diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4535 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9826498422712935,
+  "eval_steps": 80,
+  "global_step": 634,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0031545741324921135,
+      "grad_norm": 8.64667797088623,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.1089,
+      "step": 1
+    },
+    {
+      "epoch": 0.0031545741324921135,
+      "eval_loss": 1.1576200723648071,
+      "eval_runtime": 662.8793,
+      "eval_samples_per_second": 5.559,
+      "eval_steps_per_second": 1.391,
+      "step": 1
+    },
+    {
+      "epoch": 0.006309148264984227,
+      "grad_norm": 8.875336647033691,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.1522,
+      "step": 2
+    },
+    {
+      "epoch": 0.00946372239747634,
+      "grad_norm": 7.667696475982666,
+      "learning_rate": 6e-06,
+      "loss": 1.1528,
+      "step": 3
+    },
+    {
+      "epoch": 0.012618296529968454,
+      "grad_norm": 7.308061122894287,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.0438,
+      "step": 4
+    },
+    {
+      "epoch": 0.015772870662460567,
+      "grad_norm": 6.896948337554932,
+      "learning_rate": 1e-05,
+      "loss": 0.8653,
+      "step": 5
+    },
+    {
+      "epoch": 0.01892744479495268,
+      "grad_norm": 5.361591339111328,
+      "learning_rate": 1.2e-05,
+      "loss": 0.8274,
+      "step": 6
+    },
+    {
+      "epoch": 0.022082018927444796,
+      "grad_norm": 3.557570695877075,
+      "learning_rate": 1.4e-05,
+      "loss": 0.7553,
+      "step": 7
+    },
+    {
+      "epoch": 0.025236593059936908,
+      "grad_norm": 6.809145927429199,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.8477,
+      "step": 8
+    },
+    {
+      "epoch": 0.028391167192429023,
+      "grad_norm": 4.206286430358887,
+      "learning_rate": 1.8e-05,
+      "loss": 0.8306,
+      "step": 9
+    },
+    {
+      "epoch": 0.031545741324921134,
+      "grad_norm": 2.963203191757202,
+      "learning_rate": 2e-05,
+      "loss": 0.7564,
+      "step": 10
+    },
+    {
+      "epoch": 0.03470031545741325,
+      "grad_norm": 3.0027060508728027,
+      "learning_rate": 1.9999873264100076e-05,
+      "loss": 0.7356,
+      "step": 11
+    },
+    {
+      "epoch": 0.03785488958990536,
+      "grad_norm": 3.2211577892303467,
+      "learning_rate": 1.99994930596127e-05,
+      "loss": 0.7552,
+      "step": 12
+    },
+    {
+      "epoch": 0.04100946372239748,
+      "grad_norm": 3.034398078918457,
+      "learning_rate": 1.9998859396174982e-05,
+      "loss": 0.7652,
+      "step": 13
+    },
+    {
+      "epoch": 0.04416403785488959,
+      "grad_norm": 2.4853951930999756,
+      "learning_rate": 1.9997972289848505e-05,
+      "loss": 0.7347,
+      "step": 14
+    },
+    {
+      "epoch": 0.0473186119873817,
+      "grad_norm": 2.448151111602783,
+      "learning_rate": 1.999683176311891e-05,
+      "loss": 0.7483,
+      "step": 15
+    },
+    {
+      "epoch": 0.050473186119873815,
+      "grad_norm": 2.4126381874084473,
+      "learning_rate": 1.9995437844895337e-05,
+      "loss": 0.7662,
+      "step": 16
+    },
+    {
+      "epoch": 0.05362776025236593,
+      "grad_norm": 2.531477451324463,
+      "learning_rate": 1.9993790570509678e-05,
+      "loss": 0.7282,
+      "step": 17
+    },
+    {
+      "epoch": 0.056782334384858045,
+      "grad_norm": 2.363386392593384,
+      "learning_rate": 1.9991889981715696e-05,
+      "loss": 0.7401,
+      "step": 18
+    },
+    {
+      "epoch": 0.05993690851735016,
+      "grad_norm": 2.3694722652435303,
+      "learning_rate": 1.998973612668796e-05,
+      "loss": 0.7097,
+      "step": 19
+    },
+    {
+      "epoch": 0.06309148264984227,
+      "grad_norm": 2.32259464263916,
+      "learning_rate": 1.9987329060020616e-05,
+      "loss": 0.7328,
+      "step": 20
+    },
+    {
+      "epoch": 0.06624605678233439,
+      "grad_norm": 1.972579836845398,
+      "learning_rate": 1.998466884272602e-05,
+      "loss": 0.701,
+      "step": 21
+    },
+    {
+      "epoch": 0.0694006309148265,
+      "grad_norm": 2.215113878250122,
+      "learning_rate": 1.9981755542233175e-05,
+      "loss": 0.7138,
+      "step": 22
+    },
+    {
+      "epoch": 0.07255520504731862,
+      "grad_norm": 2.0755794048309326,
+      "learning_rate": 1.9978589232386036e-05,
+      "loss": 0.7001,
+      "step": 23
+    },
+    {
+      "epoch": 0.07570977917981073,
+      "grad_norm": 2.0254833698272705,
+      "learning_rate": 1.9975169993441627e-05,
+      "loss": 0.7264,
+      "step": 24
+    },
+    {
+      "epoch": 0.07886435331230283,
+      "grad_norm": 2.2342519760131836,
+      "learning_rate": 1.9971497912068014e-05,
+      "loss": 0.6705,
+      "step": 25
+    },
+    {
+      "epoch": 0.08201892744479496,
+      "grad_norm": 1.8540595769882202,
+      "learning_rate": 1.9967573081342103e-05,
+      "loss": 0.6994,
+      "step": 26
+    },
+    {
+      "epoch": 0.08517350157728706,
+      "grad_norm": 2.0183053016662598,
+      "learning_rate": 1.9963395600747285e-05,
+      "loss": 0.699,
+      "step": 27
+    },
+    {
+      "epoch": 0.08832807570977919,
+      "grad_norm": 1.9821815490722656,
+      "learning_rate": 1.995896557617091e-05,
+      "loss": 0.6842,
+      "step": 28
+    },
+    {
+      "epoch": 0.0914826498422713,
+      "grad_norm": 2.134690999984741,
+      "learning_rate": 1.9954283119901613e-05,
+      "loss": 0.7351,
+      "step": 29
+    },
+    {
+      "epoch": 0.0946372239747634,
+      "grad_norm": 2.202589750289917,
+      "learning_rate": 1.9949348350626456e-05,
+      "loss": 0.6955,
+      "step": 30
+    },
+    {
+      "epoch": 0.09779179810725552,
+      "grad_norm": 2.025588035583496,
+      "learning_rate": 1.9944161393427923e-05,
+      "loss": 0.717,
+      "step": 31
+    },
+    {
+      "epoch": 0.10094637223974763,
+      "grad_norm": 1.9175857305526733,
+      "learning_rate": 1.9938722379780747e-05,
+      "loss": 0.6919,
+      "step": 32
+    },
+    {
+      "epoch": 0.10410094637223975,
+      "grad_norm": 2.061380624771118,
+      "learning_rate": 1.9933031447548593e-05,
+      "loss": 0.6797,
+      "step": 33
+    },
+    {
+      "epoch": 0.10725552050473186,
+      "grad_norm": 1.9209778308868408,
+      "learning_rate": 1.992708874098054e-05,
+      "loss": 0.685,
+      "step": 34
+    },
+    {
+      "epoch": 0.11041009463722397,
+      "grad_norm": 1.849706768989563,
+      "learning_rate": 1.9920894410707446e-05,
+      "loss": 0.6969,
+      "step": 35
+    },
+    {
+      "epoch": 0.11356466876971609,
+      "grad_norm": 1.9812382459640503,
+      "learning_rate": 1.9914448613738107e-05,
+      "loss": 0.6919,
+      "step": 36
+    },
+    {
+      "epoch": 0.1167192429022082,
+      "grad_norm": 1.765123724937439,
+      "learning_rate": 1.99077515134553e-05,
+      "loss": 0.6954,
+      "step": 37
+    },
+    {
+      "epoch": 0.11987381703470032,
+      "grad_norm": 1.7745388746261597,
+      "learning_rate": 1.9900803279611643e-05,
+      "loss": 0.6846,
+      "step": 38
+    },
+    {
+      "epoch": 0.12302839116719243,
+      "grad_norm": 7.526177406311035,
+      "learning_rate": 1.9893604088325258e-05,
+      "loss": 0.7016,
+      "step": 39
+    },
+    {
+      "epoch": 0.12618296529968454,
+      "grad_norm": 1.893181562423706,
+      "learning_rate": 1.9886154122075344e-05,
+      "loss": 0.6665,
+      "step": 40
+    },
+    {
+      "epoch": 0.12933753943217666,
+      "grad_norm": 1.9190433025360107,
+      "learning_rate": 1.987845356969754e-05,
+      "loss": 0.7329,
+      "step": 41
+    },
+    {
+      "epoch": 0.13249211356466878,
+      "grad_norm": 1.8781452178955078,
+      "learning_rate": 1.9870502626379127e-05,
+      "loss": 0.6947,
+      "step": 42
+    },
+    {
+      "epoch": 0.13564668769716087,
+      "grad_norm": 1.9941489696502686,
+      "learning_rate": 1.986230149365411e-05,
+      "loss": 0.7134,
+      "step": 43
+    },
+    {
+      "epoch": 0.138801261829653,
+      "grad_norm": 1.83785879611969,
+      "learning_rate": 1.985385037939806e-05,
+      "loss": 0.7078,
+      "step": 44
+    },
+    {
+      "epoch": 0.14195583596214512,
+      "grad_norm": 1.7626588344573975,
+      "learning_rate": 1.9845149497822898e-05,
+      "loss": 0.6919,
+      "step": 45
+    },
+    {
+      "epoch": 0.14511041009463724,
+      "grad_norm": 1.8639565706253052,
+      "learning_rate": 1.983619906947144e-05,
+      "loss": 0.7054,
+      "step": 46
+    },
+    {
+      "epoch": 0.14826498422712933,
+      "grad_norm": 1.8324793577194214,
+      "learning_rate": 1.9826999321211793e-05,
+      "loss": 0.6579,
+      "step": 47
+    },
+    {
+      "epoch": 0.15141955835962145,
+      "grad_norm": 1.7420214414596558,
+      "learning_rate": 1.9817550486231645e-05,
+      "loss": 0.6982,
+      "step": 48
+    },
+    {
+      "epoch": 0.15457413249211358,
+      "grad_norm": 1.67805016040802,
+      "learning_rate": 1.9807852804032306e-05,
+      "loss": 0.6785,
+      "step": 49
+    },
+    {
+      "epoch": 0.15772870662460567,
+      "grad_norm": 1.8237495422363281,
+      "learning_rate": 1.979790652042268e-05,
+      "loss": 0.6634,
+      "step": 50
+    },
+    {
+      "epoch": 0.1608832807570978,
+      "grad_norm": 1.9041783809661865,
+      "learning_rate": 1.9787711887513e-05,
+      "loss": 0.6801,
+      "step": 51
+    },
+    {
+      "epoch": 0.1640378548895899,
+      "grad_norm": 1.6920254230499268,
+      "learning_rate": 1.977726916370847e-05,
+      "loss": 0.6594,
+      "step": 52
+    },
+    {
+      "epoch": 0.167192429022082,
+      "grad_norm": 1.5961421728134155,
+      "learning_rate": 1.9766578613702683e-05,
+      "loss": 0.6842,
+      "step": 53
+    },
+    {
+      "epoch": 0.17034700315457413,
+      "grad_norm": 1.8021618127822876,
+      "learning_rate": 1.975564050847094e-05,
+      "loss": 0.6848,
+      "step": 54
+    },
+    {
+      "epoch": 0.17350157728706625,
+      "grad_norm": 1.8799400329589844,
+      "learning_rate": 1.974445512526336e-05,
+      "loss": 0.6633,
+      "step": 55
+    },
+    {
+      "epoch": 0.17665615141955837,
+      "grad_norm": 1.7361634969711304,
+      "learning_rate": 1.973302274759786e-05,
+      "loss": 0.6955,
+      "step": 56
+    },
+    {
+      "epoch": 0.17981072555205047,
+      "grad_norm": 1.8239762783050537,
+      "learning_rate": 1.972134366525298e-05,
+      "loss": 0.6763,
+      "step": 57
+    },
+    {
+      "epoch": 0.1829652996845426,
+      "grad_norm": 1.8172523975372314,
+      "learning_rate": 1.9709418174260523e-05,
+      "loss": 0.6831,
+      "step": 58
+    },
+    {
+      "epoch": 0.1861198738170347,
+      "grad_norm": 1.6698800325393677,
+      "learning_rate": 1.969724657689805e-05,
+      "loss": 0.6798,
+      "step": 59
+    },
+    {
+      "epoch": 0.1892744479495268,
+      "grad_norm": 1.731540560722351,
+      "learning_rate": 1.9684829181681236e-05,
+      "loss": 0.6798,
+      "step": 60
+    },
+    {
+      "epoch": 0.19242902208201892,
+      "grad_norm": 1.7183347940444946,
+      "learning_rate": 1.967216630335603e-05,
+      "loss": 0.657,
+      "step": 61
+    },
+    {
+      "epoch": 0.19558359621451105,
+      "grad_norm": 1.7184360027313232,
+      "learning_rate": 1.9659258262890683e-05,
+      "loss": 0.6532,
+      "step": 62
+    },
+    {
+      "epoch": 0.19873817034700317,
+      "grad_norm": 1.64700448513031,
+      "learning_rate": 1.9646105387467627e-05,
+      "loss": 0.6661,
+      "step": 63
+    },
+    {
+      "epoch": 0.20189274447949526,
+      "grad_norm": 1.6115736961364746,
+      "learning_rate": 1.9632708010475166e-05,
+      "loss": 0.655,
+      "step": 64
+    },
+    {
+      "epoch": 0.20504731861198738,
+      "grad_norm": 1.7841370105743408,
+      "learning_rate": 1.9619066471499017e-05,
+      "loss": 0.6707,
+      "step": 65
+    },
+    {
+      "epoch": 0.2082018927444795,
+      "grad_norm": 1.8416322469711304,
+      "learning_rate": 1.9605181116313725e-05,
+      "loss": 0.6914,
+      "step": 66
+    },
+    {
+      "epoch": 0.2113564668769716,
+      "grad_norm": 1.6991705894470215,
+      "learning_rate": 1.959105229687389e-05,
+      "loss": 0.7034,
+      "step": 67
+    },
+    {
+      "epoch": 0.21451104100946372,
+      "grad_norm": 1.7281813621520996,
+      "learning_rate": 1.957668037130524e-05,
+      "loss": 0.6928,
+      "step": 68
+    },
+    {
+      "epoch": 0.21766561514195584,
+      "grad_norm": 1.6657763719558716,
+      "learning_rate": 1.956206570389556e-05,
+      "loss": 0.6543,
+      "step": 69
+    },
+    {
+      "epoch": 0.22082018927444794,
+      "grad_norm": 1.785522699356079,
+      "learning_rate": 1.954720866508546e-05,
+      "loss": 0.69,
+      "step": 70
+    },
+    {
+      "epoch": 0.22397476340694006,
+      "grad_norm": 1.5800981521606445,
+      "learning_rate": 1.9532109631458963e-05,
+      "loss": 0.6801,
+      "step": 71
+    },
+    {
+      "epoch": 0.22712933753943218,
+      "grad_norm": 1.63607656955719,
+      "learning_rate": 1.9516768985734008e-05,
+      "loss": 0.6673,
+      "step": 72
+    },
+    {
+      "epoch": 0.2302839116719243,
+      "grad_norm": 2.059873342514038,
+      "learning_rate": 1.9501187116752694e-05,
+      "loss": 0.687,
+      "step": 73
+    },
+    {
+      "epoch": 0.2334384858044164,
+      "grad_norm": 1.7163550853729248,
+      "learning_rate": 1.9485364419471454e-05,
+      "loss": 0.6786,
+      "step": 74
+    },
+    {
+      "epoch": 0.23659305993690852,
+      "grad_norm": 1.543913722038269,
+      "learning_rate": 1.946930129495106e-05,
+      "loss": 0.6727,
+      "step": 75
+    },
+    {
+      "epoch": 0.23974763406940064,
+      "grad_norm": 1.717105507850647,
+      "learning_rate": 1.9452998150346403e-05,
+      "loss": 0.6843,
+      "step": 76
+    },
+    {
+      "epoch": 0.24290220820189273,
+      "grad_norm": 1.7616279125213623,
+      "learning_rate": 1.9436455398896234e-05,
+      "loss": 0.6908,
+      "step": 77
+    },
+    {
+      "epoch": 0.24605678233438485,
+      "grad_norm": 1.4161309003829956,
+      "learning_rate": 1.9419673459912652e-05,
+      "loss": 0.6388,
+      "step": 78
+    },
+    {
+      "epoch": 0.24921135646687698,
+      "grad_norm": 1.5693848133087158,
+      "learning_rate": 1.9402652758770476e-05,
+      "loss": 0.6679,
+      "step": 79
+    },
+    {
+      "epoch": 0.25236593059936907,
+      "grad_norm": 1.6710033416748047,
+      "learning_rate": 1.9385393726896492e-05,
+      "loss": 0.6641,
+      "step": 80
+    },
+    {
+      "epoch": 0.25236593059936907,
+      "eval_loss": 0.6774148344993591,
+      "eval_runtime": 664.0687,
+      "eval_samples_per_second": 5.549,
+      "eval_steps_per_second": 1.388,
+      "step": 80
+    },
+    {
+      "epoch": 0.2555205047318612,
+      "grad_norm": 1.8038668632507324,
+      "learning_rate": 1.9367896801758478e-05,
+      "loss": 0.6663,
+      "step": 81
+    },
+    {
+      "epoch": 0.2586750788643533,
+      "grad_norm": 1.7743474245071411,
+      "learning_rate": 1.9350162426854152e-05,
+      "loss": 0.6951,
+      "step": 82
+    },
+    {
+      "epoch": 0.2618296529968454,
+      "grad_norm": 1.5669888257980347,
+      "learning_rate": 1.93321910516999e-05,
+      "loss": 0.6476,
+      "step": 83
+    },
+    {
+      "epoch": 0.26498422712933756,
+      "grad_norm": 1.6127671003341675,
+      "learning_rate": 1.9313983131819406e-05,
+      "loss": 0.6502,
+      "step": 84
+    },
+    {
+      "epoch": 0.26813880126182965,
+      "grad_norm": 1.6061285734176636,
+      "learning_rate": 1.9295539128732096e-05,
+      "loss": 0.687,
+      "step": 85
+    },
+    {
+      "epoch": 0.27129337539432175,
+      "grad_norm": 1.6649694442749023,
+      "learning_rate": 1.927685950994143e-05,
+      "loss": 0.6675,
+      "step": 86
+    },
+    {
+      "epoch": 0.2744479495268139,
+      "grad_norm": 1.617295503616333,
+      "learning_rate": 1.9257944748923074e-05,
+      "loss": 0.6656,
+      "step": 87
+    },
+    {
+      "epoch": 0.277602523659306,
+      "grad_norm": 1.6786539554595947,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 0.6505,
+      "step": 88
+    },
+    {
+      "epoch": 0.2807570977917981,
+      "grad_norm": 1.5207414627075195,
+      "learning_rate": 1.9219411723894717e-05,
+      "loss": 0.6734,
+      "step": 89
+    },
+    {
+      "epoch": 0.28391167192429023,
+      "grad_norm": 1.4878873825073242,
+      "learning_rate": 1.9199794436588244e-05,
+      "loss": 0.6567,
+      "step": 90
+    },
+    {
+      "epoch": 0.2870662460567823,
+      "grad_norm": 1.6213213205337524,
+      "learning_rate": 1.917994396043636e-05,
+      "loss": 0.6784,
+      "step": 91
+    },
+    {
+      "epoch": 0.2902208201892745,
+      "grad_norm": 1.5039405822753906,
+      "learning_rate": 1.915986079859266e-05,
+      "loss": 0.6711,
+      "step": 92
+    },
+    {
+      "epoch": 0.29337539432176657,
+      "grad_norm": 1.677122712135315,
+      "learning_rate": 1.9139545460108658e-05,
+      "loss": 0.6948,
+      "step": 93
+    },
+    {
+      "epoch": 0.29652996845425866,
+      "grad_norm": 1.5497688055038452,
+      "learning_rate": 1.91189984599209e-05,
+      "loss": 0.6576,
+      "step": 94
+    },
+    {
+      "epoch": 0.2996845425867508,
+      "grad_norm": 1.5921586751937866,
+      "learning_rate": 1.9098220318837897e-05,
+      "loss": 0.6873,
+      "step": 95
+    },
+    {
+      "epoch": 0.3028391167192429,
+      "grad_norm": 1.61562979221344,
+      "learning_rate": 1.9077211563526928e-05,
+      "loss": 0.7077,
+      "step": 96
+    },
+    {
+      "epoch": 0.305993690851735,
+      "grad_norm": 1.6141735315322876,
+      "learning_rate": 1.9055972726500696e-05,
+      "loss": 0.6801,
+      "step": 97
+    },
+    {
+      "epoch": 0.30914826498422715,
+      "grad_norm": 1.5078473091125488,
+      "learning_rate": 1.9034504346103825e-05,
+      "loss": 0.6281,
+      "step": 98
+    },
+    {
+      "epoch": 0.31230283911671924,
+      "grad_norm": 1.518308401107788,
+      "learning_rate": 1.9012806966499218e-05,
+      "loss": 0.678,
+      "step": 99
+    },
+    {
+      "epoch": 0.31545741324921134,
+      "grad_norm": 1.4899789094924927,
+      "learning_rate": 1.899088113765426e-05,
+      "loss": 0.6644,
+      "step": 100
+    },
+    {
+      "epoch": 0.3186119873817035,
+      "grad_norm": 1.598799228668213,
+      "learning_rate": 1.8968727415326885e-05,
+      "loss": 0.6765,
+      "step": 101
+    },
+    {
+      "epoch": 0.3217665615141956,
+      "grad_norm": 1.5242866277694702,
+      "learning_rate": 1.8946346361051474e-05,
+      "loss": 0.6445,
+      "step": 102
+    },
+    {
+      "epoch": 0.3249211356466877,
+      "grad_norm": 1.4345862865447998,
+      "learning_rate": 1.8923738542124644e-05,
+      "loss": 0.6451,
+      "step": 103
+    },
+    {
+      "epoch": 0.3280757097791798,
+      "grad_norm": 1.403699517250061,
+      "learning_rate": 1.8900904531590845e-05,
+      "loss": 0.6378,
+      "step": 104
+    },
+    {
+      "epoch": 0.3312302839116719,
+      "grad_norm": 1.473271369934082,
+      "learning_rate": 1.8877844908227857e-05,
+      "loss": 0.6849,
+      "step": 105
+    },
+    {
+      "epoch": 0.334384858044164,
+      "grad_norm": 1.5483139753341675,
+      "learning_rate": 1.8854560256532098e-05,
+      "loss": 0.624,
+      "step": 106
+    },
+    {
+      "epoch": 0.33753943217665616,
+      "grad_norm": 1.4683400392532349,
+      "learning_rate": 1.8831051166703832e-05,
+      "loss": 0.6512,
+      "step": 107
+    },
+    {
+      "epoch": 0.34069400630914826,
+      "grad_norm": 1.5488531589508057,
+      "learning_rate": 1.8807318234632187e-05,
+      "loss": 0.6423,
+      "step": 108
+    },
+    {
+      "epoch": 0.3438485804416404,
+      "grad_norm": 1.4319626092910767,
+      "learning_rate": 1.8783362061880063e-05,
+      "loss": 0.6366,
+      "step": 109
+    },
+    {
+      "epoch": 0.3470031545741325,
+      "grad_norm": 1.3048638105392456,
+      "learning_rate": 1.875918325566888e-05,
+      "loss": 0.6448,
+      "step": 110
+    },
+    {
+      "epoch": 0.3501577287066246,
+      "grad_norm": 1.5660271644592285,
+      "learning_rate": 1.8734782428863196e-05,
+      "loss": 0.6693,
+      "step": 111
+    },
+    {
+      "epoch": 0.35331230283911674,
+      "grad_norm": 1.5005526542663574,
+      "learning_rate": 1.8710160199955158e-05,
+      "loss": 0.6294,
+      "step": 112
+    },
+    {
+      "epoch": 0.35646687697160884,
+      "grad_norm": 1.4574633836746216,
+      "learning_rate": 1.8685317193048833e-05,
+      "loss": 0.6534,
+      "step": 113
+    },
+    {
+      "epoch": 0.35962145110410093,
+      "grad_norm": 1.5831985473632812,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.6787,
+      "step": 114
+    },
+    {
+      "epoch": 0.3627760252365931,
+      "grad_norm": 1.5219753980636597,
+      "learning_rate": 1.863497136962213e-05,
+      "loss": 0.6473,
+      "step": 115
+    },
+    {
+      "epoch": 0.3659305993690852,
+      "grad_norm": 1.5488148927688599,
+      "learning_rate": 1.86094698292264e-05,
+      "loss": 0.6433,
+      "step": 116
+    },
+    {
+      "epoch": 0.36908517350157727,
+      "grad_norm": 1.6819711923599243,
+      "learning_rate": 1.8583750063049338e-05,
+      "loss": 0.6693,
+      "step": 117
+    },
+    {
+      "epoch": 0.3722397476340694,
+      "grad_norm": 1.4419407844543457,
+      "learning_rate": 1.8557812723014476e-05,
+      "loss": 0.6561,
+      "step": 118
+    },
+    {
+      "epoch": 0.3753943217665615,
+      "grad_norm": 1.4941089153289795,
+      "learning_rate": 1.8531658466560248e-05,
+      "loss": 0.6374,
+      "step": 119
+    },
+    {
+      "epoch": 0.3785488958990536,
+      "grad_norm": 1.5086392164230347,
+      "learning_rate": 1.8505287956623298e-05,
+      "loss": 0.672,
+      "step": 120
+    },
+    {
+      "epoch": 0.38170347003154576,
+      "grad_norm": 1.4109466075897217,
+      "learning_rate": 1.8478701861621686e-05,
+      "loss": 0.6703,
+      "step": 121
+    },
+    {
+      "epoch": 0.38485804416403785,
+      "grad_norm": 1.858217477798462,
+      "learning_rate": 1.845190085543795e-05,
+      "loss": 0.6577,
+      "step": 122
+    },
+    {
+      "epoch": 0.38801261829652994,
+      "grad_norm": 1.5185223817825317,
+      "learning_rate": 1.8424885617402016e-05,
+      "loss": 0.6305,
+      "step": 123
+    },
+    {
+      "epoch": 0.3911671924290221,
+      "grad_norm": 1.4581152200698853,
+      "learning_rate": 1.8397656832273982e-05,
+      "loss": 0.676,
+      "step": 124
+    },
+    {
+      "epoch": 0.3943217665615142,
+      "grad_norm": 1.4872608184814453,
+      "learning_rate": 1.8370215190226768e-05,
+      "loss": 0.6644,
+      "step": 125
+    },
+    {
+      "epoch": 0.39747634069400634,
+      "grad_norm": 1.5440329313278198,
+      "learning_rate": 1.8342561386828613e-05,
+      "loss": 0.6785,
+      "step": 126
+    },
+    {
+      "epoch": 0.40063091482649843,
+      "grad_norm": 1.5344207286834717,
+      "learning_rate": 1.8314696123025456e-05,
+      "loss": 0.6828,
+      "step": 127
+    },
+    {
+      "epoch": 0.4037854889589905,
+      "grad_norm": 1.4879525899887085,
+      "learning_rate": 1.828662010512314e-05,
+      "loss": 0.6279,
+      "step": 128
+    },
+    {
+      "epoch": 0.4069400630914827,
+      "grad_norm": 1.4044585227966309,
+      "learning_rate": 1.8258334044769558e-05,
+      "loss": 0.6497,
+      "step": 129
+    },
+    {
+      "epoch": 0.41009463722397477,
+      "grad_norm": 1.3317832946777344,
+      "learning_rate": 1.8229838658936566e-05,
+      "loss": 0.6266,
+      "step": 130
+    },
+    {
+      "epoch": 0.41324921135646686,
+      "grad_norm": 1.6094805002212524,
+      "learning_rate": 1.8201134669901838e-05,
+      "loss": 0.6798,
+      "step": 131
+    },
+    {
+      "epoch": 0.416403785488959,
+      "grad_norm": 1.4129669666290283,
+      "learning_rate": 1.817222280523055e-05,
+      "loss": 0.6192,
+      "step": 132
+    },
+    {
+      "epoch": 0.4195583596214511,
+      "grad_norm": 1.4560480117797852,
+      "learning_rate": 1.8143103797756942e-05,
+      "loss": 0.6507,
+      "step": 133
+    },
+    {
+      "epoch": 0.4227129337539432,
+      "grad_norm": 1.4280908107757568,
+      "learning_rate": 1.811377838556573e-05,
+      "loss": 0.6361,
+      "step": 134
+    },
+    {
+      "epoch": 0.42586750788643535,
+      "grad_norm": 1.414389967918396,
+      "learning_rate": 1.8084247311973426e-05,
+      "loss": 0.6525,
+      "step": 135
+    },
+    {
+      "epoch": 0.42902208201892744,
+      "grad_norm": 1.464313268661499,
+      "learning_rate": 1.805451132550946e-05,
+      "loss": 0.6326,
+      "step": 136
+    },
+    {
+      "epoch": 0.43217665615141954,
+      "grad_norm": 1.3970643281936646,
+      "learning_rate": 1.8024571179897236e-05,
+      "loss": 0.6454,
+      "step": 137
+    },
+    {
+      "epoch": 0.4353312302839117,
+      "grad_norm": 1.491580605506897,
+      "learning_rate": 1.7994427634035016e-05,
+      "loss": 0.6338,
+      "step": 138
+    },
+    {
+      "epoch": 0.4384858044164038,
+      "grad_norm": 1.3385744094848633,
+      "learning_rate": 1.7964081451976673e-05,
+      "loss": 0.62,
+      "step": 139
+    },
+    {
+      "epoch": 0.4416403785488959,
+      "grad_norm": 1.394768476486206,
+      "learning_rate": 1.7933533402912354e-05,
+      "loss": 0.6456,
+      "step": 140
+    },
+    {
+      "epoch": 0.444794952681388,
+      "grad_norm": 1.2826526165008545,
+      "learning_rate": 1.7902784261148952e-05,
+      "loss": 0.6448,
+      "step": 141
+    },
+    {
+      "epoch": 0.4479495268138801,
+      "grad_norm": 1.4642585515975952,
+      "learning_rate": 1.7871834806090502e-05,
+      "loss": 0.6475,
+      "step": 142
+    },
+    {
+      "epoch": 0.45110410094637227,
+      "grad_norm": 1.4035379886627197,
+      "learning_rate": 1.784068582221841e-05,
+      "loss": 0.656,
+      "step": 143
+    },
+    {
+      "epoch": 0.45425867507886436,
+      "grad_norm": 1.3586344718933105,
+      "learning_rate": 1.780933809907158e-05,
+      "loss": 0.6348,
+      "step": 144
+    },
+    {
+      "epoch": 0.45741324921135645,
+      "grad_norm": 1.3490393161773682,
+      "learning_rate": 1.7777792431226384e-05,
+      "loss": 0.6276,
+      "step": 145
+    },
+    {
+      "epoch": 0.4605678233438486,
+      "grad_norm": 1.352192997932434,
+      "learning_rate": 1.7746049618276545e-05,
+      "loss": 0.6704,
+      "step": 146
+    },
+    {
+      "epoch": 0.4637223974763407,
+      "grad_norm": 1.5564507246017456,
+      "learning_rate": 1.7714110464812865e-05,
+      "loss": 0.6186,
+      "step": 147
+    },
+    {
+      "epoch": 0.4668769716088328,
+      "grad_norm": 1.202648639678955,
+      "learning_rate": 1.7681975780402807e-05,
+      "loss": 0.6349,
+      "step": 148
+    },
+    {
+      "epoch": 0.47003154574132494,
+      "grad_norm": 1.482734203338623,
+      "learning_rate": 1.7649646379570003e-05,
+      "loss": 0.65,
+      "step": 149
+    },
+    {
+      "epoch": 0.47318611987381703,
+      "grad_norm": 1.5167268514633179,
+      "learning_rate": 1.761712308177359e-05,
+      "loss": 0.6323,
+      "step": 150
+    },
+    {
+      "epoch": 0.47634069400630913,
+      "grad_norm": 1.400444746017456,
+      "learning_rate": 1.7584406711387462e-05,
+      "loss": 0.6367,
+      "step": 151
+    },
+    {
+      "epoch": 0.4794952681388013,
+      "grad_norm": 1.3410762548446655,
+      "learning_rate": 1.755149809767934e-05,
+      "loss": 0.6321,
+      "step": 152
+    },
+    {
+      "epoch": 0.48264984227129337,
+      "grad_norm": 1.396217703819275,
+      "learning_rate": 1.7518398074789776e-05,
+      "loss": 0.6458,
+      "step": 153
+    },
+    {
+      "epoch": 0.48580441640378547,
+      "grad_norm": 1.4141671657562256,
+      "learning_rate": 1.7485107481711014e-05,
+      "loss": 0.6399,
+      "step": 154
+    },
+    {
+      "epoch": 0.4889589905362776,
+      "grad_norm": 1.2821143865585327,
+      "learning_rate": 1.7451627162265703e-05,
+      "loss": 0.6012,
+      "step": 155
+    },
+    {
+      "epoch": 0.4921135646687697,
+      "grad_norm": 1.404240608215332,
+      "learning_rate": 1.7417957965085532e-05,
+      "loss": 0.6408,
+      "step": 156
+    },
+    {
+      "epoch": 0.4952681388012618,
+      "grad_norm": 1.405433177947998,
+      "learning_rate": 1.7384100743589698e-05,
+      "loss": 0.6367,
+      "step": 157
+    },
+    {
+      "epoch": 0.49842271293375395,
+      "grad_norm": 1.3940372467041016,
+      "learning_rate": 1.7350056355963287e-05,
+      "loss": 0.6301,
+      "step": 158
+    },
+    {
+      "epoch": 0.501577287066246,
+      "grad_norm": 1.4160747528076172,
+      "learning_rate": 1.731582566513552e-05,
+      "loss": 0.6016,
+      "step": 159
+    },
+    {
+      "epoch": 0.5047318611987381,
+      "grad_norm": 1.2445266246795654,
+      "learning_rate": 1.7281409538757886e-05,
+      "loss": 0.6434,
+      "step": 160
+    },
+    {
+      "epoch": 0.5047318611987381,
+      "eval_loss": 0.6443384289741516,
+      "eval_runtime": 664.0356,
+      "eval_samples_per_second": 5.549,
+      "eval_steps_per_second": 1.388,
+      "step": 160
+    },
+    {
+      "epoch": 0.5078864353312302,
+      "grad_norm": 1.4029924869537354,
+      "learning_rate": 1.7246808849182125e-05,
+      "loss": 0.6351,
+      "step": 161
+    },
+    {
+      "epoch": 0.5110410094637224,
+      "grad_norm": 1.2926514148712158,
+      "learning_rate": 1.7212024473438145e-05,
+      "loss": 0.6332,
+      "step": 162
+    },
+    {
+      "epoch": 0.5141955835962145,
+      "grad_norm": 1.4654884338378906,
+      "learning_rate": 1.7177057293211786e-05,
+      "loss": 0.6436,
+      "step": 163
+    },
+    {
+      "epoch": 0.5173501577287066,
+      "grad_norm": 1.3930847644805908,
+      "learning_rate": 1.7141908194822448e-05,
+      "loss": 0.6612,
+      "step": 164
+    },
+    {
+      "epoch": 0.5205047318611987,
+      "grad_norm": 1.3918216228485107,
+      "learning_rate": 1.710657806920066e-05,
+      "loss": 0.6033,
+      "step": 165
+    },
+    {
+      "epoch": 0.5236593059936908,
+      "grad_norm": 1.388616919517517,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.6347,
+      "step": 166
+    },
+    {
+      "epoch": 0.526813880126183,
+      "grad_norm": 1.4175844192504883,
+      "learning_rate": 1.703537832290178e-05,
+      "loss": 0.6121,
+      "step": 167
+    },
+    {
+      "epoch": 0.5299684542586751,
+      "grad_norm": 1.323251724243164,
+      "learning_rate": 1.6999510506937465e-05,
+      "loss": 0.6084,
+      "step": 168
+    },
+    {
+      "epoch": 0.5331230283911672,
+      "grad_norm": 1.420966625213623,
+      "learning_rate": 1.696346527312053e-05,
+      "loss": 0.6173,
+      "step": 169
+    },
+    {
+      "epoch": 0.5362776025236593,
+      "grad_norm": 1.4219274520874023,
+      "learning_rate": 1.6927243535095995e-05,
+      "loss": 0.6256,
+      "step": 170
+    },
+    {
+      "epoch": 0.5394321766561514,
+      "grad_norm": 1.3305503129959106,
+      "learning_rate": 1.6890846210982775e-05,
+      "loss": 0.6365,
+      "step": 171
+    },
+    {
+      "epoch": 0.5425867507886435,
+      "grad_norm": 1.2329329252243042,
+      "learning_rate": 1.68542742233504e-05,
+      "loss": 0.6304,
+      "step": 172
+    },
+    {
+      "epoch": 0.5457413249211357,
+      "grad_norm": 1.402936339378357,
+      "learning_rate": 1.6817528499195614e-05,
+      "loss": 0.6229,
+      "step": 173
+    },
+    {
+      "epoch": 0.5488958990536278,
+      "grad_norm": 1.372763991355896,
+      "learning_rate": 1.678060996991891e-05,
+      "loss": 0.6314,
+      "step": 174
+    },
+    {
+      "epoch": 0.5520504731861199,
+      "grad_norm": 1.3213467597961426,
+      "learning_rate": 1.674351957130089e-05,
+      "loss": 0.6338,
+      "step": 175
+    },
+    {
+      "epoch": 0.555205047318612,
+      "grad_norm": 1.2958546876907349,
+      "learning_rate": 1.670625824347856e-05,
+      "loss": 0.666,
+      "step": 176
+    },
+    {
+      "epoch": 0.5583596214511041,
+      "grad_norm": 1.2436187267303467,
+      "learning_rate": 1.666882693092151e-05,
+      "loss": 0.6075,
+      "step": 177
+    },
+    {
+      "epoch": 0.5615141955835962,
+      "grad_norm": 1.3090276718139648,
+      "learning_rate": 1.6631226582407954e-05,
+      "loss": 0.6455,
+      "step": 178
+    },
+    {
+      "epoch": 0.5646687697160884,
+      "grad_norm": 1.2850004434585571,
+      "learning_rate": 1.659345815100069e-05,
+      "loss": 0.6034,
+      "step": 179
+    },
+    {
+      "epoch": 0.5678233438485805,
+      "grad_norm": 1.3550333976745605,
+      "learning_rate": 1.655552259402295e-05,
+      "loss": 0.6171,
+      "step": 180
+    },
+    {
+      "epoch": 0.5709779179810726,
+      "grad_norm": 1.395933985710144,
+      "learning_rate": 1.651742087303412e-05,
+      "loss": 0.6533,
+      "step": 181
+    },
+    {
+      "epoch": 0.5741324921135647,
+      "grad_norm": 1.4969853162765503,
+      "learning_rate": 1.647915395380539e-05,
+      "loss": 0.6071,
+      "step": 182
+    },
+    {
+      "epoch": 0.5772870662460567,
+      "grad_norm": 1.2346265316009521,
+      "learning_rate": 1.6440722806295237e-05,
+      "loss": 0.6135,
+      "step": 183
+    },
+    {
+      "epoch": 0.580441640378549,
+      "grad_norm": 1.3327757120132446,
+      "learning_rate": 1.640212840462488e-05,
+      "loss": 0.6215,
+      "step": 184
+    },
+    {
+      "epoch": 0.583596214511041,
+      "grad_norm": 1.31157648563385,
+      "learning_rate": 1.6363371727053567e-05,
+      "loss": 0.6014,
+      "step": 185
+    },
+    {
+      "epoch": 0.5867507886435331,
+      "grad_norm": 1.4948842525482178,
+      "learning_rate": 1.6324453755953772e-05,
+      "loss": 0.6454,
+      "step": 186
+    },
+    {
+      "epoch": 0.5899053627760252,
+      "grad_norm": 1.2596051692962646,
+      "learning_rate": 1.6285375477786322e-05,
+      "loss": 0.6247,
+      "step": 187
+    },
+    {
+      "epoch": 0.5930599369085173,
+      "grad_norm": 1.3595001697540283,
+      "learning_rate": 1.6246137883075363e-05,
+      "loss": 0.6334,
+      "step": 188
+    },
+    {
+      "epoch": 0.5962145110410094,
+      "grad_norm": 1.400238037109375,
+      "learning_rate": 1.6206741966383268e-05,
+      "loss": 0.6461,
+      "step": 189
+    },
+    {
+      "epoch": 0.5993690851735016,
+      "grad_norm": 1.2766135931015015,
+      "learning_rate": 1.6167188726285433e-05,
+      "loss": 0.6278,
+      "step": 190
+    },
+    {
+      "epoch": 0.6025236593059937,
+      "grad_norm": 1.2833789587020874,
+      "learning_rate": 1.6127479165344948e-05,
+      "loss": 0.612,
+      "step": 191
+    },
+    {
+      "epoch": 0.6056782334384858,
+      "grad_norm": 1.2848033905029297,
+      "learning_rate": 1.608761429008721e-05,
+      "loss": 0.6302,
+      "step": 192
+    },
+    {
+      "epoch": 0.6088328075709779,
+      "grad_norm": 1.4177746772766113,
+      "learning_rate": 1.6047595110974376e-05,
+      "loss": 0.6073,
+      "step": 193
+    },
+    {
+      "epoch": 0.61198738170347,
+      "grad_norm": 1.344997763633728,
+      "learning_rate": 1.600742264237979e-05,
+      "loss": 0.6163,
+      "step": 194
+    },
+    {
+      "epoch": 0.6151419558359621,
+      "grad_norm": 1.304677128791809,
+      "learning_rate": 1.5967097902562242e-05,
+      "loss": 0.6306,
+      "step": 195
+    },
+    {
+      "epoch": 0.6182965299684543,
+      "grad_norm": 1.3383606672286987,
+      "learning_rate": 1.592662191364017e-05,
+      "loss": 0.6592,
+      "step": 196
+    },
+    {
+      "epoch": 0.6214511041009464,
+      "grad_norm": 1.2298630475997925,
+      "learning_rate": 1.5885995701565748e-05,
+      "loss": 0.5953,
+      "step": 197
+    },
+    {
+      "epoch": 0.6246056782334385,
+      "grad_norm": 1.338944911956787,
+      "learning_rate": 1.584522029609889e-05,
+      "loss": 0.626,
+      "step": 198
+    },
+    {
+      "epoch": 0.6277602523659306,
+      "grad_norm": 1.3825503587722778,
+      "learning_rate": 1.5804296730781134e-05,
+      "loss": 0.5928,
+      "step": 199
+    },
+    {
+      "epoch": 0.6309148264984227,
+      "grad_norm": 1.323460340499878,
+      "learning_rate": 1.5763226042909455e-05,
+      "loss": 0.6259,
+      "step": 200
+    },
+    {
+      "epoch": 0.6340694006309149,
+      "grad_norm": 1.314867615699768,
+      "learning_rate": 1.5722009273509972e-05,
+      "loss": 0.6516,
+      "step": 201
+    },
+    {
+      "epoch": 0.637223974763407,
+      "grad_norm": 1.3389687538146973,
+      "learning_rate": 1.568064746731156e-05,
+      "loss": 0.6153,
+      "step": 202
+    },
+    {
+      "epoch": 0.6403785488958991,
+      "grad_norm": 1.256998062133789,
+      "learning_rate": 1.563914167271936e-05,
+      "loss": 0.6111,
+      "step": 203
+    },
+    {
+      "epoch": 0.6435331230283912,
+      "grad_norm": 1.3791075944900513,
+      "learning_rate": 1.5597492941788225e-05,
+      "loss": 0.623,
+      "step": 204
+    },
+    {
+      "epoch": 0.6466876971608833,
+      "grad_norm": 1.2380070686340332,
+      "learning_rate": 1.5555702330196024e-05,
+      "loss": 0.6188,
+      "step": 205
+    },
+    {
+      "epoch": 0.6498422712933754,
+      "grad_norm": 1.4255740642547607,
+      "learning_rate": 1.551377089721692e-05,
+      "loss": 0.6332,
+      "step": 206
+    },
+    {
+      "epoch": 0.6529968454258676,
+      "grad_norm": 1.2124149799346924,
+      "learning_rate": 1.547169970569449e-05,
+      "loss": 0.6415,
+      "step": 207
+    },
+    {
+      "epoch": 0.6561514195583596,
+      "grad_norm": 1.2156108617782593,
+      "learning_rate": 1.542948982201479e-05,
+      "loss": 0.6267,
+      "step": 208
+    },
+    {
+      "epoch": 0.6593059936908517,
+      "grad_norm": 1.3286218643188477,
+      "learning_rate": 1.5387142316079345e-05,
+      "loss": 0.62,
+      "step": 209
+    },
+    {
+      "epoch": 0.6624605678233438,
+      "grad_norm": 1.2249011993408203,
+      "learning_rate": 1.5344658261278013e-05,
+      "loss": 0.6254,
+      "step": 210
+    },
+    {
+      "epoch": 0.6656151419558359,
+      "grad_norm": 1.2610039710998535,
+      "learning_rate": 1.530203873446177e-05,
+      "loss": 0.6071,
+      "step": 211
+    },
+    {
+      "epoch": 0.668769716088328,
+      "grad_norm": 1.19411039352417,
+      "learning_rate": 1.525928481591544e-05,
+      "loss": 0.5879,
+      "step": 212
+    },
+    {
+      "epoch": 0.6719242902208202,
+      "grad_norm": 1.2961221933364868,
+      "learning_rate": 1.5216397589330289e-05,
+      "loss": 0.6229,
+      "step": 213
+    },
+    {
+      "epoch": 0.6750788643533123,
+      "grad_norm": 1.259804368019104,
+      "learning_rate": 1.5173378141776569e-05,
+      "loss": 0.6193,
+      "step": 214
+    },
+    {
+      "epoch": 0.6782334384858044,
+      "grad_norm": 1.3336842060089111,
+      "learning_rate": 1.5130227563675957e-05,
+      "loss": 0.6213,
+      "step": 215
+    },
+    {
+      "epoch": 0.6813880126182965,
+      "grad_norm": 1.3089827299118042,
+      "learning_rate": 1.5086946948773924e-05,
+      "loss": 0.6264,
+      "step": 216
+    },
+    {
+      "epoch": 0.6845425867507886,
+      "grad_norm": 1.3578109741210938,
+      "learning_rate": 1.5043537394112008e-05,
+      "loss": 0.5961,
+      "step": 217
+    },
+    {
+      "epoch": 0.6876971608832808,
+      "grad_norm": 1.4083908796310425,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.6139,
+      "step": 218
+    },
+    {
+      "epoch": 0.6908517350157729,
+      "grad_norm": 1.3945176601409912,
+      "learning_rate": 1.495633586998807e-05,
+      "loss": 0.6085,
+      "step": 219
+    },
+    {
+      "epoch": 0.694006309148265,
+      "grad_norm": 1.1479098796844482,
+      "learning_rate": 1.4912546110838775e-05,
+      "loss": 0.6101,
+      "step": 220
+    },
+    {
+      "epoch": 0.6971608832807571,
+      "grad_norm": 1.3561136722564697,
+      "learning_rate": 1.4868631832499026e-05,
+      "loss": 0.6238,
+      "step": 221
+    },
+    {
+      "epoch": 0.7003154574132492,
+      "grad_norm": 1.2309588193893433,
+      "learning_rate": 1.4824594148071936e-05,
+      "loss": 0.6024,
+      "step": 222
+    },
+    {
+      "epoch": 0.7034700315457413,
+      "grad_norm": 1.2626235485076904,
+      "learning_rate": 1.4780434173788617e-05,
+      "loss": 0.5923,
+      "step": 223
+    },
+    {
+      "epoch": 0.7066246056782335,
+      "grad_norm": 1.399826169013977,
+      "learning_rate": 1.4736153028979893e-05,
+      "loss": 0.6386,
+      "step": 224
+    },
+    {
+      "epoch": 0.7097791798107256,
+      "grad_norm": 1.4666520357131958,
+      "learning_rate": 1.4691751836047907e-05,
+      "loss": 0.6001,
+      "step": 225
+    },
+    {
+      "epoch": 0.7129337539432177,
+      "grad_norm": 1.3348687887191772,
+      "learning_rate": 1.4647231720437687e-05,
+      "loss": 0.6057,
+      "step": 226
+    },
+    {
+      "epoch": 0.7160883280757098,
+      "grad_norm": 1.2561405897140503,
+      "learning_rate": 1.4602593810608619e-05,
+      "loss": 0.6407,
+      "step": 227
+    },
+    {
+      "epoch": 0.7192429022082019,
+      "grad_norm": 1.3740829229354858,
+      "learning_rate": 1.4557839238005831e-05,
+      "loss": 0.6099,
+      "step": 228
+    },
+    {
+      "epoch": 0.722397476340694,
+      "grad_norm": 1.402155876159668,
+      "learning_rate": 1.4512969137031538e-05,
+      "loss": 0.604,
+      "step": 229
+    },
+    {
+      "epoch": 0.7255520504731862,
+      "grad_norm": 1.2615282535552979,
+      "learning_rate": 1.4467984645016259e-05,
+      "loss": 0.6093,
+      "step": 230
+    },
+    {
+      "epoch": 0.7287066246056783,
+      "grad_norm": 1.352616786956787,
+      "learning_rate": 1.4422886902190014e-05,
+      "loss": 0.595,
+      "step": 231
+    },
+    {
+      "epoch": 0.7318611987381703,
+      "grad_norm": 1.249115228652954,
+      "learning_rate": 1.4377677051653404e-05,
+      "loss": 0.5998,
+      "step": 232
+    },
+    {
+      "epoch": 0.7350157728706624,
+      "grad_norm": 1.371975302696228,
+      "learning_rate": 1.4332356239348654e-05,
+      "loss": 0.6501,
+      "step": 233
+    },
+    {
+      "epoch": 0.7381703470031545,
+      "grad_norm": 1.2673488855361938,
+      "learning_rate": 1.4286925614030542e-05,
+      "loss": 0.5949,
+      "step": 234
+    },
+    {
+      "epoch": 0.7413249211356467,
+      "grad_norm": 1.2172908782958984,
+      "learning_rate": 1.4241386327237312e-05,
+      "loss": 0.6168,
+      "step": 235
+    },
+    {
+      "epoch": 0.7444794952681388,
+      "grad_norm": 1.286840796470642,
+      "learning_rate": 1.419573953326146e-05,
+      "loss": 0.6207,
+      "step": 236
+    },
+    {
+      "epoch": 0.7476340694006309,
+      "grad_norm": 1.1921433210372925,
+      "learning_rate": 1.4149986389120485e-05,
+      "loss": 0.5995,
+      "step": 237
+    },
+    {
+      "epoch": 0.750788643533123,
+      "grad_norm": 1.2323322296142578,
+      "learning_rate": 1.410412805452757e-05,
+      "loss": 0.5885,
+      "step": 238
+    },
+    {
+      "epoch": 0.7539432176656151,
+      "grad_norm": 1.3277860879898071,
+      "learning_rate": 1.4058165691862176e-05,
+      "loss": 0.5773,
+      "step": 239
+    },
+    {
+      "epoch": 0.7570977917981072,
+      "grad_norm": 1.2921340465545654,
+      "learning_rate": 1.4012100466140579e-05,
+      "loss": 0.5926,
+      "step": 240
+    },
+    {
+      "epoch": 0.7570977917981072,
+      "eval_loss": 0.616804301738739,
+      "eval_runtime": 663.8146,
+      "eval_samples_per_second": 5.551,
+      "eval_steps_per_second": 1.389,
+      "step": 240
+    },
+    {
+      "epoch": 0.7602523659305994,
+      "grad_norm": 1.29549241065979,
+      "learning_rate": 1.3965933544986351e-05,
+      "loss": 0.6163,
+      "step": 241
+    },
+    {
+      "epoch": 0.7634069400630915,
+      "grad_norm": 1.1697142124176025,
+      "learning_rate": 1.3919666098600753e-05,
+      "loss": 0.6154,
+      "step": 242
+    },
+    {
+      "epoch": 0.7665615141955836,
+      "grad_norm": 1.2146536111831665,
+      "learning_rate": 1.3873299299733068e-05,
+      "loss": 0.6019,
+      "step": 243
+    },
+    {
+      "epoch": 0.7697160883280757,
+      "grad_norm": 1.2958508729934692,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 0.5956,
+      "step": 244
+    },
+    {
+      "epoch": 0.7728706624605678,
+      "grad_norm": 1.1277776956558228,
+      "learning_rate": 1.3780272348110356e-05,
+      "loss": 0.6105,
+      "step": 245
+    },
+    {
+      "epoch": 0.7760252365930599,
+      "grad_norm": 1.3057700395584106,
+      "learning_rate": 1.3733614553326211e-05,
+      "loss": 0.62,
+      "step": 246
+    },
+    {
+      "epoch": 0.7791798107255521,
+      "grad_norm": 1.2456624507904053,
+      "learning_rate": 1.368686212194199e-05,
+      "loss": 0.5725,
+      "step": 247
+    },
+    {
+      "epoch": 0.7823343848580442,
+      "grad_norm": 1.2655675411224365,
+      "learning_rate": 1.3640016238999984e-05,
+      "loss": 0.5804,
+      "step": 248
+    },
+    {
+      "epoch": 0.7854889589905363,
+      "grad_norm": 1.2346055507659912,
+      "learning_rate": 1.3593078091911218e-05,
+      "loss": 0.6167,
+      "step": 249
+    },
+    {
+      "epoch": 0.7886435331230284,
+      "grad_norm": 1.2213687896728516,
+      "learning_rate": 1.3546048870425356e-05,
+      "loss": 0.5975,
+      "step": 250
+    },
+    {
+      "epoch": 0.7917981072555205,
+      "grad_norm": 1.2108865976333618,
+      "learning_rate": 1.3498929766600543e-05,
+      "loss": 0.6253,
+      "step": 251
+    },
+    {
+      "epoch": 0.7949526813880127,
+      "grad_norm": 1.2665812969207764,
+      "learning_rate": 1.3451721974773179e-05,
+      "loss": 0.617,
+      "step": 252
+    },
+    {
+      "epoch": 0.7981072555205048,
+      "grad_norm": 1.2785872220993042,
+      "learning_rate": 1.340442669152766e-05,
+      "loss": 0.5892,
+      "step": 253
+    },
+    {
+      "epoch": 0.8012618296529969,
+      "grad_norm": 1.28725004196167,
+      "learning_rate": 1.335704511566605e-05,
+      "loss": 0.6091,
+      "step": 254
+    },
+    {
+      "epoch": 0.804416403785489,
+      "grad_norm": 1.1814377307891846,
+      "learning_rate": 1.3309578448177674e-05,
+      "loss": 0.6111,
+      "step": 255
+    },
+    {
+      "epoch": 0.807570977917981,
+      "grad_norm": 1.1113842725753784,
+      "learning_rate": 1.3262027892208696e-05,
+      "loss": 0.5909,
+      "step": 256
+    },
+    {
+      "epoch": 0.8107255520504731,
+      "grad_norm": 1.260785460472107,
+      "learning_rate": 1.3214394653031616e-05,
+      "loss": 0.5868,
+      "step": 257
+    },
+    {
+      "epoch": 0.8138801261829653,
+      "grad_norm": 1.2386345863342285,
+      "learning_rate": 1.3166679938014728e-05,
+      "loss": 0.6174,
+      "step": 258
+    },
+    {
+      "epoch": 0.8170347003154574,
+      "grad_norm": 1.1698509454727173,
+      "learning_rate": 1.311888495659149e-05,
+      "loss": 0.5771,
+      "step": 259
+    },
+    {
+      "epoch": 0.8201892744479495,
+      "grad_norm": 1.1928315162658691,
+      "learning_rate": 1.3071010920229909e-05,
+      "loss": 0.5782,
+      "step": 260
+    },
+    {
+      "epoch": 0.8233438485804416,
+      "grad_norm": 1.2843685150146484,
+      "learning_rate": 1.3023059042401797e-05,
+      "loss": 0.634,
+      "step": 261
+    },
+    {
+      "epoch": 0.8264984227129337,
+      "grad_norm": 1.2431703805923462,
+      "learning_rate": 1.297503053855203e-05,
+      "loss": 0.6044,
+      "step": 262
+    },
+    {
+      "epoch": 0.8296529968454258,
+      "grad_norm": 1.2186146974563599,
+      "learning_rate": 1.2926926626067745e-05,
+      "loss": 0.5992,
+      "step": 263
+    },
+    {
+      "epoch": 0.832807570977918,
+      "grad_norm": 1.1846224069595337,
+      "learning_rate": 1.2878748524247462e-05,
+      "loss": 0.6153,
+      "step": 264
+    },
+    {
+      "epoch": 0.8359621451104101,
+      "grad_norm": 1.2342958450317383,
+      "learning_rate": 1.2830497454270206e-05,
+      "loss": 0.5964,
+      "step": 265
+    },
+    {
+      "epoch": 0.8391167192429022,
+      "grad_norm": 1.1780555248260498,
+      "learning_rate": 1.2782174639164528e-05,
+      "loss": 0.5976,
+      "step": 266
+    },
+    {
+      "epoch": 0.8422712933753943,
+      "grad_norm": 1.148030161857605,
+      "learning_rate": 1.2733781303777524e-05,
+      "loss": 0.593,
+      "step": 267
+    },
+    {
+      "epoch": 0.8454258675078864,
+      "grad_norm": 1.1693962812423706,
+      "learning_rate": 1.2685318674743769e-05,
+      "loss": 0.5967,
+      "step": 268
+    },
+    {
+      "epoch": 0.8485804416403786,
+      "grad_norm": 1.1817485094070435,
+      "learning_rate": 1.263678798045425e-05,
+      "loss": 0.602,
+      "step": 269
+    },
+    {
+      "epoch": 0.8517350157728707,
+      "grad_norm": 1.2170345783233643,
+      "learning_rate": 1.2588190451025209e-05,
+      "loss": 0.6007,
+      "step": 270
+    },
+    {
+      "epoch": 0.8548895899053628,
+      "grad_norm": 1.258149266242981,
+      "learning_rate": 1.2539527318266971e-05,
+      "loss": 0.6105,
+      "step": 271
+    },
+    {
+      "epoch": 0.8580441640378549,
+      "grad_norm": 1.240839958190918,
+      "learning_rate": 1.2490799815652719e-05,
+      "loss": 0.6192,
+      "step": 272
+    },
+    {
+      "epoch": 0.861198738170347,
+      "grad_norm": 1.1205185651779175,
+      "learning_rate": 1.244200917828723e-05,
+      "loss": 0.5872,
+      "step": 273
+    },
+    {
+      "epoch": 0.8643533123028391,
+      "grad_norm": 1.2566895484924316,
+      "learning_rate": 1.2393156642875579e-05,
+      "loss": 0.6091,
+      "step": 274
+    },
+    {
+      "epoch": 0.8675078864353313,
+      "grad_norm": 1.2157973051071167,
+      "learning_rate": 1.2344243447691767e-05,
+      "loss": 0.6024,
+      "step": 275
+    },
+    {
+      "epoch": 0.8706624605678234,
+      "grad_norm": 1.340427279472351,
+      "learning_rate": 1.2295270832547358e-05,
+      "loss": 0.6222,
+      "step": 276
+    },
+    {
+      "epoch": 0.8738170347003155,
+      "grad_norm": 1.2133880853652954,
+      "learning_rate": 1.2246240038760042e-05,
+      "loss": 0.5839,
+      "step": 277
+    },
+    {
+      "epoch": 0.8769716088328076,
+      "grad_norm": 1.2732712030410767,
+      "learning_rate": 1.2197152309122173e-05,
+      "loss": 0.5892,
+      "step": 278
+    },
+    {
+      "epoch": 0.8801261829652997,
+      "grad_norm": 1.2458168268203735,
+      "learning_rate": 1.214800888786927e-05,
+      "loss": 0.6016,
+      "step": 279
+    },
+    {
+      "epoch": 0.8832807570977917,
+      "grad_norm": 1.2807629108428955,
+      "learning_rate": 1.2098811020648475e-05,
+      "loss": 0.5992,
+      "step": 280
+    },
+    {
+      "epoch": 0.886435331230284,
+      "grad_norm": 1.227212905883789,
+      "learning_rate": 1.204955995448699e-05,
+      "loss": 0.5923,
+      "step": 281
+    },
+    {
+      "epoch": 0.889589905362776,
+      "grad_norm": 1.1762304306030273,
+      "learning_rate": 1.2000256937760446e-05,
+      "loss": 0.6033,
+      "step": 282
+    },
+    {
+      "epoch": 0.8927444794952681,
+      "grad_norm": 1.2128591537475586,
+      "learning_rate": 1.1950903220161286e-05,
+      "loss": 0.5981,
+      "step": 283
+    },
+    {
+      "epoch": 0.8958990536277602,
+      "grad_norm": 1.220958948135376,
+      "learning_rate": 1.1901500052667068e-05,
+      "loss": 0.6141,
+      "step": 284
+    },
+    {
+      "epoch": 0.8990536277602523,
+      "grad_norm": 1.2984957695007324,
+      "learning_rate": 1.1852048687508775e-05,
+      "loss": 0.5732,
+      "step": 285
+    },
+    {
+      "epoch": 0.9022082018927445,
+      "grad_norm": 1.3143209218978882,
+      "learning_rate": 1.180255037813906e-05,
+      "loss": 0.6314,
+      "step": 286
+    },
+    {
+      "epoch": 0.9053627760252366,
+      "grad_norm": 1.3318054676055908,
+      "learning_rate": 1.1753006379200478e-05,
+      "loss": 0.6521,
+      "step": 287
+    },
+    {
+      "epoch": 0.9085173501577287,
+      "grad_norm": 1.13868248462677,
+      "learning_rate": 1.1703417946493682e-05,
+      "loss": 0.5834,
+      "step": 288
+    },
+    {
+      "epoch": 0.9116719242902208,
+      "grad_norm": 1.3481686115264893,
+      "learning_rate": 1.1653786336945614e-05,
+      "loss": 0.6125,
+      "step": 289
+    },
+    {
+      "epoch": 0.9148264984227129,
+      "grad_norm": 1.1816397905349731,
+      "learning_rate": 1.1604112808577603e-05,
+      "loss": 0.5762,
+      "step": 290
+    },
+    {
+      "epoch": 0.917981072555205,
+      "grad_norm": 1.27208411693573,
+      "learning_rate": 1.1554398620473516e-05,
+      "loss": 0.618,
+      "step": 291
+    },
+    {
+      "epoch": 0.9211356466876972,
+      "grad_norm": 1.1916297674179077,
+      "learning_rate": 1.1504645032747832e-05,
+      "loss": 0.5882,
+      "step": 292
+    },
+    {
+      "epoch": 0.9242902208201893,
+      "grad_norm": 1.1507753133773804,
+      "learning_rate": 1.1454853306513687e-05,
+      "loss": 0.5897,
+      "step": 293
+    },
+    {
+      "epoch": 0.9274447949526814,
+      "grad_norm": 1.1446702480316162,
+      "learning_rate": 1.1405024703850929e-05,
+      "loss": 0.6054,
+      "step": 294
+    },
+    {
+      "epoch": 0.9305993690851735,
+      "grad_norm": 1.2258915901184082,
+      "learning_rate": 1.1355160487774119e-05,
+      "loss": 0.6226,
+      "step": 295
+    },
+    {
+      "epoch": 0.9337539432176656,
+      "grad_norm": 1.3792144060134888,
+      "learning_rate": 1.130526192220052e-05,
+      "loss": 0.6032,
+      "step": 296
+    },
+    {
+      "epoch": 0.9369085173501577,
+      "grad_norm": 1.0934364795684814,
+      "learning_rate": 1.1255330271918048e-05,
+      "loss": 0.5531,
+      "step": 297
+    },
+    {
+      "epoch": 0.9400630914826499,
+      "grad_norm": 1.2484863996505737,
+      "learning_rate": 1.1205366802553231e-05,
+      "loss": 0.6318,
+      "step": 298
+    },
+    {
+      "epoch": 0.943217665615142,
+      "grad_norm": 1.2264879941940308,
+      "learning_rate": 1.1155372780539125e-05,
+      "loss": 0.5752,
+      "step": 299
+    },
+    {
+      "epoch": 0.9463722397476341,
+      "grad_norm": 1.2283985614776611,
+      "learning_rate": 1.11053494730832e-05,
+      "loss": 0.6066,
+      "step": 300
+    },
+    {
+      "epoch": 0.9495268138801262,
+      "grad_norm": 1.202512502670288,
+      "learning_rate": 1.1055298148135236e-05,
+      "loss": 0.6071,
+      "step": 301
+    },
+    {
+      "epoch": 0.9526813880126183,
+      "grad_norm": 1.1098576784133911,
+      "learning_rate": 1.1005220074355172e-05,
+      "loss": 0.5847,
+      "step": 302
+    },
+    {
+      "epoch": 0.9558359621451105,
+      "grad_norm": 1.1731163263320923,
+      "learning_rate": 1.0955116521080959e-05,
+      "loss": 0.5566,
+      "step": 303
+    },
+    {
+      "epoch": 0.9589905362776026,
+      "grad_norm": 1.078802466392517,
+      "learning_rate": 1.090498875829638e-05,
+      "loss": 0.5951,
+      "step": 304
+    },
+    {
+      "epoch": 0.9621451104100947,
+      "grad_norm": 1.1604228019714355,
+      "learning_rate": 1.085483805659886e-05,
+      "loss": 0.581,
+      "step": 305
+    },
+    {
+      "epoch": 0.9652996845425867,
+      "grad_norm": 1.1810637712478638,
+      "learning_rate": 1.0804665687167262e-05,
+      "loss": 0.5869,
+      "step": 306
+    },
+    {
+      "epoch": 0.9684542586750788,
+      "grad_norm": 1.189583659172058,
+      "learning_rate": 1.0754472921729661e-05,
+      "loss": 0.5633,
+      "step": 307
+    },
+    {
+      "epoch": 0.9716088328075709,
+      "grad_norm": 1.166351318359375,
+      "learning_rate": 1.070426103253112e-05,
+      "loss": 0.5865,
+      "step": 308
+    },
+    {
+      "epoch": 0.9747634069400631,
+      "grad_norm": 1.4211933612823486,
+      "learning_rate": 1.0654031292301432e-05,
+      "loss": 0.5915,
+      "step": 309
+    },
+    {
+      "epoch": 0.9779179810725552,
+      "grad_norm": 1.2419383525848389,
+      "learning_rate": 1.0603784974222862e-05,
+      "loss": 0.5881,
+      "step": 310
+    },
+    {
+      "epoch": 0.9810725552050473,
+      "grad_norm": 1.1408345699310303,
+      "learning_rate": 1.0553523351897879e-05,
+      "loss": 0.5523,
+      "step": 311
+    },
+    {
+      "epoch": 0.9842271293375394,
+      "grad_norm": 1.1748343706130981,
+      "learning_rate": 1.0503247699316867e-05,
+      "loss": 0.5969,
+      "step": 312
+    },
+    {
+      "epoch": 0.9873817034700315,
+      "grad_norm": 1.317017674446106,
+      "learning_rate": 1.0452959290825846e-05,
+      "loss": 0.6046,
+      "step": 313
+    },
+    {
+      "epoch": 0.9905362776025236,
+      "grad_norm": 1.2760308980941772,
+      "learning_rate": 1.0402659401094154e-05,
+      "loss": 0.5818,
+      "step": 314
+    },
+    {
+      "epoch": 0.9936908517350158,
+      "grad_norm": 1.283344030380249,
+      "learning_rate": 1.0352349305082148e-05,
+      "loss": 0.6027,
+      "step": 315
+    },
+    {
+      "epoch": 0.9968454258675079,
+      "grad_norm": 1.2597942352294922,
+      "learning_rate": 1.030203027800889e-05,
+      "loss": 0.6036,
+      "step": 316
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.2721868753433228,
+      "learning_rate": 1.0251703595319816e-05,
+      "loss": 0.5449,
+      "step": 317
+    },
+    {
+      "epoch": 1.003154574132492,
+      "grad_norm": 1.278611183166504,
+      "learning_rate": 1.0201370532654404e-05,
+      "loss": 0.6012,
+      "step": 318
+    },
+    {
+      "epoch": 1.0063091482649842,
+      "grad_norm": 1.278645396232605,
+      "learning_rate": 1.015103236581386e-05,
+      "loss": 0.6102,
+      "step": 319
+    },
+    {
+      "epoch": 1.0094637223974763,
+      "grad_norm": 1.1869760751724243,
+      "learning_rate": 1.0100690370728756e-05,
+      "loss": 0.5517,
+      "step": 320
+    },
+    {
+      "epoch": 1.0094637223974763,
+      "eval_loss": 0.593919038772583,
+      "eval_runtime": 664.4346,
+      "eval_samples_per_second": 5.546,
+      "eval_steps_per_second": 1.388,
+      "step": 320
+    },
+    {
+      "epoch": 1.0126182965299684,
+      "grad_norm": 1.2704983949661255,
+      "learning_rate": 1.0050345823426704e-05,
+      "loss": 0.613,
+      "step": 321
+    },
+    {
+      "epoch": 1.0157728706624605,
+      "grad_norm": 1.306869626045227,
+      "learning_rate": 1e-05,
+      "loss": 0.5952,
+      "step": 322
+    },
+    {
+      "epoch": 1.001577287066246,
+      "grad_norm": 1.2639447450637817,
+      "learning_rate": 9.9496541765733e-06,
+      "loss": 0.5163,
+      "step": 323
+    },
+    {
+      "epoch": 1.0047318611987381,
+      "grad_norm": 1.4290575981140137,
+      "learning_rate": 9.899309629271246e-06,
+      "loss": 0.4636,
+      "step": 324
+    },
+    {
+      "epoch": 1.0078864353312302,
+      "grad_norm": 1.2376357316970825,
+      "learning_rate": 9.848967634186142e-06,
+      "loss": 0.4396,
+      "step": 325
+    },
+    {
+      "epoch": 1.0110410094637223,
+      "grad_norm": 1.3232203722000122,
+      "learning_rate": 9.7986294673456e-06,
+      "loss": 0.439,
+      "step": 326
+    },
+    {
+      "epoch": 1.0141955835962144,
+      "grad_norm": 1.1686787605285645,
+      "learning_rate": 9.748296404680189e-06,
+      "loss": 0.4235,
+      "step": 327
+    },
+    {
+      "epoch": 1.0173501577287065,
+      "grad_norm": 1.096146821975708,
+      "learning_rate": 9.697969721991114e-06,
+      "loss": 0.4354,
+      "step": 328
+    },
+    {
+      "epoch": 1.0205047318611988,
+      "grad_norm": 1.2047231197357178,
+      "learning_rate": 9.647650694917857e-06,
+      "loss": 0.4287,
+      "step": 329
+    },
+    {
+      "epoch": 1.023659305993691,
+      "grad_norm": 1.3038204908370972,
+      "learning_rate": 9.597340598905851e-06,
+      "loss": 0.4211,
+      "step": 330
+    },
+    {
+      "epoch": 1.026813880126183,
+      "grad_norm": 1.4500834941864014,
+      "learning_rate": 9.547040709174159e-06,
+      "loss": 0.4327,
+      "step": 331
+    },
+    {
+      "epoch": 1.0299684542586751,
+      "grad_norm": 1.2673792839050293,
+      "learning_rate": 9.496752300683133e-06,
+      "loss": 0.4075,
+      "step": 332
+    },
+    {
+      "epoch": 1.0331230283911672,
+      "grad_norm": 1.304667353630066,
+      "learning_rate": 9.446476648102123e-06,
+      "loss": 0.3996,
+      "step": 333
+    },
+    {
+      "epoch": 1.0362776025236593,
+      "grad_norm": 1.2832258939743042,
+      "learning_rate": 9.39621502577714e-06,
+      "loss": 0.4391,
+      "step": 334
+    },
+    {
+      "epoch": 1.0394321766561514,
+      "grad_norm": 1.213700294494629,
+      "learning_rate": 9.34596870769857e-06,
+      "loss": 0.4012,
+      "step": 335
+    },
+    {
+      "epoch": 1.0425867507886435,
+      "grad_norm": 1.179382562637329,
+      "learning_rate": 9.295738967468881e-06,
+      "loss": 0.4126,
+      "step": 336
+    },
+    {
+      "epoch": 1.0457413249211356,
+      "grad_norm": 1.2564669847488403,
+      "learning_rate": 9.24552707827034e-06,
+      "loss": 0.4211,
+      "step": 337
+    },
+    {
+      "epoch": 1.0488958990536277,
+      "grad_norm": 1.2675079107284546,
+      "learning_rate": 9.195334312832742e-06,
+      "loss": 0.4032,
+      "step": 338
+    },
+    {
+      "epoch": 1.0520504731861198,
+      "grad_norm": 1.2636772394180298,
+      "learning_rate": 9.145161943401143e-06,
+      "loss": 0.412,
+      "step": 339
+    },
+    {
+      "epoch": 1.055205047318612,
+      "grad_norm": 1.1221716403961182,
+      "learning_rate": 9.095011241703623e-06,
+      "loss": 0.4317,
+      "step": 340
+    },
+    {
+      "epoch": 1.0583596214511042,
+      "grad_norm": 1.1785527467727661,
+      "learning_rate": 9.044883478919043e-06,
+      "loss": 0.4034,
+      "step": 341
+    },
+    {
+      "epoch": 1.0615141955835963,
+      "grad_norm": 1.1424877643585205,
+      "learning_rate": 8.994779925644832e-06,
+      "loss": 0.4292,
+      "step": 342
+    },
+    {
+      "epoch": 1.0646687697160884,
+      "grad_norm": 1.2037312984466553,
+      "learning_rate": 8.944701851864767e-06,
+      "loss": 0.429,
+      "step": 343
+    },
+    {
+      "epoch": 1.0678233438485805,
+      "grad_norm": 1.255042314529419,
+      "learning_rate": 8.894650526916802e-06,
+      "loss": 0.4372,
+      "step": 344
+    },
+    {
+      "epoch": 1.0709779179810726,
+      "grad_norm": 1.1968024969100952,
+      "learning_rate": 8.844627219460879e-06,
+      "loss": 0.43,
+      "step": 345
+    },
+    {
+      "epoch": 1.0741324921135647,
+      "grad_norm": 1.1957961320877075,
+      "learning_rate": 8.79463319744677e-06,
+      "loss": 0.4136,
+      "step": 346
+    },
+    {
+      "epoch": 1.0772870662460567,
+      "grad_norm": 1.1776931285858154,
+      "learning_rate": 8.744669728081959e-06,
+      "loss": 0.3959,
+      "step": 347
+    },
+    {
+      "epoch": 1.0804416403785488,
+      "grad_norm": 1.2205976247787476,
+      "learning_rate": 8.694738077799487e-06,
+      "loss": 0.4152,
+      "step": 348
+    },
+    {
+      "epoch": 1.083596214511041,
+      "grad_norm": 1.2931081056594849,
+      "learning_rate": 8.644839512225886e-06,
+      "loss": 0.4117,
+      "step": 349
+    },
+    {
+      "epoch": 1.086750788643533,
+      "grad_norm": 1.2070932388305664,
+      "learning_rate": 8.594975296149076e-06,
+      "loss": 0.4008,
+      "step": 350
+    },
+    {
+      "epoch": 1.0899053627760251,
+      "grad_norm": 1.1776983737945557,
+      "learning_rate": 8.54514669348632e-06,
+      "loss": 0.4074,
+      "step": 351
+    },
+    {
+      "epoch": 1.0930599369085174,
+      "grad_norm": 1.1496163606643677,
+      "learning_rate": 8.49535496725217e-06,
+      "loss": 0.4245,
+      "step": 352
+    },
+    {
+      "epoch": 1.0962145110410095,
+      "grad_norm": 1.2373661994934082,
+      "learning_rate": 8.445601379526482e-06,
+      "loss": 0.426,
+      "step": 353
+    },
+    {
+      "epoch": 1.0993690851735016,
+      "grad_norm": 1.1729695796966553,
+      "learning_rate": 8.395887191422397e-06,
+      "loss": 0.4106,
+      "step": 354
+    },
+    {
+      "epoch": 1.1025236593059937,
+      "grad_norm": 1.281626582145691,
+      "learning_rate": 8.346213663054388e-06,
+      "loss": 0.417,
+      "step": 355
+    },
+    {
+      "epoch": 1.1056782334384858,
+      "grad_norm": 1.1353029012680054,
+      "learning_rate": 8.296582053506318e-06,
+      "loss": 0.4064,
+      "step": 356
+    },
+    {
+      "epoch": 1.108832807570978,
+      "grad_norm": 1.119582176208496,
+      "learning_rate": 8.246993620799526e-06,
+      "loss": 0.4044,
+      "step": 357
+    },
+    {
+      "epoch": 1.11198738170347,
+      "grad_norm": 1.1174126863479614,
+      "learning_rate": 8.197449621860944e-06,
+      "loss": 0.41,
+      "step": 358
+    },
+    {
+      "epoch": 1.115141955835962,
+      "grad_norm": 1.2338104248046875,
+      "learning_rate": 8.147951312491227e-06,
+      "loss": 0.42,
+      "step": 359
+    },
+    {
+      "epoch": 1.1182965299684542,
+      "grad_norm": 1.156226634979248,
+      "learning_rate": 8.098499947332935e-06,
+      "loss": 0.4116,
+      "step": 360
+    },
+    {
+      "epoch": 1.1214511041009463,
+      "grad_norm": 1.1352908611297607,
+      "learning_rate": 8.04909677983872e-06,
+      "loss": 0.4315,
+      "step": 361
+    },
+    {
+      "epoch": 1.1246056782334386,
+      "grad_norm": 1.1950355768203735,
+      "learning_rate": 7.999743062239557e-06,
+      "loss": 0.4167,
+      "step": 362
+    },
+    {
+      "epoch": 1.1277602523659307,
+      "grad_norm": 1.1333998441696167,
+      "learning_rate": 7.950440045513014e-06,
+      "loss": 0.4066,
+      "step": 363
+    },
+    {
+      "epoch": 1.1309148264984228,
+      "grad_norm": 1.3177785873413086,
+      "learning_rate": 7.901188979351527e-06,
+      "loss": 0.4258,
+      "step": 364
+    },
+    {
+      "epoch": 1.1340694006309149,
+      "grad_norm": 1.2589313983917236,
+      "learning_rate": 7.851991112130734e-06,
+      "loss": 0.4286,
+      "step": 365
+    },
+    {
+      "epoch": 1.137223974763407,
+      "grad_norm": 1.1732462644577026,
+      "learning_rate": 7.802847690877832e-06,
+      "loss": 0.4186,
+      "step": 366
+    },
+    {
+      "epoch": 1.140378548895899,
+      "grad_norm": 1.1480242013931274,
+      "learning_rate": 7.753759961239965e-06,
+      "loss": 0.4114,
+      "step": 367
+    },
+    {
+      "epoch": 1.1435331230283912,
+      "grad_norm": 1.162279486656189,
+      "learning_rate": 7.704729167452647e-06,
+      "loss": 0.4185,
+      "step": 368
+    },
+    {
+      "epoch": 1.1466876971608833,
+      "grad_norm": 1.1626859903335571,
+      "learning_rate": 7.655756552308238e-06,
+      "loss": 0.3958,
+      "step": 369
+    },
+    {
+      "epoch": 1.1498422712933754,
+      "grad_norm": 1.245093584060669,
+      "learning_rate": 7.606843357124426e-06,
+      "loss": 0.4081,
+      "step": 370
+    },
+    {
+      "epoch": 1.1529968454258674,
+      "grad_norm": 1.1781249046325684,
+      "learning_rate": 7.557990821712769e-06,
+      "loss": 0.3909,
+      "step": 371
+    },
+    {
+      "epoch": 1.1561514195583595,
+      "grad_norm": 1.1131494045257568,
+      "learning_rate": 7.5092001843472825e-06,
+      "loss": 0.4143,
+      "step": 372
+    },
+    {
+      "epoch": 1.1593059936908516,
+      "grad_norm": 1.1700150966644287,
+      "learning_rate": 7.460472681733031e-06,
+      "loss": 0.4047,
+      "step": 373
+    },
+    {
+      "epoch": 1.1624605678233437,
+      "grad_norm": 1.179157018661499,
+      "learning_rate": 7.411809548974792e-06,
+      "loss": 0.4029,
+      "step": 374
+    },
+    {
+      "epoch": 1.165615141955836,
+      "grad_norm": 1.187483787536621,
+      "learning_rate": 7.3632120195457515e-06,
+      "loss": 0.4197,
+      "step": 375
+    },
+    {
+      "epoch": 1.1687697160883281,
+      "grad_norm": 1.2395418882369995,
+      "learning_rate": 7.314681325256232e-06,
+      "loss": 0.413,
+      "step": 376
+    },
+    {
+      "epoch": 1.1719242902208202,
+      "grad_norm": 1.1669936180114746,
+      "learning_rate": 7.26621869622248e-06,
+      "loss": 0.4178,
+      "step": 377
+    },
+    {
+      "epoch": 1.1750788643533123,
+      "grad_norm": 1.1914113759994507,
+      "learning_rate": 7.217825360835475e-06,
+      "loss": 0.419,
+      "step": 378
+    },
+    {
+      "epoch": 1.1782334384858044,
+      "grad_norm": 1.221914529800415,
+      "learning_rate": 7.169502545729798e-06,
+      "loss": 0.4273,
+      "step": 379
+    },
+    {
+      "epoch": 1.1813880126182965,
+      "grad_norm": 1.2406303882598877,
+      "learning_rate": 7.12125147575254e-06,
+      "loss": 0.408,
+      "step": 380
+    },
+    {
+      "epoch": 1.1845425867507886,
+      "grad_norm": 1.104156732559204,
+      "learning_rate": 7.07307337393226e-06,
+      "loss": 0.4018,
+      "step": 381
+    },
+    {
+      "epoch": 1.1876971608832807,
+      "grad_norm": 1.0950199365615845,
+      "learning_rate": 7.024969461447973e-06,
+      "loss": 0.3922,
+      "step": 382
+    },
+    {
+      "epoch": 1.1908517350157728,
+      "grad_norm": 1.2959777116775513,
+      "learning_rate": 6.976940957598206e-06,
+      "loss": 0.4143,
+      "step": 383
+    },
+    {
+      "epoch": 1.1940063091482649,
+      "grad_norm": 1.1827858686447144,
+      "learning_rate": 6.928989079770094e-06,
+      "loss": 0.4158,
+      "step": 384
+    },
+    {
+      "epoch": 1.1971608832807572,
+      "grad_norm": 1.2715120315551758,
+      "learning_rate": 6.881115043408512e-06,
+      "loss": 0.407,
+      "step": 385
+    },
+    {
+      "epoch": 1.2003154574132493,
+      "grad_norm": 1.2652549743652344,
+      "learning_rate": 6.833320061985278e-06,
+      "loss": 0.4067,
+      "step": 386
+    },
+    {
+      "epoch": 1.2034700315457414,
+      "grad_norm": 1.2982840538024902,
+      "learning_rate": 6.785605346968387e-06,
+      "loss": 0.4197,
+      "step": 387
+    },
+    {
+      "epoch": 1.2066246056782335,
+      "grad_norm": 1.2446273565292358,
+      "learning_rate": 6.7379721077913095e-06,
+      "loss": 0.4005,
+      "step": 388
+    },
+    {
+      "epoch": 1.2097791798107256,
+      "grad_norm": 1.186058759689331,
+      "learning_rate": 6.690421551822333e-06,
+      "loss": 0.397,
+      "step": 389
+    },
+    {
+      "epoch": 1.2129337539432177,
+      "grad_norm": 1.2176367044448853,
+      "learning_rate": 6.6429548843339554e-06,
+      "loss": 0.4026,
+      "step": 390
+    },
+    {
+      "epoch": 1.2160883280757098,
+      "grad_norm": 1.2751494646072388,
+      "learning_rate": 6.595573308472338e-06,
+      "loss": 0.4221,
+      "step": 391
+    },
+    {
+      "epoch": 1.2192429022082019,
+      "grad_norm": 1.2309192419052124,
+      "learning_rate": 6.548278025226822e-06,
+      "loss": 0.3957,
+      "step": 392
+    },
+    {
+      "epoch": 1.222397476340694,
+      "grad_norm": 1.1658203601837158,
+      "learning_rate": 6.501070233399458e-06,
+      "loss": 0.4145,
+      "step": 393
+    },
+    {
+      "epoch": 1.225552050473186,
+      "grad_norm": 1.2239638566970825,
+      "learning_rate": 6.453951129574644e-06,
+      "loss": 0.4315,
+      "step": 394
+    },
+    {
+      "epoch": 1.2287066246056781,
+      "grad_norm": 1.1642601490020752,
+      "learning_rate": 6.406921908088783e-06,
+      "loss": 0.3925,
+      "step": 395
+    },
+    {
+      "epoch": 1.2318611987381702,
+      "grad_norm": 1.2048768997192383,
+      "learning_rate": 6.3599837610000194e-06,
+      "loss": 0.4141,
+      "step": 396
+    },
+    {
+      "epoch": 1.2350157728706626,
+      "grad_norm": 1.212185263633728,
+      "learning_rate": 6.3131378780580134e-06,
+      "loss": 0.4129,
+      "step": 397
+    },
+    {
+      "epoch": 1.2381703470031546,
+      "grad_norm": 1.1716859340667725,
+      "learning_rate": 6.266385446673791e-06,
+      "loss": 0.4062,
+      "step": 398
+    },
+    {
+      "epoch": 1.2413249211356467,
+      "grad_norm": 1.2397621870040894,
+      "learning_rate": 6.219727651889647e-06,
+      "loss": 0.4336,
+      "step": 399
+    },
+    {
+      "epoch": 1.2444794952681388,
+      "grad_norm": 1.3703044652938843,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.4244,
+      "step": 400
+    },
+    {
+      "epoch": 1.2444794952681388,
+      "eval_loss": 0.6105833053588867,
+      "eval_runtime": 664.4299,
+      "eval_samples_per_second": 5.546,
+      "eval_steps_per_second": 1.388,
+      "step": 400
+    },
+    {
+      "epoch": 1.247634069400631,
+      "grad_norm": 1.1632176637649536,
+      "learning_rate": 6.1267007002669345e-06,
+      "loss": 0.4016,
+      "step": 401
+    },
+    {
+      "epoch": 1.250788643533123,
+      "grad_norm": 1.1854286193847656,
+      "learning_rate": 6.080333901399252e-06,
+      "loss": 0.3862,
+      "step": 402
+    },
+    {
+      "epoch": 1.2539432176656151,
+      "grad_norm": 1.2743747234344482,
+      "learning_rate": 6.0340664550136494e-06,
+      "loss": 0.4044,
+      "step": 403
+    },
+    {
+      "epoch": 1.2570977917981072,
+      "grad_norm": 1.2288442850112915,
+      "learning_rate": 5.987899533859422e-06,
+      "loss": 0.3989,
+      "step": 404
+    },
+    {
+      "epoch": 1.2602523659305993,
+      "grad_norm": 1.2840858697891235,
+      "learning_rate": 5.941834308137828e-06,
+      "loss": 0.3944,
+      "step": 405
+    },
+    {
+      "epoch": 1.2634069400630916,
+      "grad_norm": 1.2368574142456055,
+      "learning_rate": 5.895871945472434e-06,
+      "loss": 0.3969,
+      "step": 406
+    },
+    {
+      "epoch": 1.2665615141955837,
+      "grad_norm": 1.3580341339111328,
+      "learning_rate": 5.85001361087952e-06,
+      "loss": 0.4144,
+      "step": 407
+    },
+    {
+      "epoch": 1.2697160883280758,
+      "grad_norm": 1.1498668193817139,
+      "learning_rate": 5.8042604667385475e-06,
+      "loss": 0.4003,
+      "step": 408
+    },
+    {
+      "epoch": 1.272870662460568,
+      "grad_norm": 1.314024806022644,
+      "learning_rate": 5.75861367276269e-06,
+      "loss": 0.3914,
+      "step": 409
+    },
+    {
+      "epoch": 1.27602523659306,
+      "grad_norm": 1.2495567798614502,
+      "learning_rate": 5.713074385969457e-06,
+      "loss": 0.4122,
+      "step": 410
+    },
+    {
+      "epoch": 1.279179810725552,
+      "grad_norm": 1.2432278394699097,
+      "learning_rate": 5.667643760651349e-06,
+      "loss": 0.4234,
+      "step": 411
+    },
+    {
+      "epoch": 1.2823343848580442,
+      "grad_norm": 1.187386393547058,
+      "learning_rate": 5.622322948346595e-06,
+      "loss": 0.3921,
+      "step": 412
+    },
+    {
+      "epoch": 1.2854889589905363,
+      "grad_norm": 1.2718896865844727,
+      "learning_rate": 5.5771130978099896e-06,
+      "loss": 0.4159,
+      "step": 413
+    },
+    {
+      "epoch": 1.2886435331230284,
+      "grad_norm": 1.2912015914916992,
+      "learning_rate": 5.532015354983742e-06,
+      "loss": 0.4098,
+      "step": 414
+    },
+    {
+      "epoch": 1.2917981072555205,
+      "grad_norm": 1.4019771814346313,
+      "learning_rate": 5.4870308629684675e-06,
+      "loss": 0.4077,
+      "step": 415
+    },
+    {
+      "epoch": 1.2949526813880126,
+      "grad_norm": 1.229522943496704,
+      "learning_rate": 5.442160761994171e-06,
+      "loss": 0.4022,
+      "step": 416
+    },
+    {
+      "epoch": 1.2981072555205047,
+      "grad_norm": 1.2295200824737549,
+      "learning_rate": 5.397406189391387e-06,
+      "loss": 0.4228,
+      "step": 417
+    },
+    {
+      "epoch": 1.3012618296529967,
+      "grad_norm": 1.2241239547729492,
+      "learning_rate": 5.352768279562315e-06,
+      "loss": 0.3971,
+      "step": 418
+    },
+    {
+      "epoch": 1.3044164037854888,
+      "grad_norm": 1.0674831867218018,
+      "learning_rate": 5.308248163952097e-06,
+      "loss": 0.3996,
+      "step": 419
+    },
+    {
+      "epoch": 1.307570977917981,
+      "grad_norm": 1.2485908269882202,
+      "learning_rate": 5.263846971020108e-06,
+      "loss": 0.4103,
+      "step": 420
+    },
+    {
+      "epoch": 1.3107255520504733,
+      "grad_norm": 1.1408573389053345,
+      "learning_rate": 5.219565826211382e-06,
+      "loss": 0.3956,
+      "step": 421
+    },
+    {
+      "epoch": 1.3138801261829653,
+      "grad_norm": 1.2800486087799072,
+      "learning_rate": 5.175405851928068e-06,
+      "loss": 0.4084,
+      "step": 422
+    },
+    {
+      "epoch": 1.3170347003154574,
+      "grad_norm": 1.2795509099960327,
+      "learning_rate": 5.131368167500977e-06,
+      "loss": 0.4214,
+      "step": 423
+    },
+    {
+      "epoch": 1.3201892744479495,
+      "grad_norm": 1.347074031829834,
+      "learning_rate": 5.087453889161229e-06,
+      "loss": 0.4083,
+      "step": 424
+    },
+    {
+      "epoch": 1.3233438485804416,
+      "grad_norm": 1.1154026985168457,
+      "learning_rate": 5.043664130011932e-06,
+      "loss": 0.3826,
+      "step": 425
+    },
+    {
+      "epoch": 1.3264984227129337,
+      "grad_norm": 1.1923314332962036,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.4107,
+      "step": 426
+    },
+    {
+      "epoch": 1.3296529968454258,
+      "grad_norm": 1.1018216609954834,
+      "learning_rate": 4.956462605887994e-06,
+      "loss": 0.4209,
+      "step": 427
+    },
+    {
+      "epoch": 1.332807570977918,
+      "grad_norm": 1.1312483549118042,
+      "learning_rate": 4.913053051226079e-06,
+      "loss": 0.3719,
+      "step": 428
+    },
+    {
+      "epoch": 1.3359621451104102,
+      "grad_norm": 1.2638659477233887,
+      "learning_rate": 4.869772436324045e-06,
+      "loss": 0.4253,
+      "step": 429
+    },
+    {
+      "epoch": 1.3391167192429023,
+      "grad_norm": 1.1518404483795166,
+      "learning_rate": 4.826621858223431e-06,
+      "loss": 0.401,
+      "step": 430
+    },
+    {
+      "epoch": 1.3422712933753944,
+      "grad_norm": 1.1931076049804688,
+      "learning_rate": 4.783602410669712e-06,
+      "loss": 0.3892,
+      "step": 431
+    },
+    {
+      "epoch": 1.3454258675078865,
+      "grad_norm": 1.0546995401382446,
+      "learning_rate": 4.740715184084559e-06,
+      "loss": 0.4079,
+      "step": 432
+    },
+    {
+      "epoch": 1.3485804416403786,
+      "grad_norm": 1.1187161207199097,
+      "learning_rate": 4.697961265538231e-06,
+      "loss": 0.4102,
+      "step": 433
+    },
+    {
+      "epoch": 1.3517350157728707,
+      "grad_norm": 1.1335748434066772,
+      "learning_rate": 4.655341738721989e-06,
+      "loss": 0.3976,
+      "step": 434
+    },
+    {
+      "epoch": 1.3548895899053628,
+      "grad_norm": 1.1055843830108643,
+      "learning_rate": 4.612857683920658e-06,
+      "loss": 0.3868,
+      "step": 435
+    },
+    {
+      "epoch": 1.3580441640378549,
+      "grad_norm": 1.2191433906555176,
+      "learning_rate": 4.570510177985213e-06,
+      "loss": 0.3791,
+      "step": 436
+    },
+    {
+      "epoch": 1.361198738170347,
+      "grad_norm": 1.2138489484786987,
+      "learning_rate": 4.528300294305519e-06,
+      "loss": 0.4201,
+      "step": 437
+    },
+    {
+      "epoch": 1.364353312302839,
+      "grad_norm": 1.265018105506897,
+      "learning_rate": 4.486229102783084e-06,
+      "loss": 0.4125,
+      "step": 438
+    },
+    {
+      "epoch": 1.3675078864353312,
+      "grad_norm": 1.2155956029891968,
+      "learning_rate": 4.444297669803981e-06,
+      "loss": 0.402,
+      "step": 439
+    },
+    {
+      "epoch": 1.3706624605678233,
+      "grad_norm": 1.1446729898452759,
+      "learning_rate": 4.40250705821178e-06,
+      "loss": 0.3997,
+      "step": 440
+    },
+    {
+      "epoch": 1.3738170347003154,
+      "grad_norm": 1.1707253456115723,
+      "learning_rate": 4.360858327280641e-06,
+      "loss": 0.3909,
+      "step": 441
+    },
+    {
+      "epoch": 1.3769716088328074,
+      "grad_norm": 1.1826376914978027,
+      "learning_rate": 4.319352532688444e-06,
+      "loss": 0.3989,
+      "step": 442
+    },
+    {
+      "epoch": 1.3801261829652998,
+      "grad_norm": 1.1921921968460083,
+      "learning_rate": 4.277990726490029e-06,
+      "loss": 0.4208,
+      "step": 443
+    },
+    {
+      "epoch": 1.3832807570977919,
+      "grad_norm": 1.2096469402313232,
+      "learning_rate": 4.236773957090549e-06,
+      "loss": 0.4095,
+      "step": 444
+    },
+    {
+      "epoch": 1.386435331230284,
+      "grad_norm": 1.2644251585006714,
+      "learning_rate": 4.1957032692188685e-06,
+      "loss": 0.4093,
+      "step": 445
+    },
+    {
+      "epoch": 1.389589905362776,
+      "grad_norm": 1.1463419198989868,
+      "learning_rate": 4.154779703901114e-06,
+      "loss": 0.3803,
+      "step": 446
+    },
+    {
+      "epoch": 1.3927444794952681,
+      "grad_norm": 1.2488454580307007,
+      "learning_rate": 4.1140042984342535e-06,
+      "loss": 0.402,
+      "step": 447
+    },
+    {
+      "epoch": 1.3958990536277602,
+      "grad_norm": 1.3429874181747437,
+      "learning_rate": 4.073378086359834e-06,
+      "loss": 0.3923,
+      "step": 448
+    },
+    {
+      "epoch": 1.3990536277602523,
+      "grad_norm": 1.2372740507125854,
+      "learning_rate": 4.03290209743776e-06,
+      "loss": 0.4102,
+      "step": 449
+    },
+    {
+      "epoch": 1.4022082018927444,
+      "grad_norm": 1.1483173370361328,
+      "learning_rate": 3.99257735762021e-06,
+      "loss": 0.39,
+      "step": 450
+    },
+    {
+      "epoch": 1.4053627760252365,
+      "grad_norm": 1.2342681884765625,
+      "learning_rate": 3.952404889025626e-06,
+      "loss": 0.4071,
+      "step": 451
+    },
+    {
+      "epoch": 1.4085173501577288,
+      "grad_norm": 1.1589683294296265,
+      "learning_rate": 3.912385709912794e-06,
+      "loss": 0.3892,
+      "step": 452
+    },
+    {
+      "epoch": 1.411671924290221,
+      "grad_norm": 1.1464444398880005,
+      "learning_rate": 3.8725208346550535e-06,
+      "loss": 0.388,
+      "step": 453
+    },
+    {
+      "epoch": 1.414826498422713,
+      "grad_norm": 1.1136562824249268,
+      "learning_rate": 3.832811273714569e-06,
+      "loss": 0.3977,
+      "step": 454
+    },
+    {
+      "epoch": 1.4179810725552051,
+      "grad_norm": 1.19005286693573,
+      "learning_rate": 3.7932580336167356e-06,
+      "loss": 0.395,
+      "step": 455
+    },
+    {
+      "epoch": 1.4211356466876972,
+      "grad_norm": 1.1750634908676147,
+      "learning_rate": 3.75386211692464e-06,
+      "loss": 0.3894,
+      "step": 456
+    },
+    {
+      "epoch": 1.4242902208201893,
+      "grad_norm": 1.1598374843597412,
+      "learning_rate": 3.714624522213681e-06,
+      "loss": 0.3906,
+      "step": 457
+    },
+    {
+      "epoch": 1.4274447949526814,
+      "grad_norm": 1.199683427810669,
+      "learning_rate": 3.6755462440462288e-06,
+      "loss": 0.3782,
+      "step": 458
+    },
+    {
+      "epoch": 1.4305993690851735,
+      "grad_norm": 1.3470555543899536,
+      "learning_rate": 3.636628272946434e-06,
+      "loss": 0.4017,
+      "step": 459
+    },
+    {
+      "epoch": 1.4337539432176656,
+      "grad_norm": 1.1306191682815552,
+      "learning_rate": 3.5978715953751207e-06,
+      "loss": 0.3712,
+      "step": 460
+    },
+    {
+      "epoch": 1.4369085173501577,
+      "grad_norm": 1.2298252582550049,
+      "learning_rate": 3.5592771937047633e-06,
+      "loss": 0.4142,
+      "step": 461
+    },
+    {
+      "epoch": 1.4400630914826498,
+      "grad_norm": 1.2526049613952637,
+      "learning_rate": 3.5208460461946136e-06,
+      "loss": 0.402,
+      "step": 462
+    },
+    {
+      "epoch": 1.4432176656151419,
+      "grad_norm": 1.2613544464111328,
+      "learning_rate": 3.4825791269658782e-06,
+      "loss": 0.4048,
+      "step": 463
+    },
+    {
+      "epoch": 1.446372239747634,
+      "grad_norm": 1.0886467695236206,
+      "learning_rate": 3.4444774059770537e-06,
+      "loss": 0.3776,
+      "step": 464
+    },
+    {
+      "epoch": 1.449526813880126,
+      "grad_norm": 1.1945115327835083,
+      "learning_rate": 3.4065418489993118e-06,
+      "loss": 0.4024,
+      "step": 465
+    },
+    {
+      "epoch": 1.4526813880126184,
+      "grad_norm": 1.2245378494262695,
+      "learning_rate": 3.3687734175920505e-06,
+      "loss": 0.3866,
+      "step": 466
+    },
+    {
+      "epoch": 1.4558359621451105,
+      "grad_norm": 1.2749048471450806,
+      "learning_rate": 3.3311730690784916e-06,
+      "loss": 0.4013,
+      "step": 467
+    },
+    {
+      "epoch": 1.4589905362776026,
+      "grad_norm": 1.0866283178329468,
+      "learning_rate": 3.2937417565214426e-06,
+      "loss": 0.4,
+      "step": 468
+    },
+    {
+      "epoch": 1.4621451104100947,
+      "grad_norm": 1.1891928911209106,
+      "learning_rate": 3.2564804286991137e-06,
+      "loss": 0.3867,
+      "step": 469
+    },
+    {
+      "epoch": 1.4652996845425867,
+      "grad_norm": 1.1825567483901978,
+      "learning_rate": 3.2193900300810908e-06,
+      "loss": 0.3789,
+      "step": 470
+    },
+    {
+      "epoch": 1.4684542586750788,
+      "grad_norm": 1.3308717012405396,
+      "learning_rate": 3.1824715008043874e-06,
+      "loss": 0.4014,
+      "step": 471
+    },
+    {
+      "epoch": 1.471608832807571,
+      "grad_norm": 1.1997756958007812,
+      "learning_rate": 3.145725776649602e-06,
+      "loss": 0.4114,
+      "step": 472
+    },
+    {
+      "epoch": 1.474763406940063,
+      "grad_norm": 1.1832330226898193,
+      "learning_rate": 3.1091537890172263e-06,
+      "loss": 0.4052,
+      "step": 473
+    },
+    {
+      "epoch": 1.4779179810725553,
+      "grad_norm": 1.228003740310669,
+      "learning_rate": 3.0727564649040066e-06,
+      "loss": 0.3849,
+      "step": 474
+    },
+    {
+      "epoch": 1.4810725552050474,
+      "grad_norm": 1.1024681329727173,
+      "learning_rate": 3.036534726879473e-06,
+      "loss": 0.3712,
+      "step": 475
+    },
+    {
+      "epoch": 1.4842271293375395,
+      "grad_norm": 1.2701066732406616,
+      "learning_rate": 3.000489493062535e-06,
+      "loss": 0.4225,
+      "step": 476
+    },
+    {
+      "epoch": 1.4873817034700316,
+      "grad_norm": 1.1628695726394653,
+      "learning_rate": 2.964621677098225e-06,
+      "loss": 0.391,
+      "step": 477
+    },
+    {
+      "epoch": 1.4905362776025237,
+      "grad_norm": 1.2106404304504395,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.3838,
+      "step": 478
+    },
+    {
+      "epoch": 1.4936908517350158,
+      "grad_norm": 1.1790848970413208,
+      "learning_rate": 2.8934219307993393e-06,
+      "loss": 0.4021,
+      "step": 479
+    },
+    {
+      "epoch": 1.496845425867508,
+      "grad_norm": 1.234004259109497,
+      "learning_rate": 2.8580918051775542e-06,
+      "loss": 0.4118,
+      "step": 480
+    },
+    {
+      "epoch": 1.496845425867508,
+      "eval_loss": 0.5969170331954956,
+      "eval_runtime": 665.1745,
+      "eval_samples_per_second": 5.54,
+      "eval_steps_per_second": 1.386,
+      "step": 480
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.163369059562683,
+      "learning_rate": 2.8229427067882165e-06,
+      "loss": 0.3911,
+      "step": 481
+    },
+    {
+      "epoch": 1.503154574132492,
+      "grad_norm": 1.2685822248458862,
+      "learning_rate": 2.7879755265618558e-06,
+      "loss": 0.3891,
+      "step": 482
+    },
+    {
+      "epoch": 1.5063091482649842,
+      "grad_norm": 1.04443359375,
+      "learning_rate": 2.7531911508178764e-06,
+      "loss": 0.3828,
+      "step": 483
+    },
+    {
+      "epoch": 1.5094637223974763,
+      "grad_norm": 1.1563563346862793,
+      "learning_rate": 2.7185904612421177e-06,
+      "loss": 0.4058,
+      "step": 484
+    },
+    {
+      "epoch": 1.5126182965299684,
+      "grad_norm": 1.2739872932434082,
+      "learning_rate": 2.6841743348644788e-06,
+      "loss": 0.4188,
+      "step": 485
+    },
+    {
+      "epoch": 1.5157728706624605,
+      "grad_norm": 1.2131019830703735,
+      "learning_rate": 2.6499436440367165e-06,
+      "loss": 0.3944,
+      "step": 486
+    },
+    {
+      "epoch": 1.5189274447949526,
+      "grad_norm": 1.2416787147521973,
+      "learning_rate": 2.615899256410306e-06,
+      "loss": 0.3936,
+      "step": 487
+    },
+    {
+      "epoch": 1.5220820189274447,
+      "grad_norm": 1.178523063659668,
+      "learning_rate": 2.5820420349144693e-06,
+      "loss": 0.3945,
+      "step": 488
+    },
+    {
+      "epoch": 1.5252365930599368,
+      "grad_norm": 1.2075923681259155,
+      "learning_rate": 2.548372837734299e-06,
+      "loss": 0.3671,
+      "step": 489
+    },
+    {
+      "epoch": 1.528391167192429,
+      "grad_norm": 1.1409944295883179,
+      "learning_rate": 2.514892518288988e-06,
+      "loss": 0.3952,
+      "step": 490
+    },
+    {
+      "epoch": 1.5315457413249212,
+      "grad_norm": 1.1716421842575073,
+      "learning_rate": 2.4816019252102274e-06,
+      "loss": 0.3722,
+      "step": 491
+    },
+    {
+      "epoch": 1.5347003154574133,
+      "grad_norm": 1.2308886051177979,
+      "learning_rate": 2.4485019023206637e-06,
+      "loss": 0.3976,
+      "step": 492
+    },
+    {
+      "epoch": 1.5378548895899053,
+      "grad_norm": 1.2105408906936646,
+      "learning_rate": 2.415593288612541e-06,
+      "loss": 0.3899,
+      "step": 493
+    },
+    {
+      "epoch": 1.5410094637223974,
+      "grad_norm": 1.2889288663864136,
+      "learning_rate": 2.382876918226409e-06,
+      "loss": 0.3947,
+      "step": 494
+    },
+    {
+      "epoch": 1.5441640378548895,
+      "grad_norm": 1.1646826267242432,
+      "learning_rate": 2.3503536204300017e-06,
+      "loss": 0.3935,
+      "step": 495
+    },
+    {
+      "epoch": 1.5473186119873819,
+      "grad_norm": 1.1599159240722656,
+      "learning_rate": 2.318024219597196e-06,
+      "loss": 0.4015,
+      "step": 496
+    },
+    {
+      "epoch": 1.550473186119874,
+      "grad_norm": 1.2740375995635986,
+      "learning_rate": 2.285889535187139e-06,
+      "loss": 0.3928,
+      "step": 497
+    },
+    {
+      "epoch": 1.553627760252366,
+      "grad_norm": 1.202069640159607,
+      "learning_rate": 2.2539503817234553e-06,
+      "loss": 0.3975,
+      "step": 498
+    },
+    {
+      "epoch": 1.5567823343848581,
+      "grad_norm": 1.1427030563354492,
+      "learning_rate": 2.222207568773619e-06,
+      "loss": 0.4016,
+      "step": 499
+    },
+    {
+      "epoch": 1.5599369085173502,
+      "grad_norm": 1.238408088684082,
+      "learning_rate": 2.190661900928426e-06,
+      "loss": 0.3895,
+      "step": 500
+    },
+    {
+      "epoch": 1.5630914826498423,
+      "grad_norm": 1.1599760055541992,
+      "learning_rate": 2.1593141777815906e-06,
+      "loss": 0.3767,
+      "step": 501
+    },
+    {
+      "epoch": 1.5662460567823344,
+      "grad_norm": 1.0849993228912354,
+      "learning_rate": 2.1281651939094996e-06,
+      "loss": 0.3742,
+      "step": 502
+    },
+    {
+      "epoch": 1.5694006309148265,
+      "grad_norm": 1.1617547273635864,
+      "learning_rate": 2.0972157388510484e-06,
+      "loss": 0.4007,
+      "step": 503
+    },
+    {
+      "epoch": 1.5725552050473186,
+      "grad_norm": 1.283837080001831,
+      "learning_rate": 2.0664665970876496e-06,
+      "loss": 0.4153,
+      "step": 504
+    },
+    {
+      "epoch": 1.5757097791798107,
+      "grad_norm": 1.114517331123352,
+      "learning_rate": 2.03591854802333e-06,
+      "loss": 0.3893,
+      "step": 505
+    },
+    {
+      "epoch": 1.5788643533123028,
+      "grad_norm": 1.1265556812286377,
+      "learning_rate": 2.0055723659649907e-06,
+      "loss": 0.3863,
+      "step": 506
+    },
+    {
+      "epoch": 1.5820189274447949,
+      "grad_norm": 1.1586359739303589,
+      "learning_rate": 1.975428820102766e-06,
+      "loss": 0.3909,
+      "step": 507
+    },
+    {
+      "epoch": 1.585173501577287,
+      "grad_norm": 1.0913562774658203,
+      "learning_rate": 1.94548867449054e-06,
+      "loss": 0.3885,
+      "step": 508
+    },
+    {
+      "epoch": 1.588328075709779,
+      "grad_norm": 1.1916927099227905,
+      "learning_rate": 1.9157526880265763e-06,
+      "loss": 0.4042,
+      "step": 509
+    },
+    {
+      "epoch": 1.5914826498422712,
+      "grad_norm": 1.2187116146087646,
+      "learning_rate": 1.8862216144342692e-06,
+      "loss": 0.4058,
+      "step": 510
+    },
+    {
+      "epoch": 1.5946372239747633,
+      "grad_norm": 1.0537008047103882,
+      "learning_rate": 1.8568962022430637e-06,
+      "loss": 0.3858,
+      "step": 511
+    },
+    {
+      "epoch": 1.5977917981072554,
+      "grad_norm": 1.0702356100082397,
+      "learning_rate": 1.8277771947694523e-06,
+      "loss": 0.3885,
+      "step": 512
+    },
+    {
+      "epoch": 1.6009463722397477,
+      "grad_norm": 1.1957271099090576,
+      "learning_rate": 1.7988653300981663e-06,
+      "loss": 0.3801,
+      "step": 513
+    },
+    {
+      "epoch": 1.6041009463722398,
+      "grad_norm": 1.118453025817871,
+      "learning_rate": 1.7701613410634367e-06,
+      "loss": 0.3766,
+      "step": 514
+    },
+    {
+      "epoch": 1.6072555205047319,
+      "grad_norm": 1.1698156595230103,
+      "learning_rate": 1.7416659552304449e-06,
+      "loss": 0.3786,
+      "step": 515
+    },
+    {
+      "epoch": 1.610410094637224,
+      "grad_norm": 1.1268938779830933,
+      "learning_rate": 1.7133798948768598e-06,
+      "loss": 0.3769,
+      "step": 516
+    },
+    {
+      "epoch": 1.613564668769716,
+      "grad_norm": 1.1807241439819336,
+      "learning_rate": 1.6853038769745466e-06,
+      "loss": 0.3958,
+      "step": 517
+    },
+    {
+      "epoch": 1.6167192429022084,
+      "grad_norm": 1.231959581375122,
+      "learning_rate": 1.6574386131713872e-06,
+      "loss": 0.4005,
+      "step": 518
+    },
+    {
+      "epoch": 1.6198738170347005,
+      "grad_norm": 1.1978061199188232,
+      "learning_rate": 1.6297848097732338e-06,
+      "loss": 0.4166,
+      "step": 519
+    },
+    {
+      "epoch": 1.6230283911671926,
+      "grad_norm": 1.181228518486023,
+      "learning_rate": 1.6023431677260215e-06,
+      "loss": 0.3844,
+      "step": 520
+    },
+    {
+      "epoch": 1.6261829652996846,
+      "grad_norm": 1.138068675994873,
+      "learning_rate": 1.5751143825979875e-06,
+      "loss": 0.3887,
+      "step": 521
+    },
+    {
+      "epoch": 1.6293375394321767,
+      "grad_norm": 1.2603578567504883,
+      "learning_rate": 1.5480991445620541e-06,
+      "loss": 0.3977,
+      "step": 522
+    },
+    {
+      "epoch": 1.6324921135646688,
+      "grad_norm": 1.2481218576431274,
+      "learning_rate": 1.5212981383783154e-06,
+      "loss": 0.3844,
+      "step": 523
+    },
+    {
+      "epoch": 1.635646687697161,
+      "grad_norm": 1.1707450151443481,
+      "learning_rate": 1.4947120433767048e-06,
+      "loss": 0.3605,
+      "step": 524
+    },
+    {
+      "epoch": 1.638801261829653,
+      "grad_norm": 1.1825599670410156,
+      "learning_rate": 1.4683415334397532e-06,
+      "loss": 0.3804,
+      "step": 525
+    },
+    {
+      "epoch": 1.6419558359621451,
+      "grad_norm": 1.2035417556762695,
+      "learning_rate": 1.4421872769855262e-06,
+      "loss": 0.3735,
+      "step": 526
+    },
+    {
+      "epoch": 1.6451104100946372,
+      "grad_norm": 1.179165244102478,
+      "learning_rate": 1.4162499369506655e-06,
+      "loss": 0.4025,
+      "step": 527
+    },
+    {
+      "epoch": 1.6482649842271293,
+      "grad_norm": 1.1293463706970215,
+      "learning_rate": 1.3905301707735986e-06,
+      "loss": 0.3828,
+      "step": 528
+    },
+    {
+      "epoch": 1.6514195583596214,
+      "grad_norm": 1.0890650749206543,
+      "learning_rate": 1.3650286303778715e-06,
+      "loss": 0.3944,
+      "step": 529
+    },
+    {
+      "epoch": 1.6545741324921135,
+      "grad_norm": 1.2107013463974,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.3941,
+      "step": 530
+    },
+    {
+      "epoch": 1.6577287066246056,
+      "grad_norm": 1.1603225469589233,
+      "learning_rate": 1.3146828069511708e-06,
+      "loss": 0.3941,
+      "step": 531
+    },
+    {
+      "epoch": 1.6608832807570977,
+      "grad_norm": 1.1552029848098755,
+      "learning_rate": 1.2898398000448441e-06,
+      "loss": 0.3834,
+      "step": 532
+    },
+    {
+      "epoch": 1.6640378548895898,
+      "grad_norm": 1.154603123664856,
+      "learning_rate": 1.2652175711368075e-06,
+      "loss": 0.3918,
+      "step": 533
+    },
+    {
+      "epoch": 1.6671924290220819,
+      "grad_norm": 1.1378461122512817,
+      "learning_rate": 1.2408167443311215e-06,
+      "loss": 0.3703,
+      "step": 534
+    },
+    {
+      "epoch": 1.6703470031545742,
+      "grad_norm": 1.1845020055770874,
+      "learning_rate": 1.2166379381199423e-06,
+      "loss": 0.3892,
+      "step": 535
+    },
+    {
+      "epoch": 1.6735015772870663,
+      "grad_norm": 1.1422114372253418,
+      "learning_rate": 1.1926817653678159e-06,
+      "loss": 0.3712,
+      "step": 536
+    },
+    {
+      "epoch": 1.6766561514195584,
+      "grad_norm": 1.2095820903778076,
+      "learning_rate": 1.1689488332961684e-06,
+      "loss": 0.3952,
+      "step": 537
+    },
+    {
+      "epoch": 1.6798107255520505,
+      "grad_norm": 1.217132329940796,
+      "learning_rate": 1.1454397434679022e-06,
+      "loss": 0.3986,
+      "step": 538
+    },
+    {
+      "epoch": 1.6829652996845426,
+      "grad_norm": 1.1238343715667725,
+      "learning_rate": 1.1221550917721434e-06,
+      "loss": 0.3921,
+      "step": 539
+    },
+    {
+      "epoch": 1.6861198738170347,
+      "grad_norm": 1.0890238285064697,
+      "learning_rate": 1.099095468409156e-06,
+      "loss": 0.3802,
+      "step": 540
+    },
+    {
+      "epoch": 1.689274447949527,
+      "grad_norm": 1.2087773084640503,
+      "learning_rate": 1.0762614578753571e-06,
+      "loss": 0.4019,
+      "step": 541
+    },
+    {
+      "epoch": 1.692429022082019,
+      "grad_norm": 1.165183186531067,
+      "learning_rate": 1.0536536389485275e-06,
+      "loss": 0.3653,
+      "step": 542
+    },
+    {
+      "epoch": 1.6955835962145112,
+      "grad_norm": 1.1720240116119385,
+      "learning_rate": 1.0312725846731174e-06,
+      "loss": 0.3934,
+      "step": 543
+    },
+    {
+      "epoch": 1.6987381703470033,
+      "grad_norm": 1.1540628671646118,
+      "learning_rate": 1.0091188623457415e-06,
+      "loss": 0.3734,
+      "step": 544
+    },
+    {
+      "epoch": 1.7018927444794953,
+      "grad_norm": 1.1639001369476318,
+      "learning_rate": 9.87193033500784e-07,
+      "loss": 0.3675,
+      "step": 545
+    },
+    {
+      "epoch": 1.7050473186119874,
+      "grad_norm": 1.170947790145874,
+      "learning_rate": 9.65495653896179e-07,
+      "loss": 0.3784,
+      "step": 546
+    },
+    {
+      "epoch": 1.7082018927444795,
+      "grad_norm": 1.2334061861038208,
+      "learning_rate": 9.440272734993072e-07,
+      "loss": 0.4029,
+      "step": 547
+    },
+    {
+      "epoch": 1.7113564668769716,
+      "grad_norm": 1.1448230743408203,
+      "learning_rate": 9.227884364730743e-07,
+      "loss": 0.3775,
+      "step": 548
+    },
+    {
+      "epoch": 1.7145110410094637,
+      "grad_norm": 1.183798909187317,
+      "learning_rate": 9.017796811621049e-07,
+      "loss": 0.3873,
+      "step": 549
+    },
+    {
+      "epoch": 1.7176656151419558,
+      "grad_norm": 1.0733059644699097,
+      "learning_rate": 8.810015400790994e-07,
+      "loss": 0.385,
+      "step": 550
+    },
+    {
+      "epoch": 1.720820189274448,
+      "grad_norm": 1.0583596229553223,
+      "learning_rate": 8.604545398913422e-07,
+      "loss": 0.3798,
+      "step": 551
+    },
+    {
+      "epoch": 1.72397476340694,
+      "grad_norm": 1.1049606800079346,
+      "learning_rate": 8.401392014073406e-07,
+      "loss": 0.3663,
+      "step": 552
+    },
+    {
+      "epoch": 1.727129337539432,
+      "grad_norm": 1.245421051979065,
+      "learning_rate": 8.200560395636414e-07,
+      "loss": 0.3912,
+      "step": 553
+    },
+    {
+      "epoch": 1.7302839116719242,
+      "grad_norm": 1.1275254487991333,
+      "learning_rate": 8.002055634117578e-07,
+      "loss": 0.3688,
+      "step": 554
+    },
+    {
+      "epoch": 1.7334384858044163,
+      "grad_norm": 1.2721987962722778,
+      "learning_rate": 7.805882761052852e-07,
+      "loss": 0.3913,
+      "step": 555
+    },
+    {
+      "epoch": 1.7365930599369084,
+      "grad_norm": 1.3105827569961548,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 0.3864,
+      "step": 556
+    },
+    {
+      "epoch": 1.7397476340694005,
+      "grad_norm": 1.1507766246795654,
+      "learning_rate": 7.420552510769308e-07,
+      "loss": 0.379,
+      "step": 557
+    },
+    {
+      "epoch": 1.7429022082018928,
+      "grad_norm": 1.1679091453552246,
+      "learning_rate": 7.231404900585714e-07,
+      "loss": 0.3921,
+      "step": 558
+    },
+    {
+      "epoch": 1.7460567823343849,
+      "grad_norm": 1.1608675718307495,
+      "learning_rate": 7.044608712679058e-07,
+      "loss": 0.3789,
+      "step": 559
+    },
+    {
+      "epoch": 1.749211356466877,
+      "grad_norm": 1.2368371486663818,
+      "learning_rate": 6.860168681805946e-07,
+      "loss": 0.3911,
+      "step": 560
+    },
+    {
+      "epoch": 1.749211356466877,
+      "eval_loss": 0.5901865363121033,
+      "eval_runtime": 664.9407,
+      "eval_samples_per_second": 5.542,
+      "eval_steps_per_second": 1.387,
+      "step": 560
+    },
+    {
+      "epoch": 1.752365930599369,
+      "grad_norm": 1.2072101831436157,
+      "learning_rate": 6.678089483001016e-07,
+      "loss": 0.3961,
+      "step": 561
+    },
+    {
+      "epoch": 1.7555205047318612,
+      "grad_norm": 1.0890414714813232,
+      "learning_rate": 6.498375731458529e-07,
+      "loss": 0.3583,
+      "step": 562
+    },
+    {
+      "epoch": 1.7586750788643533,
+      "grad_norm": 1.1547318696975708,
+      "learning_rate": 6.321031982415237e-07,
+      "loss": 0.3837,
+      "step": 563
+    },
+    {
+      "epoch": 1.7618296529968456,
+      "grad_norm": 1.1850864887237549,
+      "learning_rate": 6.146062731035129e-07,
+      "loss": 0.3834,
+      "step": 564
+    },
+    {
+      "epoch": 1.7649842271293377,
+      "grad_norm": 1.164358377456665,
+      "learning_rate": 5.973472412295256e-07,
+      "loss": 0.3923,
+      "step": 565
+    },
+    {
+      "epoch": 1.7681388012618298,
+      "grad_norm": 1.1400504112243652,
+      "learning_rate": 5.803265400873514e-07,
+      "loss": 0.367,
+      "step": 566
+    },
+    {
+      "epoch": 1.7712933753943219,
+      "grad_norm": 1.0794647932052612,
+      "learning_rate": 5.63544601103767e-07,
+      "loss": 0.3971,
+      "step": 567
+    },
+    {
+      "epoch": 1.774447949526814,
+      "grad_norm": 1.2106674909591675,
+      "learning_rate": 5.470018496535967e-07,
+      "loss": 0.392,
+      "step": 568
+    },
+    {
+      "epoch": 1.777602523659306,
+      "grad_norm": 1.200882911682129,
+      "learning_rate": 5.306987050489442e-07,
+      "loss": 0.3889,
+      "step": 569
+    },
+    {
+      "epoch": 1.7807570977917981,
+      "grad_norm": 1.2379915714263916,
+      "learning_rate": 5.146355805285452e-07,
+      "loss": 0.3921,
+      "step": 570
+    },
+    {
+      "epoch": 1.7839116719242902,
+      "grad_norm": 1.184643030166626,
+      "learning_rate": 4.988128832473105e-07,
+      "loss": 0.3733,
+      "step": 571
+    },
+    {
+      "epoch": 1.7870662460567823,
+      "grad_norm": 1.184597134590149,
+      "learning_rate": 4.832310142659946e-07,
+      "loss": 0.3782,
+      "step": 572
+    },
+    {
+      "epoch": 1.7902208201892744,
+      "grad_norm": 1.0617023706436157,
+      "learning_rate": 4.678903685410374e-07,
+      "loss": 0.361,
+      "step": 573
+    },
+    {
+      "epoch": 1.7933753943217665,
+      "grad_norm": 1.098720669746399,
+      "learning_rate": 4.5279133491454406e-07,
+      "loss": 0.3653,
+      "step": 574
+    },
+    {
+      "epoch": 1.7965299684542586,
+      "grad_norm": 1.1476008892059326,
+      "learning_rate": 4.3793429610444083e-07,
+      "loss": 0.3955,
+      "step": 575
+    },
+    {
+      "epoch": 1.7996845425867507,
+      "grad_norm": 1.104645013809204,
+      "learning_rate": 4.233196286947605e-07,
+      "loss": 0.3678,
+      "step": 576
+    },
+    {
+      "epoch": 1.8028391167192428,
+      "grad_norm": 1.247249960899353,
+      "learning_rate": 4.089477031261113e-07,
+      "loss": 0.3932,
+      "step": 577
+    },
+    {
+      "epoch": 1.805993690851735,
+      "grad_norm": 1.1016193628311157,
+      "learning_rate": 3.9481888368627764e-07,
+      "loss": 0.3643,
+      "step": 578
+    },
+    {
+      "epoch": 1.809148264984227,
+      "grad_norm": 1.211043357849121,
+      "learning_rate": 3.809335285009863e-07,
+      "loss": 0.3749,
+      "step": 579
+    },
+    {
+      "epoch": 1.812302839116719,
+      "grad_norm": 1.1291251182556152,
+      "learning_rate": 3.6729198952483725e-07,
+      "loss": 0.3854,
+      "step": 580
+    },
+    {
+      "epoch": 1.8154574132492114,
+      "grad_norm": 1.15535306930542,
+      "learning_rate": 3.538946125323717e-07,
+      "loss": 0.356,
+      "step": 581
+    },
+    {
+      "epoch": 1.8186119873817035,
+      "grad_norm": 1.206681489944458,
+      "learning_rate": 3.4074173710931804e-07,
+      "loss": 0.3785,
+      "step": 582
+    },
+    {
+      "epoch": 1.8217665615141956,
+      "grad_norm": 1.0925393104553223,
+      "learning_rate": 3.278336966439744e-07,
+      "loss": 0.3591,
+      "step": 583
+    },
+    {
+      "epoch": 1.8249211356466877,
+      "grad_norm": 1.289838194847107,
+      "learning_rate": 3.151708183187674e-07,
+      "loss": 0.3968,
+      "step": 584
+    },
+    {
+      "epoch": 1.8280757097791798,
+      "grad_norm": 1.131447434425354,
+      "learning_rate": 3.0275342310195133e-07,
+      "loss": 0.3826,
+      "step": 585
+    },
+    {
+      "epoch": 1.831230283911672,
+      "grad_norm": 1.0487205982208252,
+      "learning_rate": 2.905818257394799e-07,
+      "loss": 0.3614,
+      "step": 586
+    },
+    {
+      "epoch": 1.8343848580441642,
+      "grad_norm": 1.1062970161437988,
+      "learning_rate": 2.786563347470217e-07,
+      "loss": 0.386,
+      "step": 587
+    },
+    {
+      "epoch": 1.8375394321766563,
+      "grad_norm": 1.180619716644287,
+      "learning_rate": 2.6697725240214076e-07,
+      "loss": 0.3765,
+      "step": 588
+    },
+    {
+      "epoch": 1.8406940063091484,
+      "grad_norm": 1.1397796869277954,
+      "learning_rate": 2.5554487473664404e-07,
+      "loss": 0.3542,
+      "step": 589
+    },
+    {
+      "epoch": 1.8438485804416405,
+      "grad_norm": 1.1590802669525146,
+      "learning_rate": 2.4435949152906144e-07,
+      "loss": 0.3817,
+      "step": 590
+    },
+    {
+      "epoch": 1.8470031545741326,
+      "grad_norm": 1.1893994808197021,
+      "learning_rate": 2.334213862973167e-07,
+      "loss": 0.3893,
+      "step": 591
+    },
+    {
+      "epoch": 1.8501577287066246,
+      "grad_norm": 1.199181318283081,
+      "learning_rate": 2.2273083629153148e-07,
+      "loss": 0.3645,
+      "step": 592
+    },
+    {
+      "epoch": 1.8533123028391167,
+      "grad_norm": 1.1464059352874756,
+      "learning_rate": 2.122881124870002e-07,
+      "loss": 0.37,
+      "step": 593
+    },
+    {
+      "epoch": 1.8564668769716088,
+      "grad_norm": 1.2864938974380493,
+      "learning_rate": 2.0209347957732328e-07,
+      "loss": 0.3943,
+      "step": 594
+    },
+    {
+      "epoch": 1.859621451104101,
+      "grad_norm": 1.243245005607605,
+      "learning_rate": 1.921471959676957e-07,
+      "loss": 0.3814,
+      "step": 595
+    },
+    {
+      "epoch": 1.862776025236593,
+      "grad_norm": 1.3622626066207886,
+      "learning_rate": 1.8244951376835907e-07,
+      "loss": 0.3976,
+      "step": 596
+    },
+    {
+      "epoch": 1.8659305993690851,
+      "grad_norm": 1.1986589431762695,
+      "learning_rate": 1.730006787882077e-07,
+      "loss": 0.3724,
+      "step": 597
+    },
+    {
+      "epoch": 1.8690851735015772,
+      "grad_norm": 1.1340901851654053,
+      "learning_rate": 1.6380093052856482e-07,
+      "loss": 0.3675,
+      "step": 598
+    },
+    {
+      "epoch": 1.8722397476340693,
+      "grad_norm": 1.1300233602523804,
+      "learning_rate": 1.5485050217710295e-07,
+      "loss": 0.3908,
+      "step": 599
+    },
+    {
+      "epoch": 1.8753943217665614,
+      "grad_norm": 1.2240841388702393,
+      "learning_rate": 1.4614962060194303e-07,
+      "loss": 0.3898,
+      "step": 600
+    },
+    {
+      "epoch": 1.8785488958990535,
+      "grad_norm": 1.1933972835540771,
+      "learning_rate": 1.3769850634589356e-07,
+      "loss": 0.3802,
+      "step": 601
+    },
+    {
+      "epoch": 1.8817034700315456,
+      "grad_norm": 1.1771364212036133,
+      "learning_rate": 1.2949737362087156e-07,
+      "loss": 0.3664,
+      "step": 602
+    },
+    {
+      "epoch": 1.884858044164038,
+      "grad_norm": 1.2366310358047485,
+      "learning_rate": 1.2154643030246273e-07,
+      "loss": 0.3827,
+      "step": 603
+    },
+    {
+      "epoch": 1.88801261829653,
+      "grad_norm": 1.0306947231292725,
+      "learning_rate": 1.1384587792465873e-07,
+      "loss": 0.3843,
+      "step": 604
+    },
+    {
+      "epoch": 1.891167192429022,
+      "grad_norm": 1.2685048580169678,
+      "learning_rate": 1.0639591167474483e-07,
+      "loss": 0.3994,
+      "step": 605
+    },
+    {
+      "epoch": 1.8943217665615142,
+      "grad_norm": 1.2055532932281494,
+      "learning_rate": 9.919672038835926e-08,
+      "loss": 0.3845,
+      "step": 606
+    },
+    {
+      "epoch": 1.8974763406940063,
+      "grad_norm": 1.0920295715332031,
+      "learning_rate": 9.224848654469932e-08,
+      "loss": 0.38,
+      "step": 607
+    },
+    {
+      "epoch": 1.9006309148264984,
+      "grad_norm": 1.1832702159881592,
+      "learning_rate": 8.555138626189619e-08,
+      "loss": 0.3707,
+      "step": 608
+    },
+    {
+      "epoch": 1.9037854889589907,
+      "grad_norm": 1.1102246046066284,
+      "learning_rate": 7.910558929255852e-08,
+      "loss": 0.3822,
+      "step": 609
+    },
+    {
+      "epoch": 1.9069400630914828,
+      "grad_norm": 1.167602300643921,
+      "learning_rate": 7.291125901946027e-08,
+      "loss": 0.3898,
+      "step": 610
+    },
+    {
+      "epoch": 1.9100946372239749,
+      "grad_norm": 1.2090559005737305,
+      "learning_rate": 6.696855245140743e-08,
+      "loss": 0.39,
+      "step": 611
+    },
+    {
+      "epoch": 1.913249211356467,
+      "grad_norm": 1.247668743133545,
+      "learning_rate": 6.127762021925221e-08,
+      "loss": 0.3662,
+      "step": 612
+    },
+    {
+      "epoch": 1.916403785488959,
+      "grad_norm": 1.2266191244125366,
+      "learning_rate": 5.5838606572078404e-08,
+      "loss": 0.3707,
+      "step": 613
+    },
+    {
+      "epoch": 1.9195583596214512,
+      "grad_norm": 1.227341651916504,
+      "learning_rate": 5.0651649373544276e-08,
+      "loss": 0.3764,
+      "step": 614
+    },
+    {
+      "epoch": 1.9227129337539433,
+      "grad_norm": 1.174338698387146,
+      "learning_rate": 4.571688009838648e-08,
+      "loss": 0.3684,
+      "step": 615
+    },
+    {
+      "epoch": 1.9258675078864353,
+      "grad_norm": 1.1161493062973022,
+      "learning_rate": 4.103442382909051e-08,
+      "loss": 0.3653,
+      "step": 616
+    },
+    {
+      "epoch": 1.9290220820189274,
+      "grad_norm": 1.121027946472168,
+      "learning_rate": 3.660439925271764e-08,
+      "loss": 0.3614,
+      "step": 617
+    },
+    {
+      "epoch": 1.9321766561514195,
+      "grad_norm": 1.1300984621047974,
+      "learning_rate": 3.242691865790071e-08,
+      "loss": 0.3719,
+      "step": 618
+    },
+    {
+      "epoch": 1.9353312302839116,
+      "grad_norm": 1.1225863695144653,
+      "learning_rate": 2.850208793198861e-08,
+      "loss": 0.3665,
+      "step": 619
+    },
+    {
+      "epoch": 1.9384858044164037,
+      "grad_norm": 1.1923413276672363,
+      "learning_rate": 2.4830006558373975e-08,
+      "loss": 0.3882,
+      "step": 620
+    },
+    {
+      "epoch": 1.9416403785488958,
+      "grad_norm": 1.2085680961608887,
+      "learning_rate": 2.1410767613965212e-08,
+      "loss": 0.3709,
+      "step": 621
+    },
+    {
+      "epoch": 1.944794952681388,
+      "grad_norm": 1.2289031744003296,
+      "learning_rate": 1.824445776682504e-08,
+      "loss": 0.3805,
+      "step": 622
+    },
+    {
+      "epoch": 1.94794952681388,
+      "grad_norm": 1.0928528308868408,
+      "learning_rate": 1.5331157273981155e-08,
+      "loss": 0.3752,
+      "step": 623
+    },
+    {
+      "epoch": 1.951104100946372,
+      "grad_norm": 1.2201379537582397,
+      "learning_rate": 1.267093997938451e-08,
+      "loss": 0.3795,
+      "step": 624
+    },
+    {
+      "epoch": 1.9542586750788642,
+      "grad_norm": 1.23524010181427,
+      "learning_rate": 1.0263873312040818e-08,
+      "loss": 0.3747,
+      "step": 625
+    },
+    {
+      "epoch": 1.9574132492113565,
+      "grad_norm": 1.1795551776885986,
+      "learning_rate": 8.110018284304132e-09,
+      "loss": 0.386,
+      "step": 626
+    },
+    {
+      "epoch": 1.9605678233438486,
+      "grad_norm": 1.1480929851531982,
+      "learning_rate": 6.209429490323659e-09,
+      "loss": 0.3778,
+      "step": 627
+    },
+    {
+      "epoch": 1.9637223974763407,
+      "grad_norm": 1.170944094657898,
+      "learning_rate": 4.562155104665955e-09,
+      "loss": 0.3809,
+      "step": 628
+    },
+    {
+      "epoch": 1.9668769716088328,
+      "grad_norm": 1.1823114156723022,
+      "learning_rate": 3.168236881092579e-09,
+      "loss": 0.3965,
+      "step": 629
+    },
+    {
+      "epoch": 1.9700315457413249,
+      "grad_norm": 1.1339807510375977,
+      "learning_rate": 2.0277101514987184e-09,
+      "loss": 0.3691,
+      "step": 630
+    },
+    {
+      "epoch": 1.973186119873817,
+      "grad_norm": 1.177652359008789,
+      "learning_rate": 1.1406038250205699e-09,
+      "loss": 0.3842,
+      "step": 631
+    },
+    {
+      "epoch": 1.9763406940063093,
+      "grad_norm": 1.1405797004699707,
+      "learning_rate": 5.069403873025902e-10,
+      "loss": 0.3871,
+      "step": 632
+    },
+    {
+      "epoch": 1.9794952681388014,
+      "grad_norm": 1.1278390884399414,
+      "learning_rate": 1.2673589992462377e-10,
+      "loss": 0.3884,
+      "step": 633
+    },
+    {
+      "epoch": 1.9826498422712935,
+      "grad_norm": 1.1934646368026733,
+      "learning_rate": 0.0,
+      "loss": 0.3903,
+      "step": 634
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 634,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 159,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.9550153063257866e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}