diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,11132 @@
+{
+  "best_global_step": 1550,
+  "best_metric": 0.07311470806598663,
+  "best_model_checkpoint": "outputs_qwq/checkpoint-1550",
+  "epoch": 2.903891233005157,
+  "eval_steps": 50,
+  "global_step": 1550,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001875293014533521,
+      "grad_norm": 0.39477676153182983,
+      "learning_rate": 0.0,
+      "loss": 0.7899,
+      "step": 1
+    },
+    {
+      "epoch": 0.003750586029067042,
+      "grad_norm": 0.3514685332775116,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 0.9541,
+      "step": 2
+    },
+    {
+      "epoch": 0.005625879043600563,
+      "grad_norm": 0.2947579622268677,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 0.7068,
+      "step": 3
+    },
+    {
+      "epoch": 0.007501172058134084,
+      "grad_norm": 0.26537010073661804,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 0.6839,
+      "step": 4
+    },
+    {
+      "epoch": 0.009376465072667605,
+      "grad_norm": 0.26260894536972046,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.68,
+      "step": 5
+    },
+    {
+      "epoch": 0.011251758087201125,
+      "grad_norm": 0.24916508793830872,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.6905,
+      "step": 6
+    },
+    {
+      "epoch": 0.013127051101734646,
+      "grad_norm": 0.3689679801464081,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.9347,
+      "step": 7
+    },
+    {
+      "epoch": 0.015002344116268168,
+      "grad_norm": 0.27197960019111633,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 0.7164,
+      "step": 8
+    },
+    {
+      "epoch": 0.016877637130801686,
+      "grad_norm": 0.305757611989975,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.721,
+      "step": 9
+    },
+    {
+      "epoch": 0.01875293014533521,
+      "grad_norm": 0.272778183221817,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 0.6814,
+      "step": 10
+    },
+    {
+      "epoch": 0.02062822315986873,
+      "grad_norm": 0.28655725717544556,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.7508,
+      "step": 11
+    },
+    {
+      "epoch": 0.02250351617440225,
+      "grad_norm": 0.32160523533821106,
+      "learning_rate": 2.2e-06,
+      "loss": 0.8391,
+      "step": 12
+    },
+    {
+      "epoch": 0.02437880918893577,
+      "grad_norm": 0.2688426077365875,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.6594,
+      "step": 13
+    },
+    {
+      "epoch": 0.02625410220346929,
+      "grad_norm": 0.33779385685920715,
+      "learning_rate": 2.6e-06,
+      "loss": 0.7729,
+      "step": 14
+    },
+    {
+      "epoch": 0.02812939521800281,
+      "grad_norm": 0.3078613877296448,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.7372,
+      "step": 15
+    },
+    {
+      "epoch": 0.030004688232536336,
+      "grad_norm": 0.2747291922569275,
+      "learning_rate": 3e-06,
+      "loss": 0.6716,
+      "step": 16
+    },
+    {
+      "epoch": 0.03187998124706985,
+      "grad_norm": 0.30704063177108765,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.7303,
+      "step": 17
+    },
+    {
+      "epoch": 0.03375527426160337,
+      "grad_norm": 0.312160849571228,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 0.7528,
+      "step": 18
+    },
+    {
+      "epoch": 0.03563056727613689,
+      "grad_norm": 0.3484225571155548,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.728,
+      "step": 19
+    },
+    {
+      "epoch": 0.03750586029067042,
+      "grad_norm": 0.328665167093277,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 0.8129,
+      "step": 20
+    },
+    {
+      "epoch": 0.03938115330520394,
+      "grad_norm": 0.32902592420578003,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.7526,
+      "step": 21
+    },
+    {
+      "epoch": 0.04125644631973746,
+      "grad_norm": 0.3969726860523224,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.806,
+      "step": 22
+    },
+    {
+      "epoch": 0.04313173933427098,
+      "grad_norm": 0.3924497067928314,
+      "learning_rate": 4.4e-06,
+      "loss": 0.6978,
+      "step": 23
+    },
+    {
+      "epoch": 0.0450070323488045,
+      "grad_norm": 0.33922383189201355,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 0.7919,
+      "step": 24
+    },
+    {
+      "epoch": 0.04688232536333802,
+      "grad_norm": 0.3359040319919586,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.7881,
+      "step": 25
+    },
+    {
+      "epoch": 0.04875761837787154,
+      "grad_norm": 0.4279753863811493,
+      "learning_rate": 5e-06,
+      "loss": 0.9385,
+      "step": 26
+    },
+    {
+      "epoch": 0.05063291139240506,
+      "grad_norm": 0.351744145154953,
+      "learning_rate": 5.2e-06,
+      "loss": 0.8571,
+      "step": 27
+    },
+    {
+      "epoch": 0.05250820440693858,
+      "grad_norm": 0.3303033113479614,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 0.6374,
+      "step": 28
+    },
+    {
+      "epoch": 0.0543834974214721,
+      "grad_norm": 0.36658555269241333,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.7102,
+      "step": 29
+    },
+    {
+      "epoch": 0.05625879043600562,
+      "grad_norm": 0.3628396987915039,
+      "learning_rate": 5.8e-06,
+      "loss": 0.7783,
+      "step": 30
+    },
+    {
+      "epoch": 0.058134083450539144,
+      "grad_norm": 0.38539397716522217,
+      "learning_rate": 6e-06,
+      "loss": 0.798,
+      "step": 31
+    },
+    {
+      "epoch": 0.06000937646507267,
+      "grad_norm": 0.4560730755329132,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 0.716,
+      "step": 32
+    },
+    {
+      "epoch": 0.06188466947960619,
+      "grad_norm": 0.35180985927581787,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.7262,
+      "step": 33
+    },
+    {
+      "epoch": 0.0637599624941397,
+      "grad_norm": 0.4624330997467041,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 0.8614,
+      "step": 34
+    },
+    {
+      "epoch": 0.06563525550867323,
+      "grad_norm": 0.3922227621078491,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 0.7928,
+      "step": 35
+    },
+    {
+      "epoch": 0.06751054852320675,
+      "grad_norm": 0.44666793942451477,
+      "learning_rate": 7e-06,
+      "loss": 0.8177,
+      "step": 36
+    },
+    {
+      "epoch": 0.06938584153774027,
+      "grad_norm": 0.4159597158432007,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.8009,
+      "step": 37
+    },
+    {
+      "epoch": 0.07126113455227379,
+      "grad_norm": 0.442553848028183,
+      "learning_rate": 7.4e-06,
+      "loss": 0.6281,
+      "step": 38
+    },
+    {
+      "epoch": 0.07313642756680731,
+      "grad_norm": 0.36248019337654114,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 0.6649,
+      "step": 39
+    },
+    {
+      "epoch": 0.07501172058134084,
+      "grad_norm": 0.5066297650337219,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 0.8881,
+      "step": 40
+    },
+    {
+      "epoch": 0.07688701359587435,
+      "grad_norm": 0.4364047944545746,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.8244,
+      "step": 41
+    },
+    {
+      "epoch": 0.07876230661040788,
+      "grad_norm": 0.43310582637786865,
+      "learning_rate": 8.2e-06,
+      "loss": 0.6797,
+      "step": 42
+    },
+    {
+      "epoch": 0.0806375996249414,
+      "grad_norm": 0.5400950908660889,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.7903,
+      "step": 43
+    },
+    {
+      "epoch": 0.08251289263947492,
+      "grad_norm": 0.4437219500541687,
+      "learning_rate": 8.6e-06,
+      "loss": 0.6449,
+      "step": 44
+    },
+    {
+      "epoch": 0.08438818565400844,
+      "grad_norm": 0.48232921957969666,
+      "learning_rate": 8.8e-06,
+      "loss": 0.6013,
+      "step": 45
+    },
+    {
+      "epoch": 0.08626347866854196,
+      "grad_norm": 0.6040736436843872,
+      "learning_rate": 9e-06,
+      "loss": 0.5827,
+      "step": 46
+    },
+    {
+      "epoch": 0.08813877168307548,
+      "grad_norm": 0.4945738613605499,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 0.7523,
+      "step": 47
+    },
+    {
+      "epoch": 0.090014064697609,
+      "grad_norm": 0.44322240352630615,
+      "learning_rate": 9.4e-06,
+      "loss": 0.6732,
+      "step": 48
+    },
+    {
+      "epoch": 0.09188935771214252,
+      "grad_norm": 0.42256736755371094,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.6088,
+      "step": 49
+    },
+    {
+      "epoch": 0.09376465072667604,
+      "grad_norm": 0.4956076443195343,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 0.5818,
+      "step": 50
+    },
+    {
+      "epoch": 0.09376465072667604,
+      "eval_loss": 0.6773348450660706,
+      "eval_runtime": 678.2607,
+      "eval_samples_per_second": 0.29,
+      "eval_steps_per_second": 0.29,
+      "step": 50
+    },
+    {
+      "epoch": 0.09563994374120956,
+      "grad_norm": 0.5553563237190247,
+      "learning_rate": 1e-05,
+      "loss": 0.7514,
+      "step": 51
+    },
+    {
+      "epoch": 0.09751523675574308,
+      "grad_norm": 0.5681352019309998,
+      "learning_rate": 1.02e-05,
+      "loss": 0.6105,
+      "step": 52
+    },
+    {
+      "epoch": 0.09939052977027661,
+      "grad_norm": 0.5460308790206909,
+      "learning_rate": 1.04e-05,
+      "loss": 0.5824,
+      "step": 53
+    },
+    {
+      "epoch": 0.10126582278481013,
+      "grad_norm": 0.5295383930206299,
+      "learning_rate": 1.0600000000000002e-05,
+      "loss": 0.6622,
+      "step": 54
+    },
+    {
+      "epoch": 0.10314111579934365,
+      "grad_norm": 0.5573945045471191,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 0.604,
+      "step": 55
+    },
+    {
+      "epoch": 0.10501640881387717,
+      "grad_norm": 0.5254002809524536,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.5777,
+      "step": 56
+    },
+    {
+      "epoch": 0.10689170182841069,
+      "grad_norm": 0.4916251003742218,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.5605,
+      "step": 57
+    },
+    {
+      "epoch": 0.1087669948429442,
+      "grad_norm": 0.44173622131347656,
+      "learning_rate": 1.14e-05,
+      "loss": 0.4921,
+      "step": 58
+    },
+    {
+      "epoch": 0.11064228785747773,
+      "grad_norm": 0.7284818887710571,
+      "learning_rate": 1.16e-05,
+      "loss": 0.676,
+      "step": 59
+    },
+    {
+      "epoch": 0.11251758087201125,
+      "grad_norm": 0.5018178224563599,
+      "learning_rate": 1.18e-05,
+      "loss": 0.5044,
+      "step": 60
+    },
+    {
+      "epoch": 0.11439287388654477,
+      "grad_norm": 0.4238162040710449,
+      "learning_rate": 1.2e-05,
+      "loss": 0.4526,
+      "step": 61
+    },
+    {
+      "epoch": 0.11626816690107829,
+      "grad_norm": 0.43392857909202576,
+      "learning_rate": 1.22e-05,
+      "loss": 0.554,
+      "step": 62
+    },
+    {
+      "epoch": 0.11814345991561181,
+      "grad_norm": 0.37505975365638733,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 0.4167,
+      "step": 63
+    },
+    {
+      "epoch": 0.12001875293014534,
+      "grad_norm": 0.41996413469314575,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 0.4457,
+      "step": 64
+    },
+    {
+      "epoch": 0.12189404594467886,
+      "grad_norm": 0.35364946722984314,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.397,
+      "step": 65
+    },
+    {
+      "epoch": 0.12376933895921238,
+      "grad_norm": 0.3553512692451477,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.4027,
+      "step": 66
+    },
+    {
+      "epoch": 0.1256446319737459,
+      "grad_norm": 0.35016101598739624,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 0.4234,
+      "step": 67
+    },
+    {
+      "epoch": 0.1275199249882794,
+      "grad_norm": 0.28325924277305603,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 0.3099,
+      "step": 68
+    },
+    {
+      "epoch": 0.12939521800281295,
+      "grad_norm": 0.3461209237575531,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.4367,
+      "step": 69
+    },
+    {
+      "epoch": 0.13127051101734646,
+      "grad_norm": 0.28789466619491577,
+      "learning_rate": 1.38e-05,
+      "loss": 0.3175,
+      "step": 70
+    },
+    {
+      "epoch": 0.13314580403187998,
+      "grad_norm": 0.3014371693134308,
+      "learning_rate": 1.4e-05,
+      "loss": 0.3197,
+      "step": 71
+    },
+    {
+      "epoch": 0.1350210970464135,
+      "grad_norm": 0.2781355082988739,
+      "learning_rate": 1.4200000000000001e-05,
+      "loss": 0.3437,
+      "step": 72
+    },
+    {
+      "epoch": 0.13689639006094703,
+      "grad_norm": 0.2576352059841156,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.2707,
+      "step": 73
+    },
+    {
+      "epoch": 0.13877168307548055,
+      "grad_norm": 0.3152413070201874,
+      "learning_rate": 1.46e-05,
+      "loss": 0.3236,
+      "step": 74
+    },
+    {
+      "epoch": 0.14064697609001406,
+      "grad_norm": 0.6653899550437927,
+      "learning_rate": 1.48e-05,
+      "loss": 0.3465,
+      "step": 75
+    },
+    {
+      "epoch": 0.14252226910454757,
+      "grad_norm": 0.2475852519273758,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.2468,
+      "step": 76
+    },
+    {
+      "epoch": 0.1443975621190811,
+      "grad_norm": 0.28572630882263184,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 0.3287,
+      "step": 77
+    },
+    {
+      "epoch": 0.14627285513361463,
+      "grad_norm": 0.27325335144996643,
+      "learning_rate": 1.54e-05,
+      "loss": 0.2926,
+      "step": 78
+    },
+    {
+      "epoch": 0.14814814814814814,
+      "grad_norm": 0.3255375027656555,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 0.279,
+      "step": 79
+    },
+    {
+      "epoch": 0.15002344116268168,
+      "grad_norm": 0.44983330368995667,
+      "learning_rate": 1.58e-05,
+      "loss": 0.3534,
+      "step": 80
+    },
+    {
+      "epoch": 0.1518987341772152,
+      "grad_norm": 0.240912526845932,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.2541,
+      "step": 81
+    },
+    {
+      "epoch": 0.1537740271917487,
+      "grad_norm": 0.21851545572280884,
+      "learning_rate": 1.62e-05,
+      "loss": 0.2334,
+      "step": 82
+    },
+    {
+      "epoch": 0.15564932020628222,
+      "grad_norm": 0.25758886337280273,
+      "learning_rate": 1.64e-05,
+      "loss": 0.2952,
+      "step": 83
+    },
+    {
+      "epoch": 0.15752461322081576,
+      "grad_norm": 0.2865050435066223,
+      "learning_rate": 1.66e-05,
+      "loss": 0.2702,
+      "step": 84
+    },
+    {
+      "epoch": 0.15939990623534928,
+      "grad_norm": 0.21041016280651093,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.2565,
+      "step": 85
+    },
+    {
+      "epoch": 0.1612751992498828,
+      "grad_norm": 0.21632152795791626,
+      "learning_rate": 1.7e-05,
+      "loss": 0.2643,
+      "step": 86
+    },
+    {
+      "epoch": 0.1631504922644163,
+      "grad_norm": 0.1930413544178009,
+      "learning_rate": 1.72e-05,
+      "loss": 0.2386,
+      "step": 87
+    },
+    {
+      "epoch": 0.16502578527894984,
+      "grad_norm": 0.21475085616111755,
+      "learning_rate": 1.7400000000000003e-05,
+      "loss": 0.3035,
+      "step": 88
+    },
+    {
+      "epoch": 0.16690107829348336,
+      "grad_norm": 0.2499314695596695,
+      "learning_rate": 1.76e-05,
+      "loss": 0.3382,
+      "step": 89
+    },
+    {
+      "epoch": 0.16877637130801687,
+      "grad_norm": 0.22533196210861206,
+      "learning_rate": 1.7800000000000002e-05,
+      "loss": 0.2262,
+      "step": 90
+    },
+    {
+      "epoch": 0.1706516643225504,
+      "grad_norm": 0.22342143952846527,
+      "learning_rate": 1.8e-05,
+      "loss": 0.2489,
+      "step": 91
+    },
+    {
+      "epoch": 0.17252695733708392,
+      "grad_norm": 0.2666771411895752,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 0.2732,
+      "step": 92
+    },
+    {
+      "epoch": 0.17440225035161744,
+      "grad_norm": 0.22881805896759033,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 0.285,
+      "step": 93
+    },
+    {
+      "epoch": 0.17627754336615095,
+      "grad_norm": 0.3157159984111786,
+      "learning_rate": 1.86e-05,
+      "loss": 0.2406,
+      "step": 94
+    },
+    {
+      "epoch": 0.1781528363806845,
+      "grad_norm": 0.20557765662670135,
+      "learning_rate": 1.88e-05,
+      "loss": 0.2162,
+      "step": 95
+    },
+    {
+      "epoch": 0.180028129395218,
+      "grad_norm": 0.26530343294143677,
+      "learning_rate": 1.9e-05,
+      "loss": 0.2527,
+      "step": 96
+    },
+    {
+      "epoch": 0.18190342240975152,
+      "grad_norm": 0.21458247303962708,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.1677,
+      "step": 97
+    },
+    {
+      "epoch": 0.18377871542428503,
+      "grad_norm": 0.20788805186748505,
+      "learning_rate": 1.94e-05,
+      "loss": 0.2191,
+      "step": 98
+    },
+    {
+      "epoch": 0.18565400843881857,
+      "grad_norm": 0.20019683241844177,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 0.1821,
+      "step": 99
+    },
+    {
+      "epoch": 0.1875293014533521,
+      "grad_norm": 0.264813631772995,
+      "learning_rate": 1.98e-05,
+      "loss": 0.2307,
+      "step": 100
+    },
+    {
+      "epoch": 0.1875293014533521,
+      "eval_loss": 0.18159574270248413,
+      "eval_runtime": 674.6463,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 100
+    },
+    {
+      "epoch": 0.1894045944678856,
+      "grad_norm": 0.2246488779783249,
+      "learning_rate": 2e-05,
+      "loss": 0.2421,
+      "step": 101
+    },
+    {
+      "epoch": 0.19127988748241911,
+      "grad_norm": 0.24814476072788239,
+      "learning_rate": 1.9992202729044836e-05,
+      "loss": 0.1653,
+      "step": 102
+    },
+    {
+      "epoch": 0.19315518049695266,
+      "grad_norm": 0.2789003849029541,
+      "learning_rate": 1.998440545808967e-05,
+      "loss": 0.2324,
+      "step": 103
+    },
+    {
+      "epoch": 0.19503047351148617,
+      "grad_norm": 0.21182158589363098,
+      "learning_rate": 1.9976608187134504e-05,
+      "loss": 0.2226,
+      "step": 104
+    },
+    {
+      "epoch": 0.19690576652601968,
+      "grad_norm": 0.299167662858963,
+      "learning_rate": 1.996881091617934e-05,
+      "loss": 0.2122,
+      "step": 105
+    },
+    {
+      "epoch": 0.19878105954055322,
+      "grad_norm": 0.22616952657699585,
+      "learning_rate": 1.9961013645224173e-05,
+      "loss": 0.2174,
+      "step": 106
+    },
+    {
+      "epoch": 0.20065635255508674,
+      "grad_norm": 0.19206267595291138,
+      "learning_rate": 1.9953216374269007e-05,
+      "loss": 0.1776,
+      "step": 107
+    },
+    {
+      "epoch": 0.20253164556962025,
+      "grad_norm": 0.2957805097103119,
+      "learning_rate": 1.994541910331384e-05,
+      "loss": 0.1576,
+      "step": 108
+    },
+    {
+      "epoch": 0.20440693858415376,
+      "grad_norm": 0.22284749150276184,
+      "learning_rate": 1.9937621832358675e-05,
+      "loss": 0.1495,
+      "step": 109
+    },
+    {
+      "epoch": 0.2062822315986873,
+      "grad_norm": 0.21203336119651794,
+      "learning_rate": 1.992982456140351e-05,
+      "loss": 0.186,
+      "step": 110
+    },
+    {
+      "epoch": 0.20815752461322082,
+      "grad_norm": 0.18476776778697968,
+      "learning_rate": 1.9922027290448344e-05,
+      "loss": 0.1626,
+      "step": 111
+    },
+    {
+      "epoch": 0.21003281762775433,
+      "grad_norm": 0.19962036609649658,
+      "learning_rate": 1.9914230019493178e-05,
+      "loss": 0.1447,
+      "step": 112
+    },
+    {
+      "epoch": 0.21190811064228784,
+      "grad_norm": 0.1857951283454895,
+      "learning_rate": 1.9906432748538015e-05,
+      "loss": 0.175,
+      "step": 113
+    },
+    {
+      "epoch": 0.21378340365682139,
+      "grad_norm": 0.2035515159368515,
+      "learning_rate": 1.989863547758285e-05,
+      "loss": 0.1732,
+      "step": 114
+    },
+    {
+      "epoch": 0.2156586966713549,
+      "grad_norm": 0.1710767149925232,
+      "learning_rate": 1.9890838206627684e-05,
+      "loss": 0.1317,
+      "step": 115
+    },
+    {
+      "epoch": 0.2175339896858884,
+      "grad_norm": 0.2920154929161072,
+      "learning_rate": 1.9883040935672515e-05,
+      "loss": 0.1802,
+      "step": 116
+    },
+    {
+      "epoch": 0.21940928270042195,
+      "grad_norm": 0.27223774790763855,
+      "learning_rate": 1.987524366471735e-05,
+      "loss": 0.1338,
+      "step": 117
+    },
+    {
+      "epoch": 0.22128457571495547,
+      "grad_norm": 0.1965789496898651,
+      "learning_rate": 1.9867446393762183e-05,
+      "loss": 0.1751,
+      "step": 118
+    },
+    {
+      "epoch": 0.22315986872948898,
+      "grad_norm": 0.19014707207679749,
+      "learning_rate": 1.9859649122807017e-05,
+      "loss": 0.1485,
+      "step": 119
+    },
+    {
+      "epoch": 0.2250351617440225,
+      "grad_norm": 0.251869261264801,
+      "learning_rate": 1.9851851851851855e-05,
+      "loss": 0.1416,
+      "step": 120
+    },
+    {
+      "epoch": 0.22691045475855603,
+      "grad_norm": 0.2957039475440979,
+      "learning_rate": 1.984405458089669e-05,
+      "loss": 0.1892,
+      "step": 121
+    },
+    {
+      "epoch": 0.22878574777308955,
+      "grad_norm": 0.17241084575653076,
+      "learning_rate": 1.9836257309941523e-05,
+      "loss": 0.1412,
+      "step": 122
+    },
+    {
+      "epoch": 0.23066104078762306,
+      "grad_norm": 0.2205045521259308,
+      "learning_rate": 1.9828460038986357e-05,
+      "loss": 0.1602,
+      "step": 123
+    },
+    {
+      "epoch": 0.23253633380215658,
+      "grad_norm": 0.21944566071033478,
+      "learning_rate": 1.982066276803119e-05,
+      "loss": 0.146,
+      "step": 124
+    },
+    {
+      "epoch": 0.23441162681669012,
+      "grad_norm": 0.21907442808151245,
+      "learning_rate": 1.9812865497076026e-05,
+      "loss": 0.1588,
+      "step": 125
+    },
+    {
+      "epoch": 0.23628691983122363,
+      "grad_norm": 0.17742829024791718,
+      "learning_rate": 1.980506822612086e-05,
+      "loss": 0.151,
+      "step": 126
+    },
+    {
+      "epoch": 0.23816221284575714,
+      "grad_norm": 0.23079413175582886,
+      "learning_rate": 1.9797270955165694e-05,
+      "loss": 0.2186,
+      "step": 127
+    },
+    {
+      "epoch": 0.24003750586029068,
+      "grad_norm": 0.19806325435638428,
+      "learning_rate": 1.9789473684210528e-05,
+      "loss": 0.1352,
+      "step": 128
+    },
+    {
+      "epoch": 0.2419127988748242,
+      "grad_norm": 0.1862531155347824,
+      "learning_rate": 1.9781676413255362e-05,
+      "loss": 0.1456,
+      "step": 129
+    },
+    {
+      "epoch": 0.2437880918893577,
+      "grad_norm": 0.1777540147304535,
+      "learning_rate": 1.9773879142300197e-05,
+      "loss": 0.1331,
+      "step": 130
+    },
+    {
+      "epoch": 0.24566338490389122,
+      "grad_norm": 0.29459261894226074,
+      "learning_rate": 1.976608187134503e-05,
+      "loss": 0.1421,
+      "step": 131
+    },
+    {
+      "epoch": 0.24753867791842477,
+      "grad_norm": 0.20495347678661346,
+      "learning_rate": 1.9758284600389865e-05,
+      "loss": 0.1599,
+      "step": 132
+    },
+    {
+      "epoch": 0.24941397093295828,
+      "grad_norm": 0.17535296082496643,
+      "learning_rate": 1.97504873294347e-05,
+      "loss": 0.17,
+      "step": 133
+    },
+    {
+      "epoch": 0.2512892639474918,
+      "grad_norm": 0.16180671751499176,
+      "learning_rate": 1.9742690058479533e-05,
+      "loss": 0.1367,
+      "step": 134
+    },
+    {
+      "epoch": 0.25316455696202533,
+      "grad_norm": 0.4137842357158661,
+      "learning_rate": 1.9734892787524368e-05,
+      "loss": 0.1536,
+      "step": 135
+    },
+    {
+      "epoch": 0.2550398499765588,
+      "grad_norm": 0.174374058842659,
+      "learning_rate": 1.9727095516569202e-05,
+      "loss": 0.1562,
+      "step": 136
+    },
+    {
+      "epoch": 0.25691514299109236,
+      "grad_norm": 0.1792580634355545,
+      "learning_rate": 1.9719298245614036e-05,
+      "loss": 0.174,
+      "step": 137
+    },
+    {
+      "epoch": 0.2587904360056259,
+      "grad_norm": 0.40365007519721985,
+      "learning_rate": 1.971150097465887e-05,
+      "loss": 0.2197,
+      "step": 138
+    },
+    {
+      "epoch": 0.2606657290201594,
+      "grad_norm": 0.1588585376739502,
+      "learning_rate": 1.9703703703703704e-05,
+      "loss": 0.1012,
+      "step": 139
+    },
+    {
+      "epoch": 0.26254102203469293,
+      "grad_norm": 0.1797696053981781,
+      "learning_rate": 1.969590643274854e-05,
+      "loss": 0.123,
+      "step": 140
+    },
+    {
+      "epoch": 0.26441631504922647,
+      "grad_norm": 0.41170960664749146,
+      "learning_rate": 1.9688109161793373e-05,
+      "loss": 0.2208,
+      "step": 141
+    },
+    {
+      "epoch": 0.26629160806375995,
+      "grad_norm": 0.22392353415489197,
+      "learning_rate": 1.9680311890838207e-05,
+      "loss": 0.1634,
+      "step": 142
+    },
+    {
+      "epoch": 0.2681669010782935,
+      "grad_norm": 0.2506747841835022,
+      "learning_rate": 1.9672514619883044e-05,
+      "loss": 0.1539,
+      "step": 143
+    },
+    {
+      "epoch": 0.270042194092827,
+      "grad_norm": 0.3422393500804901,
+      "learning_rate": 1.966471734892788e-05,
+      "loss": 0.2705,
+      "step": 144
+    },
+    {
+      "epoch": 0.2719174871073605,
+      "grad_norm": 0.15338042378425598,
+      "learning_rate": 1.9656920077972713e-05,
+      "loss": 0.1554,
+      "step": 145
+    },
+    {
+      "epoch": 0.27379278012189406,
+      "grad_norm": 0.4089084267616272,
+      "learning_rate": 1.9649122807017544e-05,
+      "loss": 0.1756,
+      "step": 146
+    },
+    {
+      "epoch": 0.27566807313642755,
+      "grad_norm": 0.19431771337985992,
+      "learning_rate": 1.9641325536062378e-05,
+      "loss": 0.1549,
+      "step": 147
+    },
+    {
+      "epoch": 0.2775433661509611,
+      "grad_norm": 0.21856912970542908,
+      "learning_rate": 1.9633528265107212e-05,
+      "loss": 0.1854,
+      "step": 148
+    },
+    {
+      "epoch": 0.27941865916549463,
+      "grad_norm": 0.21324169635772705,
+      "learning_rate": 1.962573099415205e-05,
+      "loss": 0.1348,
+      "step": 149
+    },
+    {
+      "epoch": 0.2812939521800281,
+      "grad_norm": 0.18593250215053558,
+      "learning_rate": 1.9617933723196884e-05,
+      "loss": 0.1309,
+      "step": 150
+    },
+    {
+      "epoch": 0.2812939521800281,
+      "eval_loss": 0.11937730759382248,
+      "eval_runtime": 675.2131,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 150
+    },
+    {
+      "epoch": 0.28316924519456166,
+      "grad_norm": 0.17237016558647156,
+      "learning_rate": 1.9610136452241718e-05,
+      "loss": 0.1228,
+      "step": 151
+    },
+    {
+      "epoch": 0.28504453820909514,
+      "grad_norm": 0.20051100850105286,
+      "learning_rate": 1.9602339181286552e-05,
+      "loss": 0.1028,
+      "step": 152
+    },
+    {
+      "epoch": 0.2869198312236287,
+      "grad_norm": 0.20913535356521606,
+      "learning_rate": 1.9594541910331386e-05,
+      "loss": 0.1478,
+      "step": 153
+    },
+    {
+      "epoch": 0.2887951242381622,
+      "grad_norm": 0.22947447001934052,
+      "learning_rate": 1.958674463937622e-05,
+      "loss": 0.1199,
+      "step": 154
+    },
+    {
+      "epoch": 0.2906704172526957,
+      "grad_norm": 0.21503888070583344,
+      "learning_rate": 1.9578947368421055e-05,
+      "loss": 0.1469,
+      "step": 155
+    },
+    {
+      "epoch": 0.29254571026722925,
+      "grad_norm": 0.15914572775363922,
+      "learning_rate": 1.957115009746589e-05,
+      "loss": 0.1126,
+      "step": 156
+    },
+    {
+      "epoch": 0.2944210032817628,
+      "grad_norm": 0.1764422208070755,
+      "learning_rate": 1.9563352826510723e-05,
+      "loss": 0.142,
+      "step": 157
+    },
+    {
+      "epoch": 0.2962962962962963,
+      "grad_norm": 0.20030224323272705,
+      "learning_rate": 1.9555555555555557e-05,
+      "loss": 0.1475,
+      "step": 158
+    },
+    {
+      "epoch": 0.2981715893108298,
+      "grad_norm": 0.21742001175880432,
+      "learning_rate": 1.954775828460039e-05,
+      "loss": 0.1874,
+      "step": 159
+    },
+    {
+      "epoch": 0.30004688232536336,
+      "grad_norm": 0.2284712940454483,
+      "learning_rate": 1.9539961013645226e-05,
+      "loss": 0.1532,
+      "step": 160
+    },
+    {
+      "epoch": 0.30192217533989685,
+      "grad_norm": 0.18156972527503967,
+      "learning_rate": 1.953216374269006e-05,
+      "loss": 0.092,
+      "step": 161
+    },
+    {
+      "epoch": 0.3037974683544304,
+      "grad_norm": 0.1754453182220459,
+      "learning_rate": 1.9524366471734894e-05,
+      "loss": 0.1348,
+      "step": 162
+    },
+    {
+      "epoch": 0.3056727613689639,
+      "grad_norm": 0.16860631108283997,
+      "learning_rate": 1.9516569200779728e-05,
+      "loss": 0.1176,
+      "step": 163
+    },
+    {
+      "epoch": 0.3075480543834974,
+      "grad_norm": 0.17378303408622742,
+      "learning_rate": 1.9508771929824562e-05,
+      "loss": 0.1372,
+      "step": 164
+    },
+    {
+      "epoch": 0.30942334739803096,
+      "grad_norm": 0.23950371146202087,
+      "learning_rate": 1.9500974658869397e-05,
+      "loss": 0.1206,
+      "step": 165
+    },
+    {
+      "epoch": 0.31129864041256444,
+      "grad_norm": 0.17646121978759766,
+      "learning_rate": 1.949317738791423e-05,
+      "loss": 0.1298,
+      "step": 166
+    },
+    {
+      "epoch": 0.313173933427098,
+      "grad_norm": 0.1811673790216446,
+      "learning_rate": 1.9485380116959065e-05,
+      "loss": 0.188,
+      "step": 167
+    },
+    {
+      "epoch": 0.3150492264416315,
+      "grad_norm": 0.27424830198287964,
+      "learning_rate": 1.94775828460039e-05,
+      "loss": 0.1422,
+      "step": 168
+    },
+    {
+      "epoch": 0.316924519456165,
+      "grad_norm": 0.21313942968845367,
+      "learning_rate": 1.9469785575048733e-05,
+      "loss": 0.1201,
+      "step": 169
+    },
+    {
+      "epoch": 0.31879981247069855,
+      "grad_norm": 0.1912909597158432,
+      "learning_rate": 1.9461988304093568e-05,
+      "loss": 0.0854,
+      "step": 170
+    },
+    {
+      "epoch": 0.3206751054852321,
+      "grad_norm": 0.2451699674129486,
+      "learning_rate": 1.9454191033138402e-05,
+      "loss": 0.2009,
+      "step": 171
+    },
+    {
+      "epoch": 0.3225503984997656,
+      "grad_norm": 0.1784246265888214,
+      "learning_rate": 1.944639376218324e-05,
+      "loss": 0.1587,
+      "step": 172
+    },
+    {
+      "epoch": 0.3244256915142991,
+      "grad_norm": 0.19816836714744568,
+      "learning_rate": 1.9438596491228074e-05,
+      "loss": 0.1426,
+      "step": 173
+    },
+    {
+      "epoch": 0.3263009845288326,
+      "grad_norm": 0.19529619812965393,
+      "learning_rate": 1.9430799220272908e-05,
+      "loss": 0.1431,
+      "step": 174
+    },
+    {
+      "epoch": 0.32817627754336615,
+      "grad_norm": 0.2105475664138794,
+      "learning_rate": 1.9423001949317742e-05,
+      "loss": 0.1654,
+      "step": 175
+    },
+    {
+      "epoch": 0.3300515705578997,
+      "grad_norm": 0.17807945609092712,
+      "learning_rate": 1.9415204678362573e-05,
+      "loss": 0.1366,
+      "step": 176
+    },
+    {
+      "epoch": 0.3319268635724332,
+      "grad_norm": 0.2023920863866806,
+      "learning_rate": 1.9407407407407407e-05,
+      "loss": 0.1372,
+      "step": 177
+    },
+    {
+      "epoch": 0.3338021565869667,
+      "grad_norm": 0.1807592213153839,
+      "learning_rate": 1.939961013645224e-05,
+      "loss": 0.1188,
+      "step": 178
+    },
+    {
+      "epoch": 0.33567744960150026,
+      "grad_norm": 0.18859116733074188,
+      "learning_rate": 1.939181286549708e-05,
+      "loss": 0.1243,
+      "step": 179
+    },
+    {
+      "epoch": 0.33755274261603374,
+      "grad_norm": 0.21937714517116547,
+      "learning_rate": 1.9384015594541913e-05,
+      "loss": 0.2015,
+      "step": 180
+    },
+    {
+      "epoch": 0.3394280356305673,
+      "grad_norm": 0.1638198345899582,
+      "learning_rate": 1.9376218323586747e-05,
+      "loss": 0.1031,
+      "step": 181
+    },
+    {
+      "epoch": 0.3413033286451008,
+      "grad_norm": 0.2563665807247162,
+      "learning_rate": 1.936842105263158e-05,
+      "loss": 0.1341,
+      "step": 182
+    },
+    {
+      "epoch": 0.3431786216596343,
+      "grad_norm": 0.19014421105384827,
+      "learning_rate": 1.9360623781676415e-05,
+      "loss": 0.135,
+      "step": 183
+    },
+    {
+      "epoch": 0.34505391467416785,
+      "grad_norm": 0.19648663699626923,
+      "learning_rate": 1.935282651072125e-05,
+      "loss": 0.1239,
+      "step": 184
+    },
+    {
+      "epoch": 0.34692920768870134,
+      "grad_norm": 0.16469500958919525,
+      "learning_rate": 1.9345029239766084e-05,
+      "loss": 0.1142,
+      "step": 185
+    },
+    {
+      "epoch": 0.3488045007032349,
+      "grad_norm": 0.17405149340629578,
+      "learning_rate": 1.9337231968810918e-05,
+      "loss": 0.1025,
+      "step": 186
+    },
+    {
+      "epoch": 0.3506797937177684,
+      "grad_norm": 0.2042636275291443,
+      "learning_rate": 1.9329434697855752e-05,
+      "loss": 0.0976,
+      "step": 187
+    },
+    {
+      "epoch": 0.3525550867323019,
+      "grad_norm": 0.2545097768306732,
+      "learning_rate": 1.9321637426900586e-05,
+      "loss": 0.1707,
+      "step": 188
+    },
+    {
+      "epoch": 0.35443037974683544,
+      "grad_norm": 0.17079883813858032,
+      "learning_rate": 1.931384015594542e-05,
+      "loss": 0.1282,
+      "step": 189
+    },
+    {
+      "epoch": 0.356305672761369,
+      "grad_norm": 0.18123818933963776,
+      "learning_rate": 1.9306042884990255e-05,
+      "loss": 0.1285,
+      "step": 190
+    },
+    {
+      "epoch": 0.35818096577590247,
+      "grad_norm": 0.26533034443855286,
+      "learning_rate": 1.929824561403509e-05,
+      "loss": 0.1912,
+      "step": 191
+    },
+    {
+      "epoch": 0.360056258790436,
+      "grad_norm": 0.21302203834056854,
+      "learning_rate": 1.9290448343079923e-05,
+      "loss": 0.152,
+      "step": 192
+    },
+    {
+      "epoch": 0.36193155180496955,
+      "grad_norm": 0.16439904272556305,
+      "learning_rate": 1.9282651072124757e-05,
+      "loss": 0.1026,
+      "step": 193
+    },
+    {
+      "epoch": 0.36380684481950304,
+      "grad_norm": 0.21921955049037933,
+      "learning_rate": 1.927485380116959e-05,
+      "loss": 0.1294,
+      "step": 194
+    },
+    {
+      "epoch": 0.3656821378340366,
+      "grad_norm": 0.1894323229789734,
+      "learning_rate": 1.9267056530214426e-05,
+      "loss": 0.1307,
+      "step": 195
+    },
+    {
+      "epoch": 0.36755743084857007,
+      "grad_norm": 0.18328344821929932,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 0.1168,
+      "step": 196
+    },
+    {
+      "epoch": 0.3694327238631036,
+      "grad_norm": 0.19325962662696838,
+      "learning_rate": 1.9251461988304094e-05,
+      "loss": 0.0979,
+      "step": 197
+    },
+    {
+      "epoch": 0.37130801687763715,
+      "grad_norm": 0.16475112736225128,
+      "learning_rate": 1.924366471734893e-05,
+      "loss": 0.11,
+      "step": 198
+    },
+    {
+      "epoch": 0.37318330989217063,
+      "grad_norm": 0.24370963871479034,
+      "learning_rate": 1.9235867446393763e-05,
+      "loss": 0.1256,
+      "step": 199
+    },
+    {
+      "epoch": 0.3750586029067042,
+      "grad_norm": 0.21204431354999542,
+      "learning_rate": 1.9228070175438597e-05,
+      "loss": 0.1498,
+      "step": 200
+    },
+    {
+      "epoch": 0.3750586029067042,
+      "eval_loss": 0.10520625114440918,
+      "eval_runtime": 675.3932,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 200
+    },
+    {
+      "epoch": 0.3769338959212377,
+      "grad_norm": 0.19303874671459198,
+      "learning_rate": 1.922027290448343e-05,
+      "loss": 0.1423,
+      "step": 201
+    },
+    {
+      "epoch": 0.3788091889357712,
+      "grad_norm": 0.19636684656143188,
+      "learning_rate": 1.921247563352827e-05,
+      "loss": 0.1325,
+      "step": 202
+    },
+    {
+      "epoch": 0.38068448195030474,
+      "grad_norm": 0.23951849341392517,
+      "learning_rate": 1.9204678362573103e-05,
+      "loss": 0.1276,
+      "step": 203
+    },
+    {
+      "epoch": 0.38255977496483823,
+      "grad_norm": 0.229360893368721,
+      "learning_rate": 1.9196881091617937e-05,
+      "loss": 0.121,
+      "step": 204
+    },
+    {
+      "epoch": 0.38443506797937177,
+      "grad_norm": 0.18153493106365204,
+      "learning_rate": 1.918908382066277e-05,
+      "loss": 0.1202,
+      "step": 205
+    },
+    {
+      "epoch": 0.3863103609939053,
+      "grad_norm": 0.4398716986179352,
+      "learning_rate": 1.9181286549707602e-05,
+      "loss": 0.1189,
+      "step": 206
+    },
+    {
+      "epoch": 0.3881856540084388,
+      "grad_norm": 0.20299677550792694,
+      "learning_rate": 1.9173489278752436e-05,
+      "loss": 0.1433,
+      "step": 207
+    },
+    {
+      "epoch": 0.39006094702297234,
+      "grad_norm": 0.2434886395931244,
+      "learning_rate": 1.916569200779727e-05,
+      "loss": 0.1122,
+      "step": 208
+    },
+    {
+      "epoch": 0.3919362400375059,
+      "grad_norm": 0.16485784947872162,
+      "learning_rate": 1.9157894736842108e-05,
+      "loss": 0.1505,
+      "step": 209
+    },
+    {
+      "epoch": 0.39381153305203936,
+      "grad_norm": 0.2433251440525055,
+      "learning_rate": 1.9150097465886942e-05,
+      "loss": 0.1821,
+      "step": 210
+    },
+    {
+      "epoch": 0.3956868260665729,
+      "grad_norm": 0.2099800854921341,
+      "learning_rate": 1.9142300194931776e-05,
+      "loss": 0.143,
+      "step": 211
+    },
+    {
+      "epoch": 0.39756211908110645,
+      "grad_norm": 0.20996929705142975,
+      "learning_rate": 1.913450292397661e-05,
+      "loss": 0.1394,
+      "step": 212
+    },
+    {
+      "epoch": 0.39943741209563993,
+      "grad_norm": 0.1913692057132721,
+      "learning_rate": 1.9126705653021445e-05,
+      "loss": 0.1408,
+      "step": 213
+    },
+    {
+      "epoch": 0.4013127051101735,
+      "grad_norm": 0.1733906865119934,
+      "learning_rate": 1.911890838206628e-05,
+      "loss": 0.1024,
+      "step": 214
+    },
+    {
+      "epoch": 0.40318799812470696,
+      "grad_norm": 0.1856503039598465,
+      "learning_rate": 1.9111111111111113e-05,
+      "loss": 0.1129,
+      "step": 215
+    },
+    {
+      "epoch": 0.4050632911392405,
+      "grad_norm": 0.3084484338760376,
+      "learning_rate": 1.9103313840155947e-05,
+      "loss": 0.102,
+      "step": 216
+    },
+    {
+      "epoch": 0.40693858415377404,
+      "grad_norm": 0.42217186093330383,
+      "learning_rate": 1.909551656920078e-05,
+      "loss": 0.1372,
+      "step": 217
+    },
+    {
+      "epoch": 0.4088138771683075,
+      "grad_norm": 0.20602688193321228,
+      "learning_rate": 1.9087719298245616e-05,
+      "loss": 0.1585,
+      "step": 218
+    },
+    {
+      "epoch": 0.41068917018284107,
+      "grad_norm": 0.524498701095581,
+      "learning_rate": 1.907992202729045e-05,
+      "loss": 0.1533,
+      "step": 219
+    },
+    {
+      "epoch": 0.4125644631973746,
+      "grad_norm": 0.20144020020961761,
+      "learning_rate": 1.9072124756335284e-05,
+      "loss": 0.1539,
+      "step": 220
+    },
+    {
+      "epoch": 0.4144397562119081,
+      "grad_norm": 0.17956125736236572,
+      "learning_rate": 1.9064327485380118e-05,
+      "loss": 0.1124,
+      "step": 221
+    },
+    {
+      "epoch": 0.41631504922644164,
+      "grad_norm": 0.20834237337112427,
+      "learning_rate": 1.9056530214424952e-05,
+      "loss": 0.1151,
+      "step": 222
+    },
+    {
+      "epoch": 0.4181903422409752,
+      "grad_norm": 0.24440927803516388,
+      "learning_rate": 1.9048732943469787e-05,
+      "loss": 0.1138,
+      "step": 223
+    },
+    {
+      "epoch": 0.42006563525550866,
+      "grad_norm": 0.1995476335287094,
+      "learning_rate": 1.904093567251462e-05,
+      "loss": 0.0915,
+      "step": 224
+    },
+    {
+      "epoch": 0.4219409282700422,
+      "grad_norm": 0.18664710223674774,
+      "learning_rate": 1.9033138401559458e-05,
+      "loss": 0.1295,
+      "step": 225
+    },
+    {
+      "epoch": 0.4238162212845757,
+      "grad_norm": 0.21624010801315308,
+      "learning_rate": 1.902534113060429e-05,
+      "loss": 0.1861,
+      "step": 226
+    },
+    {
+      "epoch": 0.42569151429910923,
+      "grad_norm": 0.27115607261657715,
+      "learning_rate": 1.9017543859649123e-05,
+      "loss": 0.0858,
+      "step": 227
+    },
+    {
+      "epoch": 0.42756680731364277,
+      "grad_norm": 0.2523871660232544,
+      "learning_rate": 1.9009746588693957e-05,
+      "loss": 0.1371,
+      "step": 228
+    },
+    {
+      "epoch": 0.42944210032817626,
+      "grad_norm": 0.20124512910842896,
+      "learning_rate": 1.900194931773879e-05,
+      "loss": 0.1079,
+      "step": 229
+    },
+    {
+      "epoch": 0.4313173933427098,
+      "grad_norm": 0.15628211200237274,
+      "learning_rate": 1.8994152046783626e-05,
+      "loss": 0.0926,
+      "step": 230
+    },
+    {
+      "epoch": 0.43319268635724334,
+      "grad_norm": 0.18558505177497864,
+      "learning_rate": 1.898635477582846e-05,
+      "loss": 0.1241,
+      "step": 231
+    },
+    {
+      "epoch": 0.4350679793717768,
+      "grad_norm": 0.21066918969154358,
+      "learning_rate": 1.8978557504873298e-05,
+      "loss": 0.1148,
+      "step": 232
+    },
+    {
+      "epoch": 0.43694327238631037,
+      "grad_norm": 0.24706971645355225,
+      "learning_rate": 1.8970760233918132e-05,
+      "loss": 0.1469,
+      "step": 233
+    },
+    {
+      "epoch": 0.4388185654008439,
+      "grad_norm": 0.24075715243816376,
+      "learning_rate": 1.8962962962962966e-05,
+      "loss": 0.1078,
+      "step": 234
+    },
+    {
+      "epoch": 0.4406938584153774,
+      "grad_norm": 0.9308987855911255,
+      "learning_rate": 1.89551656920078e-05,
+      "loss": 0.1242,
+      "step": 235
+    },
+    {
+      "epoch": 0.44256915142991093,
+      "grad_norm": 0.23705029487609863,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 0.1246,
+      "step": 236
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.20728544890880585,
+      "learning_rate": 1.8939571150097465e-05,
+      "loss": 0.1216,
+      "step": 237
+    },
+    {
+      "epoch": 0.44631973745897796,
+      "grad_norm": 0.21318766474723816,
+      "learning_rate": 1.8931773879142303e-05,
+      "loss": 0.1153,
+      "step": 238
+    },
+    {
+      "epoch": 0.4481950304735115,
+      "grad_norm": 0.22454850375652313,
+      "learning_rate": 1.8923976608187137e-05,
+      "loss": 0.1337,
+      "step": 239
+    },
+    {
+      "epoch": 0.450070323488045,
+      "grad_norm": 0.23081591725349426,
+      "learning_rate": 1.891617933723197e-05,
+      "loss": 0.1014,
+      "step": 240
+    },
+    {
+      "epoch": 0.45194561650257853,
+      "grad_norm": 0.2807827293872833,
+      "learning_rate": 1.8908382066276805e-05,
+      "loss": 0.1587,
+      "step": 241
+    },
+    {
+      "epoch": 0.45382090951711207,
+      "grad_norm": 0.22218219935894012,
+      "learning_rate": 1.890058479532164e-05,
+      "loss": 0.1599,
+      "step": 242
+    },
+    {
+      "epoch": 0.45569620253164556,
+      "grad_norm": 0.23275920748710632,
+      "learning_rate": 1.8892787524366474e-05,
+      "loss": 0.1336,
+      "step": 243
+    },
+    {
+      "epoch": 0.4575714955461791,
+      "grad_norm": 0.20733550190925598,
+      "learning_rate": 1.8884990253411308e-05,
+      "loss": 0.1458,
+      "step": 244
+    },
+    {
+      "epoch": 0.45944678856071264,
+      "grad_norm": 0.19315101206302643,
+      "learning_rate": 1.8877192982456142e-05,
+      "loss": 0.0923,
+      "step": 245
+    },
+    {
+      "epoch": 0.4613220815752461,
+      "grad_norm": 0.22976315021514893,
+      "learning_rate": 1.8869395711500976e-05,
+      "loss": 0.141,
+      "step": 246
+    },
+    {
+      "epoch": 0.46319737458977966,
+      "grad_norm": 0.40744727849960327,
+      "learning_rate": 1.886159844054581e-05,
+      "loss": 0.1416,
+      "step": 247
+    },
+    {
+      "epoch": 0.46507266760431315,
+      "grad_norm": 0.16530601680278778,
+      "learning_rate": 1.8853801169590645e-05,
+      "loss": 0.0989,
+      "step": 248
+    },
+    {
+      "epoch": 0.4669479606188467,
+      "grad_norm": 0.19157516956329346,
+      "learning_rate": 1.884600389863548e-05,
+      "loss": 0.1176,
+      "step": 249
+    },
+    {
+      "epoch": 0.46882325363338023,
+      "grad_norm": 0.28270554542541504,
+      "learning_rate": 1.8838206627680313e-05,
+      "loss": 0.1039,
+      "step": 250
+    },
+    {
+      "epoch": 0.46882325363338023,
+      "eval_loss": 0.09713947772979736,
+      "eval_runtime": 675.3394,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 250
+    },
+    {
+      "epoch": 0.4706985466479137,
+      "grad_norm": 0.2453300952911377,
+      "learning_rate": 1.8830409356725147e-05,
+      "loss": 0.113,
+      "step": 251
+    },
+    {
+      "epoch": 0.47257383966244726,
+      "grad_norm": 0.2129826545715332,
+      "learning_rate": 1.882261208576998e-05,
+      "loss": 0.143,
+      "step": 252
+    },
+    {
+      "epoch": 0.4744491326769808,
+      "grad_norm": 0.20310461521148682,
+      "learning_rate": 1.8814814814814816e-05,
+      "loss": 0.1242,
+      "step": 253
+    },
+    {
+      "epoch": 0.4763244256915143,
+      "grad_norm": 0.18484044075012207,
+      "learning_rate": 1.880701754385965e-05,
+      "loss": 0.1098,
+      "step": 254
+    },
+    {
+      "epoch": 0.4781997187060478,
+      "grad_norm": 0.2547452747821808,
+      "learning_rate": 1.8799220272904487e-05,
+      "loss": 0.1461,
+      "step": 255
+    },
+    {
+      "epoch": 0.48007501172058137,
+      "grad_norm": 0.2765617072582245,
+      "learning_rate": 1.8791423001949318e-05,
+      "loss": 0.1567,
+      "step": 256
+    },
+    {
+      "epoch": 0.48195030473511485,
+      "grad_norm": 0.19878943264484406,
+      "learning_rate": 1.8783625730994152e-05,
+      "loss": 0.1102,
+      "step": 257
+    },
+    {
+      "epoch": 0.4838255977496484,
+      "grad_norm": 0.3786822259426117,
+      "learning_rate": 1.8775828460038987e-05,
+      "loss": 0.1354,
+      "step": 258
+    },
+    {
+      "epoch": 0.4857008907641819,
+      "grad_norm": 0.18465487658977509,
+      "learning_rate": 1.876803118908382e-05,
+      "loss": 0.0959,
+      "step": 259
+    },
+    {
+      "epoch": 0.4875761837787154,
+      "grad_norm": 0.206305131316185,
+      "learning_rate": 1.8760233918128655e-05,
+      "loss": 0.0962,
+      "step": 260
+    },
+    {
+      "epoch": 0.48945147679324896,
+      "grad_norm": 0.22003507614135742,
+      "learning_rate": 1.8752436647173493e-05,
+      "loss": 0.1274,
+      "step": 261
+    },
+    {
+      "epoch": 0.49132676980778245,
+      "grad_norm": 0.17606805264949799,
+      "learning_rate": 1.8744639376218327e-05,
+      "loss": 0.0998,
+      "step": 262
+    },
+    {
+      "epoch": 0.493202062822316,
+      "grad_norm": 0.27991971373558044,
+      "learning_rate": 1.873684210526316e-05,
+      "loss": 0.1564,
+      "step": 263
+    },
+    {
+      "epoch": 0.49507735583684953,
+      "grad_norm": 0.23053331673145294,
+      "learning_rate": 1.8729044834307995e-05,
+      "loss": 0.1247,
+      "step": 264
+    },
+    {
+      "epoch": 0.496952648851383,
+      "grad_norm": 0.20774593949317932,
+      "learning_rate": 1.872124756335283e-05,
+      "loss": 0.0893,
+      "step": 265
+    },
+    {
+      "epoch": 0.49882794186591656,
+      "grad_norm": 0.20343434810638428,
+      "learning_rate": 1.871345029239766e-05,
+      "loss": 0.1096,
+      "step": 266
+    },
+    {
+      "epoch": 0.5007032348804501,
+      "grad_norm": 0.2922403812408447,
+      "learning_rate": 1.8705653021442494e-05,
+      "loss": 0.151,
+      "step": 267
+    },
+    {
+      "epoch": 0.5025785278949836,
+      "grad_norm": 0.21583294868469238,
+      "learning_rate": 1.8697855750487332e-05,
+      "loss": 0.0865,
+      "step": 268
+    },
+    {
+      "epoch": 0.5044538209095171,
+      "grad_norm": 0.21538986265659332,
+      "learning_rate": 1.8690058479532166e-05,
+      "loss": 0.1181,
+      "step": 269
+    },
+    {
+      "epoch": 0.5063291139240507,
+      "grad_norm": 0.2531646192073822,
+      "learning_rate": 1.8682261208577e-05,
+      "loss": 0.1502,
+      "step": 270
+    },
+    {
+      "epoch": 0.5082044069385842,
+      "grad_norm": 0.28955358266830444,
+      "learning_rate": 1.8674463937621834e-05,
+      "loss": 0.1335,
+      "step": 271
+    },
+    {
+      "epoch": 0.5100796999531176,
+      "grad_norm": 0.27051371335983276,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.1605,
+      "step": 272
+    },
+    {
+      "epoch": 0.5119549929676512,
+      "grad_norm": 0.15731023252010345,
+      "learning_rate": 1.8658869395711503e-05,
+      "loss": 0.0682,
+      "step": 273
+    },
+    {
+      "epoch": 0.5138302859821847,
+      "grad_norm": 0.20177946984767914,
+      "learning_rate": 1.8651072124756337e-05,
+      "loss": 0.1127,
+      "step": 274
+    },
+    {
+      "epoch": 0.5157055789967182,
+      "grad_norm": 0.2223084568977356,
+      "learning_rate": 1.864327485380117e-05,
+      "loss": 0.1318,
+      "step": 275
+    },
+    {
+      "epoch": 0.5175808720112518,
+      "grad_norm": 0.2421497255563736,
+      "learning_rate": 1.8635477582846005e-05,
+      "loss": 0.1372,
+      "step": 276
+    },
+    {
+      "epoch": 0.5194561650257853,
+      "grad_norm": 0.3211030662059784,
+      "learning_rate": 1.862768031189084e-05,
+      "loss": 0.1556,
+      "step": 277
+    },
+    {
+      "epoch": 0.5213314580403188,
+      "grad_norm": 0.2547368109226227,
+      "learning_rate": 1.8619883040935674e-05,
+      "loss": 0.1249,
+      "step": 278
+    },
+    {
+      "epoch": 0.5232067510548524,
+      "grad_norm": 0.23231586813926697,
+      "learning_rate": 1.8612085769980508e-05,
+      "loss": 0.1412,
+      "step": 279
+    },
+    {
+      "epoch": 0.5250820440693859,
+      "grad_norm": 0.21071861684322357,
+      "learning_rate": 1.8604288499025342e-05,
+      "loss": 0.105,
+      "step": 280
+    },
+    {
+      "epoch": 0.5269573370839193,
+      "grad_norm": 0.22929051518440247,
+      "learning_rate": 1.8596491228070176e-05,
+      "loss": 0.11,
+      "step": 281
+    },
+    {
+      "epoch": 0.5288326300984529,
+      "grad_norm": 0.22073444724082947,
+      "learning_rate": 1.858869395711501e-05,
+      "loss": 0.1031,
+      "step": 282
+    },
+    {
+      "epoch": 0.5307079231129864,
+      "grad_norm": 0.20839901268482208,
+      "learning_rate": 1.8580896686159845e-05,
+      "loss": 0.099,
+      "step": 283
+    },
+    {
+      "epoch": 0.5325832161275199,
+      "grad_norm": 0.2510814070701599,
+      "learning_rate": 1.857309941520468e-05,
+      "loss": 0.1614,
+      "step": 284
+    },
+    {
+      "epoch": 0.5344585091420534,
+      "grad_norm": 0.23160187900066376,
+      "learning_rate": 1.8565302144249517e-05,
+      "loss": 0.1281,
+      "step": 285
+    },
+    {
+      "epoch": 0.536333802156587,
+      "grad_norm": 0.3275841474533081,
+      "learning_rate": 1.8557504873294347e-05,
+      "loss": 0.0894,
+      "step": 286
+    },
+    {
+      "epoch": 0.5382090951711205,
+      "grad_norm": 0.20695282518863678,
+      "learning_rate": 1.854970760233918e-05,
+      "loss": 0.0911,
+      "step": 287
+    },
+    {
+      "epoch": 0.540084388185654,
+      "grad_norm": 0.19326826930046082,
+      "learning_rate": 1.8541910331384016e-05,
+      "loss": 0.1097,
+      "step": 288
+    },
+    {
+      "epoch": 0.5419596812001876,
+      "grad_norm": 0.22065812349319458,
+      "learning_rate": 1.853411306042885e-05,
+      "loss": 0.1249,
+      "step": 289
+    },
+    {
+      "epoch": 0.543834974214721,
+      "grad_norm": 0.3224574327468872,
+      "learning_rate": 1.8526315789473684e-05,
+      "loss": 0.1232,
+      "step": 290
+    },
+    {
+      "epoch": 0.5457102672292545,
+      "grad_norm": 0.20936869084835052,
+      "learning_rate": 1.851851851851852e-05,
+      "loss": 0.1061,
+      "step": 291
+    },
+    {
+      "epoch": 0.5475855602437881,
+      "grad_norm": 0.2805179953575134,
+      "learning_rate": 1.8510721247563356e-05,
+      "loss": 0.1176,
+      "step": 292
+    },
+    {
+      "epoch": 0.5494608532583216,
+      "grad_norm": 0.2184048891067505,
+      "learning_rate": 1.850292397660819e-05,
+      "loss": 0.0922,
+      "step": 293
+    },
+    {
+      "epoch": 0.5513361462728551,
+      "grad_norm": 0.2294335961341858,
+      "learning_rate": 1.8495126705653024e-05,
+      "loss": 0.1065,
+      "step": 294
+    },
+    {
+      "epoch": 0.5532114392873887,
+      "grad_norm": 0.20929084718227386,
+      "learning_rate": 1.848732943469786e-05,
+      "loss": 0.1041,
+      "step": 295
+    },
+    {
+      "epoch": 0.5550867323019222,
+      "grad_norm": 0.24733637273311615,
+      "learning_rate": 1.847953216374269e-05,
+      "loss": 0.1046,
+      "step": 296
+    },
+    {
+      "epoch": 0.5569620253164557,
+      "grad_norm": 0.26208677887916565,
+      "learning_rate": 1.8471734892787523e-05,
+      "loss": 0.0933,
+      "step": 297
+    },
+    {
+      "epoch": 0.5588373183309893,
+      "grad_norm": 0.20398394763469696,
+      "learning_rate": 1.846393762183236e-05,
+      "loss": 0.0967,
+      "step": 298
+    },
+    {
+      "epoch": 0.5607126113455227,
+      "grad_norm": 0.2356574982404709,
+      "learning_rate": 1.8456140350877195e-05,
+      "loss": 0.1627,
+      "step": 299
+    },
+    {
+      "epoch": 0.5625879043600562,
+      "grad_norm": 0.32398343086242676,
+      "learning_rate": 1.844834307992203e-05,
+      "loss": 0.1334,
+      "step": 300
+    },
+    {
+      "epoch": 0.5625879043600562,
+      "eval_loss": 0.09722544252872467,
+      "eval_runtime": 675.2264,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 300
+    },
+    {
+      "epoch": 0.5644631973745898,
+      "grad_norm": 0.25528204441070557,
+      "learning_rate": 1.8440545808966864e-05,
+      "loss": 0.1637,
+      "step": 301
+    },
+    {
+      "epoch": 0.5663384903891233,
+      "grad_norm": 0.2690048813819885,
+      "learning_rate": 1.8432748538011698e-05,
+      "loss": 0.178,
+      "step": 302
+    },
+    {
+      "epoch": 0.5682137834036568,
+      "grad_norm": 0.2270357310771942,
+      "learning_rate": 1.8424951267056532e-05,
+      "loss": 0.1252,
+      "step": 303
+    },
+    {
+      "epoch": 0.5700890764181903,
+      "grad_norm": 0.3335007131099701,
+      "learning_rate": 1.8417153996101366e-05,
+      "loss": 0.1594,
+      "step": 304
+    },
+    {
+      "epoch": 0.5719643694327239,
+      "grad_norm": 0.2951895594596863,
+      "learning_rate": 1.84093567251462e-05,
+      "loss": 0.1794,
+      "step": 305
+    },
+    {
+      "epoch": 0.5738396624472574,
+      "grad_norm": 0.19075483083724976,
+      "learning_rate": 1.8401559454191035e-05,
+      "loss": 0.1237,
+      "step": 306
+    },
+    {
+      "epoch": 0.5757149554617909,
+      "grad_norm": 0.25787994265556335,
+      "learning_rate": 1.839376218323587e-05,
+      "loss": 0.1151,
+      "step": 307
+    },
+    {
+      "epoch": 0.5775902484763245,
+      "grad_norm": 0.20336200296878815,
+      "learning_rate": 1.8385964912280703e-05,
+      "loss": 0.0893,
+      "step": 308
+    },
+    {
+      "epoch": 0.5794655414908579,
+      "grad_norm": 0.2570553123950958,
+      "learning_rate": 1.8378167641325537e-05,
+      "loss": 0.1197,
+      "step": 309
+    },
+    {
+      "epoch": 0.5813408345053914,
+      "grad_norm": 0.24566973745822906,
+      "learning_rate": 1.837037037037037e-05,
+      "loss": 0.1383,
+      "step": 310
+    },
+    {
+      "epoch": 0.583216127519925,
+      "grad_norm": 0.23258450627326965,
+      "learning_rate": 1.8362573099415205e-05,
+      "loss": 0.1182,
+      "step": 311
+    },
+    {
+      "epoch": 0.5850914205344585,
+      "grad_norm": 0.2346324771642685,
+      "learning_rate": 1.835477582846004e-05,
+      "loss": 0.128,
+      "step": 312
+    },
+    {
+      "epoch": 0.586966713548992,
+      "grad_norm": 0.21992027759552002,
+      "learning_rate": 1.8346978557504874e-05,
+      "loss": 0.0972,
+      "step": 313
+    },
+    {
+      "epoch": 0.5888420065635256,
+      "grad_norm": 0.21227417886257172,
+      "learning_rate": 1.833918128654971e-05,
+      "loss": 0.1038,
+      "step": 314
+    },
+    {
+      "epoch": 0.5907172995780591,
+      "grad_norm": 0.22464649379253387,
+      "learning_rate": 1.8331384015594546e-05,
+      "loss": 0.1452,
+      "step": 315
+    },
+    {
+      "epoch": 0.5925925925925926,
+      "grad_norm": 0.22292515635490417,
+      "learning_rate": 1.8323586744639376e-05,
+      "loss": 0.1343,
+      "step": 316
+    },
+    {
+      "epoch": 0.5944678856071262,
+      "grad_norm": 0.19903215765953064,
+      "learning_rate": 1.831578947368421e-05,
+      "loss": 0.144,
+      "step": 317
+    },
+    {
+      "epoch": 0.5963431786216596,
+      "grad_norm": 0.327104389667511,
+      "learning_rate": 1.8307992202729045e-05,
+      "loss": 0.1041,
+      "step": 318
+    },
+    {
+      "epoch": 0.5982184716361931,
+      "grad_norm": 0.25846344232559204,
+      "learning_rate": 1.830019493177388e-05,
+      "loss": 0.1504,
+      "step": 319
+    },
+    {
+      "epoch": 0.6000937646507267,
+      "grad_norm": 0.23256655037403107,
+      "learning_rate": 1.8292397660818713e-05,
+      "loss": 0.1239,
+      "step": 320
+    },
+    {
+      "epoch": 0.6019690576652602,
+      "grad_norm": 0.20890699326992035,
+      "learning_rate": 1.828460038986355e-05,
+      "loss": 0.112,
+      "step": 321
+    },
+    {
+      "epoch": 0.6038443506797937,
+      "grad_norm": 0.23920990526676178,
+      "learning_rate": 1.8276803118908385e-05,
+      "loss": 0.0993,
+      "step": 322
+    },
+    {
+      "epoch": 0.6057196436943273,
+      "grad_norm": 0.4082464575767517,
+      "learning_rate": 1.826900584795322e-05,
+      "loss": 0.1376,
+      "step": 323
+    },
+    {
+      "epoch": 0.6075949367088608,
+      "grad_norm": 0.30590200424194336,
+      "learning_rate": 1.8261208576998053e-05,
+      "loss": 0.0958,
+      "step": 324
+    },
+    {
+      "epoch": 0.6094702297233943,
+      "grad_norm": 0.2254190891981125,
+      "learning_rate": 1.8253411306042888e-05,
+      "loss": 0.1343,
+      "step": 325
+    },
+    {
+      "epoch": 0.6113455227379277,
+      "grad_norm": 0.3132902979850769,
+      "learning_rate": 1.824561403508772e-05,
+      "loss": 0.1007,
+      "step": 326
+    },
+    {
+      "epoch": 0.6132208157524613,
+      "grad_norm": 0.1695740818977356,
+      "learning_rate": 1.8237816764132556e-05,
+      "loss": 0.079,
+      "step": 327
+    },
+    {
+      "epoch": 0.6150961087669948,
+      "grad_norm": 0.24086889624595642,
+      "learning_rate": 1.823001949317739e-05,
+      "loss": 0.132,
+      "step": 328
+    },
+    {
+      "epoch": 0.6169714017815283,
+      "grad_norm": 0.30719810724258423,
+      "learning_rate": 1.8222222222222224e-05,
+      "loss": 0.1589,
+      "step": 329
+    },
+    {
+      "epoch": 0.6188466947960619,
+      "grad_norm": 0.2731982171535492,
+      "learning_rate": 1.821442495126706e-05,
+      "loss": 0.1366,
+      "step": 330
+    },
+    {
+      "epoch": 0.6207219878105954,
+      "grad_norm": 0.2092353105545044,
+      "learning_rate": 1.8206627680311893e-05,
+      "loss": 0.1234,
+      "step": 331
+    },
+    {
+      "epoch": 0.6225972808251289,
+      "grad_norm": 0.21123865246772766,
+      "learning_rate": 1.8198830409356727e-05,
+      "loss": 0.129,
+      "step": 332
+    },
+    {
+      "epoch": 0.6244725738396625,
+      "grad_norm": 0.3408866226673126,
+      "learning_rate": 1.819103313840156e-05,
+      "loss": 0.1415,
+      "step": 333
+    },
+    {
+      "epoch": 0.626347866854196,
+      "grad_norm": 0.22271405160427094,
+      "learning_rate": 1.8183235867446395e-05,
+      "loss": 0.0728,
+      "step": 334
+    },
+    {
+      "epoch": 0.6282231598687295,
+      "grad_norm": 0.2094489336013794,
+      "learning_rate": 1.817543859649123e-05,
+      "loss": 0.1083,
+      "step": 335
+    },
+    {
+      "epoch": 0.630098452883263,
+      "grad_norm": 0.20245423913002014,
+      "learning_rate": 1.8167641325536064e-05,
+      "loss": 0.095,
+      "step": 336
+    },
+    {
+      "epoch": 0.6319737458977965,
+      "grad_norm": 0.23320430517196655,
+      "learning_rate": 1.8159844054580898e-05,
+      "loss": 0.1613,
+      "step": 337
+    },
+    {
+      "epoch": 0.63384903891233,
+      "grad_norm": 0.2643381655216217,
+      "learning_rate": 1.8152046783625732e-05,
+      "loss": 0.1424,
+      "step": 338
+    },
+    {
+      "epoch": 0.6357243319268636,
+      "grad_norm": 0.2379055917263031,
+      "learning_rate": 1.8144249512670566e-05,
+      "loss": 0.1042,
+      "step": 339
+    },
+    {
+      "epoch": 0.6375996249413971,
+      "grad_norm": 0.21585410833358765,
+      "learning_rate": 1.81364522417154e-05,
+      "loss": 0.1092,
+      "step": 340
+    },
+    {
+      "epoch": 0.6394749179559306,
+      "grad_norm": 0.22150596976280212,
+      "learning_rate": 1.8128654970760235e-05,
+      "loss": 0.1266,
+      "step": 341
+    },
+    {
+      "epoch": 0.6413502109704642,
+      "grad_norm": 0.3069721758365631,
+      "learning_rate": 1.812085769980507e-05,
+      "loss": 0.1232,
+      "step": 342
+    },
+    {
+      "epoch": 0.6432255039849977,
+      "grad_norm": 0.22880715131759644,
+      "learning_rate": 1.8113060428849903e-05,
+      "loss": 0.1376,
+      "step": 343
+    },
+    {
+      "epoch": 0.6451007969995312,
+      "grad_norm": 0.22147296369075775,
+      "learning_rate": 1.810526315789474e-05,
+      "loss": 0.0861,
+      "step": 344
+    },
+    {
+      "epoch": 0.6469760900140648,
+      "grad_norm": 0.18794487416744232,
+      "learning_rate": 1.8097465886939575e-05,
+      "loss": 0.1047,
+      "step": 345
+    },
+    {
+      "epoch": 0.6488513830285982,
+      "grad_norm": 0.19053122401237488,
+      "learning_rate": 1.8089668615984406e-05,
+      "loss": 0.0852,
+      "step": 346
+    },
+    {
+      "epoch": 0.6507266760431317,
+      "grad_norm": 0.2369307428598404,
+      "learning_rate": 1.808187134502924e-05,
+      "loss": 0.1077,
+      "step": 347
+    },
+    {
+      "epoch": 0.6526019690576652,
+      "grad_norm": 0.2687581479549408,
+      "learning_rate": 1.8074074074074074e-05,
+      "loss": 0.1187,
+      "step": 348
+    },
+    {
+      "epoch": 0.6544772620721988,
+      "grad_norm": 0.32420602440834045,
+      "learning_rate": 1.8066276803118908e-05,
+      "loss": 0.1453,
+      "step": 349
+    },
+    {
+      "epoch": 0.6563525550867323,
+      "grad_norm": 0.2944568693637848,
+      "learning_rate": 1.8058479532163746e-05,
+      "loss": 0.0945,
+      "step": 350
+    },
+    {
+      "epoch": 0.6563525550867323,
+      "eval_loss": 0.08979390561580658,
+      "eval_runtime": 675.9452,
+      "eval_samples_per_second": 0.291,
+      "eval_steps_per_second": 0.291,
+      "step": 350
+    },
+    {
+      "epoch": 0.6582278481012658,
+      "grad_norm": 0.21131469309329987,
+      "learning_rate": 1.805068226120858e-05,
+      "loss": 0.0927,
+      "step": 351
+    },
+    {
+      "epoch": 0.6601031411157994,
+      "grad_norm": 0.2300959676504135,
+      "learning_rate": 1.8042884990253414e-05,
+      "loss": 0.1187,
+      "step": 352
+    },
+    {
+      "epoch": 0.6619784341303329,
+      "grad_norm": 0.21290843188762665,
+      "learning_rate": 1.8035087719298248e-05,
+      "loss": 0.0714,
+      "step": 353
+    },
+    {
+      "epoch": 0.6638537271448663,
+      "grad_norm": 0.22140546143054962,
+      "learning_rate": 1.8027290448343082e-05,
+      "loss": 0.0918,
+      "step": 354
+    },
+    {
+      "epoch": 0.6657290201593999,
+      "grad_norm": 0.21354801952838898,
+      "learning_rate": 1.8019493177387917e-05,
+      "loss": 0.081,
+      "step": 355
+    },
+    {
+      "epoch": 0.6676043131739334,
+      "grad_norm": 0.25323230028152466,
+      "learning_rate": 1.8011695906432747e-05,
+      "loss": 0.1041,
+      "step": 356
+    },
+    {
+      "epoch": 0.6694796061884669,
+      "grad_norm": 0.2276594042778015,
+      "learning_rate": 1.8003898635477585e-05,
+      "loss": 0.0924,
+      "step": 357
+    },
+    {
+      "epoch": 0.6713548992030005,
+      "grad_norm": 0.35995563864707947,
+      "learning_rate": 1.799610136452242e-05,
+      "loss": 0.1373,
+      "step": 358
+    },
+    {
+      "epoch": 0.673230192217534,
+      "grad_norm": 0.2169509083032608,
+      "learning_rate": 1.7988304093567253e-05,
+      "loss": 0.1004,
+      "step": 359
+    },
+    {
+      "epoch": 0.6751054852320675,
+      "grad_norm": 0.221743643283844,
+      "learning_rate": 1.7980506822612088e-05,
+      "loss": 0.0883,
+      "step": 360
+    },
+    {
+      "epoch": 0.6769807782466011,
+      "grad_norm": 0.1859385073184967,
+      "learning_rate": 1.7972709551656922e-05,
+      "loss": 0.0863,
+      "step": 361
+    },
+    {
+      "epoch": 0.6788560712611346,
+      "grad_norm": 0.2101774513721466,
+      "learning_rate": 1.7964912280701756e-05,
+      "loss": 0.105,
+      "step": 362
+    },
+    {
+      "epoch": 0.680731364275668,
+      "grad_norm": 0.2311290204524994,
+      "learning_rate": 1.795711500974659e-05,
+      "loss": 0.096,
+      "step": 363
+    },
+    {
+      "epoch": 0.6826066572902016,
+      "grad_norm": 0.2699367105960846,
+      "learning_rate": 1.7949317738791424e-05,
+      "loss": 0.1301,
+      "step": 364
+    },
+    {
+      "epoch": 0.6844819503047351,
+      "grad_norm": 0.31696122884750366,
+      "learning_rate": 1.794152046783626e-05,
+      "loss": 0.1594,
+      "step": 365
+    },
+    {
+      "epoch": 0.6863572433192686,
+      "grad_norm": 0.31495556235313416,
+      "learning_rate": 1.7933723196881093e-05,
+      "loss": 0.094,
+      "step": 366
+    },
+    {
+      "epoch": 0.6882325363338021,
+      "grad_norm": 0.2610797882080078,
+      "learning_rate": 1.7925925925925927e-05,
+      "loss": 0.1289,
+      "step": 367
+    },
+    {
+      "epoch": 0.6901078293483357,
+      "grad_norm": 0.2139683961868286,
+      "learning_rate": 1.791812865497076e-05,
+      "loss": 0.1032,
+      "step": 368
+    },
+    {
+      "epoch": 0.6919831223628692,
+      "grad_norm": 0.2474675327539444,
+      "learning_rate": 1.7910331384015595e-05,
+      "loss": 0.088,
+      "step": 369
+    },
+    {
+      "epoch": 0.6938584153774027,
+      "grad_norm": 0.22533686459064484,
+      "learning_rate": 1.790253411306043e-05,
+      "loss": 0.0933,
+      "step": 370
+    },
+    {
+      "epoch": 0.6957337083919363,
+      "grad_norm": 0.22728174924850464,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 0.1225,
+      "step": 371
+    },
+    {
+      "epoch": 0.6976090014064698,
+      "grad_norm": 0.23081286251544952,
+      "learning_rate": 1.7886939571150098e-05,
+      "loss": 0.1099,
+      "step": 372
+    },
+    {
+      "epoch": 0.6994842944210032,
+      "grad_norm": 0.2340676188468933,
+      "learning_rate": 1.7879142300194932e-05,
+      "loss": 0.1011,
+      "step": 373
+    },
+    {
+      "epoch": 0.7013595874355368,
+      "grad_norm": 0.2684282064437866,
+      "learning_rate": 1.787134502923977e-05,
+      "loss": 0.1667,
+      "step": 374
+    },
+    {
+      "epoch": 0.7032348804500703,
+      "grad_norm": 0.19310195744037628,
+      "learning_rate": 1.7863547758284604e-05,
+      "loss": 0.0958,
+      "step": 375
+    },
+    {
+      "epoch": 0.7051101734646038,
+      "grad_norm": 0.19192589819431305,
+      "learning_rate": 1.7855750487329435e-05,
+      "loss": 0.0769,
+      "step": 376
+    },
+    {
+      "epoch": 0.7069854664791374,
+      "grad_norm": 0.25004175305366516,
+      "learning_rate": 1.784795321637427e-05,
+      "loss": 0.1112,
+      "step": 377
+    },
+    {
+      "epoch": 0.7088607594936709,
+      "grad_norm": 0.2875761389732361,
+      "learning_rate": 1.7840155945419103e-05,
+      "loss": 0.1425,
+      "step": 378
+    },
+    {
+      "epoch": 0.7107360525082044,
+      "grad_norm": 0.2135677933692932,
+      "learning_rate": 1.7832358674463937e-05,
+      "loss": 0.0945,
+      "step": 379
+    },
+    {
+      "epoch": 0.712611345522738,
+      "grad_norm": 0.38285496830940247,
+      "learning_rate": 1.7824561403508775e-05,
+      "loss": 0.0824,
+      "step": 380
+    },
+    {
+      "epoch": 0.7144866385372715,
+      "grad_norm": 0.26780804991722107,
+      "learning_rate": 1.781676413255361e-05,
+      "loss": 0.1429,
+      "step": 381
+    },
+    {
+      "epoch": 0.7163619315518049,
+      "grad_norm": 0.2771541178226471,
+      "learning_rate": 1.7808966861598443e-05,
+      "loss": 0.1449,
+      "step": 382
+    },
+    {
+      "epoch": 0.7182372245663385,
+      "grad_norm": 0.23967161774635315,
+      "learning_rate": 1.7801169590643277e-05,
+      "loss": 0.109,
+      "step": 383
+    },
+    {
+      "epoch": 0.720112517580872,
+      "grad_norm": 0.2565579414367676,
+      "learning_rate": 1.779337231968811e-05,
+      "loss": 0.143,
+      "step": 384
+    },
+    {
+      "epoch": 0.7219878105954055,
+      "grad_norm": 0.23335972428321838,
+      "learning_rate": 1.7785575048732946e-05,
+      "loss": 0.0836,
+      "step": 385
+    },
+    {
+      "epoch": 0.7238631036099391,
+      "grad_norm": 0.21833084523677826,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.0927,
+      "step": 386
+    },
+    {
+      "epoch": 0.7257383966244726,
+      "grad_norm": 0.29433101415634155,
+      "learning_rate": 1.7769980506822614e-05,
+      "loss": 0.102,
+      "step": 387
+    },
+    {
+      "epoch": 0.7276136896390061,
+      "grad_norm": 0.23135854303836823,
+      "learning_rate": 1.776218323586745e-05,
+      "loss": 0.1139,
+      "step": 388
+    },
+    {
+      "epoch": 0.7294889826535396,
+      "grad_norm": 0.24812403321266174,
+      "learning_rate": 1.7754385964912283e-05,
+      "loss": 0.1098,
+      "step": 389
+    },
+    {
+      "epoch": 0.7313642756680732,
+      "grad_norm": 0.2173132598400116,
+      "learning_rate": 1.7746588693957117e-05,
+      "loss": 0.1089,
+      "step": 390
+    },
+    {
+      "epoch": 0.7332395686826066,
+      "grad_norm": 0.24950866401195526,
+      "learning_rate": 1.773879142300195e-05,
+      "loss": 0.0805,
+      "step": 391
+    },
+    {
+      "epoch": 0.7351148616971401,
+      "grad_norm": 0.29122406244277954,
+      "learning_rate": 1.7730994152046785e-05,
+      "loss": 0.1481,
+      "step": 392
+    },
+    {
+      "epoch": 0.7369901547116737,
+      "grad_norm": 0.2415425032377243,
+      "learning_rate": 1.772319688109162e-05,
+      "loss": 0.1364,
+      "step": 393
+    },
+    {
+      "epoch": 0.7388654477262072,
+      "grad_norm": 0.2071705311536789,
+      "learning_rate": 1.7715399610136454e-05,
+      "loss": 0.1227,
+      "step": 394
+    },
+    {
+      "epoch": 0.7407407407407407,
+      "grad_norm": 0.20239049196243286,
+      "learning_rate": 1.7707602339181288e-05,
+      "loss": 0.0864,
+      "step": 395
+    },
+    {
+      "epoch": 0.7426160337552743,
+      "grad_norm": 0.25537100434303284,
+      "learning_rate": 1.7699805068226122e-05,
+      "loss": 0.1278,
+      "step": 396
+    },
+    {
+      "epoch": 0.7444913267698078,
+      "grad_norm": 0.30780285596847534,
+      "learning_rate": 1.7692007797270956e-05,
+      "loss": 0.1586,
+      "step": 397
+    },
+    {
+      "epoch": 0.7463666197843413,
+      "grad_norm": 0.36921605467796326,
+      "learning_rate": 1.768421052631579e-05,
+      "loss": 0.1244,
+      "step": 398
+    },
+    {
+      "epoch": 0.7482419127988749,
+      "grad_norm": 0.2948741614818573,
+      "learning_rate": 1.7676413255360624e-05,
+      "loss": 0.1095,
+      "step": 399
+    },
+    {
+      "epoch": 0.7501172058134083,
+      "grad_norm": 0.23762205243110657,
+      "learning_rate": 1.766861598440546e-05,
+      "loss": 0.1245,
+      "step": 400
+    },
+    {
+      "epoch": 0.7501172058134083,
+      "eval_loss": 0.08483699709177017,
+      "eval_runtime": 675.3388,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 400
+    },
+    {
+      "epoch": 0.7519924988279418,
+      "grad_norm": 0.24527454376220703,
+      "learning_rate": 1.7660818713450293e-05,
+      "loss": 0.1122,
+      "step": 401
+    },
+    {
+      "epoch": 0.7538677918424754,
+      "grad_norm": 0.22449228167533875,
+      "learning_rate": 1.7653021442495127e-05,
+      "loss": 0.1044,
+      "step": 402
+    },
+    {
+      "epoch": 0.7557430848570089,
+      "grad_norm": 0.285155713558197,
+      "learning_rate": 1.7645224171539965e-05,
+      "loss": 0.1141,
+      "step": 403
+    },
+    {
+      "epoch": 0.7576183778715424,
+      "grad_norm": 0.247589111328125,
+      "learning_rate": 1.76374269005848e-05,
+      "loss": 0.1363,
+      "step": 404
+    },
+    {
+      "epoch": 0.759493670886076,
+      "grad_norm": 0.26343780755996704,
+      "learning_rate": 1.7629629629629633e-05,
+      "loss": 0.1198,
+      "step": 405
+    },
+    {
+      "epoch": 0.7613689639006095,
+      "grad_norm": 0.2533006966114044,
+      "learning_rate": 1.7621832358674464e-05,
+      "loss": 0.1137,
+      "step": 406
+    },
+    {
+      "epoch": 0.763244256915143,
+      "grad_norm": 0.22146250307559967,
+      "learning_rate": 1.7614035087719298e-05,
+      "loss": 0.1152,
+      "step": 407
+    },
+    {
+      "epoch": 0.7651195499296765,
+      "grad_norm": 0.3412543535232544,
+      "learning_rate": 1.7606237816764132e-05,
+      "loss": 0.1838,
+      "step": 408
+    },
+    {
+      "epoch": 0.76699484294421,
+      "grad_norm": 0.2710413634777069,
+      "learning_rate": 1.7598440545808966e-05,
+      "loss": 0.1088,
+      "step": 409
+    },
+    {
+      "epoch": 0.7688701359587435,
+      "grad_norm": 0.2602677047252655,
+      "learning_rate": 1.7590643274853804e-05,
+      "loss": 0.135,
+      "step": 410
+    },
+    {
+      "epoch": 0.770745428973277,
+      "grad_norm": 0.22832591831684113,
+      "learning_rate": 1.7582846003898638e-05,
+      "loss": 0.099,
+      "step": 411
+    },
+    {
+      "epoch": 0.7726207219878106,
+      "grad_norm": 0.20037035644054413,
+      "learning_rate": 1.7575048732943472e-05,
+      "loss": 0.0988,
+      "step": 412
+    },
+    {
+      "epoch": 0.7744960150023441,
+      "grad_norm": 0.21016646921634674,
+      "learning_rate": 1.7567251461988307e-05,
+      "loss": 0.0859,
+      "step": 413
+    },
+    {
+      "epoch": 0.7763713080168776,
+      "grad_norm": 0.25442710518836975,
+      "learning_rate": 1.755945419103314e-05,
+      "loss": 0.1192,
+      "step": 414
+    },
+    {
+      "epoch": 0.7782466010314112,
+      "grad_norm": 0.2266324907541275,
+      "learning_rate": 1.7551656920077975e-05,
+      "loss": 0.106,
+      "step": 415
+    },
+    {
+      "epoch": 0.7801218940459447,
+      "grad_norm": 0.24464181065559387,
+      "learning_rate": 1.754385964912281e-05,
+      "loss": 0.116,
+      "step": 416
+    },
+    {
+      "epoch": 0.7819971870604782,
+      "grad_norm": 0.19099155068397522,
+      "learning_rate": 1.7536062378167643e-05,
+      "loss": 0.0759,
+      "step": 417
+    },
+    {
+      "epoch": 0.7838724800750118,
+      "grad_norm": 0.23417051136493683,
+      "learning_rate": 1.7528265107212477e-05,
+      "loss": 0.0815,
+      "step": 418
+    },
+    {
+      "epoch": 0.7857477730895452,
+      "grad_norm": 0.3102501630783081,
+      "learning_rate": 1.752046783625731e-05,
+      "loss": 0.1205,
+      "step": 419
+    },
+    {
+      "epoch": 0.7876230661040787,
+      "grad_norm": 0.27138587832450867,
+      "learning_rate": 1.7512670565302146e-05,
+      "loss": 0.0957,
+      "step": 420
+    },
+    {
+      "epoch": 0.7894983591186123,
+      "grad_norm": 0.3055587708950043,
+      "learning_rate": 1.750487329434698e-05,
+      "loss": 0.1181,
+      "step": 421
+    },
+    {
+      "epoch": 0.7913736521331458,
+      "grad_norm": 0.24075333774089813,
+      "learning_rate": 1.7497076023391814e-05,
+      "loss": 0.1192,
+      "step": 422
+    },
+    {
+      "epoch": 0.7932489451476793,
+      "grad_norm": 0.1806018054485321,
+      "learning_rate": 1.748927875243665e-05,
+      "loss": 0.0884,
+      "step": 423
+    },
+    {
+      "epoch": 0.7951242381622129,
+      "grad_norm": 0.2315697968006134,
+      "learning_rate": 1.7481481481481483e-05,
+      "loss": 0.0696,
+      "step": 424
+    },
+    {
+      "epoch": 0.7969995311767464,
+      "grad_norm": 0.23765219748020172,
+      "learning_rate": 1.7473684210526317e-05,
+      "loss": 0.1186,
+      "step": 425
+    },
+    {
+      "epoch": 0.7988748241912799,
+      "grad_norm": 0.18785423040390015,
+      "learning_rate": 1.746588693957115e-05,
+      "loss": 0.0727,
+      "step": 426
+    },
+    {
+      "epoch": 0.8007501172058135,
+      "grad_norm": 0.1897609829902649,
+      "learning_rate": 1.7458089668615985e-05,
+      "loss": 0.0926,
+      "step": 427
+    },
+    {
+      "epoch": 0.802625410220347,
+      "grad_norm": 0.28181466460227966,
+      "learning_rate": 1.745029239766082e-05,
+      "loss": 0.1083,
+      "step": 428
+    },
+    {
+      "epoch": 0.8045007032348804,
+      "grad_norm": 0.20873577892780304,
+      "learning_rate": 1.7442495126705654e-05,
+      "loss": 0.105,
+      "step": 429
+    },
+    {
+      "epoch": 0.8063759962494139,
+      "grad_norm": 0.23470522463321686,
+      "learning_rate": 1.7434697855750488e-05,
+      "loss": 0.1103,
+      "step": 430
+    },
+    {
+      "epoch": 0.8082512892639475,
+      "grad_norm": 0.26746684312820435,
+      "learning_rate": 1.7426900584795322e-05,
+      "loss": 0.1322,
+      "step": 431
+    },
+    {
+      "epoch": 0.810126582278481,
+      "grad_norm": 0.2638936936855316,
+      "learning_rate": 1.7419103313840156e-05,
+      "loss": 0.1228,
+      "step": 432
+    },
+    {
+      "epoch": 0.8120018752930145,
+      "grad_norm": 0.237047016620636,
+      "learning_rate": 1.7411306042884994e-05,
+      "loss": 0.1042,
+      "step": 433
+    },
+    {
+      "epoch": 0.8138771683075481,
+      "grad_norm": 0.1885695904493332,
+      "learning_rate": 1.7403508771929828e-05,
+      "loss": 0.0723,
+      "step": 434
+    },
+    {
+      "epoch": 0.8157524613220816,
+      "grad_norm": 0.21713057160377502,
+      "learning_rate": 1.7395711500974662e-05,
+      "loss": 0.0796,
+      "step": 435
+    },
+    {
+      "epoch": 0.817627754336615,
+      "grad_norm": 0.3199704885482788,
+      "learning_rate": 1.7387914230019493e-05,
+      "loss": 0.1588,
+      "step": 436
+    },
+    {
+      "epoch": 0.8195030473511487,
+      "grad_norm": 0.3154192566871643,
+      "learning_rate": 1.7380116959064327e-05,
+      "loss": 0.1365,
+      "step": 437
+    },
+    {
+      "epoch": 0.8213783403656821,
+      "grad_norm": 0.2530040144920349,
+      "learning_rate": 1.737231968810916e-05,
+      "loss": 0.0948,
+      "step": 438
+    },
+    {
+      "epoch": 0.8232536333802156,
+      "grad_norm": 0.23530352115631104,
+      "learning_rate": 1.7364522417154e-05,
+      "loss": 0.119,
+      "step": 439
+    },
+    {
+      "epoch": 0.8251289263947492,
+      "grad_norm": 0.2744591534137726,
+      "learning_rate": 1.7356725146198833e-05,
+      "loss": 0.1177,
+      "step": 440
+    },
+    {
+      "epoch": 0.8270042194092827,
+      "grad_norm": 0.21735382080078125,
+      "learning_rate": 1.7348927875243667e-05,
+      "loss": 0.0836,
+      "step": 441
+    },
+    {
+      "epoch": 0.8288795124238162,
+      "grad_norm": 0.21691051125526428,
+      "learning_rate": 1.73411306042885e-05,
+      "loss": 0.1232,
+      "step": 442
+    },
+    {
+      "epoch": 0.8307548054383498,
+      "grad_norm": 0.26144662499427795,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.115,
+      "step": 443
+    },
+    {
+      "epoch": 0.8326300984528833,
+      "grad_norm": 0.2164791226387024,
+      "learning_rate": 1.732553606237817e-05,
+      "loss": 0.1146,
+      "step": 444
+    },
+    {
+      "epoch": 0.8345053914674168,
+      "grad_norm": 0.27034950256347656,
+      "learning_rate": 1.7317738791423004e-05,
+      "loss": 0.0644,
+      "step": 445
+    },
+    {
+      "epoch": 0.8363806844819504,
+      "grad_norm": 0.19503502547740936,
+      "learning_rate": 1.7309941520467838e-05,
+      "loss": 0.0688,
+      "step": 446
+    },
+    {
+      "epoch": 0.8382559774964838,
+      "grad_norm": 0.23584571480751038,
+      "learning_rate": 1.7302144249512672e-05,
+      "loss": 0.1256,
+      "step": 447
+    },
+    {
+      "epoch": 0.8401312705110173,
+      "grad_norm": 0.2520376145839691,
+      "learning_rate": 1.7294346978557507e-05,
+      "loss": 0.1601,
+      "step": 448
+    },
+    {
+      "epoch": 0.8420065635255509,
+      "grad_norm": 0.24150918424129486,
+      "learning_rate": 1.728654970760234e-05,
+      "loss": 0.1155,
+      "step": 449
+    },
+    {
+      "epoch": 0.8438818565400844,
+      "grad_norm": 0.289753258228302,
+      "learning_rate": 1.7278752436647175e-05,
+      "loss": 0.1283,
+      "step": 450
+    },
+    {
+      "epoch": 0.8438818565400844,
+      "eval_loss": 0.08511354774236679,
+      "eval_runtime": 675.7661,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 450
+    },
+    {
+      "epoch": 0.8457571495546179,
+      "grad_norm": 0.27476435899734497,
+      "learning_rate": 1.727095516569201e-05,
+      "loss": 0.0989,
+      "step": 451
+    },
+    {
+      "epoch": 0.8476324425691514,
+      "grad_norm": 0.20759902894496918,
+      "learning_rate": 1.7263157894736843e-05,
+      "loss": 0.0963,
+      "step": 452
+    },
+    {
+      "epoch": 0.849507735583685,
+      "grad_norm": 0.3410794138908386,
+      "learning_rate": 1.7255360623781678e-05,
+      "loss": 0.1675,
+      "step": 453
+    },
+    {
+      "epoch": 0.8513830285982185,
+      "grad_norm": 0.2817666828632355,
+      "learning_rate": 1.7247563352826512e-05,
+      "loss": 0.1458,
+      "step": 454
+    },
+    {
+      "epoch": 0.853258321612752,
+      "grad_norm": 0.2924879491329193,
+      "learning_rate": 1.7239766081871346e-05,
+      "loss": 0.1393,
+      "step": 455
+    },
+    {
+      "epoch": 0.8551336146272855,
+      "grad_norm": 0.3008408844470978,
+      "learning_rate": 1.723196881091618e-05,
+      "loss": 0.1051,
+      "step": 456
+    },
+    {
+      "epoch": 0.857008907641819,
+      "grad_norm": 0.3666384518146515,
+      "learning_rate": 1.7224171539961014e-05,
+      "loss": 0.1031,
+      "step": 457
+    },
+    {
+      "epoch": 0.8588842006563525,
+      "grad_norm": 0.2442241907119751,
+      "learning_rate": 1.721637426900585e-05,
+      "loss": 0.1333,
+      "step": 458
+    },
+    {
+      "epoch": 0.8607594936708861,
+      "grad_norm": 0.3462783992290497,
+      "learning_rate": 1.7208576998050683e-05,
+      "loss": 0.115,
+      "step": 459
+    },
+    {
+      "epoch": 0.8626347866854196,
+      "grad_norm": 0.25634944438934326,
+      "learning_rate": 1.7200779727095517e-05,
+      "loss": 0.0867,
+      "step": 460
+    },
+    {
+      "epoch": 0.8645100796999531,
+      "grad_norm": 0.30906710028648376,
+      "learning_rate": 1.719298245614035e-05,
+      "loss": 0.1405,
+      "step": 461
+    },
+    {
+      "epoch": 0.8663853727144867,
+      "grad_norm": 0.2611568570137024,
+      "learning_rate": 1.7185185185185185e-05,
+      "loss": 0.1071,
+      "step": 462
+    },
+    {
+      "epoch": 0.8682606657290202,
+      "grad_norm": 0.2521813213825226,
+      "learning_rate": 1.7177387914230023e-05,
+      "loss": 0.0707,
+      "step": 463
+    },
+    {
+      "epoch": 0.8701359587435537,
+      "grad_norm": 0.25580745935440063,
+      "learning_rate": 1.7169590643274857e-05,
+      "loss": 0.0854,
+      "step": 464
+    },
+    {
+      "epoch": 0.8720112517580872,
+      "grad_norm": 0.2787662148475647,
+      "learning_rate": 1.716179337231969e-05,
+      "loss": 0.1018,
+      "step": 465
+    },
+    {
+      "epoch": 0.8738865447726207,
+      "grad_norm": 0.21670454740524292,
+      "learning_rate": 1.7153996101364522e-05,
+      "loss": 0.1214,
+      "step": 466
+    },
+    {
+      "epoch": 0.8757618377871542,
+      "grad_norm": 0.28645867109298706,
+      "learning_rate": 1.7146198830409356e-05,
+      "loss": 0.0877,
+      "step": 467
+    },
+    {
+      "epoch": 0.8776371308016878,
+      "grad_norm": 0.20811931788921356,
+      "learning_rate": 1.713840155945419e-05,
+      "loss": 0.0846,
+      "step": 468
+    },
+    {
+      "epoch": 0.8795124238162213,
+      "grad_norm": 0.301693320274353,
+      "learning_rate": 1.7130604288499028e-05,
+      "loss": 0.1215,
+      "step": 469
+    },
+    {
+      "epoch": 0.8813877168307548,
+      "grad_norm": 0.21408531069755554,
+      "learning_rate": 1.7122807017543862e-05,
+      "loss": 0.0929,
+      "step": 470
+    },
+    {
+      "epoch": 0.8832630098452883,
+      "grad_norm": 0.20539730787277222,
+      "learning_rate": 1.7115009746588696e-05,
+      "loss": 0.0689,
+      "step": 471
+    },
+    {
+      "epoch": 0.8851383028598219,
+      "grad_norm": 0.2104424089193344,
+      "learning_rate": 1.710721247563353e-05,
+      "loss": 0.0997,
+      "step": 472
+    },
+    {
+      "epoch": 0.8870135958743554,
+      "grad_norm": 0.7082309722900391,
+      "learning_rate": 1.7099415204678365e-05,
+      "loss": 0.0945,
+      "step": 473
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.19709636270999908,
+      "learning_rate": 1.70916179337232e-05,
+      "loss": 0.0781,
+      "step": 474
+    },
+    {
+      "epoch": 0.8907641819034224,
+      "grad_norm": 0.26817139983177185,
+      "learning_rate": 1.7083820662768033e-05,
+      "loss": 0.0823,
+      "step": 475
+    },
+    {
+      "epoch": 0.8926394749179559,
+      "grad_norm": 0.26233142614364624,
+      "learning_rate": 1.7076023391812867e-05,
+      "loss": 0.0938,
+      "step": 476
+    },
+    {
+      "epoch": 0.8945147679324894,
+      "grad_norm": 0.3621986508369446,
+      "learning_rate": 1.70682261208577e-05,
+      "loss": 0.1134,
+      "step": 477
+    },
+    {
+      "epoch": 0.896390060947023,
+      "grad_norm": 0.2838749289512634,
+      "learning_rate": 1.7060428849902536e-05,
+      "loss": 0.1347,
+      "step": 478
+    },
+    {
+      "epoch": 0.8982653539615565,
+      "grad_norm": 0.19387036561965942,
+      "learning_rate": 1.705263157894737e-05,
+      "loss": 0.065,
+      "step": 479
+    },
+    {
+      "epoch": 0.90014064697609,
+      "grad_norm": 0.20621807873249054,
+      "learning_rate": 1.7044834307992204e-05,
+      "loss": 0.0731,
+      "step": 480
+    },
+    {
+      "epoch": 0.9020159399906236,
+      "grad_norm": 0.2323528677225113,
+      "learning_rate": 1.7037037037037038e-05,
+      "loss": 0.0931,
+      "step": 481
+    },
+    {
+      "epoch": 0.9038912330051571,
+      "grad_norm": 0.20820242166519165,
+      "learning_rate": 1.7029239766081872e-05,
+      "loss": 0.0853,
+      "step": 482
+    },
+    {
+      "epoch": 0.9057665260196905,
+      "grad_norm": 0.27471327781677246,
+      "learning_rate": 1.7021442495126707e-05,
+      "loss": 0.1377,
+      "step": 483
+    },
+    {
+      "epoch": 0.9076418190342241,
+      "grad_norm": 0.22734041512012482,
+      "learning_rate": 1.701364522417154e-05,
+      "loss": 0.1169,
+      "step": 484
+    },
+    {
+      "epoch": 0.9095171120487576,
+      "grad_norm": 0.25026533007621765,
+      "learning_rate": 1.7005847953216375e-05,
+      "loss": 0.0976,
+      "step": 485
+    },
+    {
+      "epoch": 0.9113924050632911,
+      "grad_norm": 0.24165751039981842,
+      "learning_rate": 1.699805068226121e-05,
+      "loss": 0.1008,
+      "step": 486
+    },
+    {
+      "epoch": 0.9132676980778247,
+      "grad_norm": 0.24641317129135132,
+      "learning_rate": 1.6990253411306043e-05,
+      "loss": 0.1105,
+      "step": 487
+    },
+    {
+      "epoch": 0.9151429910923582,
+      "grad_norm": 0.18702614307403564,
+      "learning_rate": 1.6982456140350878e-05,
+      "loss": 0.0711,
+      "step": 488
+    },
+    {
+      "epoch": 0.9170182841068917,
+      "grad_norm": 0.31801968812942505,
+      "learning_rate": 1.6974658869395712e-05,
+      "loss": 0.1112,
+      "step": 489
+    },
+    {
+      "epoch": 0.9188935771214253,
+      "grad_norm": 0.311109721660614,
+      "learning_rate": 1.6966861598440546e-05,
+      "loss": 0.128,
+      "step": 490
+    },
+    {
+      "epoch": 0.9207688701359588,
+      "grad_norm": 0.24190711975097656,
+      "learning_rate": 1.695906432748538e-05,
+      "loss": 0.0873,
+      "step": 491
+    },
+    {
+      "epoch": 0.9226441631504922,
+      "grad_norm": 0.32772934436798096,
+      "learning_rate": 1.6951267056530218e-05,
+      "loss": 0.0979,
+      "step": 492
+    },
+    {
+      "epoch": 0.9245194561650257,
+      "grad_norm": 0.23775866627693176,
+      "learning_rate": 1.6943469785575052e-05,
+      "loss": 0.096,
+      "step": 493
+    },
+    {
+      "epoch": 0.9263947491795593,
+      "grad_norm": 0.38035061955451965,
+      "learning_rate": 1.6935672514619886e-05,
+      "loss": 0.1299,
+      "step": 494
+    },
+    {
+      "epoch": 0.9282700421940928,
+      "grad_norm": 0.25195756554603577,
+      "learning_rate": 1.692787524366472e-05,
+      "loss": 0.0958,
+      "step": 495
+    },
+    {
+      "epoch": 0.9301453352086263,
+      "grad_norm": 0.24769015610218048,
+      "learning_rate": 1.692007797270955e-05,
+      "loss": 0.1028,
+      "step": 496
+    },
+    {
+      "epoch": 0.9320206282231599,
+      "grad_norm": 0.2188967764377594,
+      "learning_rate": 1.6912280701754385e-05,
+      "loss": 0.0731,
+      "step": 497
+    },
+    {
+      "epoch": 0.9338959212376934,
+      "grad_norm": 0.2325516939163208,
+      "learning_rate": 1.690448343079922e-05,
+      "loss": 0.1144,
+      "step": 498
+    },
+    {
+      "epoch": 0.9357712142522269,
+      "grad_norm": 0.30458924174308777,
+      "learning_rate": 1.6896686159844057e-05,
+      "loss": 0.0813,
+      "step": 499
+    },
+    {
+      "epoch": 0.9376465072667605,
+      "grad_norm": 0.22726097702980042,
+      "learning_rate": 1.688888888888889e-05,
+      "loss": 0.0839,
+      "step": 500
+    },
+    {
+      "epoch": 0.9376465072667605,
+      "eval_loss": 0.08535026758909225,
+      "eval_runtime": 676.335,
+      "eval_samples_per_second": 0.291,
+      "eval_steps_per_second": 0.291,
+      "step": 500
+    },
+    {
+      "epoch": 0.939521800281294,
+      "grad_norm": 0.2691619098186493,
+      "learning_rate": 1.6881091617933726e-05,
+      "loss": 0.1266,
+      "step": 501
+    },
+    {
+      "epoch": 0.9413970932958274,
+      "grad_norm": 0.28599363565444946,
+      "learning_rate": 1.687329434697856e-05,
+      "loss": 0.1125,
+      "step": 502
+    },
+    {
+      "epoch": 0.943272386310361,
+      "grad_norm": 0.27564743161201477,
+      "learning_rate": 1.6865497076023394e-05,
+      "loss": 0.1029,
+      "step": 503
+    },
+    {
+      "epoch": 0.9451476793248945,
+      "grad_norm": 0.23417291045188904,
+      "learning_rate": 1.6857699805068228e-05,
+      "loss": 0.087,
+      "step": 504
+    },
+    {
+      "epoch": 0.947022972339428,
+      "grad_norm": 0.2351316511631012,
+      "learning_rate": 1.6849902534113062e-05,
+      "loss": 0.1024,
+      "step": 505
+    },
+    {
+      "epoch": 0.9488982653539616,
+      "grad_norm": 0.2007894515991211,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 0.0752,
+      "step": 506
+    },
+    {
+      "epoch": 0.9507735583684951,
+      "grad_norm": 0.2415047585964203,
+      "learning_rate": 1.683430799220273e-05,
+      "loss": 0.0912,
+      "step": 507
+    },
+    {
+      "epoch": 0.9526488513830286,
+      "grad_norm": 0.25994813442230225,
+      "learning_rate": 1.6826510721247565e-05,
+      "loss": 0.1331,
+      "step": 508
+    },
+    {
+      "epoch": 0.9545241443975622,
+      "grad_norm": 0.2513566315174103,
+      "learning_rate": 1.68187134502924e-05,
+      "loss": 0.1031,
+      "step": 509
+    },
+    {
+      "epoch": 0.9563994374120957,
+      "grad_norm": 0.2094557136297226,
+      "learning_rate": 1.6810916179337233e-05,
+      "loss": 0.0736,
+      "step": 510
+    },
+    {
+      "epoch": 0.9582747304266291,
+      "grad_norm": 0.2421177178621292,
+      "learning_rate": 1.6803118908382067e-05,
+      "loss": 0.1123,
+      "step": 511
+    },
+    {
+      "epoch": 0.9601500234411627,
+      "grad_norm": 0.2969476580619812,
+      "learning_rate": 1.67953216374269e-05,
+      "loss": 0.1175,
+      "step": 512
+    },
+    {
+      "epoch": 0.9620253164556962,
+      "grad_norm": 0.3147030770778656,
+      "learning_rate": 1.6787524366471736e-05,
+      "loss": 0.1647,
+      "step": 513
+    },
+    {
+      "epoch": 0.9639006094702297,
+      "grad_norm": 0.2841101586818695,
+      "learning_rate": 1.677972709551657e-05,
+      "loss": 0.1238,
+      "step": 514
+    },
+    {
+      "epoch": 0.9657759024847632,
+      "grad_norm": 0.28984513878822327,
+      "learning_rate": 1.6771929824561408e-05,
+      "loss": 0.1257,
+      "step": 515
+    },
+    {
+      "epoch": 0.9676511954992968,
+      "grad_norm": 0.32147666811943054,
+      "learning_rate": 1.676413255360624e-05,
+      "loss": 0.1446,
+      "step": 516
+    },
+    {
+      "epoch": 0.9695264885138303,
+      "grad_norm": 0.17480434477329254,
+      "learning_rate": 1.6756335282651073e-05,
+      "loss": 0.079,
+      "step": 517
+    },
+    {
+      "epoch": 0.9714017815283638,
+      "grad_norm": 0.2966823875904083,
+      "learning_rate": 1.6748538011695907e-05,
+      "loss": 0.1092,
+      "step": 518
+    },
+    {
+      "epoch": 0.9732770745428974,
+      "grad_norm": 0.24680288136005402,
+      "learning_rate": 1.674074074074074e-05,
+      "loss": 0.0885,
+      "step": 519
+    },
+    {
+      "epoch": 0.9751523675574308,
+      "grad_norm": 0.23044931888580322,
+      "learning_rate": 1.6732943469785575e-05,
+      "loss": 0.1615,
+      "step": 520
+    },
+    {
+      "epoch": 0.9770276605719643,
+      "grad_norm": 0.2578699588775635,
+      "learning_rate": 1.672514619883041e-05,
+      "loss": 0.1323,
+      "step": 521
+    },
+    {
+      "epoch": 0.9789029535864979,
+      "grad_norm": 0.2969329059123993,
+      "learning_rate": 1.6717348927875247e-05,
+      "loss": 0.1037,
+      "step": 522
+    },
+    {
+      "epoch": 0.9807782466010314,
+      "grad_norm": 0.2269773781299591,
+      "learning_rate": 1.670955165692008e-05,
+      "loss": 0.1047,
+      "step": 523
+    },
+    {
+      "epoch": 0.9826535396155649,
+      "grad_norm": 0.26680633425712585,
+      "learning_rate": 1.6701754385964915e-05,
+      "loss": 0.1249,
+      "step": 524
+    },
+    {
+      "epoch": 0.9845288326300985,
+      "grad_norm": 0.24191637337207794,
+      "learning_rate": 1.669395711500975e-05,
+      "loss": 0.0922,
+      "step": 525
+    },
+    {
+      "epoch": 0.986404125644632,
+      "grad_norm": 0.2023610770702362,
+      "learning_rate": 1.668615984405458e-05,
+      "loss": 0.0733,
+      "step": 526
+    },
+    {
+      "epoch": 0.9882794186591655,
+      "grad_norm": 0.34420913457870483,
+      "learning_rate": 1.6678362573099414e-05,
+      "loss": 0.0859,
+      "step": 527
+    },
+    {
+      "epoch": 0.9901547116736991,
+      "grad_norm": 0.29441869258880615,
+      "learning_rate": 1.6670565302144252e-05,
+      "loss": 0.1199,
+      "step": 528
+    },
+    {
+      "epoch": 0.9920300046882325,
+      "grad_norm": 0.20998947322368622,
+      "learning_rate": 1.6662768031189086e-05,
+      "loss": 0.0767,
+      "step": 529
+    },
+    {
+      "epoch": 0.993905297702766,
+      "grad_norm": 0.26571086049079895,
+      "learning_rate": 1.665497076023392e-05,
+      "loss": 0.1022,
+      "step": 530
+    },
+    {
+      "epoch": 0.9957805907172996,
+      "grad_norm": 0.24049051105976105,
+      "learning_rate": 1.6647173489278755e-05,
+      "loss": 0.1158,
+      "step": 531
+    },
+    {
+      "epoch": 0.9976558837318331,
+      "grad_norm": 0.3279871940612793,
+      "learning_rate": 1.663937621832359e-05,
+      "loss": 0.0895,
+      "step": 532
+    },
+    {
+      "epoch": 0.9995311767463666,
+      "grad_norm": 0.19162967801094055,
+      "learning_rate": 1.6631578947368423e-05,
+      "loss": 0.0906,
+      "step": 533
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5646827816963196,
+      "learning_rate": 1.6623781676413257e-05,
+      "loss": 0.1056,
+      "step": 534
+    },
+    {
+      "epoch": 1.0018752930145336,
+      "grad_norm": 0.22089308500289917,
+      "learning_rate": 1.661598440545809e-05,
+      "loss": 0.081,
+      "step": 535
+    },
+    {
+      "epoch": 1.003750586029067,
+      "grad_norm": 0.20040877163410187,
+      "learning_rate": 1.6608187134502926e-05,
+      "loss": 0.0746,
+      "step": 536
+    },
+    {
+      "epoch": 1.0056258790436006,
+      "grad_norm": 0.23172321915626526,
+      "learning_rate": 1.660038986354776e-05,
+      "loss": 0.1107,
+      "step": 537
+    },
+    {
+      "epoch": 1.0075011720581342,
+      "grad_norm": 0.22388094663619995,
+      "learning_rate": 1.6592592592592594e-05,
+      "loss": 0.1188,
+      "step": 538
+    },
+    {
+      "epoch": 1.0093764650726675,
+      "grad_norm": 0.20831793546676636,
+      "learning_rate": 1.6584795321637428e-05,
+      "loss": 0.0724,
+      "step": 539
+    },
+    {
+      "epoch": 1.0112517580872011,
+      "grad_norm": 0.2914353311061859,
+      "learning_rate": 1.6576998050682262e-05,
+      "loss": 0.1019,
+      "step": 540
+    },
+    {
+      "epoch": 1.0131270511017347,
+      "grad_norm": 0.2762034833431244,
+      "learning_rate": 1.6569200779727097e-05,
+      "loss": 0.095,
+      "step": 541
+    },
+    {
+      "epoch": 1.015002344116268,
+      "grad_norm": 0.22479356825351715,
+      "learning_rate": 1.656140350877193e-05,
+      "loss": 0.0883,
+      "step": 542
+    },
+    {
+      "epoch": 1.0168776371308017,
+      "grad_norm": 0.18842089176177979,
+      "learning_rate": 1.6553606237816765e-05,
+      "loss": 0.0851,
+      "step": 543
+    },
+    {
+      "epoch": 1.0187529301453353,
+      "grad_norm": 0.17845085263252258,
+      "learning_rate": 1.65458089668616e-05,
+      "loss": 0.0649,
+      "step": 544
+    },
+    {
+      "epoch": 1.0206282231598687,
+      "grad_norm": 0.22942480444908142,
+      "learning_rate": 1.6538011695906437e-05,
+      "loss": 0.0968,
+      "step": 545
+    },
+    {
+      "epoch": 1.0225035161744023,
+      "grad_norm": 0.3065761625766754,
+      "learning_rate": 1.6530214424951268e-05,
+      "loss": 0.098,
+      "step": 546
+    },
+    {
+      "epoch": 1.0243788091889359,
+      "grad_norm": 0.41475629806518555,
+      "learning_rate": 1.6522417153996102e-05,
+      "loss": 0.0882,
+      "step": 547
+    },
+    {
+      "epoch": 1.0262541022034692,
+      "grad_norm": 0.23140354454517365,
+      "learning_rate": 1.6514619883040936e-05,
+      "loss": 0.1071,
+      "step": 548
+    },
+    {
+      "epoch": 1.0281293952180028,
+      "grad_norm": 0.26001206040382385,
+      "learning_rate": 1.650682261208577e-05,
+      "loss": 0.0984,
+      "step": 549
+    },
+    {
+      "epoch": 1.0300046882325364,
+      "grad_norm": 0.27233394980430603,
+      "learning_rate": 1.6499025341130604e-05,
+      "loss": 0.1294,
+      "step": 550
+    },
+    {
+      "epoch": 1.0300046882325364,
+      "eval_loss": 0.08262171596288681,
+      "eval_runtime": 675.5684,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 550
+    },
+    {
+      "epoch": 1.0318799812470698,
+      "grad_norm": 0.28793230652809143,
+      "learning_rate": 1.649122807017544e-05,
+      "loss": 0.1191,
+      "step": 551
+    },
+    {
+      "epoch": 1.0337552742616034,
+      "grad_norm": 0.23013058304786682,
+      "learning_rate": 1.6483430799220276e-05,
+      "loss": 0.0834,
+      "step": 552
+    },
+    {
+      "epoch": 1.035630567276137,
+      "grad_norm": 0.2813061773777008,
+      "learning_rate": 1.647563352826511e-05,
+      "loss": 0.1116,
+      "step": 553
+    },
+    {
+      "epoch": 1.0375058602906704,
+      "grad_norm": 0.27876320481300354,
+      "learning_rate": 1.6467836257309944e-05,
+      "loss": 0.1027,
+      "step": 554
+    },
+    {
+      "epoch": 1.039381153305204,
+      "grad_norm": 0.2607191801071167,
+      "learning_rate": 1.646003898635478e-05,
+      "loss": 0.1119,
+      "step": 555
+    },
+    {
+      "epoch": 1.0412564463197373,
+      "grad_norm": 0.31741422414779663,
+      "learning_rate": 1.645224171539961e-05,
+      "loss": 0.1435,
+      "step": 556
+    },
+    {
+      "epoch": 1.043131739334271,
+      "grad_norm": 0.23251239955425262,
+      "learning_rate": 1.6444444444444444e-05,
+      "loss": 0.0574,
+      "step": 557
+    },
+    {
+      "epoch": 1.0450070323488045,
+      "grad_norm": 0.2328520566225052,
+      "learning_rate": 1.643664717348928e-05,
+      "loss": 0.1157,
+      "step": 558
+    },
+    {
+      "epoch": 1.046882325363338,
+      "grad_norm": 0.42328402400016785,
+      "learning_rate": 1.6428849902534115e-05,
+      "loss": 0.0988,
+      "step": 559
+    },
+    {
+      "epoch": 1.0487576183778715,
+      "grad_norm": 0.570969820022583,
+      "learning_rate": 1.642105263157895e-05,
+      "loss": 0.0954,
+      "step": 560
+    },
+    {
+      "epoch": 1.0506329113924051,
+      "grad_norm": 0.19645580649375916,
+      "learning_rate": 1.6413255360623784e-05,
+      "loss": 0.0751,
+      "step": 561
+    },
+    {
+      "epoch": 1.0525082044069385,
+      "grad_norm": 0.25933393836021423,
+      "learning_rate": 1.6405458089668618e-05,
+      "loss": 0.1031,
+      "step": 562
+    },
+    {
+      "epoch": 1.054383497421472,
+      "grad_norm": 0.2190171778202057,
+      "learning_rate": 1.6397660818713452e-05,
+      "loss": 0.0866,
+      "step": 563
+    },
+    {
+      "epoch": 1.0562587904360057,
+      "grad_norm": 0.2852676808834076,
+      "learning_rate": 1.6389863547758286e-05,
+      "loss": 0.0901,
+      "step": 564
+    },
+    {
+      "epoch": 1.058134083450539,
+      "grad_norm": 0.23386311531066895,
+      "learning_rate": 1.638206627680312e-05,
+      "loss": 0.0914,
+      "step": 565
+    },
+    {
+      "epoch": 1.0600093764650727,
+      "grad_norm": 0.23497509956359863,
+      "learning_rate": 1.6374269005847955e-05,
+      "loss": 0.0885,
+      "step": 566
+    },
+    {
+      "epoch": 1.0618846694796062,
+      "grad_norm": 0.2501085698604584,
+      "learning_rate": 1.636647173489279e-05,
+      "loss": 0.0555,
+      "step": 567
+    },
+    {
+      "epoch": 1.0637599624941396,
+      "grad_norm": 0.2505752742290497,
+      "learning_rate": 1.6358674463937623e-05,
+      "loss": 0.0836,
+      "step": 568
+    },
+    {
+      "epoch": 1.0656352555086732,
+      "grad_norm": 0.24521493911743164,
+      "learning_rate": 1.6350877192982457e-05,
+      "loss": 0.0907,
+      "step": 569
+    },
+    {
+      "epoch": 1.0675105485232068,
+      "grad_norm": 0.3060048222541809,
+      "learning_rate": 1.634307992202729e-05,
+      "loss": 0.0901,
+      "step": 570
+    },
+    {
+      "epoch": 1.0693858415377402,
+      "grad_norm": 0.2545836865901947,
+      "learning_rate": 1.6335282651072126e-05,
+      "loss": 0.1228,
+      "step": 571
+    },
+    {
+      "epoch": 1.0712611345522738,
+      "grad_norm": 0.37637385725975037,
+      "learning_rate": 1.632748538011696e-05,
+      "loss": 0.1273,
+      "step": 572
+    },
+    {
+      "epoch": 1.0731364275668074,
+      "grad_norm": 0.3298353850841522,
+      "learning_rate": 1.6319688109161794e-05,
+      "loss": 0.1297,
+      "step": 573
+    },
+    {
+      "epoch": 1.0750117205813408,
+      "grad_norm": 0.18379832804203033,
+      "learning_rate": 1.6311890838206628e-05,
+      "loss": 0.0712,
+      "step": 574
+    },
+    {
+      "epoch": 1.0768870135958744,
+      "grad_norm": 0.2380228340625763,
+      "learning_rate": 1.6304093567251466e-05,
+      "loss": 0.0645,
+      "step": 575
+    },
+    {
+      "epoch": 1.078762306610408,
+      "grad_norm": 0.38280248641967773,
+      "learning_rate": 1.6296296296296297e-05,
+      "loss": 0.1449,
+      "step": 576
+    },
+    {
+      "epoch": 1.0806375996249413,
+      "grad_norm": 0.24589546024799347,
+      "learning_rate": 1.628849902534113e-05,
+      "loss": 0.0954,
+      "step": 577
+    },
+    {
+      "epoch": 1.082512892639475,
+      "grad_norm": 0.245429128408432,
+      "learning_rate": 1.6280701754385965e-05,
+      "loss": 0.0716,
+      "step": 578
+    },
+    {
+      "epoch": 1.0843881856540085,
+      "grad_norm": 0.685253381729126,
+      "learning_rate": 1.62729044834308e-05,
+      "loss": 0.0688,
+      "step": 579
+    },
+    {
+      "epoch": 1.086263478668542,
+      "grad_norm": 0.28492575883865356,
+      "learning_rate": 1.6265107212475633e-05,
+      "loss": 0.1057,
+      "step": 580
+    },
+    {
+      "epoch": 1.0881387716830755,
+      "grad_norm": 0.2217700481414795,
+      "learning_rate": 1.625730994152047e-05,
+      "loss": 0.085,
+      "step": 581
+    },
+    {
+      "epoch": 1.090014064697609,
+      "grad_norm": 0.30696970224380493,
+      "learning_rate": 1.6249512670565305e-05,
+      "loss": 0.0976,
+      "step": 582
+    },
+    {
+      "epoch": 1.0918893577121425,
+      "grad_norm": 0.2384309470653534,
+      "learning_rate": 1.624171539961014e-05,
+      "loss": 0.0782,
+      "step": 583
+    },
+    {
+      "epoch": 1.093764650726676,
+      "grad_norm": 0.2883552610874176,
+      "learning_rate": 1.6233918128654974e-05,
+      "loss": 0.1116,
+      "step": 584
+    },
+    {
+      "epoch": 1.0956399437412097,
+      "grad_norm": 0.30440714955329895,
+      "learning_rate": 1.6226120857699808e-05,
+      "loss": 0.1204,
+      "step": 585
+    },
+    {
+      "epoch": 1.097515236755743,
+      "grad_norm": 0.9227920174598694,
+      "learning_rate": 1.621832358674464e-05,
+      "loss": 0.1289,
+      "step": 586
+    },
+    {
+      "epoch": 1.0993905297702766,
+      "grad_norm": 0.2924637198448181,
+      "learning_rate": 1.6210526315789473e-05,
+      "loss": 0.1018,
+      "step": 587
+    },
+    {
+      "epoch": 1.1012658227848102,
+      "grad_norm": 0.28547459840774536,
+      "learning_rate": 1.620272904483431e-05,
+      "loss": 0.1105,
+      "step": 588
+    },
+    {
+      "epoch": 1.1031411157993436,
+      "grad_norm": 0.2556934952735901,
+      "learning_rate": 1.6194931773879144e-05,
+      "loss": 0.0934,
+      "step": 589
+    },
+    {
+      "epoch": 1.1050164088138772,
+      "grad_norm": 0.23647433519363403,
+      "learning_rate": 1.618713450292398e-05,
+      "loss": 0.0787,
+      "step": 590
+    },
+    {
+      "epoch": 1.1068917018284108,
+      "grad_norm": 0.3279690146446228,
+      "learning_rate": 1.6179337231968813e-05,
+      "loss": 0.1185,
+      "step": 591
+    },
+    {
+      "epoch": 1.1087669948429442,
+      "grad_norm": 0.22327874600887299,
+      "learning_rate": 1.6171539961013647e-05,
+      "loss": 0.0835,
+      "step": 592
+    },
+    {
+      "epoch": 1.1106422878574778,
+      "grad_norm": 0.4429026246070862,
+      "learning_rate": 1.616374269005848e-05,
+      "loss": 0.12,
+      "step": 593
+    },
+    {
+      "epoch": 1.1125175808720114,
+      "grad_norm": 0.27783477306365967,
+      "learning_rate": 1.6155945419103315e-05,
+      "loss": 0.1396,
+      "step": 594
+    },
+    {
+      "epoch": 1.1143928738865447,
+      "grad_norm": 0.35669296979904175,
+      "learning_rate": 1.614814814814815e-05,
+      "loss": 0.0993,
+      "step": 595
+    },
+    {
+      "epoch": 1.1162681669010783,
+      "grad_norm": 0.22483180463314056,
+      "learning_rate": 1.6140350877192984e-05,
+      "loss": 0.0727,
+      "step": 596
+    },
+    {
+      "epoch": 1.1181434599156117,
+      "grad_norm": 0.2645404636859894,
+      "learning_rate": 1.6132553606237818e-05,
+      "loss": 0.1045,
+      "step": 597
+    },
+    {
+      "epoch": 1.1200187529301453,
+      "grad_norm": 0.29490724205970764,
+      "learning_rate": 1.6124756335282652e-05,
+      "loss": 0.0936,
+      "step": 598
+    },
+    {
+      "epoch": 1.121894045944679,
+      "grad_norm": 0.24751833081245422,
+      "learning_rate": 1.6116959064327486e-05,
+      "loss": 0.0827,
+      "step": 599
+    },
+    {
+      "epoch": 1.1237693389592125,
+      "grad_norm": 0.27503255009651184,
+      "learning_rate": 1.610916179337232e-05,
+      "loss": 0.1093,
+      "step": 600
+    },
+    {
+      "epoch": 1.1237693389592125,
+      "eval_loss": 0.08217223733663559,
+      "eval_runtime": 675.2558,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 600
+    },
+    {
+      "epoch": 1.1256446319737459,
+      "grad_norm": 0.30064669251441956,
+      "learning_rate": 1.6101364522417155e-05,
+      "loss": 0.0869,
+      "step": 601
+    },
+    {
+      "epoch": 1.1275199249882795,
+      "grad_norm": 0.2266855388879776,
+      "learning_rate": 1.609356725146199e-05,
+      "loss": 0.0804,
+      "step": 602
+    },
+    {
+      "epoch": 1.1293952180028128,
+      "grad_norm": 0.2856343686580658,
+      "learning_rate": 1.6085769980506823e-05,
+      "loss": 0.0893,
+      "step": 603
+    },
+    {
+      "epoch": 1.1312705110173464,
+      "grad_norm": 0.2303743064403534,
+      "learning_rate": 1.607797270955166e-05,
+      "loss": 0.0759,
+      "step": 604
+    },
+    {
+      "epoch": 1.13314580403188,
+      "grad_norm": 0.3389296531677246,
+      "learning_rate": 1.6070175438596495e-05,
+      "loss": 0.1191,
+      "step": 605
+    },
+    {
+      "epoch": 1.1350210970464134,
+      "grad_norm": 0.27957460284233093,
+      "learning_rate": 1.6062378167641326e-05,
+      "loss": 0.1205,
+      "step": 606
+    },
+    {
+      "epoch": 1.136896390060947,
+      "grad_norm": 0.24907708168029785,
+      "learning_rate": 1.605458089668616e-05,
+      "loss": 0.0942,
+      "step": 607
+    },
+    {
+      "epoch": 1.1387716830754806,
+      "grad_norm": 0.272223562002182,
+      "learning_rate": 1.6046783625730994e-05,
+      "loss": 0.1217,
+      "step": 608
+    },
+    {
+      "epoch": 1.140646976090014,
+      "grad_norm": 0.290814608335495,
+      "learning_rate": 1.603898635477583e-05,
+      "loss": 0.1062,
+      "step": 609
+    },
+    {
+      "epoch": 1.1425222691045476,
+      "grad_norm": 0.2468331754207611,
+      "learning_rate": 1.6031189083820663e-05,
+      "loss": 0.0969,
+      "step": 610
+    },
+    {
+      "epoch": 1.1443975621190812,
+      "grad_norm": 0.23120653629302979,
+      "learning_rate": 1.60233918128655e-05,
+      "loss": 0.092,
+      "step": 611
+    },
+    {
+      "epoch": 1.1462728551336145,
+      "grad_norm": 0.23789869248867035,
+      "learning_rate": 1.6015594541910334e-05,
+      "loss": 0.0992,
+      "step": 612
+    },
+    {
+      "epoch": 1.1481481481481481,
+      "grad_norm": 0.3250608444213867,
+      "learning_rate": 1.600779727095517e-05,
+      "loss": 0.12,
+      "step": 613
+    },
+    {
+      "epoch": 1.1500234411626817,
+      "grad_norm": 0.278442919254303,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.1042,
+      "step": 614
+    },
+    {
+      "epoch": 1.1518987341772151,
+      "grad_norm": 0.23886540532112122,
+      "learning_rate": 1.5992202729044833e-05,
+      "loss": 0.1048,
+      "step": 615
+    },
+    {
+      "epoch": 1.1537740271917487,
+      "grad_norm": 0.3647238612174988,
+      "learning_rate": 1.5984405458089668e-05,
+      "loss": 0.0651,
+      "step": 616
+    },
+    {
+      "epoch": 1.1556493202062823,
+      "grad_norm": 0.2553196847438812,
+      "learning_rate": 1.5976608187134505e-05,
+      "loss": 0.1011,
+      "step": 617
+    },
+    {
+      "epoch": 1.1575246132208157,
+      "grad_norm": 0.26290223002433777,
+      "learning_rate": 1.596881091617934e-05,
+      "loss": 0.1024,
+      "step": 618
+    },
+    {
+      "epoch": 1.1593999062353493,
+      "grad_norm": 0.3013964891433716,
+      "learning_rate": 1.5961013645224174e-05,
+      "loss": 0.1192,
+      "step": 619
+    },
+    {
+      "epoch": 1.1612751992498829,
+      "grad_norm": 0.30323684215545654,
+      "learning_rate": 1.5953216374269008e-05,
+      "loss": 0.1065,
+      "step": 620
+    },
+    {
+      "epoch": 1.1631504922644162,
+      "grad_norm": 0.28223493695259094,
+      "learning_rate": 1.5945419103313842e-05,
+      "loss": 0.1187,
+      "step": 621
+    },
+    {
+      "epoch": 1.1650257852789498,
+      "grad_norm": 0.3298370838165283,
+      "learning_rate": 1.5937621832358676e-05,
+      "loss": 0.1322,
+      "step": 622
+    },
+    {
+      "epoch": 1.1669010782934834,
+      "grad_norm": 0.2321697622537613,
+      "learning_rate": 1.592982456140351e-05,
+      "loss": 0.0828,
+      "step": 623
+    },
+    {
+      "epoch": 1.1687763713080168,
+      "grad_norm": 0.27766016125679016,
+      "learning_rate": 1.5922027290448345e-05,
+      "loss": 0.1183,
+      "step": 624
+    },
+    {
+      "epoch": 1.1706516643225504,
+      "grad_norm": 0.5540488362312317,
+      "learning_rate": 1.591423001949318e-05,
+      "loss": 0.1321,
+      "step": 625
+    },
+    {
+      "epoch": 1.172526957337084,
+      "grad_norm": 0.24459047615528107,
+      "learning_rate": 1.5906432748538013e-05,
+      "loss": 0.1013,
+      "step": 626
+    },
+    {
+      "epoch": 1.1744022503516174,
+      "grad_norm": 0.307108610868454,
+      "learning_rate": 1.5898635477582847e-05,
+      "loss": 0.1047,
+      "step": 627
+    },
+    {
+      "epoch": 1.176277543366151,
+      "grad_norm": 0.36074694991111755,
+      "learning_rate": 1.589083820662768e-05,
+      "loss": 0.0792,
+      "step": 628
+    },
+    {
+      "epoch": 1.1781528363806846,
+      "grad_norm": 0.28516885638237,
+      "learning_rate": 1.5883040935672516e-05,
+      "loss": 0.1324,
+      "step": 629
+    },
+    {
+      "epoch": 1.180028129395218,
+      "grad_norm": 0.3291719853878021,
+      "learning_rate": 1.587524366471735e-05,
+      "loss": 0.0955,
+      "step": 630
+    },
+    {
+      "epoch": 1.1819034224097515,
+      "grad_norm": 0.25097760558128357,
+      "learning_rate": 1.5867446393762184e-05,
+      "loss": 0.0849,
+      "step": 631
+    },
+    {
+      "epoch": 1.183778715424285,
+      "grad_norm": 0.2444642335176468,
+      "learning_rate": 1.5859649122807018e-05,
+      "loss": 0.0694,
+      "step": 632
+    },
+    {
+      "epoch": 1.1856540084388185,
+      "grad_norm": 0.27323952317237854,
+      "learning_rate": 1.5851851851851852e-05,
+      "loss": 0.0963,
+      "step": 633
+    },
+    {
+      "epoch": 1.1875293014533521,
+      "grad_norm": 0.23469096422195435,
+      "learning_rate": 1.584405458089669e-05,
+      "loss": 0.0764,
+      "step": 634
+    },
+    {
+      "epoch": 1.1894045944678857,
+      "grad_norm": 0.2885955572128296,
+      "learning_rate": 1.583625730994152e-05,
+      "loss": 0.0978,
+      "step": 635
+    },
+    {
+      "epoch": 1.191279887482419,
+      "grad_norm": 0.29515987634658813,
+      "learning_rate": 1.5828460038986355e-05,
+      "loss": 0.1237,
+      "step": 636
+    },
+    {
+      "epoch": 1.1931551804969527,
+      "grad_norm": 0.3769955039024353,
+      "learning_rate": 1.582066276803119e-05,
+      "loss": 0.0984,
+      "step": 637
+    },
+    {
+      "epoch": 1.195030473511486,
+      "grad_norm": 0.3882874846458435,
+      "learning_rate": 1.5812865497076023e-05,
+      "loss": 0.1113,
+      "step": 638
+    },
+    {
+      "epoch": 1.1969057665260197,
+      "grad_norm": 0.3134611248970032,
+      "learning_rate": 1.5805068226120857e-05,
+      "loss": 0.0973,
+      "step": 639
+    },
+    {
+      "epoch": 1.1987810595405533,
+      "grad_norm": 0.23223522305488586,
+      "learning_rate": 1.579727095516569e-05,
+      "loss": 0.054,
+      "step": 640
+    },
+    {
+      "epoch": 1.2006563525550868,
+      "grad_norm": 0.22435115277767181,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 0.0883,
+      "step": 641
+    },
+    {
+      "epoch": 1.2025316455696202,
+      "grad_norm": 0.2723855674266815,
+      "learning_rate": 1.5781676413255363e-05,
+      "loss": 0.1132,
+      "step": 642
+    },
+    {
+      "epoch": 1.2044069385841538,
+      "grad_norm": 0.32858818769454956,
+      "learning_rate": 1.5773879142300198e-05,
+      "loss": 0.0787,
+      "step": 643
+    },
+    {
+      "epoch": 1.2062822315986872,
+      "grad_norm": 0.25390857458114624,
+      "learning_rate": 1.5766081871345032e-05,
+      "loss": 0.0659,
+      "step": 644
+    },
+    {
+      "epoch": 1.2081575246132208,
+      "grad_norm": 0.24344322085380554,
+      "learning_rate": 1.5758284600389863e-05,
+      "loss": 0.0912,
+      "step": 645
+    },
+    {
+      "epoch": 1.2100328176277544,
+      "grad_norm": 0.2637319564819336,
+      "learning_rate": 1.5750487329434697e-05,
+      "loss": 0.0915,
+      "step": 646
+    },
+    {
+      "epoch": 1.2119081106422878,
+      "grad_norm": 0.31068336963653564,
+      "learning_rate": 1.5742690058479534e-05,
+      "loss": 0.1511,
+      "step": 647
+    },
+    {
+      "epoch": 1.2137834036568214,
+      "grad_norm": 0.27969691157341003,
+      "learning_rate": 1.573489278752437e-05,
+      "loss": 0.0742,
+      "step": 648
+    },
+    {
+      "epoch": 1.215658696671355,
+      "grad_norm": 0.31574878096580505,
+      "learning_rate": 1.5727095516569203e-05,
+      "loss": 0.0755,
+      "step": 649
+    },
+    {
+      "epoch": 1.2175339896858883,
+      "grad_norm": 0.24576275050640106,
+      "learning_rate": 1.5719298245614037e-05,
+      "loss": 0.0803,
+      "step": 650
+    },
+    {
+      "epoch": 1.2175339896858883,
+      "eval_loss": 0.07974553108215332,
+      "eval_runtime": 676.0594,
+      "eval_samples_per_second": 0.291,
+      "eval_steps_per_second": 0.291,
+      "step": 650
+    },
+    {
+      "epoch": 1.219409282700422,
+      "grad_norm": 0.31214672327041626,
+      "learning_rate": 1.571150097465887e-05,
+      "loss": 0.1131,
+      "step": 651
+    },
+    {
+      "epoch": 1.2212845757149555,
+      "grad_norm": 0.268762469291687,
+      "learning_rate": 1.5703703703703705e-05,
+      "loss": 0.0896,
+      "step": 652
+    },
+    {
+      "epoch": 1.223159868729489,
+      "grad_norm": 0.21026861667633057,
+      "learning_rate": 1.569590643274854e-05,
+      "loss": 0.056,
+      "step": 653
+    },
+    {
+      "epoch": 1.2250351617440225,
+      "grad_norm": 0.4084342420101166,
+      "learning_rate": 1.5688109161793374e-05,
+      "loss": 0.0833,
+      "step": 654
+    },
+    {
+      "epoch": 1.226910454758556,
+      "grad_norm": 0.28843924403190613,
+      "learning_rate": 1.5680311890838208e-05,
+      "loss": 0.0888,
+      "step": 655
+    },
+    {
+      "epoch": 1.2287857477730895,
+      "grad_norm": 0.29009735584259033,
+      "learning_rate": 1.5672514619883042e-05,
+      "loss": 0.0992,
+      "step": 656
+    },
+    {
+      "epoch": 1.230661040787623,
+      "grad_norm": 0.24224917590618134,
+      "learning_rate": 1.5664717348927876e-05,
+      "loss": 0.0624,
+      "step": 657
+    },
+    {
+      "epoch": 1.2325363338021567,
+      "grad_norm": 0.29413849115371704,
+      "learning_rate": 1.565692007797271e-05,
+      "loss": 0.095,
+      "step": 658
+    },
+    {
+      "epoch": 1.23441162681669,
+      "grad_norm": 0.21053080260753632,
+      "learning_rate": 1.5649122807017545e-05,
+      "loss": 0.0631,
+      "step": 659
+    },
+    {
+      "epoch": 1.2362869198312236,
+      "grad_norm": 0.2499096691608429,
+      "learning_rate": 1.564132553606238e-05,
+      "loss": 0.0848,
+      "step": 660
+    },
+    {
+      "epoch": 1.2381622128457572,
+      "grad_norm": 0.28722044825553894,
+      "learning_rate": 1.5633528265107213e-05,
+      "loss": 0.1162,
+      "step": 661
+    },
+    {
+      "epoch": 1.2400375058602906,
+      "grad_norm": 0.3939790427684784,
+      "learning_rate": 1.5625730994152047e-05,
+      "loss": 0.0842,
+      "step": 662
+    },
+    {
+      "epoch": 1.2419127988748242,
+      "grad_norm": 0.2812120318412781,
+      "learning_rate": 1.561793372319688e-05,
+      "loss": 0.06,
+      "step": 663
+    },
+    {
+      "epoch": 1.2437880918893578,
+      "grad_norm": 0.2532125413417816,
+      "learning_rate": 1.561013645224172e-05,
+      "loss": 0.0866,
+      "step": 664
+    },
+    {
+      "epoch": 1.2456633849038912,
+      "grad_norm": 0.3428124785423279,
+      "learning_rate": 1.560233918128655e-05,
+      "loss": 0.1119,
+      "step": 665
+    },
+    {
+      "epoch": 1.2475386779184248,
+      "grad_norm": 0.3042241930961609,
+      "learning_rate": 1.5594541910331384e-05,
+      "loss": 0.1112,
+      "step": 666
+    },
+    {
+      "epoch": 1.2494139709329584,
+      "grad_norm": 0.2961670458316803,
+      "learning_rate": 1.5586744639376218e-05,
+      "loss": 0.0964,
+      "step": 667
+    },
+    {
+      "epoch": 1.2512892639474917,
+      "grad_norm": 0.403415322303772,
+      "learning_rate": 1.5578947368421052e-05,
+      "loss": 0.1721,
+      "step": 668
+    },
+    {
+      "epoch": 1.2531645569620253,
+      "grad_norm": 0.27317744493484497,
+      "learning_rate": 1.5571150097465887e-05,
+      "loss": 0.096,
+      "step": 669
+    },
+    {
+      "epoch": 1.255039849976559,
+      "grad_norm": 0.2543340027332306,
+      "learning_rate": 1.5563352826510724e-05,
+      "loss": 0.0775,
+      "step": 670
+    },
+    {
+      "epoch": 1.2569151429910923,
+      "grad_norm": 0.4965609014034271,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 0.0798,
+      "step": 671
+    },
+    {
+      "epoch": 1.258790436005626,
+      "grad_norm": 0.2571773827075958,
+      "learning_rate": 1.5547758284600393e-05,
+      "loss": 0.0865,
+      "step": 672
+    },
+    {
+      "epoch": 1.2606657290201593,
+      "grad_norm": 0.2605682909488678,
+      "learning_rate": 1.5539961013645227e-05,
+      "loss": 0.1214,
+      "step": 673
+    },
+    {
+      "epoch": 1.2625410220346929,
+      "grad_norm": 0.2747882902622223,
+      "learning_rate": 1.553216374269006e-05,
+      "loss": 0.115,
+      "step": 674
+    },
+    {
+      "epoch": 1.2644163150492265,
+      "grad_norm": 0.2293609231710434,
+      "learning_rate": 1.5524366471734892e-05,
+      "loss": 0.0743,
+      "step": 675
+    },
+    {
+      "epoch": 1.26629160806376,
+      "grad_norm": 0.3130941092967987,
+      "learning_rate": 1.5516569200779726e-05,
+      "loss": 0.0837,
+      "step": 676
+    },
+    {
+      "epoch": 1.2681669010782934,
+      "grad_norm": 0.3052816092967987,
+      "learning_rate": 1.5508771929824563e-05,
+      "loss": 0.1217,
+      "step": 677
+    },
+    {
+      "epoch": 1.270042194092827,
+      "grad_norm": 0.24826787412166595,
+      "learning_rate": 1.5500974658869398e-05,
+      "loss": 0.0881,
+      "step": 678
+    },
+    {
+      "epoch": 1.2719174871073604,
+      "grad_norm": 0.30709272623062134,
+      "learning_rate": 1.5493177387914232e-05,
+      "loss": 0.0667,
+      "step": 679
+    },
+    {
+      "epoch": 1.273792780121894,
+      "grad_norm": 0.2654632329940796,
+      "learning_rate": 1.5485380116959066e-05,
+      "loss": 0.0915,
+      "step": 680
+    },
+    {
+      "epoch": 1.2756680731364276,
+      "grad_norm": 0.24949900805950165,
+      "learning_rate": 1.54775828460039e-05,
+      "loss": 0.0738,
+      "step": 681
+    },
+    {
+      "epoch": 1.2775433661509612,
+      "grad_norm": 0.5046447515487671,
+      "learning_rate": 1.5469785575048734e-05,
+      "loss": 0.1017,
+      "step": 682
+    },
+    {
+      "epoch": 1.2794186591654946,
+      "grad_norm": 0.31709209084510803,
+      "learning_rate": 1.546198830409357e-05,
+      "loss": 0.1129,
+      "step": 683
+    },
+    {
+      "epoch": 1.2812939521800282,
+      "grad_norm": 0.30328288674354553,
+      "learning_rate": 1.5454191033138403e-05,
+      "loss": 0.1407,
+      "step": 684
+    },
+    {
+      "epoch": 1.2831692451945615,
+      "grad_norm": 0.34941598773002625,
+      "learning_rate": 1.5446393762183237e-05,
+      "loss": 0.1595,
+      "step": 685
+    },
+    {
+      "epoch": 1.2850445382090951,
+      "grad_norm": 0.25712135434150696,
+      "learning_rate": 1.543859649122807e-05,
+      "loss": 0.082,
+      "step": 686
+    },
+    {
+      "epoch": 1.2869198312236287,
+      "grad_norm": 0.5184486508369446,
+      "learning_rate": 1.5430799220272905e-05,
+      "loss": 0.0826,
+      "step": 687
+    },
+    {
+      "epoch": 1.2887951242381623,
+      "grad_norm": 0.23768484592437744,
+      "learning_rate": 1.542300194931774e-05,
+      "loss": 0.0893,
+      "step": 688
+    },
+    {
+      "epoch": 1.2906704172526957,
+      "grad_norm": 0.2733663320541382,
+      "learning_rate": 1.5415204678362574e-05,
+      "loss": 0.1009,
+      "step": 689
+    },
+    {
+      "epoch": 1.2925457102672293,
+      "grad_norm": 0.2970256805419922,
+      "learning_rate": 1.5407407407407408e-05,
+      "loss": 0.1103,
+      "step": 690
+    },
+    {
+      "epoch": 1.2944210032817627,
+      "grad_norm": 0.29735323786735535,
+      "learning_rate": 1.5399610136452242e-05,
+      "loss": 0.1185,
+      "step": 691
+    },
+    {
+      "epoch": 1.2962962962962963,
+      "grad_norm": 0.33100953698158264,
+      "learning_rate": 1.5391812865497076e-05,
+      "loss": 0.0833,
+      "step": 692
+    },
+    {
+      "epoch": 1.2981715893108299,
+      "grad_norm": 0.30054280161857605,
+      "learning_rate": 1.5384015594541914e-05,
+      "loss": 0.058,
+      "step": 693
+    },
+    {
+      "epoch": 1.3000468823253635,
+      "grad_norm": 0.2017904669046402,
+      "learning_rate": 1.5376218323586748e-05,
+      "loss": 0.0716,
+      "step": 694
+    },
+    {
+      "epoch": 1.3019221753398968,
+      "grad_norm": 0.2905302047729492,
+      "learning_rate": 1.536842105263158e-05,
+      "loss": 0.1068,
+      "step": 695
+    },
+    {
+      "epoch": 1.3037974683544304,
+      "grad_norm": 0.26433658599853516,
+      "learning_rate": 1.5360623781676413e-05,
+      "loss": 0.0923,
+      "step": 696
+    },
+    {
+      "epoch": 1.3056727613689638,
+      "grad_norm": 0.24136176705360413,
+      "learning_rate": 1.5352826510721247e-05,
+      "loss": 0.0975,
+      "step": 697
+    },
+    {
+      "epoch": 1.3075480543834974,
+      "grad_norm": 0.2888275384902954,
+      "learning_rate": 1.534502923976608e-05,
+      "loss": 0.0736,
+      "step": 698
+    },
+    {
+      "epoch": 1.309423347398031,
+      "grad_norm": 0.28614625334739685,
+      "learning_rate": 1.5337231968810916e-05,
+      "loss": 0.0757,
+      "step": 699
+    },
+    {
+      "epoch": 1.3112986404125644,
+      "grad_norm": 1.1438792943954468,
+      "learning_rate": 1.5329434697855753e-05,
+      "loss": 0.1105,
+      "step": 700
+    },
+    {
+      "epoch": 1.3112986404125644,
+      "eval_loss": 0.08180084824562073,
+      "eval_runtime": 675.7061,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 700
+    },
+    {
+      "epoch": 1.313173933427098,
+      "grad_norm": 0.24729672074317932,
+      "learning_rate": 1.5321637426900587e-05,
+      "loss": 0.1244,
+      "step": 701
+    },
+    {
+      "epoch": 1.3150492264416316,
+      "grad_norm": 0.6661401391029358,
+      "learning_rate": 1.531384015594542e-05,
+      "loss": 0.0465,
+      "step": 702
+    },
+    {
+      "epoch": 1.316924519456165,
+      "grad_norm": 0.2756960391998291,
+      "learning_rate": 1.5306042884990256e-05,
+      "loss": 0.0845,
+      "step": 703
+    },
+    {
+      "epoch": 1.3187998124706986,
+      "grad_norm": 0.2175983488559723,
+      "learning_rate": 1.529824561403509e-05,
+      "loss": 0.0874,
+      "step": 704
+    },
+    {
+      "epoch": 1.3206751054852321,
+      "grad_norm": 0.26682281494140625,
+      "learning_rate": 1.529044834307992e-05,
+      "loss": 0.0785,
+      "step": 705
+    },
+    {
+      "epoch": 1.3225503984997655,
+      "grad_norm": 0.28904491662979126,
+      "learning_rate": 1.528265107212476e-05,
+      "loss": 0.0812,
+      "step": 706
+    },
+    {
+      "epoch": 1.3244256915142991,
+      "grad_norm": 0.37207451462745667,
+      "learning_rate": 1.5274853801169593e-05,
+      "loss": 0.118,
+      "step": 707
+    },
+    {
+      "epoch": 1.3263009845288325,
+      "grad_norm": 0.2251625806093216,
+      "learning_rate": 1.5267056530214427e-05,
+      "loss": 0.088,
+      "step": 708
+    },
+    {
+      "epoch": 1.328176277543366,
+      "grad_norm": 0.2839013934135437,
+      "learning_rate": 1.525925925925926e-05,
+      "loss": 0.089,
+      "step": 709
+    },
+    {
+      "epoch": 1.3300515705578997,
+      "grad_norm": 0.2906053960323334,
+      "learning_rate": 1.5251461988304095e-05,
+      "loss": 0.1056,
+      "step": 710
+    },
+    {
+      "epoch": 1.3319268635724333,
+      "grad_norm": 0.29006117582321167,
+      "learning_rate": 1.524366471734893e-05,
+      "loss": 0.1284,
+      "step": 711
+    },
+    {
+      "epoch": 1.3338021565869667,
+      "grad_norm": 0.23621053993701935,
+      "learning_rate": 1.5235867446393764e-05,
+      "loss": 0.0831,
+      "step": 712
+    },
+    {
+      "epoch": 1.3356774496015003,
+      "grad_norm": 0.2675006091594696,
+      "learning_rate": 1.5228070175438598e-05,
+      "loss": 0.0979,
+      "step": 713
+    },
+    {
+      "epoch": 1.3375527426160336,
+      "grad_norm": 0.25027772784233093,
+      "learning_rate": 1.5220272904483434e-05,
+      "loss": 0.0989,
+      "step": 714
+    },
+    {
+      "epoch": 1.3394280356305672,
+      "grad_norm": 0.3723505437374115,
+      "learning_rate": 1.5212475633528266e-05,
+      "loss": 0.1134,
+      "step": 715
+    },
+    {
+      "epoch": 1.3413033286451008,
+      "grad_norm": 0.2928392291069031,
+      "learning_rate": 1.52046783625731e-05,
+      "loss": 0.1118,
+      "step": 716
+    },
+    {
+      "epoch": 1.3431786216596344,
+      "grad_norm": 0.23428718745708466,
+      "learning_rate": 1.5196881091617935e-05,
+      "loss": 0.056,
+      "step": 717
+    },
+    {
+      "epoch": 1.3450539146741678,
+      "grad_norm": 0.23713499307632446,
+      "learning_rate": 1.5189083820662769e-05,
+      "loss": 0.0812,
+      "step": 718
+    },
+    {
+      "epoch": 1.3469292076887014,
+      "grad_norm": 0.27535611391067505,
+      "learning_rate": 1.5181286549707603e-05,
+      "loss": 0.0926,
+      "step": 719
+    },
+    {
+      "epoch": 1.3488045007032348,
+      "grad_norm": 0.29593226313591003,
+      "learning_rate": 1.5173489278752439e-05,
+      "loss": 0.1266,
+      "step": 720
+    },
+    {
+      "epoch": 1.3506797937177684,
+      "grad_norm": 0.32769280672073364,
+      "learning_rate": 1.5165692007797273e-05,
+      "loss": 0.1462,
+      "step": 721
+    },
+    {
+      "epoch": 1.352555086732302,
+      "grad_norm": 0.3472793996334076,
+      "learning_rate": 1.5157894736842107e-05,
+      "loss": 0.1547,
+      "step": 722
+    },
+    {
+      "epoch": 1.3544303797468356,
+      "grad_norm": 0.2513481378555298,
+      "learning_rate": 1.5150097465886941e-05,
+      "loss": 0.093,
+      "step": 723
+    },
+    {
+      "epoch": 1.356305672761369,
+      "grad_norm": 0.2549232244491577,
+      "learning_rate": 1.5142300194931776e-05,
+      "loss": 0.0858,
+      "step": 724
+    },
+    {
+      "epoch": 1.3581809657759025,
+      "grad_norm": 0.33557018637657166,
+      "learning_rate": 1.5134502923976608e-05,
+      "loss": 0.1245,
+      "step": 725
+    },
+    {
+      "epoch": 1.360056258790436,
+      "grad_norm": 0.2504093647003174,
+      "learning_rate": 1.5126705653021442e-05,
+      "loss": 0.0764,
+      "step": 726
+    },
+    {
+      "epoch": 1.3619315518049695,
+      "grad_norm": 0.20211853086948395,
+      "learning_rate": 1.5118908382066278e-05,
+      "loss": 0.0634,
+      "step": 727
+    },
+    {
+      "epoch": 1.363806844819503,
+      "grad_norm": 0.28420448303222656,
+      "learning_rate": 1.5111111111111112e-05,
+      "loss": 0.1096,
+      "step": 728
+    },
+    {
+      "epoch": 1.3656821378340367,
+      "grad_norm": 0.22307127714157104,
+      "learning_rate": 1.5103313840155947e-05,
+      "loss": 0.0603,
+      "step": 729
+    },
+    {
+      "epoch": 1.36755743084857,
+      "grad_norm": 0.21484006941318512,
+      "learning_rate": 1.509551656920078e-05,
+      "loss": 0.0784,
+      "step": 730
+    },
+    {
+      "epoch": 1.3694327238631037,
+      "grad_norm": 0.23451459407806396,
+      "learning_rate": 1.5087719298245615e-05,
+      "loss": 0.1042,
+      "step": 731
+    },
+    {
+      "epoch": 1.371308016877637,
+      "grad_norm": 0.3075193166732788,
+      "learning_rate": 1.507992202729045e-05,
+      "loss": 0.0924,
+      "step": 732
+    },
+    {
+      "epoch": 1.3731833098921706,
+      "grad_norm": 0.25598907470703125,
+      "learning_rate": 1.5072124756335285e-05,
+      "loss": 0.0911,
+      "step": 733
+    },
+    {
+      "epoch": 1.3750586029067042,
+      "grad_norm": 0.25740352272987366,
+      "learning_rate": 1.5064327485380119e-05,
+      "loss": 0.1088,
+      "step": 734
+    },
+    {
+      "epoch": 1.3769338959212378,
+      "grad_norm": 0.3566450774669647,
+      "learning_rate": 1.5056530214424952e-05,
+      "loss": 0.1148,
+      "step": 735
+    },
+    {
+      "epoch": 1.3788091889357712,
+      "grad_norm": 0.21709580719470978,
+      "learning_rate": 1.5048732943469786e-05,
+      "loss": 0.0762,
+      "step": 736
+    },
+    {
+      "epoch": 1.3806844819503048,
+      "grad_norm": 0.36706310510635376,
+      "learning_rate": 1.504093567251462e-05,
+      "loss": 0.0655,
+      "step": 737
+    },
+    {
+      "epoch": 1.3825597749648382,
+      "grad_norm": 0.2232268899679184,
+      "learning_rate": 1.5033138401559454e-05,
+      "loss": 0.0777,
+      "step": 738
+    },
+    {
+      "epoch": 1.3844350679793718,
+      "grad_norm": 0.4871269762516022,
+      "learning_rate": 1.502534113060429e-05,
+      "loss": 0.1297,
+      "step": 739
+    },
+    {
+      "epoch": 1.3863103609939054,
+      "grad_norm": 0.32851436734199524,
+      "learning_rate": 1.5017543859649124e-05,
+      "loss": 0.1113,
+      "step": 740
+    },
+    {
+      "epoch": 1.3881856540084387,
+      "grad_norm": 0.28122517466545105,
+      "learning_rate": 1.5009746588693958e-05,
+      "loss": 0.1021,
+      "step": 741
+    },
+    {
+      "epoch": 1.3900609470229723,
+      "grad_norm": 0.2887057065963745,
+      "learning_rate": 1.5001949317738793e-05,
+      "loss": 0.0871,
+      "step": 742
+    },
+    {
+      "epoch": 1.391936240037506,
+      "grad_norm": 0.2595096230506897,
+      "learning_rate": 1.4994152046783627e-05,
+      "loss": 0.0708,
+      "step": 743
+    },
+    {
+      "epoch": 1.3938115330520393,
+      "grad_norm": 0.232121542096138,
+      "learning_rate": 1.4986354775828463e-05,
+      "loss": 0.0727,
+      "step": 744
+    },
+    {
+      "epoch": 1.395686826066573,
+      "grad_norm": 0.2849893569946289,
+      "learning_rate": 1.4978557504873295e-05,
+      "loss": 0.0861,
+      "step": 745
+    },
+    {
+      "epoch": 1.3975621190811065,
+      "grad_norm": 0.2730804979801178,
+      "learning_rate": 1.497076023391813e-05,
+      "loss": 0.0973,
+      "step": 746
+    },
+    {
+      "epoch": 1.3994374120956399,
+      "grad_norm": 0.24728837609291077,
+      "learning_rate": 1.4962962962962964e-05,
+      "loss": 0.0867,
+      "step": 747
+    },
+    {
+      "epoch": 1.4013127051101735,
+      "grad_norm": 0.2925151288509369,
+      "learning_rate": 1.4955165692007798e-05,
+      "loss": 0.135,
+      "step": 748
+    },
+    {
+      "epoch": 1.4031879981247068,
+      "grad_norm": 0.35172486305236816,
+      "learning_rate": 1.4947368421052632e-05,
+      "loss": 0.1115,
+      "step": 749
+    },
+    {
+      "epoch": 1.4050632911392404,
+      "grad_norm": 0.24151886999607086,
+      "learning_rate": 1.4939571150097468e-05,
+      "loss": 0.0892,
+      "step": 750
+    },
+    {
+      "epoch": 1.4050632911392404,
+      "eval_loss": 0.07957883179187775,
+      "eval_runtime": 676.5902,
+      "eval_samples_per_second": 0.291,
+      "eval_steps_per_second": 0.291,
+      "step": 750
+    },
+    {
+      "epoch": 1.406938584153774,
+      "grad_norm": 0.2599674463272095,
+      "learning_rate": 1.4931773879142302e-05,
+      "loss": 0.073,
+      "step": 751
+    },
+    {
+      "epoch": 1.4088138771683076,
+      "grad_norm": 0.2866610586643219,
+      "learning_rate": 1.4923976608187136e-05,
+      "loss": 0.0743,
+      "step": 752
+    },
+    {
+      "epoch": 1.410689170182841,
+      "grad_norm": 0.3928741216659546,
+      "learning_rate": 1.491617933723197e-05,
+      "loss": 0.1462,
+      "step": 753
+    },
+    {
+      "epoch": 1.4125644631973746,
+      "grad_norm": 0.38174375891685486,
+      "learning_rate": 1.4908382066276805e-05,
+      "loss": 0.1081,
+      "step": 754
+    },
+    {
+      "epoch": 1.414439756211908,
+      "grad_norm": 0.2848498821258545,
+      "learning_rate": 1.4900584795321637e-05,
+      "loss": 0.1071,
+      "step": 755
+    },
+    {
+      "epoch": 1.4163150492264416,
+      "grad_norm": 0.26754605770111084,
+      "learning_rate": 1.4892787524366471e-05,
+      "loss": 0.0775,
+      "step": 756
+    },
+    {
+      "epoch": 1.4181903422409752,
+      "grad_norm": 0.3248796761035919,
+      "learning_rate": 1.4884990253411307e-05,
+      "loss": 0.1093,
+      "step": 757
+    },
+    {
+      "epoch": 1.4200656352555088,
+      "grad_norm": 0.2621667981147766,
+      "learning_rate": 1.4877192982456141e-05,
+      "loss": 0.0846,
+      "step": 758
+    },
+    {
+      "epoch": 1.4219409282700421,
+      "grad_norm": 0.3511008024215698,
+      "learning_rate": 1.4869395711500976e-05,
+      "loss": 0.1343,
+      "step": 759
+    },
+    {
+      "epoch": 1.4238162212845757,
+      "grad_norm": 0.23400309681892395,
+      "learning_rate": 1.486159844054581e-05,
+      "loss": 0.0795,
+      "step": 760
+    },
+    {
+      "epoch": 1.4256915142991091,
+      "grad_norm": 0.3075447678565979,
+      "learning_rate": 1.4853801169590644e-05,
+      "loss": 0.1026,
+      "step": 761
+    },
+    {
+      "epoch": 1.4275668073136427,
+      "grad_norm": 0.2749321162700653,
+      "learning_rate": 1.484600389863548e-05,
+      "loss": 0.0991,
+      "step": 762
+    },
+    {
+      "epoch": 1.4294421003281763,
+      "grad_norm": 0.29317227005958557,
+      "learning_rate": 1.4838206627680314e-05,
+      "loss": 0.0857,
+      "step": 763
+    },
+    {
+      "epoch": 1.43131739334271,
+      "grad_norm": 0.3031213879585266,
+      "learning_rate": 1.4830409356725148e-05,
+      "loss": 0.1196,
+      "step": 764
+    },
+    {
+      "epoch": 1.4331926863572433,
+      "grad_norm": 0.2685413658618927,
+      "learning_rate": 1.482261208576998e-05,
+      "loss": 0.0958,
+      "step": 765
+    },
+    {
+      "epoch": 1.4350679793717769,
+      "grad_norm": 0.28609031438827515,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 0.1317,
+      "step": 766
+    },
+    {
+      "epoch": 1.4369432723863103,
+      "grad_norm": 0.2767196595668793,
+      "learning_rate": 1.4807017543859649e-05,
+      "loss": 0.0985,
+      "step": 767
+    },
+    {
+      "epoch": 1.4388185654008439,
+      "grad_norm": 0.29232558608055115,
+      "learning_rate": 1.4799220272904485e-05,
+      "loss": 0.1096,
+      "step": 768
+    },
+    {
+      "epoch": 1.4406938584153774,
+      "grad_norm": 0.2960546016693115,
+      "learning_rate": 1.479142300194932e-05,
+      "loss": 0.0876,
+      "step": 769
+    },
+    {
+      "epoch": 1.442569151429911,
+      "grad_norm": 0.43392062187194824,
+      "learning_rate": 1.4783625730994153e-05,
+      "loss": 0.1089,
+      "step": 770
+    },
+    {
+      "epoch": 1.4444444444444444,
+      "grad_norm": 0.23281604051589966,
+      "learning_rate": 1.4775828460038988e-05,
+      "loss": 0.0941,
+      "step": 771
+    },
+    {
+      "epoch": 1.446319737458978,
+      "grad_norm": 0.293990820646286,
+      "learning_rate": 1.4768031189083822e-05,
+      "loss": 0.1161,
+      "step": 772
+    },
+    {
+      "epoch": 1.4481950304735114,
+      "grad_norm": 0.33511075377464294,
+      "learning_rate": 1.4760233918128658e-05,
+      "loss": 0.0917,
+      "step": 773
+    },
+    {
+      "epoch": 1.450070323488045,
+      "grad_norm": 0.32499679923057556,
+      "learning_rate": 1.4752436647173492e-05,
+      "loss": 0.0784,
+      "step": 774
+    },
+    {
+      "epoch": 1.4519456165025786,
+      "grad_norm": 0.27234625816345215,
+      "learning_rate": 1.4744639376218324e-05,
+      "loss": 0.0876,
+      "step": 775
+    },
+    {
+      "epoch": 1.4538209095171122,
+      "grad_norm": 0.38885700702667236,
+      "learning_rate": 1.4736842105263159e-05,
+      "loss": 0.1354,
+      "step": 776
+    },
+    {
+      "epoch": 1.4556962025316456,
+      "grad_norm": 0.28798046708106995,
+      "learning_rate": 1.4729044834307993e-05,
+      "loss": 0.0782,
+      "step": 777
+    },
+    {
+      "epoch": 1.4575714955461792,
+      "grad_norm": 0.2976330518722534,
+      "learning_rate": 1.4721247563352827e-05,
+      "loss": 0.0898,
+      "step": 778
+    },
+    {
+      "epoch": 1.4594467885607125,
+      "grad_norm": 0.31952551007270813,
+      "learning_rate": 1.4713450292397661e-05,
+      "loss": 0.128,
+      "step": 779
+    },
+    {
+      "epoch": 1.4613220815752461,
+      "grad_norm": 0.27772435545921326,
+      "learning_rate": 1.4705653021442497e-05,
+      "loss": 0.0893,
+      "step": 780
+    },
+    {
+      "epoch": 1.4631973745897797,
+      "grad_norm": 0.24229328334331512,
+      "learning_rate": 1.4697855750487331e-05,
+      "loss": 0.0684,
+      "step": 781
+    },
+    {
+      "epoch": 1.465072667604313,
+      "grad_norm": 0.2713194489479065,
+      "learning_rate": 1.4690058479532165e-05,
+      "loss": 0.0987,
+      "step": 782
+    },
+    {
+      "epoch": 1.4669479606188467,
+      "grad_norm": 0.37047621607780457,
+      "learning_rate": 1.4682261208577e-05,
+      "loss": 0.0901,
+      "step": 783
+    },
+    {
+      "epoch": 1.4688232536333803,
+      "grad_norm": 0.5432588458061218,
+      "learning_rate": 1.4674463937621834e-05,
+      "loss": 0.1067,
+      "step": 784
+    },
+    {
+      "epoch": 1.4706985466479137,
+      "grad_norm": 0.26890650391578674,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.0755,
+      "step": 785
+    },
+    {
+      "epoch": 1.4725738396624473,
+      "grad_norm": 0.30856049060821533,
+      "learning_rate": 1.4658869395711502e-05,
+      "loss": 0.0889,
+      "step": 786
+    },
+    {
+      "epoch": 1.4744491326769809,
+      "grad_norm": 0.2999826669692993,
+      "learning_rate": 1.4651072124756336e-05,
+      "loss": 0.134,
+      "step": 787
+    },
+    {
+      "epoch": 1.4763244256915142,
+      "grad_norm": 0.23444703221321106,
+      "learning_rate": 1.464327485380117e-05,
+      "loss": 0.0899,
+      "step": 788
+    },
+    {
+      "epoch": 1.4781997187060478,
+      "grad_norm": 0.25326988101005554,
+      "learning_rate": 1.4635477582846005e-05,
+      "loss": 0.0889,
+      "step": 789
+    },
+    {
+      "epoch": 1.4800750117205814,
+      "grad_norm": 0.4101651906967163,
+      "learning_rate": 1.4627680311890839e-05,
+      "loss": 0.0708,
+      "step": 790
+    },
+    {
+      "epoch": 1.4819503047351148,
+      "grad_norm": 0.3386366069316864,
+      "learning_rate": 1.4619883040935675e-05,
+      "loss": 0.079,
+      "step": 791
+    },
+    {
+      "epoch": 1.4838255977496484,
+      "grad_norm": 0.2348432093858719,
+      "learning_rate": 1.4612085769980509e-05,
+      "loss": 0.071,
+      "step": 792
+    },
+    {
+      "epoch": 1.485700890764182,
+      "grad_norm": 0.2450290322303772,
+      "learning_rate": 1.4604288499025343e-05,
+      "loss": 0.0811,
+      "step": 793
+    },
+    {
+      "epoch": 1.4875761837787154,
+      "grad_norm": 0.30053895711898804,
+      "learning_rate": 1.4596491228070177e-05,
+      "loss": 0.092,
+      "step": 794
+    },
+    {
+      "epoch": 1.489451476793249,
+      "grad_norm": 0.3324088752269745,
+      "learning_rate": 1.458869395711501e-05,
+      "loss": 0.1003,
+      "step": 795
+    },
+    {
+      "epoch": 1.4913267698077823,
+      "grad_norm": 0.31995347142219543,
+      "learning_rate": 1.4580896686159844e-05,
+      "loss": 0.1084,
+      "step": 796
+    },
+    {
+      "epoch": 1.493202062822316,
+      "grad_norm": 0.2602950930595398,
+      "learning_rate": 1.4573099415204678e-05,
+      "loss": 0.0751,
+      "step": 797
+    },
+    {
+      "epoch": 1.4950773558368495,
+      "grad_norm": 0.2805696725845337,
+      "learning_rate": 1.4565302144249514e-05,
+      "loss": 0.1025,
+      "step": 798
+    },
+    {
+      "epoch": 1.4969526488513831,
+      "grad_norm": 0.29233241081237793,
+      "learning_rate": 1.4557504873294348e-05,
+      "loss": 0.1161,
+      "step": 799
+    },
+    {
+      "epoch": 1.4988279418659165,
+      "grad_norm": 0.31460389494895935,
+      "learning_rate": 1.4549707602339183e-05,
+      "loss": 0.0975,
+      "step": 800
+    },
+    {
+      "epoch": 1.4988279418659165,
+      "eval_loss": 0.07664016634225845,
+      "eval_runtime": 675.2627,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 800
+    },
+    {
+      "epoch": 1.50070323488045,
+      "grad_norm": 0.3660965859889984,
+      "learning_rate": 1.4541910331384017e-05,
+      "loss": 0.1098,
+      "step": 801
+    },
+    {
+      "epoch": 1.5025785278949835,
+      "grad_norm": 0.26959264278411865,
+      "learning_rate": 1.4534113060428851e-05,
+      "loss": 0.0855,
+      "step": 802
+    },
+    {
+      "epoch": 1.504453820909517,
+      "grad_norm": 0.3515307605266571,
+      "learning_rate": 1.4526315789473687e-05,
+      "loss": 0.1272,
+      "step": 803
+    },
+    {
+      "epoch": 1.5063291139240507,
+      "grad_norm": 0.2386811077594757,
+      "learning_rate": 1.4518518518518521e-05,
+      "loss": 0.0612,
+      "step": 804
+    },
+    {
+      "epoch": 1.5082044069385843,
+      "grad_norm": 0.34232282638549805,
+      "learning_rate": 1.4510721247563353e-05,
+      "loss": 0.061,
+      "step": 805
+    },
+    {
+      "epoch": 1.5100796999531176,
+      "grad_norm": 0.6131661534309387,
+      "learning_rate": 1.4502923976608188e-05,
+      "loss": 0.0817,
+      "step": 806
+    },
+    {
+      "epoch": 1.5119549929676512,
+      "grad_norm": 0.20144400000572205,
+      "learning_rate": 1.4495126705653022e-05,
+      "loss": 0.0456,
+      "step": 807
+    },
+    {
+      "epoch": 1.5138302859821846,
+      "grad_norm": 0.2816990315914154,
+      "learning_rate": 1.4487329434697856e-05,
+      "loss": 0.0858,
+      "step": 808
+    },
+    {
+      "epoch": 1.5157055789967182,
+      "grad_norm": 0.3411908447742462,
+      "learning_rate": 1.447953216374269e-05,
+      "loss": 0.0866,
+      "step": 809
+    },
+    {
+      "epoch": 1.5175808720112518,
+      "grad_norm": 0.2759881615638733,
+      "learning_rate": 1.4471734892787526e-05,
+      "loss": 0.0947,
+      "step": 810
+    },
+    {
+      "epoch": 1.5194561650257854,
+      "grad_norm": 0.26385390758514404,
+      "learning_rate": 1.446393762183236e-05,
+      "loss": 0.0853,
+      "step": 811
+    },
+    {
+      "epoch": 1.5213314580403188,
+      "grad_norm": 0.20965765416622162,
+      "learning_rate": 1.4456140350877195e-05,
+      "loss": 0.0716,
+      "step": 812
+    },
+    {
+      "epoch": 1.5232067510548524,
+      "grad_norm": 0.33767369389533997,
+      "learning_rate": 1.4448343079922029e-05,
+      "loss": 0.1221,
+      "step": 813
+    },
+    {
+      "epoch": 1.5250820440693857,
+      "grad_norm": 0.9217659831047058,
+      "learning_rate": 1.4440545808966863e-05,
+      "loss": 0.0984,
+      "step": 814
+    },
+    {
+      "epoch": 1.5269573370839193,
+      "grad_norm": 0.2534090280532837,
+      "learning_rate": 1.4432748538011695e-05,
+      "loss": 0.0744,
+      "step": 815
+    },
+    {
+      "epoch": 1.528832630098453,
+      "grad_norm": 0.27146193385124207,
+      "learning_rate": 1.4424951267056531e-05,
+      "loss": 0.0876,
+      "step": 816
+    },
+    {
+      "epoch": 1.5307079231129865,
+      "grad_norm": 0.23886770009994507,
+      "learning_rate": 1.4417153996101365e-05,
+      "loss": 0.0823,
+      "step": 817
+    },
+    {
+      "epoch": 1.53258321612752,
+      "grad_norm": 0.29205018281936646,
+      "learning_rate": 1.44093567251462e-05,
+      "loss": 0.1128,
+      "step": 818
+    },
+    {
+      "epoch": 1.5344585091420533,
+      "grad_norm": 0.2402828186750412,
+      "learning_rate": 1.4401559454191034e-05,
+      "loss": 0.0853,
+      "step": 819
+    },
+    {
+      "epoch": 1.5363338021565869,
+      "grad_norm": 0.28632691502571106,
+      "learning_rate": 1.4393762183235868e-05,
+      "loss": 0.0977,
+      "step": 820
+    },
+    {
+      "epoch": 1.5382090951711205,
+      "grad_norm": 0.296055406332016,
+      "learning_rate": 1.4385964912280704e-05,
+      "loss": 0.1045,
+      "step": 821
+    },
+    {
+      "epoch": 1.540084388185654,
+      "grad_norm": 0.2865302264690399,
+      "learning_rate": 1.4378167641325538e-05,
+      "loss": 0.0779,
+      "step": 822
+    },
+    {
+      "epoch": 1.5419596812001877,
+      "grad_norm": 0.25754040479660034,
+      "learning_rate": 1.4370370370370372e-05,
+      "loss": 0.0892,
+      "step": 823
+    },
+    {
+      "epoch": 1.543834974214721,
+      "grad_norm": 0.2635495960712433,
+      "learning_rate": 1.4362573099415207e-05,
+      "loss": 0.0716,
+      "step": 824
+    },
+    {
+      "epoch": 1.5457102672292544,
+      "grad_norm": 0.42626357078552246,
+      "learning_rate": 1.4354775828460039e-05,
+      "loss": 0.1188,
+      "step": 825
+    },
+    {
+      "epoch": 1.547585560243788,
+      "grad_norm": 0.2752715051174164,
+      "learning_rate": 1.4346978557504873e-05,
+      "loss": 0.0927,
+      "step": 826
+    },
+    {
+      "epoch": 1.5494608532583216,
+      "grad_norm": 0.20132949948310852,
+      "learning_rate": 1.4339181286549707e-05,
+      "loss": 0.0541,
+      "step": 827
+    },
+    {
+      "epoch": 1.5513361462728552,
+      "grad_norm": 0.3923582434654236,
+      "learning_rate": 1.4331384015594543e-05,
+      "loss": 0.0871,
+      "step": 828
+    },
+    {
+      "epoch": 1.5532114392873888,
+      "grad_norm": 0.2865050137042999,
+      "learning_rate": 1.4323586744639377e-05,
+      "loss": 0.0802,
+      "step": 829
+    },
+    {
+      "epoch": 1.5550867323019222,
+      "grad_norm": 0.2447250485420227,
+      "learning_rate": 1.4315789473684212e-05,
+      "loss": 0.0699,
+      "step": 830
+    },
+    {
+      "epoch": 1.5569620253164556,
+      "grad_norm": 0.2569245398044586,
+      "learning_rate": 1.4307992202729046e-05,
+      "loss": 0.0736,
+      "step": 831
+    },
+    {
+      "epoch": 1.5588373183309892,
+      "grad_norm": 0.27975255250930786,
+      "learning_rate": 1.430019493177388e-05,
+      "loss": 0.0841,
+      "step": 832
+    },
+    {
+      "epoch": 1.5607126113455227,
+      "grad_norm": 0.246305450797081,
+      "learning_rate": 1.4292397660818716e-05,
+      "loss": 0.0722,
+      "step": 833
+    },
+    {
+      "epoch": 1.5625879043600563,
+      "grad_norm": 0.3482731580734253,
+      "learning_rate": 1.428460038986355e-05,
+      "loss": 0.1454,
+      "step": 834
+    },
+    {
+      "epoch": 1.56446319737459,
+      "grad_norm": 0.8472810983657837,
+      "learning_rate": 1.4276803118908383e-05,
+      "loss": 0.1386,
+      "step": 835
+    },
+    {
+      "epoch": 1.5663384903891233,
+      "grad_norm": 0.2856524884700775,
+      "learning_rate": 1.4269005847953217e-05,
+      "loss": 0.0976,
+      "step": 836
+    },
+    {
+      "epoch": 1.5682137834036567,
+      "grad_norm": 0.22626835107803345,
+      "learning_rate": 1.4261208576998051e-05,
+      "loss": 0.07,
+      "step": 837
+    },
+    {
+      "epoch": 1.5700890764181903,
+      "grad_norm": 0.28921768069267273,
+      "learning_rate": 1.4253411306042885e-05,
+      "loss": 0.0653,
+      "step": 838
+    },
+    {
+      "epoch": 1.5719643694327239,
+      "grad_norm": 0.3061649799346924,
+      "learning_rate": 1.4245614035087721e-05,
+      "loss": 0.1024,
+      "step": 839
+    },
+    {
+      "epoch": 1.5738396624472575,
+      "grad_norm": 0.23194481432437897,
+      "learning_rate": 1.4237816764132555e-05,
+      "loss": 0.0542,
+      "step": 840
+    },
+    {
+      "epoch": 1.5757149554617909,
+      "grad_norm": 0.21323058009147644,
+      "learning_rate": 1.423001949317739e-05,
+      "loss": 0.0413,
+      "step": 841
+    },
+    {
+      "epoch": 1.5775902484763245,
+      "grad_norm": 0.6626904010772705,
+      "learning_rate": 1.4222222222222224e-05,
+      "loss": 0.061,
+      "step": 842
+    },
+    {
+      "epoch": 1.5794655414908578,
+      "grad_norm": 0.279397577047348,
+      "learning_rate": 1.4214424951267058e-05,
+      "loss": 0.069,
+      "step": 843
+    },
+    {
+      "epoch": 1.5813408345053914,
+      "grad_norm": 0.35950684547424316,
+      "learning_rate": 1.4206627680311894e-05,
+      "loss": 0.0628,
+      "step": 844
+    },
+    {
+      "epoch": 1.583216127519925,
+      "grad_norm": 0.35051649808883667,
+      "learning_rate": 1.4198830409356725e-05,
+      "loss": 0.1164,
+      "step": 845
+    },
+    {
+      "epoch": 1.5850914205344586,
+      "grad_norm": 0.23908878862857819,
+      "learning_rate": 1.419103313840156e-05,
+      "loss": 0.0745,
+      "step": 846
+    },
+    {
+      "epoch": 1.586966713548992,
+      "grad_norm": 0.293130487203598,
+      "learning_rate": 1.4183235867446395e-05,
+      "loss": 0.0875,
+      "step": 847
+    },
+    {
+      "epoch": 1.5888420065635256,
+      "grad_norm": 0.2919827699661255,
+      "learning_rate": 1.4175438596491229e-05,
+      "loss": 0.0886,
+      "step": 848
+    },
+    {
+      "epoch": 1.590717299578059,
+      "grad_norm": 0.3345141112804413,
+      "learning_rate": 1.4167641325536063e-05,
+      "loss": 0.1071,
+      "step": 849
+    },
+    {
+      "epoch": 1.5925925925925926,
+      "grad_norm": 0.6707221269607544,
+      "learning_rate": 1.4159844054580897e-05,
+      "loss": 0.1002,
+      "step": 850
+    },
+    {
+      "epoch": 1.5925925925925926,
+      "eval_loss": 0.07790966331958771,
+      "eval_runtime": 676.4736,
+      "eval_samples_per_second": 0.291,
+      "eval_steps_per_second": 0.291,
+      "step": 850
+    },
+    {
+      "epoch": 1.5944678856071262,
+      "grad_norm": 0.27356359362602234,
+      "learning_rate": 1.4152046783625733e-05,
+      "loss": 0.0801,
+      "step": 851
+    },
+    {
+      "epoch": 1.5963431786216598,
+      "grad_norm": 0.3435034155845642,
+      "learning_rate": 1.4144249512670567e-05,
+      "loss": 0.1108,
+      "step": 852
+    },
+    {
+      "epoch": 1.5982184716361931,
+      "grad_norm": 0.32657772302627563,
+      "learning_rate": 1.4136452241715401e-05,
+      "loss": 0.113,
+      "step": 853
+    },
+    {
+      "epoch": 1.6000937646507267,
+      "grad_norm": 0.3669753074645996,
+      "learning_rate": 1.4128654970760236e-05,
+      "loss": 0.1343,
+      "step": 854
+    },
+    {
+      "epoch": 1.60196905766526,
+      "grad_norm": 0.23169514536857605,
+      "learning_rate": 1.4120857699805068e-05,
+      "loss": 0.0894,
+      "step": 855
+    },
+    {
+      "epoch": 1.6038443506797937,
+      "grad_norm": 0.2689126431941986,
+      "learning_rate": 1.4113060428849902e-05,
+      "loss": 0.0753,
+      "step": 856
+    },
+    {
+      "epoch": 1.6057196436943273,
+      "grad_norm": 0.2116389125585556,
+      "learning_rate": 1.4105263157894738e-05,
+      "loss": 0.0592,
+      "step": 857
+    },
+    {
+      "epoch": 1.6075949367088609,
+      "grad_norm": 0.2922620475292206,
+      "learning_rate": 1.4097465886939572e-05,
+      "loss": 0.0944,
+      "step": 858
+    },
+    {
+      "epoch": 1.6094702297233943,
+      "grad_norm": 0.26720038056373596,
+      "learning_rate": 1.4089668615984407e-05,
+      "loss": 0.0805,
+      "step": 859
+    },
+    {
+      "epoch": 1.6113455227379276,
+      "grad_norm": 0.36932405829429626,
+      "learning_rate": 1.408187134502924e-05,
+      "loss": 0.0871,
+      "step": 860
+    },
+    {
+      "epoch": 1.6132208157524612,
+      "grad_norm": 0.29181620478630066,
+      "learning_rate": 1.4074074074074075e-05,
+      "loss": 0.0903,
+      "step": 861
+    },
+    {
+      "epoch": 1.6150961087669948,
+      "grad_norm": 0.24898619949817657,
+      "learning_rate": 1.4066276803118911e-05,
+      "loss": 0.0612,
+      "step": 862
+    },
+    {
+      "epoch": 1.6169714017815284,
+      "grad_norm": 0.3132479786872864,
+      "learning_rate": 1.4058479532163745e-05,
+      "loss": 0.0962,
+      "step": 863
+    },
+    {
+      "epoch": 1.618846694796062,
+      "grad_norm": 0.27776429057121277,
+      "learning_rate": 1.405068226120858e-05,
+      "loss": 0.0734,
+      "step": 864
+    },
+    {
+      "epoch": 1.6207219878105954,
+      "grad_norm": 0.3017600476741791,
+      "learning_rate": 1.4042884990253412e-05,
+      "loss": 0.0802,
+      "step": 865
+    },
+    {
+      "epoch": 1.6225972808251288,
+      "grad_norm": 0.2887416183948517,
+      "learning_rate": 1.4035087719298246e-05,
+      "loss": 0.098,
+      "step": 866
+    },
+    {
+      "epoch": 1.6244725738396624,
+      "grad_norm": 0.3379286527633667,
+      "learning_rate": 1.402729044834308e-05,
+      "loss": 0.107,
+      "step": 867
+    },
+    {
+      "epoch": 1.626347866854196,
+      "grad_norm": 0.33945974707603455,
+      "learning_rate": 1.4019493177387914e-05,
+      "loss": 0.0949,
+      "step": 868
+    },
+    {
+      "epoch": 1.6282231598687296,
+      "grad_norm": 0.5351037979125977,
+      "learning_rate": 1.401169590643275e-05,
+      "loss": 0.0974,
+      "step": 869
+    },
+    {
+      "epoch": 1.6300984528832632,
+      "grad_norm": 0.28477218747138977,
+      "learning_rate": 1.4003898635477584e-05,
+      "loss": 0.1049,
+      "step": 870
+    },
+    {
+      "epoch": 1.6319737458977965,
+      "grad_norm": 0.34814324975013733,
+      "learning_rate": 1.3996101364522419e-05,
+      "loss": 0.1052,
+      "step": 871
+    },
+    {
+      "epoch": 1.63384903891233,
+      "grad_norm": 0.3389275074005127,
+      "learning_rate": 1.3988304093567253e-05,
+      "loss": 0.1081,
+      "step": 872
+    },
+    {
+      "epoch": 1.6357243319268635,
+      "grad_norm": 0.3551498353481293,
+      "learning_rate": 1.3980506822612087e-05,
+      "loss": 0.1016,
+      "step": 873
+    },
+    {
+      "epoch": 1.637599624941397,
+      "grad_norm": 0.43945154547691345,
+      "learning_rate": 1.3972709551656923e-05,
+      "loss": 0.0631,
+      "step": 874
+    },
+    {
+      "epoch": 1.6394749179559307,
+      "grad_norm": 0.7714282870292664,
+      "learning_rate": 1.3964912280701755e-05,
+      "loss": 0.0969,
+      "step": 875
+    },
+    {
+      "epoch": 1.6413502109704643,
+      "grad_norm": 0.2278236597776413,
+      "learning_rate": 1.395711500974659e-05,
+      "loss": 0.0711,
+      "step": 876
+    },
+    {
+      "epoch": 1.6432255039849977,
+      "grad_norm": 0.3024490773677826,
+      "learning_rate": 1.3949317738791424e-05,
+      "loss": 0.0727,
+      "step": 877
+    },
+    {
+      "epoch": 1.645100796999531,
+      "grad_norm": 0.31009918451309204,
+      "learning_rate": 1.3941520467836258e-05,
+      "loss": 0.1222,
+      "step": 878
+    },
+    {
+      "epoch": 1.6469760900140646,
+      "grad_norm": 0.402402400970459,
+      "learning_rate": 1.3933723196881092e-05,
+      "loss": 0.1404,
+      "step": 879
+    },
+    {
+      "epoch": 1.6488513830285982,
+      "grad_norm": 0.3172832727432251,
+      "learning_rate": 1.3925925925925928e-05,
+      "loss": 0.1052,
+      "step": 880
+    },
+    {
+      "epoch": 1.6507266760431318,
+      "grad_norm": 0.6237524151802063,
+      "learning_rate": 1.3918128654970762e-05,
+      "loss": 0.0804,
+      "step": 881
+    },
+    {
+      "epoch": 1.6526019690576652,
+      "grad_norm": 0.27731725573539734,
+      "learning_rate": 1.3910331384015596e-05,
+      "loss": 0.0988,
+      "step": 882
+    },
+    {
+      "epoch": 1.6544772620721988,
+      "grad_norm": 0.2532290518283844,
+      "learning_rate": 1.390253411306043e-05,
+      "loss": 0.1025,
+      "step": 883
+    },
+    {
+      "epoch": 1.6563525550867322,
+      "grad_norm": 0.3791520297527313,
+      "learning_rate": 1.3894736842105265e-05,
+      "loss": 0.1178,
+      "step": 884
+    },
+    {
+      "epoch": 1.6582278481012658,
+      "grad_norm": 0.24422794580459595,
+      "learning_rate": 1.3886939571150097e-05,
+      "loss": 0.0764,
+      "step": 885
+    },
+    {
+      "epoch": 1.6601031411157994,
+      "grad_norm": 0.3019620478153229,
+      "learning_rate": 1.3879142300194931e-05,
+      "loss": 0.103,
+      "step": 886
+    },
+    {
+      "epoch": 1.661978434130333,
+      "grad_norm": 0.33067408204078674,
+      "learning_rate": 1.3871345029239767e-05,
+      "loss": 0.114,
+      "step": 887
+    },
+    {
+      "epoch": 1.6638537271448663,
+      "grad_norm": 0.30137330293655396,
+      "learning_rate": 1.3863547758284602e-05,
+      "loss": 0.0538,
+      "step": 888
+    },
+    {
+      "epoch": 1.6657290201594,
+      "grad_norm": 0.3999065160751343,
+      "learning_rate": 1.3855750487329436e-05,
+      "loss": 0.0878,
+      "step": 889
+    },
+    {
+      "epoch": 1.6676043131739333,
+      "grad_norm": 0.23727497458457947,
+      "learning_rate": 1.384795321637427e-05,
+      "loss": 0.0762,
+      "step": 890
+    },
+    {
+      "epoch": 1.669479606188467,
+      "grad_norm": 0.3073793053627014,
+      "learning_rate": 1.3840155945419104e-05,
+      "loss": 0.1009,
+      "step": 891
+    },
+    {
+      "epoch": 1.6713548992030005,
+      "grad_norm": 0.3737054467201233,
+      "learning_rate": 1.383235867446394e-05,
+      "loss": 0.118,
+      "step": 892
+    },
+    {
+      "epoch": 1.673230192217534,
+      "grad_norm": 0.2168496549129486,
+      "learning_rate": 1.3824561403508774e-05,
+      "loss": 0.0733,
+      "step": 893
+    },
+    {
+      "epoch": 1.6751054852320675,
+      "grad_norm": 0.3298211395740509,
+      "learning_rate": 1.3816764132553608e-05,
+      "loss": 0.1167,
+      "step": 894
+    },
+    {
+      "epoch": 1.676980778246601,
+      "grad_norm": 0.2308388352394104,
+      "learning_rate": 1.3808966861598441e-05,
+      "loss": 0.0744,
+      "step": 895
+    },
+    {
+      "epoch": 1.6788560712611345,
+      "grad_norm": 0.27769342064857483,
+      "learning_rate": 1.3801169590643275e-05,
+      "loss": 0.0651,
+      "step": 896
+    },
+    {
+      "epoch": 1.680731364275668,
+      "grad_norm": 0.2548845708370209,
+      "learning_rate": 1.379337231968811e-05,
+      "loss": 0.0993,
+      "step": 897
+    },
+    {
+      "epoch": 1.6826066572902016,
+      "grad_norm": 0.28797003626823425,
+      "learning_rate": 1.3785575048732943e-05,
+      "loss": 0.0773,
+      "step": 898
+    },
+    {
+      "epoch": 1.6844819503047352,
+      "grad_norm": 0.32892584800720215,
+      "learning_rate": 1.377777777777778e-05,
+      "loss": 0.0944,
+      "step": 899
+    },
+    {
+      "epoch": 1.6863572433192686,
+      "grad_norm": 0.40558120608329773,
+      "learning_rate": 1.3769980506822614e-05,
+      "loss": 0.1006,
+      "step": 900
+    },
+    {
+      "epoch": 1.6863572433192686,
+      "eval_loss": 0.07565333694219589,
+      "eval_runtime": 675.1107,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 900
+    },
+    {
+      "epoch": 1.688232536333802,
+      "grad_norm": 0.3913367986679077,
+      "learning_rate": 1.3762183235867448e-05,
+      "loss": 0.0936,
+      "step": 901
+    },
+    {
+      "epoch": 1.6901078293483356,
+      "grad_norm": 0.2699122726917267,
+      "learning_rate": 1.3754385964912282e-05,
+      "loss": 0.0829,
+      "step": 902
+    },
+    {
+      "epoch": 1.6919831223628692,
+      "grad_norm": 0.35245972871780396,
+      "learning_rate": 1.3746588693957116e-05,
+      "loss": 0.0932,
+      "step": 903
+    },
+    {
+      "epoch": 1.6938584153774028,
+      "grad_norm": 0.25840988755226135,
+      "learning_rate": 1.3738791423001952e-05,
+      "loss": 0.086,
+      "step": 904
+    },
+    {
+      "epoch": 1.6957337083919364,
+      "grad_norm": 0.44727715849876404,
+      "learning_rate": 1.3730994152046784e-05,
+      "loss": 0.0939,
+      "step": 905
+    },
+    {
+      "epoch": 1.6976090014064698,
+      "grad_norm": 0.2222200334072113,
+      "learning_rate": 1.3723196881091619e-05,
+      "loss": 0.0859,
+      "step": 906
+    },
+    {
+      "epoch": 1.6994842944210031,
+      "grad_norm": 0.27306899428367615,
+      "learning_rate": 1.3715399610136453e-05,
+      "loss": 0.0877,
+      "step": 907
+    },
+    {
+      "epoch": 1.7013595874355367,
+      "grad_norm": 0.41138342022895813,
+      "learning_rate": 1.3707602339181287e-05,
+      "loss": 0.0812,
+      "step": 908
+    },
+    {
+      "epoch": 1.7032348804500703,
+      "grad_norm": 0.29449108242988586,
+      "learning_rate": 1.3699805068226121e-05,
+      "loss": 0.0887,
+      "step": 909
+    },
+    {
+      "epoch": 1.705110173464604,
+      "grad_norm": 0.25317761301994324,
+      "learning_rate": 1.3692007797270957e-05,
+      "loss": 0.0921,
+      "step": 910
+    },
+    {
+      "epoch": 1.7069854664791375,
+      "grad_norm": 0.24713647365570068,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 0.0646,
+      "step": 911
+    },
+    {
+      "epoch": 1.7088607594936709,
+      "grad_norm": 0.2317000776529312,
+      "learning_rate": 1.3676413255360625e-05,
+      "loss": 0.0596,
+      "step": 912
+    },
+    {
+      "epoch": 1.7107360525082043,
+      "grad_norm": 0.2771907150745392,
+      "learning_rate": 1.366861598440546e-05,
+      "loss": 0.091,
+      "step": 913
+    },
+    {
+      "epoch": 1.7126113455227379,
+      "grad_norm": 0.23459120094776154,
+      "learning_rate": 1.3660818713450294e-05,
+      "loss": 0.0701,
+      "step": 914
+    },
+    {
+      "epoch": 1.7144866385372715,
+      "grad_norm": 0.30127042531967163,
+      "learning_rate": 1.3653021442495126e-05,
+      "loss": 0.088,
+      "step": 915
+    },
+    {
+      "epoch": 1.716361931551805,
+      "grad_norm": 0.24772456288337708,
+      "learning_rate": 1.364522417153996e-05,
+      "loss": 0.0774,
+      "step": 916
+    },
+    {
+      "epoch": 1.7182372245663387,
+      "grad_norm": 0.2913759648799896,
+      "learning_rate": 1.3637426900584796e-05,
+      "loss": 0.0826,
+      "step": 917
+    },
+    {
+      "epoch": 1.720112517580872,
+      "grad_norm": 0.28088027238845825,
+      "learning_rate": 1.362962962962963e-05,
+      "loss": 0.0881,
+      "step": 918
+    },
+    {
+      "epoch": 1.7219878105954054,
+      "grad_norm": 0.30184197425842285,
+      "learning_rate": 1.3621832358674465e-05,
+      "loss": 0.0942,
+      "step": 919
+    },
+    {
+      "epoch": 1.723863103609939,
+      "grad_norm": 0.2570498287677765,
+      "learning_rate": 1.3614035087719299e-05,
+      "loss": 0.0693,
+      "step": 920
+    },
+    {
+      "epoch": 1.7257383966244726,
+      "grad_norm": 0.31009960174560547,
+      "learning_rate": 1.3606237816764133e-05,
+      "loss": 0.068,
+      "step": 921
+    },
+    {
+      "epoch": 1.7276136896390062,
+      "grad_norm": 0.23270297050476074,
+      "learning_rate": 1.3598440545808969e-05,
+      "loss": 0.0619,
+      "step": 922
+    },
+    {
+      "epoch": 1.7294889826535396,
+      "grad_norm": 0.33265721797943115,
+      "learning_rate": 1.3590643274853803e-05,
+      "loss": 0.0995,
+      "step": 923
+    },
+    {
+      "epoch": 1.7313642756680732,
+      "grad_norm": 0.38637009263038635,
+      "learning_rate": 1.3582846003898637e-05,
+      "loss": 0.1298,
+      "step": 924
+    },
+    {
+      "epoch": 1.7332395686826065,
+      "grad_norm": 0.27367690205574036,
+      "learning_rate": 1.357504873294347e-05,
+      "loss": 0.0852,
+      "step": 925
+    },
+    {
+      "epoch": 1.7351148616971401,
+      "grad_norm": 0.3776531517505646,
+      "learning_rate": 1.3567251461988304e-05,
+      "loss": 0.1081,
+      "step": 926
+    },
+    {
+      "epoch": 1.7369901547116737,
+      "grad_norm": 0.25184816122055054,
+      "learning_rate": 1.3559454191033138e-05,
+      "loss": 0.0747,
+      "step": 927
+    },
+    {
+      "epoch": 1.7388654477262073,
+      "grad_norm": 0.26781925559043884,
+      "learning_rate": 1.3551656920077974e-05,
+      "loss": 0.0974,
+      "step": 928
+    },
+    {
+      "epoch": 1.7407407407407407,
+      "grad_norm": 0.38023102283477783,
+      "learning_rate": 1.3543859649122808e-05,
+      "loss": 0.0989,
+      "step": 929
+    },
+    {
+      "epoch": 1.7426160337552743,
+      "grad_norm": 0.30040284991264343,
+      "learning_rate": 1.3536062378167643e-05,
+      "loss": 0.098,
+      "step": 930
+    },
+    {
+      "epoch": 1.7444913267698077,
+      "grad_norm": 0.41722437739372253,
+      "learning_rate": 1.3528265107212477e-05,
+      "loss": 0.0788,
+      "step": 931
+    },
+    {
+      "epoch": 1.7463666197843413,
+      "grad_norm": 0.31635239720344543,
+      "learning_rate": 1.3520467836257311e-05,
+      "loss": 0.131,
+      "step": 932
+    },
+    {
+      "epoch": 1.7482419127988749,
+      "grad_norm": 0.34196072816848755,
+      "learning_rate": 1.3512670565302147e-05,
+      "loss": 0.1302,
+      "step": 933
+    },
+    {
+      "epoch": 1.7501172058134085,
+      "grad_norm": 0.24944040179252625,
+      "learning_rate": 1.3504873294346981e-05,
+      "loss": 0.078,
+      "step": 934
+    },
+    {
+      "epoch": 1.7519924988279418,
+      "grad_norm": 0.2601966857910156,
+      "learning_rate": 1.3497076023391814e-05,
+      "loss": 0.0774,
+      "step": 935
+    },
+    {
+      "epoch": 1.7538677918424754,
+      "grad_norm": 0.27114635705947876,
+      "learning_rate": 1.3489278752436648e-05,
+      "loss": 0.0871,
+      "step": 936
+    },
+    {
+      "epoch": 1.7557430848570088,
+      "grad_norm": 0.27622026205062866,
+      "learning_rate": 1.3481481481481482e-05,
+      "loss": 0.0668,
+      "step": 937
+    },
+    {
+      "epoch": 1.7576183778715424,
+      "grad_norm": 0.32424378395080566,
+      "learning_rate": 1.3473684210526316e-05,
+      "loss": 0.1001,
+      "step": 938
+    },
+    {
+      "epoch": 1.759493670886076,
+      "grad_norm": 0.36716216802597046,
+      "learning_rate": 1.346588693957115e-05,
+      "loss": 0.1308,
+      "step": 939
+    },
+    {
+      "epoch": 1.7613689639006096,
+      "grad_norm": 0.29691779613494873,
+      "learning_rate": 1.3458089668615986e-05,
+      "loss": 0.0736,
+      "step": 940
+    },
+    {
+      "epoch": 1.763244256915143,
+      "grad_norm": 0.36653903126716614,
+      "learning_rate": 1.345029239766082e-05,
+      "loss": 0.1169,
+      "step": 941
+    },
+    {
+      "epoch": 1.7651195499296763,
+      "grad_norm": 0.21092906594276428,
+      "learning_rate": 1.3442495126705655e-05,
+      "loss": 0.0722,
+      "step": 942
+    },
+    {
+      "epoch": 1.76699484294421,
+      "grad_norm": 0.34096020460128784,
+      "learning_rate": 1.3434697855750489e-05,
+      "loss": 0.1135,
+      "step": 943
+    },
+    {
+      "epoch": 1.7688701359587435,
+      "grad_norm": 0.2584519386291504,
+      "learning_rate": 1.3426900584795323e-05,
+      "loss": 0.0712,
+      "step": 944
+    },
+    {
+      "epoch": 1.7707454289732771,
+      "grad_norm": 0.24627022445201874,
+      "learning_rate": 1.3419103313840155e-05,
+      "loss": 0.0683,
+      "step": 945
+    },
+    {
+      "epoch": 1.7726207219878107,
+      "grad_norm": 0.2898278534412384,
+      "learning_rate": 1.3411306042884991e-05,
+      "loss": 0.0833,
+      "step": 946
+    },
+    {
+      "epoch": 1.774496015002344,
+      "grad_norm": 0.2810303568840027,
+      "learning_rate": 1.3403508771929826e-05,
+      "loss": 0.1173,
+      "step": 947
+    },
+    {
+      "epoch": 1.7763713080168775,
+      "grad_norm": 0.2736395597457886,
+      "learning_rate": 1.339571150097466e-05,
+      "loss": 0.0681,
+      "step": 948
+    },
+    {
+      "epoch": 1.778246601031411,
+      "grad_norm": 0.31029412150382996,
+      "learning_rate": 1.3387914230019494e-05,
+      "loss": 0.0915,
+      "step": 949
+    },
+    {
+      "epoch": 1.7801218940459447,
+      "grad_norm": 0.24978873133659363,
+      "learning_rate": 1.3380116959064328e-05,
+      "loss": 0.091,
+      "step": 950
+    },
+    {
+      "epoch": 1.7801218940459447,
+      "eval_loss": 0.07832063734531403,
+      "eval_runtime": 676.2932,
+      "eval_samples_per_second": 0.291,
+      "eval_steps_per_second": 0.291,
+      "step": 950
+    },
+    {
+      "epoch": 1.7819971870604783,
+      "grad_norm": 0.3023127615451813,
+      "learning_rate": 1.3372319688109164e-05,
+      "loss": 0.0887,
+      "step": 951
+    },
+    {
+      "epoch": 1.7838724800750119,
+      "grad_norm": 0.4247094988822937,
+      "learning_rate": 1.3364522417153998e-05,
+      "loss": 0.1705,
+      "step": 952
+    },
+    {
+      "epoch": 1.7857477730895452,
+      "grad_norm": 0.3702313005924225,
+      "learning_rate": 1.3356725146198832e-05,
+      "loss": 0.1087,
+      "step": 953
+    },
+    {
+      "epoch": 1.7876230661040786,
+      "grad_norm": 0.26381444931030273,
+      "learning_rate": 1.3348927875243667e-05,
+      "loss": 0.072,
+      "step": 954
+    },
+    {
+      "epoch": 1.7894983591186122,
+      "grad_norm": 0.32169631123542786,
+      "learning_rate": 1.3341130604288499e-05,
+      "loss": 0.0828,
+      "step": 955
+    },
+    {
+      "epoch": 1.7913736521331458,
+      "grad_norm": 0.37914732098579407,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.0998,
+      "step": 956
+    },
+    {
+      "epoch": 1.7932489451476794,
+      "grad_norm": 0.2996065020561218,
+      "learning_rate": 1.3325536062378167e-05,
+      "loss": 0.1056,
+      "step": 957
+    },
+    {
+      "epoch": 1.795124238162213,
+      "grad_norm": 0.2926497161388397,
+      "learning_rate": 1.3317738791423003e-05,
+      "loss": 0.1248,
+      "step": 958
+    },
+    {
+      "epoch": 1.7969995311767464,
+      "grad_norm": 0.26383379101753235,
+      "learning_rate": 1.3309941520467838e-05,
+      "loss": 0.095,
+      "step": 959
+    },
+    {
+      "epoch": 1.7988748241912798,
+      "grad_norm": 0.334276020526886,
+      "learning_rate": 1.3302144249512672e-05,
+      "loss": 0.0806,
+      "step": 960
+    },
+    {
+      "epoch": 1.8007501172058133,
+      "grad_norm": 0.31996652483940125,
+      "learning_rate": 1.3294346978557506e-05,
+      "loss": 0.0691,
+      "step": 961
+    },
+    {
+      "epoch": 1.802625410220347,
+      "grad_norm": 0.3317575752735138,
+      "learning_rate": 1.328654970760234e-05,
+      "loss": 0.0823,
+      "step": 962
+    },
+    {
+      "epoch": 1.8045007032348805,
+      "grad_norm": 0.307790070772171,
+      "learning_rate": 1.3278752436647176e-05,
+      "loss": 0.0916,
+      "step": 963
+    },
+    {
+      "epoch": 1.806375996249414,
+      "grad_norm": 0.3371288478374481,
+      "learning_rate": 1.327095516569201e-05,
+      "loss": 0.0917,
+      "step": 964
+    },
+    {
+      "epoch": 1.8082512892639475,
+      "grad_norm": 0.4027099907398224,
+      "learning_rate": 1.3263157894736843e-05,
+      "loss": 0.1157,
+      "step": 965
+    },
+    {
+      "epoch": 1.810126582278481,
+      "grad_norm": 0.28519630432128906,
+      "learning_rate": 1.3255360623781677e-05,
+      "loss": 0.0841,
+      "step": 966
+    },
+    {
+      "epoch": 1.8120018752930145,
+      "grad_norm": 0.2463994026184082,
+      "learning_rate": 1.3247563352826511e-05,
+      "loss": 0.0894,
+      "step": 967
+    },
+    {
+      "epoch": 1.813877168307548,
+      "grad_norm": 0.2853533625602722,
+      "learning_rate": 1.3239766081871345e-05,
+      "loss": 0.0667,
+      "step": 968
+    },
+    {
+      "epoch": 1.8157524613220817,
+      "grad_norm": 0.25355201959609985,
+      "learning_rate": 1.3231968810916181e-05,
+      "loss": 0.0765,
+      "step": 969
+    },
+    {
+      "epoch": 1.817627754336615,
+      "grad_norm": 0.3773306608200073,
+      "learning_rate": 1.3224171539961015e-05,
+      "loss": 0.0854,
+      "step": 970
+    },
+    {
+      "epoch": 1.8195030473511487,
+      "grad_norm": 0.29040902853012085,
+      "learning_rate": 1.321637426900585e-05,
+      "loss": 0.0893,
+      "step": 971
+    },
+    {
+      "epoch": 1.821378340365682,
+      "grad_norm": 0.35382455587387085,
+      "learning_rate": 1.3208576998050684e-05,
+      "loss": 0.1021,
+      "step": 972
+    },
+    {
+      "epoch": 1.8232536333802156,
+      "grad_norm": 0.4576570391654968,
+      "learning_rate": 1.3200779727095518e-05,
+      "loss": 0.1179,
+      "step": 973
+    },
+    {
+      "epoch": 1.8251289263947492,
+      "grad_norm": 0.37235504388809204,
+      "learning_rate": 1.3192982456140354e-05,
+      "loss": 0.1344,
+      "step": 974
+    },
+    {
+      "epoch": 1.8270042194092828,
+      "grad_norm": 0.30962350964546204,
+      "learning_rate": 1.3185185185185185e-05,
+      "loss": 0.0751,
+      "step": 975
+    },
+    {
+      "epoch": 1.8288795124238162,
+      "grad_norm": 0.2378237396478653,
+      "learning_rate": 1.317738791423002e-05,
+      "loss": 0.0714,
+      "step": 976
+    },
+    {
+      "epoch": 1.8307548054383498,
+      "grad_norm": 0.2367488443851471,
+      "learning_rate": 1.3169590643274855e-05,
+      "loss": 0.0599,
+      "step": 977
+    },
+    {
+      "epoch": 1.8326300984528832,
+      "grad_norm": 0.2833852171897888,
+      "learning_rate": 1.3161793372319689e-05,
+      "loss": 0.0995,
+      "step": 978
+    },
+    {
+      "epoch": 1.8345053914674168,
+      "grad_norm": 1.0794099569320679,
+      "learning_rate": 1.3153996101364523e-05,
+      "loss": 0.1319,
+      "step": 979
+    },
+    {
+      "epoch": 1.8363806844819504,
+      "grad_norm": 0.29689860343933105,
+      "learning_rate": 1.3146198830409357e-05,
+      "loss": 0.0849,
+      "step": 980
+    },
+    {
+      "epoch": 1.838255977496484,
+      "grad_norm": 0.3836843967437744,
+      "learning_rate": 1.3138401559454193e-05,
+      "loss": 0.1075,
+      "step": 981
+    },
+    {
+      "epoch": 1.8401312705110173,
+      "grad_norm": 0.27197182178497314,
+      "learning_rate": 1.3130604288499027e-05,
+      "loss": 0.08,
+      "step": 982
+    },
+    {
+      "epoch": 1.842006563525551,
+      "grad_norm": 0.3288334012031555,
+      "learning_rate": 1.3122807017543862e-05,
+      "loss": 0.0604,
+      "step": 983
+    },
+    {
+      "epoch": 1.8438818565400843,
+      "grad_norm": 0.3458631932735443,
+      "learning_rate": 1.3115009746588696e-05,
+      "loss": 0.1051,
+      "step": 984
+    },
+    {
+      "epoch": 1.845757149554618,
+      "grad_norm": 0.28444722294807434,
+      "learning_rate": 1.3107212475633528e-05,
+      "loss": 0.0876,
+      "step": 985
+    },
+    {
+      "epoch": 1.8476324425691515,
+      "grad_norm": 0.3890294134616852,
+      "learning_rate": 1.3099415204678362e-05,
+      "loss": 0.1209,
+      "step": 986
+    },
+    {
+      "epoch": 1.849507735583685,
+      "grad_norm": 0.2536354064941406,
+      "learning_rate": 1.3091617933723197e-05,
+      "loss": 0.0615,
+      "step": 987
+    },
+    {
+      "epoch": 1.8513830285982185,
+      "grad_norm": 0.2301822453737259,
+      "learning_rate": 1.3083820662768032e-05,
+      "loss": 0.0802,
+      "step": 988
+    },
+    {
+      "epoch": 1.8532583216127518,
+      "grad_norm": 0.5876390337944031,
+      "learning_rate": 1.3076023391812867e-05,
+      "loss": 0.134,
+      "step": 989
+    },
+    {
+      "epoch": 1.8551336146272854,
+      "grad_norm": 0.2957145571708679,
+      "learning_rate": 1.3068226120857701e-05,
+      "loss": 0.0834,
+      "step": 990
+    },
+    {
+      "epoch": 1.857008907641819,
+      "grad_norm": 0.321350634098053,
+      "learning_rate": 1.3060428849902535e-05,
+      "loss": 0.0998,
+      "step": 991
+    },
+    {
+      "epoch": 1.8588842006563526,
+      "grad_norm": 0.32503458857536316,
+      "learning_rate": 1.305263157894737e-05,
+      "loss": 0.1065,
+      "step": 992
+    },
+    {
+      "epoch": 1.8607594936708862,
+      "grad_norm": 0.2906481623649597,
+      "learning_rate": 1.3044834307992205e-05,
+      "loss": 0.0878,
+      "step": 993
+    },
+    {
+      "epoch": 1.8626347866854196,
+      "grad_norm": 0.25748249888420105,
+      "learning_rate": 1.303703703703704e-05,
+      "loss": 0.0784,
+      "step": 994
+    },
+    {
+      "epoch": 1.864510079699953,
+      "grad_norm": 0.2645077407360077,
+      "learning_rate": 1.3029239766081872e-05,
+      "loss": 0.0827,
+      "step": 995
+    },
+    {
+      "epoch": 1.8663853727144866,
+      "grad_norm": 0.28453439474105835,
+      "learning_rate": 1.3021442495126706e-05,
+      "loss": 0.0792,
+      "step": 996
+    },
+    {
+      "epoch": 1.8682606657290202,
+      "grad_norm": 0.2839300036430359,
+      "learning_rate": 1.301364522417154e-05,
+      "loss": 0.069,
+      "step": 997
+    },
+    {
+      "epoch": 1.8701359587435538,
+      "grad_norm": 0.40391039848327637,
+      "learning_rate": 1.3005847953216374e-05,
+      "loss": 0.068,
+      "step": 998
+    },
+    {
+      "epoch": 1.8720112517580874,
+      "grad_norm": 0.29251784086227417,
+      "learning_rate": 1.299805068226121e-05,
+      "loss": 0.0731,
+      "step": 999
+    },
+    {
+      "epoch": 1.8738865447726207,
+      "grad_norm": 0.2935086786746979,
+      "learning_rate": 1.2990253411306044e-05,
+      "loss": 0.0984,
+      "step": 1000
+    },
+    {
+      "epoch": 1.8738865447726207,
+      "eval_loss": 0.07738383859395981,
+      "eval_runtime": 676.7683,
+      "eval_samples_per_second": 0.291,
+      "eval_steps_per_second": 0.291,
+      "step": 1000
+    },
+    {
+      "epoch": 1.875761837787154,
+      "grad_norm": 0.5241721868515015,
+      "learning_rate": 1.2982456140350879e-05,
+      "loss": 0.1288,
+      "step": 1001
+    },
+    {
+      "epoch": 1.8776371308016877,
+      "grad_norm": 0.2855282425880432,
+      "learning_rate": 1.2974658869395713e-05,
+      "loss": 0.0962,
+      "step": 1002
+    },
+    {
+      "epoch": 1.8795124238162213,
+      "grad_norm": 0.26559796929359436,
+      "learning_rate": 1.2966861598440547e-05,
+      "loss": 0.0838,
+      "step": 1003
+    },
+    {
+      "epoch": 1.881387716830755,
+      "grad_norm": 0.31060707569122314,
+      "learning_rate": 1.2959064327485383e-05,
+      "loss": 0.1008,
+      "step": 1004
+    },
+    {
+      "epoch": 1.8832630098452883,
+      "grad_norm": 0.30378690361976624,
+      "learning_rate": 1.2951267056530214e-05,
+      "loss": 0.0783,
+      "step": 1005
+    },
+    {
+      "epoch": 1.8851383028598219,
+      "grad_norm": 0.3393579125404358,
+      "learning_rate": 1.294346978557505e-05,
+      "loss": 0.075,
+      "step": 1006
+    },
+    {
+      "epoch": 1.8870135958743552,
+      "grad_norm": 0.36604905128479004,
+      "learning_rate": 1.2935672514619884e-05,
+      "loss": 0.0771,
+      "step": 1007
+    },
+    {
+      "epoch": 1.8888888888888888,
+      "grad_norm": 0.3377784192562103,
+      "learning_rate": 1.2927875243664718e-05,
+      "loss": 0.1062,
+      "step": 1008
+    },
+    {
+      "epoch": 1.8907641819034224,
+      "grad_norm": 0.29350435733795166,
+      "learning_rate": 1.2920077972709552e-05,
+      "loss": 0.0703,
+      "step": 1009
+    },
+    {
+      "epoch": 1.892639474917956,
+      "grad_norm": 0.2543967366218567,
+      "learning_rate": 1.2912280701754386e-05,
+      "loss": 0.0736,
+      "step": 1010
+    },
+    {
+      "epoch": 1.8945147679324894,
+      "grad_norm": 0.29992133378982544,
+      "learning_rate": 1.2904483430799222e-05,
+      "loss": 0.0945,
+      "step": 1011
+    },
+    {
+      "epoch": 1.896390060947023,
+      "grad_norm": 0.26045867800712585,
+      "learning_rate": 1.2896686159844056e-05,
+      "loss": 0.0823,
+      "step": 1012
+    },
+    {
+      "epoch": 1.8982653539615564,
+      "grad_norm": 0.37396302819252014,
+      "learning_rate": 1.288888888888889e-05,
+      "loss": 0.1008,
+      "step": 1013
+    },
+    {
+      "epoch": 1.90014064697609,
+      "grad_norm": 0.3353418707847595,
+      "learning_rate": 1.2881091617933725e-05,
+      "loss": 0.1059,
+      "step": 1014
+    },
+    {
+      "epoch": 1.9020159399906236,
+      "grad_norm": 0.2745090126991272,
+      "learning_rate": 1.2873294346978557e-05,
+      "loss": 0.0838,
+      "step": 1015
+    },
+    {
+      "epoch": 1.9038912330051572,
+      "grad_norm": 0.25266072154045105,
+      "learning_rate": 1.2865497076023392e-05,
+      "loss": 0.0638,
+      "step": 1016
+    },
+    {
+      "epoch": 1.9057665260196905,
+      "grad_norm": 0.41703373193740845,
+      "learning_rate": 1.2857699805068227e-05,
+      "loss": 0.1085,
+      "step": 1017
+    },
+    {
+      "epoch": 1.9076418190342241,
+      "grad_norm": 0.35806798934936523,
+      "learning_rate": 1.2849902534113062e-05,
+      "loss": 0.0803,
+      "step": 1018
+    },
+    {
+      "epoch": 1.9095171120487575,
+      "grad_norm": 0.36233675479888916,
+      "learning_rate": 1.2842105263157896e-05,
+      "loss": 0.1004,
+      "step": 1019
+    },
+    {
+      "epoch": 1.9113924050632911,
+      "grad_norm": 0.2726714611053467,
+      "learning_rate": 1.283430799220273e-05,
+      "loss": 0.0771,
+      "step": 1020
+    },
+    {
+      "epoch": 1.9132676980778247,
+      "grad_norm": 0.25889918208122253,
+      "learning_rate": 1.2826510721247564e-05,
+      "loss": 0.0729,
+      "step": 1021
+    },
+    {
+      "epoch": 1.9151429910923583,
+      "grad_norm": 0.27719736099243164,
+      "learning_rate": 1.28187134502924e-05,
+      "loss": 0.0926,
+      "step": 1022
+    },
+    {
+      "epoch": 1.9170182841068917,
+      "grad_norm": 0.6307101249694824,
+      "learning_rate": 1.2810916179337234e-05,
+      "loss": 0.1319,
+      "step": 1023
+    },
+    {
+      "epoch": 1.9188935771214253,
+      "grad_norm": 0.3898720443248749,
+      "learning_rate": 1.2803118908382068e-05,
+      "loss": 0.0881,
+      "step": 1024
+    },
+    {
+      "epoch": 1.9207688701359587,
+      "grad_norm": 0.33780285716056824,
+      "learning_rate": 1.2795321637426901e-05,
+      "loss": 0.1037,
+      "step": 1025
+    },
+    {
+      "epoch": 1.9226441631504922,
+      "grad_norm": 0.28749212622642517,
+      "learning_rate": 1.2787524366471735e-05,
+      "loss": 0.093,
+      "step": 1026
+    },
+    {
+      "epoch": 1.9245194561650258,
+      "grad_norm": 0.33152034878730774,
+      "learning_rate": 1.277972709551657e-05,
+      "loss": 0.0945,
+      "step": 1027
+    },
+    {
+      "epoch": 1.9263947491795594,
+      "grad_norm": 0.28936097025871277,
+      "learning_rate": 1.2771929824561404e-05,
+      "loss": 0.0916,
+      "step": 1028
+    },
+    {
+      "epoch": 1.9282700421940928,
+      "grad_norm": 0.2941216826438904,
+      "learning_rate": 1.276413255360624e-05,
+      "loss": 0.0994,
+      "step": 1029
+    },
+    {
+      "epoch": 1.9301453352086262,
+      "grad_norm": 0.30549120903015137,
+      "learning_rate": 1.2756335282651074e-05,
+      "loss": 0.0899,
+      "step": 1030
+    },
+    {
+      "epoch": 1.9320206282231598,
+      "grad_norm": 0.24775011837482452,
+      "learning_rate": 1.2748538011695908e-05,
+      "loss": 0.0623,
+      "step": 1031
+    },
+    {
+      "epoch": 1.9338959212376934,
+      "grad_norm": 0.27423885464668274,
+      "learning_rate": 1.2740740740740742e-05,
+      "loss": 0.0614,
+      "step": 1032
+    },
+    {
+      "epoch": 1.935771214252227,
+      "grad_norm": 0.33300426602363586,
+      "learning_rate": 1.2732943469785576e-05,
+      "loss": 0.1014,
+      "step": 1033
+    },
+    {
+      "epoch": 1.9376465072667606,
+      "grad_norm": 0.318990558385849,
+      "learning_rate": 1.2725146198830412e-05,
+      "loss": 0.0904,
+      "step": 1034
+    },
+    {
+      "epoch": 1.939521800281294,
+      "grad_norm": 0.2824667692184448,
+      "learning_rate": 1.2717348927875245e-05,
+      "loss": 0.0735,
+      "step": 1035
+    },
+    {
+      "epoch": 1.9413970932958273,
+      "grad_norm": 0.2873893976211548,
+      "learning_rate": 1.2709551656920079e-05,
+      "loss": 0.066,
+      "step": 1036
+    },
+    {
+      "epoch": 1.943272386310361,
+      "grad_norm": 0.33205756545066833,
+      "learning_rate": 1.2701754385964913e-05,
+      "loss": 0.1017,
+      "step": 1037
+    },
+    {
+      "epoch": 1.9451476793248945,
+      "grad_norm": 0.5657170414924622,
+      "learning_rate": 1.2693957115009747e-05,
+      "loss": 0.08,
+      "step": 1038
+    },
+    {
+      "epoch": 1.9470229723394281,
+      "grad_norm": 0.5216059684753418,
+      "learning_rate": 1.2686159844054581e-05,
+      "loss": 0.0942,
+      "step": 1039
+    },
+    {
+      "epoch": 1.9488982653539617,
+      "grad_norm": 0.45388665795326233,
+      "learning_rate": 1.2678362573099417e-05,
+      "loss": 0.0876,
+      "step": 1040
+    },
+    {
+      "epoch": 1.950773558368495,
+      "grad_norm": 0.28133487701416016,
+      "learning_rate": 1.2670565302144251e-05,
+      "loss": 0.0829,
+      "step": 1041
+    },
+    {
+      "epoch": 1.9526488513830285,
+      "grad_norm": 0.32736194133758545,
+      "learning_rate": 1.2662768031189086e-05,
+      "loss": 0.1283,
+      "step": 1042
+    },
+    {
+      "epoch": 1.954524144397562,
+      "grad_norm": 0.307423859834671,
+      "learning_rate": 1.265497076023392e-05,
+      "loss": 0.0954,
+      "step": 1043
+    },
+    {
+      "epoch": 1.9563994374120957,
+      "grad_norm": 0.3167324662208557,
+      "learning_rate": 1.2647173489278754e-05,
+      "loss": 0.0948,
+      "step": 1044
+    },
+    {
+      "epoch": 1.9582747304266293,
+      "grad_norm": 0.2761521637439728,
+      "learning_rate": 1.2639376218323586e-05,
+      "loss": 0.0632,
+      "step": 1045
+    },
+    {
+      "epoch": 1.9601500234411628,
+      "grad_norm": 0.3254704475402832,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 0.068,
+      "step": 1046
+    },
+    {
+      "epoch": 1.9620253164556962,
+      "grad_norm": 0.41552290320396423,
+      "learning_rate": 1.2623781676413257e-05,
+      "loss": 0.1075,
+      "step": 1047
+    },
+    {
+      "epoch": 1.9639006094702296,
+      "grad_norm": 0.31078511476516724,
+      "learning_rate": 1.261598440545809e-05,
+      "loss": 0.0812,
+      "step": 1048
+    },
+    {
+      "epoch": 1.9657759024847632,
+      "grad_norm": 0.3442966639995575,
+      "learning_rate": 1.2608187134502925e-05,
+      "loss": 0.0926,
+      "step": 1049
+    },
+    {
+      "epoch": 1.9676511954992968,
+      "grad_norm": 0.2290525734424591,
+      "learning_rate": 1.2600389863547759e-05,
+      "loss": 0.069,
+      "step": 1050
+    },
+    {
+      "epoch": 1.9676511954992968,
+      "eval_loss": 0.07501588761806488,
+      "eval_runtime": 675.7387,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 1050
+    },
+    {
+      "epoch": 1.9695264885138304,
+      "grad_norm": 0.2715151309967041,
+      "learning_rate": 1.2592592592592593e-05,
+      "loss": 0.0701,
+      "step": 1051
+    },
+    {
+      "epoch": 1.9714017815283638,
+      "grad_norm": 0.3169614374637604,
+      "learning_rate": 1.258479532163743e-05,
+      "loss": 0.1119,
+      "step": 1052
+    },
+    {
+      "epoch": 1.9732770745428974,
+      "grad_norm": 0.6188083291053772,
+      "learning_rate": 1.2576998050682263e-05,
+      "loss": 0.0969,
+      "step": 1053
+    },
+    {
+      "epoch": 1.9751523675574307,
+      "grad_norm": 0.28429678082466125,
+      "learning_rate": 1.2569200779727098e-05,
+      "loss": 0.058,
+      "step": 1054
+    },
+    {
+      "epoch": 1.9770276605719643,
+      "grad_norm": 0.7146753072738647,
+      "learning_rate": 1.256140350877193e-05,
+      "loss": 0.0636,
+      "step": 1055
+    },
+    {
+      "epoch": 1.978902953586498,
+      "grad_norm": 0.5578083992004395,
+      "learning_rate": 1.2553606237816764e-05,
+      "loss": 0.0935,
+      "step": 1056
+    },
+    {
+      "epoch": 1.9807782466010315,
+      "grad_norm": 0.30978304147720337,
+      "learning_rate": 1.2545808966861598e-05,
+      "loss": 0.0784,
+      "step": 1057
+    },
+    {
+      "epoch": 1.982653539615565,
+      "grad_norm": 0.42586255073547363,
+      "learning_rate": 1.2538011695906434e-05,
+      "loss": 0.0982,
+      "step": 1058
+    },
+    {
+      "epoch": 1.9845288326300985,
+      "grad_norm": 0.3126954138278961,
+      "learning_rate": 1.2530214424951269e-05,
+      "loss": 0.0776,
+      "step": 1059
+    },
+    {
+      "epoch": 1.9864041256446319,
+      "grad_norm": 0.37667107582092285,
+      "learning_rate": 1.2522417153996103e-05,
+      "loss": 0.1329,
+      "step": 1060
+    },
+    {
+      "epoch": 1.9882794186591655,
+      "grad_norm": 0.484842985868454,
+      "learning_rate": 1.2514619883040937e-05,
+      "loss": 0.0538,
+      "step": 1061
+    },
+    {
+      "epoch": 1.990154711673699,
+      "grad_norm": 0.3426309823989868,
+      "learning_rate": 1.2506822612085771e-05,
+      "loss": 0.1018,
+      "step": 1062
+    },
+    {
+      "epoch": 1.9920300046882327,
+      "grad_norm": 0.2878342866897583,
+      "learning_rate": 1.2499025341130607e-05,
+      "loss": 0.094,
+      "step": 1063
+    },
+    {
+      "epoch": 1.993905297702766,
+      "grad_norm": 0.30334293842315674,
+      "learning_rate": 1.2491228070175441e-05,
+      "loss": 0.0705,
+      "step": 1064
+    },
+    {
+      "epoch": 1.9957805907172996,
+      "grad_norm": 0.31606927514076233,
+      "learning_rate": 1.2483430799220274e-05,
+      "loss": 0.0955,
+      "step": 1065
+    },
+    {
+      "epoch": 1.997655883731833,
+      "grad_norm": 0.3537469804286957,
+      "learning_rate": 1.2475633528265108e-05,
+      "loss": 0.0897,
+      "step": 1066
+    },
+    {
+      "epoch": 1.9995311767463666,
+      "grad_norm": 0.30905476212501526,
+      "learning_rate": 1.2467836257309942e-05,
+      "loss": 0.1066,
+      "step": 1067
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.5351856350898743,
+      "learning_rate": 1.2460038986354776e-05,
+      "loss": 0.0439,
+      "step": 1068
+    },
+    {
+      "epoch": 2.0018752930145336,
+      "grad_norm": 0.3859151601791382,
+      "learning_rate": 1.245224171539961e-05,
+      "loss": 0.086,
+      "step": 1069
+    },
+    {
+      "epoch": 2.003750586029067,
+      "grad_norm": 0.3007124066352844,
+      "learning_rate": 1.2444444444444446e-05,
+      "loss": 0.0868,
+      "step": 1070
+    },
+    {
+      "epoch": 2.0056258790436003,
+      "grad_norm": 0.26716360449790955,
+      "learning_rate": 1.243664717348928e-05,
+      "loss": 0.1024,
+      "step": 1071
+    },
+    {
+      "epoch": 2.007501172058134,
+      "grad_norm": 0.33358505368232727,
+      "learning_rate": 1.2428849902534115e-05,
+      "loss": 0.097,
+      "step": 1072
+    },
+    {
+      "epoch": 2.0093764650726675,
+      "grad_norm": 0.30429476499557495,
+      "learning_rate": 1.2421052631578949e-05,
+      "loss": 0.0773,
+      "step": 1073
+    },
+    {
+      "epoch": 2.011251758087201,
+      "grad_norm": 0.2670055627822876,
+      "learning_rate": 1.2413255360623783e-05,
+      "loss": 0.0705,
+      "step": 1074
+    },
+    {
+      "epoch": 2.0131270511017347,
+      "grad_norm": 0.30095306038856506,
+      "learning_rate": 1.2405458089668616e-05,
+      "loss": 0.0999,
+      "step": 1075
+    },
+    {
+      "epoch": 2.0150023441162683,
+      "grad_norm": 0.3201741874217987,
+      "learning_rate": 1.239766081871345e-05,
+      "loss": 0.1148,
+      "step": 1076
+    },
+    {
+      "epoch": 2.0168776371308015,
+      "grad_norm": 0.31769925355911255,
+      "learning_rate": 1.2389863547758286e-05,
+      "loss": 0.0756,
+      "step": 1077
+    },
+    {
+      "epoch": 2.018752930145335,
+      "grad_norm": 0.3209584951400757,
+      "learning_rate": 1.238206627680312e-05,
+      "loss": 0.077,
+      "step": 1078
+    },
+    {
+      "epoch": 2.0206282231598687,
+      "grad_norm": 0.2504936754703522,
+      "learning_rate": 1.2374269005847954e-05,
+      "loss": 0.0819,
+      "step": 1079
+    },
+    {
+      "epoch": 2.0225035161744023,
+      "grad_norm": 0.31677699089050293,
+      "learning_rate": 1.2366471734892788e-05,
+      "loss": 0.0874,
+      "step": 1080
+    },
+    {
+      "epoch": 2.024378809188936,
+      "grad_norm": 0.35840892791748047,
+      "learning_rate": 1.2358674463937622e-05,
+      "loss": 0.0859,
+      "step": 1081
+    },
+    {
+      "epoch": 2.0262541022034695,
+      "grad_norm": 0.3263484537601471,
+      "learning_rate": 1.2350877192982458e-05,
+      "loss": 0.1069,
+      "step": 1082
+    },
+    {
+      "epoch": 2.0281293952180026,
+      "grad_norm": 0.23933690786361694,
+      "learning_rate": 1.2343079922027293e-05,
+      "loss": 0.0628,
+      "step": 1083
+    },
+    {
+      "epoch": 2.030004688232536,
+      "grad_norm": 0.2623092532157898,
+      "learning_rate": 1.2335282651072127e-05,
+      "loss": 0.0698,
+      "step": 1084
+    },
+    {
+      "epoch": 2.03187998124707,
+      "grad_norm": 0.30900081992149353,
+      "learning_rate": 1.232748538011696e-05,
+      "loss": 0.0713,
+      "step": 1085
+    },
+    {
+      "epoch": 2.0337552742616034,
+      "grad_norm": 0.3421262502670288,
+      "learning_rate": 1.2319688109161793e-05,
+      "loss": 0.1117,
+      "step": 1086
+    },
+    {
+      "epoch": 2.035630567276137,
+      "grad_norm": 0.22926147282123566,
+      "learning_rate": 1.2311890838206628e-05,
+      "loss": 0.0522,
+      "step": 1087
+    },
+    {
+      "epoch": 2.0375058602906706,
+      "grad_norm": 0.2777685821056366,
+      "learning_rate": 1.2304093567251463e-05,
+      "loss": 0.0524,
+      "step": 1088
+    },
+    {
+      "epoch": 2.0393811533052038,
+      "grad_norm": 0.3484315276145935,
+      "learning_rate": 1.2296296296296298e-05,
+      "loss": 0.0844,
+      "step": 1089
+    },
+    {
+      "epoch": 2.0412564463197373,
+      "grad_norm": 0.3067864775657654,
+      "learning_rate": 1.2288499025341132e-05,
+      "loss": 0.0889,
+      "step": 1090
+    },
+    {
+      "epoch": 2.043131739334271,
+      "grad_norm": 0.3049250543117523,
+      "learning_rate": 1.2280701754385966e-05,
+      "loss": 0.0874,
+      "step": 1091
+    },
+    {
+      "epoch": 2.0450070323488045,
+      "grad_norm": 0.371250718832016,
+      "learning_rate": 1.22729044834308e-05,
+      "loss": 0.0608,
+      "step": 1092
+    },
+    {
+      "epoch": 2.046882325363338,
+      "grad_norm": 0.35286757349967957,
+      "learning_rate": 1.2265107212475636e-05,
+      "loss": 0.0617,
+      "step": 1093
+    },
+    {
+      "epoch": 2.0487576183778717,
+      "grad_norm": 0.31813135743141174,
+      "learning_rate": 1.225730994152047e-05,
+      "loss": 0.072,
+      "step": 1094
+    },
+    {
+      "epoch": 2.050632911392405,
+      "grad_norm": 0.3504825830459595,
+      "learning_rate": 1.2249512670565303e-05,
+      "loss": 0.0917,
+      "step": 1095
+    },
+    {
+      "epoch": 2.0525082044069385,
+      "grad_norm": 0.3622145354747772,
+      "learning_rate": 1.2241715399610137e-05,
+      "loss": 0.1072,
+      "step": 1096
+    },
+    {
+      "epoch": 2.054383497421472,
+      "grad_norm": 0.2720160186290741,
+      "learning_rate": 1.2233918128654971e-05,
+      "loss": 0.0644,
+      "step": 1097
+    },
+    {
+      "epoch": 2.0562587904360057,
+      "grad_norm": 0.2766894996166229,
+      "learning_rate": 1.2226120857699805e-05,
+      "loss": 0.0764,
+      "step": 1098
+    },
+    {
+      "epoch": 2.0581340834505393,
+      "grad_norm": 0.3748610019683838,
+      "learning_rate": 1.221832358674464e-05,
+      "loss": 0.074,
+      "step": 1099
+    },
+    {
+      "epoch": 2.060009376465073,
+      "grad_norm": 0.2557353377342224,
+      "learning_rate": 1.2210526315789475e-05,
+      "loss": 0.0731,
+      "step": 1100
+    },
+    {
+      "epoch": 2.060009376465073,
+      "eval_loss": 0.07578001916408539,
+      "eval_runtime": 676.4414,
+      "eval_samples_per_second": 0.291,
+      "eval_steps_per_second": 0.291,
+      "step": 1100
+    },
+    {
+      "epoch": 2.061884669479606,
+      "grad_norm": 0.4064513146877289,
+      "learning_rate": 1.220272904483431e-05,
+      "loss": 0.0999,
+      "step": 1101
+    },
+    {
+      "epoch": 2.0637599624941396,
+      "grad_norm": 0.3794358968734741,
+      "learning_rate": 1.2194931773879144e-05,
+      "loss": 0.1305,
+      "step": 1102
+    },
+    {
+      "epoch": 2.065635255508673,
+      "grad_norm": 0.3337356448173523,
+      "learning_rate": 1.2187134502923978e-05,
+      "loss": 0.073,
+      "step": 1103
+    },
+    {
+      "epoch": 2.067510548523207,
+      "grad_norm": 0.310234010219574,
+      "learning_rate": 1.2179337231968812e-05,
+      "loss": 0.0899,
+      "step": 1104
+    },
+    {
+      "epoch": 2.0693858415377404,
+      "grad_norm": 0.28629547357559204,
+      "learning_rate": 1.2171539961013645e-05,
+      "loss": 0.0674,
+      "step": 1105
+    },
+    {
+      "epoch": 2.071261134552274,
+      "grad_norm": 0.31278061866760254,
+      "learning_rate": 1.216374269005848e-05,
+      "loss": 0.0724,
+      "step": 1106
+    },
+    {
+      "epoch": 2.073136427566807,
+      "grad_norm": 0.3247126638889313,
+      "learning_rate": 1.2155945419103315e-05,
+      "loss": 0.0771,
+      "step": 1107
+    },
+    {
+      "epoch": 2.0750117205813408,
+      "grad_norm": 0.3694022297859192,
+      "learning_rate": 1.2148148148148149e-05,
+      "loss": 0.0897,
+      "step": 1108
+    },
+    {
+      "epoch": 2.0768870135958744,
+      "grad_norm": 0.3070647120475769,
+      "learning_rate": 1.2140350877192983e-05,
+      "loss": 0.0817,
+      "step": 1109
+    },
+    {
+      "epoch": 2.078762306610408,
+      "grad_norm": 0.3457229733467102,
+      "learning_rate": 1.2132553606237817e-05,
+      "loss": 0.0743,
+      "step": 1110
+    },
+    {
+      "epoch": 2.0806375996249415,
+      "grad_norm": 0.31315433979034424,
+      "learning_rate": 1.2124756335282653e-05,
+      "loss": 0.0742,
+      "step": 1111
+    },
+    {
+      "epoch": 2.0825128926394747,
+      "grad_norm": 0.31154918670654297,
+      "learning_rate": 1.2116959064327487e-05,
+      "loss": 0.0748,
+      "step": 1112
+    },
+    {
+      "epoch": 2.0843881856540083,
+      "grad_norm": 0.3989027738571167,
+      "learning_rate": 1.2109161793372322e-05,
+      "loss": 0.1073,
+      "step": 1113
+    },
+    {
+      "epoch": 2.086263478668542,
+      "grad_norm": 0.3276286721229553,
+      "learning_rate": 1.2101364522417156e-05,
+      "loss": 0.1067,
+      "step": 1114
+    },
+    {
+      "epoch": 2.0881387716830755,
+      "grad_norm": 0.32193440198898315,
+      "learning_rate": 1.2093567251461988e-05,
+      "loss": 0.0636,
+      "step": 1115
+    },
+    {
+      "epoch": 2.090014064697609,
+      "grad_norm": 0.34374260902404785,
+      "learning_rate": 1.2085769980506823e-05,
+      "loss": 0.092,
+      "step": 1116
+    },
+    {
+      "epoch": 2.0918893577121427,
+      "grad_norm": 0.24487826228141785,
+      "learning_rate": 1.2077972709551657e-05,
+      "loss": 0.0617,
+      "step": 1117
+    },
+    {
+      "epoch": 2.093764650726676,
+      "grad_norm": 0.3080878257751465,
+      "learning_rate": 1.2070175438596493e-05,
+      "loss": 0.0682,
+      "step": 1118
+    },
+    {
+      "epoch": 2.0956399437412094,
+      "grad_norm": 0.2777683138847351,
+      "learning_rate": 1.2062378167641327e-05,
+      "loss": 0.0716,
+      "step": 1119
+    },
+    {
+      "epoch": 2.097515236755743,
+      "grad_norm": 0.3162023723125458,
+      "learning_rate": 1.2054580896686161e-05,
+      "loss": 0.0841,
+      "step": 1120
+    },
+    {
+      "epoch": 2.0993905297702766,
+      "grad_norm": 0.4254573881626129,
+      "learning_rate": 1.2046783625730995e-05,
+      "loss": 0.0511,
+      "step": 1121
+    },
+    {
+      "epoch": 2.1012658227848102,
+      "grad_norm": 0.33648309111595154,
+      "learning_rate": 1.203898635477583e-05,
+      "loss": 0.0949,
+      "step": 1122
+    },
+    {
+      "epoch": 2.103141115799344,
+      "grad_norm": 0.24535001814365387,
+      "learning_rate": 1.2031189083820665e-05,
+      "loss": 0.0637,
+      "step": 1123
+    },
+    {
+      "epoch": 2.105016408813877,
+      "grad_norm": 0.2758919894695282,
+      "learning_rate": 1.20233918128655e-05,
+      "loss": 0.0825,
+      "step": 1124
+    },
+    {
+      "epoch": 2.1068917018284106,
+      "grad_norm": 0.29626914858818054,
+      "learning_rate": 1.2015594541910332e-05,
+      "loss": 0.0829,
+      "step": 1125
+    },
+    {
+      "epoch": 2.108766994842944,
+      "grad_norm": 0.43505001068115234,
+      "learning_rate": 1.2007797270955166e-05,
+      "loss": 0.1068,
+      "step": 1126
+    },
+    {
+      "epoch": 2.1106422878574778,
+      "grad_norm": 0.29316994547843933,
+      "learning_rate": 1.2e-05,
+      "loss": 0.0632,
+      "step": 1127
+    },
+    {
+      "epoch": 2.1125175808720114,
+      "grad_norm": 0.3550972640514374,
+      "learning_rate": 1.1992202729044834e-05,
+      "loss": 0.0824,
+      "step": 1128
+    },
+    {
+      "epoch": 2.114392873886545,
+      "grad_norm": 0.34816864132881165,
+      "learning_rate": 1.198440545808967e-05,
+      "loss": 0.0885,
+      "step": 1129
+    },
+    {
+      "epoch": 2.116268166901078,
+      "grad_norm": 0.3325449526309967,
+      "learning_rate": 1.1976608187134505e-05,
+      "loss": 0.0932,
+      "step": 1130
+    },
+    {
+      "epoch": 2.1181434599156117,
+      "grad_norm": 0.3157937228679657,
+      "learning_rate": 1.1968810916179339e-05,
+      "loss": 0.0735,
+      "step": 1131
+    },
+    {
+      "epoch": 2.1200187529301453,
+      "grad_norm": 0.32261866331100464,
+      "learning_rate": 1.1961013645224173e-05,
+      "loss": 0.0657,
+      "step": 1132
+    },
+    {
+      "epoch": 2.121894045944679,
+      "grad_norm": 0.36068305373191833,
+      "learning_rate": 1.1953216374269007e-05,
+      "loss": 0.0977,
+      "step": 1133
+    },
+    {
+      "epoch": 2.1237693389592125,
+      "grad_norm": 0.27126020193099976,
+      "learning_rate": 1.194541910331384e-05,
+      "loss": 0.0554,
+      "step": 1134
+    },
+    {
+      "epoch": 2.125644631973746,
+      "grad_norm": 0.3004317879676819,
+      "learning_rate": 1.1937621832358674e-05,
+      "loss": 0.0741,
+      "step": 1135
+    },
+    {
+      "epoch": 2.1275199249882792,
+      "grad_norm": 0.33724191784858704,
+      "learning_rate": 1.192982456140351e-05,
+      "loss": 0.1176,
+      "step": 1136
+    },
+    {
+      "epoch": 2.129395218002813,
+      "grad_norm": 0.34397128224372864,
+      "learning_rate": 1.1922027290448344e-05,
+      "loss": 0.0782,
+      "step": 1137
+    },
+    {
+      "epoch": 2.1312705110173464,
+      "grad_norm": 0.34579211473464966,
+      "learning_rate": 1.1914230019493178e-05,
+      "loss": 0.091,
+      "step": 1138
+    },
+    {
+      "epoch": 2.13314580403188,
+      "grad_norm": 0.27181315422058105,
+      "learning_rate": 1.1906432748538012e-05,
+      "loss": 0.0536,
+      "step": 1139
+    },
+    {
+      "epoch": 2.1350210970464136,
+      "grad_norm": 0.3022139370441437,
+      "learning_rate": 1.1898635477582846e-05,
+      "loss": 0.0658,
+      "step": 1140
+    },
+    {
+      "epoch": 2.1368963900609472,
+      "grad_norm": 0.38249850273132324,
+      "learning_rate": 1.1890838206627682e-05,
+      "loss": 0.0708,
+      "step": 1141
+    },
+    {
+      "epoch": 2.1387716830754804,
+      "grad_norm": 0.3176434636116028,
+      "learning_rate": 1.1883040935672517e-05,
+      "loss": 0.0811,
+      "step": 1142
+    },
+    {
+      "epoch": 2.140646976090014,
+      "grad_norm": 0.40170273184776306,
+      "learning_rate": 1.187524366471735e-05,
+      "loss": 0.1152,
+      "step": 1143
+    },
+    {
+      "epoch": 2.1425222691045476,
+      "grad_norm": 0.3950449824333191,
+      "learning_rate": 1.1867446393762183e-05,
+      "loss": 0.1099,
+      "step": 1144
+    },
+    {
+      "epoch": 2.144397562119081,
+      "grad_norm": 0.28338244557380676,
+      "learning_rate": 1.1859649122807017e-05,
+      "loss": 0.0821,
+      "step": 1145
+    },
+    {
+      "epoch": 2.1462728551336148,
+      "grad_norm": 0.6058647036552429,
+      "learning_rate": 1.1851851851851852e-05,
+      "loss": 0.0856,
+      "step": 1146
+    },
+    {
+      "epoch": 2.148148148148148,
+      "grad_norm": 0.28024426102638245,
+      "learning_rate": 1.1844054580896688e-05,
+      "loss": 0.0695,
+      "step": 1147
+    },
+    {
+      "epoch": 2.1500234411626815,
+      "grad_norm": 0.3102171719074249,
+      "learning_rate": 1.1836257309941522e-05,
+      "loss": 0.0646,
+      "step": 1148
+    },
+    {
+      "epoch": 2.151898734177215,
+      "grad_norm": 0.30384719371795654,
+      "learning_rate": 1.1828460038986356e-05,
+      "loss": 0.0779,
+      "step": 1149
+    },
+    {
+      "epoch": 2.1537740271917487,
+      "grad_norm": 0.31122729182243347,
+      "learning_rate": 1.182066276803119e-05,
+      "loss": 0.0802,
+      "step": 1150
+    },
+    {
+      "epoch": 2.1537740271917487,
+      "eval_loss": 0.07541442662477493,
+      "eval_runtime": 675.8929,
+      "eval_samples_per_second": 0.291,
+      "eval_steps_per_second": 0.291,
+      "step": 1150
+    },
+    {
+      "epoch": 2.1556493202062823,
+      "grad_norm": 0.39089834690093994,
+      "learning_rate": 1.1812865497076024e-05,
+      "loss": 0.1295,
+      "step": 1151
+    },
+    {
+      "epoch": 2.157524613220816,
+      "grad_norm": 0.286997526884079,
+      "learning_rate": 1.180506822612086e-05,
+      "loss": 0.0582,
+      "step": 1152
+    },
+    {
+      "epoch": 2.1593999062353495,
+      "grad_norm": 0.2970065772533417,
+      "learning_rate": 1.1797270955165694e-05,
+      "loss": 0.0862,
+      "step": 1153
+    },
+    {
+      "epoch": 2.1612751992498827,
+      "grad_norm": 0.2717902660369873,
+      "learning_rate": 1.1789473684210527e-05,
+      "loss": 0.055,
+      "step": 1154
+    },
+    {
+      "epoch": 2.1631504922644162,
+      "grad_norm": 0.3172938823699951,
+      "learning_rate": 1.1781676413255361e-05,
+      "loss": 0.0869,
+      "step": 1155
+    },
+    {
+      "epoch": 2.16502578527895,
+      "grad_norm": 0.42429736256599426,
+      "learning_rate": 1.1773879142300195e-05,
+      "loss": 0.0802,
+      "step": 1156
+    },
+    {
+      "epoch": 2.1669010782934834,
+      "grad_norm": 0.31430354714393616,
+      "learning_rate": 1.176608187134503e-05,
+      "loss": 0.0894,
+      "step": 1157
+    },
+    {
+      "epoch": 2.168776371308017,
+      "grad_norm": 0.33213984966278076,
+      "learning_rate": 1.1758284600389864e-05,
+      "loss": 0.1002,
+      "step": 1158
+    },
+    {
+      "epoch": 2.17065166432255,
+      "grad_norm": 0.35632964968681335,
+      "learning_rate": 1.17504873294347e-05,
+      "loss": 0.0793,
+      "step": 1159
+    },
+    {
+      "epoch": 2.172526957337084,
+      "grad_norm": 0.376068651676178,
+      "learning_rate": 1.1742690058479534e-05,
+      "loss": 0.0858,
+      "step": 1160
+    },
+    {
+      "epoch": 2.1744022503516174,
+      "grad_norm": 0.3594323694705963,
+      "learning_rate": 1.1734892787524368e-05,
+      "loss": 0.0589,
+      "step": 1161
+    },
+    {
+      "epoch": 2.176277543366151,
+      "grad_norm": 0.37725338339805603,
+      "learning_rate": 1.1727095516569202e-05,
+      "loss": 0.1046,
+      "step": 1162
+    },
+    {
+      "epoch": 2.1781528363806846,
+      "grad_norm": 0.28712713718414307,
+      "learning_rate": 1.1719298245614036e-05,
+      "loss": 0.0629,
+      "step": 1163
+    },
+    {
+      "epoch": 2.180028129395218,
+      "grad_norm": 0.31834840774536133,
+      "learning_rate": 1.1711500974658869e-05,
+      "loss": 0.0971,
+      "step": 1164
+    },
+    {
+      "epoch": 2.1819034224097513,
+      "grad_norm": 0.38184770941734314,
+      "learning_rate": 1.1703703703703703e-05,
+      "loss": 0.0971,
+      "step": 1165
+    },
+    {
+      "epoch": 2.183778715424285,
+      "grad_norm": 0.3888046145439148,
+      "learning_rate": 1.1695906432748539e-05,
+      "loss": 0.1172,
+      "step": 1166
+    },
+    {
+      "epoch": 2.1856540084388185,
+      "grad_norm": 0.3301868736743927,
+      "learning_rate": 1.1688109161793373e-05,
+      "loss": 0.0927,
+      "step": 1167
+    },
+    {
+      "epoch": 2.187529301453352,
+      "grad_norm": 0.3630843460559845,
+      "learning_rate": 1.1680311890838207e-05,
+      "loss": 0.0839,
+      "step": 1168
+    },
+    {
+      "epoch": 2.1894045944678857,
+      "grad_norm": 0.6889258623123169,
+      "learning_rate": 1.1672514619883041e-05,
+      "loss": 0.1217,
+      "step": 1169
+    },
+    {
+      "epoch": 2.1912798874824193,
+      "grad_norm": 0.6070849299430847,
+      "learning_rate": 1.1664717348927876e-05,
+      "loss": 0.0926,
+      "step": 1170
+    },
+    {
+      "epoch": 2.1931551804969525,
+      "grad_norm": 0.3465521037578583,
+      "learning_rate": 1.1656920077972711e-05,
+      "loss": 0.086,
+      "step": 1171
+    },
+    {
+      "epoch": 2.195030473511486,
+      "grad_norm": 0.3908824622631073,
+      "learning_rate": 1.1649122807017546e-05,
+      "loss": 0.1146,
+      "step": 1172
+    },
+    {
+      "epoch": 2.1969057665260197,
+      "grad_norm": 0.28279078006744385,
+      "learning_rate": 1.164132553606238e-05,
+      "loss": 0.072,
+      "step": 1173
+    },
+    {
+      "epoch": 2.1987810595405533,
+      "grad_norm": 0.31403693556785583,
+      "learning_rate": 1.1633528265107212e-05,
+      "loss": 0.0871,
+      "step": 1174
+    },
+    {
+      "epoch": 2.200656352555087,
+      "grad_norm": 0.33991819620132446,
+      "learning_rate": 1.1625730994152047e-05,
+      "loss": 0.0876,
+      "step": 1175
+    },
+    {
+      "epoch": 2.2025316455696204,
+      "grad_norm": 0.3142626881599426,
+      "learning_rate": 1.161793372319688e-05,
+      "loss": 0.0724,
+      "step": 1176
+    },
+    {
+      "epoch": 2.2044069385841536,
+      "grad_norm": 0.3219679594039917,
+      "learning_rate": 1.1610136452241717e-05,
+      "loss": 0.0659,
+      "step": 1177
+    },
+    {
+      "epoch": 2.206282231598687,
+      "grad_norm": 0.3814351260662079,
+      "learning_rate": 1.160233918128655e-05,
+      "loss": 0.0708,
+      "step": 1178
+    },
+    {
+      "epoch": 2.208157524613221,
+      "grad_norm": 0.380666583776474,
+      "learning_rate": 1.1594541910331385e-05,
+      "loss": 0.0952,
+      "step": 1179
+    },
+    {
+      "epoch": 2.2100328176277544,
+      "grad_norm": 0.2872433662414551,
+      "learning_rate": 1.158674463937622e-05,
+      "loss": 0.0794,
+      "step": 1180
+    },
+    {
+      "epoch": 2.211908110642288,
+      "grad_norm": 0.28936418890953064,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 0.0606,
+      "step": 1181
+    },
+    {
+      "epoch": 2.2137834036568216,
+      "grad_norm": 0.3666747808456421,
+      "learning_rate": 1.157115009746589e-05,
+      "loss": 0.0965,
+      "step": 1182
+    },
+    {
+      "epoch": 2.2156586966713547,
+      "grad_norm": 0.3257901072502136,
+      "learning_rate": 1.1563352826510723e-05,
+      "loss": 0.0796,
+      "step": 1183
+    },
+    {
+      "epoch": 2.2175339896858883,
+      "grad_norm": 0.47882843017578125,
+      "learning_rate": 1.1555555555555556e-05,
+      "loss": 0.0816,
+      "step": 1184
+    },
+    {
+      "epoch": 2.219409282700422,
+      "grad_norm": 0.2858003079891205,
+      "learning_rate": 1.154775828460039e-05,
+      "loss": 0.069,
+      "step": 1185
+    },
+    {
+      "epoch": 2.2212845757149555,
+      "grad_norm": 0.31669747829437256,
+      "learning_rate": 1.1539961013645224e-05,
+      "loss": 0.0842,
+      "step": 1186
+    },
+    {
+      "epoch": 2.223159868729489,
+      "grad_norm": 0.3197777271270752,
+      "learning_rate": 1.1532163742690059e-05,
+      "loss": 0.099,
+      "step": 1187
+    },
+    {
+      "epoch": 2.2250351617440227,
+      "grad_norm": 0.34864693880081177,
+      "learning_rate": 1.1524366471734893e-05,
+      "loss": 0.0849,
+      "step": 1188
+    },
+    {
+      "epoch": 2.226910454758556,
+      "grad_norm": 0.7603491544723511,
+      "learning_rate": 1.1516569200779729e-05,
+      "loss": 0.0865,
+      "step": 1189
+    },
+    {
+      "epoch": 2.2287857477730895,
+      "grad_norm": 0.3353845775127411,
+      "learning_rate": 1.1508771929824563e-05,
+      "loss": 0.0724,
+      "step": 1190
+    },
+    {
+      "epoch": 2.230661040787623,
+      "grad_norm": 0.27294620871543884,
+      "learning_rate": 1.1500974658869397e-05,
+      "loss": 0.0718,
+      "step": 1191
+    },
+    {
+      "epoch": 2.2325363338021567,
+      "grad_norm": 0.30225276947021484,
+      "learning_rate": 1.1493177387914231e-05,
+      "loss": 0.0752,
+      "step": 1192
+    },
+    {
+      "epoch": 2.2344116268166903,
+      "grad_norm": 0.30490773916244507,
+      "learning_rate": 1.1485380116959065e-05,
+      "loss": 0.0653,
+      "step": 1193
+    },
+    {
+      "epoch": 2.2362869198312234,
+      "grad_norm": 0.306281715631485,
+      "learning_rate": 1.1477582846003898e-05,
+      "loss": 0.0747,
+      "step": 1194
+    },
+    {
+      "epoch": 2.238162212845757,
+      "grad_norm": 0.3963259756565094,
+      "learning_rate": 1.1469785575048734e-05,
+      "loss": 0.1006,
+      "step": 1195
+    },
+    {
+      "epoch": 2.2400375058602906,
+      "grad_norm": 0.3330558240413666,
+      "learning_rate": 1.1461988304093568e-05,
+      "loss": 0.0925,
+      "step": 1196
+    },
+    {
+      "epoch": 2.241912798874824,
+      "grad_norm": 0.27601566910743713,
+      "learning_rate": 1.1454191033138402e-05,
+      "loss": 0.0686,
+      "step": 1197
+    },
+    {
+      "epoch": 2.243788091889358,
+      "grad_norm": 0.412579208612442,
+      "learning_rate": 1.1446393762183236e-05,
+      "loss": 0.0789,
+      "step": 1198
+    },
+    {
+      "epoch": 2.2456633849038914,
+      "grad_norm": 0.39170923829078674,
+      "learning_rate": 1.143859649122807e-05,
+      "loss": 0.095,
+      "step": 1199
+    },
+    {
+      "epoch": 2.247538677918425,
+      "grad_norm": 0.3268628716468811,
+      "learning_rate": 1.1430799220272906e-05,
+      "loss": 0.0772,
+      "step": 1200
+    },
+    {
+      "epoch": 2.247538677918425,
+      "eval_loss": 0.07522810995578766,
+      "eval_runtime": 675.3854,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 1200
+    },
+    {
+      "epoch": 2.249413970932958,
+      "grad_norm": 0.34962981939315796,
+      "learning_rate": 1.142300194931774e-05,
+      "loss": 0.0791,
+      "step": 1201
+    },
+    {
+      "epoch": 2.2512892639474917,
+      "grad_norm": 0.3399597704410553,
+      "learning_rate": 1.1415204678362575e-05,
+      "loss": 0.0807,
+      "step": 1202
+    },
+    {
+      "epoch": 2.2531645569620253,
+      "grad_norm": 0.27295050024986267,
+      "learning_rate": 1.1407407407407409e-05,
+      "loss": 0.0607,
+      "step": 1203
+    },
+    {
+      "epoch": 2.255039849976559,
+      "grad_norm": 0.32791128754615784,
+      "learning_rate": 1.1399610136452241e-05,
+      "loss": 0.0548,
+      "step": 1204
+    },
+    {
+      "epoch": 2.2569151429910925,
+      "grad_norm": 0.35572582483291626,
+      "learning_rate": 1.1391812865497076e-05,
+      "loss": 0.0806,
+      "step": 1205
+    },
+    {
+      "epoch": 2.2587904360056257,
+      "grad_norm": 0.3451082408428192,
+      "learning_rate": 1.138401559454191e-05,
+      "loss": 0.0599,
+      "step": 1206
+    },
+    {
+      "epoch": 2.2606657290201593,
+      "grad_norm": 0.25782617926597595,
+      "learning_rate": 1.1376218323586746e-05,
+      "loss": 0.0584,
+      "step": 1207
+    },
+    {
+      "epoch": 2.262541022034693,
+      "grad_norm": 0.2698267996311188,
+      "learning_rate": 1.136842105263158e-05,
+      "loss": 0.067,
+      "step": 1208
+    },
+    {
+      "epoch": 2.2644163150492265,
+      "grad_norm": 0.2824949026107788,
+      "learning_rate": 1.1360623781676414e-05,
+      "loss": 0.0325,
+      "step": 1209
+    },
+    {
+      "epoch": 2.26629160806376,
+      "grad_norm": 0.30985549092292786,
+      "learning_rate": 1.1352826510721248e-05,
+      "loss": 0.0591,
+      "step": 1210
+    },
+    {
+      "epoch": 2.2681669010782937,
+      "grad_norm": 0.3595530688762665,
+      "learning_rate": 1.1345029239766083e-05,
+      "loss": 0.0821,
+      "step": 1211
+    },
+    {
+      "epoch": 2.270042194092827,
+      "grad_norm": 0.3911292552947998,
+      "learning_rate": 1.1337231968810918e-05,
+      "loss": 0.0942,
+      "step": 1212
+    },
+    {
+      "epoch": 2.2719174871073604,
+      "grad_norm": 0.31286779046058655,
+      "learning_rate": 1.1329434697855753e-05,
+      "loss": 0.0703,
+      "step": 1213
+    },
+    {
+      "epoch": 2.273792780121894,
+      "grad_norm": 0.6129016876220703,
+      "learning_rate": 1.1321637426900585e-05,
+      "loss": 0.0904,
+      "step": 1214
+    },
+    {
+      "epoch": 2.2756680731364276,
+      "grad_norm": 0.35265544056892395,
+      "learning_rate": 1.131384015594542e-05,
+      "loss": 0.102,
+      "step": 1215
+    },
+    {
+      "epoch": 2.277543366150961,
+      "grad_norm": 0.49612244963645935,
+      "learning_rate": 1.1306042884990253e-05,
+      "loss": 0.0851,
+      "step": 1216
+    },
+    {
+      "epoch": 2.279418659165495,
+      "grad_norm": 2.8707265853881836,
+      "learning_rate": 1.1298245614035088e-05,
+      "loss": 0.0575,
+      "step": 1217
+    },
+    {
+      "epoch": 2.281293952180028,
+      "grad_norm": 0.676102340221405,
+      "learning_rate": 1.1290448343079924e-05,
+      "loss": 0.0946,
+      "step": 1218
+    },
+    {
+      "epoch": 2.2831692451945615,
+      "grad_norm": 0.3254333436489105,
+      "learning_rate": 1.1282651072124758e-05,
+      "loss": 0.0722,
+      "step": 1219
+    },
+    {
+      "epoch": 2.285044538209095,
+      "grad_norm": 0.2682390809059143,
+      "learning_rate": 1.1274853801169592e-05,
+      "loss": 0.0674,
+      "step": 1220
+    },
+    {
+      "epoch": 2.2869198312236287,
+      "grad_norm": 0.2639862298965454,
+      "learning_rate": 1.1267056530214426e-05,
+      "loss": 0.0515,
+      "step": 1221
+    },
+    {
+      "epoch": 2.2887951242381623,
+      "grad_norm": 0.30463695526123047,
+      "learning_rate": 1.125925925925926e-05,
+      "loss": 0.0683,
+      "step": 1222
+    },
+    {
+      "epoch": 2.290670417252696,
+      "grad_norm": 0.2600836753845215,
+      "learning_rate": 1.1251461988304096e-05,
+      "loss": 0.0588,
+      "step": 1223
+    },
+    {
+      "epoch": 2.292545710267229,
+      "grad_norm": 0.2697446644306183,
+      "learning_rate": 1.1243664717348927e-05,
+      "loss": 0.0519,
+      "step": 1224
+    },
+    {
+      "epoch": 2.2944210032817627,
+      "grad_norm": 0.34204694628715515,
+      "learning_rate": 1.1235867446393763e-05,
+      "loss": 0.0839,
+      "step": 1225
+    },
+    {
+      "epoch": 2.2962962962962963,
+      "grad_norm": 0.30241382122039795,
+      "learning_rate": 1.1228070175438597e-05,
+      "loss": 0.0793,
+      "step": 1226
+    },
+    {
+      "epoch": 2.29817158931083,
+      "grad_norm": 0.36129674315452576,
+      "learning_rate": 1.1220272904483431e-05,
+      "loss": 0.0572,
+      "step": 1227
+    },
+    {
+      "epoch": 2.3000468823253635,
+      "grad_norm": 0.3861963450908661,
+      "learning_rate": 1.1212475633528265e-05,
+      "loss": 0.1096,
+      "step": 1228
+    },
+    {
+      "epoch": 2.3019221753398966,
+      "grad_norm": 0.2883913516998291,
+      "learning_rate": 1.12046783625731e-05,
+      "loss": 0.0614,
+      "step": 1229
+    },
+    {
+      "epoch": 2.3037974683544302,
+      "grad_norm": 0.3693699538707733,
+      "learning_rate": 1.1196881091617936e-05,
+      "loss": 0.0957,
+      "step": 1230
+    },
+    {
+      "epoch": 2.305672761368964,
+      "grad_norm": 0.35850393772125244,
+      "learning_rate": 1.118908382066277e-05,
+      "loss": 0.0849,
+      "step": 1231
+    },
+    {
+      "epoch": 2.3075480543834974,
+      "grad_norm": 0.35511448979377747,
+      "learning_rate": 1.1181286549707604e-05,
+      "loss": 0.1032,
+      "step": 1232
+    },
+    {
+      "epoch": 2.309423347398031,
+      "grad_norm": 0.5250701308250427,
+      "learning_rate": 1.1173489278752438e-05,
+      "loss": 0.1174,
+      "step": 1233
+    },
+    {
+      "epoch": 2.3112986404125646,
+      "grad_norm": 0.28343549370765686,
+      "learning_rate": 1.116569200779727e-05,
+      "loss": 0.0702,
+      "step": 1234
+    },
+    {
+      "epoch": 2.313173933427098,
+      "grad_norm": 0.3174719512462616,
+      "learning_rate": 1.1157894736842105e-05,
+      "loss": 0.0628,
+      "step": 1235
+    },
+    {
+      "epoch": 2.3150492264416314,
+      "grad_norm": 0.4021863341331482,
+      "learning_rate": 1.115009746588694e-05,
+      "loss": 0.0856,
+      "step": 1236
+    },
+    {
+      "epoch": 2.316924519456165,
+      "grad_norm": 0.4704308807849884,
+      "learning_rate": 1.1142300194931775e-05,
+      "loss": 0.0977,
+      "step": 1237
+    },
+    {
+      "epoch": 2.3187998124706986,
+      "grad_norm": 0.319234162569046,
+      "learning_rate": 1.1134502923976609e-05,
+      "loss": 0.0647,
+      "step": 1238
+    },
+    {
+      "epoch": 2.320675105485232,
+      "grad_norm": 0.3543676435947418,
+      "learning_rate": 1.1126705653021443e-05,
+      "loss": 0.0791,
+      "step": 1239
+    },
+    {
+      "epoch": 2.3225503984997657,
+      "grad_norm": 0.32436403632164,
+      "learning_rate": 1.1118908382066277e-05,
+      "loss": 0.0613,
+      "step": 1240
+    },
+    {
+      "epoch": 2.324425691514299,
+      "grad_norm": 0.33297768235206604,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 0.0684,
+      "step": 1241
+    },
+    {
+      "epoch": 2.3263009845288325,
+      "grad_norm": 0.46087008714675903,
+      "learning_rate": 1.1103313840155948e-05,
+      "loss": 0.0744,
+      "step": 1242
+    },
+    {
+      "epoch": 2.328176277543366,
+      "grad_norm": 0.3725678622722626,
+      "learning_rate": 1.1095516569200782e-05,
+      "loss": 0.0878,
+      "step": 1243
+    },
+    {
+      "epoch": 2.3300515705578997,
+      "grad_norm": 0.6072844862937927,
+      "learning_rate": 1.1087719298245614e-05,
+      "loss": 0.1077,
+      "step": 1244
+    },
+    {
+      "epoch": 2.3319268635724333,
+      "grad_norm": 0.32860544323921204,
+      "learning_rate": 1.1079922027290448e-05,
+      "loss": 0.0694,
+      "step": 1245
+    },
+    {
+      "epoch": 2.333802156586967,
+      "grad_norm": 0.2839890122413635,
+      "learning_rate": 1.1072124756335283e-05,
+      "loss": 0.0402,
+      "step": 1246
+    },
+    {
+      "epoch": 2.3356774496015005,
+      "grad_norm": 0.4085495173931122,
+      "learning_rate": 1.1064327485380117e-05,
+      "loss": 0.0641,
+      "step": 1247
+    },
+    {
+      "epoch": 2.3375527426160336,
+      "grad_norm": 0.3415165841579437,
+      "learning_rate": 1.1056530214424953e-05,
+      "loss": 0.0643,
+      "step": 1248
+    },
+    {
+      "epoch": 2.3394280356305672,
+      "grad_norm": 0.34704262018203735,
+      "learning_rate": 1.1048732943469787e-05,
+      "loss": 0.0792,
+      "step": 1249
+    },
+    {
+      "epoch": 2.341303328645101,
+      "grad_norm": 0.2980154752731323,
+      "learning_rate": 1.1040935672514621e-05,
+      "loss": 0.0673,
+      "step": 1250
+    },
+    {
+      "epoch": 2.341303328645101,
+      "eval_loss": 0.07502703368663788,
+      "eval_runtime": 675.207,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 1250
+    },
+    {
+      "epoch": 2.3431786216596344,
+      "grad_norm": 0.27935025095939636,
+      "learning_rate": 1.1033138401559455e-05,
+      "loss": 0.0512,
+      "step": 1251
+    },
+    {
+      "epoch": 2.345053914674168,
+      "grad_norm": 0.42243000864982605,
+      "learning_rate": 1.102534113060429e-05,
+      "loss": 0.1033,
+      "step": 1252
+    },
+    {
+      "epoch": 2.346929207688701,
+      "grad_norm": 0.36166539788246155,
+      "learning_rate": 1.1017543859649125e-05,
+      "loss": 0.0787,
+      "step": 1253
+    },
+    {
+      "epoch": 2.3488045007032348,
+      "grad_norm": 1.874434232711792,
+      "learning_rate": 1.1009746588693956e-05,
+      "loss": 0.1049,
+      "step": 1254
+    },
+    {
+      "epoch": 2.3506797937177684,
+      "grad_norm": 0.30916956067085266,
+      "learning_rate": 1.1001949317738792e-05,
+      "loss": 0.0895,
+      "step": 1255
+    },
+    {
+      "epoch": 2.352555086732302,
+      "grad_norm": 0.37780481576919556,
+      "learning_rate": 1.0994152046783626e-05,
+      "loss": 0.089,
+      "step": 1256
+    },
+    {
+      "epoch": 2.3544303797468356,
+      "grad_norm": 0.3470967411994934,
+      "learning_rate": 1.098635477582846e-05,
+      "loss": 0.0827,
+      "step": 1257
+    },
+    {
+      "epoch": 2.356305672761369,
+      "grad_norm": 0.28173500299453735,
+      "learning_rate": 1.0978557504873295e-05,
+      "loss": 0.0831,
+      "step": 1258
+    },
+    {
+      "epoch": 2.3581809657759023,
+      "grad_norm": 0.32230475544929504,
+      "learning_rate": 1.0970760233918129e-05,
+      "loss": 0.0706,
+      "step": 1259
+    },
+    {
+      "epoch": 2.360056258790436,
+      "grad_norm": 0.2897712290287018,
+      "learning_rate": 1.0962962962962965e-05,
+      "loss": 0.0614,
+      "step": 1260
+    },
+    {
+      "epoch": 2.3619315518049695,
+      "grad_norm": 0.46456751227378845,
+      "learning_rate": 1.0955165692007799e-05,
+      "loss": 0.1074,
+      "step": 1261
+    },
+    {
+      "epoch": 2.363806844819503,
+      "grad_norm": 0.35259947180747986,
+      "learning_rate": 1.0947368421052633e-05,
+      "loss": 0.09,
+      "step": 1262
+    },
+    {
+      "epoch": 2.3656821378340367,
+      "grad_norm": 0.4690414071083069,
+      "learning_rate": 1.0939571150097467e-05,
+      "loss": 0.0667,
+      "step": 1263
+    },
+    {
+      "epoch": 2.36755743084857,
+      "grad_norm": 0.3024637997150421,
+      "learning_rate": 1.09317738791423e-05,
+      "loss": 0.0723,
+      "step": 1264
+    },
+    {
+      "epoch": 2.3694327238631034,
+      "grad_norm": 0.2506920397281647,
+      "learning_rate": 1.0923976608187134e-05,
+      "loss": 0.0584,
+      "step": 1265
+    },
+    {
+      "epoch": 2.371308016877637,
+      "grad_norm": 0.3608229160308838,
+      "learning_rate": 1.091617933723197e-05,
+      "loss": 0.0659,
+      "step": 1266
+    },
+    {
+      "epoch": 2.3731833098921706,
+      "grad_norm": 0.3218965232372284,
+      "learning_rate": 1.0908382066276804e-05,
+      "loss": 0.0803,
+      "step": 1267
+    },
+    {
+      "epoch": 2.3750586029067042,
+      "grad_norm": 0.8374189734458923,
+      "learning_rate": 1.0900584795321638e-05,
+      "loss": 0.0654,
+      "step": 1268
+    },
+    {
+      "epoch": 2.376933895921238,
+      "grad_norm": 0.3754996955394745,
+      "learning_rate": 1.0892787524366472e-05,
+      "loss": 0.0868,
+      "step": 1269
+    },
+    {
+      "epoch": 2.3788091889357714,
+      "grad_norm": 0.928810179233551,
+      "learning_rate": 1.0884990253411307e-05,
+      "loss": 0.1044,
+      "step": 1270
+    },
+    {
+      "epoch": 2.3806844819503046,
+      "grad_norm": 0.3189548850059509,
+      "learning_rate": 1.0877192982456142e-05,
+      "loss": 0.0773,
+      "step": 1271
+    },
+    {
+      "epoch": 2.382559774964838,
+      "grad_norm": 0.3129134178161621,
+      "learning_rate": 1.0869395711500977e-05,
+      "loss": 0.0769,
+      "step": 1272
+    },
+    {
+      "epoch": 2.3844350679793718,
+      "grad_norm": 0.2928052544593811,
+      "learning_rate": 1.086159844054581e-05,
+      "loss": 0.0783,
+      "step": 1273
+    },
+    {
+      "epoch": 2.3863103609939054,
+      "grad_norm": 0.380834698677063,
+      "learning_rate": 1.0853801169590643e-05,
+      "loss": 0.0734,
+      "step": 1274
+    },
+    {
+      "epoch": 2.388185654008439,
+      "grad_norm": 0.28873157501220703,
+      "learning_rate": 1.0846003898635478e-05,
+      "loss": 0.0697,
+      "step": 1275
+    },
+    {
+      "epoch": 2.390060947022972,
+      "grad_norm": 0.2910199463367462,
+      "learning_rate": 1.0838206627680312e-05,
+      "loss": 0.0691,
+      "step": 1276
+    },
+    {
+      "epoch": 2.3919362400375057,
+      "grad_norm": 0.3059178292751312,
+      "learning_rate": 1.0830409356725146e-05,
+      "loss": 0.064,
+      "step": 1277
+    },
+    {
+      "epoch": 2.3938115330520393,
+      "grad_norm": 0.33945995569229126,
+      "learning_rate": 1.0822612085769982e-05,
+      "loss": 0.0589,
+      "step": 1278
+    },
+    {
+      "epoch": 2.395686826066573,
+      "grad_norm": 0.35901668667793274,
+      "learning_rate": 1.0814814814814816e-05,
+      "loss": 0.0798,
+      "step": 1279
+    },
+    {
+      "epoch": 2.3975621190811065,
+      "grad_norm": 0.2913598418235779,
+      "learning_rate": 1.080701754385965e-05,
+      "loss": 0.0609,
+      "step": 1280
+    },
+    {
+      "epoch": 2.39943741209564,
+      "grad_norm": 0.30235838890075684,
+      "learning_rate": 1.0799220272904484e-05,
+      "loss": 0.0696,
+      "step": 1281
+    },
+    {
+      "epoch": 2.4013127051101737,
+      "grad_norm": 0.29583993554115295,
+      "learning_rate": 1.0791423001949319e-05,
+      "loss": 0.0813,
+      "step": 1282
+    },
+    {
+      "epoch": 2.403187998124707,
+      "grad_norm": 0.34998810291290283,
+      "learning_rate": 1.0783625730994154e-05,
+      "loss": 0.0782,
+      "step": 1283
+    },
+    {
+      "epoch": 2.4050632911392404,
+      "grad_norm": 0.3048896789550781,
+      "learning_rate": 1.0775828460038987e-05,
+      "loss": 0.0579,
+      "step": 1284
+    },
+    {
+      "epoch": 2.406938584153774,
+      "grad_norm": 0.22718416154384613,
+      "learning_rate": 1.0768031189083821e-05,
+      "loss": 0.0662,
+      "step": 1285
+    },
+    {
+      "epoch": 2.4088138771683076,
+      "grad_norm": 0.2834322154521942,
+      "learning_rate": 1.0760233918128655e-05,
+      "loss": 0.0615,
+      "step": 1286
+    },
+    {
+      "epoch": 2.4106891701828412,
+      "grad_norm": 0.36230021715164185,
+      "learning_rate": 1.075243664717349e-05,
+      "loss": 0.0743,
+      "step": 1287
+    },
+    {
+      "epoch": 2.4125644631973744,
+      "grad_norm": 0.2996331453323364,
+      "learning_rate": 1.0744639376218324e-05,
+      "loss": 0.0763,
+      "step": 1288
+    },
+    {
+      "epoch": 2.414439756211908,
+      "grad_norm": 0.3829980790615082,
+      "learning_rate": 1.073684210526316e-05,
+      "loss": 0.06,
+      "step": 1289
+    },
+    {
+      "epoch": 2.4163150492264416,
+      "grad_norm": 0.36401882767677307,
+      "learning_rate": 1.0729044834307994e-05,
+      "loss": 0.091,
+      "step": 1290
+    },
+    {
+      "epoch": 2.418190342240975,
+      "grad_norm": 0.38041242957115173,
+      "learning_rate": 1.0721247563352828e-05,
+      "loss": 0.073,
+      "step": 1291
+    },
+    {
+      "epoch": 2.4200656352555088,
+      "grad_norm": 0.4183158576488495,
+      "learning_rate": 1.0713450292397662e-05,
+      "loss": 0.0889,
+      "step": 1292
+    },
+    {
+      "epoch": 2.4219409282700424,
+      "grad_norm": 0.30840641260147095,
+      "learning_rate": 1.0705653021442496e-05,
+      "loss": 0.0878,
+      "step": 1293
+    },
+    {
+      "epoch": 2.4238162212845755,
+      "grad_norm": 1.0156400203704834,
+      "learning_rate": 1.0697855750487329e-05,
+      "loss": 0.0942,
+      "step": 1294
+    },
+    {
+      "epoch": 2.425691514299109,
+      "grad_norm": 0.26748737692832947,
+      "learning_rate": 1.0690058479532163e-05,
+      "loss": 0.0535,
+      "step": 1295
+    },
+    {
+      "epoch": 2.4275668073136427,
+      "grad_norm": 0.3611546456813812,
+      "learning_rate": 1.0682261208576999e-05,
+      "loss": 0.0684,
+      "step": 1296
+    },
+    {
+      "epoch": 2.4294421003281763,
+      "grad_norm": 0.4616575539112091,
+      "learning_rate": 1.0674463937621833e-05,
+      "loss": 0.0916,
+      "step": 1297
+    },
+    {
+      "epoch": 2.43131739334271,
+      "grad_norm": 0.37064328789711,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.0871,
+      "step": 1298
+    },
+    {
+      "epoch": 2.4331926863572435,
+      "grad_norm": 0.30871546268463135,
+      "learning_rate": 1.0658869395711502e-05,
+      "loss": 0.0725,
+      "step": 1299
+    },
+    {
+      "epoch": 2.4350679793717767,
+      "grad_norm": 0.45006898045539856,
+      "learning_rate": 1.0651072124756336e-05,
+      "loss": 0.1114,
+      "step": 1300
+    },
+    {
+      "epoch": 2.4350679793717767,
+      "eval_loss": 0.07475950568914413,
+      "eval_runtime": 674.4121,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 1300
+    },
+    {
+      "epoch": 2.4369432723863103,
+      "grad_norm": 0.410637229681015,
+      "learning_rate": 1.0643274853801172e-05,
+      "loss": 0.0854,
+      "step": 1301
+    },
+    {
+      "epoch": 2.438818565400844,
+      "grad_norm": 0.40461596846580505,
+      "learning_rate": 1.0635477582846006e-05,
+      "loss": 0.1023,
+      "step": 1302
+    },
+    {
+      "epoch": 2.4406938584153774,
+      "grad_norm": 0.47095608711242676,
+      "learning_rate": 1.062768031189084e-05,
+      "loss": 0.097,
+      "step": 1303
+    },
+    {
+      "epoch": 2.442569151429911,
+      "grad_norm": 0.3717558979988098,
+      "learning_rate": 1.0619883040935672e-05,
+      "loss": 0.076,
+      "step": 1304
+    },
+    {
+      "epoch": 2.4444444444444446,
+      "grad_norm": 0.3064993619918823,
+      "learning_rate": 1.0612085769980507e-05,
+      "loss": 0.06,
+      "step": 1305
+    },
+    {
+      "epoch": 2.446319737458978,
+      "grad_norm": 0.380657821893692,
+      "learning_rate": 1.060428849902534e-05,
+      "loss": 0.0997,
+      "step": 1306
+    },
+    {
+      "epoch": 2.4481950304735114,
+      "grad_norm": 0.3160291612148285,
+      "learning_rate": 1.0596491228070177e-05,
+      "loss": 0.0548,
+      "step": 1307
+    },
+    {
+      "epoch": 2.450070323488045,
+      "grad_norm": 0.377288818359375,
+      "learning_rate": 1.0588693957115011e-05,
+      "loss": 0.0803,
+      "step": 1308
+    },
+    {
+      "epoch": 2.4519456165025786,
+      "grad_norm": 0.27472472190856934,
+      "learning_rate": 1.0580896686159845e-05,
+      "loss": 0.0477,
+      "step": 1309
+    },
+    {
+      "epoch": 2.453820909517112,
+      "grad_norm": 0.3660586476325989,
+      "learning_rate": 1.057309941520468e-05,
+      "loss": 0.075,
+      "step": 1310
+    },
+    {
+      "epoch": 2.4556962025316453,
+      "grad_norm": 1.1128591299057007,
+      "learning_rate": 1.0565302144249513e-05,
+      "loss": 0.1068,
+      "step": 1311
+    },
+    {
+      "epoch": 2.457571495546179,
+      "grad_norm": 0.3397810757160187,
+      "learning_rate": 1.055750487329435e-05,
+      "loss": 0.0689,
+      "step": 1312
+    },
+    {
+      "epoch": 2.4594467885607125,
+      "grad_norm": 0.28418809175491333,
+      "learning_rate": 1.0549707602339184e-05,
+      "loss": 0.078,
+      "step": 1313
+    },
+    {
+      "epoch": 2.461322081575246,
+      "grad_norm": 0.30030253529548645,
+      "learning_rate": 1.0541910331384016e-05,
+      "loss": 0.0504,
+      "step": 1314
+    },
+    {
+      "epoch": 2.4631973745897797,
+      "grad_norm": 0.306923508644104,
+      "learning_rate": 1.053411306042885e-05,
+      "loss": 0.0516,
+      "step": 1315
+    },
+    {
+      "epoch": 2.4650726676043133,
+      "grad_norm": 0.2753802239894867,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.0733,
+      "step": 1316
+    },
+    {
+      "epoch": 2.466947960618847,
+      "grad_norm": 0.4096115231513977,
+      "learning_rate": 1.0518518518518519e-05,
+      "loss": 0.1417,
+      "step": 1317
+    },
+    {
+      "epoch": 2.46882325363338,
+      "grad_norm": 0.45057594776153564,
+      "learning_rate": 1.0510721247563353e-05,
+      "loss": 0.082,
+      "step": 1318
+    },
+    {
+      "epoch": 2.4706985466479137,
+      "grad_norm": 0.2660753130912781,
+      "learning_rate": 1.0502923976608189e-05,
+      "loss": 0.0762,
+      "step": 1319
+    },
+    {
+      "epoch": 2.4725738396624473,
+      "grad_norm": 0.3073916435241699,
+      "learning_rate": 1.0495126705653023e-05,
+      "loss": 0.0603,
+      "step": 1320
+    },
+    {
+      "epoch": 2.474449132676981,
+      "grad_norm": 0.45714837312698364,
+      "learning_rate": 1.0487329434697857e-05,
+      "loss": 0.0975,
+      "step": 1321
+    },
+    {
+      "epoch": 2.4763244256915145,
+      "grad_norm": 0.34037506580352783,
+      "learning_rate": 1.0479532163742691e-05,
+      "loss": 0.0837,
+      "step": 1322
+    },
+    {
+      "epoch": 2.4781997187060476,
+      "grad_norm": 0.41261813044548035,
+      "learning_rate": 1.0471734892787525e-05,
+      "loss": 0.1105,
+      "step": 1323
+    },
+    {
+      "epoch": 2.480075011720581,
+      "grad_norm": 0.3210998475551605,
+      "learning_rate": 1.0463937621832358e-05,
+      "loss": 0.0582,
+      "step": 1324
+    },
+    {
+      "epoch": 2.481950304735115,
+      "grad_norm": 0.37079358100891113,
+      "learning_rate": 1.0456140350877194e-05,
+      "loss": 0.0842,
+      "step": 1325
+    },
+    {
+      "epoch": 2.4838255977496484,
+      "grad_norm": 0.3284546434879303,
+      "learning_rate": 1.0448343079922028e-05,
+      "loss": 0.0634,
+      "step": 1326
+    },
+    {
+      "epoch": 2.485700890764182,
+      "grad_norm": 0.28334537148475647,
+      "learning_rate": 1.0440545808966862e-05,
+      "loss": 0.0451,
+      "step": 1327
+    },
+    {
+      "epoch": 2.4875761837787156,
+      "grad_norm": 0.3325977027416229,
+      "learning_rate": 1.0432748538011696e-05,
+      "loss": 0.0798,
+      "step": 1328
+    },
+    {
+      "epoch": 2.489451476793249,
+      "grad_norm": 0.4519757330417633,
+      "learning_rate": 1.042495126705653e-05,
+      "loss": 0.1076,
+      "step": 1329
+    },
+    {
+      "epoch": 2.4913267698077823,
+      "grad_norm": 0.313209593296051,
+      "learning_rate": 1.0417153996101367e-05,
+      "loss": 0.068,
+      "step": 1330
+    },
+    {
+      "epoch": 2.493202062822316,
+      "grad_norm": 0.35849788784980774,
+      "learning_rate": 1.04093567251462e-05,
+      "loss": 0.1017,
+      "step": 1331
+    },
+    {
+      "epoch": 2.4950773558368495,
+      "grad_norm": 0.34800082445144653,
+      "learning_rate": 1.0401559454191035e-05,
+      "loss": 0.0732,
+      "step": 1332
+    },
+    {
+      "epoch": 2.496952648851383,
+      "grad_norm": 0.41697457432746887,
+      "learning_rate": 1.0393762183235869e-05,
+      "loss": 0.0898,
+      "step": 1333
+    },
+    {
+      "epoch": 2.4988279418659167,
+      "grad_norm": 0.9164847731590271,
+      "learning_rate": 1.0385964912280702e-05,
+      "loss": 0.0843,
+      "step": 1334
+    },
+    {
+      "epoch": 2.50070323488045,
+      "grad_norm": 0.3565974533557892,
+      "learning_rate": 1.0378167641325536e-05,
+      "loss": 0.0821,
+      "step": 1335
+    },
+    {
+      "epoch": 2.5025785278949835,
+      "grad_norm": 0.37837761640548706,
+      "learning_rate": 1.037037037037037e-05,
+      "loss": 0.0807,
+      "step": 1336
+    },
+    {
+      "epoch": 2.504453820909517,
+      "grad_norm": 0.30958202481269836,
+      "learning_rate": 1.0362573099415206e-05,
+      "loss": 0.0623,
+      "step": 1337
+    },
+    {
+      "epoch": 2.5063291139240507,
+      "grad_norm": 0.29366958141326904,
+      "learning_rate": 1.035477582846004e-05,
+      "loss": 0.0592,
+      "step": 1338
+    },
+    {
+      "epoch": 2.5082044069385843,
+      "grad_norm": 0.3304692506790161,
+      "learning_rate": 1.0346978557504874e-05,
+      "loss": 0.0606,
+      "step": 1339
+    },
+    {
+      "epoch": 2.510079699953118,
+      "grad_norm": 0.3314734697341919,
+      "learning_rate": 1.0339181286549708e-05,
+      "loss": 0.078,
+      "step": 1340
+    },
+    {
+      "epoch": 2.5119549929676515,
+      "grad_norm": 0.2976376712322235,
+      "learning_rate": 1.0331384015594543e-05,
+      "loss": 0.0737,
+      "step": 1341
+    },
+    {
+      "epoch": 2.5138302859821846,
+      "grad_norm": 0.3934020400047302,
+      "learning_rate": 1.0323586744639378e-05,
+      "loss": 0.0567,
+      "step": 1342
+    },
+    {
+      "epoch": 2.515705578996718,
+      "grad_norm": 0.35475224256515503,
+      "learning_rate": 1.0315789473684213e-05,
+      "loss": 0.0785,
+      "step": 1343
+    },
+    {
+      "epoch": 2.517580872011252,
+      "grad_norm": 0.2962830364704132,
+      "learning_rate": 1.0307992202729045e-05,
+      "loss": 0.0693,
+      "step": 1344
+    },
+    {
+      "epoch": 2.5194561650257854,
+      "grad_norm": 0.4114619195461273,
+      "learning_rate": 1.030019493177388e-05,
+      "loss": 0.1023,
+      "step": 1345
+    },
+    {
+      "epoch": 2.5213314580403186,
+      "grad_norm": 0.3815486431121826,
+      "learning_rate": 1.0292397660818714e-05,
+      "loss": 0.079,
+      "step": 1346
+    },
+    {
+      "epoch": 2.523206751054852,
+      "grad_norm": 0.4123891592025757,
+      "learning_rate": 1.0284600389863548e-05,
+      "loss": 0.1021,
+      "step": 1347
+    },
+    {
+      "epoch": 2.5250820440693857,
+      "grad_norm": 0.3692156672477722,
+      "learning_rate": 1.0276803118908382e-05,
+      "loss": 0.0983,
+      "step": 1348
+    },
+    {
+      "epoch": 2.5269573370839193,
+      "grad_norm": 0.3262293338775635,
+      "learning_rate": 1.0269005847953218e-05,
+      "loss": 0.096,
+      "step": 1349
+    },
+    {
+      "epoch": 2.528832630098453,
+      "grad_norm": 1.2034567594528198,
+      "learning_rate": 1.0261208576998052e-05,
+      "loss": 0.0772,
+      "step": 1350
+    },
+    {
+      "epoch": 2.528832630098453,
+      "eval_loss": 0.07395470887422562,
+      "eval_runtime": 674.8172,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 1350
+    },
+    {
+      "epoch": 2.5307079231129865,
+      "grad_norm": 0.3334695100784302,
+      "learning_rate": 1.0253411306042886e-05,
+      "loss": 0.0581,
+      "step": 1351
+    },
+    {
+      "epoch": 2.53258321612752,
+      "grad_norm": 0.4798605442047119,
+      "learning_rate": 1.024561403508772e-05,
+      "loss": 0.0992,
+      "step": 1352
+    },
+    {
+      "epoch": 2.5344585091420533,
+      "grad_norm": 0.3962152600288391,
+      "learning_rate": 1.0237816764132555e-05,
+      "loss": 0.0782,
+      "step": 1353
+    },
+    {
+      "epoch": 2.536333802156587,
+      "grad_norm": 0.3626305162906647,
+      "learning_rate": 1.0230019493177387e-05,
+      "loss": 0.1039,
+      "step": 1354
+    },
+    {
+      "epoch": 2.5382090951711205,
+      "grad_norm": 0.6571470499038696,
+      "learning_rate": 1.0222222222222223e-05,
+      "loss": 0.0758,
+      "step": 1355
+    },
+    {
+      "epoch": 2.540084388185654,
+      "grad_norm": 0.26701807975769043,
+      "learning_rate": 1.0214424951267057e-05,
+      "loss": 0.0533,
+      "step": 1356
+    },
+    {
+      "epoch": 2.5419596812001877,
+      "grad_norm": 0.23627153038978577,
+      "learning_rate": 1.0206627680311891e-05,
+      "loss": 0.0486,
+      "step": 1357
+    },
+    {
+      "epoch": 2.543834974214721,
+      "grad_norm": 0.2959972620010376,
+      "learning_rate": 1.0198830409356726e-05,
+      "loss": 0.0725,
+      "step": 1358
+    },
+    {
+      "epoch": 2.5457102672292544,
+      "grad_norm": 0.9190336465835571,
+      "learning_rate": 1.019103313840156e-05,
+      "loss": 0.0876,
+      "step": 1359
+    },
+    {
+      "epoch": 2.547585560243788,
+      "grad_norm": 0.3755362033843994,
+      "learning_rate": 1.0183235867446396e-05,
+      "loss": 0.088,
+      "step": 1360
+    },
+    {
+      "epoch": 2.5494608532583216,
+      "grad_norm": 0.3958858847618103,
+      "learning_rate": 1.017543859649123e-05,
+      "loss": 0.1089,
+      "step": 1361
+    },
+    {
+      "epoch": 2.551336146272855,
+      "grad_norm": 0.26009872555732727,
+      "learning_rate": 1.0167641325536064e-05,
+      "loss": 0.0635,
+      "step": 1362
+    },
+    {
+      "epoch": 2.553211439287389,
+      "grad_norm": 0.3636474311351776,
+      "learning_rate": 1.0159844054580898e-05,
+      "loss": 0.0619,
+      "step": 1363
+    },
+    {
+      "epoch": 2.5550867323019224,
+      "grad_norm": 0.4022858440876007,
+      "learning_rate": 1.015204678362573e-05,
+      "loss": 0.0611,
+      "step": 1364
+    },
+    {
+      "epoch": 2.5569620253164556,
+      "grad_norm": 0.44928935170173645,
+      "learning_rate": 1.0144249512670565e-05,
+      "loss": 0.0787,
+      "step": 1365
+    },
+    {
+      "epoch": 2.558837318330989,
+      "grad_norm": 0.40881165862083435,
+      "learning_rate": 1.0136452241715399e-05,
+      "loss": 0.0693,
+      "step": 1366
+    },
+    {
+      "epoch": 2.5607126113455227,
+      "grad_norm": 0.3600151836872101,
+      "learning_rate": 1.0128654970760235e-05,
+      "loss": 0.1056,
+      "step": 1367
+    },
+    {
+      "epoch": 2.5625879043600563,
+      "grad_norm": 0.4044286608695984,
+      "learning_rate": 1.012085769980507e-05,
+      "loss": 0.1205,
+      "step": 1368
+    },
+    {
+      "epoch": 2.56446319737459,
+      "grad_norm": 0.3752521872520447,
+      "learning_rate": 1.0113060428849903e-05,
+      "loss": 0.0906,
+      "step": 1369
+    },
+    {
+      "epoch": 2.566338490389123,
+      "grad_norm": 0.4615623652935028,
+      "learning_rate": 1.0105263157894738e-05,
+      "loss": 0.1037,
+      "step": 1370
+    },
+    {
+      "epoch": 2.5682137834036567,
+      "grad_norm": 0.27481046319007874,
+      "learning_rate": 1.0097465886939572e-05,
+      "loss": 0.0464,
+      "step": 1371
+    },
+    {
+      "epoch": 2.5700890764181903,
+      "grad_norm": 0.4228805601596832,
+      "learning_rate": 1.0089668615984408e-05,
+      "loss": 0.1014,
+      "step": 1372
+    },
+    {
+      "epoch": 2.571964369432724,
+      "grad_norm": 0.33466285467147827,
+      "learning_rate": 1.0081871345029242e-05,
+      "loss": 0.0808,
+      "step": 1373
+    },
+    {
+      "epoch": 2.5738396624472575,
+      "grad_norm": 0.2553812861442566,
+      "learning_rate": 1.0074074074074074e-05,
+      "loss": 0.0516,
+      "step": 1374
+    },
+    {
+      "epoch": 2.575714955461791,
+      "grad_norm": 0.400551438331604,
+      "learning_rate": 1.0066276803118908e-05,
+      "loss": 0.07,
+      "step": 1375
+    },
+    {
+      "epoch": 2.5775902484763247,
+      "grad_norm": 0.3270658552646637,
+      "learning_rate": 1.0058479532163743e-05,
+      "loss": 0.0657,
+      "step": 1376
+    },
+    {
+      "epoch": 2.579465541490858,
+      "grad_norm": 0.3232525587081909,
+      "learning_rate": 1.0050682261208577e-05,
+      "loss": 0.0741,
+      "step": 1377
+    },
+    {
+      "epoch": 2.5813408345053914,
+      "grad_norm": 0.3367152810096741,
+      "learning_rate": 1.0042884990253413e-05,
+      "loss": 0.0884,
+      "step": 1378
+    },
+    {
+      "epoch": 2.583216127519925,
+      "grad_norm": 0.4030762314796448,
+      "learning_rate": 1.0035087719298247e-05,
+      "loss": 0.0941,
+      "step": 1379
+    },
+    {
+      "epoch": 2.5850914205344586,
+      "grad_norm": 0.31161463260650635,
+      "learning_rate": 1.0027290448343081e-05,
+      "loss": 0.0782,
+      "step": 1380
+    },
+    {
+      "epoch": 2.5869667135489918,
+      "grad_norm": 0.39888834953308105,
+      "learning_rate": 1.0019493177387915e-05,
+      "loss": 0.0821,
+      "step": 1381
+    },
+    {
+      "epoch": 2.5888420065635254,
+      "grad_norm": 0.3201531767845154,
+      "learning_rate": 1.001169590643275e-05,
+      "loss": 0.0638,
+      "step": 1382
+    },
+    {
+      "epoch": 2.590717299578059,
+      "grad_norm": 0.32830819487571716,
+      "learning_rate": 1.0003898635477585e-05,
+      "loss": 0.0777,
+      "step": 1383
+    },
+    {
+      "epoch": 2.5925925925925926,
+      "grad_norm": 0.2944835424423218,
+      "learning_rate": 9.996101364522418e-06,
+      "loss": 0.0653,
+      "step": 1384
+    },
+    {
+      "epoch": 2.594467885607126,
+      "grad_norm": 0.45982295274734497,
+      "learning_rate": 9.988304093567252e-06,
+      "loss": 0.1088,
+      "step": 1385
+    },
+    {
+      "epoch": 2.5963431786216598,
+      "grad_norm": 0.3153984248638153,
+      "learning_rate": 9.980506822612086e-06,
+      "loss": 0.0653,
+      "step": 1386
+    },
+    {
+      "epoch": 2.5982184716361933,
+      "grad_norm": 0.41158658266067505,
+      "learning_rate": 9.97270955165692e-06,
+      "loss": 0.0848,
+      "step": 1387
+    },
+    {
+      "epoch": 2.600093764650727,
+      "grad_norm": 0.4037891924381256,
+      "learning_rate": 9.964912280701755e-06,
+      "loss": 0.1319,
+      "step": 1388
+    },
+    {
+      "epoch": 2.60196905766526,
+      "grad_norm": 0.4108717739582062,
+      "learning_rate": 9.957115009746589e-06,
+      "loss": 0.1257,
+      "step": 1389
+    },
+    {
+      "epoch": 2.6038443506797937,
+      "grad_norm": 0.30166369676589966,
+      "learning_rate": 9.949317738791425e-06,
+      "loss": 0.0695,
+      "step": 1390
+    },
+    {
+      "epoch": 2.6057196436943273,
+      "grad_norm": 0.3084668517112732,
+      "learning_rate": 9.941520467836257e-06,
+      "loss": 0.0787,
+      "step": 1391
+    },
+    {
+      "epoch": 2.607594936708861,
+      "grad_norm": 0.5487121939659119,
+      "learning_rate": 9.933723196881091e-06,
+      "loss": 0.1363,
+      "step": 1392
+    },
+    {
+      "epoch": 2.609470229723394,
+      "grad_norm": 0.3409046530723572,
+      "learning_rate": 9.925925925925927e-06,
+      "loss": 0.0781,
+      "step": 1393
+    },
+    {
+      "epoch": 2.6113455227379276,
+      "grad_norm": 0.35762616991996765,
+      "learning_rate": 9.918128654970762e-06,
+      "loss": 0.077,
+      "step": 1394
+    },
+    {
+      "epoch": 2.6132208157524612,
+      "grad_norm": 0.3606860935688019,
+      "learning_rate": 9.910331384015596e-06,
+      "loss": 0.0701,
+      "step": 1395
+    },
+    {
+      "epoch": 2.615096108766995,
+      "grad_norm": 0.33565056324005127,
+      "learning_rate": 9.90253411306043e-06,
+      "loss": 0.0705,
+      "step": 1396
+    },
+    {
+      "epoch": 2.6169714017815284,
+      "grad_norm": 0.35233068466186523,
+      "learning_rate": 9.894736842105264e-06,
+      "loss": 0.0727,
+      "step": 1397
+    },
+    {
+      "epoch": 2.618846694796062,
+      "grad_norm": 0.7528280019760132,
+      "learning_rate": 9.886939571150098e-06,
+      "loss": 0.0846,
+      "step": 1398
+    },
+    {
+      "epoch": 2.6207219878105956,
+      "grad_norm": 0.3784114420413971,
+      "learning_rate": 9.879142300194932e-06,
+      "loss": 0.0894,
+      "step": 1399
+    },
+    {
+      "epoch": 2.6225972808251288,
+      "grad_norm": 0.2998042106628418,
+      "learning_rate": 9.871345029239767e-06,
+      "loss": 0.0729,
+      "step": 1400
+    },
+    {
+      "epoch": 2.6225972808251288,
+      "eval_loss": 0.07494800537824631,
+      "eval_runtime": 675.6374,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 1400
+    },
+    {
+      "epoch": 2.6244725738396624,
+      "grad_norm": 0.45033907890319824,
+      "learning_rate": 9.863547758284601e-06,
+      "loss": 0.0543,
+      "step": 1401
+    },
+    {
+      "epoch": 2.626347866854196,
+      "grad_norm": 0.33740511536598206,
+      "learning_rate": 9.855750487329435e-06,
+      "loss": 0.0887,
+      "step": 1402
+    },
+    {
+      "epoch": 2.6282231598687296,
+      "grad_norm": 0.4127645492553711,
+      "learning_rate": 9.84795321637427e-06,
+      "loss": 0.0915,
+      "step": 1403
+    },
+    {
+      "epoch": 2.630098452883263,
+      "grad_norm": 0.3304232656955719,
+      "learning_rate": 9.840155945419103e-06,
+      "loss": 0.0788,
+      "step": 1404
+    },
+    {
+      "epoch": 2.6319737458977963,
+      "grad_norm": 0.36797669529914856,
+      "learning_rate": 9.83235867446394e-06,
+      "loss": 0.0873,
+      "step": 1405
+    },
+    {
+      "epoch": 2.63384903891233,
+      "grad_norm": 0.2819061577320099,
+      "learning_rate": 9.824561403508772e-06,
+      "loss": 0.0618,
+      "step": 1406
+    },
+    {
+      "epoch": 2.6357243319268635,
+      "grad_norm": 0.52513587474823,
+      "learning_rate": 9.816764132553606e-06,
+      "loss": 0.0824,
+      "step": 1407
+    },
+    {
+      "epoch": 2.637599624941397,
+      "grad_norm": 0.37794506549835205,
+      "learning_rate": 9.808966861598442e-06,
+      "loss": 0.0629,
+      "step": 1408
+    },
+    {
+      "epoch": 2.6394749179559307,
+      "grad_norm": 0.35513389110565186,
+      "learning_rate": 9.801169590643276e-06,
+      "loss": 0.1138,
+      "step": 1409
+    },
+    {
+      "epoch": 2.6413502109704643,
+      "grad_norm": 0.36185532808303833,
+      "learning_rate": 9.79337231968811e-06,
+      "loss": 0.0829,
+      "step": 1410
+    },
+    {
+      "epoch": 2.643225503984998,
+      "grad_norm": 0.3326586186885834,
+      "learning_rate": 9.785575048732944e-06,
+      "loss": 0.0579,
+      "step": 1411
+    },
+    {
+      "epoch": 2.645100796999531,
+      "grad_norm": 0.364624947309494,
+      "learning_rate": 9.777777777777779e-06,
+      "loss": 0.095,
+      "step": 1412
+    },
+    {
+      "epoch": 2.6469760900140646,
+      "grad_norm": 0.3595339059829712,
+      "learning_rate": 9.769980506822613e-06,
+      "loss": 0.0816,
+      "step": 1413
+    },
+    {
+      "epoch": 2.6488513830285982,
+      "grad_norm": 0.25556787848472595,
+      "learning_rate": 9.762183235867447e-06,
+      "loss": 0.0403,
+      "step": 1414
+    },
+    {
+      "epoch": 2.650726676043132,
+      "grad_norm": 0.3111402094364166,
+      "learning_rate": 9.754385964912281e-06,
+      "loss": 0.0652,
+      "step": 1415
+    },
+    {
+      "epoch": 2.652601969057665,
+      "grad_norm": 0.3296256363391876,
+      "learning_rate": 9.746588693957115e-06,
+      "loss": 0.0722,
+      "step": 1416
+    },
+    {
+      "epoch": 2.6544772620721986,
+      "grad_norm": 0.38634583353996277,
+      "learning_rate": 9.73879142300195e-06,
+      "loss": 0.0659,
+      "step": 1417
+    },
+    {
+      "epoch": 2.656352555086732,
+      "grad_norm": 0.35213443636894226,
+      "learning_rate": 9.730994152046784e-06,
+      "loss": 0.0671,
+      "step": 1418
+    },
+    {
+      "epoch": 2.6582278481012658,
+      "grad_norm": 0.3570035696029663,
+      "learning_rate": 9.72319688109162e-06,
+      "loss": 0.061,
+      "step": 1419
+    },
+    {
+      "epoch": 2.6601031411157994,
+      "grad_norm": 0.33103039860725403,
+      "learning_rate": 9.715399610136454e-06,
+      "loss": 0.0693,
+      "step": 1420
+    },
+    {
+      "epoch": 2.661978434130333,
+      "grad_norm": 0.33688971400260925,
+      "learning_rate": 9.707602339181286e-06,
+      "loss": 0.0913,
+      "step": 1421
+    },
+    {
+      "epoch": 2.6638537271448666,
+      "grad_norm": 0.4284787178039551,
+      "learning_rate": 9.69980506822612e-06,
+      "loss": 0.0908,
+      "step": 1422
+    },
+    {
+      "epoch": 2.6657290201594,
+      "grad_norm": 0.46354052424430847,
+      "learning_rate": 9.692007797270956e-06,
+      "loss": 0.0702,
+      "step": 1423
+    },
+    {
+      "epoch": 2.6676043131739333,
+      "grad_norm": 0.32590043544769287,
+      "learning_rate": 9.68421052631579e-06,
+      "loss": 0.0664,
+      "step": 1424
+    },
+    {
+      "epoch": 2.669479606188467,
+      "grad_norm": 0.39951497316360474,
+      "learning_rate": 9.676413255360625e-06,
+      "loss": 0.1049,
+      "step": 1425
+    },
+    {
+      "epoch": 2.6713548992030005,
+      "grad_norm": 0.38672441244125366,
+      "learning_rate": 9.668615984405459e-06,
+      "loss": 0.0707,
+      "step": 1426
+    },
+    {
+      "epoch": 2.673230192217534,
+      "grad_norm": 0.33832356333732605,
+      "learning_rate": 9.660818713450293e-06,
+      "loss": 0.0766,
+      "step": 1427
+    },
+    {
+      "epoch": 2.6751054852320673,
+      "grad_norm": 0.38057902455329895,
+      "learning_rate": 9.653021442495127e-06,
+      "loss": 0.0622,
+      "step": 1428
+    },
+    {
+      "epoch": 2.676980778246601,
+      "grad_norm": 0.2721557021141052,
+      "learning_rate": 9.645224171539962e-06,
+      "loss": 0.0585,
+      "step": 1429
+    },
+    {
+      "epoch": 2.6788560712611345,
+      "grad_norm": 0.3114902973175049,
+      "learning_rate": 9.637426900584796e-06,
+      "loss": 0.0599,
+      "step": 1430
+    },
+    {
+      "epoch": 2.680731364275668,
+      "grad_norm": 0.4293680787086487,
+      "learning_rate": 9.62962962962963e-06,
+      "loss": 0.095,
+      "step": 1431
+    },
+    {
+      "epoch": 2.6826066572902016,
+      "grad_norm": 0.4264834523200989,
+      "learning_rate": 9.621832358674464e-06,
+      "loss": 0.0989,
+      "step": 1432
+    },
+    {
+      "epoch": 2.6844819503047352,
+      "grad_norm": 0.45689934492111206,
+      "learning_rate": 9.614035087719298e-06,
+      "loss": 0.1043,
+      "step": 1433
+    },
+    {
+      "epoch": 2.686357243319269,
+      "grad_norm": 0.3924713432788849,
+      "learning_rate": 9.606237816764134e-06,
+      "loss": 0.0931,
+      "step": 1434
+    },
+    {
+      "epoch": 2.688232536333802,
+      "grad_norm": 0.4446573853492737,
+      "learning_rate": 9.598440545808968e-06,
+      "loss": 0.1063,
+      "step": 1435
+    },
+    {
+      "epoch": 2.6901078293483356,
+      "grad_norm": 0.45865702629089355,
+      "learning_rate": 9.590643274853801e-06,
+      "loss": 0.1034,
+      "step": 1436
+    },
+    {
+      "epoch": 2.691983122362869,
+      "grad_norm": 0.39059358835220337,
+      "learning_rate": 9.582846003898635e-06,
+      "loss": 0.0587,
+      "step": 1437
+    },
+    {
+      "epoch": 2.693858415377403,
+      "grad_norm": 0.678318202495575,
+      "learning_rate": 9.575048732943471e-06,
+      "loss": 0.1233,
+      "step": 1438
+    },
+    {
+      "epoch": 2.6957337083919364,
+      "grad_norm": 0.3813665211200714,
+      "learning_rate": 9.567251461988305e-06,
+      "loss": 0.1088,
+      "step": 1439
+    },
+    {
+      "epoch": 2.6976090014064695,
+      "grad_norm": 0.43635404109954834,
+      "learning_rate": 9.55945419103314e-06,
+      "loss": 0.111,
+      "step": 1440
+    },
+    {
+      "epoch": 2.699484294421003,
+      "grad_norm": 0.5471066236495972,
+      "learning_rate": 9.551656920077974e-06,
+      "loss": 0.082,
+      "step": 1441
+    },
+    {
+      "epoch": 2.7013595874355367,
+      "grad_norm": 0.31651976704597473,
+      "learning_rate": 9.543859649122808e-06,
+      "loss": 0.0533,
+      "step": 1442
+    },
+    {
+      "epoch": 2.7032348804500703,
+      "grad_norm": 0.3988685607910156,
+      "learning_rate": 9.536062378167642e-06,
+      "loss": 0.0786,
+      "step": 1443
+    },
+    {
+      "epoch": 2.705110173464604,
+      "grad_norm": 1.229805827140808,
+      "learning_rate": 9.528265107212476e-06,
+      "loss": 0.0704,
+      "step": 1444
+    },
+    {
+      "epoch": 2.7069854664791375,
+      "grad_norm": 0.3627307415008545,
+      "learning_rate": 9.52046783625731e-06,
+      "loss": 0.0857,
+      "step": 1445
+    },
+    {
+      "epoch": 2.708860759493671,
+      "grad_norm": 0.2482980489730835,
+      "learning_rate": 9.512670565302145e-06,
+      "loss": 0.0583,
+      "step": 1446
+    },
+    {
+      "epoch": 2.7107360525082043,
+      "grad_norm": 0.33016133308410645,
+      "learning_rate": 9.504873294346979e-06,
+      "loss": 0.0678,
+      "step": 1447
+    },
+    {
+      "epoch": 2.712611345522738,
+      "grad_norm": 0.3658045828342438,
+      "learning_rate": 9.497076023391813e-06,
+      "loss": 0.0926,
+      "step": 1448
+    },
+    {
+      "epoch": 2.7144866385372715,
+      "grad_norm": 0.3729887008666992,
+      "learning_rate": 9.489278752436649e-06,
+      "loss": 0.0698,
+      "step": 1449
+    },
+    {
+      "epoch": 2.716361931551805,
+      "grad_norm": 0.36187222599983215,
+      "learning_rate": 9.481481481481483e-06,
+      "loss": 0.0651,
+      "step": 1450
+    },
+    {
+      "epoch": 2.716361931551805,
+      "eval_loss": 0.07403497397899628,
+      "eval_runtime": 673.9675,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 1450
+    },
+    {
+      "epoch": 2.7182372245663387,
+      "grad_norm": 0.3599952757358551,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.0779,
+      "step": 1451
+    },
+    {
+      "epoch": 2.720112517580872,
+      "grad_norm": 0.6131613850593567,
+      "learning_rate": 9.465886939571151e-06,
+      "loss": 0.1098,
+      "step": 1452
+    },
+    {
+      "epoch": 2.7219878105954054,
+      "grad_norm": 0.32765620946884155,
+      "learning_rate": 9.458089668615986e-06,
+      "loss": 0.0596,
+      "step": 1453
+    },
+    {
+      "epoch": 2.723863103609939,
+      "grad_norm": 0.34976986050605774,
+      "learning_rate": 9.45029239766082e-06,
+      "loss": 0.0649,
+      "step": 1454
+    },
+    {
+      "epoch": 2.7257383966244726,
+      "grad_norm": 0.39676281809806824,
+      "learning_rate": 9.442495126705654e-06,
+      "loss": 0.0835,
+      "step": 1455
+    },
+    {
+      "epoch": 2.727613689639006,
+      "grad_norm": 0.3270690143108368,
+      "learning_rate": 9.434697855750488e-06,
+      "loss": 0.0987,
+      "step": 1456
+    },
+    {
+      "epoch": 2.72948898265354,
+      "grad_norm": 0.40614816546440125,
+      "learning_rate": 9.426900584795322e-06,
+      "loss": 0.0847,
+      "step": 1457
+    },
+    {
+      "epoch": 2.7313642756680734,
+      "grad_norm": 0.42256006598472595,
+      "learning_rate": 9.419103313840157e-06,
+      "loss": 0.08,
+      "step": 1458
+    },
+    {
+      "epoch": 2.7332395686826065,
+      "grad_norm": 0.34114959836006165,
+      "learning_rate": 9.41130604288499e-06,
+      "loss": 0.0618,
+      "step": 1459
+    },
+    {
+      "epoch": 2.73511486169714,
+      "grad_norm": 0.30372029542922974,
+      "learning_rate": 9.403508771929825e-06,
+      "loss": 0.0598,
+      "step": 1460
+    },
+    {
+      "epoch": 2.7369901547116737,
+      "grad_norm": 0.439656525850296,
+      "learning_rate": 9.395711500974659e-06,
+      "loss": 0.0857,
+      "step": 1461
+    },
+    {
+      "epoch": 2.7388654477262073,
+      "grad_norm": 0.36488088965415955,
+      "learning_rate": 9.387914230019493e-06,
+      "loss": 0.081,
+      "step": 1462
+    },
+    {
+      "epoch": 2.7407407407407405,
+      "grad_norm": 0.2475002259016037,
+      "learning_rate": 9.380116959064327e-06,
+      "loss": 0.0479,
+      "step": 1463
+    },
+    {
+      "epoch": 2.742616033755274,
+      "grad_norm": 0.3404149115085602,
+      "learning_rate": 9.372319688109163e-06,
+      "loss": 0.07,
+      "step": 1464
+    },
+    {
+      "epoch": 2.7444913267698077,
+      "grad_norm": 0.4210808575153351,
+      "learning_rate": 9.364522417153998e-06,
+      "loss": 0.094,
+      "step": 1465
+    },
+    {
+      "epoch": 2.7463666197843413,
+      "grad_norm": 0.4397883713245392,
+      "learning_rate": 9.35672514619883e-06,
+      "loss": 0.0938,
+      "step": 1466
+    },
+    {
+      "epoch": 2.748241912798875,
+      "grad_norm": 0.4252859950065613,
+      "learning_rate": 9.348927875243666e-06,
+      "loss": 0.0688,
+      "step": 1467
+    },
+    {
+      "epoch": 2.7501172058134085,
+      "grad_norm": 0.39042210578918457,
+      "learning_rate": 9.3411306042885e-06,
+      "loss": 0.0929,
+      "step": 1468
+    },
+    {
+      "epoch": 2.751992498827942,
+      "grad_norm": 0.29327455163002014,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.0375,
+      "step": 1469
+    },
+    {
+      "epoch": 2.7538677918424757,
+      "grad_norm": 0.5631712675094604,
+      "learning_rate": 9.325536062378169e-06,
+      "loss": 0.0884,
+      "step": 1470
+    },
+    {
+      "epoch": 2.755743084857009,
+      "grad_norm": 0.9923895001411438,
+      "learning_rate": 9.317738791423003e-06,
+      "loss": 0.1044,
+      "step": 1471
+    },
+    {
+      "epoch": 2.7576183778715424,
+      "grad_norm": 0.4112682640552521,
+      "learning_rate": 9.309941520467837e-06,
+      "loss": 0.1121,
+      "step": 1472
+    },
+    {
+      "epoch": 2.759493670886076,
+      "grad_norm": 0.3854548931121826,
+      "learning_rate": 9.302144249512671e-06,
+      "loss": 0.0859,
+      "step": 1473
+    },
+    {
+      "epoch": 2.7613689639006096,
+      "grad_norm": 0.46496057510375977,
+      "learning_rate": 9.294346978557505e-06,
+      "loss": 0.1247,
+      "step": 1474
+    },
+    {
+      "epoch": 2.7632442569151427,
+      "grad_norm": 0.4549020230770111,
+      "learning_rate": 9.28654970760234e-06,
+      "loss": 0.0826,
+      "step": 1475
+    },
+    {
+      "epoch": 2.7651195499296763,
+      "grad_norm": 0.3205399811267853,
+      "learning_rate": 9.278752436647174e-06,
+      "loss": 0.0791,
+      "step": 1476
+    },
+    {
+      "epoch": 2.76699484294421,
+      "grad_norm": 0.41249391436576843,
+      "learning_rate": 9.270955165692008e-06,
+      "loss": 0.101,
+      "step": 1477
+    },
+    {
+      "epoch": 2.7688701359587435,
+      "grad_norm": 0.3202146291732788,
+      "learning_rate": 9.263157894736842e-06,
+      "loss": 0.0538,
+      "step": 1478
+    },
+    {
+      "epoch": 2.770745428973277,
+      "grad_norm": 0.34748685359954834,
+      "learning_rate": 9.255360623781678e-06,
+      "loss": 0.0803,
+      "step": 1479
+    },
+    {
+      "epoch": 2.7726207219878107,
+      "grad_norm": 0.3739032745361328,
+      "learning_rate": 9.247563352826512e-06,
+      "loss": 0.0908,
+      "step": 1480
+    },
+    {
+      "epoch": 2.7744960150023443,
+      "grad_norm": 0.420016884803772,
+      "learning_rate": 9.239766081871345e-06,
+      "loss": 0.0991,
+      "step": 1481
+    },
+    {
+      "epoch": 2.7763713080168775,
+      "grad_norm": 0.4077822268009186,
+      "learning_rate": 9.23196881091618e-06,
+      "loss": 0.0954,
+      "step": 1482
+    },
+    {
+      "epoch": 2.778246601031411,
+      "grad_norm": 0.30607473850250244,
+      "learning_rate": 9.224171539961015e-06,
+      "loss": 0.0585,
+      "step": 1483
+    },
+    {
+      "epoch": 2.7801218940459447,
+      "grad_norm": 0.4178578853607178,
+      "learning_rate": 9.216374269005849e-06,
+      "loss": 0.0904,
+      "step": 1484
+    },
+    {
+      "epoch": 2.7819971870604783,
+      "grad_norm": 0.37330979108810425,
+      "learning_rate": 9.208576998050683e-06,
+      "loss": 0.0743,
+      "step": 1485
+    },
+    {
+      "epoch": 2.783872480075012,
+      "grad_norm": 0.3077203631401062,
+      "learning_rate": 9.200779727095517e-06,
+      "loss": 0.0803,
+      "step": 1486
+    },
+    {
+      "epoch": 2.785747773089545,
+      "grad_norm": 0.3057253360748291,
+      "learning_rate": 9.192982456140351e-06,
+      "loss": 0.0621,
+      "step": 1487
+    },
+    {
+      "epoch": 2.7876230661040786,
+      "grad_norm": 0.31388983130455017,
+      "learning_rate": 9.185185185185186e-06,
+      "loss": 0.074,
+      "step": 1488
+    },
+    {
+      "epoch": 2.789498359118612,
+      "grad_norm": 0.3483256995677948,
+      "learning_rate": 9.17738791423002e-06,
+      "loss": 0.0828,
+      "step": 1489
+    },
+    {
+      "epoch": 2.791373652133146,
+      "grad_norm": 0.2892071306705475,
+      "learning_rate": 9.169590643274856e-06,
+      "loss": 0.0716,
+      "step": 1490
+    },
+    {
+      "epoch": 2.7932489451476794,
+      "grad_norm": 0.39768117666244507,
+      "learning_rate": 9.161793372319688e-06,
+      "loss": 0.093,
+      "step": 1491
+    },
+    {
+      "epoch": 2.795124238162213,
+      "grad_norm": 0.408591628074646,
+      "learning_rate": 9.153996101364522e-06,
+      "loss": 0.0962,
+      "step": 1492
+    },
+    {
+      "epoch": 2.7969995311767466,
+      "grad_norm": 1.1204290390014648,
+      "learning_rate": 9.146198830409357e-06,
+      "loss": 0.0881,
+      "step": 1493
+    },
+    {
+      "epoch": 2.7988748241912798,
+      "grad_norm": 0.38295555114746094,
+      "learning_rate": 9.138401559454192e-06,
+      "loss": 0.0959,
+      "step": 1494
+    },
+    {
+      "epoch": 2.8007501172058133,
+      "grad_norm": 0.33753833174705505,
+      "learning_rate": 9.130604288499027e-06,
+      "loss": 0.1013,
+      "step": 1495
+    },
+    {
+      "epoch": 2.802625410220347,
+      "grad_norm": 0.32718968391418457,
+      "learning_rate": 9.12280701754386e-06,
+      "loss": 0.0674,
+      "step": 1496
+    },
+    {
+      "epoch": 2.8045007032348805,
+      "grad_norm": 0.3198351562023163,
+      "learning_rate": 9.115009746588695e-06,
+      "loss": 0.0626,
+      "step": 1497
+    },
+    {
+      "epoch": 2.8063759962494137,
+      "grad_norm": 0.5296013951301575,
+      "learning_rate": 9.10721247563353e-06,
+      "loss": 0.1014,
+      "step": 1498
+    },
+    {
+      "epoch": 2.8082512892639473,
+      "grad_norm": 0.4616522789001465,
+      "learning_rate": 9.099415204678363e-06,
+      "loss": 0.0927,
+      "step": 1499
+    },
+    {
+      "epoch": 2.810126582278481,
+      "grad_norm": 0.41021260619163513,
+      "learning_rate": 9.091617933723198e-06,
+      "loss": 0.0658,
+      "step": 1500
+    },
+    {
+      "epoch": 2.810126582278481,
+      "eval_loss": 0.07440079748630524,
+      "eval_runtime": 671.7286,
+      "eval_samples_per_second": 0.293,
+      "eval_steps_per_second": 0.293,
+      "step": 1500
+    },
+    {
+      "epoch": 2.8120018752930145,
+      "grad_norm": 0.30366870760917664,
+      "learning_rate": 9.083820662768032e-06,
+      "loss": 0.0542,
+      "step": 1501
+    },
+    {
+      "epoch": 2.813877168307548,
+      "grad_norm": 0.484829843044281,
+      "learning_rate": 9.076023391812866e-06,
+      "loss": 0.1142,
+      "step": 1502
+    },
+    {
+      "epoch": 2.8157524613220817,
+      "grad_norm": 0.4134244918823242,
+      "learning_rate": 9.0682261208577e-06,
+      "loss": 0.1089,
+      "step": 1503
+    },
+    {
+      "epoch": 2.8176277543366153,
+      "grad_norm": 0.34767600893974304,
+      "learning_rate": 9.060428849902534e-06,
+      "loss": 0.0834,
+      "step": 1504
+    },
+    {
+      "epoch": 2.819503047351149,
+      "grad_norm": 0.32123830914497375,
+      "learning_rate": 9.05263157894737e-06,
+      "loss": 0.0647,
+      "step": 1505
+    },
+    {
+      "epoch": 2.821378340365682,
+      "grad_norm": 0.2995455861091614,
+      "learning_rate": 9.044834307992203e-06,
+      "loss": 0.051,
+      "step": 1506
+    },
+    {
+      "epoch": 2.8232536333802156,
+      "grad_norm": 0.2944452464580536,
+      "learning_rate": 9.037037037037037e-06,
+      "loss": 0.0508,
+      "step": 1507
+    },
+    {
+      "epoch": 2.825128926394749,
+      "grad_norm": 0.4798865020275116,
+      "learning_rate": 9.029239766081873e-06,
+      "loss": 0.0894,
+      "step": 1508
+    },
+    {
+      "epoch": 2.827004219409283,
+      "grad_norm": 0.3530484437942505,
+      "learning_rate": 9.021442495126707e-06,
+      "loss": 0.0788,
+      "step": 1509
+    },
+    {
+      "epoch": 2.828879512423816,
+      "grad_norm": 0.328976571559906,
+      "learning_rate": 9.013645224171541e-06,
+      "loss": 0.0723,
+      "step": 1510
+    },
+    {
+      "epoch": 2.8307548054383496,
+      "grad_norm": 0.4133010506629944,
+      "learning_rate": 9.005847953216374e-06,
+      "loss": 0.041,
+      "step": 1511
+    },
+    {
+      "epoch": 2.832630098452883,
+      "grad_norm": 0.407266765832901,
+      "learning_rate": 8.99805068226121e-06,
+      "loss": 0.0664,
+      "step": 1512
+    },
+    {
+      "epoch": 2.8345053914674168,
+      "grad_norm": 0.3167998194694519,
+      "learning_rate": 8.990253411306044e-06,
+      "loss": 0.0653,
+      "step": 1513
+    },
+    {
+      "epoch": 2.8363806844819504,
+      "grad_norm": 0.36208438873291016,
+      "learning_rate": 8.982456140350878e-06,
+      "loss": 0.0538,
+      "step": 1514
+    },
+    {
+      "epoch": 2.838255977496484,
+      "grad_norm": 0.5785827040672302,
+      "learning_rate": 8.974658869395712e-06,
+      "loss": 0.1072,
+      "step": 1515
+    },
+    {
+      "epoch": 2.8401312705110175,
+      "grad_norm": 0.4494905471801758,
+      "learning_rate": 8.966861598440546e-06,
+      "loss": 0.0961,
+      "step": 1516
+    },
+    {
+      "epoch": 2.842006563525551,
+      "grad_norm": 0.2944582998752594,
+      "learning_rate": 8.95906432748538e-06,
+      "loss": 0.0481,
+      "step": 1517
+    },
+    {
+      "epoch": 2.8438818565400843,
+      "grad_norm": 0.31694504618644714,
+      "learning_rate": 8.951267056530215e-06,
+      "loss": 0.07,
+      "step": 1518
+    },
+    {
+      "epoch": 2.845757149554618,
+      "grad_norm": 0.33348628878593445,
+      "learning_rate": 8.943469785575049e-06,
+      "loss": 0.0719,
+      "step": 1519
+    },
+    {
+      "epoch": 2.8476324425691515,
+      "grad_norm": 0.26419350504875183,
+      "learning_rate": 8.935672514619885e-06,
+      "loss": 0.0496,
+      "step": 1520
+    },
+    {
+      "epoch": 2.849507735583685,
+      "grad_norm": 0.422234445810318,
+      "learning_rate": 8.927875243664717e-06,
+      "loss": 0.0972,
+      "step": 1521
+    },
+    {
+      "epoch": 2.8513830285982182,
+      "grad_norm": 0.4871450364589691,
+      "learning_rate": 8.920077972709552e-06,
+      "loss": 0.1091,
+      "step": 1522
+    },
+    {
+      "epoch": 2.853258321612752,
+      "grad_norm": 0.39795181155204773,
+      "learning_rate": 8.912280701754387e-06,
+      "loss": 0.0972,
+      "step": 1523
+    },
+    {
+      "epoch": 2.8551336146272854,
+      "grad_norm": 0.3796606957912445,
+      "learning_rate": 8.904483430799222e-06,
+      "loss": 0.071,
+      "step": 1524
+    },
+    {
+      "epoch": 2.857008907641819,
+      "grad_norm": 0.3433877229690552,
+      "learning_rate": 8.896686159844056e-06,
+      "loss": 0.087,
+      "step": 1525
+    },
+    {
+      "epoch": 2.8588842006563526,
+      "grad_norm": 0.3441942036151886,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 0.0537,
+      "step": 1526
+    },
+    {
+      "epoch": 2.8607594936708862,
+      "grad_norm": 0.5122886896133423,
+      "learning_rate": 8.881091617933724e-06,
+      "loss": 0.104,
+      "step": 1527
+    },
+    {
+      "epoch": 2.86263478668542,
+      "grad_norm": 0.3636258542537689,
+      "learning_rate": 8.873294346978558e-06,
+      "loss": 0.0622,
+      "step": 1528
+    },
+    {
+      "epoch": 2.864510079699953,
+      "grad_norm": 0.39862120151519775,
+      "learning_rate": 8.865497076023393e-06,
+      "loss": 0.0827,
+      "step": 1529
+    },
+    {
+      "epoch": 2.8663853727144866,
+      "grad_norm": 0.4351373612880707,
+      "learning_rate": 8.857699805068227e-06,
+      "loss": 0.1058,
+      "step": 1530
+    },
+    {
+      "epoch": 2.86826066572902,
+      "grad_norm": 1.1896051168441772,
+      "learning_rate": 8.849902534113061e-06,
+      "loss": 0.0846,
+      "step": 1531
+    },
+    {
+      "epoch": 2.8701359587435538,
+      "grad_norm": 0.4099143147468567,
+      "learning_rate": 8.842105263157895e-06,
+      "loss": 0.0858,
+      "step": 1532
+    },
+    {
+      "epoch": 2.8720112517580874,
+      "grad_norm": 0.4123772978782654,
+      "learning_rate": 8.83430799220273e-06,
+      "loss": 0.0738,
+      "step": 1533
+    },
+    {
+      "epoch": 2.8738865447726205,
+      "grad_norm": 0.39457976818084717,
+      "learning_rate": 8.826510721247564e-06,
+      "loss": 0.0598,
+      "step": 1534
+    },
+    {
+      "epoch": 2.875761837787154,
+      "grad_norm": 0.36988794803619385,
+      "learning_rate": 8.8187134502924e-06,
+      "loss": 0.0887,
+      "step": 1535
+    },
+    {
+      "epoch": 2.8776371308016877,
+      "grad_norm": 0.4902806282043457,
+      "learning_rate": 8.810916179337232e-06,
+      "loss": 0.0715,
+      "step": 1536
+    },
+    {
+      "epoch": 2.8795124238162213,
+      "grad_norm": 0.31671905517578125,
+      "learning_rate": 8.803118908382066e-06,
+      "loss": 0.0586,
+      "step": 1537
+    },
+    {
+      "epoch": 2.881387716830755,
+      "grad_norm": 0.33224424719810486,
+      "learning_rate": 8.795321637426902e-06,
+      "loss": 0.0644,
+      "step": 1538
+    },
+    {
+      "epoch": 2.8832630098452885,
+      "grad_norm": 0.3493894338607788,
+      "learning_rate": 8.787524366471736e-06,
+      "loss": 0.0737,
+      "step": 1539
+    },
+    {
+      "epoch": 2.885138302859822,
+      "grad_norm": 0.3095395863056183,
+      "learning_rate": 8.77972709551657e-06,
+      "loss": 0.0632,
+      "step": 1540
+    },
+    {
+      "epoch": 2.8870135958743552,
+      "grad_norm": 0.283744215965271,
+      "learning_rate": 8.771929824561405e-06,
+      "loss": 0.0652,
+      "step": 1541
+    },
+    {
+      "epoch": 2.888888888888889,
+      "grad_norm": 0.487179696559906,
+      "learning_rate": 8.764132553606239e-06,
+      "loss": 0.0984,
+      "step": 1542
+    },
+    {
+      "epoch": 2.8907641819034224,
+      "grad_norm": 0.287124365568161,
+      "learning_rate": 8.756335282651073e-06,
+      "loss": 0.0511,
+      "step": 1543
+    },
+    {
+      "epoch": 2.892639474917956,
+      "grad_norm": 0.3281533718109131,
+      "learning_rate": 8.748538011695907e-06,
+      "loss": 0.0639,
+      "step": 1544
+    },
+    {
+      "epoch": 2.894514767932489,
+      "grad_norm": 0.399172306060791,
+      "learning_rate": 8.740740740740741e-06,
+      "loss": 0.0539,
+      "step": 1545
+    },
+    {
+      "epoch": 2.896390060947023,
+      "grad_norm": 0.35018855333328247,
+      "learning_rate": 8.732943469785576e-06,
+      "loss": 0.0751,
+      "step": 1546
+    },
+    {
+      "epoch": 2.8982653539615564,
+      "grad_norm": 0.3269667327404022,
+      "learning_rate": 8.72514619883041e-06,
+      "loss": 0.0528,
+      "step": 1547
+    },
+    {
+      "epoch": 2.90014064697609,
+      "grad_norm": 0.4104063808917999,
+      "learning_rate": 8.717348927875244e-06,
+      "loss": 0.0827,
+      "step": 1548
+    },
+    {
+      "epoch": 2.9020159399906236,
+      "grad_norm": 0.4464809000492096,
+      "learning_rate": 8.709551656920078e-06,
+      "loss": 0.1076,
+      "step": 1549
+    },
+    {
+      "epoch": 2.903891233005157,
+      "grad_norm": 0.3798385560512543,
+      "learning_rate": 8.701754385964914e-06,
+      "loss": 0.0625,
+      "step": 1550
+    },
+    {
+      "epoch": 2.903891233005157,
+      "eval_loss": 0.07311470806598663,
+      "eval_runtime": 675.3623,
+      "eval_samples_per_second": 0.292,
+      "eval_steps_per_second": 0.292,
+      "step": 1550
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2665,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 50,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.3036945883587664e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}