diff --git "a/v1/checkpoint-850/trainer_state.json" "b/v1/checkpoint-850/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/v1/checkpoint-850/trainer_state.json"
@@ -0,0 +1,5971 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.5546410608139003,
+  "eval_steps": 500,
+  "global_step": 850,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.849355824164473,
+      "learning_rate": 4.878048780487805e-07,
+      "loss": 1.3655,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.01567518957158,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 1.5767,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6466000875559635,
+      "learning_rate": 1.4634146341463414e-06,
+      "loss": 1.3913,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6644565932010504,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 1.3218,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.571354207588475,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 1.3597,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.31036262839244955,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 1.2832,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.2622135027188184,
+      "learning_rate": 3.414634146341464e-06,
+      "loss": 1.2161,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.296824630261661,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 1.2985,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.2557267467361569,
+      "learning_rate": 4.390243902439025e-06,
+      "loss": 1.3175,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.23418939513890769,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 1.2617,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.2364760983285843,
+      "learning_rate": 5.365853658536586e-06,
+      "loss": 1.3103,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.23893034721889,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 1.2405,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.25563593295485887,
+      "learning_rate": 6.341463414634147e-06,
+      "loss": 1.2831,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.23239975352661665,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 1.3125,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.3092813858209507,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 1.2422,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.282563380367434,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 1.2453,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.22065680088315018,
+      "learning_rate": 8.292682926829268e-06,
+      "loss": 1.2491,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.22777800877980184,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 1.2655,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.22145212540177928,
+      "learning_rate": 9.268292682926831e-06,
+      "loss": 1.2413,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.22482351883112714,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 1.2653,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.20823080508385733,
+      "learning_rate": 1.024390243902439e-05,
+      "loss": 1.2374,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.26025492562935737,
+      "learning_rate": 1.0731707317073172e-05,
+      "loss": 1.2065,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.2150252124176173,
+      "learning_rate": 1.1219512195121953e-05,
+      "loss": 1.2782,
+      "step": 23
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.2505915177425618,
+      "learning_rate": 1.1707317073170731e-05,
+      "loss": 1.2742,
+      "step": 24
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.20129223044786942,
+      "learning_rate": 1.2195121951219513e-05,
+      "loss": 1.3366,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.1973508510397107,
+      "learning_rate": 1.2682926829268294e-05,
+      "loss": 1.2476,
+      "step": 26
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.27103325392437194,
+      "learning_rate": 1.3170731707317076e-05,
+      "loss": 1.2325,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17954976411006285,
+      "learning_rate": 1.3658536585365855e-05,
+      "loss": 1.2523,
+      "step": 28
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.22216997851088888,
+      "learning_rate": 1.4146341463414635e-05,
+      "loss": 1.3297,
+      "step": 29
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.2071458864548587,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 1.2127,
+      "step": 30
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.18039422081622164,
+      "learning_rate": 1.5121951219512196e-05,
+      "loss": 1.2509,
+      "step": 31
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.18631254372974412,
+      "learning_rate": 1.5609756097560978e-05,
+      "loss": 1.2247,
+      "step": 32
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.18843872523649827,
+      "learning_rate": 1.6097560975609757e-05,
+      "loss": 1.195,
+      "step": 33
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.2163847267778325,
+      "learning_rate": 1.6585365853658537e-05,
+      "loss": 1.2179,
+      "step": 34
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.19687688475496104,
+      "learning_rate": 1.7073170731707317e-05,
+      "loss": 1.2763,
+      "step": 35
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.20409643064887947,
+      "learning_rate": 1.75609756097561e-05,
+      "loss": 1.253,
+      "step": 36
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.1879182661759335,
+      "learning_rate": 1.804878048780488e-05,
+      "loss": 1.2586,
+      "step": 37
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19400648948514373,
+      "learning_rate": 1.8536585365853663e-05,
+      "loss": 1.2154,
+      "step": 38
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.1878879343148452,
+      "learning_rate": 1.902439024390244e-05,
+      "loss": 1.2304,
+      "step": 39
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.17687475469924052,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 1.2351,
+      "step": 40
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.18223935625384885,
+      "learning_rate": 2e-05,
+      "loss": 1.2222,
+      "step": 41
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.1943061629408338,
+      "learning_rate": 2.048780487804878e-05,
+      "loss": 1.2044,
+      "step": 42
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.17027514338700078,
+      "learning_rate": 2.0975609756097564e-05,
+      "loss": 1.1548,
+      "step": 43
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.18553769630586192,
+      "learning_rate": 2.1463414634146344e-05,
+      "loss": 1.2721,
+      "step": 44
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.19732826914228765,
+      "learning_rate": 2.1951219512195124e-05,
+      "loss": 1.3097,
+      "step": 45
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.18714230986631472,
+      "learning_rate": 2.2439024390243907e-05,
+      "loss": 1.2662,
+      "step": 46
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.19988987568002223,
+      "learning_rate": 2.2926829268292683e-05,
+      "loss": 1.2904,
+      "step": 47
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.17744650133390918,
+      "learning_rate": 2.3414634146341463e-05,
+      "loss": 1.1825,
+      "step": 48
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.16576734763834533,
+      "learning_rate": 2.3902439024390246e-05,
+      "loss": 1.1858,
+      "step": 49
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.179591794065527,
+      "learning_rate": 2.4390243902439026e-05,
+      "loss": 1.2711,
+      "step": 50
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.17923464471176911,
+      "learning_rate": 2.4878048780487805e-05,
+      "loss": 1.2289,
+      "step": 51
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.18991742907836837,
+      "learning_rate": 2.536585365853659e-05,
+      "loss": 1.3097,
+      "step": 52
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.19849796137254636,
+      "learning_rate": 2.5853658536585368e-05,
+      "loss": 1.2489,
+      "step": 53
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.17452371110976383,
+      "learning_rate": 2.634146341463415e-05,
+      "loss": 1.2461,
+      "step": 54
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.17671022353085036,
+      "learning_rate": 2.682926829268293e-05,
+      "loss": 1.153,
+      "step": 55
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.36820559192096686,
+      "learning_rate": 2.731707317073171e-05,
+      "loss": 1.2431,
+      "step": 56
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.20331468526494198,
+      "learning_rate": 2.7804878048780487e-05,
+      "loss": 1.2575,
+      "step": 57
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2402486598118377,
+      "learning_rate": 2.829268292682927e-05,
+      "loss": 1.2538,
+      "step": 58
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2549409484173144,
+      "learning_rate": 2.878048780487805e-05,
+      "loss": 1.2065,
+      "step": 59
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2053105349872685,
+      "learning_rate": 2.926829268292683e-05,
+      "loss": 1.2094,
+      "step": 60
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.17971910872957886,
+      "learning_rate": 2.9756097560975613e-05,
+      "loss": 1.228,
+      "step": 61
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1885853654992973,
+      "learning_rate": 3.0243902439024392e-05,
+      "loss": 1.2286,
+      "step": 62
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.1848524571968613,
+      "learning_rate": 3.073170731707317e-05,
+      "loss": 1.2718,
+      "step": 63
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.18734105883548513,
+      "learning_rate": 3.1219512195121955e-05,
+      "loss": 1.2357,
+      "step": 64
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.17774668052121825,
+      "learning_rate": 3.170731707317074e-05,
+      "loss": 1.1509,
+      "step": 65
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.17890968008080646,
+      "learning_rate": 3.2195121951219514e-05,
+      "loss": 1.1924,
+      "step": 66
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.18249273371332375,
+      "learning_rate": 3.268292682926829e-05,
+      "loss": 1.2545,
+      "step": 67
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.21064122671902577,
+      "learning_rate": 3.3170731707317074e-05,
+      "loss": 1.2832,
+      "step": 68
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.1820064171955093,
+      "learning_rate": 3.365853658536586e-05,
+      "loss": 1.2071,
+      "step": 69
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.16996662800553433,
+      "learning_rate": 3.414634146341463e-05,
+      "loss": 1.2073,
+      "step": 70
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.1618669302922445,
+      "learning_rate": 3.4634146341463416e-05,
+      "loss": 1.1289,
+      "step": 71
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.18948744950985544,
+      "learning_rate": 3.51219512195122e-05,
+      "loss": 1.2915,
+      "step": 72
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.18326143691603383,
+      "learning_rate": 3.5609756097560976e-05,
+      "loss": 1.2238,
+      "step": 73
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.17410704510700503,
+      "learning_rate": 3.609756097560976e-05,
+      "loss": 1.1784,
+      "step": 74
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.1983667344995625,
+      "learning_rate": 3.658536585365854e-05,
+      "loss": 1.2452,
+      "step": 75
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3416310763369357,
+      "learning_rate": 3.7073170731707325e-05,
+      "loss": 1.1972,
+      "step": 76
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2776466983511955,
+      "learning_rate": 3.75609756097561e-05,
+      "loss": 1.3121,
+      "step": 77
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.20026129636576834,
+      "learning_rate": 3.804878048780488e-05,
+      "loss": 1.2436,
+      "step": 78
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.21064549243917835,
+      "learning_rate": 3.853658536585366e-05,
+      "loss": 1.2064,
+      "step": 79
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.22119482175714267,
+      "learning_rate": 3.9024390243902444e-05,
+      "loss": 1.2715,
+      "step": 80
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.23047133748844142,
+      "learning_rate": 3.951219512195122e-05,
+      "loss": 1.2888,
+      "step": 81
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.18741863156973176,
+      "learning_rate": 4e-05,
+      "loss": 1.248,
+      "step": 82
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.1747859810629604,
+      "learning_rate": 4.0487804878048786e-05,
+      "loss": 1.1683,
+      "step": 83
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.1896944798413341,
+      "learning_rate": 4.097560975609756e-05,
+      "loss": 1.2155,
+      "step": 84
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.18724128114363303,
+      "learning_rate": 4.1463414634146346e-05,
+      "loss": 1.2273,
+      "step": 85
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.17368125504855478,
+      "learning_rate": 4.195121951219513e-05,
+      "loss": 1.224,
+      "step": 86
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.18371141013625703,
+      "learning_rate": 4.2439024390243905e-05,
+      "loss": 1.2294,
+      "step": 87
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.1791029365673714,
+      "learning_rate": 4.292682926829269e-05,
+      "loss": 1.2895,
+      "step": 88
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.20259974283859655,
+      "learning_rate": 4.341463414634147e-05,
+      "loss": 1.1841,
+      "step": 89
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.17457456183272174,
+      "learning_rate": 4.390243902439025e-05,
+      "loss": 1.2357,
+      "step": 90
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.1815824380789748,
+      "learning_rate": 4.439024390243903e-05,
+      "loss": 1.2304,
+      "step": 91
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.17566480599583392,
+      "learning_rate": 4.4878048780487814e-05,
+      "loss": 1.242,
+      "step": 92
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.18422975005984474,
+      "learning_rate": 4.536585365853658e-05,
+      "loss": 1.2177,
+      "step": 93
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.16796781877940678,
+      "learning_rate": 4.5853658536585366e-05,
+      "loss": 1.1482,
+      "step": 94
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.18636131653783305,
+      "learning_rate": 4.634146341463415e-05,
+      "loss": 1.1758,
+      "step": 95
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.1823665700289814,
+      "learning_rate": 4.6829268292682926e-05,
+      "loss": 1.289,
+      "step": 96
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.1719900691262439,
+      "learning_rate": 4.731707317073171e-05,
+      "loss": 1.1626,
+      "step": 97
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.17937994168039778,
+      "learning_rate": 4.780487804878049e-05,
+      "loss": 1.175,
+      "step": 98
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.16631851422106986,
+      "learning_rate": 4.829268292682927e-05,
+      "loss": 1.2177,
+      "step": 99
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.19143696232800309,
+      "learning_rate": 4.878048780487805e-05,
+      "loss": 1.3071,
+      "step": 100
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.17859506638780318,
+      "learning_rate": 4.9268292682926835e-05,
+      "loss": 1.2351,
+      "step": 101
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.18381520321248196,
+      "learning_rate": 4.975609756097561e-05,
+      "loss": 1.2342,
+      "step": 102
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.17968218683773912,
+      "learning_rate": 5.0243902439024394e-05,
+      "loss": 1.2074,
+      "step": 103
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.18139489969339018,
+      "learning_rate": 5.073170731707318e-05,
+      "loss": 1.1558,
+      "step": 104
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.17366624842514394,
+      "learning_rate": 5.121951219512195e-05,
+      "loss": 1.1897,
+      "step": 105
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.16034845455223745,
+      "learning_rate": 5.1707317073170736e-05,
+      "loss": 1.179,
+      "step": 106
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.17583069577827776,
+      "learning_rate": 5.219512195121952e-05,
+      "loss": 1.1856,
+      "step": 107
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.1853758076989552,
+      "learning_rate": 5.26829268292683e-05,
+      "loss": 1.2072,
+      "step": 108
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.19597443965936462,
+      "learning_rate": 5.317073170731708e-05,
+      "loss": 1.2271,
+      "step": 109
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.1899206334098331,
+      "learning_rate": 5.365853658536586e-05,
+      "loss": 1.1961,
+      "step": 110
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.17463763837757018,
+      "learning_rate": 5.4146341463414645e-05,
+      "loss": 1.2049,
+      "step": 111
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.20431371701229986,
+      "learning_rate": 5.463414634146342e-05,
+      "loss": 1.2891,
+      "step": 112
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.1814475107638498,
+      "learning_rate": 5.51219512195122e-05,
+      "loss": 1.2346,
+      "step": 113
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.1883849423207823,
+      "learning_rate": 5.5609756097560974e-05,
+      "loss": 1.244,
+      "step": 114
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.1857258128640568,
+      "learning_rate": 5.609756097560976e-05,
+      "loss": 1.2669,
+      "step": 115
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.1740768514118401,
+      "learning_rate": 5.658536585365854e-05,
+      "loss": 1.2414,
+      "step": 116
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.1919320335584178,
+      "learning_rate": 5.7073170731707317e-05,
+      "loss": 1.2886,
+      "step": 117
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.18288775167828136,
+      "learning_rate": 5.75609756097561e-05,
+      "loss": 1.1875,
+      "step": 118
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.18208588867750863,
+      "learning_rate": 5.804878048780488e-05,
+      "loss": 1.2388,
+      "step": 119
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.1743260015658331,
+      "learning_rate": 5.853658536585366e-05,
+      "loss": 1.1762,
+      "step": 120
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.17856046291517946,
+      "learning_rate": 5.902439024390244e-05,
+      "loss": 1.2888,
+      "step": 121
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.17493794870966536,
+      "learning_rate": 5.9512195121951225e-05,
+      "loss": 1.2222,
+      "step": 122
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.1909202655203384,
+      "learning_rate": 6.000000000000001e-05,
+      "loss": 1.2414,
+      "step": 123
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.18345819482834988,
+      "learning_rate": 6.0487804878048785e-05,
+      "loss": 1.2756,
+      "step": 124
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2057069352956621,
+      "learning_rate": 6.097560975609757e-05,
+      "loss": 1.261,
+      "step": 125
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.299775882469108,
+      "learning_rate": 6.146341463414634e-05,
+      "loss": 1.2566,
+      "step": 126
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.1869687633018095,
+      "learning_rate": 6.195121951219513e-05,
+      "loss": 1.3039,
+      "step": 127
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.17747149926197442,
+      "learning_rate": 6.243902439024391e-05,
+      "loss": 1.2524,
+      "step": 128
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.17885157788044242,
+      "learning_rate": 6.29268292682927e-05,
+      "loss": 1.2455,
+      "step": 129
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.17617298187845123,
+      "learning_rate": 6.341463414634148e-05,
+      "loss": 1.2009,
+      "step": 130
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.20164176323497066,
+      "learning_rate": 6.390243902439025e-05,
+      "loss": 1.2634,
+      "step": 131
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.20459903417307612,
+      "learning_rate": 6.439024390243903e-05,
+      "loss": 1.1963,
+      "step": 132
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.1863755486334296,
+      "learning_rate": 6.487804878048781e-05,
+      "loss": 1.2387,
+      "step": 133
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.19265866140295207,
+      "learning_rate": 6.536585365853658e-05,
+      "loss": 1.2688,
+      "step": 134
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.1823425868969493,
+      "learning_rate": 6.585365853658536e-05,
+      "loss": 1.2041,
+      "step": 135
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2016853266472781,
+      "learning_rate": 6.634146341463415e-05,
+      "loss": 1.1223,
+      "step": 136
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.17282675192463448,
+      "learning_rate": 6.682926829268293e-05,
+      "loss": 1.1879,
+      "step": 137
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.17398811693399288,
+      "learning_rate": 6.731707317073171e-05,
+      "loss": 1.2682,
+      "step": 138
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.18516916965434696,
+      "learning_rate": 6.78048780487805e-05,
+      "loss": 1.1666,
+      "step": 139
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.1852213129647933,
+      "learning_rate": 6.829268292682927e-05,
+      "loss": 1.2501,
+      "step": 140
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.17915948766591883,
+      "learning_rate": 6.878048780487805e-05,
+      "loss": 1.2264,
+      "step": 141
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.21599939417233183,
+      "learning_rate": 6.926829268292683e-05,
+      "loss": 1.2376,
+      "step": 142
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.17839304459521851,
+      "learning_rate": 6.975609756097562e-05,
+      "loss": 1.2353,
+      "step": 143
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.20826913231380875,
+      "learning_rate": 7.02439024390244e-05,
+      "loss": 1.1901,
+      "step": 144
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.20788894913361589,
+      "learning_rate": 7.073170731707318e-05,
+      "loss": 1.2577,
+      "step": 145
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.18420055842301297,
+      "learning_rate": 7.121951219512195e-05,
+      "loss": 1.1393,
+      "step": 146
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.19903048468685589,
+      "learning_rate": 7.170731707317073e-05,
+      "loss": 1.2321,
+      "step": 147
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.19074116314985748,
+      "learning_rate": 7.219512195121952e-05,
+      "loss": 1.1912,
+      "step": 148
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2353816469403903,
+      "learning_rate": 7.26829268292683e-05,
+      "loss": 1.28,
+      "step": 149
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.21634875684769345,
+      "learning_rate": 7.317073170731708e-05,
+      "loss": 1.3312,
+      "step": 150
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.18290969006743918,
+      "learning_rate": 7.365853658536587e-05,
+      "loss": 1.2214,
+      "step": 151
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.18484243897545208,
+      "learning_rate": 7.414634146341465e-05,
+      "loss": 1.1895,
+      "step": 152
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.21882343112978872,
+      "learning_rate": 7.463414634146342e-05,
+      "loss": 1.2219,
+      "step": 153
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.19868284379241205,
+      "learning_rate": 7.51219512195122e-05,
+      "loss": 1.2176,
+      "step": 154
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.20912516312950613,
+      "learning_rate": 7.560975609756097e-05,
+      "loss": 1.242,
+      "step": 155
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.23811880045549916,
+      "learning_rate": 7.609756097560976e-05,
+      "loss": 1.2838,
+      "step": 156
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.19511077122033713,
+      "learning_rate": 7.658536585365854e-05,
+      "loss": 1.1594,
+      "step": 157
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.20094129399534238,
+      "learning_rate": 7.707317073170732e-05,
+      "loss": 1.2966,
+      "step": 158
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.19366245038292418,
+      "learning_rate": 7.75609756097561e-05,
+      "loss": 1.2246,
+      "step": 159
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.19409570223867306,
+      "learning_rate": 7.804878048780489e-05,
+      "loss": 1.2312,
+      "step": 160
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.2087258457033805,
+      "learning_rate": 7.853658536585366e-05,
+      "loss": 1.2169,
+      "step": 161
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.18765223996270428,
+      "learning_rate": 7.902439024390244e-05,
+      "loss": 1.2383,
+      "step": 162
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.20734180224147242,
+      "learning_rate": 7.951219512195122e-05,
+      "loss": 1.2587,
+      "step": 163
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.24690929540287834,
+      "learning_rate": 8e-05,
+      "loss": 1.1951,
+      "step": 164
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2003538797619543,
+      "learning_rate": 7.999990914797545e-05,
+      "loss": 1.1982,
+      "step": 165
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.22469075613510484,
+      "learning_rate": 7.99996365923145e-05,
+      "loss": 1.2355,
+      "step": 166
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.21870100788336058,
+      "learning_rate": 7.999918233425526e-05,
+      "loss": 1.1103,
+      "step": 167
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.20939989594131886,
+      "learning_rate": 7.999854637586122e-05,
+      "loss": 1.1966,
+      "step": 168
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.43108211416237796,
+      "learning_rate": 7.999772872002132e-05,
+      "loss": 1.2882,
+      "step": 169
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.27045413432174487,
+      "learning_rate": 7.999672937044984e-05,
+      "loss": 1.2399,
+      "step": 170
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.19700483036740515,
+      "learning_rate": 7.999554833168642e-05,
+      "loss": 1.202,
+      "step": 171
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3335979493370708,
+      "learning_rate": 7.999418560909604e-05,
+      "loss": 1.1995,
+      "step": 172
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3165803974474567,
+      "learning_rate": 7.999264120886902e-05,
+      "loss": 1.1569,
+      "step": 173
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.1951699080346223,
+      "learning_rate": 7.999091513802093e-05,
+      "loss": 1.1778,
+      "step": 174
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.2087559121749787,
+      "learning_rate": 7.998900740439265e-05,
+      "loss": 1.1736,
+      "step": 175
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20345180977460478,
+      "learning_rate": 7.998691801665024e-05,
+      "loss": 1.2281,
+      "step": 176
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.24617644827252333,
+      "learning_rate": 7.998464698428495e-05,
+      "loss": 1.2072,
+      "step": 177
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.2469050959356265,
+      "learning_rate": 7.998219431761318e-05,
+      "loss": 1.2242,
+      "step": 178
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.19529317748460623,
+      "learning_rate": 7.997956002777642e-05,
+      "loss": 1.2567,
+      "step": 179
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.19048389491381376,
+      "learning_rate": 7.99767441267412e-05,
+      "loss": 1.2982,
+      "step": 180
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.2085799116493225,
+      "learning_rate": 7.997374662729904e-05,
+      "loss": 1.1254,
+      "step": 181
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.20636853256378995,
+      "learning_rate": 7.997056754306636e-05,
+      "loss": 1.2435,
+      "step": 182
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.20590016382290252,
+      "learning_rate": 7.99672068884845e-05,
+      "loss": 1.2658,
+      "step": 183
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1931166169764433,
+      "learning_rate": 7.996366467881955e-05,
+      "loss": 1.1637,
+      "step": 184
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.18873318157988098,
+      "learning_rate": 7.995994093016237e-05,
+      "loss": 1.1335,
+      "step": 185
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.19210254625199108,
+      "learning_rate": 7.995603565942846e-05,
+      "loss": 1.1928,
+      "step": 186
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.2130986479765664,
+      "learning_rate": 7.995194888435792e-05,
+      "loss": 1.2158,
+      "step": 187
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.22003854501814088,
+      "learning_rate": 7.994768062351532e-05,
+      "loss": 1.2288,
+      "step": 188
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.20330803191993058,
+      "learning_rate": 7.994323089628968e-05,
+      "loss": 1.2426,
+      "step": 189
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.20567314642208634,
+      "learning_rate": 7.993859972289434e-05,
+      "loss": 1.2649,
+      "step": 190
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.21556663727342962,
+      "learning_rate": 7.993378712436686e-05,
+      "loss": 1.2545,
+      "step": 191
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.20309165469109888,
+      "learning_rate": 7.992879312256897e-05,
+      "loss": 1.3338,
+      "step": 192
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.19574356669421325,
+      "learning_rate": 7.992361774018641e-05,
+      "loss": 1.278,
+      "step": 193
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.2763613746722313,
+      "learning_rate": 7.991826100072891e-05,
+      "loss": 1.2571,
+      "step": 194
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.19346552479915102,
+      "learning_rate": 7.991272292852996e-05,
+      "loss": 1.2027,
+      "step": 195
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.2281167812123908,
+      "learning_rate": 7.990700354874683e-05,
+      "loss": 1.2586,
+      "step": 196
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.19699013712137542,
+      "learning_rate": 7.990110288736042e-05,
+      "loss": 1.1371,
+      "step": 197
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21768209981475933,
+      "learning_rate": 7.989502097117503e-05,
+      "loss": 1.2522,
+      "step": 198
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21335427847754582,
+      "learning_rate": 7.988875782781838e-05,
+      "loss": 1.2437,
+      "step": 199
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.21856710629066897,
+      "learning_rate": 7.988231348574147e-05,
+      "loss": 1.2135,
+      "step": 200
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.20482062658774797,
+      "learning_rate": 7.987568797421836e-05,
+      "loss": 1.1755,
+      "step": 201
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.2017756813960897,
+      "learning_rate": 7.986888132334608e-05,
+      "loss": 1.1699,
+      "step": 202
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.20496443848153809,
+      "learning_rate": 7.986189356404458e-05,
+      "loss": 1.2125,
+      "step": 203
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.2134603800558358,
+      "learning_rate": 7.985472472805643e-05,
+      "loss": 1.2391,
+      "step": 204
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.2364175573420861,
+      "learning_rate": 7.98473748479468e-05,
+      "loss": 1.2384,
+      "step": 205
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1872419861598724,
+      "learning_rate": 7.983984395710326e-05,
+      "loss": 1.1457,
+      "step": 206
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.28222194007095774,
+      "learning_rate": 7.983213208973566e-05,
+      "loss": 1.2952,
+      "step": 207
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1916094851162064,
+      "learning_rate": 7.982423928087593e-05,
+      "loss": 1.1763,
+      "step": 208
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.18446245256166657,
+      "learning_rate": 7.981616556637795e-05,
+      "loss": 1.1863,
+      "step": 209
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.195191961022491,
+      "learning_rate": 7.980791098291737e-05,
+      "loss": 1.2036,
+      "step": 210
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.2652439657825496,
+      "learning_rate": 7.979947556799151e-05,
+      "loss": 1.2834,
+      "step": 211
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.24308438957843412,
+      "learning_rate": 7.979085935991906e-05,
+      "loss": 1.234,
+      "step": 212
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.21294701043622016,
+      "learning_rate": 7.978206239784004e-05,
+      "loss": 1.3006,
+      "step": 213
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.25809277041859524,
+      "learning_rate": 7.977308472171553e-05,
+      "loss": 1.2272,
+      "step": 214
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.193463860107294,
+      "learning_rate": 7.976392637232754e-05,
+      "loss": 1.2295,
+      "step": 215
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.2150023760609626,
+      "learning_rate": 7.975458739127877e-05,
+      "loss": 1.2135,
+      "step": 216
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.22590495955605894,
+      "learning_rate": 7.974506782099253e-05,
+      "loss": 1.2532,
+      "step": 217
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.21023744668403702,
+      "learning_rate": 7.973536770471242e-05,
+      "loss": 1.2472,
+      "step": 218
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.2345749799511543,
+      "learning_rate": 7.972548708650218e-05,
+      "loss": 1.1791,
+      "step": 219
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.2158876734005217,
+      "learning_rate": 7.971542601124553e-05,
+      "loss": 1.2483,
+      "step": 220
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.29455339949432446,
+      "learning_rate": 7.970518452464593e-05,
+      "loss": 1.2894,
+      "step": 221
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23983708730626851,
+      "learning_rate": 7.969476267322636e-05,
+      "loss": 1.271,
+      "step": 222
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.1922400905426158,
+      "learning_rate": 7.968416050432912e-05,
+      "loss": 1.2139,
+      "step": 223
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.2238136844422931,
+      "learning_rate": 7.967337806611568e-05,
+      "loss": 1.2655,
+      "step": 224
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.21230292828267672,
+      "learning_rate": 7.966241540756631e-05,
+      "loss": 1.2406,
+      "step": 225
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.26656119419070456,
+      "learning_rate": 7.965127257848004e-05,
+      "loss": 1.2595,
+      "step": 226
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.22381385502992684,
+      "learning_rate": 7.963994962947426e-05,
+      "loss": 1.1737,
+      "step": 227
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.20056702203994298,
+      "learning_rate": 7.962844661198462e-05,
+      "loss": 1.1969,
+      "step": 228
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.20148701321526885,
+      "learning_rate": 7.961676357826478e-05,
+      "loss": 1.2151,
+      "step": 229
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.20034834807028637,
+      "learning_rate": 7.960490058138604e-05,
+      "loss": 1.1455,
+      "step": 230
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.21050838521846033,
+      "learning_rate": 7.959285767523732e-05,
+      "loss": 1.2223,
+      "step": 231
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.20904772138969777,
+      "learning_rate": 7.95806349145247e-05,
+      "loss": 1.2534,
+      "step": 232
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.20307877304792957,
+      "learning_rate": 7.956823235477134e-05,
+      "loss": 1.1352,
+      "step": 233
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.20501105270897094,
+      "learning_rate": 7.95556500523171e-05,
+      "loss": 1.2031,
+      "step": 234
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.19800586972038586,
+      "learning_rate": 7.954288806431838e-05,
+      "loss": 1.2567,
+      "step": 235
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.2175102450594135,
+      "learning_rate": 7.952994644874777e-05,
+      "loss": 1.2538,
+      "step": 236
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.22698189300067595,
+      "learning_rate": 7.951682526439391e-05,
+      "loss": 1.3088,
+      "step": 237
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.19208392014975315,
+      "learning_rate": 7.950352457086109e-05,
+      "loss": 1.2336,
+      "step": 238
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.27004086334319655,
+      "learning_rate": 7.949004442856905e-05,
+      "loss": 1.2012,
+      "step": 239
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.23420974954538043,
+      "learning_rate": 7.947638489875272e-05,
+      "loss": 1.2244,
+      "step": 240
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.20514399124802024,
+      "learning_rate": 7.946254604346186e-05,
+      "loss": 1.2548,
+      "step": 241
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.19334973602372896,
+      "learning_rate": 7.944852792556092e-05,
+      "loss": 1.2104,
+      "step": 242
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.1992640714537956,
+      "learning_rate": 7.943433060872858e-05,
+      "loss": 1.2628,
+      "step": 243
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.203284617090413,
+      "learning_rate": 7.941995415745761e-05,
+      "loss": 1.2002,
+      "step": 244
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.22795306969682058,
+      "learning_rate": 7.94053986370545e-05,
+      "loss": 1.2215,
+      "step": 245
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.20789041346838505,
+      "learning_rate": 7.939066411363915e-05,
+      "loss": 1.0998,
+      "step": 246
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.22354868884742066,
+      "learning_rate": 7.937575065414464e-05,
+      "loss": 1.2564,
+      "step": 247
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.21176392726647736,
+      "learning_rate": 7.936065832631687e-05,
+      "loss": 1.2816,
+      "step": 248
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.19967179557235587,
+      "learning_rate": 7.934538719871427e-05,
+      "loss": 1.1961,
+      "step": 249
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.210819577350627,
+      "learning_rate": 7.932993734070747e-05,
+      "loss": 1.2167,
+      "step": 250
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.21537794551756187,
+      "learning_rate": 7.931430882247903e-05,
+      "loss": 1.2341,
+      "step": 251
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.22850872387256574,
+      "learning_rate": 7.929850171502304e-05,
+      "loss": 1.1686,
+      "step": 252
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.22380366415076383,
+      "learning_rate": 7.928251609014493e-05,
+      "loss": 1.1462,
+      "step": 253
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.22426923149036065,
+      "learning_rate": 7.926635202046102e-05,
+      "loss": 1.1792,
+      "step": 254
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.42082703321103965,
+      "learning_rate": 7.925000957939822e-05,
+      "loss": 1.2718,
+      "step": 255
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2235432774854074,
+      "learning_rate": 7.92334888411937e-05,
+      "loss": 1.2598,
+      "step": 256
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.281644028934108,
+      "learning_rate": 7.92167898808946e-05,
+      "loss": 1.2205,
+      "step": 257
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2037705143888748,
+      "learning_rate": 7.919991277435763e-05,
+      "loss": 1.1737,
+      "step": 258
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.20917419230028977,
+      "learning_rate": 7.918285759824879e-05,
+      "loss": 1.2035,
+      "step": 259
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.20510847570635518,
+      "learning_rate": 7.916562443004292e-05,
+      "loss": 1.2135,
+      "step": 260
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.25172483071092466,
+      "learning_rate": 7.914821334802342e-05,
+      "loss": 1.2218,
+      "step": 261
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.21102706700634313,
+      "learning_rate": 7.91306244312819e-05,
+      "loss": 1.1738,
+      "step": 262
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.22626060872645815,
+      "learning_rate": 7.911285775971781e-05,
+      "loss": 1.238,
+      "step": 263
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.22448567539778486,
+      "learning_rate": 7.909491341403805e-05,
+      "loss": 1.2404,
+      "step": 264
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2019099786139193,
+      "learning_rate": 7.907679147575661e-05,
+      "loss": 1.213,
+      "step": 265
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.24307234839096267,
+      "learning_rate": 7.905849202719422e-05,
+      "loss": 1.2322,
+      "step": 266
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.19801890521743487,
+      "learning_rate": 7.904001515147802e-05,
+      "loss": 1.2448,
+      "step": 267
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2102742273575385,
+      "learning_rate": 7.902136093254106e-05,
+      "loss": 1.1657,
+      "step": 268
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2173464476815016,
+      "learning_rate": 7.900252945512201e-05,
+      "loss": 1.2549,
+      "step": 269
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.20957275458699595,
+      "learning_rate": 7.898352080476479e-05,
+      "loss": 1.2536,
+      "step": 270
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20691966388952363,
+      "learning_rate": 7.896433506781811e-05,
+      "loss": 1.2661,
+      "step": 271
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.2276662275112648,
+      "learning_rate": 7.894497233143509e-05,
+      "loss": 1.2409,
+      "step": 272
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.23854109569301263,
+      "learning_rate": 7.892543268357297e-05,
+      "loss": 1.2681,
+      "step": 273
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.2233864156677627,
+      "learning_rate": 7.890571621299252e-05,
+      "loss": 1.1687,
+      "step": 274
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20114129147925475,
+      "learning_rate": 7.888582300925787e-05,
+      "loss": 1.2184,
+      "step": 275
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.2154654670569462,
+      "learning_rate": 7.886575316273586e-05,
+      "loss": 1.1982,
+      "step": 276
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2292982209343639,
+      "learning_rate": 7.884550676459583e-05,
+      "loss": 1.2129,
+      "step": 277
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.21302713135229548,
+      "learning_rate": 7.882508390680908e-05,
+      "loss": 1.1605,
+      "step": 278
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2123661020671048,
+      "learning_rate": 7.88044846821485e-05,
+      "loss": 1.2308,
+      "step": 279
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2080577410800404,
+      "learning_rate": 7.878370918418818e-05,
+      "loss": 1.2195,
+      "step": 280
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.19663901881127385,
+      "learning_rate": 7.876275750730289e-05,
+      "loss": 1.1591,
+      "step": 281
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.20534502031312163,
+      "learning_rate": 7.874162974666776e-05,
+      "loss": 1.2664,
+      "step": 282
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.23240445399513837,
+      "learning_rate": 7.872032599825779e-05,
+      "loss": 1.2151,
+      "step": 283
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2672527316717507,
+      "learning_rate": 7.86988463588474e-05,
+      "loss": 1.2406,
+      "step": 284
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.19893903058743695,
+      "learning_rate": 7.867719092601003e-05,
+      "loss": 1.1291,
+      "step": 285
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.33275268109930917,
+      "learning_rate": 7.865535979811768e-05,
+      "loss": 1.1406,
+      "step": 286
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2373619455690358,
+      "learning_rate": 7.863335307434045e-05,
+      "loss": 1.2799,
+      "step": 287
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.263235735390858,
+      "learning_rate": 7.861117085464612e-05,
+      "loss": 1.2415,
+      "step": 288
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.25884281780784324,
+      "learning_rate": 7.858881323979965e-05,
+      "loss": 1.3919,
+      "step": 289
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.25426288332255736,
+      "learning_rate": 7.85662803313628e-05,
+      "loss": 1.174,
+      "step": 290
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.26655405527881243,
+      "learning_rate": 7.854357223169356e-05,
+      "loss": 1.2806,
+      "step": 291
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.20909844432349833,
+      "learning_rate": 7.852068904394579e-05,
+      "loss": 1.2627,
+      "step": 292
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.21307115068935759,
+      "learning_rate": 7.849763087206866e-05,
+      "loss": 1.1879,
+      "step": 293
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.25009949471398946,
+      "learning_rate": 7.847439782080628e-05,
+      "loss": 1.2881,
+      "step": 294
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.20960783418679174,
+      "learning_rate": 7.845098999569712e-05,
+      "loss": 1.2723,
+      "step": 295
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.24968832437925104,
+      "learning_rate": 7.842740750307362e-05,
+      "loss": 1.2029,
+      "step": 296
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.22981196585125677,
+      "learning_rate": 7.84036504500616e-05,
+      "loss": 1.1695,
+      "step": 297
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2320606844751365,
+      "learning_rate": 7.837971894457991e-05,
+      "loss": 1.2317,
+      "step": 298
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.23051459673906124,
+      "learning_rate": 7.835561309533981e-05,
+      "loss": 1.2046,
+      "step": 299
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2510027231060586,
+      "learning_rate": 7.833133301184457e-05,
+      "loss": 1.199,
+      "step": 300
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.23601180466018787,
+      "learning_rate": 7.830687880438895e-05,
+      "loss": 1.1755,
+      "step": 301
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24740820934385369,
+      "learning_rate": 7.828225058405864e-05,
+      "loss": 1.2054,
+      "step": 302
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.23065372979111173,
+      "learning_rate": 7.825744846272984e-05,
+      "loss": 1.2066,
+      "step": 303
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.22385077334838213,
+      "learning_rate": 7.823247255306866e-05,
+      "loss": 1.2147,
+      "step": 304
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.42981213948386104,
+      "learning_rate": 7.820732296853074e-05,
+      "loss": 1.2314,
+      "step": 305
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.21122844902751076,
+      "learning_rate": 7.818199982336058e-05,
+      "loss": 1.1462,
+      "step": 306
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.23374869692118933,
+      "learning_rate": 7.815650323259117e-05,
+      "loss": 1.2051,
+      "step": 307
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.21662363795962128,
+      "learning_rate": 7.813083331204332e-05,
+      "loss": 1.1575,
+      "step": 308
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2088315773384112,
+      "learning_rate": 7.810499017832526e-05,
+      "loss": 1.1316,
+      "step": 309
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2095238410730976,
+      "learning_rate": 7.807897394883203e-05,
+      "loss": 1.2087,
+      "step": 310
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.22672932127256515,
+      "learning_rate": 7.805278474174499e-05,
+      "loss": 1.2512,
+      "step": 311
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.21873052340922736,
+      "learning_rate": 7.802642267603126e-05,
+      "loss": 1.1909,
+      "step": 312
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.219814521916342,
+      "learning_rate": 7.79998878714432e-05,
+      "loss": 1.1669,
+      "step": 313
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3049426027257317,
+      "learning_rate": 7.797318044851786e-05,
+      "loss": 1.1797,
+      "step": 314
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.22309435690065985,
+      "learning_rate": 7.794630052857638e-05,
+      "loss": 1.1417,
+      "step": 315
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3891885169154885,
+      "learning_rate": 7.791924823372354e-05,
+      "loss": 1.2369,
+      "step": 316
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.24780269452456372,
+      "learning_rate": 7.789202368684711e-05,
+      "loss": 1.2521,
+      "step": 317
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.21660460720269362,
+      "learning_rate": 7.786462701161738e-05,
+      "loss": 1.2151,
+      "step": 318
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.23635409466561857,
+      "learning_rate": 7.783705833248649e-05,
+      "loss": 1.2363,
+      "step": 319
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.2616135839903218,
+      "learning_rate": 7.780931777468797e-05,
+      "loss": 1.2428,
+      "step": 320
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21461059159245083,
+      "learning_rate": 7.77814054642361e-05,
+      "loss": 1.1434,
+      "step": 321
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.25348824286656163,
+      "learning_rate": 7.775332152792539e-05,
+      "loss": 1.2368,
+      "step": 322
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.22275034726331247,
+      "learning_rate": 7.772506609332995e-05,
+      "loss": 1.1827,
+      "step": 323
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.25030821228147526,
+      "learning_rate": 7.769663928880298e-05,
+      "loss": 1.2428,
+      "step": 324
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.22251804398745534,
+      "learning_rate": 7.766804124347608e-05,
+      "loss": 1.1889,
+      "step": 325
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.23381455520411995,
+      "learning_rate": 7.763927208725879e-05,
+      "loss": 1.2115,
+      "step": 326
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.27341902651946226,
+      "learning_rate": 7.761033195083791e-05,
+      "loss": 1.2535,
+      "step": 327
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.24862471659814522,
+      "learning_rate": 7.758122096567694e-05,
+      "loss": 1.2128,
+      "step": 328
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2251357082045494,
+      "learning_rate": 7.755193926401547e-05,
+      "loss": 1.2334,
+      "step": 329
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3173274941622932,
+      "learning_rate": 7.752248697886857e-05,
+      "loss": 1.226,
+      "step": 330
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.23056440717672175,
+      "learning_rate": 7.74928642440263e-05,
+      "loss": 1.2339,
+      "step": 331
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2801507500859342,
+      "learning_rate": 7.746307119405286e-05,
+      "loss": 1.287,
+      "step": 332
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2267818430426272,
+      "learning_rate": 7.743310796428622e-05,
+      "loss": 1.1916,
+      "step": 333
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2777329160365585,
+      "learning_rate": 7.74029746908374e-05,
+      "loss": 1.252,
+      "step": 334
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.25289169762353,
+      "learning_rate": 7.737267151058983e-05,
+      "loss": 1.2153,
+      "step": 335
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2424670686901653,
+      "learning_rate": 7.734219856119875e-05,
+      "loss": 1.2227,
+      "step": 336
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.22747092217441645,
+      "learning_rate": 7.731155598109067e-05,
+      "loss": 1.19,
+      "step": 337
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2307810940100189,
+      "learning_rate": 7.728074390946257e-05,
+      "loss": 1.1818,
+      "step": 338
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2583402574655623,
+      "learning_rate": 7.724976248628142e-05,
+      "loss": 1.1608,
+      "step": 339
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.22140209760890694,
+      "learning_rate": 7.721861185228347e-05,
+      "loss": 1.1245,
+      "step": 340
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.25859310758244686,
+      "learning_rate": 7.718729214897362e-05,
+      "loss": 1.2247,
+      "step": 341
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.26371179531372124,
+      "learning_rate": 7.715580351862482e-05,
+      "loss": 1.2128,
+      "step": 342
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.26575541302851047,
+      "learning_rate": 7.712414610427733e-05,
+      "loss": 1.2443,
+      "step": 343
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.269978305197599,
+      "learning_rate": 7.709232004973816e-05,
+      "loss": 1.2231,
+      "step": 344
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.26583998705977047,
+      "learning_rate": 7.70603254995804e-05,
+      "loss": 1.2476,
+      "step": 345
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.24256062164066097,
+      "learning_rate": 7.702816259914253e-05,
+      "loss": 1.2901,
+      "step": 346
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3463123472658915,
+      "learning_rate": 7.699583149452779e-05,
+      "loss": 1.3277,
+      "step": 347
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2269096590531878,
+      "learning_rate": 7.696333233260345e-05,
+      "loss": 1.2047,
+      "step": 348
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25136883001050025,
+      "learning_rate": 7.693066526100031e-05,
+      "loss": 1.1619,
+      "step": 349
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2565112571116145,
+      "learning_rate": 7.68978304281118e-05,
+      "loss": 1.2389,
+      "step": 350
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.22175779550828703,
+      "learning_rate": 7.686482798309349e-05,
+      "loss": 1.2238,
+      "step": 351
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.22588304332216555,
+      "learning_rate": 7.683165807586234e-05,
+      "loss": 1.174,
+      "step": 352
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.24889474296529737,
+      "learning_rate": 7.6798320857096e-05,
+      "loss": 1.2366,
+      "step": 353
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.27339703806525034,
+      "learning_rate": 7.676481647823214e-05,
+      "loss": 1.2356,
+      "step": 354
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.23424666722888365,
+      "learning_rate": 7.673114509146782e-05,
+      "loss": 1.2089,
+      "step": 355
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.27978285392461766,
+      "learning_rate": 7.66973068497587e-05,
+      "loss": 1.2609,
+      "step": 356
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.2509423350138824,
+      "learning_rate": 7.666330190681844e-05,
+      "loss": 1.1777,
+      "step": 357
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.23007730927468031,
+      "learning_rate": 7.662913041711793e-05,
+      "loss": 1.154,
+      "step": 358
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.2438648674953112,
+      "learning_rate": 7.659479253588462e-05,
+      "loss": 1.2257,
+      "step": 359
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.28816093242092233,
+      "learning_rate": 7.65602884191018e-05,
+      "loss": 1.2558,
+      "step": 360
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24972815300596035,
+      "learning_rate": 7.652561822350793e-05,
+      "loss": 1.2837,
+      "step": 361
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.2543189139697063,
+      "learning_rate": 7.649078210659587e-05,
+      "loss": 1.2193,
+      "step": 362
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.2237937956718952,
+      "learning_rate": 7.645578022661224e-05,
+      "loss": 1.2237,
+      "step": 363
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.29742029408787396,
+      "learning_rate": 7.642061274255657e-05,
+      "loss": 1.2116,
+      "step": 364
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.2462883147335493,
+      "learning_rate": 7.638527981418075e-05,
+      "loss": 1.1827,
+      "step": 365
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.2647802498907096,
+      "learning_rate": 7.634978160198817e-05,
+      "loss": 1.2739,
+      "step": 366
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.22360398779217264,
+      "learning_rate": 7.631411826723306e-05,
+      "loss": 1.2185,
+      "step": 367
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.2635048004593543,
+      "learning_rate": 7.627828997191973e-05,
+      "loss": 1.2317,
+      "step": 368
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.2764803449917684,
+      "learning_rate": 7.624229687880184e-05,
+      "loss": 1.1923,
+      "step": 369
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.25724943233414527,
+      "learning_rate": 7.620613915138166e-05,
+      "loss": 1.2218,
+      "step": 370
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2858318045794755,
+      "learning_rate": 7.61698169539093e-05,
+      "loss": 1.1496,
+      "step": 371
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.23547216647460364,
+      "learning_rate": 7.613333045138206e-05,
+      "loss": 1.1905,
+      "step": 372
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.22984814903684375,
+      "learning_rate": 7.609667980954355e-05,
+      "loss": 1.2009,
+      "step": 373
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2551903754079084,
+      "learning_rate": 7.605986519488301e-05,
+      "loss": 1.2042,
+      "step": 374
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.2508257410125616,
+      "learning_rate": 7.602288677463457e-05,
+      "loss": 1.2468,
+      "step": 375
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.25324577774935964,
+      "learning_rate": 7.598574471677644e-05,
+      "loss": 1.2603,
+      "step": 376
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.35888776531769967,
+      "learning_rate": 7.59484391900302e-05,
+      "loss": 1.1929,
+      "step": 377
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.22048517191014724,
+      "learning_rate": 7.591097036385994e-05,
+      "loss": 1.1783,
+      "step": 378
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.2781160412746083,
+      "learning_rate": 7.587333840847162e-05,
+      "loss": 1.3397,
+      "step": 379
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.24033046830332258,
+      "learning_rate": 7.583554349481222e-05,
+      "loss": 1.2436,
+      "step": 380
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.26413762380260003,
+      "learning_rate": 7.579758579456893e-05,
+      "loss": 1.1917,
+      "step": 381
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.2390937887338632,
+      "learning_rate": 7.575946548016847e-05,
+      "loss": 1.2186,
+      "step": 382
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.25131263043429275,
+      "learning_rate": 7.572118272477622e-05,
+      "loss": 1.2538,
+      "step": 383
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.223974104870702,
+      "learning_rate": 7.568273770229546e-05,
+      "loss": 1.2165,
+      "step": 384
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.25840356830252875,
+      "learning_rate": 7.564413058736663e-05,
+      "loss": 1.1848,
+      "step": 385
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2723156683076603,
+      "learning_rate": 7.560536155536641e-05,
+      "loss": 1.1982,
+      "step": 386
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.265687427976889,
+      "learning_rate": 7.556643078240708e-05,
+      "loss": 1.231,
+      "step": 387
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.25152762080976077,
+      "learning_rate": 7.552733844533562e-05,
+      "loss": 1.1974,
+      "step": 388
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2366049485053541,
+      "learning_rate": 7.548808472173292e-05,
+      "loss": 1.3119,
+      "step": 389
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.22092196577077122,
+      "learning_rate": 7.5448669789913e-05,
+      "loss": 1.195,
+      "step": 390
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.22667521540462374,
+      "learning_rate": 7.540909382892217e-05,
+      "loss": 1.1431,
+      "step": 391
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.25432207282646513,
+      "learning_rate": 7.536935701853823e-05,
+      "loss": 1.2173,
+      "step": 392
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.29950506457923864,
+      "learning_rate": 7.53294595392697e-05,
+      "loss": 1.1962,
+      "step": 393
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.24735689607229913,
+      "learning_rate": 7.528940157235487e-05,
+      "loss": 1.2053,
+      "step": 394
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.24394198607459663,
+      "learning_rate": 7.524918329976114e-05,
+      "loss": 1.1979,
+      "step": 395
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.2630369372689188,
+      "learning_rate": 7.520880490418409e-05,
+      "loss": 1.2111,
+      "step": 396
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.26275028416291457,
+      "learning_rate": 7.516826656904664e-05,
+      "loss": 1.2133,
+      "step": 397
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.23938074620956928,
+      "learning_rate": 7.512756847849831e-05,
+      "loss": 1.1355,
+      "step": 398
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3724960610098138,
+      "learning_rate": 7.508671081741428e-05,
+      "loss": 1.2572,
+      "step": 399
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24161685847894723,
+      "learning_rate": 7.504569377139462e-05,
+      "loss": 1.1706,
+      "step": 400
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.26121591322670523,
+      "learning_rate": 7.50045175267634e-05,
+      "loss": 1.2135,
+      "step": 401
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.2465579498164775,
+      "learning_rate": 7.496318227056788e-05,
+      "loss": 1.1641,
+      "step": 402
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.2556288696122787,
+      "learning_rate": 7.492168819057767e-05,
+      "loss": 1.2939,
+      "step": 403
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.261481216336303,
+      "learning_rate": 7.488003547528382e-05,
+      "loss": 1.2026,
+      "step": 404
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.2389415135676362,
+      "learning_rate": 7.483822431389799e-05,
+      "loss": 1.2131,
+      "step": 405
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.2559201956627192,
+      "learning_rate": 7.479625489635162e-05,
+      "loss": 1.1246,
+      "step": 406
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.27127932491822604,
+      "learning_rate": 7.475412741329504e-05,
+      "loss": 1.2429,
+      "step": 407
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.27006004008695594,
+      "learning_rate": 7.47118420560966e-05,
+      "loss": 1.2388,
+      "step": 408
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.23716823297200537,
+      "learning_rate": 7.466939901684182e-05,
+      "loss": 1.1264,
+      "step": 409
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.2885373898669248,
+      "learning_rate": 7.462679848833252e-05,
+      "loss": 1.2786,
+      "step": 410
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.49215227598639927,
+      "learning_rate": 7.458404066408588e-05,
+      "loss": 1.2386,
+      "step": 411
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.24235735604947403,
+      "learning_rate": 7.454112573833368e-05,
+      "loss": 1.1423,
+      "step": 412
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.2584614748054343,
+      "learning_rate": 7.449805390602127e-05,
+      "loss": 1.2669,
+      "step": 413
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.23806123085998873,
+      "learning_rate": 7.445482536280684e-05,
+      "loss": 1.1763,
+      "step": 414
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.24459517607786851,
+      "learning_rate": 7.441144030506043e-05,
+      "loss": 1.198,
+      "step": 415
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.25801616402700395,
+      "learning_rate": 7.436789892986304e-05,
+      "loss": 1.2136,
+      "step": 416
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.2814819942392514,
+      "learning_rate": 7.432420143500578e-05,
+      "loss": 1.2398,
+      "step": 417
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.22134709322606153,
+      "learning_rate": 7.428034801898893e-05,
+      "loss": 1.1592,
+      "step": 418
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.2899677536995633,
+      "learning_rate": 7.42363388810211e-05,
+      "loss": 1.2296,
+      "step": 419
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.24005943230262294,
+      "learning_rate": 7.419217422101822e-05,
+      "loss": 1.2223,
+      "step": 420
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.26417562369496167,
+      "learning_rate": 7.414785423960275e-05,
+      "loss": 1.2261,
+      "step": 421
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.2580815883535521,
+      "learning_rate": 7.410337913810271e-05,
+      "loss": 1.2021,
+      "step": 422
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.25242217589496435,
+      "learning_rate": 7.405874911855071e-05,
+      "loss": 1.239,
+      "step": 423
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.21991733999839932,
+      "learning_rate": 7.401396438368315e-05,
+      "loss": 1.1716,
+      "step": 424
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.40116538322720213,
+      "learning_rate": 7.396902513693924e-05,
+      "loss": 1.2773,
+      "step": 425
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.277333939455099,
+      "learning_rate": 7.392393158246002e-05,
+      "loss": 1.2574,
+      "step": 426
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.27146087746385755,
+      "learning_rate": 7.387868392508756e-05,
+      "loss": 1.2243,
+      "step": 427
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.255881055620786,
+      "learning_rate": 7.38332823703639e-05,
+      "loss": 1.223,
+      "step": 428
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.24807364856677255,
+      "learning_rate": 7.378772712453021e-05,
+      "loss": 1.1985,
+      "step": 429
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.25746257617764423,
+      "learning_rate": 7.37420183945258e-05,
+      "loss": 1.2502,
+      "step": 430
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.28851991049982234,
+      "learning_rate": 7.369615638798722e-05,
+      "loss": 1.2535,
+      "step": 431
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.24113389811604363,
+      "learning_rate": 7.365014131324725e-05,
+      "loss": 1.2227,
+      "step": 432
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.2414465151257969,
+      "learning_rate": 7.360397337933405e-05,
+      "loss": 1.1884,
+      "step": 433
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.2735463134699831,
+      "learning_rate": 7.355765279597011e-05,
+      "loss": 1.2756,
+      "step": 434
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2588437452987293,
+      "learning_rate": 7.351117977357139e-05,
+      "loss": 1.2108,
+      "step": 435
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.26573294117796553,
+      "learning_rate": 7.346455452324629e-05,
+      "loss": 1.1821,
+      "step": 436
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2555476577827304,
+      "learning_rate": 7.341777725679473e-05,
+      "loss": 1.1937,
+      "step": 437
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2867704132108098,
+      "learning_rate": 7.337084818670716e-05,
+      "loss": 1.2272,
+      "step": 438
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.27726678115981157,
+      "learning_rate": 7.332376752616367e-05,
+      "loss": 1.2331,
+      "step": 439
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.26955338021079955,
+      "learning_rate": 7.32765354890329e-05,
+      "loss": 1.1731,
+      "step": 440
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.25250321202536524,
+      "learning_rate": 7.322915228987116e-05,
+      "loss": 1.2653,
+      "step": 441
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.24748844179765395,
+      "learning_rate": 7.318161814392143e-05,
+      "loss": 1.24,
+      "step": 442
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.28177805247356325,
+      "learning_rate": 7.313393326711239e-05,
+      "loss": 1.185,
+      "step": 443
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.24093242000396312,
+      "learning_rate": 7.30860978760574e-05,
+      "loss": 1.1994,
+      "step": 444
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.26277803901457075,
+      "learning_rate": 7.30381121880536e-05,
+      "loss": 1.212,
+      "step": 445
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2506524258682433,
+      "learning_rate": 7.298997642108079e-05,
+      "loss": 1.2421,
+      "step": 446
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2840599700015824,
+      "learning_rate": 7.294169079380061e-05,
+      "loss": 1.1818,
+      "step": 447
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.24892184038117549,
+      "learning_rate": 7.289325552555538e-05,
+      "loss": 1.1916,
+      "step": 448
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2700898428541357,
+      "learning_rate": 7.284467083636722e-05,
+      "loss": 1.2517,
+      "step": 449
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2617848546539419,
+      "learning_rate": 7.279593694693698e-05,
+      "loss": 1.2063,
+      "step": 450
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2698278585334131,
+      "learning_rate": 7.274705407864332e-05,
+      "loss": 1.194,
+      "step": 451
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.23678313024953834,
+      "learning_rate": 7.26980224535416e-05,
+      "loss": 1.2349,
+      "step": 452
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.24851875792002978,
+      "learning_rate": 7.264884229436293e-05,
+      "loss": 1.1758,
+      "step": 453
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.24122080121681125,
+      "learning_rate": 7.259951382451318e-05,
+      "loss": 1.1962,
+      "step": 454
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.22741322959884405,
+      "learning_rate": 7.25500372680719e-05,
+      "loss": 1.1702,
+      "step": 455
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.2297475610861458,
+      "learning_rate": 7.250041284979137e-05,
+      "loss": 1.1466,
+      "step": 456
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3057605989721467,
+      "learning_rate": 7.245064079509553e-05,
+      "loss": 1.246,
+      "step": 457
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.2719638501597136,
+      "learning_rate": 7.240072133007899e-05,
+      "loss": 1.2184,
+      "step": 458
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.2436807816414479,
+      "learning_rate": 7.235065468150593e-05,
+      "loss": 1.2324,
+      "step": 459
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.23436349430255515,
+      "learning_rate": 7.23004410768092e-05,
+      "loss": 1.1813,
+      "step": 460
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.2398940990211377,
+      "learning_rate": 7.22500807440892e-05,
+      "loss": 1.1924,
+      "step": 461
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.2605716625062531,
+      "learning_rate": 7.219957391211281e-05,
+      "loss": 1.182,
+      "step": 462
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.260462524570941,
+      "learning_rate": 7.214892081031244e-05,
+      "loss": 1.2136,
+      "step": 463
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.21979766512306334,
+      "learning_rate": 7.209812166878491e-05,
+      "loss": 1.2066,
+      "step": 464
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.23324453647530663,
+      "learning_rate": 7.204717671829051e-05,
+      "loss": 1.1657,
+      "step": 465
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.2529434935507481,
+      "learning_rate": 7.199608619025177e-05,
+      "loss": 1.2093,
+      "step": 466
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.25371701891720116,
+      "learning_rate": 7.194485031675265e-05,
+      "loss": 1.2225,
+      "step": 467
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.23272423066292103,
+      "learning_rate": 7.189346933053725e-05,
+      "loss": 1.1721,
+      "step": 468
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.25122928735587546,
+      "learning_rate": 7.184194346500892e-05,
+      "loss": 1.2537,
+      "step": 469
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.2159270875490409,
+      "learning_rate": 7.179027295422913e-05,
+      "loss": 1.197,
+      "step": 470
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.2633111059076544,
+      "learning_rate": 7.173845803291636e-05,
+      "loss": 1.1721,
+      "step": 471
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.30555936322098703,
+      "learning_rate": 7.168649893644517e-05,
+      "loss": 1.3011,
+      "step": 472
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.23492670111453726,
+      "learning_rate": 7.163439590084502e-05,
+      "loss": 1.1601,
+      "step": 473
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.26602734263721806,
+      "learning_rate": 7.158214916279923e-05,
+      "loss": 1.2808,
+      "step": 474
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3182695007856262,
+      "learning_rate": 7.152975895964386e-05,
+      "loss": 1.2967,
+      "step": 475
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.2785021674736721,
+      "learning_rate": 7.147722552936673e-05,
+      "loss": 1.1789,
+      "step": 476
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.279474303138652,
+      "learning_rate": 7.142454911060627e-05,
+      "loss": 1.2596,
+      "step": 477
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.2556980144910755,
+      "learning_rate": 7.137172994265044e-05,
+      "loss": 1.2426,
+      "step": 478
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3311256331993533,
+      "learning_rate": 7.131876826543565e-05,
+      "loss": 1.2059,
+      "step": 479
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.26467296197775253,
+      "learning_rate": 7.12656643195457e-05,
+      "loss": 1.2482,
+      "step": 480
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.27444885274652553,
+      "learning_rate": 7.121241834621064e-05,
+      "loss": 1.2528,
+      "step": 481
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.2572283861115396,
+      "learning_rate": 7.115903058730567e-05,
+      "loss": 1.1849,
+      "step": 482
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.2677065778235683,
+      "learning_rate": 7.11055012853501e-05,
+      "loss": 1.2011,
+      "step": 483
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.29470622036742816,
+      "learning_rate": 7.105183068350619e-05,
+      "loss": 1.2398,
+      "step": 484
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.27609230248969197,
+      "learning_rate": 7.099801902557811e-05,
+      "loss": 1.2259,
+      "step": 485
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24248634168099284,
+      "learning_rate": 7.094406655601073e-05,
+      "loss": 1.2282,
+      "step": 486
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.2765941767688746,
+      "learning_rate": 7.088997351988865e-05,
+      "loss": 1.2319,
+      "step": 487
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.29347776909858947,
+      "learning_rate": 7.083574016293493e-05,
+      "loss": 1.1765,
+      "step": 488
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.285370295424537,
+      "learning_rate": 7.078136673151008e-05,
+      "loss": 1.26,
+      "step": 489
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.29408734903836536,
+      "learning_rate": 7.072685347261093e-05,
+      "loss": 1.226,
+      "step": 490
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.27437470239205813,
+      "learning_rate": 7.067220063386947e-05,
+      "loss": 1.1976,
+      "step": 491
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.2680770258777871,
+      "learning_rate": 7.061740846355176e-05,
+      "loss": 1.1915,
+      "step": 492
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.27200362879502954,
+      "learning_rate": 7.056247721055678e-05,
+      "loss": 1.2002,
+      "step": 493
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.2637811092577037,
+      "learning_rate": 7.050740712441528e-05,
+      "loss": 1.287,
+      "step": 494
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.24657959209271266,
+      "learning_rate": 7.045219845528875e-05,
+      "loss": 1.2284,
+      "step": 495
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.25311992110358666,
+      "learning_rate": 7.039685145396812e-05,
+      "loss": 1.1616,
+      "step": 496
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2564633694193358,
+      "learning_rate": 7.034136637187275e-05,
+      "loss": 1.2067,
+      "step": 497
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2446797651174144,
+      "learning_rate": 7.028574346104926e-05,
+      "loss": 1.2284,
+      "step": 498
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2592751463399255,
+      "learning_rate": 7.022998297417034e-05,
+      "loss": 1.2371,
+      "step": 499
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2500713943206808,
+      "learning_rate": 7.017408516453365e-05,
+      "loss": 1.1061,
+      "step": 500
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.2812266276040743,
+      "learning_rate": 7.011805028606064e-05,
+      "loss": 1.1949,
+      "step": 501
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.298829667668083,
+      "learning_rate": 7.006187859329544e-05,
+      "loss": 1.2313,
+      "step": 502
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.26518768159745104,
+      "learning_rate": 7.000557034140361e-05,
+      "loss": 1.2246,
+      "step": 503
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3037280360760458,
+      "learning_rate": 6.994912578617113e-05,
+      "loss": 1.1617,
+      "step": 504
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.2726903109255714,
+      "learning_rate": 6.989254518400309e-05,
+      "loss": 1.2415,
+      "step": 505
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.25568082003046966,
+      "learning_rate": 6.98358287919226e-05,
+      "loss": 1.1817,
+      "step": 506
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.25633294893705044,
+      "learning_rate": 6.97789768675696e-05,
+      "loss": 1.2149,
+      "step": 507
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28291439435087123,
+      "learning_rate": 6.972198966919972e-05,
+      "loss": 1.1578,
+      "step": 508
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.27195184756655516,
+      "learning_rate": 6.966486745568308e-05,
+      "loss": 1.2355,
+      "step": 509
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.239159568376005,
+      "learning_rate": 6.960761048650312e-05,
+      "loss": 1.1688,
+      "step": 510
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.22961475425949177,
+      "learning_rate": 6.955021902175543e-05,
+      "loss": 1.2094,
+      "step": 511
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.27443773600741117,
+      "learning_rate": 6.949269332214651e-05,
+      "loss": 1.2559,
+      "step": 512
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.26230551832002097,
+      "learning_rate": 6.94350336489927e-05,
+      "loss": 1.2121,
+      "step": 513
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.2716742985303849,
+      "learning_rate": 6.937724026421892e-05,
+      "loss": 1.2444,
+      "step": 514
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.2537850139439542,
+      "learning_rate": 6.931931343035742e-05,
+      "loss": 1.1327,
+      "step": 515
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.28599587967496826,
+      "learning_rate": 6.926125341054676e-05,
+      "loss": 1.2236,
+      "step": 516
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.26780654378470103,
+      "learning_rate": 6.920306046853043e-05,
+      "loss": 1.2295,
+      "step": 517
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.23606296888412015,
+      "learning_rate": 6.914473486865577e-05,
+      "loss": 1.1543,
+      "step": 518
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.34976881174240837,
+      "learning_rate": 6.90862768758727e-05,
+      "loss": 1.2067,
+      "step": 519
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.2481257873494882,
+      "learning_rate": 6.902768675573258e-05,
+      "loss": 1.2188,
+      "step": 520
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.2996395778117021,
+      "learning_rate": 6.896896477438699e-05,
+      "loss": 1.2326,
+      "step": 521
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8839768816333193,
+      "learning_rate": 6.891011119858643e-05,
+      "loss": 1.2435,
+      "step": 522
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.2851882482058998,
+      "learning_rate": 6.885112629567927e-05,
+      "loss": 1.2644,
+      "step": 523
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.2813663482913699,
+      "learning_rate": 6.879201033361035e-05,
+      "loss": 1.2309,
+      "step": 524
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3257551560135454,
+      "learning_rate": 6.873276358091996e-05,
+      "loss": 1.2755,
+      "step": 525
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.28930479952494365,
+      "learning_rate": 6.867338630674247e-05,
+      "loss": 1.1962,
+      "step": 526
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3077462996938649,
+      "learning_rate": 6.861387878080511e-05,
+      "loss": 1.2402,
+      "step": 527
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.2848900193452761,
+      "learning_rate": 6.855424127342688e-05,
+      "loss": 1.2748,
+      "step": 528
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4765938812802202,
+      "learning_rate": 6.849447405551718e-05,
+      "loss": 1.2226,
+      "step": 529
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.53184473292579,
+      "learning_rate": 6.843457739857467e-05,
+      "loss": 1.2347,
+      "step": 530
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6416239346492343,
+      "learning_rate": 6.837455157468596e-05,
+      "loss": 1.2429,
+      "step": 531
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3188092712502773,
+      "learning_rate": 6.831439685652442e-05,
+      "loss": 1.216,
+      "step": 532
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3527495731006385,
+      "learning_rate": 6.825411351734895e-05,
+      "loss": 1.1682,
+      "step": 533
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.29603753744741856,
+      "learning_rate": 6.819370183100274e-05,
+      "loss": 1.1434,
+      "step": 534
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5252450389976622,
+      "learning_rate": 6.813316207191198e-05,
+      "loss": 1.1943,
+      "step": 535
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.32999419558659937,
+      "learning_rate": 6.807249451508466e-05,
+      "loss": 1.192,
+      "step": 536
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3650175469778724,
+      "learning_rate": 6.801169943610929e-05,
+      "loss": 1.2141,
+      "step": 537
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.0643532150783557,
+      "learning_rate": 6.795077711115368e-05,
+      "loss": 1.2253,
+      "step": 538
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5041310609130145,
+      "learning_rate": 6.788972781696363e-05,
+      "loss": 1.278,
+      "step": 539
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5123058164360991,
+      "learning_rate": 6.782855183086177e-05,
+      "loss": 1.2231,
+      "step": 540
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3533015702394419,
+      "learning_rate": 6.776724943074619e-05,
+      "loss": 1.2072,
+      "step": 541
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.30253964625417207,
+      "learning_rate": 6.770582089508927e-05,
+      "loss": 1.1382,
+      "step": 542
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.348991618828202,
+      "learning_rate": 6.764426650293633e-05,
+      "loss": 1.2079,
+      "step": 543
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.46017440578788743,
+      "learning_rate": 6.758258653390444e-05,
+      "loss": 1.1813,
+      "step": 544
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.31962101755594885,
+      "learning_rate": 6.75207812681811e-05,
+      "loss": 1.1339,
+      "step": 545
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.37092024548285923,
+      "learning_rate": 6.745885098652298e-05,
+      "loss": 1.2591,
+      "step": 546
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.32347106450715835,
+      "learning_rate": 6.739679597025466e-05,
+      "loss": 1.2017,
+      "step": 547
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.39250187112342494,
+      "learning_rate": 6.733461650126733e-05,
+      "loss": 1.0933,
+      "step": 548
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.473522452217324,
+      "learning_rate": 6.727231286201752e-05,
+      "loss": 1.1124,
+      "step": 549
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.4809062179622052,
+      "learning_rate": 6.720988533552582e-05,
+      "loss": 1.1585,
+      "step": 550
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.3529662801059162,
+      "learning_rate": 6.714733420537559e-05,
+      "loss": 1.0501,
+      "step": 551
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5958247214391118,
+      "learning_rate": 6.708465975571168e-05,
+      "loss": 1.1086,
+      "step": 552
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5341364205022454,
+      "learning_rate": 6.70218622712391e-05,
+      "loss": 1.0518,
+      "step": 553
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.3601805724462006,
+      "learning_rate": 6.695894203722181e-05,
+      "loss": 1.1779,
+      "step": 554
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.43410190338280613,
+      "learning_rate": 6.68958993394813e-05,
+      "loss": 1.093,
+      "step": 555
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.46217742572873594,
+      "learning_rate": 6.683273446439546e-05,
+      "loss": 1.0117,
+      "step": 556
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.8591682373623357,
+      "learning_rate": 6.676944769889708e-05,
+      "loss": 1.1002,
+      "step": 557
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.7383229487622726,
+      "learning_rate": 6.670603933047272e-05,
+      "loss": 1.0779,
+      "step": 558
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5965305891207813,
+      "learning_rate": 6.664250964716131e-05,
+      "loss": 1.0889,
+      "step": 559
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.6030858606684543,
+      "learning_rate": 6.657885893755288e-05,
+      "loss": 1.0982,
+      "step": 560
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.4644510682398409,
+      "learning_rate": 6.65150874907872e-05,
+      "loss": 1.1004,
+      "step": 561
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.43943285132452564,
+      "learning_rate": 6.645119559655254e-05,
+      "loss": 1.0536,
+      "step": 562
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.4456395978600012,
+      "learning_rate": 6.638718354508427e-05,
+      "loss": 1.0733,
+      "step": 563
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.3303824433217466,
+      "learning_rate": 6.632305162716365e-05,
+      "loss": 1.0552,
+      "step": 564
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.3617704823170143,
+      "learning_rate": 6.62588001341164e-05,
+      "loss": 1.1092,
+      "step": 565
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.4465013349903427,
+      "learning_rate": 6.619442935781141e-05,
+      "loss": 1.0781,
+      "step": 566
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.48516780613791277,
+      "learning_rate": 6.612993959065947e-05,
+      "loss": 1.0686,
+      "step": 567
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.38867820318536633,
+      "learning_rate": 6.606533112561186e-05,
+      "loss": 1.1215,
+      "step": 568
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.38566119820378336,
+      "learning_rate": 6.600060425615907e-05,
+      "loss": 1.1213,
+      "step": 569
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.35534855445058544,
+      "learning_rate": 6.593575927632947e-05,
+      "loss": 1.0955,
+      "step": 570
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.38124406233349717,
+      "learning_rate": 6.587079648068795e-05,
+      "loss": 1.0659,
+      "step": 571
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.454750160923548,
+      "learning_rate": 6.580571616433457e-05,
+      "loss": 1.1149,
+      "step": 572
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.35353190088025255,
+      "learning_rate": 6.574051862290325e-05,
+      "loss": 1.0388,
+      "step": 573
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.3249395594793626,
+      "learning_rate": 6.567520415256045e-05,
+      "loss": 1.0784,
+      "step": 574
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.40078898818247227,
+      "learning_rate": 6.560977305000375e-05,
+      "loss": 1.0859,
+      "step": 575
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.4115264795060035,
+      "learning_rate": 6.554422561246054e-05,
+      "loss": 1.1828,
+      "step": 576
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.30090229228069215,
+      "learning_rate": 6.54785621376867e-05,
+      "loss": 1.0901,
+      "step": 577
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.28827860350299206,
+      "learning_rate": 6.541278292396523e-05,
+      "loss": 1.0277,
+      "step": 578
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.34690404488996757,
+      "learning_rate": 6.534688827010484e-05,
+      "loss": 1.048,
+      "step": 579
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.29943113556644785,
+      "learning_rate": 6.528087847543867e-05,
+      "loss": 1.0646,
+      "step": 580
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.37318202575874415,
+      "learning_rate": 6.521475383982291e-05,
+      "loss": 1.1091,
+      "step": 581
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.3049663659203959,
+      "learning_rate": 6.51485146636354e-05,
+      "loss": 1.0552,
+      "step": 582
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.3342407867509692,
+      "learning_rate": 6.508216124777431e-05,
+      "loss": 1.2227,
+      "step": 583
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.3348396047855952,
+      "learning_rate": 6.501569389365674e-05,
+      "loss": 1.0861,
+      "step": 584
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.30951429367513383,
+      "learning_rate": 6.494911290321737e-05,
+      "loss": 1.0461,
+      "step": 585
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.33898401361064606,
+      "learning_rate": 6.488241857890711e-05,
+      "loss": 1.0854,
+      "step": 586
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.4901462068263497,
+      "learning_rate": 6.481561122369164e-05,
+      "loss": 1.1012,
+      "step": 587
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.3179574879809652,
+      "learning_rate": 6.474869114105018e-05,
+      "loss": 1.0451,
+      "step": 588
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.32159328915060714,
+      "learning_rate": 6.468165863497395e-05,
+      "loss": 1.0458,
+      "step": 589
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.36462235008537297,
+      "learning_rate": 6.461451400996491e-05,
+      "loss": 1.1247,
+      "step": 590
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5373862753611778,
+      "learning_rate": 6.454725757103432e-05,
+      "loss": 1.0542,
+      "step": 591
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.3160409270291303,
+      "learning_rate": 6.447988962370133e-05,
+      "loss": 1.0829,
+      "step": 592
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.390452102978435,
+      "learning_rate": 6.441241047399169e-05,
+      "loss": 1.192,
+      "step": 593
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3802122712014928,
+      "learning_rate": 6.434482042843627e-05,
+      "loss": 1.1153,
+      "step": 594
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.4081584328242501,
+      "learning_rate": 6.427711979406966e-05,
+      "loss": 1.1635,
+      "step": 595
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3791962989638633,
+      "learning_rate": 6.420930887842889e-05,
+      "loss": 1.1581,
+      "step": 596
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.33239440056484193,
+      "learning_rate": 6.414138798955189e-05,
+      "loss": 1.0926,
+      "step": 597
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3279881540815014,
+      "learning_rate": 6.407335743597616e-05,
+      "loss": 1.1386,
+      "step": 598
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.30309644763750837,
+      "learning_rate": 6.40052175267374e-05,
+      "loss": 1.0523,
+      "step": 599
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.3349097308403333,
+      "learning_rate": 6.393696857136801e-05,
+      "loss": 1.0815,
+      "step": 600
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.3288227593556618,
+      "learning_rate": 6.386861087989581e-05,
+      "loss": 1.015,
+      "step": 601
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.36685586740843157,
+      "learning_rate": 6.380014476284255e-05,
+      "loss": 1.1232,
+      "step": 602
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.3620977714204643,
+      "learning_rate": 6.373157053122243e-05,
+      "loss": 1.1138,
+      "step": 603
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.3130587018197183,
+      "learning_rate": 6.366288849654091e-05,
+      "loss": 1.1255,
+      "step": 604
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.3602737087072766,
+      "learning_rate": 6.359409897079303e-05,
+      "loss": 1.0282,
+      "step": 605
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.31168852571991945,
+      "learning_rate": 6.352520226646222e-05,
+      "loss": 1.0779,
+      "step": 606
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.3516045580189353,
+      "learning_rate": 6.345619869651871e-05,
+      "loss": 1.1028,
+      "step": 607
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.3231857927563657,
+      "learning_rate": 6.33870885744182e-05,
+      "loss": 1.1202,
+      "step": 608
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.30205205129701157,
+      "learning_rate": 6.331787221410041e-05,
+      "loss": 1.1369,
+      "step": 609
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.3198359813888166,
+      "learning_rate": 6.32485499299877e-05,
+      "loss": 1.1763,
+      "step": 610
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.3128641370321787,
+      "learning_rate": 6.31791220369835e-05,
+      "loss": 1.0223,
+      "step": 611
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.2989105616213649,
+      "learning_rate": 6.31095888504711e-05,
+      "loss": 1.0358,
+      "step": 612
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.3103537906853337,
+      "learning_rate": 6.303995068631203e-05,
+      "loss": 1.1261,
+      "step": 613
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.28598715532508207,
+      "learning_rate": 6.297020786084467e-05,
+      "loss": 1.0629,
+      "step": 614
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.29809789918093255,
+      "learning_rate": 6.290036069088288e-05,
+      "loss": 1.035,
+      "step": 615
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.33765270252261453,
+      "learning_rate": 6.283040949371451e-05,
+      "loss": 1.1221,
+      "step": 616
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.3424617501293415,
+      "learning_rate": 6.276035458709993e-05,
+      "loss": 1.155,
+      "step": 617
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.3799189737987811,
+      "learning_rate": 6.269019628927067e-05,
+      "loss": 1.0701,
+      "step": 618
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.3358898935253196,
+      "learning_rate": 6.261993491892791e-05,
+      "loss": 1.1649,
+      "step": 619
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.31569979424117356,
+      "learning_rate": 6.254957079524099e-05,
+      "loss": 1.0633,
+      "step": 620
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.3002168156888237,
+      "learning_rate": 6.247910423784609e-05,
+      "loss": 1.0846,
+      "step": 621
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.3097238823450595,
+      "learning_rate": 6.24085355668447e-05,
+      "loss": 1.0808,
+      "step": 622
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.3120312761417578,
+      "learning_rate": 6.233786510280212e-05,
+      "loss": 1.0142,
+      "step": 623
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.3335343015064923,
+      "learning_rate": 6.22670931667461e-05,
+      "loss": 1.0674,
+      "step": 624
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.3234062304634526,
+      "learning_rate": 6.219622008016533e-05,
+      "loss": 1.0981,
+      "step": 625
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.32152678786547273,
+      "learning_rate": 6.212524616500798e-05,
+      "loss": 1.0244,
+      "step": 626
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.39031977608147594,
+      "learning_rate": 6.205417174368023e-05,
+      "loss": 1.1205,
+      "step": 627
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.3806189090017157,
+      "learning_rate": 6.198299713904485e-05,
+      "loss": 1.1134,
+      "step": 628
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.2978349276971668,
+      "learning_rate": 6.191172267441967e-05,
+      "loss": 1.0088,
+      "step": 629
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.3190354077382501,
+      "learning_rate": 6.184034867357617e-05,
+      "loss": 1.108,
+      "step": 630
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.32633048665038994,
+      "learning_rate": 6.176887546073797e-05,
+      "loss": 1.0825,
+      "step": 631
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.3428026413020903,
+      "learning_rate": 6.169730336057939e-05,
+      "loss": 1.0765,
+      "step": 632
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.3475737151929015,
+      "learning_rate": 6.162563269822391e-05,
+      "loss": 1.0693,
+      "step": 633
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.3870252154591392,
+      "learning_rate": 6.15538637992428e-05,
+      "loss": 1.1081,
+      "step": 634
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.33597355193652834,
+      "learning_rate": 6.148199698965352e-05,
+      "loss": 1.0893,
+      "step": 635
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.30805894179787247,
+      "learning_rate": 6.141003259591834e-05,
+      "loss": 1.0995,
+      "step": 636
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.3025073882734066,
+      "learning_rate": 6.133797094494281e-05,
+      "loss": 1.0388,
+      "step": 637
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.3524395196391662,
+      "learning_rate": 6.126581236407429e-05,
+      "loss": 1.1196,
+      "step": 638
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.3377646188130345,
+      "learning_rate": 6.119355718110039e-05,
+      "loss": 1.0382,
+      "step": 639
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.35508400659785483,
+      "learning_rate": 6.112120572424763e-05,
+      "loss": 1.1402,
+      "step": 640
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.3454418793700457,
+      "learning_rate": 6.104875832217982e-05,
+      "loss": 1.1032,
+      "step": 641
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.32629806837059866,
+      "learning_rate": 6.097621530399661e-05,
+      "loss": 1.0959,
+      "step": 642
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3329536837751315,
+      "learning_rate": 6.090357699923202e-05,
+      "loss": 1.0467,
+      "step": 643
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.32302233828349475,
+      "learning_rate": 6.083084373785287e-05,
+      "loss": 1.0858,
+      "step": 644
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3310358826507611,
+      "learning_rate": 6.075801585025739e-05,
+      "loss": 1.0715,
+      "step": 645
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.319322035854079,
+      "learning_rate": 6.068509366727362e-05,
+      "loss": 1.177,
+      "step": 646
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3065230667302707,
+      "learning_rate": 6.061207752015797e-05,
+      "loss": 1.0649,
+      "step": 647
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.29926795565748227,
+      "learning_rate": 6.053896774059368e-05,
+      "loss": 1.1325,
+      "step": 648
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.3556069634279046,
+      "learning_rate": 6.046576466068931e-05,
+      "loss": 1.1366,
+      "step": 649
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.3189191131461966,
+      "learning_rate": 6.039246861297727e-05,
+      "loss": 1.0693,
+      "step": 650
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.3347197156648834,
+      "learning_rate": 6.031907993041227e-05,
+      "loss": 1.1009,
+      "step": 651
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.32274156348185445,
+      "learning_rate": 6.0245598946369826e-05,
+      "loss": 1.1675,
+      "step": 652
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.35534089035455224,
+      "learning_rate": 6.017202599464476e-05,
+      "loss": 1.1723,
+      "step": 653
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.3106026578570133,
+      "learning_rate": 6.009836140944965e-05,
+      "loss": 1.0954,
+      "step": 654
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.3309144454564729,
+      "learning_rate": 6.002460552541331e-05,
+      "loss": 1.0209,
+      "step": 655
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.3023619281400003,
+      "learning_rate": 5.9950758677579345e-05,
+      "loss": 1.0363,
+      "step": 656
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.3311182880219704,
+      "learning_rate": 5.987682120140451e-05,
+      "loss": 1.0515,
+      "step": 657
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.33396486010030413,
+      "learning_rate": 5.980279343275729e-05,
+      "loss": 1.1251,
+      "step": 658
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.3465764556678002,
+      "learning_rate": 5.97286757079163e-05,
+      "loss": 1.165,
+      "step": 659
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.304193441363374,
+      "learning_rate": 5.965446836356882e-05,
+      "loss": 1.0228,
+      "step": 660
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.3415149030413082,
+      "learning_rate": 5.9580171736809224e-05,
+      "loss": 1.0742,
+      "step": 661
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.33138658321132064,
+      "learning_rate": 5.950578616513746e-05,
+      "loss": 1.0843,
+      "step": 662
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.30774403421162994,
+      "learning_rate": 5.943131198645752e-05,
+      "loss": 1.065,
+      "step": 663
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.3428877492183819,
+      "learning_rate": 5.9356749539075885e-05,
+      "loss": 1.1101,
+      "step": 664
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.3621290546130101,
+      "learning_rate": 5.928209916170003e-05,
+      "loss": 1.1372,
+      "step": 665
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.3482375945469884,
+      "learning_rate": 5.9207361193436865e-05,
+      "loss": 1.132,
+      "step": 666
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.31754384974068384,
+      "learning_rate": 5.9132535973791156e-05,
+      "loss": 1.148,
+      "step": 667
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.36003834782050365,
+      "learning_rate": 5.9057623842664044e-05,
+      "loss": 1.1099,
+      "step": 668
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.2963701622969662,
+      "learning_rate": 5.8982625140351464e-05,
+      "loss": 1.0755,
+      "step": 669
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.32579569606066516,
+      "learning_rate": 5.8907540207542616e-05,
+      "loss": 1.0809,
+      "step": 670
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.4247563451753457,
+      "learning_rate": 5.8832369385318416e-05,
+      "loss": 1.097,
+      "step": 671
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.33076932102169776,
+      "learning_rate": 5.875711301514992e-05,
+      "loss": 1.1078,
+      "step": 672
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.3609238032332309,
+      "learning_rate": 5.8681771438896815e-05,
+      "loss": 1.1031,
+      "step": 673
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.325159585649425,
+      "learning_rate": 5.860634499880583e-05,
+      "loss": 1.0707,
+      "step": 674
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.4620687271068983,
+      "learning_rate": 5.853083403750922e-05,
+      "loss": 1.1017,
+      "step": 675
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.33485279064365936,
+      "learning_rate": 5.845523889802316e-05,
+      "loss": 1.0989,
+      "step": 676
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.30952573170841513,
+      "learning_rate": 5.8379559923746214e-05,
+      "loss": 1.0393,
+      "step": 677
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.33498605810588283,
+      "learning_rate": 5.830379745845781e-05,
+      "loss": 1.1259,
+      "step": 678
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.35771921163037307,
+      "learning_rate": 5.822795184631659e-05,
+      "loss": 1.0815,
+      "step": 679
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.3329650192347647,
+      "learning_rate": 5.815202343185894e-05,
+      "loss": 1.1344,
+      "step": 680
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.3356634465845771,
+      "learning_rate": 5.807601255999736e-05,
+      "loss": 1.1297,
+      "step": 681
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.3289442034151235,
+      "learning_rate": 5.7999919576018934e-05,
+      "loss": 1.022,
+      "step": 682
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.3207007334784113,
+      "learning_rate": 5.7923744825583745e-05,
+      "loss": 1.0571,
+      "step": 683
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.3582460325329284,
+      "learning_rate": 5.7847488654723304e-05,
+      "loss": 1.0778,
+      "step": 684
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.3563317666176927,
+      "learning_rate": 5.777115140983899e-05,
+      "loss": 1.1003,
+      "step": 685
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 3.4694912945702105,
+      "learning_rate": 5.769473343770047e-05,
+      "loss": 1.121,
+      "step": 686
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.43002349520483113,
+      "learning_rate": 5.761823508544411e-05,
+      "loss": 1.0765,
+      "step": 687
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.39467783104839754,
+      "learning_rate": 5.754165670057142e-05,
+      "loss": 1.0788,
+      "step": 688
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.39629029674867916,
+      "learning_rate": 5.7464998630947464e-05,
+      "loss": 1.0812,
+      "step": 689
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.3880152093965208,
+      "learning_rate": 5.738826122479929e-05,
+      "loss": 1.1228,
+      "step": 690
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.3777874121959188,
+      "learning_rate": 5.7311444830714324e-05,
+      "loss": 1.0907,
+      "step": 691
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.38004041653523696,
+      "learning_rate": 5.723454979763882e-05,
+      "loss": 1.1263,
+      "step": 692
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.37049672627797636,
+      "learning_rate": 5.7157576474876246e-05,
+      "loss": 1.1438,
+      "step": 693
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.32973606103437614,
+      "learning_rate": 5.7080525212085725e-05,
+      "loss": 1.0553,
+      "step": 694
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.31674639252070325,
+      "learning_rate": 5.700339635928038e-05,
+      "loss": 1.06,
+      "step": 695
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.32282199426553837,
+      "learning_rate": 5.692619026682588e-05,
+      "loss": 1.0841,
+      "step": 696
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.4810882958061859,
+      "learning_rate": 5.684890728543869e-05,
+      "loss": 1.0803,
+      "step": 697
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.3995638550178378,
+      "learning_rate": 5.6771547766184566e-05,
+      "loss": 1.1187,
+      "step": 698
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.35264932960583484,
+      "learning_rate": 5.669411206047699e-05,
+      "loss": 1.0641,
+      "step": 699
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.35240640524733,
+      "learning_rate": 5.661660052007547e-05,
+      "loss": 1.076,
+      "step": 700
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.3540694609860389,
+      "learning_rate": 5.653901349708401e-05,
+      "loss": 1.1369,
+      "step": 701
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.3196055112925304,
+      "learning_rate": 5.646135134394955e-05,
+      "loss": 1.0677,
+      "step": 702
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.4214141007955914,
+      "learning_rate": 5.6383614413460266e-05,
+      "loss": 1.1139,
+      "step": 703
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.3625611311798579,
+      "learning_rate": 5.630580305874402e-05,
+      "loss": 1.1845,
+      "step": 704
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.3425208672181188,
+      "learning_rate": 5.62279176332668e-05,
+      "loss": 1.174,
+      "step": 705
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.3108419862818321,
+      "learning_rate": 5.6149958490830996e-05,
+      "loss": 1.0331,
+      "step": 706
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.3274644181571904,
+      "learning_rate": 5.607192598557394e-05,
+      "loss": 1.0664,
+      "step": 707
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.346218197215145,
+      "learning_rate": 5.599382047196617e-05,
+      "loss": 1.2088,
+      "step": 708
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.328497632267458,
+      "learning_rate": 5.591564230480989e-05,
+      "loss": 1.0287,
+      "step": 709
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.3708173720611468,
+      "learning_rate": 5.583739183923732e-05,
+      "loss": 1.0883,
+      "step": 710
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.3631427403535479,
+      "learning_rate": 5.575906943070915e-05,
+      "loss": 1.1155,
+      "step": 711
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.3305201458598695,
+      "learning_rate": 5.5680675435012834e-05,
+      "loss": 1.0958,
+      "step": 712
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.34978833532083714,
+      "learning_rate": 5.5602210208261036e-05,
+      "loss": 1.1437,
+      "step": 713
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.3510553882510229,
+      "learning_rate": 5.552367410688999e-05,
+      "loss": 1.0941,
+      "step": 714
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.3523747462465078,
+      "learning_rate": 5.544506748765789e-05,
+      "loss": 1.1289,
+      "step": 715
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.38262637783927445,
+      "learning_rate": 5.5366390707643266e-05,
+      "loss": 1.099,
+      "step": 716
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.38620065989073454,
+      "learning_rate": 5.528764412424334e-05,
+      "loss": 1.083,
+      "step": 717
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.3401355276121096,
+      "learning_rate": 5.520882809517245e-05,
+      "loss": 1.028,
+      "step": 718
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.3392061008943934,
+      "learning_rate": 5.512994297846039e-05,
+      "loss": 1.1083,
+      "step": 719
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.34219480421015414,
+      "learning_rate": 5.505098913245077e-05,
+      "loss": 1.1108,
+      "step": 720
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.3275058061553761,
+      "learning_rate": 5.497196691579945e-05,
+      "loss": 1.111,
+      "step": 721
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.36800249746509384,
+      "learning_rate": 5.489287668747283e-05,
+      "loss": 1.1221,
+      "step": 722
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.4129005533101575,
+      "learning_rate": 5.481371880674628e-05,
+      "loss": 1.0966,
+      "step": 723
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.36563906596251655,
+      "learning_rate": 5.4734493633202505e-05,
+      "loss": 1.0927,
+      "step": 724
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.3614650536839971,
+      "learning_rate": 5.465520152672986e-05,
+      "loss": 1.13,
+      "step": 725
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.36419665098633497,
+      "learning_rate": 5.4575842847520765e-05,
+      "loss": 1.1183,
+      "step": 726
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.34490689807258995,
+      "learning_rate": 5.449641795607005e-05,
+      "loss": 1.0919,
+      "step": 727
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.3627643746876298,
+      "learning_rate": 5.441692721317334e-05,
+      "loss": 1.0411,
+      "step": 728
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.323620411949565,
+      "learning_rate": 5.433737097992537e-05,
+      "loss": 1.0725,
+      "step": 729
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.3521599501824965,
+      "learning_rate": 5.425774961771838e-05,
+      "loss": 1.0926,
+      "step": 730
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.3302390546764222,
+      "learning_rate": 5.417806348824047e-05,
+      "loss": 1.0468,
+      "step": 731
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.3833325802616019,
+      "learning_rate": 5.4098312953473956e-05,
+      "loss": 1.1291,
+      "step": 732
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.3708621126835512,
+      "learning_rate": 5.401849837569372e-05,
+      "loss": 1.0887,
+      "step": 733
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.3625834373416278,
+      "learning_rate": 5.393862011746555e-05,
+      "loss": 1.0981,
+      "step": 734
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.3583343965080617,
+      "learning_rate": 5.385867854164451e-05,
+      "loss": 1.1021,
+      "step": 735
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.34598320594096066,
+      "learning_rate": 5.377867401137332e-05,
+      "loss": 1.1376,
+      "step": 736
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.3046382791315433,
+      "learning_rate": 5.369860689008066e-05,
+      "loss": 1.0206,
+      "step": 737
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.34464948380043725,
+      "learning_rate": 5.3618477541479505e-05,
+      "loss": 1.1084,
+      "step": 738
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.3203242519627101,
+      "learning_rate": 5.353828632956557e-05,
+      "loss": 1.0731,
+      "step": 739
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.3431169960355163,
+      "learning_rate": 5.3458033618615516e-05,
+      "loss": 1.091,
+      "step": 740
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.33492074521678705,
+      "learning_rate": 5.337771977318543e-05,
+      "loss": 1.1112,
+      "step": 741
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.32576546585541344,
+      "learning_rate": 5.3297345158109086e-05,
+      "loss": 1.0993,
+      "step": 742
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.3410007245037574,
+      "learning_rate": 5.3216910138496286e-05,
+      "loss": 1.094,
+      "step": 743
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.34891180680896833,
+      "learning_rate": 5.313641507973128e-05,
+      "loss": 1.1331,
+      "step": 744
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.37135766946717214,
+      "learning_rate": 5.3055860347471006e-05,
+      "loss": 1.1,
+      "step": 745
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.3465019415478411,
+      "learning_rate": 5.297524630764349e-05,
+      "loss": 1.1256,
+      "step": 746
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.37035388481626563,
+      "learning_rate": 5.289457332644615e-05,
+      "loss": 1.0366,
+      "step": 747
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.33853883270759155,
+      "learning_rate": 5.281384177034421e-05,
+      "loss": 1.0547,
+      "step": 748
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.364306618627317,
+      "learning_rate": 5.2733052006068897e-05,
+      "loss": 1.0768,
+      "step": 749
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.4021754315731627,
+      "learning_rate": 5.2652204400615916e-05,
+      "loss": 1.1382,
+      "step": 750
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3332185389039008,
+      "learning_rate": 5.257129932124368e-05,
+      "loss": 1.0815,
+      "step": 751
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.3453105709879854,
+      "learning_rate": 5.249033713547173e-05,
+      "loss": 1.1109,
+      "step": 752
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.3385397539717797,
+      "learning_rate": 5.2409318211078966e-05,
+      "loss": 1.0529,
+      "step": 753
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.33197994450130447,
+      "learning_rate": 5.232824291610206e-05,
+      "loss": 1.0721,
+      "step": 754
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.32836289576124167,
+      "learning_rate": 5.224711161883375e-05,
+      "loss": 1.0459,
+      "step": 755
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.32491620058831744,
+      "learning_rate": 5.216592468782117e-05,
+      "loss": 1.0897,
+      "step": 756
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.3137879047811153,
+      "learning_rate": 5.2084682491864155e-05,
+      "loss": 1.096,
+      "step": 757
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.3356938043023012,
+      "learning_rate": 5.200338540001364e-05,
+      "loss": 1.0827,
+      "step": 758
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.36044340490819055,
+      "learning_rate": 5.192203378156984e-05,
+      "loss": 1.0617,
+      "step": 759
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34674262047888293,
+      "learning_rate": 5.184062800608077e-05,
+      "loss": 1.1267,
+      "step": 760
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.32469442322149333,
+      "learning_rate": 5.1759168443340375e-05,
+      "loss": 1.1483,
+      "step": 761
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.3290384307774216,
+      "learning_rate": 5.167765546338698e-05,
+      "loss": 1.047,
+      "step": 762
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.31637612188770403,
+      "learning_rate": 5.1596089436501525e-05,
+      "loss": 1.0311,
+      "step": 763
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.3168693829641207,
+      "learning_rate": 5.151447073320597e-05,
+      "loss": 1.1405,
+      "step": 764
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.34322421571238926,
+      "learning_rate": 5.143279972426153e-05,
+      "loss": 1.1428,
+      "step": 765
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.3291030435830325,
+      "learning_rate": 5.1351076780667026e-05,
+      "loss": 1.0473,
+      "step": 766
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.33772039158758044,
+      "learning_rate": 5.1269302273657195e-05,
+      "loss": 1.0909,
+      "step": 767
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.3802031736890876,
+      "learning_rate": 5.118747657470102e-05,
+      "loss": 1.1482,
+      "step": 768
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.3296067628997962,
+      "learning_rate": 5.1105600055500025e-05,
+      "loss": 1.0085,
+      "step": 769
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.3707139982828035,
+      "learning_rate": 5.102367308798658e-05,
+      "loss": 1.0746,
+      "step": 770
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.3378537316757011,
+      "learning_rate": 5.094169604432225e-05,
+      "loss": 1.0482,
+      "step": 771
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.4008417246255145,
+      "learning_rate": 5.085966929689601e-05,
+      "loss": 1.1065,
+      "step": 772
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.3244385106988064,
+      "learning_rate": 5.077759321832271e-05,
+      "loss": 1.0827,
+      "step": 773
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.37228575732812336,
+      "learning_rate": 5.0695468181441215e-05,
+      "loss": 1.1146,
+      "step": 774
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.33761714797540276,
+      "learning_rate": 5.061329455931283e-05,
+      "loss": 1.092,
+      "step": 775
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.3158158390913494,
+      "learning_rate": 5.053107272521955e-05,
+      "loss": 1.1058,
+      "step": 776
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.3691501929738938,
+      "learning_rate": 5.044880305266239e-05,
+      "loss": 1.1599,
+      "step": 777
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.33730914019805525,
+      "learning_rate": 5.0366485915359645e-05,
+      "loss": 1.0615,
+      "step": 778
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.34970059240017,
+      "learning_rate": 5.0284121687245257e-05,
+      "loss": 1.1475,
+      "step": 779
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.3374028029407197,
+      "learning_rate": 5.020171074246707e-05,
+      "loss": 1.0926,
+      "step": 780
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.3350020681123992,
+      "learning_rate": 5.011925345538514e-05,
+      "loss": 1.1276,
+      "step": 781
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.3224228965786606,
+      "learning_rate": 5.003675020057003e-05,
+      "loss": 1.0183,
+      "step": 782
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.3357310714740298,
+      "learning_rate": 4.995420135280114e-05,
+      "loss": 1.1114,
+      "step": 783
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.3590203255363759,
+      "learning_rate": 4.9871607287064966e-05,
+      "loss": 1.1504,
+      "step": 784
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.33011195419611655,
+      "learning_rate": 4.9788968378553396e-05,
+      "loss": 1.0826,
+      "step": 785
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.31088868195439445,
+      "learning_rate": 4.970628500266207e-05,
+      "loss": 1.0704,
+      "step": 786
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.3144996103179409,
+      "learning_rate": 4.962355753498858e-05,
+      "loss": 1.1403,
+      "step": 787
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.3147269555419068,
+      "learning_rate": 4.954078635133081e-05,
+      "loss": 1.0898,
+      "step": 788
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.3280151747783868,
+      "learning_rate": 4.945797182768524e-05,
+      "loss": 1.1115,
+      "step": 789
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.3551996569232493,
+      "learning_rate": 4.937511434024524e-05,
+      "loss": 1.1731,
+      "step": 790
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.343863208057807,
+      "learning_rate": 4.9292214265399336e-05,
+      "loss": 1.0866,
+      "step": 791
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.37316699385322466,
+      "learning_rate": 4.920927197972949e-05,
+      "loss": 1.1083,
+      "step": 792
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.3635739774067832,
+      "learning_rate": 4.9126287860009453e-05,
+      "loss": 1.1393,
+      "step": 793
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.3755910554972886,
+      "learning_rate": 4.9043262283202974e-05,
+      "loss": 1.1624,
+      "step": 794
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.3635899120146823,
+      "learning_rate": 4.8960195626462145e-05,
+      "loss": 1.2095,
+      "step": 795
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.3642202684342816,
+      "learning_rate": 4.8877088267125664e-05,
+      "loss": 1.1099,
+      "step": 796
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.3339946548799316,
+      "learning_rate": 4.879394058271712e-05,
+      "loss": 1.1157,
+      "step": 797
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.3457189703100475,
+      "learning_rate": 4.871075295094329e-05,
+      "loss": 1.129,
+      "step": 798
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.3550931839691424,
+      "learning_rate": 4.862752574969241e-05,
+      "loss": 1.076,
+      "step": 799
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.36139108917966734,
+      "learning_rate": 4.8544259357032475e-05,
+      "loss": 1.1577,
+      "step": 800
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.32133637199528886,
+      "learning_rate": 4.8460954151209486e-05,
+      "loss": 1.0148,
+      "step": 801
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.043434839098495,
+      "learning_rate": 4.837761051064579e-05,
+      "loss": 1.1145,
+      "step": 802
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.36683258883118186,
+      "learning_rate": 4.8294228813938285e-05,
+      "loss": 1.0952,
+      "step": 803
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.4664840097151361,
+      "learning_rate": 4.8210809439856804e-05,
+      "loss": 1.1644,
+      "step": 804
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.355267647636684,
+      "learning_rate": 4.8127352767342276e-05,
+      "loss": 1.1062,
+      "step": 805
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.369440033402877,
+      "learning_rate": 4.8043859175505095e-05,
+      "loss": 1.0846,
+      "step": 806
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.3530796042687365,
+      "learning_rate": 4.7960329043623344e-05,
+      "loss": 1.1608,
+      "step": 807
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.4321129927586221,
+      "learning_rate": 4.787676275114111e-05,
+      "loss": 1.1339,
+      "step": 808
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.349782342376565,
+      "learning_rate": 4.779316067766673e-05,
+      "loss": 1.0685,
+      "step": 809
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.36418077406249405,
+      "learning_rate": 4.770952320297109e-05,
+      "loss": 1.1467,
+      "step": 810
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.35831016374198177,
+      "learning_rate": 4.7625850706985886e-05,
+      "loss": 1.064,
+      "step": 811
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.3596998725696811,
+      "learning_rate": 4.7542143569801894e-05,
+      "loss": 1.119,
+      "step": 812
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.3437403316058801,
+      "learning_rate": 4.745840217166725e-05,
+      "loss": 1.0451,
+      "step": 813
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.36614738184483053,
+      "learning_rate": 4.737462689298577e-05,
+      "loss": 1.1388,
+      "step": 814
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.4127884784033637,
+      "learning_rate": 4.7290818114315086e-05,
+      "loss": 1.1786,
+      "step": 815
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.4110838984805364,
+      "learning_rate": 4.72069762163651e-05,
+      "loss": 1.0709,
+      "step": 816
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.42581764215872087,
+      "learning_rate": 4.7123101579996106e-05,
+      "loss": 1.1019,
+      "step": 817
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.39442487793142056,
+      "learning_rate": 4.7039194586217136e-05,
+      "loss": 1.1532,
+      "step": 818
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3362730343659587,
+      "learning_rate": 4.695525561618418e-05,
+      "loss": 1.1149,
+      "step": 819
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5554581175413662,
+      "learning_rate": 4.687128505119853e-05,
+      "loss": 1.0697,
+      "step": 820
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.369476979421362,
+      "learning_rate": 4.6787283272704966e-05,
+      "loss": 1.1038,
+      "step": 821
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.4002553035181225,
+      "learning_rate": 4.670325066229009e-05,
+      "loss": 1.0714,
+      "step": 822
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.5196234078753172,
+      "learning_rate": 4.661918760168052e-05,
+      "loss": 1.1027,
+      "step": 823
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.39318954902323083,
+      "learning_rate": 4.653509447274121e-05,
+      "loss": 1.1562,
+      "step": 824
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.3822202885742058,
+      "learning_rate": 4.6450971657473743e-05,
+      "loss": 1.0662,
+      "step": 825
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.4203426346301331,
+      "learning_rate": 4.63668195380145e-05,
+      "loss": 1.0713,
+      "step": 826
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.3850673744429753,
+      "learning_rate": 4.628263849663301e-05,
+      "loss": 1.123,
+      "step": 827
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.3615517608466556,
+      "learning_rate": 4.619842891573016e-05,
+      "loss": 1.0565,
+      "step": 828
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.34448279386839303,
+      "learning_rate": 4.6114191177836514e-05,
+      "loss": 1.0703,
+      "step": 829
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.3649438531637055,
+      "learning_rate": 4.6029925665610524e-05,
+      "loss": 1.061,
+      "step": 830
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.360215034359632,
+      "learning_rate": 4.59456327618368e-05,
+      "loss": 1.075,
+      "step": 831
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.3297320285815758,
+      "learning_rate": 4.5861312849424386e-05,
+      "loss": 1.0177,
+      "step": 832
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.3687310180731473,
+      "learning_rate": 4.5776966311405035e-05,
+      "loss": 1.078,
+      "step": 833
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.40273770874860215,
+      "learning_rate": 4.5692593530931416e-05,
+      "loss": 1.1237,
+      "step": 834
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.41430404774660234,
+      "learning_rate": 4.560819489127545e-05,
+      "loss": 1.0891,
+      "step": 835
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.39133407251294566,
+      "learning_rate": 4.552377077582646e-05,
+      "loss": 1.1244,
+      "step": 836
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.45340195323000915,
+      "learning_rate": 4.543932156808959e-05,
+      "loss": 1.1659,
+      "step": 837
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.37488623560038786,
+      "learning_rate": 4.535484765168386e-05,
+      "loss": 1.1026,
+      "step": 838
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.36994669175389283,
+      "learning_rate": 4.527034941034063e-05,
+      "loss": 1.1481,
+      "step": 839
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 19.06482385633655,
+      "learning_rate": 4.51858272279017e-05,
+      "loss": 1.1897,
+      "step": 840
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.7974039643314161,
+      "learning_rate": 4.5101281488317634e-05,
+      "loss": 1.1467,
+      "step": 841
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1211.6099030928651,
+      "learning_rate": 4.501671257564602e-05,
+      "loss": 1.9989,
+      "step": 842
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.438002431348562,
+      "learning_rate": 4.49321208740497e-05,
+      "loss": 1.1715,
+      "step": 843
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 5.653373858566177,
+      "learning_rate": 4.484750676779504e-05,
+      "loss": 1.1828,
+      "step": 844
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 625.3410428452096,
+      "learning_rate": 4.4762870641250185e-05,
+      "loss": 1.3283,
+      "step": 845
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 5.653780923251844,
+      "learning_rate": 4.467821287888331e-05,
+      "loss": 1.1532,
+      "step": 846
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 177.17210534716736,
+      "learning_rate": 4.459353386526086e-05,
+      "loss": 1.1178,
+      "step": 847
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 4.805542680722976,
+      "learning_rate": 4.450883398504584e-05,
+      "loss": 1.1057,
+      "step": 848
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.055060058990183,
+      "learning_rate": 4.442411362299602e-05,
+      "loss": 1.1423,
+      "step": 849
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 750.168576149713,
+      "learning_rate": 4.433937316396224e-05,
+      "loss": 1.6724,
+      "step": 850
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1638,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 50,
+  "total_flos": 881390103035904.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}