      "epoch": 0.2322503008423586,
+      "grad_norm": 9.5,
+      "learning_rate": 1.7454172008957417e-05,
+      "loss": 1.55,
+      "step": 193
+    },
+    {
+      "epoch": 0.23345367027677497,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.7428918077811802e-05,
+      "loss": 1.8567,
+      "step": 194
+    },
+    {
+      "epoch": 0.23465703971119134,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.740355797154881e-05,
+      "loss": 1.6953,
+      "step": 195
+    },
+    {
+      "epoch": 0.2358604091456077,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.7378092052618565e-05,
+      "loss": 1.781,
+      "step": 196
+    },
+    {
+      "epoch": 0.23706377858002406,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.7352520684983474e-05,
+      "loss": 1.9668,
+      "step": 197
+    },
+    {
+      "epoch": 0.23826714801444043,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7326844234113037e-05,
+      "loss": 1.8419,
+      "step": 198
+    },
+    {
+      "epoch": 0.2394705174488568,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.7301063066978617e-05,
+      "loss": 1.7109,
+      "step": 199
+    },
+    {
+      "epoch": 0.24067388688327315,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.727517755204819e-05,
+      "loss": 1.8454,
+      "step": 200
+    },
+    {
+      "epoch": 0.24187725631768953,
+      "grad_norm": 7.65625,
+      "learning_rate": 1.72491880592811e-05,
+      "loss": 1.6736,
+      "step": 201
+    },
+    {
+      "epoch": 0.2430806257521059,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.7223094960122733e-05,
+      "loss": 1.8484,
+      "step": 202
+    },
+    {
+      "epoch": 0.24428399518652227,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.719689862749926e-05,
+      "loss": 2.0111,
+      "step": 203
+    },
+    {
+      "epoch": 0.24548736462093862,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.7170599435812253e-05,
+      "loss": 1.8034,
+      "step": 204
+    },
+    {
+      "epoch": 0.246690734055355,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.714419776093338e-05,
+      "loss": 1.6535,
+      "step": 205
+    },
+    {
+      "epoch": 0.24789410348977137,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.7117693980198996e-05,
+      "loss": 1.7763,
+      "step": 206
+    },
+    {
+      "epoch": 0.2490974729241877,
+      "grad_norm": 9.5625,
+      "learning_rate": 1.709108847240478e-05,
+      "loss": 1.4858,
+      "step": 207
+    },
+    {
+      "epoch": 0.2503008423586041,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7064381617800302e-05,
+      "loss": 1.8895,
+      "step": 208
+    },
+    {
+      "epoch": 0.25150421179302046,
+      "grad_norm": 9.375,
+      "learning_rate": 1.7037573798083598e-05,
+      "loss": 1.7483,
+      "step": 209
+    },
+    {
+      "epoch": 0.2527075812274368,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.7010665396395706e-05,
+      "loss": 1.8537,
+      "step": 210
+    },
+    {
+      "epoch": 0.2539109506618532,
+      "grad_norm": 7.625,
+      "learning_rate": 1.6983656797315197e-05,
+      "loss": 1.4985,
+      "step": 211
+    },
+    {
+      "epoch": 0.25511432009626955,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.6956548386852684e-05,
+      "loss": 1.8389,
+      "step": 212
+    },
+    {
+      "epoch": 0.2563176895306859,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.6929340552445283e-05,
+      "loss": 2.014,
+      "step": 213
+    },
+    {
+      "epoch": 0.2575210589651023,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6902033682951104e-05,
+      "loss": 1.8123,
+      "step": 214
+    },
+    {
+      "epoch": 0.25872442839951865,
+      "grad_norm": 8.75,
+      "learning_rate": 1.6874628168643683e-05,
+      "loss": 1.4427,
+      "step": 215
+    },
+    {
+      "epoch": 0.259927797833935,
+      "grad_norm": 1.703125,
+      "learning_rate": 1.6847124401206384e-05,
+      "loss": 1.8153,
+      "step": 216
+    },
+    {
+      "epoch": 0.2611311672683514,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.681952277372683e-05,
+      "loss": 1.7849,
+      "step": 217
+    },
+    {
+      "epoch": 0.26233453670276774,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.6791823680691276e-05,
+      "loss": 1.8759,
+      "step": 218
+    },
+    {
+      "epoch": 0.26353790613718414,
+      "grad_norm": 7.96875,
+      "learning_rate": 1.676402751797896e-05,
+      "loss": 1.5881,
+      "step": 219
+    },
+    {
+      "epoch": 0.2647412755716005,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.673613468285646e-05,
+      "loss": 1.7757,
+      "step": 220
+    },
+    {
+      "epoch": 0.26594464500601683,
+      "grad_norm": 9.125,
+      "learning_rate": 1.6708145573972005e-05,
+      "loss": 1.6333,
+      "step": 221
+    },
+    {
+      "epoch": 0.26714801444043323,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6680060591349774e-05,
+      "loss": 1.7887,
+      "step": 222
+    },
+    {
+      "epoch": 0.2683513838748496,
+      "grad_norm": 11.125,
+      "learning_rate": 1.6651880136384215e-05,
+      "loss": 1.8873,
+      "step": 223
+    },
+    {
+      "epoch": 0.2695547533092659,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.662360461183424e-05,
+      "loss": 1.8092,
+      "step": 224
+    },
+    {
+      "epoch": 0.27075812274368233,
+      "grad_norm": 8.5,
+      "learning_rate": 1.659523442181754e-05,
+      "loss": 1.7207,
+      "step": 225
+    },
+    {
+      "epoch": 0.2719614921780987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.6566769971804763e-05,
+      "loss": 1.7972,
+      "step": 226
+    },
+    {
+      "epoch": 0.273164861612515,
+      "grad_norm": 9.5,
+      "learning_rate": 1.653821166861374e-05,
+      "loss": 1.6875,
+      "step": 227
+    },
+    {
+      "epoch": 0.2743682310469314,
+      "grad_norm": 1.859375,
+      "learning_rate": 1.6509559920403663e-05,
+      "loss": 1.8418,
+      "step": 228
+    },
+    {
+      "epoch": 0.27557160048134777,
+      "grad_norm": 7.71875,
+      "learning_rate": 1.6480815136669248e-05,
+      "loss": 1.5936,
+      "step": 229
+    },
+    {
+      "epoch": 0.2767749699157641,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.6451977728234894e-05,
+      "loss": 1.8199,
+      "step": 230
+    },
+    {
+      "epoch": 0.2779783393501805,
+      "grad_norm": 8.75,
+      "learning_rate": 1.64230481072488e-05,
+      "loss": 1.7345,
+      "step": 231
+    },
+    {
+      "epoch": 0.27918170878459686,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.639402668717709e-05,
+      "loss": 1.8464,
+      "step": 232
+    },
+    {
+      "epoch": 0.28038507821901326,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.6364913882797875e-05,
+      "loss": 1.6354,
+      "step": 233
+    },
+    {
+      "epoch": 0.2815884476534296,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.633571011019536e-05,
+      "loss": 1.809,
+      "step": 234
+    },
+    {
+      "epoch": 0.28279181708784595,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.630641578675387e-05,
+      "loss": 1.9231,
+      "step": 235
+    },
+    {
+      "epoch": 0.28399518652226236,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.62770313311519e-05,
+      "loss": 1.8013,
+      "step": 236
+    },
+    {
+      "epoch": 0.2851985559566787,
+      "grad_norm": 8.75,
+      "learning_rate": 1.6247557163356127e-05,
+      "loss": 1.5578,
+      "step": 237
+    },
+    {
+      "epoch": 0.28640192539109505,
+      "grad_norm": 2.46875,
+      "learning_rate": 1.62179937046154e-05,
+      "loss": 1.77,
+      "step": 238
+    },
+    {
+      "epoch": 0.28760529482551145,
+      "grad_norm": 7.53125,
+      "learning_rate": 1.6188341377454735e-05,
+      "loss": 1.4921,
+      "step": 239
+    },
+    {
+      "epoch": 0.2888086642599278,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.6158600605669264e-05,
+      "loss": 1.7778,
+      "step": 240
+    },
+    {
+      "epoch": 0.29001203369434414,
+      "grad_norm": 10.875,
+      "learning_rate": 1.6128771814318178e-05,
+      "loss": 1.7163,
+      "step": 241
+    },
+    {
+      "epoch": 0.29121540312876054,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.6098855429718662e-05,
+      "loss": 1.7586,
+      "step": 242
+    },
+    {
+      "epoch": 0.2924187725631769,
+      "grad_norm": 8.5,
+      "learning_rate": 1.606885187943979e-05,
+      "loss": 1.5912,
+      "step": 243
+    },
+    {
+      "epoch": 0.29362214199759323,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.6038761592296435e-05,
+      "loss": 1.7934,
+      "step": 244
+    },
+    {
+      "epoch": 0.29482551143200963,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.60085849983431e-05,
+      "loss": 1.6993,
+      "step": 245
+    },
+    {
+      "epoch": 0.296028880866426,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.597832252886781e-05,
+      "loss": 1.8313,
+      "step": 246
+    },
+    {
+      "epoch": 0.2972322503008424,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.594797461638594e-05,
+      "loss": 1.8327,
+      "step": 247
+    },
+    {
+      "epoch": 0.29843561973525873,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.591754169463402e-05,
+      "loss": 1.747,
+      "step": 248
+    },
+    {
+      "epoch": 0.2996389891696751,
+      "grad_norm": 8.25,
+      "learning_rate": 1.5887024198563552e-05,
+      "loss": 1.5689,
+      "step": 249
+    },
+    {
+      "epoch": 0.3008423586040915,
+      "grad_norm": 1.625,
+      "learning_rate": 1.5856422564334772e-05,
+      "loss": 1.7359,
+      "step": 250
+    },
+    {
+      "epoch": 0.3020457280385078,
+      "grad_norm": 9.75,
+      "learning_rate": 1.5825737229310448e-05,
+      "loss": 1.636,
+      "step": 251
+    },
+    {
+      "epoch": 0.30324909747292417,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.5794968632049598e-05,
+      "loss": 1.8081,
+      "step": 252
+    },
+    {
+      "epoch": 0.30445246690734057,
+      "grad_norm": 7.59375,
+      "learning_rate": 1.576411721230124e-05,
+      "loss": 1.9024,
+      "step": 253
+    },
+    {
+      "epoch": 0.3056558363417569,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.57331834109981e-05,
+      "loss": 1.8062,
+      "step": 254
+    },
+    {
+      "epoch": 0.30685920577617326,
+      "grad_norm": 7.5,
+      "learning_rate": 1.570216767025032e-05,
+      "loss": 1.7436,
+      "step": 255
+    },
+    {
+      "epoch": 0.30806257521058966,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.5671070433339116e-05,
+      "loss": 1.8097,
+      "step": 256
+    },
+    {
+      "epoch": 0.309265944645006,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.5639892144710477e-05,
+      "loss": 1.5525,
+      "step": 257
+    },
+    {
+      "epoch": 0.3104693140794224,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.5608633249968783e-05,
+      "loss": 1.824,
+      "step": 258
+    },
+    {
+      "epoch": 0.31167268351383876,
+      "grad_norm": 8.125,
+      "learning_rate": 1.557729419587045e-05,
+      "loss": 1.8183,
+      "step": 259
+    },
+    {
+      "epoch": 0.3128760529482551,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5545875430317546e-05,
+      "loss": 1.7785,
+      "step": 260
+    },
+    {
+      "epoch": 0.3140794223826715,
+      "grad_norm": 9.125,
+      "learning_rate": 1.5514377402351376e-05,
+      "loss": 1.7519,
+      "step": 261
+    },
+    {
+      "epoch": 0.31528279181708785,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.548280056214609e-05,
+      "loss": 1.8864,
+      "step": 262
+    },
+    {
+      "epoch": 0.3164861612515042,
+      "grad_norm": 9.4375,
+      "learning_rate": 1.545114536100222e-05,
+      "loss": 1.7366,
+      "step": 263
+    },
+    {
+      "epoch": 0.3176895306859206,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.541941225134025e-05,
+      "loss": 1.8211,
+      "step": 264
+    },
+    {
+      "epoch": 0.31889290012033694,
+      "grad_norm": 7.5,
+      "learning_rate": 1.5387601686694134e-05,
+      "loss": 1.9038,
+      "step": 265
+    },
+    {
+      "epoch": 0.3200962695547533,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.5355714121704846e-05,
+      "loss": 1.7855,
+      "step": 266
+    },
+    {
+      "epoch": 0.3212996389891697,
+      "grad_norm": 7.5625,
+      "learning_rate": 1.532375001211383e-05,
+      "loss": 2.1078,
+      "step": 267
+    },
+    {
+      "epoch": 0.32250300842358604,
+      "grad_norm": 1.9140625,
+      "learning_rate": 1.529170981475653e-05,
+      "loss": 1.8535,
+      "step": 268
+    },
+    {
+      "epoch": 0.3237063778580024,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.525959398755585e-05,
+      "loss": 2.0128,
+      "step": 269
+    },
+    {
+      "epoch": 0.3249097472924188,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.5227402989515607e-05,
+      "loss": 1.8164,
+      "step": 270
+    },
+    {
+      "epoch": 0.32611311672683513,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.519513728071396e-05,
+      "loss": 1.7586,
+      "step": 271
+    },
+    {
+      "epoch": 0.32731648616125153,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.5162797322296855e-05,
+      "loss": 1.8448,
+      "step": 272
+    },
+    {
+      "epoch": 0.3285198555956679,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.5130383576471415e-05,
+      "loss": 1.5937,
+      "step": 273
+    },
+    {
+      "epoch": 0.3297232250300842,
+      "grad_norm": 1.875,
+      "learning_rate": 1.5097896506499349e-05,
+      "loss": 1.8645,
+      "step": 274
+    },
+    {
+      "epoch": 0.3309265944645006,
+      "grad_norm": 6.875,
+      "learning_rate": 1.5065336576690318e-05,
+      "loss": 1.5509,
+      "step": 275
+    },
+    {
+      "epoch": 0.33212996389891697,
+      "grad_norm": 1.6796875,
+      "learning_rate": 1.5032704252395315e-05,
+      "loss": 1.7767,
+      "step": 276
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.5383,
+      "step": 277
+    },
+    {
+      "epoch": 0.3345367027677497,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.496722428691804e-05,
+      "loss": 1.8775,
+      "step": 278
+    },
+    {
+      "epoch": 0.33574007220216606,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.4934377581584425e-05,
+      "loss": 1.6032,
+      "step": 279
+    },
+    {
+      "epoch": 0.3369434416365824,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.490146035344878e-05,
+      "loss": 1.8144,
+      "step": 280
+    },
+    {
+      "epoch": 0.3381468110709988,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.4868473072968645e-05,
+      "loss": 2.3891,
+      "step": 281
+    },
+    {
+      "epoch": 0.33935018050541516,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.4835416211602771e-05,
+      "loss": 1.7891,
+      "step": 282
+    },
+    {
+      "epoch": 0.3405535499398315,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.4802290241804355e-05,
+      "loss": 1.9642,
+      "step": 283
+    },
+    {
+      "epoch": 0.3417569193742479,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.4769095637014308e-05,
+      "loss": 1.784,
+      "step": 284
+    },
+    {
+      "epoch": 0.34296028880866425,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.473583287165448e-05,
+      "loss": 1.9608,
+      "step": 285
+    },
+    {
+      "epoch": 0.34416365824308065,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4702502421120884e-05,
+      "loss": 1.7739,
+      "step": 286
+    },
+    {
+      "epoch": 0.345367027677497,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.4669104761776892e-05,
+      "loss": 1.7284,
+      "step": 287
+    },
+    {
+      "epoch": 0.34657039711191334,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.463564037094644e-05,
+      "loss": 1.7732,
+      "step": 288
+    },
+    {
+      "epoch": 0.34777376654632974,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.4602109726907197e-05,
+      "loss": 1.5802,
+      "step": 289
+    },
+    {
+      "epoch": 0.3489771359807461,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.4568513308883732e-05,
+      "loss": 1.7499,
+      "step": 290
+    },
+    {
+      "epoch": 0.35018050541516244,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.4534851597040666e-05,
+      "loss": 1.7473,
+      "step": 291
+    },
+    {
+      "epoch": 0.35138387484957884,
+      "grad_norm": 2.375,
+      "learning_rate": 1.4501125072475804e-05,
+      "loss": 1.7983,
+      "step": 292
+    },
+    {
+      "epoch": 0.3525872442839952,
+      "grad_norm": 10.625,
+      "learning_rate": 1.4467334217213274e-05,
+      "loss": 2.0496,
+      "step": 293
+    },
+    {
+      "epoch": 0.35379061371841153,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.4433479514196615e-05,
+      "loss": 1.7396,
+      "step": 294
+    },
+    {
+      "epoch": 0.35499398315282793,
+      "grad_norm": 6.75,
+      "learning_rate": 1.439956144728189e-05,
+      "loss": 1.4716,
+      "step": 295
+    },
+    {
+      "epoch": 0.3561973525872443,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.4365580501230776e-05,
+      "loss": 1.8019,
+      "step": 296
+    },
+    {
+      "epoch": 0.3574007220216607,
+      "grad_norm": 9.875,
+      "learning_rate": 1.4331537161703612e-05,
+      "loss": 2.123,
+      "step": 297
+    },
+    {
+      "epoch": 0.358604091456077,
+      "grad_norm": 1.921875,
+      "learning_rate": 1.4297431915252487e-05,
+      "loss": 1.7762,
+      "step": 298
+    },
+    {
+      "epoch": 0.35980746089049337,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.4263265249314269e-05,
+      "loss": 1.5948,
+      "step": 299
+    },
+    {
+      "epoch": 0.36101083032490977,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.422903765220363e-05,
+      "loss": 1.8318,
+      "step": 300
+    },
+    {
+      "epoch": 0.3622141997593261,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.41947496131061e-05,
+      "loss": 1.6696,
+      "step": 301
+    },
+    {
+      "epoch": 0.36341756919374246,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4160401622071039e-05,
+      "loss": 1.7549,
+      "step": 302
+    },
+    {
+      "epoch": 0.36462093862815886,
+      "grad_norm": 9.75,
+      "learning_rate": 1.4125994170004644e-05,
+      "loss": 1.749,
+      "step": 303
+    },
+    {
+      "epoch": 0.3658243080625752,
+      "grad_norm": 2.6875,
+      "learning_rate": 1.4091527748662957e-05,
+      "loss": 1.8525,
+      "step": 304
+    },
+    {
+      "epoch": 0.36702767749699156,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.4057002850644796e-05,
+      "loss": 1.5852,
+      "step": 305
+    },
+    {
+      "epoch": 0.36823104693140796,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.402241996938475e-05,
+      "loss": 1.7885,
+      "step": 306
+    },
+    {
+      "epoch": 0.3694344163658243,
+      "grad_norm": 9.4375,
+      "learning_rate": 1.3987779599146105e-05,
+      "loss": 1.7472,
+      "step": 307
+    },
+    {
+      "epoch": 0.37063778580024065,
+      "grad_norm": 1.859375,
+      "learning_rate": 1.3953082235013788e-05,
+      "loss": 1.7533,
+      "step": 308
+    },
+    {
+      "epoch": 0.37184115523465705,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.3918328372887295e-05,
+      "loss": 1.6258,
+      "step": 309
+    },
+    {
+      "epoch": 0.3730445246690734,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.3883518509473598e-05,
+      "loss": 1.7508,
+      "step": 310
+    },
+    {
+      "epoch": 0.3742478941034898,
+      "grad_norm": 8.125,
+      "learning_rate": 1.3848653142280037e-05,
+      "loss": 1.9308,
+      "step": 311
+    },
+    {
+      "epoch": 0.37545126353790614,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.381373276960724e-05,
+      "loss": 1.8421,
+      "step": 312
+    },
+    {
+      "epoch": 0.3766546329723225,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.377875789054196e-05,
+      "loss": 1.6391,
+      "step": 313
+    },
+    {
+      "epoch": 0.3778580024067389,
+      "grad_norm": 1.6640625,
+      "learning_rate": 1.3743729004949972e-05,
+      "loss": 1.7977,
+      "step": 314
+    },
+    {
+      "epoch": 0.37906137184115524,
+      "grad_norm": 11.875,
+      "learning_rate": 1.3708646613468925e-05,
+      "loss": 1.6365,
+      "step": 315
+    },
+    {
+      "epoch": 0.3802647412755716,
+      "grad_norm": 1.9921875,
+      "learning_rate": 1.3673511217501172e-05,
+      "loss": 1.7404,
+      "step": 316
+    },
+    {
+      "epoch": 0.381468110709988,
+      "grad_norm": 8.25,
+      "learning_rate": 1.3638323319206617e-05,
+      "loss": 1.4718,
+      "step": 317
+    },
+    {
+      "epoch": 0.38267148014440433,
+      "grad_norm": 2.59375,
+      "learning_rate": 1.3603083421495535e-05,
+      "loss": 1.8494,
+      "step": 318
+    },
+    {
+      "epoch": 0.3838748495788207,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.3567792028021382e-05,
+      "loss": 1.7267,
+      "step": 319
+    },
+    {
+      "epoch": 0.3850782190132371,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.3532449643173604e-05,
+      "loss": 1.7702,
+      "step": 320
+    },
+    {
+      "epoch": 0.3862815884476534,
+      "grad_norm": 6.25,
+      "learning_rate": 1.3497056772070417e-05,
+      "loss": 1.6415,
+      "step": 321
+    },
+    {
+      "epoch": 0.38748495788206977,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.3461613920551598e-05,
+      "loss": 1.7866,
+      "step": 322
+    },
+    {
+      "epoch": 0.38868832731648617,
+      "grad_norm": 10.375,
+      "learning_rate": 1.3426121595171242e-05,
+      "loss": 1.7979,
+      "step": 323
+    },
+    {
+      "epoch": 0.3898916967509025,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.3390580303190541e-05,
+      "loss": 1.7856,
+      "step": 324
+    },
+    {
+      "epoch": 0.3910950661853189,
+      "grad_norm": 9.75,
+      "learning_rate": 1.335499055257052e-05,
+      "loss": 1.7654,
+      "step": 325
+    },
+    {
+      "epoch": 0.39229843561973526,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.3319352851964787e-05,
+      "loss": 1.8444,
+      "step": 326
+    },
+    {
+      "epoch": 0.3935018050541516,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.3283667710712245e-05,
+      "loss": 1.3942,
+      "step": 327
+    },
+    {
+      "epoch": 0.394705174488568,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.3247935638829838e-05,
+      "loss": 1.7739,
+      "step": 328
+    },
+    {
+      "epoch": 0.39590854392298436,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.3212157147005244e-05,
+      "loss": 1.7018,
+      "step": 329
+    },
+    {
+      "epoch": 0.3971119133574007,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.3176332746589587e-05,
+      "loss": 1.7783,
+      "step": 330
+    },
+    {
+      "epoch": 0.3983152827918171,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.3140462949590107e-05,
+      "loss": 2.0592,
+      "step": 331
+    },
+    {
+      "epoch": 0.39951865222623345,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.3104548268662873e-05,
+      "loss": 1.75,
+      "step": 332
+    },
+    {
+      "epoch": 0.4007220216606498,
+      "grad_norm": 10.375,
+      "learning_rate": 1.3068589217105441e-05,
+      "loss": 1.879,
+      "step": 333
+    },
+    {
+      "epoch": 0.4019253910950662,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.3032586308849512e-05,
+      "loss": 1.875,
+      "step": 334
+    },
+    {
+      "epoch": 0.40312876052948254,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.2996540058453589e-05,
+      "loss": 1.745,
+      "step": 335
+    },
+    {
+      "epoch": 0.4043321299638989,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.2960450981095643e-05,
+      "loss": 1.75,
+      "step": 336
+    },
+    {
+      "epoch": 0.4055354993983153,
+      "grad_norm": 10.375,
+      "learning_rate": 1.2924319592565713e-05,
+      "loss": 1.891,
+      "step": 337
+    },
+    {
+      "epoch": 0.40673886883273164,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.2888146409258575e-05,
+      "loss": 1.7498,
+      "step": 338
+    },
+    {
+      "epoch": 0.40794223826714804,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.2851931948166328e-05,
+      "loss": 1.7661,
+      "step": 339
+    },
+    {
+      "epoch": 0.4091456077015644,
+      "grad_norm": 2.125,
+      "learning_rate": 1.281567672687102e-05,
+      "loss": 1.7895,
+      "step": 340
+    },
+    {
+      "epoch": 0.41034897713598073,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.2779381263537262e-05,
+      "loss": 1.4515,
+      "step": 341
+    },
+    {
+      "epoch": 0.41155234657039713,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.2743046076904795e-05,
+      "loss": 1.7657,
+      "step": 342
+    },
+    {
+      "epoch": 0.4127557160048135,
+      "grad_norm": 11.625,
+      "learning_rate": 1.2706671686281094e-05,
+      "loss": 2.1425,
+      "step": 343
+    },
+    {
+      "epoch": 0.4139590854392298,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.2670258611533947e-05,
+      "loss": 1.7438,
+      "step": 344
+    },
+    {
+      "epoch": 0.4151624548736462,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.2633807373084022e-05,
+      "loss": 1.7648,
+      "step": 345
+    },
+    {
+      "epoch": 0.41636582430806257,
+      "grad_norm": 1.828125,
+      "learning_rate": 1.2597318491897416e-05,
+      "loss": 1.8152,
+      "step": 346
+    },
+    {
+      "epoch": 0.4175691937424789,
+      "grad_norm": 8.0,
+      "learning_rate": 1.2560792489478244e-05,
+      "loss": 1.6368,
+      "step": 347
+    },
+    {
+      "epoch": 0.4187725631768953,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.2524229887861132e-05,
+      "loss": 1.7436,
+      "step": 348
+    },
+    {
+      "epoch": 0.41997593261131166,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.2487631209603819e-05,
+      "loss": 1.6639,
+      "step": 349
+    },
+    {
+      "epoch": 0.42117930204572807,
+      "grad_norm": 1.921875,
+      "learning_rate": 1.245099697777963e-05,
+      "loss": 1.766,
+      "step": 350
+    },
+    {
+      "epoch": 0.4223826714801444,
+      "grad_norm": 8.125,
+      "learning_rate": 1.241432771597004e-05,
+      "loss": 1.4623,
+      "step": 351
+    },
+    {
+      "epoch": 0.42358604091456076,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.237762394825718e-05,
+      "loss": 1.8042,
+      "step": 352
+    },
+    {
+      "epoch": 0.42478941034897716,
+      "grad_norm": 7.8125,
+      "learning_rate": 1.234088619921633e-05,
+      "loss": 1.6498,
+      "step": 353
+    },
+    {
+      "epoch": 0.4259927797833935,
+      "grad_norm": 2.53125,
+      "learning_rate": 1.230411499390845e-05,
+      "loss": 1.7521,
+      "step": 354
+    },
+    {
+      "epoch": 0.42719614921780985,
+      "grad_norm": 7.59375,
+      "learning_rate": 1.2267310857872654e-05,
+      "loss": 1.6759,
+      "step": 355
+    },
+    {
+      "epoch": 0.42839951865222625,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.2230474317118708e-05,
+      "loss": 1.8461,
+      "step": 356
+    },
+    {
+      "epoch": 0.4296028880866426,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.2193605898119513e-05,
+      "loss": 1.7665,
+      "step": 357
+    },
+    {
+      "epoch": 0.43080625752105894,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.2156706127803578e-05,
+      "loss": 1.7748,
+      "step": 358
+    },
+    {
+      "epoch": 0.43200962695547535,
+      "grad_norm": 8.25,
+      "learning_rate": 1.2119775533547482e-05,
+      "loss": 1.6212,
+      "step": 359
+    },
+    {
+      "epoch": 0.4332129963898917,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.2082814643168357e-05,
+      "loss": 1.7921,
+      "step": 360
+    },
+    {
+      "epoch": 0.43441636582430804,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.2045823984916317e-05,
+      "loss": 1.6924,
+      "step": 361
+    },
+    {
+      "epoch": 0.43561973525872444,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.2008804087466931e-05,
+      "loss": 1.8069,
+      "step": 362
+    },
+    {
+      "epoch": 0.4368231046931408,
+      "grad_norm": 8.625,
+      "learning_rate": 1.1971755479913665e-05,
+      "loss": 1.656,
+      "step": 363
+    },
+    {
+      "epoch": 0.4380264741275572,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.1934678691760296e-05,
+      "loss": 1.7714,
+      "step": 364
+    },
+    {
+      "epoch": 0.43922984356197353,
+      "grad_norm": 9.375,
+      "learning_rate": 1.1897574252913377e-05,
+      "loss": 1.5384,
+      "step": 365
+    },
+    {
+      "epoch": 0.4404332129963899,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.1860442693674648e-05,
+      "loss": 1.7601,
+      "step": 366
+    },
+    {
+      "epoch": 0.4416365824308063,
+      "grad_norm": 10.125,
+      "learning_rate": 1.182328454473344e-05,
+      "loss": 1.7158,
+      "step": 367
+    },
+    {
+      "epoch": 0.4428399518652226,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.1786100337159132e-05,
+      "loss": 1.8104,
+      "step": 368
+    },
+    {
+      "epoch": 0.44404332129963897,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.1748890602393521e-05,
+      "loss": 1.9543,
+      "step": 369
+    },
+    {
+      "epoch": 0.4452466907340554,
+      "grad_norm": 1.625,
+      "learning_rate": 1.1711655872243247e-05,
+      "loss": 1.6937,
+      "step": 370
+    },
+    {
+      "epoch": 0.4464500601684717,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.1674396678872186e-05,
+      "loss": 1.8733,
+      "step": 371
+    },
+    {
+      "epoch": 0.44765342960288806,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.1637113554793846e-05,
+      "loss": 1.871,
+      "step": 372
+    },
+    {
+      "epoch": 0.44885679903730447,
+      "grad_norm": 7.1875,
+      "learning_rate": 1.1599807032863756e-05,
+      "loss": 1.685,
+      "step": 373
+    },
+    {
+      "epoch": 0.4500601684717208,
+      "grad_norm": 1.859375,
+      "learning_rate": 1.1562477646271856e-05,
+      "loss": 1.7347,
+      "step": 374
+    },
+    {
+      "epoch": 0.45126353790613716,
+      "grad_norm": 7.65625,
+      "learning_rate": 1.152512592853486e-05,
+      "loss": 1.7329,
+      "step": 375
+    },
+    {
+      "epoch": 0.45246690734055356,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.1487752413488646e-05,
+      "loss": 1.7809,
+      "step": 376
+    },
+    {
+      "epoch": 0.4536702767749699,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.1450357635280628e-05,
+      "loss": 1.4815,
+      "step": 377
+    },
+    {
+      "epoch": 0.4548736462093863,
+      "grad_norm": 1.875,
+      "learning_rate": 1.141294212836211e-05,
+      "loss": 1.7814,
+      "step": 378
+    },
+    {
+      "epoch": 0.45607701564380265,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.1375506427480658e-05,
+      "loss": 1.5316,
+      "step": 379
+    },
+    {
+      "epoch": 0.457280385078219,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.1338051067672444e-05,
+      "loss": 1.7501,
+      "step": 380
+    },
+    {
+      "epoch": 0.4584837545126354,
+      "grad_norm": 7.875,
+      "learning_rate": 1.1300576584254617e-05,
+      "loss": 1.7606,
+      "step": 381
+    },
+    {
+      "epoch": 0.45968712394705175,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.1263083512817644e-05,
+      "loss": 1.7445,
+      "step": 382
+    },
+    {
+      "epoch": 0.4608904933814681,
+      "grad_norm": 8.5,
+      "learning_rate": 1.1225572389217643e-05,
+      "loss": 1.7416,
+      "step": 383
+    },
+    {
+      "epoch": 0.4620938628158845,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.1188043749568752e-05,
+      "loss": 1.7761,
+      "step": 384
+    },
+    {
+      "epoch": 0.46329723225030084,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.1150498130235435e-05,
+      "loss": 1.8021,
+      "step": 385
+    },
+    {
+      "epoch": 0.4645006016847172,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.1112936067824847e-05,
+      "loss": 1.8076,
+      "step": 386
+    },
+    {
+      "epoch": 0.4657039711191336,
+      "grad_norm": 9.0,
+      "learning_rate": 1.1075358099179136e-05,
+      "loss": 1.8252,
+      "step": 387
+    },
+    {
+      "epoch": 0.46690734055354993,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.1037764761367795e-05,
+      "loss": 1.7513,
+      "step": 388
+    },
+    {
+      "epoch": 0.4681107099879663,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.1000156591679971e-05,
+      "loss": 1.6586,
+      "step": 389
+    },
+    {
+      "epoch": 0.4693140794223827,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.0962534127616784e-05,
+      "loss": 1.8155,
+      "step": 390
+    },
+    {
+      "epoch": 0.470517448856799,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.0924897906883663e-05,
+      "loss": 1.5809,
+      "step": 391
+    },
+    {
+      "epoch": 0.4717208182912154,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.088724846738264e-05,
+      "loss": 1.7288,
+      "step": 392
+    },
+    {
+      "epoch": 0.4729241877256318,
+      "grad_norm": 9.0,
+      "learning_rate": 1.0849586347204677e-05,
+      "loss": 1.8313,
+      "step": 393
+    },
+    {
+      "epoch": 0.4741275571600481,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.0811912084621968e-05,
+      "loss": 1.7223,
+      "step": 394
+    },
+    {
+      "epoch": 0.4753309265944645,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.0774226218080244e-05,
+      "loss": 1.5261,
+      "step": 395
+    },
+    {
+      "epoch": 0.47653429602888087,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.0736529286191087e-05,
+      "loss": 1.8465,
+      "step": 396
+    },
+    {
+      "epoch": 0.4777376654632972,
+      "grad_norm": 9.0,
+      "learning_rate": 1.0698821827724225e-05,
+      "loss": 1.6954,
+      "step": 397
+    },
+    {
+      "epoch": 0.4789410348977136,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.0661104381599833e-05,
+      "loss": 1.7999,
+      "step": 398
+    },
+    {
+      "epoch": 0.48014440433212996,
+      "grad_norm": 10.75,
+      "learning_rate": 1.0623377486880831e-05,
+      "loss": 1.7214,
+      "step": 399
+    },
+    {
+      "epoch": 0.4813477737665463,
+      "grad_norm": 1.9921875,
+      "learning_rate": 1.058564168276518e-05,
+      "loss": 1.7619,
+      "step": 400
+    },
+    {
+      "epoch": 0.4825511432009627,
+      "grad_norm": 11.375,
+      "learning_rate": 1.054789750857817e-05,
+      "loss": 1.9203,
+      "step": 401
+    },
+    {
+      "epoch": 0.48375451263537905,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.0510145503764727e-05,
+      "loss": 1.7714,
+      "step": 402
+    },
+    {
+      "epoch": 0.48495788206979545,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.0472386207881684e-05,
+      "loss": 1.748,
+      "step": 403
+    },
+    {
+      "epoch": 0.4861612515042118,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.0434620160590086e-05,
+      "loss": 1.7704,
+      "step": 404
+    },
+    {
+      "epoch": 0.48736462093862815,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.0396847901647469e-05,
+      "loss": 1.8945,
+      "step": 405
+    },
+    {
+      "epoch": 0.48856799037304455,
+      "grad_norm": 1.703125,
+      "learning_rate": 1.0359069970900139e-05,
+      "loss": 1.6917,
+      "step": 406
+    },
+    {
+      "epoch": 0.4897713598074609,
+      "grad_norm": 7.875,
+      "learning_rate": 1.0321286908275476e-05,
+      "loss": 1.6683,
+      "step": 407
+    },
+    {
+      "epoch": 0.49097472924187724,
+      "grad_norm": 1.921875,
+      "learning_rate": 1.0283499253774201e-05,
+      "loss": 1.7628,
+      "step": 408
+    },
+    {
+      "epoch": 0.49217809867629364,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.0245707547462654e-05,
+      "loss": 1.7848,
+      "step": 409
+    },
+    {
+      "epoch": 0.49338146811071,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.0207912329465097e-05,
+      "loss": 1.7637,
+      "step": 410
+    },
+    {
+      "epoch": 0.49458483754512633,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.0170114139955975e-05,
+      "loss": 1.8309,
+      "step": 411
+    },
+    {
+      "epoch": 0.49578820697954273,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.01323135191522e-05,
+      "loss": 1.7202,
+      "step": 412
+    },
+    {
+      "epoch": 0.4969915764139591,
+      "grad_norm": 7.9375,
+      "learning_rate": 1.0094511007305445e-05,
+      "loss": 1.4065,
+      "step": 413
+    },
+    {
+      "epoch": 0.4981949458483754,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.005670714469439e-05,
+      "loss": 1.8112,
+      "step": 414
+    },
+    {
+      "epoch": 0.4993983152827918,
+      "grad_norm": 8.75,
+      "learning_rate": 1.0018902471617037e-05,
+      "loss": 1.9798,
+      "step": 415
+    },
+    {
+      "epoch": 0.5006016847172082,
+      "grad_norm": 2.0,
+      "learning_rate": 9.981097528382964e-06,
+      "loss": 1.7527,
+      "step": 416
+    },
+    {
+      "epoch": 0.5018050541516246,
+      "grad_norm": 9.75,
+      "learning_rate": 9.943292855305611e-06,
+      "loss": 1.4855,
+      "step": 417
+    },
+    {
+      "epoch": 0.5030084235860409,
+      "grad_norm": 2.203125,
+      "learning_rate": 9.905488992694558e-06,
+      "loss": 1.7516,
+      "step": 418
+    },
+    {
+      "epoch": 0.5042117930204573,
+      "grad_norm": 7.21875,
+      "learning_rate": 9.867686480847801e-06,
+      "loss": 1.5753,
+      "step": 419
+    },
+    {
+      "epoch": 0.5054151624548736,
+      "grad_norm": 1.875,
+      "learning_rate": 9.829885860044028e-06,
+      "loss": 1.7993,
+      "step": 420
+    },
+    {
+      "epoch": 0.50661853188929,
+      "grad_norm": 7.6875,
+      "learning_rate": 9.792087670534908e-06,
+      "loss": 1.6177,
+      "step": 421
+    },
+    {
+      "epoch": 0.5078219013237064,
+      "grad_norm": 2.296875,
+      "learning_rate": 9.754292452537348e-06,
+      "loss": 1.7433,
+      "step": 422
+    },
+    {
+      "epoch": 0.5090252707581228,
+      "grad_norm": 9.375,
+      "learning_rate": 9.716500746225802e-06,
+      "loss": 1.6121,
+      "step": 423
+    },
+    {
+      "epoch": 0.5102286401925391,
+      "grad_norm": 3.15625,
+      "learning_rate": 9.678713091724527e-06,
+      "loss": 1.8498,
+      "step": 424
+    },
+    {
+      "epoch": 0.5114320096269555,
+      "grad_norm": 8.4375,
+      "learning_rate": 9.640930029099863e-06,
+      "loss": 2.0478,
+      "step": 425
+    },
+    {
+      "epoch": 0.5126353790613718,
+      "grad_norm": 1.8359375,
+      "learning_rate": 9.603152098352538e-06,
+      "loss": 1.7512,
+      "step": 426
+    },
+    {
+      "epoch": 0.5138387484957883,
+      "grad_norm": 8.75,
+      "learning_rate": 9.565379839409916e-06,
+      "loss": 1.9938,
+      "step": 427
+    },
+    {
+      "epoch": 0.5150421179302046,
+      "grad_norm": 1.8515625,
+      "learning_rate": 9.527613792118318e-06,
+      "loss": 1.7771,
+      "step": 428
+    },
+    {
+      "epoch": 0.516245487364621,
+      "grad_norm": 8.6875,
+      "learning_rate": 9.489854496235278e-06,
+      "loss": 1.6869,
+      "step": 429
+    },
+    {
+      "epoch": 0.5174488567990373,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.452102491421835e-06,
+      "loss": 1.7586,
+      "step": 430
+    },
+    {
+      "epoch": 0.5186522262334536,
+      "grad_norm": 8.625,
+      "learning_rate": 9.414358317234826e-06,
+      "loss": 1.4468,
+      "step": 431
+    },
+    {
+      "epoch": 0.51985559566787,
+      "grad_norm": 2.296875,
+      "learning_rate": 9.376622513119174e-06,
+      "loss": 1.8391,
+      "step": 432
+    },
+    {
+      "epoch": 0.5210589651022864,
+      "grad_norm": 7.71875,
+      "learning_rate": 9.338895618400168e-06,
+      "loss": 1.6689,
+      "step": 433
+    },
+    {
+      "epoch": 0.5222623345367028,
+      "grad_norm": 1.84375,
+      "learning_rate": 9.301178172275776e-06,
+      "loss": 1.793,
+      "step": 434
+    },
+    {
+      "epoch": 0.5234657039711191,
+      "grad_norm": 7.1875,
+      "learning_rate": 9.263470713808917e-06,
+      "loss": 1.7688,
+      "step": 435
+    },
+    {
+      "epoch": 0.5246690734055355,
+      "grad_norm": 1.8125,
+      "learning_rate": 9.22577378191976e-06,
+      "loss": 1.7634,
+      "step": 436
+    },
+    {
+      "epoch": 0.5258724428399518,
+      "grad_norm": 6.75,
+      "learning_rate": 9.188087915378037e-06,
+      "loss": 1.5367,
+      "step": 437
+    },
+    {
+      "epoch": 0.5270758122743683,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.150413652795325e-06,
+      "loss": 1.7128,
+      "step": 438
+    },
+    {
+      "epoch": 0.5282791817087846,
+      "grad_norm": 8.3125,
+      "learning_rate": 9.112751532617361e-06,
+      "loss": 1.776,
+      "step": 439
+    },
+    {
+      "epoch": 0.529482551143201,
+      "grad_norm": 1.7109375,
+      "learning_rate": 9.07510209311634e-06,
+      "loss": 1.7408,
+      "step": 440
+    },
+    {
+      "epoch": 0.5306859205776173,
+      "grad_norm": 8.875,
+      "learning_rate": 9.037465872383219e-06,
+      "loss": 1.732,
+      "step": 441
+    },
+    {
+      "epoch": 0.5318892900120337,
+      "grad_norm": 1.765625,
+      "learning_rate": 8.999843408320034e-06,
+      "loss": 1.7082,
+      "step": 442
+    },
+    {
+      "epoch": 0.53309265944645,
+      "grad_norm": 8.5625,
+      "learning_rate": 8.962235238632208e-06,
+      "loss": 1.4712,
+      "step": 443
+    },
+    {
+      "epoch": 0.5342960288808665,
+      "grad_norm": 2.15625,
+      "learning_rate": 8.924641900820864e-06,
+      "loss": 1.7716,
+      "step": 444
+    },
+    {
+      "epoch": 0.5354993983152828,
+      "grad_norm": 7.71875,
+      "learning_rate": 8.887063932175156e-06,
+      "loss": 1.729,
+      "step": 445
+    },
+    {
+      "epoch": 0.5367027677496992,
+      "grad_norm": 1.96875,
+      "learning_rate": 8.849501869764569e-06,
+      "loss": 1.7342,
+      "step": 446
+    },
+    {
+      "epoch": 0.5379061371841155,
+      "grad_norm": 9.3125,
+      "learning_rate": 8.811956250431253e-06,
+      "loss": 1.7085,
+      "step": 447
+    },
+    {
+      "epoch": 0.5391095066185319,
+      "grad_norm": 2.015625,
+      "learning_rate": 8.77442761078236e-06,
+      "loss": 1.77,
+      "step": 448
+    },
+    {
+      "epoch": 0.5403128760529483,
+      "grad_norm": 8.375,
+      "learning_rate": 8.73691648718236e-06,
+      "loss": 1.9969,
+      "step": 449
+    },
+    {
+      "epoch": 0.5415162454873647,
+      "grad_norm": 2.671875,
+      "learning_rate": 8.699423415745383e-06,
+      "loss": 1.7322,
+      "step": 450
+    },
+    {
+      "epoch": 0.542719614921781,
+      "grad_norm": 10.1875,
+      "learning_rate": 8.661948932327558e-06,
+      "loss": 1.7757,
+      "step": 451
+    },
+    {
+      "epoch": 0.5439229843561973,
+      "grad_norm": 1.6484375,
+      "learning_rate": 8.624493572519345e-06,
+      "loss": 1.7314,
+      "step": 452
+    },
+    {
+      "epoch": 0.5451263537906137,
+      "grad_norm": 10.25,
+      "learning_rate": 8.587057871637891e-06,
+      "loss": 1.517,
+      "step": 453
+    },
+    {
+      "epoch": 0.54632972322503,
+      "grad_norm": 1.9609375,
+      "learning_rate": 8.549642364719373e-06,
+      "loss": 1.8205,
+      "step": 454
+    },
+    {
+      "epoch": 0.5475330926594465,
+      "grad_norm": 19.75,
+      "learning_rate": 8.512247586511354e-06,
+      "loss": 2.0706,
+      "step": 455
+    },
+    {
+      "epoch": 0.5487364620938628,
+      "grad_norm": 2.078125,
+      "learning_rate": 8.474874071465144e-06,
+      "loss": 1.7824,
+      "step": 456
+    },
+    {
+      "epoch": 0.5499398315282792,
+      "grad_norm": 10.0625,
+      "learning_rate": 8.437522353728147e-06,
+      "loss": 1.5444,
+      "step": 457
+    },
+    {
+      "epoch": 0.5511432009626955,
+      "grad_norm": 2.09375,
+      "learning_rate": 8.400192967136245e-06,
+      "loss": 1.7308,
+      "step": 458
+    },
+    {
+      "epoch": 0.5523465703971119,
+      "grad_norm": 7.1875,
+      "learning_rate": 8.36288644520616e-06,
+      "loss": 1.3597,
+      "step": 459
+    },
+    {
+      "epoch": 0.5535499398315282,
+      "grad_norm": 1.875,
+      "learning_rate": 8.325603321127819e-06,
+      "loss": 1.7225,
+      "step": 460
+    },
+    {
+      "epoch": 0.5547533092659447,
+      "grad_norm": 9.25,
+      "learning_rate": 8.288344127756755e-06,
+      "loss": 1.15,
+      "step": 461
+    },
+    {
+      "epoch": 0.555956678700361,
+      "grad_norm": 1.75,
+      "learning_rate": 8.251109397606482e-06,
+      "loss": 1.8114,
+      "step": 462
+    },
+    {
+      "epoch": 0.5571600481347774,
+      "grad_norm": 13.5625,
+      "learning_rate": 8.213899662840871e-06,
+      "loss": 1.638,
+      "step": 463
+    },
+    {
+      "epoch": 0.5583634175691937,
+      "grad_norm": 2.140625,
+      "learning_rate": 8.176715455266564e-06,
+      "loss": 1.7039,
+      "step": 464
+    },
+    {
+      "epoch": 0.5595667870036101,
+      "grad_norm": 9.375,
+      "learning_rate": 8.139557306325359e-06,
+      "loss": 1.7151,
+      "step": 465
+    },
+    {
+      "epoch": 0.5607701564380265,
+      "grad_norm": 2.09375,
+      "learning_rate": 8.102425747086623e-06,
+      "loss": 1.7974,
+      "step": 466
+    },
+    {
+      "epoch": 0.5619735258724429,
+      "grad_norm": 8.6875,
+      "learning_rate": 8.065321308239706e-06,
+      "loss": 1.5067,
+      "step": 467
+    },
+    {
+      "epoch": 0.5631768953068592,
+      "grad_norm": 2.296875,
+      "learning_rate": 8.028244520086338e-06,
+      "loss": 1.7269,
+      "step": 468
+    },
+    {
+      "epoch": 0.5643802647412756,
+      "grad_norm": 10.6875,
+      "learning_rate": 7.99119591253307e-06,
+      "loss": 1.3304,
+      "step": 469
+    },
+    {
+      "epoch": 0.5655836341756919,
+      "grad_norm": 2.21875,
+      "learning_rate": 7.954176015083687e-06,
+      "loss": 1.7198,
+      "step": 470
+    },
+    {
+      "epoch": 0.5667870036101083,
+      "grad_norm": 11.3125,
+      "learning_rate": 7.91718535683165e-06,
+      "loss": 1.7933,
+      "step": 471
+    },
+    {
+      "epoch": 0.5679903730445247,
+      "grad_norm": 2.28125,
+      "learning_rate": 7.88022446645252e-06,
+      "loss": 1.7562,
+      "step": 472
+    },
+    {
+      "epoch": 0.5691937424789411,
+      "grad_norm": 8.6875,
+      "learning_rate": 7.843293872196425e-06,
+      "loss": 1.7976,
+      "step": 473
+    },
+    {
+      "epoch": 0.5703971119133574,
+      "grad_norm": 1.734375,
+      "learning_rate": 7.806394101880488e-06,
+      "loss": 1.8179,
+      "step": 474
+    },
+    {
+      "epoch": 0.5716004813477737,
+      "grad_norm": 8.6875,
+      "learning_rate": 7.769525682881295e-06,
+      "loss": 1.7221,
+      "step": 475
+    },
+    {
+      "epoch": 0.5728038507821901,
+      "grad_norm": 1.4921875,
+      "learning_rate": 7.73268914212735e-06,
+      "loss": 1.786,
+      "step": 476
+    },
+    {
+      "epoch": 0.5740072202166066,
+      "grad_norm": 9.625,
+      "learning_rate": 7.695885006091552e-06,
+      "loss": 1.8582,
+      "step": 477
+    },
+    {
+      "epoch": 0.5752105896510229,
+      "grad_norm": 2.015625,
+      "learning_rate": 7.659113800783672e-06,
+      "loss": 1.7797,
+      "step": 478
+    },
+    {
+      "epoch": 0.5764139590854392,
+      "grad_norm": 7.03125,
+      "learning_rate": 7.622376051742824e-06,
+      "loss": 1.1962,
+      "step": 479
+    },
+    {
+      "epoch": 0.5776173285198556,
+      "grad_norm": 1.9453125,
+      "learning_rate": 7.585672284029962e-06,
+      "loss": 1.7732,
+      "step": 480
+    },
+    {
+      "epoch": 0.5788206979542719,
+      "grad_norm": 8.0,
+      "learning_rate": 7.549003022220374e-06,
+      "loss": 1.4953,
+      "step": 481
+    },
+    {
+      "epoch": 0.5800240673886883,
+      "grad_norm": 3.171875,
+      "learning_rate": 7.512368790396186e-06,
+      "loss": 1.75,
+      "step": 482
+    },
+    {
+      "epoch": 0.5812274368231047,
+      "grad_norm": 8.25,
+      "learning_rate": 7.475770112138867e-06,
+      "loss": 1.7405,
+      "step": 483
+    },
+    {
+      "epoch": 0.5824308062575211,
+      "grad_norm": 1.796875,
+      "learning_rate": 7.43920751052176e-06,
+      "loss": 1.7973,
+      "step": 484
+    },
+    {
+      "epoch": 0.5836341756919374,
+      "grad_norm": 8.875,
+      "learning_rate": 7.402681508102585e-06,
+      "loss": 1.9704,
+      "step": 485
+    },
+    {
+      "epoch": 0.5848375451263538,
+      "grad_norm": 1.7578125,
+      "learning_rate": 7.366192626915982e-06,
+      "loss": 1.6921,
+      "step": 486
+    },
+    {
+      "epoch": 0.5860409145607701,
+      "grad_norm": 8.8125,
+      "learning_rate": 7.329741388466056e-06,
+      "loss": 1.6404,
+      "step": 487
+    },
+    {
+      "epoch": 0.5872442839951865,
+      "grad_norm": 1.703125,
+      "learning_rate": 7.293328313718912e-06,
+      "loss": 1.7947,
+      "step": 488
+    },
+    {
+      "epoch": 0.5884476534296029,
+      "grad_norm": 9.5625,
+      "learning_rate": 7.256953923095209e-06,
+      "loss": 1.6754,
+      "step": 489
+    },
+    {
+      "epoch": 0.5896510228640193,
+      "grad_norm": 1.84375,
+      "learning_rate": 7.220618736462739e-06,
+      "loss": 1.7515,
+      "step": 490
+    },
+    {
+      "epoch": 0.5908543922984356,
+      "grad_norm": 8.625,
+      "learning_rate": 7.184323273128981e-06,
+      "loss": 1.612,
+      "step": 491
+    },
+    {
+      "epoch": 0.592057761732852,
+      "grad_norm": 1.703125,
+      "learning_rate": 7.1480680518336766e-06,
+      "loss": 1.7887,
+      "step": 492
+    },
+    {
+      "epoch": 0.5932611311672683,
+      "grad_norm": 11.375,
+      "learning_rate": 7.111853590741431e-06,
+      "loss": 2.0435,
+      "step": 493
+    },
+    {
+      "epoch": 0.5944645006016848,
+      "grad_norm": 1.8828125,
+      "learning_rate": 7.075680407434289e-06,
+      "loss": 1.7261,
+      "step": 494
+    },
+    {
+      "epoch": 0.5956678700361011,
+      "grad_norm": 7.96875,
+      "learning_rate": 7.039549018904362e-06,
+      "loss": 1.8012,
+      "step": 495
+    },
+    {
+      "epoch": 0.5968712394705175,
+      "grad_norm": 1.7734375,
+      "learning_rate": 7.0034599415464135e-06,
+      "loss": 1.77,
+      "step": 496
+    },
+    {
+      "epoch": 0.5980746089049338,
+      "grad_norm": 10.625,
+      "learning_rate": 6.967413691150493e-06,
+      "loss": 2.0642,
+      "step": 497
+    },
+    {
+      "epoch": 0.5992779783393501,
+      "grad_norm": 2.828125,
+      "learning_rate": 6.931410782894563e-06,
+      "loss": 1.806,
+      "step": 498
+    },
+    {
+      "epoch": 0.6004813477737665,
+      "grad_norm": 11.8125,
+      "learning_rate": 6.895451731337129e-06,
+      "loss": 1.8019,
+      "step": 499
+    },
+    {
+      "epoch": 0.601684717208183,
+      "grad_norm": 2.484375,
+      "learning_rate": 6.859537050409895e-06,
+      "loss": 1.7507,
+      "step": 500
+    },
+    {
+      "epoch": 0.6028880866425993,
+      "grad_norm": 8.5625,
+      "learning_rate": 6.823667253410417e-06,
+      "loss": 1.726,
+      "step": 501
+    },
+    {
+      "epoch": 0.6040914560770156,
+      "grad_norm": 1.9375,
+      "learning_rate": 6.787842852994757e-06,
+      "loss": 1.7218,
+      "step": 502
+    },
+    {
+      "epoch": 0.605294825511432,
+      "grad_norm": 9.6875,
+      "learning_rate": 6.752064361170165e-06,
+      "loss": 1.4246,
+      "step": 503
+    },
+    {
+      "epoch": 0.6064981949458483,
+      "grad_norm": 2.078125,
+      "learning_rate": 6.716332289287759e-06,
+      "loss": 1.7802,
+      "step": 504
+    },
+    {
+      "epoch": 0.6077015643802648,
+      "grad_norm": 9.3125,
+      "learning_rate": 6.6806471480352175e-06,
+      "loss": 1.8945,
+      "step": 505
+    },
+    {
+      "epoch": 0.6089049338146811,
+      "grad_norm": 1.6328125,
+      "learning_rate": 6.64500944742948e-06,
+      "loss": 1.7197,
+      "step": 506
+    },
+    {
+      "epoch": 0.6101083032490975,
+      "grad_norm": 13.0,
+      "learning_rate": 6.609419696809463e-06,
+      "loss": 1.7151,
+      "step": 507
+    },
+    {
+      "epoch": 0.6113116726835138,
+      "grad_norm": 1.9609375,
+      "learning_rate": 6.5738784048287615e-06,
+      "loss": 1.7689,
+      "step": 508
+    },
+    {
+      "epoch": 0.6125150421179302,
+      "grad_norm": 13.1875,
+      "learning_rate": 6.5383860794484065e-06,
+      "loss": 1.9819,
+      "step": 509
+    },
+    {
+      "epoch": 0.6137184115523465,
+      "grad_norm": 2.359375,
+      "learning_rate": 6.502943227929586e-06,
+      "loss": 1.7717,
+      "step": 510
+    },
+    {
+      "epoch": 0.614921780986763,
+      "grad_norm": 6.9375,
+      "learning_rate": 6.4675503568263955e-06,
+      "loss": 1.5605,
+      "step": 511
+    },
+    {
+      "epoch": 0.6161251504211793,
+      "grad_norm": 1.6328125,
+      "learning_rate": 6.432207971978619e-06,
+      "loss": 1.7743,
+      "step": 512
+    },
+    {
+      "epoch": 0.6173285198555957,
+      "grad_norm": 7.875,
+      "learning_rate": 6.396916578504468e-06,
+      "loss": 1.5416,
+      "step": 513
+    },
+    {
+      "epoch": 0.618531889290012,
+      "grad_norm": 1.78125,
+      "learning_rate": 6.3616766807933875e-06,
+      "loss": 1.6756,
+      "step": 514
+    },
+    {
+      "epoch": 0.6197352587244284,
+      "grad_norm": 10.5625,
+      "learning_rate": 6.326488782498831e-06,
+      "loss": 1.5457,
+      "step": 515
+    },
+    {
+      "epoch": 0.6209386281588448,
+      "grad_norm": 1.9609375,
+      "learning_rate": 6.291353386531074e-06,
+      "loss": 1.705,
+      "step": 516
+    },
+    {
+      "epoch": 0.6221419975932612,
+      "grad_norm": 9.125,
+      "learning_rate": 6.256270995050026e-06,
+      "loss": 1.7022,
+      "step": 517
+    },
+    {
+      "epoch": 0.6233453670276775,
+      "grad_norm": 2.03125,
+      "learning_rate": 6.221242109458043e-06,
+      "loss": 1.825,
+      "step": 518
+    },
+    {
+      "epoch": 0.6245487364620939,
+      "grad_norm": 8.0,
+      "learning_rate": 6.186267230392762e-06,
+      "loss": 1.5879,
+      "step": 519
+    },
+    {
+      "epoch": 0.6257521058965102,
+      "grad_norm": 2.21875,
+      "learning_rate": 6.151346857719964e-06,
+      "loss": 1.8472,
+      "step": 520
+    },
+    {
+      "epoch": 0.6269554753309265,
+      "grad_norm": 6.375,
+      "learning_rate": 6.116481490526407e-06,
+      "loss": 1.6695,
+      "step": 521
+    },
+    {
+      "epoch": 0.628158844765343,
+      "grad_norm": 1.7890625,
+      "learning_rate": 6.081671627112704e-06,
+      "loss": 1.7865,
+      "step": 522
+    },
+    {
+      "epoch": 0.6293622141997594,
+      "grad_norm": 8.8125,
+      "learning_rate": 6.046917764986213e-06,
+      "loss": 1.6878,
+      "step": 523
+    },
+    {
+      "epoch": 0.6305655836341757,
+      "grad_norm": 1.78125,
+      "learning_rate": 6.012220400853899e-06,
+      "loss": 1.835,
+      "step": 524
+    },
+    {
+      "epoch": 0.631768953068592,
+      "grad_norm": 8.0625,
+      "learning_rate": 5.977580030615254e-06,
+      "loss": 1.5589,
+      "step": 525
+    },
+    {
+      "epoch": 0.6329723225030084,
+      "grad_norm": 2.09375,
+      "learning_rate": 5.942997149355208e-06,
+      "loss": 1.7318,
+      "step": 526
+    },
+    {
+      "epoch": 0.6341756919374247,
+      "grad_norm": 11.25,
+      "learning_rate": 5.9084722513370485e-06,
+      "loss": 1.6077,
+      "step": 527
+    },
+    {
+      "epoch": 0.6353790613718412,
+      "grad_norm": 1.5703125,
+      "learning_rate": 5.874005829995358e-06,
+      "loss": 1.7789,
+      "step": 528
+    },
+    {
+      "epoch": 0.6365824308062575,
+      "grad_norm": 7.6875,
+      "learning_rate": 5.839598377928964e-06,
+      "loss": 1.3707,
+      "step": 529
+    },
+    {
+      "epoch": 0.6377858002406739,
+      "grad_norm": 2.328125,
+      "learning_rate": 5.8052503868939005e-06,
+      "loss": 1.7943,
+      "step": 530
+    },
+    {
+      "epoch": 0.6389891696750902,
+      "grad_norm": 8.9375,
+      "learning_rate": 5.7709623477963696e-06,
+      "loss": 1.7155,
+      "step": 531
+    },
+    {
+      "epoch": 0.6401925391095066,
+      "grad_norm": 1.4921875,
+      "learning_rate": 5.736734750685737e-06,
+      "loss": 1.7964,
+      "step": 532
+    },
+    {
+      "epoch": 0.641395908543923,
+      "grad_norm": 7.53125,
+      "learning_rate": 5.702568084747513e-06,
+      "loss": 1.529,
+      "step": 533
+    },
+    {
+      "epoch": 0.6425992779783394,
+      "grad_norm": 2.0625,
+      "learning_rate": 5.6684628382963905e-06,
+      "loss": 1.7383,
+      "step": 534
+    },
+    {
+      "epoch": 0.6438026474127557,
+      "grad_norm": 8.875,
+      "learning_rate": 5.6344194987692304e-06,
+      "loss": 1.7915,
+      "step": 535
+    },
+    {
+      "epoch": 0.6450060168471721,
+      "grad_norm": 1.8671875,
+      "learning_rate": 5.60043855271811e-06,
+      "loss": 1.7456,
+      "step": 536
+    },
+    {
+      "epoch": 0.6462093862815884,
+      "grad_norm": 8.25,
+      "learning_rate": 5.566520485803388e-06,
+      "loss": 1.8949,
+      "step": 537
+    },
+    {
+      "epoch": 0.6474127557160048,
+      "grad_norm": 2.03125,
+      "learning_rate": 5.53266578278673e-06,
+      "loss": 1.7912,
+      "step": 538
+    },
+    {
+      "epoch": 0.6486161251504212,
+      "grad_norm": 11.3125,
+      "learning_rate": 5.498874927524196e-06,
+      "loss": 1.7826,
+      "step": 539
+    },
+    {
+      "epoch": 0.6498194945848376,
+      "grad_norm": 1.7890625,
+      "learning_rate": 5.465148402959339e-06,
+      "loss": 1.76,
+      "step": 540
+    },
+    {
+      "epoch": 0.6510228640192539,
+      "grad_norm": 6.96875,
+      "learning_rate": 5.43148669111627e-06,
+      "loss": 1.4643,
+      "step": 541
+    },
+    {
+      "epoch": 0.6522262334536703,
+      "grad_norm": 2.1875,
+      "learning_rate": 5.397890273092807e-06,
+      "loss": 1.7544,
+      "step": 542
+    },
+    {
+      "epoch": 0.6534296028880866,
+      "grad_norm": 8.5625,
+      "learning_rate": 5.364359629053566e-06,
+      "loss": 1.5244,
+      "step": 543
+    },
+    {
+      "epoch": 0.6546329723225031,
+      "grad_norm": 2.140625,
+      "learning_rate": 5.33089523822311e-06,
+      "loss": 1.7565,
+      "step": 544
+    },
+    {
+      "epoch": 0.6558363417569194,
+      "grad_norm": 7.40625,
+      "learning_rate": 5.29749757887912e-06,
+      "loss": 1.675,
+      "step": 545
+    },
+    {
+      "epoch": 0.6570397111913358,
+      "grad_norm": 2.171875,
+      "learning_rate": 5.264167128345523e-06,
+      "loss": 1.7383,
+      "step": 546
+    },
+    {
+      "epoch": 0.6582430806257521,
+      "grad_norm": 8.8125,
+      "learning_rate": 5.230904362985694e-06,
+      "loss": 1.6509,
+      "step": 547
+    },
+    {
+      "epoch": 0.6594464500601684,
+      "grad_norm": 1.8515625,
+      "learning_rate": 5.197709758195648e-06,
+      "loss": 1.7565,
+      "step": 548
+    },
+    {
+      "epoch": 0.6606498194945848,
+      "grad_norm": 8.9375,
+      "learning_rate": 5.164583788397234e-06,
+      "loss": 1.5013,
+      "step": 549
+    },
+    {
+      "epoch": 0.6618531889290012,
+      "grad_norm": 1.84375,
+      "learning_rate": 5.131526927031356e-06,
+      "loss": 1.7796,
+      "step": 550
+    },
+    {
+      "epoch": 0.6630565583634176,
+      "grad_norm": 7.0625,
+      "learning_rate": 5.098539646551226e-06,
+      "loss": 1.6768,
+      "step": 551
+    },
+    {
+      "epoch": 0.6642599277978339,
+      "grad_norm": 2.34375,
+      "learning_rate": 5.0656224184155764e-06,
+      "loss": 1.8201,
+      "step": 552
+    },
+    {
+      "epoch": 0.6654632972322503,
+      "grad_norm": 7.09375,
+      "learning_rate": 5.032775713081963e-06,
+      "loss": 1.6698,
+      "step": 553
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 1.9140625,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 1.7584,
+      "step": 554
+    },
+    {
+      "epoch": 0.6678700361010831,
+      "grad_norm": 8.125,
+      "learning_rate": 4.967295747604685e-06,
+      "loss": 1.7075,
+      "step": 555
+    },
+    {
+      "epoch": 0.6690734055354994,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.934663423309685e-06,
+      "loss": 1.8019,
+      "step": 556
+    },
+    {
+      "epoch": 0.6702767749699158,
+      "grad_norm": 8.9375,
+      "learning_rate": 4.902103493500654e-06,
+      "loss": 1.7358,
+      "step": 557
+    },
+    {
+      "epoch": 0.6714801444043321,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.869616423528588e-06,
+      "loss": 1.8208,
+      "step": 558
+    },
+    {
+      "epoch": 0.6726835138387485,
+      "grad_norm": 8.9375,
+      "learning_rate": 4.837202677703149e-06,
+      "loss": 1.66,
+      "step": 559
+    },
+    {
+      "epoch": 0.6738868832731648,
+      "grad_norm": 2.21875,
+      "learning_rate": 4.804862719286044e-06,
+      "loss": 1.7842,
+      "step": 560
+    },
+    {
+      "epoch": 0.6750902527075813,
+      "grad_norm": 9.6875,
+      "learning_rate": 4.772597010484396e-06,
+      "loss": 1.7979,
+      "step": 561
+    },
+    {
+      "epoch": 0.6762936221419976,
+      "grad_norm": 1.7890625,
+      "learning_rate": 4.740406012444153e-06,
+      "loss": 1.7748,
+      "step": 562
+    },
+    {
+      "epoch": 0.677496991576414,
+      "grad_norm": 9.75,
+      "learning_rate": 4.7082901852434734e-06,
+      "loss": 1.78,
+      "step": 563
+    },
+    {
+      "epoch": 0.6787003610108303,
+      "grad_norm": 2.125,
+      "learning_rate": 4.6762499878861764e-06,
+      "loss": 1.7373,
+      "step": 564
+    },
+    {
+      "epoch": 0.6799037304452467,
+      "grad_norm": 16.375,
+      "learning_rate": 4.644285878295161e-06,
+      "loss": 1.8036,
+      "step": 565
+    },
+    {
+      "epoch": 0.681107099879663,
+      "grad_norm": 1.875,
+      "learning_rate": 4.612398313305867e-06,
+      "loss": 1.8106,
+      "step": 566
+    },
+    {
+      "epoch": 0.6823104693140795,
+      "grad_norm": 20.75,
+      "learning_rate": 4.580587748659753e-06,
+      "loss": 1.8474,
+      "step": 567
+    },
+    {
+      "epoch": 0.6835138387484958,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.548854638997778e-06,
+      "loss": 1.8194,
+      "step": 568
+    },
+    {
+      "epoch": 0.6847172081829122,
+      "grad_norm": 8.5,
+      "learning_rate": 4.517199437853909e-06,
+      "loss": 1.7905,
+      "step": 569
+    },
+    {
+      "epoch": 0.6859205776173285,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.485622597648624e-06,
+      "loss": 1.8076,
+      "step": 570
+    },
+    {
+      "epoch": 0.6871239470517448,
+      "grad_norm": 7.09375,
+      "learning_rate": 4.454124569682459e-06,
+      "loss": 1.5847,
+      "step": 571
+    },
+    {
+      "epoch": 0.6883273164861613,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.4227058041295515e-06,
+      "loss": 1.7217,
+      "step": 572
+    },
+    {
+      "epoch": 0.6895306859205776,
+      "grad_norm": 7.40625,
+      "learning_rate": 4.391366750031217e-06,
+      "loss": 1.6804,
+      "step": 573
+    },
+    {
+      "epoch": 0.690734055354994,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.3601078552895245e-06,
+      "loss": 1.747,
+      "step": 574
+    },
+    {
+      "epoch": 0.6919374247894103,
+      "grad_norm": 8.875,
+      "learning_rate": 4.3289295666608865e-06,
+      "loss": 1.7562,
+      "step": 575
+    },
+    {
+      "epoch": 0.6931407942238267,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.297832329749687e-06,
+      "loss": 1.7482,
+      "step": 576
+    },
+    {
+      "epoch": 0.694344163658243,
+      "grad_norm": 10.0625,
+      "learning_rate": 4.2668165890019044e-06,
+      "loss": 1.6526,
+      "step": 577
+    },
+    {
+      "epoch": 0.6955475330926595,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.235882787698763e-06,
+      "loss": 1.7288,
+      "step": 578
+    },
+    {
+      "epoch": 0.6967509025270758,
+      "grad_norm": 8.8125,
+      "learning_rate": 4.205031367950402e-06,
+      "loss": 1.9839,
+      "step": 579
+    },
+    {
+      "epoch": 0.6979542719614922,
+      "grad_norm": 1.7109375,
+      "learning_rate": 4.174262770689552e-06,
+      "loss": 1.8001,
+      "step": 580
+    },
+    {
+      "epoch": 0.6991576413959085,
+      "grad_norm": 8.125,
+      "learning_rate": 4.143577435665229e-06,
+      "loss": 1.7003,
+      "step": 581
+    },
+    {
+      "epoch": 0.7003610108303249,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.112975801436454e-06,
+      "loss": 1.757,
+      "step": 582
+    },
+    {
+      "epoch": 0.7015643802647413,
+      "grad_norm": 9.0,
+      "learning_rate": 4.082458305365982e-06,
+      "loss": 1.8433,
+      "step": 583
+    },
+    {
+      "epoch": 0.7027677496991577,
+      "grad_norm": 1.9296875,
+      "learning_rate": 4.052025383614061e-06,
+      "loss": 1.7098,
+      "step": 584
+    },
+    {
+      "epoch": 0.703971119133574,
+      "grad_norm": 7.625,
+      "learning_rate": 4.021677471132192e-06,
+      "loss": 1.5424,
+      "step": 585
+    },
+    {
+      "epoch": 0.7051744885679904,
+      "grad_norm": 1.796875,
+      "learning_rate": 3.991415001656906e-06,
+      "loss": 1.7097,
+      "step": 586
+    },
+    {
+      "epoch": 0.7063778580024067,
+      "grad_norm": 7.9375,
+      "learning_rate": 3.9612384077035705e-06,
+      "loss": 1.8125,
+      "step": 587
+    },
+    {
+      "epoch": 0.7075812274368231,
+      "grad_norm": 1.734375,
+      "learning_rate": 3.931148120560211e-06,
+      "loss": 1.7072,
+      "step": 588
+    },
+    {
+      "epoch": 0.7087845968712395,
+      "grad_norm": 9.6875,
+      "learning_rate": 3.90114457028134e-06,
+      "loss": 1.7082,
+      "step": 589
+    },
+    {
+      "epoch": 0.7099879663056559,
+      "grad_norm": 2.421875,
+      "learning_rate": 3.871228185681822e-06,
+      "loss": 1.7654,
+      "step": 590
+    },
+    {
+      "epoch": 0.7111913357400722,
+      "grad_norm": 7.46875,
+      "learning_rate": 3.84139939433074e-06,
+      "loss": 1.66,
+      "step": 591
+    },
+    {
+      "epoch": 0.7123947051744886,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.811658622545268e-06,
+      "loss": 1.6727,
+      "step": 592
+    },
+    {
+      "epoch": 0.7135980746089049,
+      "grad_norm": 9.125,
+      "learning_rate": 3.782006295384604e-06,
+      "loss": 1.4466,
+      "step": 593
+    },
+    {
+      "epoch": 0.7148014440433214,
+      "grad_norm": 2.046875,
+      "learning_rate": 3.7524428366438757e-06,
+      "loss": 1.7239,
+      "step": 594
+    },
+    {
+      "epoch": 0.7160048134777377,
+      "grad_norm": 7.9375,
+      "learning_rate": 3.722968668848098e-06,
+      "loss": 1.8132,
+      "step": 595
+    },
+    {
+      "epoch": 0.717208182912154,
+      "grad_norm": 1.7890625,
+      "learning_rate": 3.6935842132461307e-06,
+      "loss": 1.7111,
+      "step": 596
+    },
+    {
+      "epoch": 0.7184115523465704,
+      "grad_norm": 10.25,
+      "learning_rate": 3.664289889804643e-06,
+      "loss": 1.9077,
+      "step": 597
+    },
+    {
+      "epoch": 0.7196149217809867,
+      "grad_norm": 2.828125,
+      "learning_rate": 3.635086117202128e-06,
+      "loss": 1.8594,
+      "step": 598
+    },
+    {
+      "epoch": 0.7208182912154031,
+      "grad_norm": 7.4375,
+      "learning_rate": 3.6059733128229125e-06,
+      "loss": 1.6636,
+      "step": 599
+    },
+    {
+      "epoch": 0.7220216606498195,
+      "grad_norm": 2.421875,
+      "learning_rate": 3.576951892751197e-06,
+      "loss": 1.7859,
+      "step": 600
+    },
+    {
+      "epoch": 0.7232250300842359,
+      "grad_norm": 7.28125,
+      "learning_rate": 3.548022271765107e-06,
+      "loss": 1.6168,
+      "step": 601
+    },
+    {
+      "epoch": 0.7244283995186522,
+      "grad_norm": 1.9140625,
+      "learning_rate": 3.5191848633307545e-06,
+      "loss": 1.7513,
+      "step": 602
+    },
+    {
+      "epoch": 0.7256317689530686,
+      "grad_norm": 6.96875,
+      "learning_rate": 3.490440079596341e-06,
+      "loss": 1.413,
+      "step": 603
+    },
+    {
+      "epoch": 0.7268351383874849,
+      "grad_norm": 2.125,
+      "learning_rate": 3.4617883313862633e-06,
+      "loss": 1.7607,
+      "step": 604
+    },
+    {
+      "epoch": 0.7280385078219013,
+      "grad_norm": 8.0625,
+      "learning_rate": 3.433230028195239e-06,
+      "loss": 1.4379,
+      "step": 605
+    },
+    {
+      "epoch": 0.7292418772563177,
+      "grad_norm": 2.21875,
+      "learning_rate": 3.4047655781824605e-06,
+      "loss": 1.7282,
+      "step": 606
+    },
+    {
+      "epoch": 0.7304452466907341,
+      "grad_norm": 8.3125,
+      "learning_rate": 3.376395388165762e-06,
+      "loss": 1.6588,
+      "step": 607
+    },
+    {
+      "epoch": 0.7316486161251504,
+      "grad_norm": 1.8359375,
+      "learning_rate": 3.3481198636157908e-06,
+      "loss": 1.7713,
+      "step": 608
+    },
+    {
+      "epoch": 0.7328519855595668,
+      "grad_norm": 7.75,
+      "learning_rate": 3.3199394086502257e-06,
+      "loss": 1.1923,
+      "step": 609
+    },
+    {
+      "epoch": 0.7340553549939831,
+      "grad_norm": 2.546875,
+      "learning_rate": 3.2918544260279985e-06,
+      "loss": 1.773,
+      "step": 610
+    },
+    {
+      "epoch": 0.7352587244283996,
+      "grad_norm": 8.625,
+      "learning_rate": 3.2638653171435387e-06,
+      "loss": 1.6534,
+      "step": 611
+    },
+    {
+      "epoch": 0.7364620938628159,
+      "grad_norm": 2.71875,
+      "learning_rate": 3.2359724820210394e-06,
+      "loss": 1.8335,
+      "step": 612
+    },
+    {
+      "epoch": 0.7376654632972323,
+      "grad_norm": 7.875,
+      "learning_rate": 3.2081763193087247e-06,
+      "loss": 1.5612,
+      "step": 613
+    },
+    {
+      "epoch": 0.7388688327316486,
+      "grad_norm": 1.84375,
+      "learning_rate": 3.180477226273172e-06,
+      "loss": 1.7428,
+      "step": 614
+    },
+    {
+      "epoch": 0.740072202166065,
+      "grad_norm": 8.25,
+      "learning_rate": 3.1528755987936188e-06,
+      "loss": 1.465,
+      "step": 615
+    },
+    {
+      "epoch": 0.7412755716004813,
+      "grad_norm": 2.21875,
+      "learning_rate": 3.1253718313563207e-06,
+      "loss": 1.7722,
+      "step": 616
+    },
+    {
+      "epoch": 0.7424789410348978,
+      "grad_norm": 8.4375,
+      "learning_rate": 3.097966317048895e-06,
+      "loss": 1.6246,
+      "step": 617
+    },
+    {
+      "epoch": 0.7436823104693141,
+      "grad_norm": 2.09375,
+      "learning_rate": 3.070659447554719e-06,
+      "loss": 1.7443,
+      "step": 618
+    },
+    {
+      "epoch": 0.7448856799037304,
+      "grad_norm": 9.1875,
+      "learning_rate": 3.0434516131473214e-06,
+      "loss": 1.6198,
+      "step": 619
+    },
+    {
+      "epoch": 0.7460890493381468,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3.016343202684807e-06,
+      "loss": 1.7547,
+      "step": 620
+    },
+    {
+      "epoch": 0.7472924187725631,
+      "grad_norm": 7.0625,
+      "learning_rate": 2.9893346036042968e-06,
+      "loss": 1.6661,
+      "step": 621
+    },
+    {
+      "epoch": 0.7484957882069796,
+      "grad_norm": 1.4765625,
+      "learning_rate": 2.962426201916402e-06,
+      "loss": 1.7177,
+      "step": 622
+    },
+    {
+      "epoch": 0.7496991576413959,
+      "grad_norm": 8.0625,
+      "learning_rate": 2.9356183821996976e-06,
+      "loss": 1.6522,
+      "step": 623
+    },
+    {
+      "epoch": 0.7509025270758123,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.9089115275952217e-06,
+      "loss": 1.7324,
+      "step": 624
+    },
+    {
+      "epoch": 0.7521058965102286,
+      "grad_norm": 11.875,
+      "learning_rate": 2.882306019801008e-06,
+      "loss": 1.4715,
+      "step": 625
+    },
+    {
+      "epoch": 0.753309265944645,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.855802239066623e-06,
+      "loss": 1.7237,
+      "step": 626
+    },
+    {
+      "epoch": 0.7545126353790613,
+      "grad_norm": 9.5,
+      "learning_rate": 2.8294005641877486e-06,
+      "loss": 1.8183,
+      "step": 627
+    },
+    {
+      "epoch": 0.7557160048134778,
+      "grad_norm": 2.234375,
+      "learning_rate": 2.8031013725007415e-06,
+      "loss": 1.7527,
+      "step": 628
+    },
+    {
+      "epoch": 0.7569193742478941,
+      "grad_norm": 8.125,
+      "learning_rate": 2.776905039877268e-06,
+      "loss": 1.7001,
+      "step": 629
+    },
+    {
+      "epoch": 0.7581227436823105,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.750811940718906e-06,
+      "loss": 1.6861,
+      "step": 630
+    },
+    {
+      "epoch": 0.7593261131167268,
+      "grad_norm": 8.75,
+      "learning_rate": 2.724822447951814e-06,
+      "loss": 1.7897,
+      "step": 631
+    },
+    {
+      "epoch": 0.7605294825511432,
+      "grad_norm": 1.828125,
+      "learning_rate": 2.6989369330213865e-06,
+      "loss": 1.7686,
+      "step": 632
+    },
+    {
+      "epoch": 0.7617328519855595,
+      "grad_norm": 7.5,
+      "learning_rate": 2.6731557658869668e-06,
+      "loss": 1.7332,
+      "step": 633
+    },
+    {
+      "epoch": 0.762936221419976,
+      "grad_norm": 2.265625,
+      "learning_rate": 2.647479315016528e-06,
+      "loss": 1.7653,
+      "step": 634
+    },
+    {
+      "epoch": 0.7641395908543923,
+      "grad_norm": 8.125,
+      "learning_rate": 2.621907947381438e-06,
+      "loss": 1.5852,
+      "step": 635
+    },
+    {
+      "epoch": 0.7653429602888087,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.596442028451194e-06,
+      "loss": 1.7552,
+      "step": 636
+    },
+    {
+      "epoch": 0.766546329723225,
+      "grad_norm": 9.1875,
+      "learning_rate": 2.5710819221882e-06,
+      "loss": 1.5638,
+      "step": 637
+    },
+    {
+      "epoch": 0.7677496991576414,
+      "grad_norm": 1.8359375,
+      "learning_rate": 2.5458279910425865e-06,
+      "loss": 1.7789,
+      "step": 638
+    },
+    {
+      "epoch": 0.7689530685920578,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.5206805959469984e-06,
+      "loss": 1.8226,
+      "step": 639
+    },
+    {
+      "epoch": 0.7701564380264742,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.4956400963114647e-06,
+      "loss": 1.7214,
+      "step": 640
+    },
+    {
+      "epoch": 0.7713598074608905,
+      "grad_norm": 8.125,
+      "learning_rate": 2.4707068500182442e-06,
+      "loss": 1.5876,
+      "step": 641
+    },
+    {
+      "epoch": 0.7725631768953068,
+      "grad_norm": 2.078125,
+      "learning_rate": 2.445881213416713e-06,
+      "loss": 1.7379,
+      "step": 642
+    },
+    {
+      "epoch": 0.7737665463297232,
+      "grad_norm": 10.3125,
+      "learning_rate": 2.4211635413182845e-06,
+      "loss": 1.7739,
+      "step": 643
+    },
+    {
+      "epoch": 0.7749699157641395,
+      "grad_norm": 2.515625,
+      "learning_rate": 2.3965541869913188e-06,
+      "loss": 1.7963,
+      "step": 644
+    },
+    {
+      "epoch": 0.776173285198556,
+      "grad_norm": 8.0,
+      "learning_rate": 2.3720535021560864e-06,
+      "loss": 1.8607,
+      "step": 645
+    },
+    {
+      "epoch": 0.7773766546329723,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.3476618369797457e-06,
+      "loss": 1.7694,
+      "step": 646
+    },
+    {
+      "epoch": 0.7785800240673887,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.32337954007132e-06,
+      "loss": 1.8392,
+      "step": 647
+    },
+    {
+      "epoch": 0.779783393501805,
+      "grad_norm": 2.15625,
+      "learning_rate": 2.299206958476731e-06,
+      "loss": 1.7709,
+      "step": 648
+    },
+    {
+      "epoch": 0.7809867629362214,
+      "grad_norm": 7.15625,
+      "learning_rate": 2.2751444376738373e-06,
+      "loss": 1.7899,
+      "step": 649
+    },
+    {
+      "epoch": 0.7821901323706378,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.251192321567488e-06,
+      "loss": 1.7695,
+      "step": 650
+    },
+    {
+      "epoch": 0.7833935018050542,
+      "grad_norm": 9.625,
+      "learning_rate": 2.2273509524846193e-06,
+      "loss": 1.5654,
+      "step": 651
+    },
+    {
+      "epoch": 0.7845968712394705,
+      "grad_norm": 2.03125,
+      "learning_rate": 2.2036206711693508e-06,
+      "loss": 1.734,
+      "step": 652
+    },
+    {
+      "epoch": 0.7858002406738869,
+      "grad_norm": 9.4375,
+      "learning_rate": 2.180001816778118e-06,
+      "loss": 1.7517,
+      "step": 653
+    },
+    {
+      "epoch": 0.7870036101083032,
+      "grad_norm": 1.8359375,
+      "learning_rate": 2.1564947268748382e-06,
+      "loss": 1.7432,
+      "step": 654
+    },
+    {
+      "epoch": 0.7882069795427196,
+      "grad_norm": 9.125,
+      "learning_rate": 2.133099737426064e-06,
+      "loss": 1.8342,
+      "step": 655
+    },
+    {
+      "epoch": 0.789410348977136,
+      "grad_norm": 1.75,
+      "learning_rate": 2.1098171827961965e-06,
+      "loss": 1.7589,
+      "step": 656
+    },
+    {
+      "epoch": 0.7906137184115524,
+      "grad_norm": 7.0,
+      "learning_rate": 2.086647395742709e-06,
+      "loss": 1.7399,
+      "step": 657
+    },
+    {
+      "epoch": 0.7918170878459687,
+      "grad_norm": 2.03125,
+      "learning_rate": 2.0635907074113737e-06,
+      "loss": 1.6705,
+      "step": 658
+    },
+    {
+      "epoch": 0.7930204572803851,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.040647447331553e-06,
+      "loss": 1.6823,
+      "step": 659
+    },
+    {
+      "epoch": 0.7942238267148014,
+      "grad_norm": 1.71875,
+      "learning_rate": 2.0178179434114674e-06,
+      "loss": 1.7502,
+      "step": 660
+    },
+    {
+      "epoch": 0.7954271961492179,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.9951025219335183e-06,
+      "loss": 1.7313,
+      "step": 661
+    },
+    {
+      "epoch": 0.7966305655836342,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.972501507549637e-06,
+      "loss": 1.7425,
+      "step": 662
+    },
+    {
+      "epoch": 0.7978339350180506,
+      "grad_norm": 12.5,
+      "learning_rate": 1.95001522327662e-06,
+      "loss": 1.6735,
+      "step": 663
+    },
+    {
+      "epoch": 0.7990373044524669,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.927643990491528e-06,
+      "loss": 1.7718,
+      "step": 664
+    },
+    {
+      "epoch": 0.8002406738868832,
+      "grad_norm": 11.125,
+      "learning_rate": 1.905388128927098e-06,
+      "loss": 1.7697,
+      "step": 665
+    },
+    {
+      "epoch": 0.8014440433212996,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.883247956667157e-06,
+      "loss": 1.7468,
+      "step": 666
+    },
+    {
+      "epoch": 0.802647412755716,
+      "grad_norm": 6.53125,
+      "learning_rate": 1.8612237901420838e-06,
+      "loss": 1.455,
+      "step": 667
+    },
+    {
+      "epoch": 0.8038507821901324,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.839315944124298e-06,
+      "loss": 1.7169,
+      "step": 668
+    },
+    {
+      "epoch": 0.8050541516245487,
+      "grad_norm": 7.25,
+      "learning_rate": 1.8175247317237365e-06,
+      "loss": 1.4329,
+      "step": 669
+    },
+    {
+      "epoch": 0.8062575210589651,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.7958504643834062e-06,
+      "loss": 1.7497,
+      "step": 670
+    },
+    {
+      "epoch": 0.8074608904933814,
+      "grad_norm": 7.375,
+      "learning_rate": 1.774293451874909e-06,
+      "loss": 1.4798,
+      "step": 671
+    },
+    {
+      "epoch": 0.8086642599277978,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.7528540022940288e-06,
+      "loss": 1.7344,
+      "step": 672
+    },
+    {
+      "epoch": 0.8098676293622142,
+      "grad_norm": 10.125,
+      "learning_rate": 1.731532422056319e-06,
+      "loss": 1.9758,
+      "step": 673
+    },
+    {
+      "epoch": 0.8110709987966306,
+      "grad_norm": 1.9921875,
+      "learning_rate": 1.71032901589274e-06,
+      "loss": 1.7087,
+      "step": 674
+    },
+    {
+      "epoch": 0.8122743682310469,
+      "grad_norm": 11.25,
+      "learning_rate": 1.6892440868452763e-06,
+      "loss": 1.5607,
+      "step": 675
+    },
+    {
+      "epoch": 0.8134777376654633,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.6682779362626378e-06,
+      "loss": 1.7157,
+      "step": 676
+    },
+    {
+      "epoch": 0.8146811070998796,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.6474308637959235e-06,
+      "loss": 1.7396,
+      "step": 677
+    },
+    {
+      "epoch": 0.8158844765342961,
+      "grad_norm": 1.875,
+      "learning_rate": 1.6267031673943546e-06,
+      "loss": 1.7405,
+      "step": 678
+    },
+    {
+      "epoch": 0.8170878459687124,
+      "grad_norm": 8.0,
+      "learning_rate": 1.6060951433010186e-06,
+      "loss": 1.5432,
+      "step": 679
+    },
+    {
+      "epoch": 0.8182912154031288,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.5856070860486205e-06,
+      "loss": 1.8063,
+      "step": 680
+    },
+    {
+      "epoch": 0.8194945848375451,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.5652392884552947e-06,
+      "loss": 1.6438,
+      "step": 681
+    },
+    {
+      "epoch": 0.8206979542719615,
+      "grad_norm": 1.375,
+      "learning_rate": 1.544992041620398e-06,
+      "loss": 1.7676,
+      "step": 682
+    },
+    {
+      "epoch": 0.8219013237063778,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.5248656349203628e-06,
+      "loss": 1.6503,
+      "step": 683
+    },
+    {
+      "epoch": 0.8231046931407943,
+      "grad_norm": 1.9140625,
+      "learning_rate": 1.5048603560045549e-06,
+      "loss": 1.7596,
+      "step": 684
+    },
+    {
+      "epoch": 0.8243080625752106,
+      "grad_norm": 10.375,
+      "learning_rate": 1.4849764907911712e-06,
+      "loss": 1.867,
+      "step": 685
+    },
+    {
+      "epoch": 0.825511432009627,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.4652143234631465e-06,
+      "loss": 1.7849,
+      "step": 686
+    },
+    {
+      "epoch": 0.8267148014440433,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.4455741364640863e-06,
+      "loss": 1.7476,
+      "step": 687
+    },
+    {
+      "epoch": 0.8279181708784596,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.426056210494241e-06,
+      "loss": 1.7788,
+      "step": 688
+    },
+    {
+      "epoch": 0.8291215403128761,
+      "grad_norm": 7.5,
+      "learning_rate": 1.4066608245064872e-06,
+      "loss": 1.5044,
+      "step": 689
+    },
+    {
+      "epoch": 0.8303249097472925,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.3873882557023488e-06,
+      "loss": 1.7591,
+      "step": 690
+    },
+    {
+      "epoch": 0.8315282791817088,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.3682387795280228e-06,
+      "loss": 1.868,
+      "step": 691
+    },
+    {
+      "epoch": 0.8327316486161251,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.3492126696704544e-06,
+      "loss": 1.7034,
+      "step": 692
+    },
+    {
+      "epoch": 0.8339350180505415,
+      "grad_norm": 7.9375,
+      "learning_rate": 1.3303101980534183e-06,
+      "loss": 1.4884,
+      "step": 693
+    },
+    {
+      "epoch": 0.8351383874849578,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.3115316348336348e-06,
+      "loss": 1.7203,
+      "step": 694
+    },
+    {
+      "epoch": 0.8363417569193743,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.2928772483969054e-06,
+      "loss": 1.865,
+      "step": 695
+    },
+    {
+      "epoch": 0.8375451263537906,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.2743473053542842e-06,
+      "loss": 1.8178,
+      "step": 696
+    },
+    {
+      "epoch": 0.838748495788207,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.2559420705382664e-06,
+      "loss": 1.6496,
+      "step": 697
+    },
+    {
+      "epoch": 0.8399518652226233,
+      "grad_norm": 1.859375,
+      "learning_rate": 1.237661806998991e-06,
+      "loss": 1.8212,
+      "step": 698
+    },
+    {
+      "epoch": 0.8411552346570397,
+      "grad_norm": 8.5,
+      "learning_rate": 1.2195067760004952e-06,
+      "loss": 1.4821,
+      "step": 699
+    },
+    {
+      "epoch": 0.8423586040914561,
+      "grad_norm": 1.859375,
+      "learning_rate": 1.2014772370169747e-06,
+      "loss": 1.793,
+      "step": 700
+    },
+    {
+      "epoch": 0.8435619735258725,
+      "grad_norm": 9.5625,
+      "learning_rate": 1.1835734477290784e-06,
+      "loss": 1.6435,
+      "step": 701
+    },
+    {
+      "epoch": 0.8447653429602888,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.1657956640202217e-06,
+      "loss": 1.7819,
+      "step": 702
+    },
+    {
+      "epoch": 0.8459687123947052,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.148144139972931e-06,
+      "loss": 1.4722,
+      "step": 703
+    },
+    {
+      "epoch": 0.8471720818291215,
+      "grad_norm": 1.9921875,
+      "learning_rate": 1.1306191278652112e-06,
+      "loss": 1.7809,
+      "step": 704
+    },
+    {
+      "epoch": 0.8483754512635379,
+      "grad_norm": 9.375,
+      "learning_rate": 1.1132208781669418e-06,
+      "loss": 1.7592,
+      "step": 705
+    },
+    {
+      "epoch": 0.8495788206979543,
+      "grad_norm": 1.6796875,
+      "learning_rate": 1.0959496395362946e-06,
+      "loss": 1.8302,
+      "step": 706
+    },
+    {
+      "epoch": 0.8507821901323707,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.0788056588161854e-06,
+      "loss": 1.3777,
+      "step": 707
+    },
+    {
+      "epoch": 0.851985559566787,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.0617891810307458e-06,
+      "loss": 1.7717,
+      "step": 708
+    },
+    {
+      "epoch": 0.8531889290012034,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.0449004493818083e-06,
+      "loss": 1.6946,
+      "step": 709
+    },
+    {
+      "epoch": 0.8543922984356197,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.0281397052454457e-06,
+      "loss": 1.7454,
+      "step": 710
+    },
+    {
+      "epoch": 0.855595667870036,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.0115071881685134e-06,
+      "loss": 1.7094,
+      "step": 711
+    },
+    {
+      "epoch": 0.8567990373044525,
+      "grad_norm": 1.7265625,
+      "learning_rate": 9.950031358652313e-07,
+      "loss": 1.7548,
+      "step": 712
+    },
+    {
+      "epoch": 0.8580024067388689,
+      "grad_norm": 6.5625,
+      "learning_rate": 9.786277842137837e-07,
+      "loss": 1.4388,
+      "step": 713
+    },
+    {
+      "epoch": 0.8592057761732852,
+      "grad_norm": 2.546875,
+      "learning_rate": 9.623813672529437e-07,
+      "loss": 1.861,
+      "step": 714
+    },
+    {
+      "epoch": 0.8604091456077015,
+      "grad_norm": 7.9375,
+      "learning_rate": 9.462641171787313e-07,
+      "loss": 1.6033,
+      "step": 715
+    },
+    {
+      "epoch": 0.8616125150421179,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.302762643411e-07,
+      "loss": 1.7625,
+      "step": 716
+    },
+    {
+      "epoch": 0.8628158844765343,
+      "grad_norm": 8.0625,
+      "learning_rate": 9.144180372406342e-07,
+      "loss": 1.5054,
+      "step": 717
+    },
+    {
+      "epoch": 0.8640192539109507,
+      "grad_norm": 1.4921875,
+      "learning_rate": 8.986896625253006e-07,
+      "loss": 1.7148,
+      "step": 718
+    },
+    {
+      "epoch": 0.865222623345367,
+      "grad_norm": 6.78125,
+      "learning_rate": 8.830913649871875e-07,
+      "loss": 1.7363,
+      "step": 719
+    },
+    {
+      "epoch": 0.8664259927797834,
+      "grad_norm": 1.84375,
+      "learning_rate": 8.676233675593038e-07,
+      "loss": 1.7645,
+      "step": 720
+    },
+    {
+      "epoch": 0.8676293622141997,
+      "grad_norm": 9.3125,
+      "learning_rate": 8.522858913123944e-07,
+      "loss": 1.7745,
+      "step": 721
+    },
+    {
+      "epoch": 0.8688327316486161,
+      "grad_norm": 1.984375,
+      "learning_rate": 8.370791554517743e-07,
+      "loss": 1.8315,
+      "step": 722
+    },
+    {
+      "epoch": 0.8700361010830325,
+      "grad_norm": 8.125,
+      "learning_rate": 8.220033773142022e-07,
+      "loss": 1.7244,
+      "step": 723
+    },
+    {
+      "epoch": 0.8712394705174489,
+      "grad_norm": 1.9140625,
+      "learning_rate": 8.070587723647705e-07,
+      "loss": 1.8274,
+      "step": 724
+    },
+    {
+      "epoch": 0.8724428399518652,
+      "grad_norm": 7.59375,
+      "learning_rate": 7.922455541938245e-07,
+      "loss": 1.7885,
+      "step": 725
+    },
+    {
+      "epoch": 0.8736462093862816,
+      "grad_norm": 1.921875,
+      "learning_rate": 7.77563934513913e-07,
+      "loss": 1.7429,
+      "step": 726
+    },
+    {
+      "epoch": 0.8748495788206979,
+      "grad_norm": 8.9375,
+      "learning_rate": 7.630141231567589e-07,
+      "loss": 1.5492,
+      "step": 727
+    },
+    {
+      "epoch": 0.8760529482551144,
+      "grad_norm": 1.625,
+      "learning_rate": 7.485963280702646e-07,
+      "loss": 1.7734,
+      "step": 728
+    },
+    {
+      "epoch": 0.8772563176895307,
+      "grad_norm": 11.5,
+      "learning_rate": 7.343107553155404e-07,
+      "loss": 1.7769,
+      "step": 729
+    },
+    {
+      "epoch": 0.8784596871239471,
+      "grad_norm": 2.15625,
+      "learning_rate": 7.201576090639529e-07,
+      "loss": 1.7938,
+      "step": 730
+    },
+    {
+      "epoch": 0.8796630565583634,
+      "grad_norm": 7.46875,
+      "learning_rate": 7.061370915942101e-07,
+      "loss": 1.6373,
+      "step": 731
+    },
+    {
+      "epoch": 0.8808664259927798,
+      "grad_norm": 1.796875,
+      "learning_rate": 6.922494032894744e-07,
+      "loss": 1.7239,
+      "step": 732
+    },
+    {
+      "epoch": 0.8820697954271961,
+      "grad_norm": 9.0625,
+      "learning_rate": 6.784947426344923e-07,
+      "loss": 1.7693,
+      "step": 733
+    },
+    {
+      "epoch": 0.8832731648616126,
+      "grad_norm": 2.015625,
+      "learning_rate": 6.648733062127643e-07,
+      "loss": 1.7668,
+      "step": 734
+    },
+    {
+      "epoch": 0.8844765342960289,
+      "grad_norm": 9.0,
+      "learning_rate": 6.513852887037319e-07,
+      "loss": 1.5196,
+      "step": 735
+    },
+    {
+      "epoch": 0.8856799037304453,
+      "grad_norm": 1.6875,
+      "learning_rate": 6.380308828799919e-07,
+      "loss": 1.735,
+      "step": 736
+    },
+    {
+      "epoch": 0.8868832731648616,
+      "grad_norm": 8.25,
+      "learning_rate": 6.248102796045475e-07,
+      "loss": 1.7167,
+      "step": 737
+    },
+    {
+      "epoch": 0.8880866425992779,
+      "grad_norm": 2.0625,
+      "learning_rate": 6.117236678280736e-07,
+      "loss": 1.7921,
+      "step": 738
+    },
+    {
+      "epoch": 0.8892900120336944,
+      "grad_norm": 13.5625,
+      "learning_rate": 5.98771234586224e-07,
+      "loss": 2.1283,
+      "step": 739
+    },
+    {
+      "epoch": 0.8904933814681107,
+      "grad_norm": 2.078125,
+      "learning_rate": 5.859531649969563e-07,
+      "loss": 1.7826,
+      "step": 740
+    },
+    {
+      "epoch": 0.8916967509025271,
+      "grad_norm": 9.3125,
+      "learning_rate": 5.732696422578787e-07,
+      "loss": 1.4974,
+      "step": 741
+    },
+    {
+      "epoch": 0.8929001203369434,
+      "grad_norm": 1.7421875,
+      "learning_rate": 5.60720847643641e-07,
+      "loss": 1.7384,
+      "step": 742
+    },
+    {
+      "epoch": 0.8941034897713598,
+      "grad_norm": 8.6875,
+      "learning_rate": 5.483069605033365e-07,
+      "loss": 1.7097,
+      "step": 743
+    },
+    {
+      "epoch": 0.8953068592057761,
+      "grad_norm": 1.921875,
+      "learning_rate": 5.360281582579474e-07,
+      "loss": 1.7171,
+      "step": 744
+    },
+    {
+      "epoch": 0.8965102286401926,
+      "grad_norm": 8.3125,
+      "learning_rate": 5.238846163978018e-07,
+      "loss": 1.5687,
+      "step": 745
+    },
+    {
+      "epoch": 0.8977135980746089,
+      "grad_norm": 2.28125,
+      "learning_rate": 5.11876508480067e-07,
+      "loss": 1.7238,
+      "step": 746
+    },
+    {
+      "epoch": 0.8989169675090253,
+      "grad_norm": 8.5,
+      "learning_rate": 5.000040061262712e-07,
+      "loss": 1.6323,
+      "step": 747
+    },
+    {
+      "epoch": 0.9001203369434416,
+      "grad_norm": 1.75,
+      "learning_rate": 4.882672790198473e-07,
+      "loss": 1.7586,
+      "step": 748
+    },
+    {
+      "epoch": 0.901323706377858,
+      "grad_norm": 9.3125,
+      "learning_rate": 4.766664949037103e-07,
+      "loss": 1.4364,
+      "step": 749
+    },
+    {
+      "epoch": 0.9025270758122743,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.652018195778629e-07,
+      "loss": 1.753,
+      "step": 750
+    },
+    {
+      "epoch": 0.9037304452466908,
+      "grad_norm": 9.875,
+      "learning_rate": 4.538734168970149e-07,
+      "loss": 1.9226,
+      "step": 751
+    },
+    {
+      "epoch": 0.9049338146811071,
+      "grad_norm": 1.7578125,
+      "learning_rate": 4.4268144876825846e-07,
+      "loss": 1.7851,
+      "step": 752
+    },
+    {
+      "epoch": 0.9061371841155235,
+      "grad_norm": 9.25,
+      "learning_rate": 4.3162607514873556e-07,
+      "loss": 1.483,
+      "step": 753
+    },
+    {
+      "epoch": 0.9073405535499398,
+      "grad_norm": 1.765625,
+      "learning_rate": 4.207074540433631e-07,
+      "loss": 1.7761,
+      "step": 754
+    },
+    {
+      "epoch": 0.9085439229843562,
+      "grad_norm": 7.8125,
+      "learning_rate": 4.09925741502577e-07,
+      "loss": 1.559,
+      "step": 755
+    },
+    {
+      "epoch": 0.9097472924187726,
+      "grad_norm": 1.65625,
+      "learning_rate": 3.9928109162008953e-07,
+      "loss": 1.6966,
+      "step": 756
+    },
+    {
+      "epoch": 0.910950661853189,
+      "grad_norm": 6.875,
+      "learning_rate": 3.887736565307032e-07,
+      "loss": 1.6578,
+      "step": 757
+    },
+    {
+      "epoch": 0.9121540312876053,
+      "grad_norm": 2.328125,
+      "learning_rate": 3.7840358640812036e-07,
+      "loss": 1.7345,
+      "step": 758
+    },
+    {
+      "epoch": 0.9133574007220217,
+      "grad_norm": 10.125,
+      "learning_rate": 3.68171029462806e-07,
+      "loss": 1.6894,
+      "step": 759
+    },
+    {
+      "epoch": 0.914560770156438,
+      "grad_norm": 1.84375,
+      "learning_rate": 3.580761319398729e-07,
+      "loss": 1.8096,
+      "step": 760
+    },
+    {
+      "epoch": 0.9157641395908543,
+      "grad_norm": 8.625,
+      "learning_rate": 3.481190381169808e-07,
+      "loss": 1.755,
+      "step": 761
+    },
+    {
+      "epoch": 0.9169675090252708,
+      "grad_norm": 2.046875,
+      "learning_rate": 3.3829989030228163e-07,
+      "loss": 1.817,
+      "step": 762
+    },
+    {
+      "epoch": 0.9181708784596871,
+      "grad_norm": 6.90625,
+      "learning_rate": 3.286188288323844e-07,
+      "loss": 1.5882,
+      "step": 763
+    },
+    {
+      "epoch": 0.9193742478941035,
+      "grad_norm": 1.7421875,
+      "learning_rate": 3.190759920703512e-07,
+      "loss": 1.7517,
+      "step": 764
+    },
+    {
+      "epoch": 0.9205776173285198,
+      "grad_norm": 7.3125,
+      "learning_rate": 3.096715164037123e-07,
+      "loss": 1.9038,
+      "step": 765
+    },
+    {
+      "epoch": 0.9217809867629362,
+      "grad_norm": 1.7421875,
+      "learning_rate": 3.0040553624252844e-07,
+      "loss": 1.7996,
+      "step": 766
+    },
+    {
+      "epoch": 0.9229843561973526,
+      "grad_norm": 8.75,
+      "learning_rate": 2.9127818401745833e-07,
+      "loss": 1.4735,
+      "step": 767
+    },
+    {
+      "epoch": 0.924187725631769,
+      "grad_norm": 1.5703125,
+      "learning_rate": 2.822895901778744e-07,
+      "loss": 1.7877,
+      "step": 768
+    },
+    {
+      "epoch": 0.9253910950661853,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.7343988318999536e-07,
+      "loss": 1.6788,
+      "step": 769
+    },
+    {
+      "epoch": 0.9265944645006017,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.6472918953504566e-07,
+      "loss": 1.7739,
+      "step": 770
+    },
+    {
+      "epoch": 0.927797833935018,
+      "grad_norm": 7.0,
+      "learning_rate": 2.5615763370745894e-07,
+      "loss": 1.5671,
+      "step": 771
+    },
+    {
+      "epoch": 0.9290012033694344,
+      "grad_norm": 1.828125,
+      "learning_rate": 2.477253382130862e-07,
+      "loss": 1.7519,
+      "step": 772
+    },
+    {
+      "epoch": 0.9302045728038508,
+      "grad_norm": 8.4375,
+      "learning_rate": 2.394324235674517e-07,
+      "loss": 1.7307,
+      "step": 773
+    },
+    {
+      "epoch": 0.9314079422382672,
+      "grad_norm": 3.8125,
+      "learning_rate": 2.3127900829403305e-07,
+      "loss": 1.776,
+      "step": 774
+    },
+    {
+      "epoch": 0.9326113116726835,
+      "grad_norm": 13.375,
+      "learning_rate": 2.2326520892255953e-07,
+      "loss": 1.7122,
+      "step": 775
+    },
+    {
+      "epoch": 0.9338146811070999,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.1539113998735094e-07,
+      "loss": 1.7175,
+      "step": 776
+    },
+    {
+      "epoch": 0.9350180505415162,
+      "grad_norm": 8.875,
+      "learning_rate": 2.0765691402568455e-07,
+      "loss": 1.575,
+      "step": 777
+    },
+    {
+      "epoch": 0.9362214199759326,
+      "grad_norm": 2.328125,
+      "learning_rate": 2.000626415761786e-07,
+      "loss": 1.7733,
+      "step": 778
+    },
+    {
+      "epoch": 0.937424789410349,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.9260843117721695e-07,
+      "loss": 1.5334,
+      "step": 779
+    },
+    {
+      "epoch": 0.9386281588447654,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.8529438936540022e-07,
+      "loss": 1.7955,
+      "step": 780
+    },
+    {
+      "epoch": 0.9398315282791817,
+      "grad_norm": 9.625,
+      "learning_rate": 1.7812062067401713e-07,
+      "loss": 1.0221,
+      "step": 781
+    },
+    {
+      "epoch": 0.941034897713598,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.710872276315556e-07,
+      "loss": 1.7049,
+      "step": 782
+    },
+    {
+      "epoch": 0.9422382671480144,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.6419431076023505e-07,
+      "loss": 1.8431,
+      "step": 783
+    },
+    {
+      "epoch": 0.9434416365824309,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.5744196857456874e-07,
+      "loss": 1.7725,
+      "step": 784
+    },
+    {
+      "epoch": 0.9446450060168472,
+      "grad_norm": 6.65625,
+      "learning_rate": 1.5083029757995914e-07,
+      "loss": 1.5299,
+      "step": 785
+    },
+    {
+      "epoch": 0.9458483754512635,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.4435939227131712e-07,
+      "loss": 1.7676,
+      "step": 786
+    },
+    {
+      "epoch": 0.9470517448856799,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.3802934513170828e-07,
+      "loss": 1.6324,
+      "step": 787
+    },
+    {
+      "epoch": 0.9482551143200962,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.3184024663103755e-07,
+      "loss": 1.7018,
+      "step": 788
+    },
+    {
+      "epoch": 0.9494584837545126,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.25792185224749e-07,
+      "loss": 1.3774,
+      "step": 789
+    },
+    {
+      "epoch": 0.950661853188929,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.198852473525669e-07,
+      "loss": 1.7961,
+      "step": 790
+    },
+    {
+      "epoch": 0.9518652226233454,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.1411951743726002e-07,
+      "loss": 1.8785,
+      "step": 791
+    },
+    {
+      "epoch": 0.9530685920577617,
+      "grad_norm": 1.921875,
+      "learning_rate": 1.0849507788343038e-07,
+      "loss": 1.7485,
+      "step": 792
+    },
+    {
+      "epoch": 0.9542719614921781,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.030120090763409e-07,
+      "loss": 1.7257,
+      "step": 793
+    },
+    {
+      "epoch": 0.9554753309265944,
+      "grad_norm": 1.6640625,
+      "learning_rate": 9.767038938076511e-08,
+      "loss": 1.7644,
+      "step": 794
+    },
+    {
+      "epoch": 0.9566787003610109,
+      "grad_norm": 9.375,
+      "learning_rate": 9.247029513986482e-08,
+      "loss": 1.9214,
+      "step": 795
+    },
+    {
+      "epoch": 0.9578820697954272,
+      "grad_norm": 1.703125,
+      "learning_rate": 8.741180067409982e-08,
+      "loss": 1.7783,
+      "step": 796
+    },
+    {
+      "epoch": 0.9590854392298436,
+      "grad_norm": 8.625,
+      "learning_rate": 8.249497828016872e-08,
+      "loss": 1.6367,
+      "step": 797
+    },
+    {
+      "epoch": 0.9602888086642599,
+      "grad_norm": 1.9765625,
+      "learning_rate": 7.771989822997206e-08,
+      "loss": 1.789,
+      "step": 798
+    },
+    {
+      "epoch": 0.9614921780986763,
+      "grad_norm": 6.75,
+      "learning_rate": 7.30866287696097e-08,
+      "loss": 1.5397,
+      "step": 799
+    },
+    {
+      "epoch": 0.9626955475330926,
+      "grad_norm": 1.9609375,
+      "learning_rate": 6.859523611840612e-08,
+      "loss": 1.727,
+      "step": 800
+    },
+    {
+      "epoch": 0.9638989169675091,
+      "grad_norm": 7.3125,
+      "learning_rate": 6.424578446796004e-08,
+      "loss": 1.7173,
+      "step": 801
+    },
+    {
+      "epoch": 0.9651022864019254,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.003833598123287e-08,
+      "loss": 1.7525,
+      "step": 802
+    },
+    {
+      "epoch": 0.9663056558363418,
+      "grad_norm": 7.90625,
+      "learning_rate": 5.597295079165621e-08,
+      "loss": 1.4858,
+      "step": 803
+    },
+    {
+      "epoch": 0.9675090252707581,
+      "grad_norm": 1.890625,
+      "learning_rate": 5.204968700227242e-08,
+      "loss": 1.7038,
+      "step": 804
+    },
+    {
+      "epoch": 0.9687123947051745,
+      "grad_norm": 9.4375,
+      "learning_rate": 4.826860068490868e-08,
+      "loss": 1.6603,
+      "step": 805
+    },
+    {
+      "epoch": 0.9699157641395909,
+      "grad_norm": 1.8671875,
+      "learning_rate": 4.4629745879367634e-08,
+      "loss": 1.7481,
+      "step": 806
+    },
+    {
+      "epoch": 0.9711191335740073,
+      "grad_norm": 8.75,
+      "learning_rate": 4.113317459266242e-08,
+      "loss": 1.605,
+      "step": 807
+    },
+    {
+      "epoch": 0.9723225030084236,
+      "grad_norm": 1.7578125,
+      "learning_rate": 3.777893679827061e-08,
+      "loss": 1.7071,
+      "step": 808
+    },
+    {
+      "epoch": 0.97352587244284,
+      "grad_norm": 8.4375,
+      "learning_rate": 3.456708043541812e-08,
+      "loss": 1.9034,
+      "step": 809
+    },
+    {
+      "epoch": 0.9747292418772563,
+      "grad_norm": 1.59375,
+      "learning_rate": 3.1497651408399774e-08,
+      "loss": 1.7136,
+      "step": 810
+    },
+    {
+      "epoch": 0.9759326113116726,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.8570693585914246e-08,
+      "loss": 1.8149,
+      "step": 811
+    },
+    {
+      "epoch": 0.9771359807460891,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.578624880044567e-08,
+      "loss": 1.8252,
+      "step": 812
+    },
+    {
+      "epoch": 0.9783393501805054,
+      "grad_norm": 7.875,
+      "learning_rate": 2.314435684766081e-08,
+      "loss": 1.8397,
+      "step": 813
+    },
+    {
+      "epoch": 0.9795427196149218,
+      "grad_norm": 1.890625,
+      "learning_rate": 2.0645055485842837e-08,
+      "loss": 1.7662,
+      "step": 814
+    },
+    {
+      "epoch": 0.9807460890493381,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.8288380435349527e-08,
+      "loss": 1.67,
+      "step": 815
+    },
+    {
+      "epoch": 0.9819494584837545,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.6074365378105915e-08,
+      "loss": 1.7614,
+      "step": 816
+    },
+    {
+      "epoch": 0.9831528279181708,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.400304195711688e-08,
+      "loss": 1.5358,
+      "step": 817
+    },
+    {
+      "epoch": 0.9843561973525873,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.2074439776021962e-08,
+      "loss": 1.7641,
+      "step": 818
+    },
+    {
+      "epoch": 0.9855595667870036,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.0288586398670141e-08,
+      "loss": 1.5657,
+      "step": 819
+    },
+    {
+      "epoch": 0.98676293622142,
+      "grad_norm": 1.828125,
+      "learning_rate": 8.64550734872016e-09,
+      "loss": 1.7538,
+      "step": 820
+    },
+    {
+      "epoch": 0.9879663056558363,
+      "grad_norm": 7.875,
+      "learning_rate": 7.145226109286363e-09,
+      "loss": 1.5632,
+      "step": 821
+    },
+    {
+      "epoch": 0.9891696750902527,
+      "grad_norm": 2.203125,
+      "learning_rate": 5.787764122592299e-09,
+      "loss": 1.8164,
+      "step": 822
+    },
+    {
+      "epoch": 0.9903730445246691,
+      "grad_norm": 8.9375,
+      "learning_rate": 4.573140789672082e-09,
+      "loss": 1.886,
+      "step": 823
+    },
+    {
+      "epoch": 0.9915764139590855,
+      "grad_norm": 1.828125,
+      "learning_rate": 3.5013734700883874e-09,
+      "loss": 1.7194,
+      "step": 824
+    },
+    {
+      "epoch": 0.9927797833935018,
+      "grad_norm": 9.125,
+      "learning_rate": 2.5724774816870966e-09,
+      "loss": 1.3841,
+      "step": 825
+    },
+    {
+      "epoch": 0.9939831528279182,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7864661003774708e-09,
+      "loss": 1.7312,
+      "step": 826
+    },
+    {
+      "epoch": 0.9951865222623345,
+      "grad_norm": 7.09375,
+      "learning_rate": 1.1433505599434126e-09,
+      "loss": 1.4796,
+      "step": 827
+    },
+    {
+      "epoch": 0.9963898916967509,
+      "grad_norm": 1.4921875,
+      "learning_rate": 6.431400518780439e-10,
+      "loss": 1.6919,
+      "step": 828
+    },
+    {
+      "epoch": 0.9975932611311673,
+      "grad_norm": 7.53125,
+      "learning_rate": 2.8584172526047173e-10,
+      "loss": 1.6439,
+      "step": 829
+    },
+    {
+      "epoch": 0.9987966305655837,
+      "grad_norm": 1.90625,
+      "learning_rate": 7.146068664698469e-11,
+      "loss": 1.7275,
+      "step": 830
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 8.5,
+      "learning_rate": 0.0,
+      "loss": 1.8022,
+      "step": 831
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 831,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.843230151337574e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}