diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8577 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.352018816150529,
+  "eval_steps": 700,
+  "global_step": 12000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001960015680125441,
+      "grad_norm": 7.928689002990723,
+      "learning_rate": 8.333333333333334e-08,
+      "loss": 1.4388,
+      "step": 10
+    },
+    {
+      "epoch": 0.003920031360250882,
+      "grad_norm": 9.54440689086914,
+      "learning_rate": 1.6666666666666668e-07,
+      "loss": 1.4362,
+      "step": 20
+    },
+    {
+      "epoch": 0.005880047040376323,
+      "grad_norm": 9.008377075195312,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 1.4445,
+      "step": 30
+    },
+    {
+      "epoch": 0.007840062720501764,
+      "grad_norm": 7.416823863983154,
+      "learning_rate": 3.3333333333333335e-07,
+      "loss": 1.4217,
+      "step": 40
+    },
+    {
+      "epoch": 0.009800078400627205,
+      "grad_norm": 7.977886199951172,
+      "learning_rate": 4.1666666666666667e-07,
+      "loss": 1.4243,
+      "step": 50
+    },
+    {
+      "epoch": 0.011760094080752646,
+      "grad_norm": 8.24797248840332,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 1.3976,
+      "step": 60
+    },
+    {
+      "epoch": 0.013720109760878087,
+      "grad_norm": 7.240636348724365,
+      "learning_rate": 5.833333333333334e-07,
+      "loss": 1.3853,
+      "step": 70
+    },
+    {
+      "epoch": 0.01568012544100353,
+      "grad_norm": 5.709108352661133,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 1.3353,
+      "step": 80
+    },
+    {
+      "epoch": 0.01764014112112897,
+      "grad_norm": 9.35693073272705,
+      "learning_rate": 7.5e-07,
+      "loss": 1.3051,
+      "step": 90
+    },
+    {
+      "epoch": 0.01960015680125441,
+      "grad_norm": 47.39125442504883,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 1.265,
+      "step": 100
+    },
+    {
+      "epoch": 0.02156017248137985,
+      "grad_norm": 5.6352434158325195,
+      "learning_rate": 9.166666666666666e-07,
+      "loss": 1.2583,
+      "step": 110
+    },
+    {
+      "epoch": 0.023520188161505293,
+      "grad_norm": 6.022407531738281,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.2398,
+      "step": 120
+    },
+    {
+      "epoch": 0.025480203841630734,
+      "grad_norm": 6.143012046813965,
+      "learning_rate": 1.0833333333333335e-06,
+      "loss": 1.2237,
+      "step": 130
+    },
+    {
+      "epoch": 0.027440219521756175,
+      "grad_norm": 6.173267841339111,
+      "learning_rate": 1.1666666666666668e-06,
+      "loss": 1.2246,
+      "step": 140
+    },
+    {
+      "epoch": 0.029400235201881616,
+      "grad_norm": 5.4503092765808105,
+      "learning_rate": 1.25e-06,
+      "loss": 1.1993,
+      "step": 150
+    },
+    {
+      "epoch": 0.03136025088200706,
+      "grad_norm": 17.596582412719727,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.1889,
+      "step": 160
+    },
+    {
+      "epoch": 0.033320266562132494,
+      "grad_norm": 6.131169319152832,
+      "learning_rate": 1.4166666666666667e-06,
+      "loss": 1.1884,
+      "step": 170
+    },
+    {
+      "epoch": 0.03528028224225794,
+      "grad_norm": 5.450586795806885,
+      "learning_rate": 1.5e-06,
+      "loss": 1.1622,
+      "step": 180
+    },
+    {
+      "epoch": 0.03724029792238338,
+      "grad_norm": 14.713933944702148,
+      "learning_rate": 1.5833333333333333e-06,
+      "loss": 1.1675,
+      "step": 190
+    },
+    {
+      "epoch": 0.03920031360250882,
+      "grad_norm": 8.510261535644531,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.1584,
+      "step": 200
+    },
+    {
+      "epoch": 0.04116032928263426,
+      "grad_norm": 6.9118971824646,
+      "learning_rate": 1.75e-06,
+      "loss": 1.1696,
+      "step": 210
+    },
+    {
+      "epoch": 0.0431203449627597,
+      "grad_norm": 5.986424922943115,
+      "learning_rate": 1.8333333333333333e-06,
+      "loss": 1.1734,
+      "step": 220
+    },
+    {
+      "epoch": 0.04508036064288514,
+      "grad_norm": 7.179569244384766,
+      "learning_rate": 1.916666666666667e-06,
+      "loss": 1.1563,
+      "step": 230
+    },
+    {
+      "epoch": 0.047040376323010585,
+      "grad_norm": 7.15562629699707,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.1443,
+      "step": 240
+    },
+    {
+      "epoch": 0.04900039200313602,
+      "grad_norm": 5.163909912109375,
+      "learning_rate": 2.0833333333333334e-06,
+      "loss": 1.1567,
+      "step": 250
+    },
+    {
+      "epoch": 0.05096040768326147,
+      "grad_norm": 5.861210346221924,
+      "learning_rate": 2.166666666666667e-06,
+      "loss": 1.1596,
+      "step": 260
+    },
+    {
+      "epoch": 0.052920423363386905,
+      "grad_norm": 5.1880292892456055,
+      "learning_rate": 2.25e-06,
+      "loss": 1.1428,
+      "step": 270
+    },
+    {
+      "epoch": 0.05488043904351235,
+      "grad_norm": 6.203003406524658,
+      "learning_rate": 2.3333333333333336e-06,
+      "loss": 1.1463,
+      "step": 280
+    },
+    {
+      "epoch": 0.05684045472363779,
+      "grad_norm": 6.191583156585693,
+      "learning_rate": 2.4166666666666667e-06,
+      "loss": 1.1538,
+      "step": 290
+    },
+    {
+      "epoch": 0.05880047040376323,
+      "grad_norm": 9.0908203125,
+      "learning_rate": 2.5e-06,
+      "loss": 1.1277,
+      "step": 300
+    },
+    {
+      "epoch": 0.06076048608388867,
+      "grad_norm": 5.685515880584717,
+      "learning_rate": 2.5833333333333337e-06,
+      "loss": 1.1471,
+      "step": 310
+    },
+    {
+      "epoch": 0.06272050176401411,
+      "grad_norm": 6.010474681854248,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.1232,
+      "step": 320
+    },
+    {
+      "epoch": 0.06468051744413955,
+      "grad_norm": 9.0894193649292,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 1.1286,
+      "step": 330
+    },
+    {
+      "epoch": 0.06664053312426499,
+      "grad_norm": 5.936020851135254,
+      "learning_rate": 2.8333333333333335e-06,
+      "loss": 1.1116,
+      "step": 340
+    },
+    {
+      "epoch": 0.06860054880439044,
+      "grad_norm": 6.055352210998535,
+      "learning_rate": 2.916666666666667e-06,
+      "loss": 1.1167,
+      "step": 350
+    },
+    {
+      "epoch": 0.07056056448451588,
+      "grad_norm": 5.160061836242676,
+      "learning_rate": 3e-06,
+      "loss": 1.1225,
+      "step": 360
+    },
+    {
+      "epoch": 0.07252058016464132,
+      "grad_norm": 5.748876571655273,
+      "learning_rate": 3.0833333333333336e-06,
+      "loss": 1.1298,
+      "step": 370
+    },
+    {
+      "epoch": 0.07448059584476675,
+      "grad_norm": 5.9786057472229,
+      "learning_rate": 3.1666666666666667e-06,
+      "loss": 1.1394,
+      "step": 380
+    },
+    {
+      "epoch": 0.0764406115248922,
+      "grad_norm": 5.474339008331299,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 1.138,
+      "step": 390
+    },
+    {
+      "epoch": 0.07840062720501764,
+      "grad_norm": 6.028172016143799,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.1243,
+      "step": 400
+    },
+    {
+      "epoch": 0.08036064288514308,
+      "grad_norm": 11.877293586730957,
+      "learning_rate": 3.416666666666667e-06,
+      "loss": 1.1086,
+      "step": 410
+    },
+    {
+      "epoch": 0.08232065856526852,
+      "grad_norm": 6.457036972045898,
+      "learning_rate": 3.5e-06,
+      "loss": 1.1113,
+      "step": 420
+    },
+    {
+      "epoch": 0.08428067424539397,
+      "grad_norm": 8.589061737060547,
+      "learning_rate": 3.5833333333333335e-06,
+      "loss": 1.1207,
+      "step": 430
+    },
+    {
+      "epoch": 0.0862406899255194,
+      "grad_norm": 6.367908000946045,
+      "learning_rate": 3.6666666666666666e-06,
+      "loss": 1.1192,
+      "step": 440
+    },
+    {
+      "epoch": 0.08820070560564484,
+      "grad_norm": 6.087027549743652,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 1.114,
+      "step": 450
+    },
+    {
+      "epoch": 0.09016072128577028,
+      "grad_norm": 6.365817546844482,
+      "learning_rate": 3.833333333333334e-06,
+      "loss": 1.1104,
+      "step": 460
+    },
+    {
+      "epoch": 0.09212073696589573,
+      "grad_norm": 6.38438606262207,
+      "learning_rate": 3.916666666666667e-06,
+      "loss": 1.1222,
+      "step": 470
+    },
+    {
+      "epoch": 0.09408075264602117,
+      "grad_norm": 5.983357906341553,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.1082,
+      "step": 480
+    },
+    {
+      "epoch": 0.09604076832614661,
+      "grad_norm": 7.514024257659912,
+      "learning_rate": 4.083333333333334e-06,
+      "loss": 1.1305,
+      "step": 490
+    },
+    {
+      "epoch": 0.09800078400627205,
+      "grad_norm": 5.6476922035217285,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.1183,
+      "step": 500
+    },
+    {
+      "epoch": 0.0999607996863975,
+      "grad_norm": 6.5463128089904785,
+      "learning_rate": 4.25e-06,
+      "loss": 1.1126,
+      "step": 510
+    },
+    {
+      "epoch": 0.10192081536652293,
+      "grad_norm": 6.156862258911133,
+      "learning_rate": 4.333333333333334e-06,
+      "loss": 1.1161,
+      "step": 520
+    },
+    {
+      "epoch": 0.10388083104664837,
+      "grad_norm": 5.682535648345947,
+      "learning_rate": 4.416666666666667e-06,
+      "loss": 1.1096,
+      "step": 530
+    },
+    {
+      "epoch": 0.10584084672677381,
+      "grad_norm": 7.1881489753723145,
+      "learning_rate": 4.5e-06,
+      "loss": 1.1136,
+      "step": 540
+    },
+    {
+      "epoch": 0.10780086240689926,
+      "grad_norm": 5.990985870361328,
+      "learning_rate": 4.583333333333333e-06,
+      "loss": 1.1224,
+      "step": 550
+    },
+    {
+      "epoch": 0.1097608780870247,
+      "grad_norm": 5.474586009979248,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 1.1078,
+      "step": 560
+    },
+    {
+      "epoch": 0.11172089376715014,
+      "grad_norm": 5.7349371910095215,
+      "learning_rate": 4.75e-06,
+      "loss": 1.1238,
+      "step": 570
+    },
+    {
+      "epoch": 0.11368090944727557,
+      "grad_norm": 7.513430595397949,
+      "learning_rate": 4.833333333333333e-06,
+      "loss": 1.1158,
+      "step": 580
+    },
+    {
+      "epoch": 0.11564092512740103,
+      "grad_norm": 5.410061836242676,
+      "learning_rate": 4.9166666666666665e-06,
+      "loss": 1.1102,
+      "step": 590
+    },
+    {
+      "epoch": 0.11760094080752646,
+      "grad_norm": 6.409933567047119,
+      "learning_rate": 5e-06,
+      "loss": 1.1035,
+      "step": 600
+    },
+    {
+      "epoch": 0.1195609564876519,
+      "grad_norm": 14.141221046447754,
+      "learning_rate": 5.0833333333333335e-06,
+      "loss": 1.1066,
+      "step": 610
+    },
+    {
+      "epoch": 0.12152097216777734,
+      "grad_norm": 5.547974586486816,
+      "learning_rate": 5.1666666666666675e-06,
+      "loss": 1.1218,
+      "step": 620
+    },
+    {
+      "epoch": 0.12348098784790279,
+      "grad_norm": 5.946651935577393,
+      "learning_rate": 5.2500000000000006e-06,
+      "loss": 1.112,
+      "step": 630
+    },
+    {
+      "epoch": 0.12544100352802823,
+      "grad_norm": 6.724623680114746,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.0986,
+      "step": 640
+    },
+    {
+      "epoch": 0.12740101920815367,
+      "grad_norm": 8.897869110107422,
+      "learning_rate": 5.416666666666667e-06,
+      "loss": 1.1077,
+      "step": 650
+    },
+    {
+      "epoch": 0.1293610348882791,
+      "grad_norm": 6.75730037689209,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 1.1018,
+      "step": 660
+    },
+    {
+      "epoch": 0.13132105056840454,
+      "grad_norm": 5.3133111000061035,
+      "learning_rate": 5.583333333333334e-06,
+      "loss": 1.1089,
+      "step": 670
+    },
+    {
+      "epoch": 0.13328106624852998,
+      "grad_norm": 5.644856929779053,
+      "learning_rate": 5.666666666666667e-06,
+      "loss": 1.1159,
+      "step": 680
+    },
+    {
+      "epoch": 0.13524108192865542,
+      "grad_norm": 5.264536380767822,
+      "learning_rate": 5.75e-06,
+      "loss": 1.1068,
+      "step": 690
+    },
+    {
+      "epoch": 0.13720109760878088,
+      "grad_norm": 5.970139026641846,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 1.1214,
+      "step": 700
+    },
+    {
+      "epoch": 0.13720109760878088,
+      "eval_loss": 1.0996023416519165,
+      "eval_runtime": 14.1011,
+      "eval_samples_per_second": 46.379,
+      "eval_steps_per_second": 5.815,
+      "step": 700
+    },
+    {
+      "epoch": 0.13916111328890632,
+      "grad_norm": 5.035984039306641,
+      "learning_rate": 5.916666666666667e-06,
+      "loss": 1.1078,
+      "step": 710
+    },
+    {
+      "epoch": 0.14112112896903176,
+      "grad_norm": 5.657273292541504,
+      "learning_rate": 6e-06,
+      "loss": 1.0998,
+      "step": 720
+    },
+    {
+      "epoch": 0.1430811446491572,
+      "grad_norm": 5.413987636566162,
+      "learning_rate": 6.083333333333333e-06,
+      "loss": 1.1023,
+      "step": 730
+    },
+    {
+      "epoch": 0.14504116032928263,
+      "grad_norm": 7.05068302154541,
+      "learning_rate": 6.166666666666667e-06,
+      "loss": 1.1053,
+      "step": 740
+    },
+    {
+      "epoch": 0.14700117600940807,
+      "grad_norm": 6.170547008514404,
+      "learning_rate": 6.25e-06,
+      "loss": 1.0998,
+      "step": 750
+    },
+    {
+      "epoch": 0.1489611916895335,
+      "grad_norm": 5.373660087585449,
+      "learning_rate": 6.333333333333333e-06,
+      "loss": 1.0899,
+      "step": 760
+    },
+    {
+      "epoch": 0.15092120736965894,
+      "grad_norm": 6.981970310211182,
+      "learning_rate": 6.416666666666667e-06,
+      "loss": 1.113,
+      "step": 770
+    },
+    {
+      "epoch": 0.1528812230497844,
+      "grad_norm": 4.918264865875244,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 1.0884,
+      "step": 780
+    },
+    {
+      "epoch": 0.15484123872990985,
+      "grad_norm": 13.080545425415039,
+      "learning_rate": 6.5833333333333335e-06,
+      "loss": 1.1012,
+      "step": 790
+    },
+    {
+      "epoch": 0.15680125441003528,
+      "grad_norm": 5.957456111907959,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.1044,
+      "step": 800
+    },
+    {
+      "epoch": 0.15876127009016072,
+      "grad_norm": 5.442764759063721,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 1.1051,
+      "step": 810
+    },
+    {
+      "epoch": 0.16072128577028616,
+      "grad_norm": 5.379378318786621,
+      "learning_rate": 6.833333333333334e-06,
+      "loss": 1.1092,
+      "step": 820
+    },
+    {
+      "epoch": 0.1626813014504116,
+      "grad_norm": 6.137138366699219,
+      "learning_rate": 6.916666666666667e-06,
+      "loss": 1.0835,
+      "step": 830
+    },
+    {
+      "epoch": 0.16464131713053703,
+      "grad_norm": 5.297956943511963,
+      "learning_rate": 7e-06,
+      "loss": 1.1089,
+      "step": 840
+    },
+    {
+      "epoch": 0.16660133281066247,
+      "grad_norm": 5.63748025894165,
+      "learning_rate": 7.083333333333335e-06,
+      "loss": 1.1024,
+      "step": 850
+    },
+    {
+      "epoch": 0.16856134849078794,
+      "grad_norm": 5.432644844055176,
+      "learning_rate": 7.166666666666667e-06,
+      "loss": 1.0712,
+      "step": 860
+    },
+    {
+      "epoch": 0.17052136417091338,
+      "grad_norm": 5.4105305671691895,
+      "learning_rate": 7.25e-06,
+      "loss": 1.1122,
+      "step": 870
+    },
+    {
+      "epoch": 0.1724813798510388,
+      "grad_norm": 6.026330471038818,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 1.1049,
+      "step": 880
+    },
+    {
+      "epoch": 0.17444139553116425,
+      "grad_norm": 6.143797397613525,
+      "learning_rate": 7.416666666666668e-06,
+      "loss": 1.1038,
+      "step": 890
+    },
+    {
+      "epoch": 0.1764014112112897,
+      "grad_norm": 5.558448791503906,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.095,
+      "step": 900
+    },
+    {
+      "epoch": 0.17836142689141513,
+      "grad_norm": 6.965857028961182,
+      "learning_rate": 7.583333333333333e-06,
+      "loss": 1.0979,
+      "step": 910
+    },
+    {
+      "epoch": 0.18032144257154056,
+      "grad_norm": 6.795146942138672,
+      "learning_rate": 7.666666666666667e-06,
+      "loss": 1.103,
+      "step": 920
+    },
+    {
+      "epoch": 0.182281458251666,
+      "grad_norm": 5.742541313171387,
+      "learning_rate": 7.75e-06,
+      "loss": 1.1031,
+      "step": 930
+    },
+    {
+      "epoch": 0.18424147393179147,
+      "grad_norm": 5.67035436630249,
+      "learning_rate": 7.833333333333333e-06,
+      "loss": 1.1018,
+      "step": 940
+    },
+    {
+      "epoch": 0.1862014896119169,
+      "grad_norm": 8.501360893249512,
+      "learning_rate": 7.916666666666667e-06,
+      "loss": 1.0916,
+      "step": 950
+    },
+    {
+      "epoch": 0.18816150529204234,
+      "grad_norm": 5.092253684997559,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.0886,
+      "step": 960
+    },
+    {
+      "epoch": 0.19012152097216778,
+      "grad_norm": 6.499626636505127,
+      "learning_rate": 8.083333333333334e-06,
+      "loss": 1.1008,
+      "step": 970
+    },
+    {
+      "epoch": 0.19208153665229322,
+      "grad_norm": 5.9327216148376465,
+      "learning_rate": 8.166666666666668e-06,
+      "loss": 1.0911,
+      "step": 980
+    },
+    {
+      "epoch": 0.19404155233241865,
+      "grad_norm": 5.990231990814209,
+      "learning_rate": 8.25e-06,
+      "loss": 1.0955,
+      "step": 990
+    },
+    {
+      "epoch": 0.1960015680125441,
+      "grad_norm": 32.17470932006836,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.1087,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19796158369266953,
+      "grad_norm": 5.19512414932251,
+      "learning_rate": 8.416666666666667e-06,
+      "loss": 1.1302,
+      "step": 1010
+    },
+    {
+      "epoch": 0.199921599372795,
+      "grad_norm": 5.765759468078613,
+      "learning_rate": 8.5e-06,
+      "loss": 1.1089,
+      "step": 1020
+    },
+    {
+      "epoch": 0.20188161505292043,
+      "grad_norm": 5.5894927978515625,
+      "learning_rate": 8.583333333333333e-06,
+      "loss": 1.0914,
+      "step": 1030
+    },
+    {
+      "epoch": 0.20384163073304587,
+      "grad_norm": 6.387049198150635,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 1.095,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2058016464131713,
+      "grad_norm": 5.629969596862793,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 1.0999,
+      "step": 1050
+    },
+    {
+      "epoch": 0.20776166209329675,
+      "grad_norm": 22.919910430908203,
+      "learning_rate": 8.833333333333334e-06,
+      "loss": 1.096,
+      "step": 1060
+    },
+    {
+      "epoch": 0.20972167777342218,
+      "grad_norm": 5.267335414886475,
+      "learning_rate": 8.916666666666667e-06,
+      "loss": 1.1006,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21168169345354762,
+      "grad_norm": 53.39712905883789,
+      "learning_rate": 9e-06,
+      "loss": 1.106,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21364170913367306,
+      "grad_norm": 5.858781814575195,
+      "learning_rate": 9.083333333333333e-06,
+      "loss": 1.2549,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21560172481379852,
+      "grad_norm": 6.5703935623168945,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 1.094,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21756174049392396,
+      "grad_norm": 5.414449214935303,
+      "learning_rate": 9.250000000000001e-06,
+      "loss": 1.0891,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2195217561740494,
+      "grad_norm": 7.185405731201172,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 1.089,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22148177185417484,
+      "grad_norm": 6.4889092445373535,
+      "learning_rate": 9.416666666666667e-06,
+      "loss": 1.0874,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22344178753430027,
+      "grad_norm": 6.4478759765625,
+      "learning_rate": 9.5e-06,
+      "loss": 1.1111,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2254018032144257,
+      "grad_norm": 6.49618673324585,
+      "learning_rate": 9.583333333333335e-06,
+      "loss": 1.1001,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22736181889455115,
+      "grad_norm": 7.893542766571045,
+      "learning_rate": 9.666666666666667e-06,
+      "loss": 1.097,
+      "step": 1160
+    },
+    {
+      "epoch": 0.2293218345746766,
+      "grad_norm": 11.165901184082031,
+      "learning_rate": 9.75e-06,
+      "loss": 1.092,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23128185025480205,
+      "grad_norm": 21.967939376831055,
+      "learning_rate": 9.833333333333333e-06,
+      "loss": 1.1216,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2332418659349275,
+      "grad_norm": 14.317709922790527,
+      "learning_rate": 9.916666666666668e-06,
+      "loss": 1.1047,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23520188161505293,
+      "grad_norm": 6.240993499755859,
+      "learning_rate": 1e-05,
+      "loss": 1.1045,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23716189729517836,
+      "grad_norm": 5.804283142089844,
+      "learning_rate": 9.990740740740741e-06,
+      "loss": 1.0933,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2391219129753038,
+      "grad_norm": 5.778935432434082,
+      "learning_rate": 9.981481481481482e-06,
+      "loss": 1.1098,
+      "step": 1220
+    },
+    {
+      "epoch": 0.24108192865542924,
+      "grad_norm": 6.391726016998291,
+      "learning_rate": 9.972222222222224e-06,
+      "loss": 1.0993,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24304194433555468,
+      "grad_norm": 4.967405319213867,
+      "learning_rate": 9.962962962962964e-06,
+      "loss": 1.0835,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24500196001568011,
+      "grad_norm": 11.337072372436523,
+      "learning_rate": 9.953703703703704e-06,
+      "loss": 1.1057,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24696197569580558,
+      "grad_norm": 5.182584285736084,
+      "learning_rate": 9.944444444444445e-06,
+      "loss": 1.0913,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24892199137593102,
+      "grad_norm": 6.305624961853027,
+      "learning_rate": 9.935185185185185e-06,
+      "loss": 1.0963,
+      "step": 1270
+    },
+    {
+      "epoch": 0.25088200705605646,
+      "grad_norm": 5.263504981994629,
+      "learning_rate": 9.925925925925927e-06,
+      "loss": 1.1097,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2528420227361819,
+      "grad_norm": 5.2210493087768555,
+      "learning_rate": 9.916666666666668e-06,
+      "loss": 1.0934,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25480203841630733,
+      "grad_norm": 6.707633972167969,
+      "learning_rate": 9.907407407407408e-06,
+      "loss": 1.1049,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25676205409643277,
+      "grad_norm": 5.195355415344238,
+      "learning_rate": 9.898148148148148e-06,
+      "loss": 1.0866,
+      "step": 1310
+    },
+    {
+      "epoch": 0.2587220697765582,
+      "grad_norm": 7.137194633483887,
+      "learning_rate": 9.88888888888889e-06,
+      "loss": 1.1032,
+      "step": 1320
+    },
+    {
+      "epoch": 0.26068208545668364,
+      "grad_norm": 5.271111488342285,
+      "learning_rate": 9.87962962962963e-06,
+      "loss": 1.0925,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2626421011368091,
+      "grad_norm": 6.781525135040283,
+      "learning_rate": 9.870370370370371e-06,
+      "loss": 1.0966,
+      "step": 1340
+    },
+    {
+      "epoch": 0.2646021168169345,
+      "grad_norm": 8.270143508911133,
+      "learning_rate": 9.861111111111112e-06,
+      "loss": 1.0991,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26656213249705996,
+      "grad_norm": 6.368267059326172,
+      "learning_rate": 9.851851851851852e-06,
+      "loss": 1.1005,
+      "step": 1360
+    },
+    {
+      "epoch": 0.2685221481771854,
+      "grad_norm": 5.750402927398682,
+      "learning_rate": 9.842592592592594e-06,
+      "loss": 1.0907,
+      "step": 1370
+    },
+    {
+      "epoch": 0.27048216385731083,
+      "grad_norm": 5.706563472747803,
+      "learning_rate": 9.833333333333333e-06,
+      "loss": 1.1133,
+      "step": 1380
+    },
+    {
+      "epoch": 0.2724421795374363,
+      "grad_norm": 5.816508769989014,
+      "learning_rate": 9.824074074074075e-06,
+      "loss": 1.1074,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27440219521756176,
+      "grad_norm": 4.7041449546813965,
+      "learning_rate": 9.814814814814815e-06,
+      "loss": 1.0944,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27440219521756176,
+      "eval_loss": 1.089659571647644,
+      "eval_runtime": 14.1203,
+      "eval_samples_per_second": 46.316,
+      "eval_steps_per_second": 5.807,
+      "step": 1400
+    },
+    {
+      "epoch": 0.2763622108976872,
+      "grad_norm": 4.5806355476379395,
+      "learning_rate": 9.805555555555556e-06,
+      "loss": 1.0868,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27832222657781264,
+      "grad_norm": 6.664804458618164,
+      "learning_rate": 9.796296296296298e-06,
+      "loss": 1.1031,
+      "step": 1420
+    },
+    {
+      "epoch": 0.2802822422579381,
+      "grad_norm": 5.814957141876221,
+      "learning_rate": 9.787037037037038e-06,
+      "loss": 1.0818,
+      "step": 1430
+    },
+    {
+      "epoch": 0.2822422579380635,
+      "grad_norm": 11.043229103088379,
+      "learning_rate": 9.777777777777779e-06,
+      "loss": 1.1147,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28420227361818895,
+      "grad_norm": 5.907972812652588,
+      "learning_rate": 9.768518518518519e-06,
+      "loss": 1.0949,
+      "step": 1450
+    },
+    {
+      "epoch": 0.2861622892983144,
+      "grad_norm": 8.000860214233398,
+      "learning_rate": 9.759259259259261e-06,
+      "loss": 1.107,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2881223049784398,
+      "grad_norm": 5.279980182647705,
+      "learning_rate": 9.75e-06,
+      "loss": 1.0988,
+      "step": 1470
+    },
+    {
+      "epoch": 0.29008232065856526,
+      "grad_norm": 6.427779674530029,
+      "learning_rate": 9.740740740740742e-06,
+      "loss": 1.0969,
+      "step": 1480
+    },
+    {
+      "epoch": 0.2920423363386907,
+      "grad_norm": 5.733342170715332,
+      "learning_rate": 9.731481481481482e-06,
+      "loss": 1.0832,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29400235201881614,
+      "grad_norm": 5.370787620544434,
+      "learning_rate": 9.722222222222223e-06,
+      "loss": 1.0724,
+      "step": 1500
+    },
+    {
+      "epoch": 0.2959623676989416,
+      "grad_norm": 4.780156135559082,
+      "learning_rate": 9.712962962962965e-06,
+      "loss": 1.0911,
+      "step": 1510
+    },
+    {
+      "epoch": 0.297922383379067,
+      "grad_norm": 5.770766258239746,
+      "learning_rate": 9.703703703703703e-06,
+      "loss": 1.0913,
+      "step": 1520
+    },
+    {
+      "epoch": 0.29988239905919245,
+      "grad_norm": 5.799376487731934,
+      "learning_rate": 9.694444444444446e-06,
+      "loss": 1.1072,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3018424147393179,
+      "grad_norm": 4.571477890014648,
+      "learning_rate": 9.685185185185186e-06,
+      "loss": 1.0977,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3038024304194434,
+      "grad_norm": 6.020542144775391,
+      "learning_rate": 9.675925925925926e-06,
+      "loss": 1.1056,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3057624460995688,
+      "grad_norm": 6.760439395904541,
+      "learning_rate": 9.666666666666667e-06,
+      "loss": 1.0874,
+      "step": 1560
+    },
+    {
+      "epoch": 0.30772246177969426,
+      "grad_norm": 4.900004863739014,
+      "learning_rate": 9.657407407407409e-06,
+      "loss": 1.0946,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3096824774598197,
+      "grad_norm": 4.82609748840332,
+      "learning_rate": 9.64814814814815e-06,
+      "loss": 1.1017,
+      "step": 1580
+    },
+    {
+      "epoch": 0.31164249313994513,
+      "grad_norm": 5.115661144256592,
+      "learning_rate": 9.63888888888889e-06,
+      "loss": 1.1073,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31360250882007057,
+      "grad_norm": 4.987372875213623,
+      "learning_rate": 9.62962962962963e-06,
+      "loss": 1.084,
+      "step": 1600
+    },
+    {
+      "epoch": 0.315562524500196,
+      "grad_norm": 5.010407447814941,
+      "learning_rate": 9.62037037037037e-06,
+      "loss": 1.1031,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31752254018032144,
+      "grad_norm": 4.8331122398376465,
+      "learning_rate": 9.611111111111112e-06,
+      "loss": 1.0995,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3194825558604469,
+      "grad_norm": 5.5287909507751465,
+      "learning_rate": 9.601851851851853e-06,
+      "loss": 1.0884,
+      "step": 1630
+    },
+    {
+      "epoch": 0.3214425715405723,
+      "grad_norm": 7.749701023101807,
+      "learning_rate": 9.592592592592593e-06,
+      "loss": 1.0786,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32340258722069776,
+      "grad_norm": 4.908470153808594,
+      "learning_rate": 9.583333333333335e-06,
+      "loss": 1.1049,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3253626029008232,
+      "grad_norm": 8.760143280029297,
+      "learning_rate": 9.574074074074074e-06,
+      "loss": 1.1,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32732261858094863,
+      "grad_norm": 4.7910590171813965,
+      "learning_rate": 9.564814814814816e-06,
+      "loss": 1.0936,
+      "step": 1670
+    },
+    {
+      "epoch": 0.32928263426107407,
+      "grad_norm": 10.59984302520752,
+      "learning_rate": 9.555555555555556e-06,
+      "loss": 1.0888,
+      "step": 1680
+    },
+    {
+      "epoch": 0.3312426499411995,
+      "grad_norm": 5.18034029006958,
+      "learning_rate": 9.546296296296297e-06,
+      "loss": 1.1015,
+      "step": 1690
+    },
+    {
+      "epoch": 0.33320266562132494,
+      "grad_norm": 4.835277557373047,
+      "learning_rate": 9.537037037037037e-06,
+      "loss": 1.0934,
+      "step": 1700
+    },
+    {
+      "epoch": 0.33516268130145044,
+      "grad_norm": 5.690189838409424,
+      "learning_rate": 9.527777777777778e-06,
+      "loss": 1.0981,
+      "step": 1710
+    },
+    {
+      "epoch": 0.3371226969815759,
+      "grad_norm": 5.126386642456055,
+      "learning_rate": 9.51851851851852e-06,
+      "loss": 1.0731,
+      "step": 1720
+    },
+    {
+      "epoch": 0.3390827126617013,
+      "grad_norm": 5.041740894317627,
+      "learning_rate": 9.50925925925926e-06,
+      "loss": 1.0822,
+      "step": 1730
+    },
+    {
+      "epoch": 0.34104272834182675,
+      "grad_norm": 5.251774787902832,
+      "learning_rate": 9.5e-06,
+      "loss": 1.0914,
+      "step": 1740
+    },
+    {
+      "epoch": 0.3430027440219522,
+      "grad_norm": 4.976339817047119,
+      "learning_rate": 9.490740740740741e-06,
+      "loss": 1.0994,
+      "step": 1750
+    },
+    {
+      "epoch": 0.3449627597020776,
+      "grad_norm": 4.93164587020874,
+      "learning_rate": 9.481481481481483e-06,
+      "loss": 1.1029,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34692277538220306,
+      "grad_norm": 4.964654445648193,
+      "learning_rate": 9.472222222222223e-06,
+      "loss": 1.0846,
+      "step": 1770
+    },
+    {
+      "epoch": 0.3488827910623285,
+      "grad_norm": 5.763545036315918,
+      "learning_rate": 9.462962962962964e-06,
+      "loss": 1.0861,
+      "step": 1780
+    },
+    {
+      "epoch": 0.35084280674245394,
+      "grad_norm": 7.256554126739502,
+      "learning_rate": 9.453703703703704e-06,
+      "loss": 1.0878,
+      "step": 1790
+    },
+    {
+      "epoch": 0.3528028224225794,
+      "grad_norm": 5.118113994598389,
+      "learning_rate": 9.444444444444445e-06,
+      "loss": 1.0814,
+      "step": 1800
+    },
+    {
+      "epoch": 0.3547628381027048,
+      "grad_norm": 5.370246410369873,
+      "learning_rate": 9.435185185185187e-06,
+      "loss": 1.0977,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35672285378283025,
+      "grad_norm": 4.952902793884277,
+      "learning_rate": 9.425925925925925e-06,
+      "loss": 1.0889,
+      "step": 1820
+    },
+    {
+      "epoch": 0.3586828694629557,
+      "grad_norm": 4.842602252960205,
+      "learning_rate": 9.416666666666667e-06,
+      "loss": 1.0926,
+      "step": 1830
+    },
+    {
+      "epoch": 0.3606428851430811,
+      "grad_norm": 4.805694103240967,
+      "learning_rate": 9.407407407407408e-06,
+      "loss": 1.1087,
+      "step": 1840
+    },
+    {
+      "epoch": 0.36260290082320656,
+      "grad_norm": 5.138493061065674,
+      "learning_rate": 9.398148148148148e-06,
+      "loss": 1.0862,
+      "step": 1850
+    },
+    {
+      "epoch": 0.364562916503332,
+      "grad_norm": 7.099350929260254,
+      "learning_rate": 9.38888888888889e-06,
+      "loss": 1.0887,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3665229321834575,
+      "grad_norm": 5.295989990234375,
+      "learning_rate": 9.37962962962963e-06,
+      "loss": 1.0715,
+      "step": 1870
+    },
+    {
+      "epoch": 0.36848294786358293,
+      "grad_norm": 5.222052097320557,
+      "learning_rate": 9.370370370370371e-06,
+      "loss": 1.0979,
+      "step": 1880
+    },
+    {
+      "epoch": 0.37044296354370837,
+      "grad_norm": 4.347951889038086,
+      "learning_rate": 9.361111111111111e-06,
+      "loss": 1.1036,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3724029792238338,
+      "grad_norm": 5.062961101531982,
+      "learning_rate": 9.351851851851854e-06,
+      "loss": 1.0864,
+      "step": 1900
+    },
+    {
+      "epoch": 0.37436299490395925,
+      "grad_norm": 5.308873176574707,
+      "learning_rate": 9.342592592592594e-06,
+      "loss": 1.0877,
+      "step": 1910
+    },
+    {
+      "epoch": 0.3763230105840847,
+      "grad_norm": 4.578522682189941,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 1.0961,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3782830262642101,
+      "grad_norm": 4.571235179901123,
+      "learning_rate": 9.324074074074075e-06,
+      "loss": 1.0838,
+      "step": 1930
+    },
+    {
+      "epoch": 0.38024304194433556,
+      "grad_norm": 4.522839069366455,
+      "learning_rate": 9.314814814814815e-06,
+      "loss": 1.0786,
+      "step": 1940
+    },
+    {
+      "epoch": 0.382203057624461,
+      "grad_norm": 4.68695068359375,
+      "learning_rate": 9.305555555555557e-06,
+      "loss": 1.1146,
+      "step": 1950
+    },
+    {
+      "epoch": 0.38416307330458643,
+      "grad_norm": 4.976430416107178,
+      "learning_rate": 9.296296296296296e-06,
+      "loss": 1.0937,
+      "step": 1960
+    },
+    {
+      "epoch": 0.38612308898471187,
+      "grad_norm": 5.086645126342773,
+      "learning_rate": 9.287037037037038e-06,
+      "loss": 1.097,
+      "step": 1970
+    },
+    {
+      "epoch": 0.3880831046648373,
+      "grad_norm": 4.946636199951172,
+      "learning_rate": 9.277777777777778e-06,
+      "loss": 1.0986,
+      "step": 1980
+    },
+    {
+      "epoch": 0.39004312034496275,
+      "grad_norm": 4.823138236999512,
+      "learning_rate": 9.268518518518519e-06,
+      "loss": 1.1034,
+      "step": 1990
+    },
+    {
+      "epoch": 0.3920031360250882,
+      "grad_norm": 5.26657772064209,
+      "learning_rate": 9.25925925925926e-06,
+      "loss": 1.0798,
+      "step": 2000
+    },
+    {
+      "epoch": 0.3939631517052136,
+      "grad_norm": 5.231563568115234,
+      "learning_rate": 9.250000000000001e-06,
+      "loss": 1.0944,
+      "step": 2010
+    },
+    {
+      "epoch": 0.39592316738533906,
+      "grad_norm": 4.791154384613037,
+      "learning_rate": 9.240740740740742e-06,
+      "loss": 1.0949,
+      "step": 2020
+    },
+    {
+      "epoch": 0.3978831830654645,
+      "grad_norm": 4.691703796386719,
+      "learning_rate": 9.231481481481482e-06,
+      "loss": 1.0856,
+      "step": 2030
+    },
+    {
+      "epoch": 0.39984319874559,
+      "grad_norm": 7.213362693786621,
+      "learning_rate": 9.222222222222224e-06,
+      "loss": 1.0922,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4018032144257154,
+      "grad_norm": 6.0655646324157715,
+      "learning_rate": 9.212962962962963e-06,
+      "loss": 1.0948,
+      "step": 2050
+    },
+    {
+      "epoch": 0.40376323010584086,
+      "grad_norm": 4.249541759490967,
+      "learning_rate": 9.203703703703705e-06,
+      "loss": 1.091,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4057232457859663,
+      "grad_norm": 4.832751750946045,
+      "learning_rate": 9.194444444444445e-06,
+      "loss": 1.0907,
+      "step": 2070
+    },
+    {
+      "epoch": 0.40768326146609174,
+      "grad_norm": 5.216608047485352,
+      "learning_rate": 9.185185185185186e-06,
+      "loss": 1.0802,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4096432771462172,
+      "grad_norm": 5.988987445831299,
+      "learning_rate": 9.175925925925928e-06,
+      "loss": 1.0938,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4116032928263426,
+      "grad_norm": 4.744123458862305,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 1.0889,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4116032928263426,
+      "eval_loss": 1.08055579662323,
+      "eval_runtime": 14.0694,
+      "eval_samples_per_second": 46.484,
+      "eval_steps_per_second": 5.828,
+      "step": 2100
+    },
+    {
+      "epoch": 0.41356330850646805,
+      "grad_norm": 5.0730485916137695,
+      "learning_rate": 9.157407407407409e-06,
+      "loss": 1.0956,
+      "step": 2110
+    },
+    {
+      "epoch": 0.4155233241865935,
+      "grad_norm": 4.156872749328613,
+      "learning_rate": 9.148148148148149e-06,
+      "loss": 1.0899,
+      "step": 2120
+    },
+    {
+      "epoch": 0.4174833398667189,
+      "grad_norm": 5.194650173187256,
+      "learning_rate": 9.13888888888889e-06,
+      "loss": 1.0772,
+      "step": 2130
+    },
+    {
+      "epoch": 0.41944335554684437,
+      "grad_norm": 5.057788848876953,
+      "learning_rate": 9.12962962962963e-06,
+      "loss": 1.0886,
+      "step": 2140
+    },
+    {
+      "epoch": 0.4214033712269698,
+      "grad_norm": 4.849510192871094,
+      "learning_rate": 9.120370370370372e-06,
+      "loss": 1.1027,
+      "step": 2150
+    },
+    {
+      "epoch": 0.42336338690709524,
+      "grad_norm": 4.8577704429626465,
+      "learning_rate": 9.111111111111112e-06,
+      "loss": 1.0809,
+      "step": 2160
+    },
+    {
+      "epoch": 0.4253234025872207,
+      "grad_norm": 4.937686443328857,
+      "learning_rate": 9.101851851851853e-06,
+      "loss": 1.0819,
+      "step": 2170
+    },
+    {
+      "epoch": 0.4272834182673461,
+      "grad_norm": 5.295153617858887,
+      "learning_rate": 9.092592592592593e-06,
+      "loss": 1.0734,
+      "step": 2180
+    },
+    {
+      "epoch": 0.42924343394747155,
+      "grad_norm": 5.876978874206543,
+      "learning_rate": 9.083333333333333e-06,
+      "loss": 1.078,
+      "step": 2190
+    },
+    {
+      "epoch": 0.43120344962759705,
+      "grad_norm": 4.725461006164551,
+      "learning_rate": 9.074074074074075e-06,
+      "loss": 1.0787,
+      "step": 2200
+    },
+    {
+      "epoch": 0.4331634653077225,
+      "grad_norm": 4.425513744354248,
+      "learning_rate": 9.064814814814816e-06,
+      "loss": 1.1066,
+      "step": 2210
+    },
+    {
+      "epoch": 0.4351234809878479,
+      "grad_norm": 4.781891822814941,
+      "learning_rate": 9.055555555555556e-06,
+      "loss": 1.083,
+      "step": 2220
+    },
+    {
+      "epoch": 0.43708349666797336,
+      "grad_norm": 5.292329788208008,
+      "learning_rate": 9.046296296296298e-06,
+      "loss": 1.094,
+      "step": 2230
+    },
+    {
+      "epoch": 0.4390435123480988,
+      "grad_norm": 5.246599197387695,
+      "learning_rate": 9.037037037037037e-06,
+      "loss": 1.0903,
+      "step": 2240
+    },
+    {
+      "epoch": 0.44100352802822423,
+      "grad_norm": 4.701263427734375,
+      "learning_rate": 9.027777777777779e-06,
+      "loss": 1.0811,
+      "step": 2250
+    },
+    {
+      "epoch": 0.44296354370834967,
+      "grad_norm": 5.434783458709717,
+      "learning_rate": 9.01851851851852e-06,
+      "loss": 1.0862,
+      "step": 2260
+    },
+    {
+      "epoch": 0.4449235593884751,
+      "grad_norm": 5.898828506469727,
+      "learning_rate": 9.00925925925926e-06,
+      "loss": 1.0972,
+      "step": 2270
+    },
+    {
+      "epoch": 0.44688357506860055,
+      "grad_norm": 5.089272975921631,
+      "learning_rate": 9e-06,
+      "loss": 1.091,
+      "step": 2280
+    },
+    {
+      "epoch": 0.448843590748726,
+      "grad_norm": 6.887242317199707,
+      "learning_rate": 8.99074074074074e-06,
+      "loss": 1.0785,
+      "step": 2290
+    },
+    {
+      "epoch": 0.4508036064288514,
+      "grad_norm": 5.111120700836182,
+      "learning_rate": 8.981481481481483e-06,
+      "loss": 1.0682,
+      "step": 2300
+    },
+    {
+      "epoch": 0.45276362210897686,
+      "grad_norm": 5.331275939941406,
+      "learning_rate": 8.972222222222223e-06,
+      "loss": 1.0854,
+      "step": 2310
+    },
+    {
+      "epoch": 0.4547236377891023,
+      "grad_norm": 5.59519100189209,
+      "learning_rate": 8.962962962962963e-06,
+      "loss": 1.0883,
+      "step": 2320
+    },
+    {
+      "epoch": 0.45668365346922773,
+      "grad_norm": 5.6016364097595215,
+      "learning_rate": 8.953703703703704e-06,
+      "loss": 1.0799,
+      "step": 2330
+    },
+    {
+      "epoch": 0.4586436691493532,
+      "grad_norm": 4.625690460205078,
+      "learning_rate": 8.944444444444446e-06,
+      "loss": 1.0716,
+      "step": 2340
+    },
+    {
+      "epoch": 0.4606036848294786,
+      "grad_norm": 4.6536688804626465,
+      "learning_rate": 8.935185185185186e-06,
+      "loss": 1.0624,
+      "step": 2350
+    },
+    {
+      "epoch": 0.4625637005096041,
+      "grad_norm": 4.807366847991943,
+      "learning_rate": 8.925925925925927e-06,
+      "loss": 1.0829,
+      "step": 2360
+    },
+    {
+      "epoch": 0.46452371618972954,
+      "grad_norm": 4.950624942779541,
+      "learning_rate": 8.916666666666667e-06,
+      "loss": 1.0861,
+      "step": 2370
+    },
+    {
+      "epoch": 0.466483731869855,
+      "grad_norm": 4.980051517486572,
+      "learning_rate": 8.907407407407408e-06,
+      "loss": 1.0772,
+      "step": 2380
+    },
+    {
+      "epoch": 0.4684437475499804,
+      "grad_norm": 5.208264350891113,
+      "learning_rate": 8.89814814814815e-06,
+      "loss": 1.0929,
+      "step": 2390
+    },
+    {
+      "epoch": 0.47040376323010585,
+      "grad_norm": 5.6281657218933105,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 1.0818,
+      "step": 2400
+    },
+    {
+      "epoch": 0.4723637789102313,
+      "grad_norm": 6.125396251678467,
+      "learning_rate": 8.87962962962963e-06,
+      "loss": 1.0865,
+      "step": 2410
+    },
+    {
+      "epoch": 0.47432379459035673,
+      "grad_norm": 5.989163398742676,
+      "learning_rate": 8.87037037037037e-06,
+      "loss": 1.0685,
+      "step": 2420
+    },
+    {
+      "epoch": 0.47628381027048217,
+      "grad_norm": 9.247689247131348,
+      "learning_rate": 8.861111111111111e-06,
+      "loss": 1.0879,
+      "step": 2430
+    },
+    {
+      "epoch": 0.4782438259506076,
+      "grad_norm": 4.848996639251709,
+      "learning_rate": 8.851851851851853e-06,
+      "loss": 1.0783,
+      "step": 2440
+    },
+    {
+      "epoch": 0.48020384163073304,
+      "grad_norm": 5.04819917678833,
+      "learning_rate": 8.842592592592594e-06,
+      "loss": 1.0679,
+      "step": 2450
+    },
+    {
+      "epoch": 0.4821638573108585,
+      "grad_norm": 5.755705833435059,
+      "learning_rate": 8.833333333333334e-06,
+      "loss": 1.0939,
+      "step": 2460
+    },
+    {
+      "epoch": 0.4841238729909839,
+      "grad_norm": 4.5991058349609375,
+      "learning_rate": 8.824074074074074e-06,
+      "loss": 1.0775,
+      "step": 2470
+    },
+    {
+      "epoch": 0.48608388867110935,
+      "grad_norm": 5.101771354675293,
+      "learning_rate": 8.814814814814817e-06,
+      "loss": 1.084,
+      "step": 2480
+    },
+    {
+      "epoch": 0.4880439043512348,
+      "grad_norm": 4.9321441650390625,
+      "learning_rate": 8.805555555555557e-06,
+      "loss": 1.0762,
+      "step": 2490
+    },
+    {
+      "epoch": 0.49000392003136023,
+      "grad_norm": 5.363485813140869,
+      "learning_rate": 8.796296296296297e-06,
+      "loss": 1.0595,
+      "step": 2500
+    },
+    {
+      "epoch": 0.49196393571148567,
+      "grad_norm": 5.353972434997559,
+      "learning_rate": 8.787037037037038e-06,
+      "loss": 1.0689,
+      "step": 2510
+    },
+    {
+      "epoch": 0.49392395139161116,
+      "grad_norm": 5.158483505249023,
+      "learning_rate": 8.777777777777778e-06,
+      "loss": 1.0711,
+      "step": 2520
+    },
+    {
+      "epoch": 0.4958839670717366,
+      "grad_norm": 5.160126686096191,
+      "learning_rate": 8.76851851851852e-06,
+      "loss": 1.0886,
+      "step": 2530
+    },
+    {
+      "epoch": 0.49784398275186204,
+      "grad_norm": 4.926945686340332,
+      "learning_rate": 8.759259259259259e-06,
+      "loss": 1.0856,
+      "step": 2540
+    },
+    {
+      "epoch": 0.4998039984319875,
+      "grad_norm": 5.253294467926025,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 1.0876,
+      "step": 2550
+    },
+    {
+      "epoch": 0.5017640141121129,
+      "grad_norm": 11.674654960632324,
+      "learning_rate": 8.740740740740741e-06,
+      "loss": 1.0784,
+      "step": 2560
+    },
+    {
+      "epoch": 0.5037240297922383,
+      "grad_norm": 5.032644271850586,
+      "learning_rate": 8.731481481481482e-06,
+      "loss": 1.07,
+      "step": 2570
+    },
+    {
+      "epoch": 0.5056840454723638,
+      "grad_norm": 5.1500372886657715,
+      "learning_rate": 8.722222222222224e-06,
+      "loss": 1.0932,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5076440611524892,
+      "grad_norm": 5.761635780334473,
+      "learning_rate": 8.712962962962964e-06,
+      "loss": 1.0816,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5096040768326147,
+      "grad_norm": 5.257477760314941,
+      "learning_rate": 8.703703703703705e-06,
+      "loss": 1.0712,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5115640925127402,
+      "grad_norm": 5.4931416511535645,
+      "learning_rate": 8.694444444444445e-06,
+      "loss": 1.0832,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5135241081928655,
+      "grad_norm": 6.102149963378906,
+      "learning_rate": 8.685185185185185e-06,
+      "loss": 1.076,
+      "step": 2620
+    },
+    {
+      "epoch": 0.515484123872991,
+      "grad_norm": 5.596585273742676,
+      "learning_rate": 8.675925925925926e-06,
+      "loss": 1.0972,
+      "step": 2630
+    },
+    {
+      "epoch": 0.5174441395531164,
+      "grad_norm": 4.6653523445129395,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 1.0872,
+      "step": 2640
+    },
+    {
+      "epoch": 0.5194041552332419,
+      "grad_norm": 4.938266277313232,
+      "learning_rate": 8.657407407407408e-06,
+      "loss": 1.0717,
+      "step": 2650
+    },
+    {
+      "epoch": 0.5213641709133673,
+      "grad_norm": 5.11916446685791,
+      "learning_rate": 8.648148148148149e-06,
+      "loss": 1.0955,
+      "step": 2660
+    },
+    {
+      "epoch": 0.5233241865934928,
+      "grad_norm": 4.867392539978027,
+      "learning_rate": 8.63888888888889e-06,
+      "loss": 1.0769,
+      "step": 2670
+    },
+    {
+      "epoch": 0.5252842022736182,
+      "grad_norm": 5.589994430541992,
+      "learning_rate": 8.62962962962963e-06,
+      "loss": 1.0808,
+      "step": 2680
+    },
+    {
+      "epoch": 0.5272442179537437,
+      "grad_norm": 5.221999645233154,
+      "learning_rate": 8.620370370370371e-06,
+      "loss": 1.0627,
+      "step": 2690
+    },
+    {
+      "epoch": 0.529204233633869,
+      "grad_norm": 4.844895839691162,
+      "learning_rate": 8.611111111111112e-06,
+      "loss": 1.0951,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5311642493139945,
+      "grad_norm": 4.937107563018799,
+      "learning_rate": 8.601851851851852e-06,
+      "loss": 1.0898,
+      "step": 2710
+    },
+    {
+      "epoch": 0.5331242649941199,
+      "grad_norm": 4.758755207061768,
+      "learning_rate": 8.592592592592593e-06,
+      "loss": 1.1017,
+      "step": 2720
+    },
+    {
+      "epoch": 0.5350842806742454,
+      "grad_norm": 5.145150661468506,
+      "learning_rate": 8.583333333333333e-06,
+      "loss": 1.0702,
+      "step": 2730
+    },
+    {
+      "epoch": 0.5370442963543708,
+      "grad_norm": 5.257258415222168,
+      "learning_rate": 8.574074074074075e-06,
+      "loss": 1.0822,
+      "step": 2740
+    },
+    {
+      "epoch": 0.5390043120344963,
+      "grad_norm": 5.826773166656494,
+      "learning_rate": 8.564814814814816e-06,
+      "loss": 1.0714,
+      "step": 2750
+    },
+    {
+      "epoch": 0.5409643277146217,
+      "grad_norm": 4.5895304679870605,
+      "learning_rate": 8.555555555555556e-06,
+      "loss": 1.0866,
+      "step": 2760
+    },
+    {
+      "epoch": 0.5429243433947472,
+      "grad_norm": 6.960358619689941,
+      "learning_rate": 8.546296296296296e-06,
+      "loss": 1.0942,
+      "step": 2770
+    },
+    {
+      "epoch": 0.5448843590748726,
+      "grad_norm": 5.870516300201416,
+      "learning_rate": 8.537037037037038e-06,
+      "loss": 1.0937,
+      "step": 2780
+    },
+    {
+      "epoch": 0.546844374754998,
+      "grad_norm": 4.99727725982666,
+      "learning_rate": 8.527777777777779e-06,
+      "loss": 1.0851,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5488043904351235,
+      "grad_norm": 4.9571003913879395,
+      "learning_rate": 8.518518518518519e-06,
+      "loss": 1.085,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5488043904351235,
+      "eval_loss": 1.073391318321228,
+      "eval_runtime": 14.0283,
+      "eval_samples_per_second": 46.62,
+      "eval_steps_per_second": 5.845,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5507644061152489,
+      "grad_norm": 4.457803726196289,
+      "learning_rate": 8.509259259259261e-06,
+      "loss": 1.0787,
+      "step": 2810
+    },
+    {
+      "epoch": 0.5527244217953744,
+      "grad_norm": 5.098880290985107,
+      "learning_rate": 8.5e-06,
+      "loss": 1.0672,
+      "step": 2820
+    },
+    {
+      "epoch": 0.5546844374754998,
+      "grad_norm": 5.3498735427856445,
+      "learning_rate": 8.490740740740742e-06,
+      "loss": 1.0852,
+      "step": 2830
+    },
+    {
+      "epoch": 0.5566444531556253,
+      "grad_norm": 5.1486921310424805,
+      "learning_rate": 8.481481481481482e-06,
+      "loss": 1.0564,
+      "step": 2840
+    },
+    {
+      "epoch": 0.5586044688357507,
+      "grad_norm": 5.613958358764648,
+      "learning_rate": 8.472222222222223e-06,
+      "loss": 1.0864,
+      "step": 2850
+    },
+    {
+      "epoch": 0.5605644845158761,
+      "grad_norm": 5.687408924102783,
+      "learning_rate": 8.462962962962963e-06,
+      "loss": 1.0871,
+      "step": 2860
+    },
+    {
+      "epoch": 0.5625245001960015,
+      "grad_norm": 5.822880268096924,
+      "learning_rate": 8.453703703703704e-06,
+      "loss": 1.0708,
+      "step": 2870
+    },
+    {
+      "epoch": 0.564484515876127,
+      "grad_norm": 5.982561111450195,
+      "learning_rate": 8.444444444444446e-06,
+      "loss": 1.0633,
+      "step": 2880
+    },
+    {
+      "epoch": 0.5664445315562524,
+      "grad_norm": 4.887526035308838,
+      "learning_rate": 8.435185185185186e-06,
+      "loss": 1.0823,
+      "step": 2890
+    },
+    {
+      "epoch": 0.5684045472363779,
+      "grad_norm": 5.810245037078857,
+      "learning_rate": 8.425925925925926e-06,
+      "loss": 1.0683,
+      "step": 2900
+    },
+    {
+      "epoch": 0.5703645629165033,
+      "grad_norm": 4.757987022399902,
+      "learning_rate": 8.416666666666667e-06,
+      "loss": 1.0703,
+      "step": 2910
+    },
+    {
+      "epoch": 0.5723245785966288,
+      "grad_norm": 5.597621440887451,
+      "learning_rate": 8.407407407407409e-06,
+      "loss": 1.0861,
+      "step": 2920
+    },
+    {
+      "epoch": 0.5742845942767543,
+      "grad_norm": 9.786075592041016,
+      "learning_rate": 8.39814814814815e-06,
+      "loss": 1.0755,
+      "step": 2930
+    },
+    {
+      "epoch": 0.5762446099568796,
+      "grad_norm": 4.969895362854004,
+      "learning_rate": 8.38888888888889e-06,
+      "loss": 1.0907,
+      "step": 2940
+    },
+    {
+      "epoch": 0.5782046256370051,
+      "grad_norm": 5.833816051483154,
+      "learning_rate": 8.37962962962963e-06,
+      "loss": 1.0679,
+      "step": 2950
+    },
+    {
+      "epoch": 0.5801646413171305,
+      "grad_norm": 5.07546329498291,
+      "learning_rate": 8.37037037037037e-06,
+      "loss": 1.0705,
+      "step": 2960
+    },
+    {
+      "epoch": 0.582124656997256,
+      "grad_norm": 5.735626220703125,
+      "learning_rate": 8.361111111111113e-06,
+      "loss": 1.0676,
+      "step": 2970
+    },
+    {
+      "epoch": 0.5840846726773814,
+      "grad_norm": 4.732352256774902,
+      "learning_rate": 8.351851851851851e-06,
+      "loss": 1.0809,
+      "step": 2980
+    },
+    {
+      "epoch": 0.5860446883575069,
+      "grad_norm": 6.006665229797363,
+      "learning_rate": 8.342592592592593e-06,
+      "loss": 1.0624,
+      "step": 2990
+    },
+    {
+      "epoch": 0.5880047040376323,
+      "grad_norm": 6.649717330932617,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.072,
+      "step": 3000
+    },
+    {
+      "epoch": 0.5899647197177578,
+      "grad_norm": 5.156081199645996,
+      "learning_rate": 8.324074074074074e-06,
+      "loss": 1.0791,
+      "step": 3010
+    },
+    {
+      "epoch": 0.5919247353978832,
+      "grad_norm": 5.489958763122559,
+      "learning_rate": 8.314814814814816e-06,
+      "loss": 1.0803,
+      "step": 3020
+    },
+    {
+      "epoch": 0.5938847510780086,
+      "grad_norm": 5.912964344024658,
+      "learning_rate": 8.305555555555557e-06,
+      "loss": 1.0888,
+      "step": 3030
+    },
+    {
+      "epoch": 0.595844766758134,
+      "grad_norm": 5.113432884216309,
+      "learning_rate": 8.296296296296297e-06,
+      "loss": 1.0728,
+      "step": 3040
+    },
+    {
+      "epoch": 0.5978047824382595,
+      "grad_norm": 4.774835109710693,
+      "learning_rate": 8.287037037037037e-06,
+      "loss": 1.0772,
+      "step": 3050
+    },
+    {
+      "epoch": 0.5997647981183849,
+      "grad_norm": 5.954660892486572,
+      "learning_rate": 8.277777777777778e-06,
+      "loss": 1.0805,
+      "step": 3060
+    },
+    {
+      "epoch": 0.6017248137985104,
+      "grad_norm": 4.933722019195557,
+      "learning_rate": 8.26851851851852e-06,
+      "loss": 1.0678,
+      "step": 3070
+    },
+    {
+      "epoch": 0.6036848294786358,
+      "grad_norm": 5.152841567993164,
+      "learning_rate": 8.25925925925926e-06,
+      "loss": 1.0831,
+      "step": 3080
+    },
+    {
+      "epoch": 0.6056448451587613,
+      "grad_norm": 5.011537551879883,
+      "learning_rate": 8.25e-06,
+      "loss": 1.0841,
+      "step": 3090
+    },
+    {
+      "epoch": 0.6076048608388868,
+      "grad_norm": 4.196855545043945,
+      "learning_rate": 8.240740740740741e-06,
+      "loss": 1.0647,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6095648765190121,
+      "grad_norm": 5.21035099029541,
+      "learning_rate": 8.231481481481483e-06,
+      "loss": 1.049,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6115248921991376,
+      "grad_norm": 26.132904052734375,
+      "learning_rate": 8.222222222222222e-06,
+      "loss": 1.0799,
+      "step": 3120
+    },
+    {
+      "epoch": 0.613484907879263,
+      "grad_norm": 6.4696760177612305,
+      "learning_rate": 8.212962962962964e-06,
+      "loss": 1.0666,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6154449235593885,
+      "grad_norm": 4.687003135681152,
+      "learning_rate": 8.203703703703704e-06,
+      "loss": 1.0732,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6174049392395139,
+      "grad_norm": 5.188929557800293,
+      "learning_rate": 8.194444444444445e-06,
+      "loss": 1.0911,
+      "step": 3150
+    },
+    {
+      "epoch": 0.6193649549196394,
+      "grad_norm": 4.781430244445801,
+      "learning_rate": 8.185185185185187e-06,
+      "loss": 1.0705,
+      "step": 3160
+    },
+    {
+      "epoch": 0.6213249705997648,
+      "grad_norm": 4.717843055725098,
+      "learning_rate": 8.175925925925925e-06,
+      "loss": 1.0822,
+      "step": 3170
+    },
+    {
+      "epoch": 0.6232849862798903,
+      "grad_norm": 5.404903888702393,
+      "learning_rate": 8.166666666666668e-06,
+      "loss": 1.0917,
+      "step": 3180
+    },
+    {
+      "epoch": 0.6252450019600156,
+      "grad_norm": 5.684728145599365,
+      "learning_rate": 8.157407407407408e-06,
+      "loss": 1.0842,
+      "step": 3190
+    },
+    {
+      "epoch": 0.6272050176401411,
+      "grad_norm": 4.7895965576171875,
+      "learning_rate": 8.148148148148148e-06,
+      "loss": 1.0652,
+      "step": 3200
+    },
+    {
+      "epoch": 0.6291650333202665,
+      "grad_norm": 6.572015285491943,
+      "learning_rate": 8.138888888888889e-06,
+      "loss": 1.0959,
+      "step": 3210
+    },
+    {
+      "epoch": 0.631125049000392,
+      "grad_norm": 4.640923976898193,
+      "learning_rate": 8.12962962962963e-06,
+      "loss": 1.0896,
+      "step": 3220
+    },
+    {
+      "epoch": 0.6330850646805174,
+      "grad_norm": 4.732535362243652,
+      "learning_rate": 8.120370370370371e-06,
+      "loss": 1.0646,
+      "step": 3230
+    },
+    {
+      "epoch": 0.6350450803606429,
+      "grad_norm": 5.692999362945557,
+      "learning_rate": 8.111111111111112e-06,
+      "loss": 1.0774,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6370050960407684,
+      "grad_norm": 5.28448486328125,
+      "learning_rate": 8.101851851851854e-06,
+      "loss": 1.0844,
+      "step": 3250
+    },
+    {
+      "epoch": 0.6389651117208938,
+      "grad_norm": 3.9686877727508545,
+      "learning_rate": 8.092592592592592e-06,
+      "loss": 1.0626,
+      "step": 3260
+    },
+    {
+      "epoch": 0.6409251274010193,
+      "grad_norm": 6.364278793334961,
+      "learning_rate": 8.083333333333334e-06,
+      "loss": 1.0696,
+      "step": 3270
+    },
+    {
+      "epoch": 0.6428851430811446,
+      "grad_norm": 4.789191722869873,
+      "learning_rate": 8.074074074074075e-06,
+      "loss": 1.0696,
+      "step": 3280
+    },
+    {
+      "epoch": 0.6448451587612701,
+      "grad_norm": 4.61089563369751,
+      "learning_rate": 8.064814814814815e-06,
+      "loss": 1.0701,
+      "step": 3290
+    },
+    {
+      "epoch": 0.6468051744413955,
+      "grad_norm": 4.959714889526367,
+      "learning_rate": 8.055555555555557e-06,
+      "loss": 1.0852,
+      "step": 3300
+    },
+    {
+      "epoch": 0.648765190121521,
+      "grad_norm": 4.559225082397461,
+      "learning_rate": 8.046296296296296e-06,
+      "loss": 1.0708,
+      "step": 3310
+    },
+    {
+      "epoch": 0.6507252058016464,
+      "grad_norm": 4.9751691818237305,
+      "learning_rate": 8.037037037037038e-06,
+      "loss": 1.0545,
+      "step": 3320
+    },
+    {
+      "epoch": 0.6526852214817719,
+      "grad_norm": 5.153299808502197,
+      "learning_rate": 8.027777777777778e-06,
+      "loss": 1.074,
+      "step": 3330
+    },
+    {
+      "epoch": 0.6546452371618973,
+      "grad_norm": 5.507195472717285,
+      "learning_rate": 8.018518518518519e-06,
+      "loss": 1.0772,
+      "step": 3340
+    },
+    {
+      "epoch": 0.6566052528420228,
+      "grad_norm": 4.8530449867248535,
+      "learning_rate": 8.00925925925926e-06,
+      "loss": 1.087,
+      "step": 3350
+    },
+    {
+      "epoch": 0.6585652685221481,
+      "grad_norm": 4.754231929779053,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.0792,
+      "step": 3360
+    },
+    {
+      "epoch": 0.6605252842022736,
+      "grad_norm": 4.642824172973633,
+      "learning_rate": 7.990740740740742e-06,
+      "loss": 1.0681,
+      "step": 3370
+    },
+    {
+      "epoch": 0.662485299882399,
+      "grad_norm": 5.338252067565918,
+      "learning_rate": 7.981481481481482e-06,
+      "loss": 1.0579,
+      "step": 3380
+    },
+    {
+      "epoch": 0.6644453155625245,
+      "grad_norm": 5.290416240692139,
+      "learning_rate": 7.972222222222224e-06,
+      "loss": 1.0523,
+      "step": 3390
+    },
+    {
+      "epoch": 0.6664053312426499,
+      "grad_norm": 4.5010294914245605,
+      "learning_rate": 7.962962962962963e-06,
+      "loss": 1.0694,
+      "step": 3400
+    },
+    {
+      "epoch": 0.6683653469227754,
+      "grad_norm": 5.253542900085449,
+      "learning_rate": 7.953703703703705e-06,
+      "loss": 1.0745,
+      "step": 3410
+    },
+    {
+      "epoch": 0.6703253626029009,
+      "grad_norm": 4.743661880493164,
+      "learning_rate": 7.944444444444445e-06,
+      "loss": 1.0586,
+      "step": 3420
+    },
+    {
+      "epoch": 0.6722853782830263,
+      "grad_norm": 5.1141839027404785,
+      "learning_rate": 7.935185185185186e-06,
+      "loss": 1.0767,
+      "step": 3430
+    },
+    {
+      "epoch": 0.6742453939631518,
+      "grad_norm": 4.729719638824463,
+      "learning_rate": 7.925925925925926e-06,
+      "loss": 1.0841,
+      "step": 3440
+    },
+    {
+      "epoch": 0.6762054096432771,
+      "grad_norm": 5.445465087890625,
+      "learning_rate": 7.916666666666667e-06,
+      "loss": 1.0901,
+      "step": 3450
+    },
+    {
+      "epoch": 0.6781654253234026,
+      "grad_norm": 4.362582683563232,
+      "learning_rate": 7.907407407407409e-06,
+      "loss": 1.0634,
+      "step": 3460
+    },
+    {
+      "epoch": 0.680125441003528,
+      "grad_norm": 4.604685306549072,
+      "learning_rate": 7.898148148148149e-06,
+      "loss": 1.0631,
+      "step": 3470
+    },
+    {
+      "epoch": 0.6820854566836535,
+      "grad_norm": 5.426617622375488,
+      "learning_rate": 7.88888888888889e-06,
+      "loss": 1.0691,
+      "step": 3480
+    },
+    {
+      "epoch": 0.6840454723637789,
+      "grad_norm": 4.578293323516846,
+      "learning_rate": 7.87962962962963e-06,
+      "loss": 1.0761,
+      "step": 3490
+    },
+    {
+      "epoch": 0.6860054880439044,
+      "grad_norm": 5.362606525421143,
+      "learning_rate": 7.870370370370372e-06,
+      "loss": 1.0836,
+      "step": 3500
+    },
+    {
+      "epoch": 0.6860054880439044,
+      "eval_loss": 1.0641947984695435,
+      "eval_runtime": 14.0879,
+      "eval_samples_per_second": 46.423,
+      "eval_steps_per_second": 5.821,
+      "step": 3500
+    },
+    {
+      "epoch": 0.6879655037240298,
+      "grad_norm": 4.338893890380859,
+      "learning_rate": 7.861111111111112e-06,
+      "loss": 1.0604,
+      "step": 3510
+    },
+    {
+      "epoch": 0.6899255194041553,
+      "grad_norm": 5.606161594390869,
+      "learning_rate": 7.851851851851853e-06,
+      "loss": 1.0869,
+      "step": 3520
+    },
+    {
+      "epoch": 0.6918855350842806,
+      "grad_norm": 4.934096336364746,
+      "learning_rate": 7.842592592592593e-06,
+      "loss": 1.07,
+      "step": 3530
+    },
+    {
+      "epoch": 0.6938455507644061,
+      "grad_norm": 4.351364612579346,
+      "learning_rate": 7.833333333333333e-06,
+      "loss": 1.0682,
+      "step": 3540
+    },
+    {
+      "epoch": 0.6958055664445315,
+      "grad_norm": 4.8932976722717285,
+      "learning_rate": 7.824074074074076e-06,
+      "loss": 1.0766,
+      "step": 3550
+    },
+    {
+      "epoch": 0.697765582124657,
+      "grad_norm": 4.501565456390381,
+      "learning_rate": 7.814814814814816e-06,
+      "loss": 1.0612,
+      "step": 3560
+    },
+    {
+      "epoch": 0.6997255978047824,
+      "grad_norm": 4.929937362670898,
+      "learning_rate": 7.805555555555556e-06,
+      "loss": 1.0609,
+      "step": 3570
+    },
+    {
+      "epoch": 0.7016856134849079,
+      "grad_norm": 5.501171112060547,
+      "learning_rate": 7.796296296296297e-06,
+      "loss": 1.0583,
+      "step": 3580
+    },
+    {
+      "epoch": 0.7036456291650334,
+      "grad_norm": 4.9919891357421875,
+      "learning_rate": 7.787037037037037e-06,
+      "loss": 1.0603,
+      "step": 3590
+    },
+    {
+      "epoch": 0.7056056448451588,
+      "grad_norm": 4.892634868621826,
+      "learning_rate": 7.77777777777778e-06,
+      "loss": 1.0718,
+      "step": 3600
+    },
+    {
+      "epoch": 0.7075656605252842,
+      "grad_norm": 5.0666632652282715,
+      "learning_rate": 7.76851851851852e-06,
+      "loss": 1.0733,
+      "step": 3610
+    },
+    {
+      "epoch": 0.7095256762054096,
+      "grad_norm": 4.797816753387451,
+      "learning_rate": 7.75925925925926e-06,
+      "loss": 1.0664,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7114856918855351,
+      "grad_norm": 5.004721164703369,
+      "learning_rate": 7.75e-06,
+      "loss": 1.0847,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7134457075656605,
+      "grad_norm": 4.80883264541626,
+      "learning_rate": 7.74074074074074e-06,
+      "loss": 1.075,
+      "step": 3640
+    },
+    {
+      "epoch": 0.715405723245786,
+      "grad_norm": 5.439119815826416,
+      "learning_rate": 7.731481481481483e-06,
+      "loss": 1.0517,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7173657389259114,
+      "grad_norm": 4.244631290435791,
+      "learning_rate": 7.722222222222223e-06,
+      "loss": 1.0836,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7193257546060369,
+      "grad_norm": 4.710737705230713,
+      "learning_rate": 7.712962962962964e-06,
+      "loss": 1.0748,
+      "step": 3670
+    },
+    {
+      "epoch": 0.7212857702861623,
+      "grad_norm": 5.23133659362793,
+      "learning_rate": 7.703703703703704e-06,
+      "loss": 1.0843,
+      "step": 3680
+    },
+    {
+      "epoch": 0.7232457859662877,
+      "grad_norm": 4.706710338592529,
+      "learning_rate": 7.694444444444446e-06,
+      "loss": 1.0637,
+      "step": 3690
+    },
+    {
+      "epoch": 0.7252058016464131,
+      "grad_norm": 4.4408955574035645,
+      "learning_rate": 7.685185185185185e-06,
+      "loss": 1.0945,
+      "step": 3700
+    },
+    {
+      "epoch": 0.7271658173265386,
+      "grad_norm": 5.1011199951171875,
+      "learning_rate": 7.675925925925927e-06,
+      "loss": 1.0573,
+      "step": 3710
+    },
+    {
+      "epoch": 0.729125833006664,
+      "grad_norm": 4.85494327545166,
+      "learning_rate": 7.666666666666667e-06,
+      "loss": 1.0686,
+      "step": 3720
+    },
+    {
+      "epoch": 0.7310858486867895,
+      "grad_norm": 4.990062236785889,
+      "learning_rate": 7.657407407407408e-06,
+      "loss": 1.052,
+      "step": 3730
+    },
+    {
+      "epoch": 0.733045864366915,
+      "grad_norm": 4.779956817626953,
+      "learning_rate": 7.64814814814815e-06,
+      "loss": 1.0619,
+      "step": 3740
+    },
+    {
+      "epoch": 0.7350058800470404,
+      "grad_norm": 4.921076774597168,
+      "learning_rate": 7.638888888888888e-06,
+      "loss": 1.0834,
+      "step": 3750
+    },
+    {
+      "epoch": 0.7369658957271659,
+      "grad_norm": 4.729534149169922,
+      "learning_rate": 7.62962962962963e-06,
+      "loss": 1.0643,
+      "step": 3760
+    },
+    {
+      "epoch": 0.7389259114072912,
+      "grad_norm": 4.321104049682617,
+      "learning_rate": 7.620370370370372e-06,
+      "loss": 1.0736,
+      "step": 3770
+    },
+    {
+      "epoch": 0.7408859270874167,
+      "grad_norm": 5.252387523651123,
+      "learning_rate": 7.611111111111111e-06,
+      "loss": 1.0515,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7428459427675421,
+      "grad_norm": 4.770890712738037,
+      "learning_rate": 7.6018518518518525e-06,
+      "loss": 1.0865,
+      "step": 3790
+    },
+    {
+      "epoch": 0.7448059584476676,
+      "grad_norm": 4.786081790924072,
+      "learning_rate": 7.592592592592594e-06,
+      "loss": 1.0727,
+      "step": 3800
+    },
+    {
+      "epoch": 0.746765974127793,
+      "grad_norm": 5.215248107910156,
+      "learning_rate": 7.583333333333333e-06,
+      "loss": 1.0657,
+      "step": 3810
+    },
+    {
+      "epoch": 0.7487259898079185,
+      "grad_norm": 8.386008262634277,
+      "learning_rate": 7.5740740740740745e-06,
+      "loss": 1.0774,
+      "step": 3820
+    },
+    {
+      "epoch": 0.7506860054880439,
+      "grad_norm": 5.009541034698486,
+      "learning_rate": 7.564814814814816e-06,
+      "loss": 1.0517,
+      "step": 3830
+    },
+    {
+      "epoch": 0.7526460211681694,
+      "grad_norm": 5.512828826904297,
+      "learning_rate": 7.555555555555556e-06,
+      "loss": 1.0661,
+      "step": 3840
+    },
+    {
+      "epoch": 0.7546060368482947,
+      "grad_norm": 4.983898162841797,
+      "learning_rate": 7.546296296296297e-06,
+      "loss": 1.0561,
+      "step": 3850
+    },
+    {
+      "epoch": 0.7565660525284202,
+      "grad_norm": 4.687797546386719,
+      "learning_rate": 7.537037037037037e-06,
+      "loss": 1.0623,
+      "step": 3860
+    },
+    {
+      "epoch": 0.7585260682085456,
+      "grad_norm": 4.924409866333008,
+      "learning_rate": 7.527777777777778e-06,
+      "loss": 1.0557,
+      "step": 3870
+    },
+    {
+      "epoch": 0.7604860838886711,
+      "grad_norm": 4.704287528991699,
+      "learning_rate": 7.518518518518519e-06,
+      "loss": 1.0732,
+      "step": 3880
+    },
+    {
+      "epoch": 0.7624460995687965,
+      "grad_norm": 5.610473155975342,
+      "learning_rate": 7.50925925925926e-06,
+      "loss": 1.0827,
+      "step": 3890
+    },
+    {
+      "epoch": 0.764406115248922,
+      "grad_norm": 5.7728590965271,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.0738,
+      "step": 3900
+    },
+    {
+      "epoch": 0.7663661309290475,
+      "grad_norm": 4.7245659828186035,
+      "learning_rate": 7.4907407407407414e-06,
+      "loss": 1.0665,
+      "step": 3910
+    },
+    {
+      "epoch": 0.7683261466091729,
+      "grad_norm": 6.519947528839111,
+      "learning_rate": 7.481481481481482e-06,
+      "loss": 1.0631,
+      "step": 3920
+    },
+    {
+      "epoch": 0.7702861622892984,
+      "grad_norm": 4.989842891693115,
+      "learning_rate": 7.472222222222223e-06,
+      "loss": 1.053,
+      "step": 3930
+    },
+    {
+      "epoch": 0.7722461779694237,
+      "grad_norm": 4.687224864959717,
+      "learning_rate": 7.462962962962964e-06,
+      "loss": 1.0593,
+      "step": 3940
+    },
+    {
+      "epoch": 0.7742061936495492,
+      "grad_norm": 4.693930625915527,
+      "learning_rate": 7.453703703703704e-06,
+      "loss": 1.0665,
+      "step": 3950
+    },
+    {
+      "epoch": 0.7761662093296746,
+      "grad_norm": 5.265392303466797,
+      "learning_rate": 7.444444444444445e-06,
+      "loss": 1.042,
+      "step": 3960
+    },
+    {
+      "epoch": 0.7781262250098001,
+      "grad_norm": 5.122781753540039,
+      "learning_rate": 7.4351851851851855e-06,
+      "loss": 1.0748,
+      "step": 3970
+    },
+    {
+      "epoch": 0.7800862406899255,
+      "grad_norm": 4.934235095977783,
+      "learning_rate": 7.425925925925927e-06,
+      "loss": 1.0603,
+      "step": 3980
+    },
+    {
+      "epoch": 0.782046256370051,
+      "grad_norm": 4.4583587646484375,
+      "learning_rate": 7.416666666666668e-06,
+      "loss": 1.0724,
+      "step": 3990
+    },
+    {
+      "epoch": 0.7840062720501764,
+      "grad_norm": 5.3386030197143555,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 1.0651,
+      "step": 4000
+    },
+    {
+      "epoch": 0.7859662877303019,
+      "grad_norm": 4.659664154052734,
+      "learning_rate": 7.398148148148149e-06,
+      "loss": 1.0525,
+      "step": 4010
+    },
+    {
+      "epoch": 0.7879263034104272,
+      "grad_norm": 4.794917106628418,
+      "learning_rate": 7.38888888888889e-06,
+      "loss": 1.0561,
+      "step": 4020
+    },
+    {
+      "epoch": 0.7898863190905527,
+      "grad_norm": 5.210708141326904,
+      "learning_rate": 7.3796296296296295e-06,
+      "loss": 1.0641,
+      "step": 4030
+    },
+    {
+      "epoch": 0.7918463347706781,
+      "grad_norm": 4.441596031188965,
+      "learning_rate": 7.370370370370371e-06,
+      "loss": 1.0696,
+      "step": 4040
+    },
+    {
+      "epoch": 0.7938063504508036,
+      "grad_norm": 5.239253044128418,
+      "learning_rate": 7.361111111111112e-06,
+      "loss": 1.0576,
+      "step": 4050
+    },
+    {
+      "epoch": 0.795766366130929,
+      "grad_norm": 5.195384502410889,
+      "learning_rate": 7.351851851851852e-06,
+      "loss": 1.0556,
+      "step": 4060
+    },
+    {
+      "epoch": 0.7977263818110545,
+      "grad_norm": 5.0446648597717285,
+      "learning_rate": 7.342592592592594e-06,
+      "loss": 1.067,
+      "step": 4070
+    },
+    {
+      "epoch": 0.79968639749118,
+      "grad_norm": 5.061046600341797,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 1.0587,
+      "step": 4080
+    },
+    {
+      "epoch": 0.8016464131713054,
+      "grad_norm": 7.661867141723633,
+      "learning_rate": 7.324074074074074e-06,
+      "loss": 1.0694,
+      "step": 4090
+    },
+    {
+      "epoch": 0.8036064288514309,
+      "grad_norm": 4.821944236755371,
+      "learning_rate": 7.314814814814816e-06,
+      "loss": 1.0675,
+      "step": 4100
+    },
+    {
+      "epoch": 0.8055664445315562,
+      "grad_norm": 4.810634613037109,
+      "learning_rate": 7.305555555555556e-06,
+      "loss": 1.0589,
+      "step": 4110
+    },
+    {
+      "epoch": 0.8075264602116817,
+      "grad_norm": 5.0643157958984375,
+      "learning_rate": 7.296296296296297e-06,
+      "loss": 1.0714,
+      "step": 4120
+    },
+    {
+      "epoch": 0.8094864758918071,
+      "grad_norm": 4.812649250030518,
+      "learning_rate": 7.287037037037038e-06,
+      "loss": 1.0635,
+      "step": 4130
+    },
+    {
+      "epoch": 0.8114464915719326,
+      "grad_norm": 4.7638840675354,
+      "learning_rate": 7.277777777777778e-06,
+      "loss": 1.064,
+      "step": 4140
+    },
+    {
+      "epoch": 0.813406507252058,
+      "grad_norm": 4.386782646179199,
+      "learning_rate": 7.268518518518519e-06,
+      "loss": 1.0726,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8153665229321835,
+      "grad_norm": 4.593876838684082,
+      "learning_rate": 7.2592592592592605e-06,
+      "loss": 1.0542,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8173265386123089,
+      "grad_norm": 4.822241306304932,
+      "learning_rate": 7.25e-06,
+      "loss": 1.0541,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8192865542924344,
+      "grad_norm": 4.806110858917236,
+      "learning_rate": 7.240740740740741e-06,
+      "loss": 1.0522,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8212465699725597,
+      "grad_norm": 4.5700788497924805,
+      "learning_rate": 7.231481481481482e-06,
+      "loss": 1.0672,
+      "step": 4190
+    },
+    {
+      "epoch": 0.8232065856526852,
+      "grad_norm": 4.629611492156982,
+      "learning_rate": 7.222222222222223e-06,
+      "loss": 1.072,
+      "step": 4200
+    },
+    {
+      "epoch": 0.8232065856526852,
+      "eval_loss": 1.0551141500473022,
+      "eval_runtime": 13.9585,
+      "eval_samples_per_second": 46.853,
+      "eval_steps_per_second": 5.875,
+      "step": 4200
+    },
+    {
+      "epoch": 0.8251666013328106,
+      "grad_norm": 5.06083345413208,
+      "learning_rate": 7.212962962962964e-06,
+      "loss": 1.0641,
+      "step": 4210
+    },
+    {
+      "epoch": 0.8271266170129361,
+      "grad_norm": 4.53200626373291,
+      "learning_rate": 7.203703703703704e-06,
+      "loss": 1.0764,
+      "step": 4220
+    },
+    {
+      "epoch": 0.8290866326930616,
+      "grad_norm": 4.572458744049072,
+      "learning_rate": 7.194444444444445e-06,
+      "loss": 1.0616,
+      "step": 4230
+    },
+    {
+      "epoch": 0.831046648373187,
+      "grad_norm": 21.508056640625,
+      "learning_rate": 7.185185185185186e-06,
+      "loss": 1.0606,
+      "step": 4240
+    },
+    {
+      "epoch": 0.8330066640533125,
+      "grad_norm": 5.574312210083008,
+      "learning_rate": 7.1759259259259266e-06,
+      "loss": 1.0736,
+      "step": 4250
+    },
+    {
+      "epoch": 0.8349666797334379,
+      "grad_norm": 4.778395175933838,
+      "learning_rate": 7.166666666666667e-06,
+      "loss": 1.0561,
+      "step": 4260
+    },
+    {
+      "epoch": 0.8369266954135633,
+      "grad_norm": 4.696834564208984,
+      "learning_rate": 7.157407407407408e-06,
+      "loss": 1.0753,
+      "step": 4270
+    },
+    {
+      "epoch": 0.8388867110936887,
+      "grad_norm": 4.8193769454956055,
+      "learning_rate": 7.1481481481481486e-06,
+      "loss": 1.0454,
+      "step": 4280
+    },
+    {
+      "epoch": 0.8408467267738142,
+      "grad_norm": 4.693731784820557,
+      "learning_rate": 7.13888888888889e-06,
+      "loss": 1.0634,
+      "step": 4290
+    },
+    {
+      "epoch": 0.8428067424539396,
+      "grad_norm": 5.045989990234375,
+      "learning_rate": 7.129629629629629e-06,
+      "loss": 1.057,
+      "step": 4300
+    },
+    {
+      "epoch": 0.8447667581340651,
+      "grad_norm": 5.245114326477051,
+      "learning_rate": 7.120370370370371e-06,
+      "loss": 1.0435,
+      "step": 4310
+    },
+    {
+      "epoch": 0.8467267738141905,
+      "grad_norm": 5.094894886016846,
+      "learning_rate": 7.111111111111112e-06,
+      "loss": 1.0657,
+      "step": 4320
+    },
+    {
+      "epoch": 0.848686789494316,
+      "grad_norm": 4.630204200744629,
+      "learning_rate": 7.101851851851852e-06,
+      "loss": 1.0729,
+      "step": 4330
+    },
+    {
+      "epoch": 0.8506468051744414,
+      "grad_norm": 4.463089942932129,
+      "learning_rate": 7.0925925925925935e-06,
+      "loss": 1.0573,
+      "step": 4340
+    },
+    {
+      "epoch": 0.8526068208545668,
+      "grad_norm": 4.479245185852051,
+      "learning_rate": 7.083333333333335e-06,
+      "loss": 1.0534,
+      "step": 4350
+    },
+    {
+      "epoch": 0.8545668365346922,
+      "grad_norm": 4.832218170166016,
+      "learning_rate": 7.074074074074074e-06,
+      "loss": 1.0499,
+      "step": 4360
+    },
+    {
+      "epoch": 0.8565268522148177,
+      "grad_norm": 4.618699550628662,
+      "learning_rate": 7.0648148148148155e-06,
+      "loss": 1.073,
+      "step": 4370
+    },
+    {
+      "epoch": 0.8584868678949431,
+      "grad_norm": 4.646695137023926,
+      "learning_rate": 7.055555555555557e-06,
+      "loss": 1.0662,
+      "step": 4380
+    },
+    {
+      "epoch": 0.8604468835750686,
+      "grad_norm": 4.74224853515625,
+      "learning_rate": 7.046296296296296e-06,
+      "loss": 1.0516,
+      "step": 4390
+    },
+    {
+      "epoch": 0.8624068992551941,
+      "grad_norm": 4.836453914642334,
+      "learning_rate": 7.0370370370370375e-06,
+      "loss": 1.0622,
+      "step": 4400
+    },
+    {
+      "epoch": 0.8643669149353195,
+      "grad_norm": 5.0554046630859375,
+      "learning_rate": 7.027777777777778e-06,
+      "loss": 1.0725,
+      "step": 4410
+    },
+    {
+      "epoch": 0.866326930615445,
+      "grad_norm": 4.8123579025268555,
+      "learning_rate": 7.018518518518519e-06,
+      "loss": 1.0451,
+      "step": 4420
+    },
+    {
+      "epoch": 0.8682869462955703,
+      "grad_norm": 4.815413951873779,
+      "learning_rate": 7.00925925925926e-06,
+      "loss": 1.057,
+      "step": 4430
+    },
+    {
+      "epoch": 0.8702469619756958,
+      "grad_norm": 5.103263854980469,
+      "learning_rate": 7e-06,
+      "loss": 1.073,
+      "step": 4440
+    },
+    {
+      "epoch": 0.8722069776558212,
+      "grad_norm": 5.115508556365967,
+      "learning_rate": 6.990740740740741e-06,
+      "loss": 1.058,
+      "step": 4450
+    },
+    {
+      "epoch": 0.8741669933359467,
+      "grad_norm": 4.976015567779541,
+      "learning_rate": 6.981481481481482e-06,
+      "loss": 1.0601,
+      "step": 4460
+    },
+    {
+      "epoch": 0.8761270090160721,
+      "grad_norm": 5.111723899841309,
+      "learning_rate": 6.972222222222223e-06,
+      "loss": 1.0735,
+      "step": 4470
+    },
+    {
+      "epoch": 0.8780870246961976,
+      "grad_norm": 4.711007118225098,
+      "learning_rate": 6.962962962962964e-06,
+      "loss": 1.0723,
+      "step": 4480
+    },
+    {
+      "epoch": 0.880047040376323,
+      "grad_norm": 4.8820881843566895,
+      "learning_rate": 6.953703703703704e-06,
+      "loss": 1.0562,
+      "step": 4490
+    },
+    {
+      "epoch": 0.8820070560564485,
+      "grad_norm": 5.674580097198486,
+      "learning_rate": 6.944444444444445e-06,
+      "loss": 1.0657,
+      "step": 4500
+    },
+    {
+      "epoch": 0.8839670717365739,
+      "grad_norm": 4.517936706542969,
+      "learning_rate": 6.935185185185186e-06,
+      "loss": 1.0604,
+      "step": 4510
+    },
+    {
+      "epoch": 0.8859270874166993,
+      "grad_norm": 5.491835594177246,
+      "learning_rate": 6.9259259259259256e-06,
+      "loss": 1.0635,
+      "step": 4520
+    },
+    {
+      "epoch": 0.8878871030968247,
+      "grad_norm": 5.377597808837891,
+      "learning_rate": 6.916666666666667e-06,
+      "loss": 1.0567,
+      "step": 4530
+    },
+    {
+      "epoch": 0.8898471187769502,
+      "grad_norm": 5.251382350921631,
+      "learning_rate": 6.907407407407408e-06,
+      "loss": 1.0564,
+      "step": 4540
+    },
+    {
+      "epoch": 0.8918071344570757,
+      "grad_norm": 4.623981952667236,
+      "learning_rate": 6.898148148148148e-06,
+      "loss": 1.0663,
+      "step": 4550
+    },
+    {
+      "epoch": 0.8937671501372011,
+      "grad_norm": 4.460480690002441,
+      "learning_rate": 6.88888888888889e-06,
+      "loss": 1.0773,
+      "step": 4560
+    },
+    {
+      "epoch": 0.8957271658173266,
+      "grad_norm": 4.909152984619141,
+      "learning_rate": 6.879629629629631e-06,
+      "loss": 1.0462,
+      "step": 4570
+    },
+    {
+      "epoch": 0.897687181497452,
+      "grad_norm": 4.5406951904296875,
+      "learning_rate": 6.8703703703703704e-06,
+      "loss": 1.0619,
+      "step": 4580
+    },
+    {
+      "epoch": 0.8996471971775775,
+      "grad_norm": 4.8268890380859375,
+      "learning_rate": 6.861111111111112e-06,
+      "loss": 1.0749,
+      "step": 4590
+    },
+    {
+      "epoch": 0.9016072128577028,
+      "grad_norm": 5.024204254150391,
+      "learning_rate": 6.851851851851853e-06,
+      "loss": 1.082,
+      "step": 4600
+    },
+    {
+      "epoch": 0.9035672285378283,
+      "grad_norm": 4.6898908615112305,
+      "learning_rate": 6.842592592592593e-06,
+      "loss": 1.0527,
+      "step": 4610
+    },
+    {
+      "epoch": 0.9055272442179537,
+      "grad_norm": 5.445582866668701,
+      "learning_rate": 6.833333333333334e-06,
+      "loss": 1.05,
+      "step": 4620
+    },
+    {
+      "epoch": 0.9074872598980792,
+      "grad_norm": 4.625185966491699,
+      "learning_rate": 6.824074074074075e-06,
+      "loss": 1.0623,
+      "step": 4630
+    },
+    {
+      "epoch": 0.9094472755782046,
+      "grad_norm": 4.991087913513184,
+      "learning_rate": 6.814814814814815e-06,
+      "loss": 1.048,
+      "step": 4640
+    },
+    {
+      "epoch": 0.9114072912583301,
+      "grad_norm": 5.890523433685303,
+      "learning_rate": 6.8055555555555566e-06,
+      "loss": 1.0539,
+      "step": 4650
+    },
+    {
+      "epoch": 0.9133673069384555,
+      "grad_norm": 4.583831787109375,
+      "learning_rate": 6.796296296296296e-06,
+      "loss": 1.0616,
+      "step": 4660
+    },
+    {
+      "epoch": 0.915327322618581,
+      "grad_norm": 5.098804473876953,
+      "learning_rate": 6.787037037037037e-06,
+      "loss": 1.0555,
+      "step": 4670
+    },
+    {
+      "epoch": 0.9172873382987063,
+      "grad_norm": 4.51339864730835,
+      "learning_rate": 6.777777777777779e-06,
+      "loss": 1.0563,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9192473539788318,
+      "grad_norm": 5.437107086181641,
+      "learning_rate": 6.768518518518519e-06,
+      "loss": 1.0639,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9212073696589572,
+      "grad_norm": 5.220668315887451,
+      "learning_rate": 6.75925925925926e-06,
+      "loss": 1.0335,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9231673853390827,
+      "grad_norm": 5.214885711669922,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 1.0616,
+      "step": 4710
+    },
+    {
+      "epoch": 0.9251274010192082,
+      "grad_norm": 4.807358264923096,
+      "learning_rate": 6.740740740740741e-06,
+      "loss": 1.0556,
+      "step": 4720
+    },
+    {
+      "epoch": 0.9270874166993336,
+      "grad_norm": 4.536919116973877,
+      "learning_rate": 6.731481481481482e-06,
+      "loss": 1.0427,
+      "step": 4730
+    },
+    {
+      "epoch": 0.9290474323794591,
+      "grad_norm": 4.960407733917236,
+      "learning_rate": 6.7222222222222235e-06,
+      "loss": 1.0573,
+      "step": 4740
+    },
+    {
+      "epoch": 0.9310074480595845,
+      "grad_norm": 5.058611869812012,
+      "learning_rate": 6.712962962962963e-06,
+      "loss": 1.0529,
+      "step": 4750
+    },
+    {
+      "epoch": 0.93296746373971,
+      "grad_norm": 4.611677169799805,
+      "learning_rate": 6.703703703703704e-06,
+      "loss": 1.0609,
+      "step": 4760
+    },
+    {
+      "epoch": 0.9349274794198353,
+      "grad_norm": 4.662081241607666,
+      "learning_rate": 6.694444444444445e-06,
+      "loss": 1.0585,
+      "step": 4770
+    },
+    {
+      "epoch": 0.9368874950999608,
+      "grad_norm": 4.845633029937744,
+      "learning_rate": 6.685185185185186e-06,
+      "loss": 1.0591,
+      "step": 4780
+    },
+    {
+      "epoch": 0.9388475107800862,
+      "grad_norm": 4.63094425201416,
+      "learning_rate": 6.675925925925927e-06,
+      "loss": 1.051,
+      "step": 4790
+    },
+    {
+      "epoch": 0.9408075264602117,
+      "grad_norm": 5.272851467132568,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.0643,
+      "step": 4800
+    },
+    {
+      "epoch": 0.9427675421403371,
+      "grad_norm": 5.056580066680908,
+      "learning_rate": 6.657407407407408e-06,
+      "loss": 1.0483,
+      "step": 4810
+    },
+    {
+      "epoch": 0.9447275578204626,
+      "grad_norm": 4.985904693603516,
+      "learning_rate": 6.648148148148149e-06,
+      "loss": 1.0526,
+      "step": 4820
+    },
+    {
+      "epoch": 0.946687573500588,
+      "grad_norm": 5.002072334289551,
+      "learning_rate": 6.6388888888888895e-06,
+      "loss": 1.0693,
+      "step": 4830
+    },
+    {
+      "epoch": 0.9486475891807135,
+      "grad_norm": 5.099184036254883,
+      "learning_rate": 6.62962962962963e-06,
+      "loss": 1.0629,
+      "step": 4840
+    },
+    {
+      "epoch": 0.9506076048608388,
+      "grad_norm": 4.710622310638428,
+      "learning_rate": 6.620370370370371e-06,
+      "loss": 1.0458,
+      "step": 4850
+    },
+    {
+      "epoch": 0.9525676205409643,
+      "grad_norm": 5.143291473388672,
+      "learning_rate": 6.6111111111111115e-06,
+      "loss": 1.0552,
+      "step": 4860
+    },
+    {
+      "epoch": 0.9545276362210897,
+      "grad_norm": 5.28003454208374,
+      "learning_rate": 6.601851851851853e-06,
+      "loss": 1.0643,
+      "step": 4870
+    },
+    {
+      "epoch": 0.9564876519012152,
+      "grad_norm": 5.275862216949463,
+      "learning_rate": 6.592592592592592e-06,
+      "loss": 1.0566,
+      "step": 4880
+    },
+    {
+      "epoch": 0.9584476675813407,
+      "grad_norm": 4.868998050689697,
+      "learning_rate": 6.5833333333333335e-06,
+      "loss": 1.0703,
+      "step": 4890
+    },
+    {
+      "epoch": 0.9604076832614661,
+      "grad_norm": 4.964428424835205,
+      "learning_rate": 6.574074074074075e-06,
+      "loss": 1.0564,
+      "step": 4900
+    },
+    {
+      "epoch": 0.9604076832614661,
+      "eval_loss": 1.0497612953186035,
+      "eval_runtime": 14.0356,
+      "eval_samples_per_second": 46.596,
+      "eval_steps_per_second": 5.842,
+      "step": 4900
+    },
+    {
+      "epoch": 0.9623676989415916,
+      "grad_norm": 4.545341968536377,
+      "learning_rate": 6.564814814814815e-06,
+      "loss": 1.0741,
+      "step": 4910
+    },
+    {
+      "epoch": 0.964327714621717,
+      "grad_norm": 5.679270267486572,
+      "learning_rate": 6.555555555555556e-06,
+      "loss": 1.0528,
+      "step": 4920
+    },
+    {
+      "epoch": 0.9662877303018425,
+      "grad_norm": 4.5773749351501465,
+      "learning_rate": 6.546296296296298e-06,
+      "loss": 1.048,
+      "step": 4930
+    },
+    {
+      "epoch": 0.9682477459819678,
+      "grad_norm": 5.288699150085449,
+      "learning_rate": 6.537037037037037e-06,
+      "loss": 1.0487,
+      "step": 4940
+    },
+    {
+      "epoch": 0.9702077616620933,
+      "grad_norm": 5.2343363761901855,
+      "learning_rate": 6.5277777777777784e-06,
+      "loss": 1.0523,
+      "step": 4950
+    },
+    {
+      "epoch": 0.9721677773422187,
+      "grad_norm": 4.845068454742432,
+      "learning_rate": 6.51851851851852e-06,
+      "loss": 1.0519,
+      "step": 4960
+    },
+    {
+      "epoch": 0.9741277930223442,
+      "grad_norm": 5.101285457611084,
+      "learning_rate": 6.509259259259259e-06,
+      "loss": 1.0618,
+      "step": 4970
+    },
+    {
+      "epoch": 0.9760878087024696,
+      "grad_norm": 4.684749126434326,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 1.0506,
+      "step": 4980
+    },
+    {
+      "epoch": 0.9780478243825951,
+      "grad_norm": 4.986086368560791,
+      "learning_rate": 6.490740740740741e-06,
+      "loss": 1.0621,
+      "step": 4990
+    },
+    {
+      "epoch": 0.9800078400627205,
+      "grad_norm": 4.658474922180176,
+      "learning_rate": 6.481481481481482e-06,
+      "loss": 1.0727,
+      "step": 5000
+    },
+    {
+      "epoch": 0.981967855742846,
+      "grad_norm": 5.1348419189453125,
+      "learning_rate": 6.472222222222223e-06,
+      "loss": 1.0542,
+      "step": 5010
+    },
+    {
+      "epoch": 0.9839278714229713,
+      "grad_norm": 5.368491172790527,
+      "learning_rate": 6.462962962962963e-06,
+      "loss": 1.0656,
+      "step": 5020
+    },
+    {
+      "epoch": 0.9858878871030968,
+      "grad_norm": 5.312760829925537,
+      "learning_rate": 6.453703703703704e-06,
+      "loss": 1.0587,
+      "step": 5030
+    },
+    {
+      "epoch": 0.9878479027832223,
+      "grad_norm": 5.235229969024658,
+      "learning_rate": 6.444444444444445e-06,
+      "loss": 1.0588,
+      "step": 5040
+    },
+    {
+      "epoch": 0.9898079184633477,
+      "grad_norm": 4.761176586151123,
+      "learning_rate": 6.435185185185186e-06,
+      "loss": 1.0507,
+      "step": 5050
+    },
+    {
+      "epoch": 0.9917679341434732,
+      "grad_norm": 5.440311431884766,
+      "learning_rate": 6.425925925925927e-06,
+      "loss": 1.0566,
+      "step": 5060
+    },
+    {
+      "epoch": 0.9937279498235986,
+      "grad_norm": 4.793478488922119,
+      "learning_rate": 6.416666666666667e-06,
+      "loss": 1.0607,
+      "step": 5070
+    },
+    {
+      "epoch": 0.9956879655037241,
+      "grad_norm": 5.079543113708496,
+      "learning_rate": 6.407407407407408e-06,
+      "loss": 1.0494,
+      "step": 5080
+    },
+    {
+      "epoch": 0.9976479811838495,
+      "grad_norm": 4.377906322479248,
+      "learning_rate": 6.398148148148149e-06,
+      "loss": 1.0558,
+      "step": 5090
+    },
+    {
+      "epoch": 0.999607996863975,
+      "grad_norm": 4.611093521118164,
+      "learning_rate": 6.3888888888888885e-06,
+      "loss": 1.0693,
+      "step": 5100
+    },
+    {
+      "epoch": 1.0015680125441004,
+      "grad_norm": 4.401882171630859,
+      "learning_rate": 6.37962962962963e-06,
+      "loss": 1.0297,
+      "step": 5110
+    },
+    {
+      "epoch": 1.0035280282242258,
+      "grad_norm": 4.928489685058594,
+      "learning_rate": 6.370370370370371e-06,
+      "loss": 1.0276,
+      "step": 5120
+    },
+    {
+      "epoch": 1.0054880439043512,
+      "grad_norm": 5.5292487144470215,
+      "learning_rate": 6.361111111111111e-06,
+      "loss": 1.0267,
+      "step": 5130
+    },
+    {
+      "epoch": 1.0074480595844766,
+      "grad_norm": 5.00799036026001,
+      "learning_rate": 6.351851851851853e-06,
+      "loss": 1.0173,
+      "step": 5140
+    },
+    {
+      "epoch": 1.0094080752646022,
+      "grad_norm": 4.399227619171143,
+      "learning_rate": 6.342592592592594e-06,
+      "loss": 1.0106,
+      "step": 5150
+    },
+    {
+      "epoch": 1.0113680909447276,
+      "grad_norm": 5.132420539855957,
+      "learning_rate": 6.333333333333333e-06,
+      "loss": 1.0212,
+      "step": 5160
+    },
+    {
+      "epoch": 1.013328106624853,
+      "grad_norm": 5.298924446105957,
+      "learning_rate": 6.324074074074075e-06,
+      "loss": 1.0122,
+      "step": 5170
+    },
+    {
+      "epoch": 1.0152881223049783,
+      "grad_norm": 5.066485404968262,
+      "learning_rate": 6.314814814814816e-06,
+      "loss": 1.0114,
+      "step": 5180
+    },
+    {
+      "epoch": 1.017248137985104,
+      "grad_norm": 6.66790771484375,
+      "learning_rate": 6.305555555555556e-06,
+      "loss": 1.0128,
+      "step": 5190
+    },
+    {
+      "epoch": 1.0192081536652293,
+      "grad_norm": 4.905239105224609,
+      "learning_rate": 6.296296296296297e-06,
+      "loss": 1.0118,
+      "step": 5200
+    },
+    {
+      "epoch": 1.0211681693453547,
+      "grad_norm": 5.377009868621826,
+      "learning_rate": 6.287037037037037e-06,
+      "loss": 1.0229,
+      "step": 5210
+    },
+    {
+      "epoch": 1.0231281850254803,
+      "grad_norm": 5.1396942138671875,
+      "learning_rate": 6.277777777777778e-06,
+      "loss": 1.0221,
+      "step": 5220
+    },
+    {
+      "epoch": 1.0250882007056057,
+      "grad_norm": 5.001946449279785,
+      "learning_rate": 6.2685185185185195e-06,
+      "loss": 1.0461,
+      "step": 5230
+    },
+    {
+      "epoch": 1.027048216385731,
+      "grad_norm": 6.100106716156006,
+      "learning_rate": 6.259259259259259e-06,
+      "loss": 1.0169,
+      "step": 5240
+    },
+    {
+      "epoch": 1.0290082320658565,
+      "grad_norm": 5.1383209228515625,
+      "learning_rate": 6.25e-06,
+      "loss": 1.0282,
+      "step": 5250
+    },
+    {
+      "epoch": 1.030968247745982,
+      "grad_norm": 5.477851390838623,
+      "learning_rate": 6.2407407407407415e-06,
+      "loss": 1.0266,
+      "step": 5260
+    },
+    {
+      "epoch": 1.0329282634261074,
+      "grad_norm": 5.047489643096924,
+      "learning_rate": 6.231481481481482e-06,
+      "loss": 1.0256,
+      "step": 5270
+    },
+    {
+      "epoch": 1.0348882791062328,
+      "grad_norm": 4.994131565093994,
+      "learning_rate": 6.222222222222223e-06,
+      "loss": 1.0224,
+      "step": 5280
+    },
+    {
+      "epoch": 1.0368482947863582,
+      "grad_norm": 5.558932781219482,
+      "learning_rate": 6.2129629629629636e-06,
+      "loss": 1.0186,
+      "step": 5290
+    },
+    {
+      "epoch": 1.0388083104664838,
+      "grad_norm": 5.365056037902832,
+      "learning_rate": 6.203703703703704e-06,
+      "loss": 1.0095,
+      "step": 5300
+    },
+    {
+      "epoch": 1.0407683261466092,
+      "grad_norm": 5.53687858581543,
+      "learning_rate": 6.194444444444445e-06,
+      "loss": 1.0132,
+      "step": 5310
+    },
+    {
+      "epoch": 1.0427283418267346,
+      "grad_norm": 5.936549663543701,
+      "learning_rate": 6.1851851851851856e-06,
+      "loss": 1.0347,
+      "step": 5320
+    },
+    {
+      "epoch": 1.04468835750686,
+      "grad_norm": 5.228058338165283,
+      "learning_rate": 6.175925925925926e-06,
+      "loss": 1.0234,
+      "step": 5330
+    },
+    {
+      "epoch": 1.0466483731869856,
+      "grad_norm": 4.896886348724365,
+      "learning_rate": 6.166666666666667e-06,
+      "loss": 1.0065,
+      "step": 5340
+    },
+    {
+      "epoch": 1.048608388867111,
+      "grad_norm": 6.313221454620361,
+      "learning_rate": 6.157407407407408e-06,
+      "loss": 1.0191,
+      "step": 5350
+    },
+    {
+      "epoch": 1.0505684045472363,
+      "grad_norm": 5.8848114013671875,
+      "learning_rate": 6.148148148148149e-06,
+      "loss": 1.0378,
+      "step": 5360
+    },
+    {
+      "epoch": 1.052528420227362,
+      "grad_norm": 5.468878746032715,
+      "learning_rate": 6.13888888888889e-06,
+      "loss": 1.0056,
+      "step": 5370
+    },
+    {
+      "epoch": 1.0544884359074873,
+      "grad_norm": 5.208171367645264,
+      "learning_rate": 6.12962962962963e-06,
+      "loss": 1.0194,
+      "step": 5380
+    },
+    {
+      "epoch": 1.0564484515876127,
+      "grad_norm": 5.1294450759887695,
+      "learning_rate": 6.120370370370371e-06,
+      "loss": 1.0084,
+      "step": 5390
+    },
+    {
+      "epoch": 1.058408467267738,
+      "grad_norm": 5.981677532196045,
+      "learning_rate": 6.111111111111112e-06,
+      "loss": 1.0148,
+      "step": 5400
+    },
+    {
+      "epoch": 1.0603684829478637,
+      "grad_norm": 5.662644863128662,
+      "learning_rate": 6.1018518518518525e-06,
+      "loss": 1.0163,
+      "step": 5410
+    },
+    {
+      "epoch": 1.062328498627989,
+      "grad_norm": 5.452698230743408,
+      "learning_rate": 6.092592592592593e-06,
+      "loss": 1.0047,
+      "step": 5420
+    },
+    {
+      "epoch": 1.0642885143081144,
+      "grad_norm": 4.969189167022705,
+      "learning_rate": 6.083333333333333e-06,
+      "loss": 1.0031,
+      "step": 5430
+    },
+    {
+      "epoch": 1.0662485299882398,
+      "grad_norm": 5.396177768707275,
+      "learning_rate": 6.0740740740740745e-06,
+      "loss": 1.0109,
+      "step": 5440
+    },
+    {
+      "epoch": 1.0682085456683654,
+      "grad_norm": 6.180945873260498,
+      "learning_rate": 6.064814814814816e-06,
+      "loss": 1.0318,
+      "step": 5450
+    },
+    {
+      "epoch": 1.0701685613484908,
+      "grad_norm": 4.878726959228516,
+      "learning_rate": 6.055555555555555e-06,
+      "loss": 1.0257,
+      "step": 5460
+    },
+    {
+      "epoch": 1.0721285770286162,
+      "grad_norm": 5.634671688079834,
+      "learning_rate": 6.0462962962962965e-06,
+      "loss": 1.0241,
+      "step": 5470
+    },
+    {
+      "epoch": 1.0740885927087416,
+      "grad_norm": 5.971144199371338,
+      "learning_rate": 6.037037037037038e-06,
+      "loss": 1.0247,
+      "step": 5480
+    },
+    {
+      "epoch": 1.0760486083888672,
+      "grad_norm": 5.1925950050354,
+      "learning_rate": 6.027777777777778e-06,
+      "loss": 1.0094,
+      "step": 5490
+    },
+    {
+      "epoch": 1.0780086240689926,
+      "grad_norm": 4.921753406524658,
+      "learning_rate": 6.018518518518519e-06,
+      "loss": 1.0233,
+      "step": 5500
+    },
+    {
+      "epoch": 1.079968639749118,
+      "grad_norm": 5.969301700592041,
+      "learning_rate": 6.009259259259261e-06,
+      "loss": 0.994,
+      "step": 5510
+    },
+    {
+      "epoch": 1.0819286554292433,
+      "grad_norm": 5.875487804412842,
+      "learning_rate": 6e-06,
+      "loss": 1.0171,
+      "step": 5520
+    },
+    {
+      "epoch": 1.083888671109369,
+      "grad_norm": 5.540329933166504,
+      "learning_rate": 5.990740740740741e-06,
+      "loss": 1.0198,
+      "step": 5530
+    },
+    {
+      "epoch": 1.0858486867894943,
+      "grad_norm": 5.154452323913574,
+      "learning_rate": 5.981481481481482e-06,
+      "loss": 1.0174,
+      "step": 5540
+    },
+    {
+      "epoch": 1.0878087024696197,
+      "grad_norm": 5.469849586486816,
+      "learning_rate": 5.972222222222222e-06,
+      "loss": 1.0141,
+      "step": 5550
+    },
+    {
+      "epoch": 1.0897687181497453,
+      "grad_norm": 6.041581153869629,
+      "learning_rate": 5.962962962962963e-06,
+      "loss": 1.0216,
+      "step": 5560
+    },
+    {
+      "epoch": 1.0917287338298707,
+      "grad_norm": 5.2390522956848145,
+      "learning_rate": 5.953703703703704e-06,
+      "loss": 1.0202,
+      "step": 5570
+    },
+    {
+      "epoch": 1.093688749509996,
+      "grad_norm": 5.63343620300293,
+      "learning_rate": 5.944444444444445e-06,
+      "loss": 1.0234,
+      "step": 5580
+    },
+    {
+      "epoch": 1.0956487651901214,
+      "grad_norm": 14.769401550292969,
+      "learning_rate": 5.935185185185186e-06,
+      "loss": 1.019,
+      "step": 5590
+    },
+    {
+      "epoch": 1.097608780870247,
+      "grad_norm": 5.917840957641602,
+      "learning_rate": 5.925925925925926e-06,
+      "loss": 1.0259,
+      "step": 5600
+    },
+    {
+      "epoch": 1.097608780870247,
+      "eval_loss": 1.0469825267791748,
+      "eval_runtime": 14.056,
+      "eval_samples_per_second": 46.528,
+      "eval_steps_per_second": 5.834,
+      "step": 5600
+    },
+    {
+      "epoch": 1.0995687965503724,
+      "grad_norm": 4.929704666137695,
+      "learning_rate": 5.916666666666667e-06,
+      "loss": 1.0146,
+      "step": 5610
+    },
+    {
+      "epoch": 1.1015288122304978,
+      "grad_norm": 6.432901382446289,
+      "learning_rate": 5.907407407407408e-06,
+      "loss": 1.0268,
+      "step": 5620
+    },
+    {
+      "epoch": 1.1034888279106232,
+      "grad_norm": 5.020516872406006,
+      "learning_rate": 5.898148148148149e-06,
+      "loss": 1.0272,
+      "step": 5630
+    },
+    {
+      "epoch": 1.1054488435907488,
+      "grad_norm": 5.231268882751465,
+      "learning_rate": 5.88888888888889e-06,
+      "loss": 1.0046,
+      "step": 5640
+    },
+    {
+      "epoch": 1.1074088592708742,
+      "grad_norm": 5.203913688659668,
+      "learning_rate": 5.8796296296296295e-06,
+      "loss": 1.015,
+      "step": 5650
+    },
+    {
+      "epoch": 1.1093688749509996,
+      "grad_norm": 5.593926906585693,
+      "learning_rate": 5.870370370370371e-06,
+      "loss": 1.0185,
+      "step": 5660
+    },
+    {
+      "epoch": 1.1113288906311252,
+      "grad_norm": 5.165170669555664,
+      "learning_rate": 5.861111111111112e-06,
+      "loss": 1.0056,
+      "step": 5670
+    },
+    {
+      "epoch": 1.1132889063112505,
+      "grad_norm": 5.521978855133057,
+      "learning_rate": 5.8518518518518515e-06,
+      "loss": 1.022,
+      "step": 5680
+    },
+    {
+      "epoch": 1.115248921991376,
+      "grad_norm": 5.280319690704346,
+      "learning_rate": 5.842592592592593e-06,
+      "loss": 1.0357,
+      "step": 5690
+    },
+    {
+      "epoch": 1.1172089376715013,
+      "grad_norm": 6.761595726013184,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 1.0229,
+      "step": 5700
+    },
+    {
+      "epoch": 1.119168953351627,
+      "grad_norm": 5.606836318969727,
+      "learning_rate": 5.824074074074074e-06,
+      "loss": 1.019,
+      "step": 5710
+    },
+    {
+      "epoch": 1.1211289690317523,
+      "grad_norm": 5.351015567779541,
+      "learning_rate": 5.814814814814816e-06,
+      "loss": 1.0343,
+      "step": 5720
+    },
+    {
+      "epoch": 1.1230889847118777,
+      "grad_norm": 4.853587627410889,
+      "learning_rate": 5.805555555555557e-06,
+      "loss": 1.0239,
+      "step": 5730
+    },
+    {
+      "epoch": 1.125049000392003,
+      "grad_norm": 16.355466842651367,
+      "learning_rate": 5.796296296296296e-06,
+      "loss": 1.0013,
+      "step": 5740
+    },
+    {
+      "epoch": 1.1270090160721287,
+      "grad_norm": 5.404659748077393,
+      "learning_rate": 5.787037037037038e-06,
+      "loss": 1.0181,
+      "step": 5750
+    },
+    {
+      "epoch": 1.128969031752254,
+      "grad_norm": 5.103781700134277,
+      "learning_rate": 5.777777777777778e-06,
+      "loss": 1.0213,
+      "step": 5760
+    },
+    {
+      "epoch": 1.1309290474323794,
+      "grad_norm": 5.420756816864014,
+      "learning_rate": 5.768518518518519e-06,
+      "loss": 1.0072,
+      "step": 5770
+    },
+    {
+      "epoch": 1.1328890631125048,
+      "grad_norm": 5.301304817199707,
+      "learning_rate": 5.75925925925926e-06,
+      "loss": 0.9901,
+      "step": 5780
+    },
+    {
+      "epoch": 1.1348490787926304,
+      "grad_norm": 4.864381790161133,
+      "learning_rate": 5.75e-06,
+      "loss": 1.0129,
+      "step": 5790
+    },
+    {
+      "epoch": 1.1368090944727558,
+      "grad_norm": 6.017991542816162,
+      "learning_rate": 5.740740740740741e-06,
+      "loss": 1.0113,
+      "step": 5800
+    },
+    {
+      "epoch": 1.1387691101528812,
+      "grad_norm": 5.6904520988464355,
+      "learning_rate": 5.7314814814814825e-06,
+      "loss": 1.0211,
+      "step": 5810
+    },
+    {
+      "epoch": 1.1407291258330066,
+      "grad_norm": 5.778028964996338,
+      "learning_rate": 5.722222222222222e-06,
+      "loss": 1.0247,
+      "step": 5820
+    },
+    {
+      "epoch": 1.1426891415131322,
+      "grad_norm": 5.682682991027832,
+      "learning_rate": 5.712962962962963e-06,
+      "loss": 1.0309,
+      "step": 5830
+    },
+    {
+      "epoch": 1.1446491571932575,
+      "grad_norm": 5.886664867401123,
+      "learning_rate": 5.7037037037037045e-06,
+      "loss": 1.0244,
+      "step": 5840
+    },
+    {
+      "epoch": 1.146609172873383,
+      "grad_norm": 5.014996528625488,
+      "learning_rate": 5.694444444444445e-06,
+      "loss": 1.0207,
+      "step": 5850
+    },
+    {
+      "epoch": 1.1485691885535085,
+      "grad_norm": 5.563379287719727,
+      "learning_rate": 5.685185185185186e-06,
+      "loss": 1.0276,
+      "step": 5860
+    },
+    {
+      "epoch": 1.150529204233634,
+      "grad_norm": 5.241649627685547,
+      "learning_rate": 5.675925925925926e-06,
+      "loss": 1.022,
+      "step": 5870
+    },
+    {
+      "epoch": 1.1524892199137593,
+      "grad_norm": 5.154331207275391,
+      "learning_rate": 5.666666666666667e-06,
+      "loss": 1.0232,
+      "step": 5880
+    },
+    {
+      "epoch": 1.1544492355938847,
+      "grad_norm": 5.823696136474609,
+      "learning_rate": 5.657407407407408e-06,
+      "loss": 1.0364,
+      "step": 5890
+    },
+    {
+      "epoch": 1.1564092512740103,
+      "grad_norm": 5.451704978942871,
+      "learning_rate": 5.6481481481481485e-06,
+      "loss": 1.02,
+      "step": 5900
+    },
+    {
+      "epoch": 1.1583692669541357,
+      "grad_norm": 5.540503978729248,
+      "learning_rate": 5.638888888888889e-06,
+      "loss": 1.0127,
+      "step": 5910
+    },
+    {
+      "epoch": 1.160329282634261,
+      "grad_norm": 5.463021278381348,
+      "learning_rate": 5.62962962962963e-06,
+      "loss": 1.0205,
+      "step": 5920
+    },
+    {
+      "epoch": 1.1622892983143864,
+      "grad_norm": 5.372107028961182,
+      "learning_rate": 5.6203703703703705e-06,
+      "loss": 1.0085,
+      "step": 5930
+    },
+    {
+      "epoch": 1.164249313994512,
+      "grad_norm": 5.14231538772583,
+      "learning_rate": 5.611111111111112e-06,
+      "loss": 1.0189,
+      "step": 5940
+    },
+    {
+      "epoch": 1.1662093296746374,
+      "grad_norm": 5.416358470916748,
+      "learning_rate": 5.601851851851853e-06,
+      "loss": 1.0066,
+      "step": 5950
+    },
+    {
+      "epoch": 1.1681693453547628,
+      "grad_norm": 5.824409484863281,
+      "learning_rate": 5.5925925925925926e-06,
+      "loss": 1.0294,
+      "step": 5960
+    },
+    {
+      "epoch": 1.1701293610348882,
+      "grad_norm": 5.362888336181641,
+      "learning_rate": 5.583333333333334e-06,
+      "loss": 1.011,
+      "step": 5970
+    },
+    {
+      "epoch": 1.1720893767150138,
+      "grad_norm": 5.583401679992676,
+      "learning_rate": 5.574074074074075e-06,
+      "loss": 1.0218,
+      "step": 5980
+    },
+    {
+      "epoch": 1.1740493923951392,
+      "grad_norm": 5.501253128051758,
+      "learning_rate": 5.5648148148148154e-06,
+      "loss": 1.0198,
+      "step": 5990
+    },
+    {
+      "epoch": 1.1760094080752646,
+      "grad_norm": 5.536640167236328,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 1.0243,
+      "step": 6000
+    },
+    {
+      "epoch": 1.17796942375539,
+      "grad_norm": 5.525614261627197,
+      "learning_rate": 5.546296296296296e-06,
+      "loss": 1.0234,
+      "step": 6010
+    },
+    {
+      "epoch": 1.1799294394355155,
+      "grad_norm": 5.270368576049805,
+      "learning_rate": 5.5370370370370374e-06,
+      "loss": 1.0214,
+      "step": 6020
+    },
+    {
+      "epoch": 1.181889455115641,
+      "grad_norm": 5.895664691925049,
+      "learning_rate": 5.527777777777779e-06,
+      "loss": 1.0203,
+      "step": 6030
+    },
+    {
+      "epoch": 1.1838494707957663,
+      "grad_norm": 5.530089378356934,
+      "learning_rate": 5.518518518518518e-06,
+      "loss": 1.0254,
+      "step": 6040
+    },
+    {
+      "epoch": 1.185809486475892,
+      "grad_norm": 5.275720119476318,
+      "learning_rate": 5.5092592592592595e-06,
+      "loss": 1.015,
+      "step": 6050
+    },
+    {
+      "epoch": 1.1877695021560173,
+      "grad_norm": 5.464200496673584,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 1.0134,
+      "step": 6060
+    },
+    {
+      "epoch": 1.1897295178361427,
+      "grad_norm": 5.803837299346924,
+      "learning_rate": 5.490740740740741e-06,
+      "loss": 1.0125,
+      "step": 6070
+    },
+    {
+      "epoch": 1.191689533516268,
+      "grad_norm": 5.1351752281188965,
+      "learning_rate": 5.481481481481482e-06,
+      "loss": 1.0061,
+      "step": 6080
+    },
+    {
+      "epoch": 1.1936495491963937,
+      "grad_norm": 5.232799530029297,
+      "learning_rate": 5.4722222222222236e-06,
+      "loss": 1.0213,
+      "step": 6090
+    },
+    {
+      "epoch": 1.195609564876519,
+      "grad_norm": 5.996954917907715,
+      "learning_rate": 5.462962962962963e-06,
+      "loss": 1.0177,
+      "step": 6100
+    },
+    {
+      "epoch": 1.1975695805566444,
+      "grad_norm": 7.579484939575195,
+      "learning_rate": 5.453703703703704e-06,
+      "loss": 1.0133,
+      "step": 6110
+    },
+    {
+      "epoch": 1.1995295962367698,
+      "grad_norm": 5.351324081420898,
+      "learning_rate": 5.444444444444445e-06,
+      "loss": 1.0019,
+      "step": 6120
+    },
+    {
+      "epoch": 1.2014896119168954,
+      "grad_norm": 5.5653581619262695,
+      "learning_rate": 5.435185185185186e-06,
+      "loss": 1.0069,
+      "step": 6130
+    },
+    {
+      "epoch": 1.2034496275970208,
+      "grad_norm": 5.198008060455322,
+      "learning_rate": 5.425925925925926e-06,
+      "loss": 1.0157,
+      "step": 6140
+    },
+    {
+      "epoch": 1.2054096432771462,
+      "grad_norm": 5.67602014541626,
+      "learning_rate": 5.416666666666667e-06,
+      "loss": 1.0327,
+      "step": 6150
+    },
+    {
+      "epoch": 1.2073696589572718,
+      "grad_norm": 5.185479164123535,
+      "learning_rate": 5.407407407407408e-06,
+      "loss": 1.0118,
+      "step": 6160
+    },
+    {
+      "epoch": 1.2093296746373972,
+      "grad_norm": 5.254889965057373,
+      "learning_rate": 5.398148148148149e-06,
+      "loss": 1.0282,
+      "step": 6170
+    },
+    {
+      "epoch": 1.2112896903175225,
+      "grad_norm": 5.635478496551514,
+      "learning_rate": 5.388888888888889e-06,
+      "loss": 1.0211,
+      "step": 6180
+    },
+    {
+      "epoch": 1.213249705997648,
+      "grad_norm": 5.388778209686279,
+      "learning_rate": 5.37962962962963e-06,
+      "loss": 1.0171,
+      "step": 6190
+    },
+    {
+      "epoch": 1.2152097216777733,
+      "grad_norm": 5.546689987182617,
+      "learning_rate": 5.370370370370371e-06,
+      "loss": 1.0113,
+      "step": 6200
+    },
+    {
+      "epoch": 1.217169737357899,
+      "grad_norm": 5.316011428833008,
+      "learning_rate": 5.361111111111112e-06,
+      "loss": 1.0183,
+      "step": 6210
+    },
+    {
+      "epoch": 1.2191297530380243,
+      "grad_norm": 6.057391166687012,
+      "learning_rate": 5.351851851851853e-06,
+      "loss": 1.0149,
+      "step": 6220
+    },
+    {
+      "epoch": 1.2210897687181497,
+      "grad_norm": 5.451624393463135,
+      "learning_rate": 5.342592592592592e-06,
+      "loss": 1.0129,
+      "step": 6230
+    },
+    {
+      "epoch": 1.2230497843982753,
+      "grad_norm": 5.58920955657959,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.0152,
+      "step": 6240
+    },
+    {
+      "epoch": 1.2250098000784007,
+      "grad_norm": 5.132778167724609,
+      "learning_rate": 5.324074074074075e-06,
+      "loss": 1.0059,
+      "step": 6250
+    },
+    {
+      "epoch": 1.226969815758526,
+      "grad_norm": 4.807291030883789,
+      "learning_rate": 5.314814814814815e-06,
+      "loss": 1.0146,
+      "step": 6260
+    },
+    {
+      "epoch": 1.2289298314386514,
+      "grad_norm": 5.3906474113464355,
+      "learning_rate": 5.305555555555556e-06,
+      "loss": 1.0045,
+      "step": 6270
+    },
+    {
+      "epoch": 1.230889847118777,
+      "grad_norm": 5.229481220245361,
+      "learning_rate": 5.296296296296297e-06,
+      "loss": 1.0018,
+      "step": 6280
+    },
+    {
+      "epoch": 1.2328498627989024,
+      "grad_norm": 5.5774126052856445,
+      "learning_rate": 5.287037037037037e-06,
+      "loss": 1.0096,
+      "step": 6290
+    },
+    {
+      "epoch": 1.2348098784790278,
+      "grad_norm": 5.752622127532959,
+      "learning_rate": 5.2777777777777785e-06,
+      "loss": 1.0173,
+      "step": 6300
+    },
+    {
+      "epoch": 1.2348098784790278,
+      "eval_loss": 1.0454390048980713,
+      "eval_runtime": 14.0845,
+      "eval_samples_per_second": 46.434,
+      "eval_steps_per_second": 5.822,
+      "step": 6300
+    },
+    {
+      "epoch": 1.2367698941591532,
+      "grad_norm": 4.995856285095215,
+      "learning_rate": 5.26851851851852e-06,
+      "loss": 1.0274,
+      "step": 6310
+    },
+    {
+      "epoch": 1.2387299098392788,
+      "grad_norm": 5.631256103515625,
+      "learning_rate": 5.259259259259259e-06,
+      "loss": 1.029,
+      "step": 6320
+    },
+    {
+      "epoch": 1.2406899255194042,
+      "grad_norm": 5.617944717407227,
+      "learning_rate": 5.2500000000000006e-06,
+      "loss": 1.0087,
+      "step": 6330
+    },
+    {
+      "epoch": 1.2426499411995295,
+      "grad_norm": 5.174662113189697,
+      "learning_rate": 5.240740740740741e-06,
+      "loss": 1.0114,
+      "step": 6340
+    },
+    {
+      "epoch": 1.2446099568796551,
+      "grad_norm": 5.841761112213135,
+      "learning_rate": 5.231481481481482e-06,
+      "loss": 1.0113,
+      "step": 6350
+    },
+    {
+      "epoch": 1.2465699725597805,
+      "grad_norm": 5.008285999298096,
+      "learning_rate": 5.2222222222222226e-06,
+      "loss": 1.0058,
+      "step": 6360
+    },
+    {
+      "epoch": 1.248529988239906,
+      "grad_norm": 5.529172420501709,
+      "learning_rate": 5.212962962962963e-06,
+      "loss": 1.0225,
+      "step": 6370
+    },
+    {
+      "epoch": 1.2504900039200313,
+      "grad_norm": 6.809742450714111,
+      "learning_rate": 5.203703703703704e-06,
+      "loss": 1.0194,
+      "step": 6380
+    },
+    {
+      "epoch": 1.2524500196001567,
+      "grad_norm": 5.778404712677002,
+      "learning_rate": 5.1944444444444454e-06,
+      "loss": 1.0312,
+      "step": 6390
+    },
+    {
+      "epoch": 1.2544100352802823,
+      "grad_norm": 6.095964431762695,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 1.0155,
+      "step": 6400
+    },
+    {
+      "epoch": 1.2563700509604077,
+      "grad_norm": 5.453831672668457,
+      "learning_rate": 5.175925925925926e-06,
+      "loss": 1.0155,
+      "step": 6410
+    },
+    {
+      "epoch": 1.258330066640533,
+      "grad_norm": 5.425611972808838,
+      "learning_rate": 5.1666666666666675e-06,
+      "loss": 1.0353,
+      "step": 6420
+    },
+    {
+      "epoch": 1.2602900823206586,
+      "grad_norm": 6.835920810699463,
+      "learning_rate": 5.157407407407408e-06,
+      "loss": 1.0089,
+      "step": 6430
+    },
+    {
+      "epoch": 1.262250098000784,
+      "grad_norm": 8.35204029083252,
+      "learning_rate": 5.148148148148149e-06,
+      "loss": 1.0235,
+      "step": 6440
+    },
+    {
+      "epoch": 1.2642101136809094,
+      "grad_norm": 5.1433234214782715,
+      "learning_rate": 5.138888888888889e-06,
+      "loss": 1.0226,
+      "step": 6450
+    },
+    {
+      "epoch": 1.266170129361035,
+      "grad_norm": 5.907522201538086,
+      "learning_rate": 5.12962962962963e-06,
+      "loss": 1.0161,
+      "step": 6460
+    },
+    {
+      "epoch": 1.2681301450411604,
+      "grad_norm": 5.631725788116455,
+      "learning_rate": 5.120370370370371e-06,
+      "loss": 1.014,
+      "step": 6470
+    },
+    {
+      "epoch": 1.2700901607212858,
+      "grad_norm": 5.635436058044434,
+      "learning_rate": 5.1111111111111115e-06,
+      "loss": 1.0216,
+      "step": 6480
+    },
+    {
+      "epoch": 1.2720501764014112,
+      "grad_norm": 6.376223087310791,
+      "learning_rate": 5.101851851851852e-06,
+      "loss": 1.0135,
+      "step": 6490
+    },
+    {
+      "epoch": 1.2740101920815365,
+      "grad_norm": 6.083188533782959,
+      "learning_rate": 5.092592592592593e-06,
+      "loss": 1.0117,
+      "step": 6500
+    },
+    {
+      "epoch": 1.2759702077616621,
+      "grad_norm": 5.436441898345947,
+      "learning_rate": 5.0833333333333335e-06,
+      "loss": 1.0079,
+      "step": 6510
+    },
+    {
+      "epoch": 1.2779302234417875,
+      "grad_norm": 5.303137302398682,
+      "learning_rate": 5.074074074074075e-06,
+      "loss": 1.02,
+      "step": 6520
+    },
+    {
+      "epoch": 1.279890239121913,
+      "grad_norm": 5.048837661743164,
+      "learning_rate": 5.064814814814816e-06,
+      "loss": 1.0045,
+      "step": 6530
+    },
+    {
+      "epoch": 1.2818502548020385,
+      "grad_norm": 6.105396747589111,
+      "learning_rate": 5.0555555555555555e-06,
+      "loss": 1.0105,
+      "step": 6540
+    },
+    {
+      "epoch": 1.283810270482164,
+      "grad_norm": 4.97545051574707,
+      "learning_rate": 5.046296296296297e-06,
+      "loss": 1.0128,
+      "step": 6550
+    },
+    {
+      "epoch": 1.2857702861622893,
+      "grad_norm": 6.359137535095215,
+      "learning_rate": 5.037037037037037e-06,
+      "loss": 1.0127,
+      "step": 6560
+    },
+    {
+      "epoch": 1.2877303018424147,
+      "grad_norm": 6.185931205749512,
+      "learning_rate": 5.027777777777778e-06,
+      "loss": 1.0156,
+      "step": 6570
+    },
+    {
+      "epoch": 1.28969031752254,
+      "grad_norm": 6.333951950073242,
+      "learning_rate": 5.01851851851852e-06,
+      "loss": 1.0113,
+      "step": 6580
+    },
+    {
+      "epoch": 1.2916503332026656,
+      "grad_norm": 5.14535665512085,
+      "learning_rate": 5.009259259259259e-06,
+      "loss": 1.0062,
+      "step": 6590
+    },
+    {
+      "epoch": 1.293610348882791,
+      "grad_norm": 6.086668491363525,
+      "learning_rate": 5e-06,
+      "loss": 1.0123,
+      "step": 6600
+    },
+    {
+      "epoch": 1.2955703645629164,
+      "grad_norm": 5.336708068847656,
+      "learning_rate": 4.990740740740741e-06,
+      "loss": 1.0058,
+      "step": 6610
+    },
+    {
+      "epoch": 1.297530380243042,
+      "grad_norm": 5.65183687210083,
+      "learning_rate": 4.981481481481482e-06,
+      "loss": 1.0232,
+      "step": 6620
+    },
+    {
+      "epoch": 1.2994903959231674,
+      "grad_norm": 5.356630802154541,
+      "learning_rate": 4.9722222222222224e-06,
+      "loss": 1.0248,
+      "step": 6630
+    },
+    {
+      "epoch": 1.3014504116032928,
+      "grad_norm": 5.61469841003418,
+      "learning_rate": 4.962962962962964e-06,
+      "loss": 1.0049,
+      "step": 6640
+    },
+    {
+      "epoch": 1.3034104272834184,
+      "grad_norm": 5.575668811798096,
+      "learning_rate": 4.953703703703704e-06,
+      "loss": 1.0046,
+      "step": 6650
+    },
+    {
+      "epoch": 1.3053704429635438,
+      "grad_norm": 5.938807010650635,
+      "learning_rate": 4.944444444444445e-06,
+      "loss": 1.0254,
+      "step": 6660
+    },
+    {
+      "epoch": 1.3073304586436691,
+      "grad_norm": 5.57589054107666,
+      "learning_rate": 4.935185185185186e-06,
+      "loss": 1.016,
+      "step": 6670
+    },
+    {
+      "epoch": 1.3092904743237945,
+      "grad_norm": 5.684128284454346,
+      "learning_rate": 4.925925925925926e-06,
+      "loss": 1.0102,
+      "step": 6680
+    },
+    {
+      "epoch": 1.31125049000392,
+      "grad_norm": 5.43093729019165,
+      "learning_rate": 4.9166666666666665e-06,
+      "loss": 1.0092,
+      "step": 6690
+    },
+    {
+      "epoch": 1.3132105056840455,
+      "grad_norm": 5.595739841461182,
+      "learning_rate": 4.907407407407408e-06,
+      "loss": 1.0103,
+      "step": 6700
+    },
+    {
+      "epoch": 1.315170521364171,
+      "grad_norm": 6.116888046264648,
+      "learning_rate": 4.898148148148149e-06,
+      "loss": 1.0177,
+      "step": 6710
+    },
+    {
+      "epoch": 1.3171305370442963,
+      "grad_norm": 5.478735446929932,
+      "learning_rate": 4.888888888888889e-06,
+      "loss": 1.0147,
+      "step": 6720
+    },
+    {
+      "epoch": 1.3190905527244219,
+      "grad_norm": 5.878492832183838,
+      "learning_rate": 4.8796296296296306e-06,
+      "loss": 1.0189,
+      "step": 6730
+    },
+    {
+      "epoch": 1.3210505684045473,
+      "grad_norm": 5.890417098999023,
+      "learning_rate": 4.870370370370371e-06,
+      "loss": 1.0246,
+      "step": 6740
+    },
+    {
+      "epoch": 1.3230105840846726,
+      "grad_norm": 5.2328410148620605,
+      "learning_rate": 4.861111111111111e-06,
+      "loss": 1.0116,
+      "step": 6750
+    },
+    {
+      "epoch": 1.3249705997647983,
+      "grad_norm": 6.1324005126953125,
+      "learning_rate": 4.851851851851852e-06,
+      "loss": 1.0028,
+      "step": 6760
+    },
+    {
+      "epoch": 1.3269306154449236,
+      "grad_norm": 5.392694473266602,
+      "learning_rate": 4.842592592592593e-06,
+      "loss": 1.0192,
+      "step": 6770
+    },
+    {
+      "epoch": 1.328890631125049,
+      "grad_norm": 5.902957439422607,
+      "learning_rate": 4.833333333333333e-06,
+      "loss": 1.0145,
+      "step": 6780
+    },
+    {
+      "epoch": 1.3308506468051744,
+      "grad_norm": 5.553819179534912,
+      "learning_rate": 4.824074074074075e-06,
+      "loss": 1.0091,
+      "step": 6790
+    },
+    {
+      "epoch": 1.3328106624852998,
+      "grad_norm": 6.013294219970703,
+      "learning_rate": 4.814814814814815e-06,
+      "loss": 1.0198,
+      "step": 6800
+    },
+    {
+      "epoch": 1.3347706781654254,
+      "grad_norm": 5.731135368347168,
+      "learning_rate": 4.805555555555556e-06,
+      "loss": 1.014,
+      "step": 6810
+    },
+    {
+      "epoch": 1.3367306938455508,
+      "grad_norm": 5.954074382781982,
+      "learning_rate": 4.796296296296297e-06,
+      "loss": 1.0128,
+      "step": 6820
+    },
+    {
+      "epoch": 1.3386907095256761,
+      "grad_norm": 5.550869464874268,
+      "learning_rate": 4.787037037037037e-06,
+      "loss": 1.0151,
+      "step": 6830
+    },
+    {
+      "epoch": 1.3406507252058018,
+      "grad_norm": 6.043259620666504,
+      "learning_rate": 4.777777777777778e-06,
+      "loss": 1.0069,
+      "step": 6840
+    },
+    {
+      "epoch": 1.3426107408859271,
+      "grad_norm": 5.002622127532959,
+      "learning_rate": 4.768518518518519e-06,
+      "loss": 1.0219,
+      "step": 6850
+    },
+    {
+      "epoch": 1.3445707565660525,
+      "grad_norm": 6.016299247741699,
+      "learning_rate": 4.75925925925926e-06,
+      "loss": 1.0158,
+      "step": 6860
+    },
+    {
+      "epoch": 1.346530772246178,
+      "grad_norm": 5.6830220222473145,
+      "learning_rate": 4.75e-06,
+      "loss": 1.0092,
+      "step": 6870
+    },
+    {
+      "epoch": 1.3484907879263033,
+      "grad_norm": 6.092140197753906,
+      "learning_rate": 4.7407407407407415e-06,
+      "loss": 1.0241,
+      "step": 6880
+    },
+    {
+      "epoch": 1.3504508036064289,
+      "grad_norm": 5.837907791137695,
+      "learning_rate": 4.731481481481482e-06,
+      "loss": 1.0259,
+      "step": 6890
+    },
+    {
+      "epoch": 1.3524108192865543,
+      "grad_norm": 5.598001480102539,
+      "learning_rate": 4.722222222222222e-06,
+      "loss": 1.0043,
+      "step": 6900
+    },
+    {
+      "epoch": 1.3543708349666796,
+      "grad_norm": 5.820865631103516,
+      "learning_rate": 4.712962962962963e-06,
+      "loss": 1.0325,
+      "step": 6910
+    },
+    {
+      "epoch": 1.3563308506468053,
+      "grad_norm": 6.162250518798828,
+      "learning_rate": 4.703703703703704e-06,
+      "loss": 1.0118,
+      "step": 6920
+    },
+    {
+      "epoch": 1.3582908663269306,
+      "grad_norm": 5.53606653213501,
+      "learning_rate": 4.694444444444445e-06,
+      "loss": 1.0099,
+      "step": 6930
+    },
+    {
+      "epoch": 1.360250882007056,
+      "grad_norm": 6.468924045562744,
+      "learning_rate": 4.6851851851851855e-06,
+      "loss": 1.0172,
+      "step": 6940
+    },
+    {
+      "epoch": 1.3622108976871816,
+      "grad_norm": 6.867157936096191,
+      "learning_rate": 4.675925925925927e-06,
+      "loss": 1.0021,
+      "step": 6950
+    },
+    {
+      "epoch": 1.364170913367307,
+      "grad_norm": 6.068783760070801,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 1.0083,
+      "step": 6960
+    },
+    {
+      "epoch": 1.3661309290474324,
+      "grad_norm": 5.502538204193115,
+      "learning_rate": 4.6574074074074076e-06,
+      "loss": 1.0173,
+      "step": 6970
+    },
+    {
+      "epoch": 1.3680909447275578,
+      "grad_norm": 5.83467960357666,
+      "learning_rate": 4.648148148148148e-06,
+      "loss": 1.031,
+      "step": 6980
+    },
+    {
+      "epoch": 1.3700509604076831,
+      "grad_norm": 5.383768558502197,
+      "learning_rate": 4.638888888888889e-06,
+      "loss": 1.0187,
+      "step": 6990
+    },
+    {
+      "epoch": 1.3720109760878088,
+      "grad_norm": 5.172204494476318,
+      "learning_rate": 4.62962962962963e-06,
+      "loss": 1.0079,
+      "step": 7000
+    },
+    {
+      "epoch": 1.3720109760878088,
+      "eval_loss": 1.0383639335632324,
+      "eval_runtime": 14.0609,
+      "eval_samples_per_second": 46.512,
+      "eval_steps_per_second": 5.832,
+      "step": 7000
+    },
+    {
+      "epoch": 1.3739709917679341,
+      "grad_norm": 5.811285495758057,
+      "learning_rate": 4.620370370370371e-06,
+      "loss": 1.0153,
+      "step": 7010
+    },
+    {
+      "epoch": 1.3759310074480595,
+      "grad_norm": 5.887601375579834,
+      "learning_rate": 4.611111111111112e-06,
+      "loss": 1.0104,
+      "step": 7020
+    },
+    {
+      "epoch": 1.3778910231281851,
+      "grad_norm": 5.245418548583984,
+      "learning_rate": 4.6018518518518524e-06,
+      "loss": 1.0063,
+      "step": 7030
+    },
+    {
+      "epoch": 1.3798510388083105,
+      "grad_norm": 6.191716194152832,
+      "learning_rate": 4.592592592592593e-06,
+      "loss": 1.0254,
+      "step": 7040
+    },
+    {
+      "epoch": 1.3818110544884359,
+      "grad_norm": 6.047815799713135,
+      "learning_rate": 4.583333333333333e-06,
+      "loss": 1.0291,
+      "step": 7050
+    },
+    {
+      "epoch": 1.3837710701685613,
+      "grad_norm": 5.58579158782959,
+      "learning_rate": 4.5740740740740745e-06,
+      "loss": 1.0126,
+      "step": 7060
+    },
+    {
+      "epoch": 1.3857310858486869,
+      "grad_norm": 5.651394844055176,
+      "learning_rate": 4.564814814814815e-06,
+      "loss": 1.013,
+      "step": 7070
+    },
+    {
+      "epoch": 1.3876911015288123,
+      "grad_norm": 5.703254699707031,
+      "learning_rate": 4.555555555555556e-06,
+      "loss": 1.0209,
+      "step": 7080
+    },
+    {
+      "epoch": 1.3896511172089376,
+      "grad_norm": 6.428845405578613,
+      "learning_rate": 4.5462962962962965e-06,
+      "loss": 1.0065,
+      "step": 7090
+    },
+    {
+      "epoch": 1.391611132889063,
+      "grad_norm": 5.854006767272949,
+      "learning_rate": 4.537037037037038e-06,
+      "loss": 1.0192,
+      "step": 7100
+    },
+    {
+      "epoch": 1.3935711485691886,
+      "grad_norm": 5.257528781890869,
+      "learning_rate": 4.527777777777778e-06,
+      "loss": 1.0125,
+      "step": 7110
+    },
+    {
+      "epoch": 1.395531164249314,
+      "grad_norm": 5.832418441772461,
+      "learning_rate": 4.5185185185185185e-06,
+      "loss": 1.0077,
+      "step": 7120
+    },
+    {
+      "epoch": 1.3974911799294394,
+      "grad_norm": 18.90093994140625,
+      "learning_rate": 4.50925925925926e-06,
+      "loss": 1.0039,
+      "step": 7130
+    },
+    {
+      "epoch": 1.399451195609565,
+      "grad_norm": 6.230194568634033,
+      "learning_rate": 4.5e-06,
+      "loss": 1.0168,
+      "step": 7140
+    },
+    {
+      "epoch": 1.4014112112896904,
+      "grad_norm": 5.916494846343994,
+      "learning_rate": 4.490740740740741e-06,
+      "loss": 1.0367,
+      "step": 7150
+    },
+    {
+      "epoch": 1.4033712269698158,
+      "grad_norm": 6.498172760009766,
+      "learning_rate": 4.481481481481482e-06,
+      "loss": 1.0226,
+      "step": 7160
+    },
+    {
+      "epoch": 1.4053312426499411,
+      "grad_norm": 5.681482791900635,
+      "learning_rate": 4.472222222222223e-06,
+      "loss": 1.0135,
+      "step": 7170
+    },
+    {
+      "epoch": 1.4072912583300665,
+      "grad_norm": 5.3359880447387695,
+      "learning_rate": 4.462962962962963e-06,
+      "loss": 1.0222,
+      "step": 7180
+    },
+    {
+      "epoch": 1.4092512740101921,
+      "grad_norm": 6.866293430328369,
+      "learning_rate": 4.453703703703704e-06,
+      "loss": 1.015,
+      "step": 7190
+    },
+    {
+      "epoch": 1.4112112896903175,
+      "grad_norm": 6.149731159210205,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 0.998,
+      "step": 7200
+    },
+    {
+      "epoch": 1.4131713053704429,
+      "grad_norm": 5.832232475280762,
+      "learning_rate": 4.435185185185185e-06,
+      "loss": 1.0097,
+      "step": 7210
+    },
+    {
+      "epoch": 1.4151313210505685,
+      "grad_norm": 5.8602986335754395,
+      "learning_rate": 4.425925925925927e-06,
+      "loss": 1.0133,
+      "step": 7220
+    },
+    {
+      "epoch": 1.4170913367306939,
+      "grad_norm": 6.716607093811035,
+      "learning_rate": 4.416666666666667e-06,
+      "loss": 1.0181,
+      "step": 7230
+    },
+    {
+      "epoch": 1.4190513524108193,
+      "grad_norm": 5.983255863189697,
+      "learning_rate": 4.407407407407408e-06,
+      "loss": 1.0152,
+      "step": 7240
+    },
+    {
+      "epoch": 1.4210113680909449,
+      "grad_norm": 5.979559898376465,
+      "learning_rate": 4.398148148148149e-06,
+      "loss": 1.0187,
+      "step": 7250
+    },
+    {
+      "epoch": 1.4229713837710702,
+      "grad_norm": 5.463419437408447,
+      "learning_rate": 4.388888888888889e-06,
+      "loss": 1.0018,
+      "step": 7260
+    },
+    {
+      "epoch": 1.4249313994511956,
+      "grad_norm": 5.58013916015625,
+      "learning_rate": 4.379629629629629e-06,
+      "loss": 1.0161,
+      "step": 7270
+    },
+    {
+      "epoch": 1.426891415131321,
+      "grad_norm": 5.896617412567139,
+      "learning_rate": 4.370370370370371e-06,
+      "loss": 1.0234,
+      "step": 7280
+    },
+    {
+      "epoch": 1.4288514308114464,
+      "grad_norm": 6.59088659286499,
+      "learning_rate": 4.361111111111112e-06,
+      "loss": 1.0116,
+      "step": 7290
+    },
+    {
+      "epoch": 1.430811446491572,
+      "grad_norm": 5.4909348487854,
+      "learning_rate": 4.351851851851852e-06,
+      "loss": 1.0241,
+      "step": 7300
+    },
+    {
+      "epoch": 1.4327714621716974,
+      "grad_norm": 5.706267356872559,
+      "learning_rate": 4.342592592592593e-06,
+      "loss": 1.0113,
+      "step": 7310
+    },
+    {
+      "epoch": 1.4347314778518228,
+      "grad_norm": 5.893515586853027,
+      "learning_rate": 4.333333333333334e-06,
+      "loss": 1.0074,
+      "step": 7320
+    },
+    {
+      "epoch": 1.4366914935319484,
+      "grad_norm": 5.337756156921387,
+      "learning_rate": 4.324074074074074e-06,
+      "loss": 1.0086,
+      "step": 7330
+    },
+    {
+      "epoch": 1.4386515092120737,
+      "grad_norm": 6.063902854919434,
+      "learning_rate": 4.314814814814815e-06,
+      "loss": 1.0135,
+      "step": 7340
+    },
+    {
+      "epoch": 1.4406115248921991,
+      "grad_norm": 5.862417221069336,
+      "learning_rate": 4.305555555555556e-06,
+      "loss": 1.0218,
+      "step": 7350
+    },
+    {
+      "epoch": 1.4425715405723245,
+      "grad_norm": 6.047614574432373,
+      "learning_rate": 4.296296296296296e-06,
+      "loss": 1.0147,
+      "step": 7360
+    },
+    {
+      "epoch": 1.4445315562524499,
+      "grad_norm": 7.018215656280518,
+      "learning_rate": 4.2870370370370376e-06,
+      "loss": 1.029,
+      "step": 7370
+    },
+    {
+      "epoch": 1.4464915719325755,
+      "grad_norm": 5.391064167022705,
+      "learning_rate": 4.277777777777778e-06,
+      "loss": 1.007,
+      "step": 7380
+    },
+    {
+      "epoch": 1.4484515876127009,
+      "grad_norm": 5.364874362945557,
+      "learning_rate": 4.268518518518519e-06,
+      "loss": 1.0188,
+      "step": 7390
+    },
+    {
+      "epoch": 1.4504116032928263,
+      "grad_norm": 5.990734577178955,
+      "learning_rate": 4.2592592592592596e-06,
+      "loss": 1.0003,
+      "step": 7400
+    },
+    {
+      "epoch": 1.4523716189729519,
+      "grad_norm": 5.727319717407227,
+      "learning_rate": 4.25e-06,
+      "loss": 1.0198,
+      "step": 7410
+    },
+    {
+      "epoch": 1.4543316346530772,
+      "grad_norm": 5.658199310302734,
+      "learning_rate": 4.240740740740741e-06,
+      "loss": 1.0221,
+      "step": 7420
+    },
+    {
+      "epoch": 1.4562916503332026,
+      "grad_norm": 5.460375785827637,
+      "learning_rate": 4.231481481481482e-06,
+      "loss": 1.0127,
+      "step": 7430
+    },
+    {
+      "epoch": 1.4582516660133282,
+      "grad_norm": 6.161991596221924,
+      "learning_rate": 4.222222222222223e-06,
+      "loss": 1.0025,
+      "step": 7440
+    },
+    {
+      "epoch": 1.4602116816934536,
+      "grad_norm": 5.7905192375183105,
+      "learning_rate": 4.212962962962963e-06,
+      "loss": 1.0071,
+      "step": 7450
+    },
+    {
+      "epoch": 1.462171697373579,
+      "grad_norm": 5.652228832244873,
+      "learning_rate": 4.2037037037037045e-06,
+      "loss": 1.0112,
+      "step": 7460
+    },
+    {
+      "epoch": 1.4641317130537044,
+      "grad_norm": 5.817644119262695,
+      "learning_rate": 4.194444444444445e-06,
+      "loss": 1.0191,
+      "step": 7470
+    },
+    {
+      "epoch": 1.4660917287338298,
+      "grad_norm": 6.155873775482178,
+      "learning_rate": 4.185185185185185e-06,
+      "loss": 1.013,
+      "step": 7480
+    },
+    {
+      "epoch": 1.4680517444139554,
+      "grad_norm": 5.775312423706055,
+      "learning_rate": 4.175925925925926e-06,
+      "loss": 1.0107,
+      "step": 7490
+    },
+    {
+      "epoch": 1.4700117600940807,
+      "grad_norm": 5.765095233917236,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.0151,
+      "step": 7500
+    },
+    {
+      "epoch": 1.4719717757742061,
+      "grad_norm": 5.336995601654053,
+      "learning_rate": 4.157407407407408e-06,
+      "loss": 1.0048,
+      "step": 7510
+    },
+    {
+      "epoch": 1.4739317914543317,
+      "grad_norm": 6.261089324951172,
+      "learning_rate": 4.1481481481481485e-06,
+      "loss": 1.0002,
+      "step": 7520
+    },
+    {
+      "epoch": 1.4758918071344571,
+      "grad_norm": 5.8046746253967285,
+      "learning_rate": 4.138888888888889e-06,
+      "loss": 1.0079,
+      "step": 7530
+    },
+    {
+      "epoch": 1.4778518228145825,
+      "grad_norm": 5.5930867195129395,
+      "learning_rate": 4.12962962962963e-06,
+      "loss": 1.0094,
+      "step": 7540
+    },
+    {
+      "epoch": 1.4798118384947079,
+      "grad_norm": 5.929795742034912,
+      "learning_rate": 4.1203703703703705e-06,
+      "loss": 1.0122,
+      "step": 7550
+    },
+    {
+      "epoch": 1.4817718541748335,
+      "grad_norm": 5.746526718139648,
+      "learning_rate": 4.111111111111111e-06,
+      "loss": 1.0186,
+      "step": 7560
+    },
+    {
+      "epoch": 1.4837318698549589,
+      "grad_norm": 6.316400051116943,
+      "learning_rate": 4.101851851851852e-06,
+      "loss": 1.0019,
+      "step": 7570
+    },
+    {
+      "epoch": 1.4856918855350842,
+      "grad_norm": 5.561999320983887,
+      "learning_rate": 4.092592592592593e-06,
+      "loss": 1.0157,
+      "step": 7580
+    },
+    {
+      "epoch": 1.4876519012152096,
+      "grad_norm": 6.012383937835693,
+      "learning_rate": 4.083333333333334e-06,
+      "loss": 1.0108,
+      "step": 7590
+    },
+    {
+      "epoch": 1.4896119168953352,
+      "grad_norm": 5.594381809234619,
+      "learning_rate": 4.074074074074074e-06,
+      "loss": 0.9917,
+      "step": 7600
+    },
+    {
+      "epoch": 1.4915719325754606,
+      "grad_norm": 6.158176898956299,
+      "learning_rate": 4.064814814814815e-06,
+      "loss": 1.0063,
+      "step": 7610
+    },
+    {
+      "epoch": 1.493531948255586,
+      "grad_norm": 5.4881510734558105,
+      "learning_rate": 4.055555555555556e-06,
+      "loss": 1.0122,
+      "step": 7620
+    },
+    {
+      "epoch": 1.4954919639357116,
+      "grad_norm": 6.7185869216918945,
+      "learning_rate": 4.046296296296296e-06,
+      "loss": 0.9927,
+      "step": 7630
+    },
+    {
+      "epoch": 1.497451979615837,
+      "grad_norm": 6.015646934509277,
+      "learning_rate": 4.037037037037037e-06,
+      "loss": 1.0054,
+      "step": 7640
+    },
+    {
+      "epoch": 1.4994119952959624,
+      "grad_norm": 5.855646133422852,
+      "learning_rate": 4.027777777777779e-06,
+      "loss": 0.9988,
+      "step": 7650
+    },
+    {
+      "epoch": 1.501372010976088,
+      "grad_norm": 6.158780574798584,
+      "learning_rate": 4.018518518518519e-06,
+      "loss": 1.027,
+      "step": 7660
+    },
+    {
+      "epoch": 1.5033320266562131,
+      "grad_norm": 5.943362712860107,
+      "learning_rate": 4.0092592592592594e-06,
+      "loss": 1.0285,
+      "step": 7670
+    },
+    {
+      "epoch": 1.5052920423363387,
+      "grad_norm": 5.775484085083008,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.0093,
+      "step": 7680
+    },
+    {
+      "epoch": 1.5072520580164641,
+      "grad_norm": 6.551724910736084,
+      "learning_rate": 3.990740740740741e-06,
+      "loss": 1.0185,
+      "step": 7690
+    },
+    {
+      "epoch": 1.5092120736965895,
+      "grad_norm": 5.986613750457764,
+      "learning_rate": 3.9814814814814814e-06,
+      "loss": 1.0058,
+      "step": 7700
+    },
+    {
+      "epoch": 1.5092120736965895,
+      "eval_loss": 1.0326706171035767,
+      "eval_runtime": 13.9953,
+      "eval_samples_per_second": 46.73,
+      "eval_steps_per_second": 5.859,
+      "step": 7700
+    },
+    {
+      "epoch": 1.511172089376715,
+      "grad_norm": 5.968577861785889,
+      "learning_rate": 3.972222222222223e-06,
+      "loss": 1.0111,
+      "step": 7710
+    },
+    {
+      "epoch": 1.5131321050568405,
+      "grad_norm": 5.956306457519531,
+      "learning_rate": 3.962962962962963e-06,
+      "loss": 0.9822,
+      "step": 7720
+    },
+    {
+      "epoch": 1.5150921207369659,
+      "grad_norm": 6.391535758972168,
+      "learning_rate": 3.953703703703704e-06,
+      "loss": 1.0073,
+      "step": 7730
+    },
+    {
+      "epoch": 1.5170521364170915,
+      "grad_norm": 6.031282901763916,
+      "learning_rate": 3.944444444444445e-06,
+      "loss": 0.9992,
+      "step": 7740
+    },
+    {
+      "epoch": 1.5190121520972166,
+      "grad_norm": 6.070298194885254,
+      "learning_rate": 3.935185185185186e-06,
+      "loss": 0.9995,
+      "step": 7750
+    },
+    {
+      "epoch": 1.5209721677773422,
+      "grad_norm": 6.470989227294922,
+      "learning_rate": 3.925925925925926e-06,
+      "loss": 1.0041,
+      "step": 7760
+    },
+    {
+      "epoch": 1.5229321834574676,
+      "grad_norm": 6.339069843292236,
+      "learning_rate": 3.916666666666667e-06,
+      "loss": 1.0145,
+      "step": 7770
+    },
+    {
+      "epoch": 1.524892199137593,
+      "grad_norm": 6.204465866088867,
+      "learning_rate": 3.907407407407408e-06,
+      "loss": 1.0202,
+      "step": 7780
+    },
+    {
+      "epoch": 1.5268522148177186,
+      "grad_norm": 5.39393424987793,
+      "learning_rate": 3.898148148148148e-06,
+      "loss": 1.0144,
+      "step": 7790
+    },
+    {
+      "epoch": 1.528812230497844,
+      "grad_norm": 6.083856105804443,
+      "learning_rate": 3.88888888888889e-06,
+      "loss": 1.0087,
+      "step": 7800
+    },
+    {
+      "epoch": 1.5307722461779694,
+      "grad_norm": 6.141456604003906,
+      "learning_rate": 3.87962962962963e-06,
+      "loss": 1.0189,
+      "step": 7810
+    },
+    {
+      "epoch": 1.532732261858095,
+      "grad_norm": 6.509696960449219,
+      "learning_rate": 3.87037037037037e-06,
+      "loss": 1.0136,
+      "step": 7820
+    },
+    {
+      "epoch": 1.5346922775382204,
+      "grad_norm": 5.372650146484375,
+      "learning_rate": 3.861111111111112e-06,
+      "loss": 1.0009,
+      "step": 7830
+    },
+    {
+      "epoch": 1.5366522932183457,
+      "grad_norm": 5.931579113006592,
+      "learning_rate": 3.851851851851852e-06,
+      "loss": 1.0144,
+      "step": 7840
+    },
+    {
+      "epoch": 1.5386123088984713,
+      "grad_norm": 6.2442626953125,
+      "learning_rate": 3.842592592592592e-06,
+      "loss": 1.0132,
+      "step": 7850
+    },
+    {
+      "epoch": 1.5405723245785965,
+      "grad_norm": 6.475978851318359,
+      "learning_rate": 3.833333333333334e-06,
+      "loss": 1.0058,
+      "step": 7860
+    },
+    {
+      "epoch": 1.542532340258722,
+      "grad_norm": 5.5199103355407715,
+      "learning_rate": 3.824074074074075e-06,
+      "loss": 1.0017,
+      "step": 7870
+    },
+    {
+      "epoch": 1.5444923559388475,
+      "grad_norm": 5.857003211975098,
+      "learning_rate": 3.814814814814815e-06,
+      "loss": 1.0108,
+      "step": 7880
+    },
+    {
+      "epoch": 1.5464523716189729,
+      "grad_norm": 5.81916618347168,
+      "learning_rate": 3.8055555555555556e-06,
+      "loss": 1.0185,
+      "step": 7890
+    },
+    {
+      "epoch": 1.5484123872990985,
+      "grad_norm": 6.099959850311279,
+      "learning_rate": 3.796296296296297e-06,
+      "loss": 1.0109,
+      "step": 7900
+    },
+    {
+      "epoch": 1.5503724029792239,
+      "grad_norm": 5.837829113006592,
+      "learning_rate": 3.7870370370370373e-06,
+      "loss": 1.0004,
+      "step": 7910
+    },
+    {
+      "epoch": 1.5523324186593492,
+      "grad_norm": 6.274796485900879,
+      "learning_rate": 3.777777777777778e-06,
+      "loss": 1.0165,
+      "step": 7920
+    },
+    {
+      "epoch": 1.5542924343394748,
+      "grad_norm": 5.926800727844238,
+      "learning_rate": 3.7685185185185185e-06,
+      "loss": 1.0067,
+      "step": 7930
+    },
+    {
+      "epoch": 1.5562524500196,
+      "grad_norm": 6.2746381759643555,
+      "learning_rate": 3.7592592592592597e-06,
+      "loss": 1.0078,
+      "step": 7940
+    },
+    {
+      "epoch": 1.5582124656997256,
+      "grad_norm": 6.315319538116455,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 1.0085,
+      "step": 7950
+    },
+    {
+      "epoch": 1.560172481379851,
+      "grad_norm": 6.36542272567749,
+      "learning_rate": 3.740740740740741e-06,
+      "loss": 1.0073,
+      "step": 7960
+    },
+    {
+      "epoch": 1.5621324970599764,
+      "grad_norm": 6.047231197357178,
+      "learning_rate": 3.731481481481482e-06,
+      "loss": 1.0019,
+      "step": 7970
+    },
+    {
+      "epoch": 1.564092512740102,
+      "grad_norm": 6.49619722366333,
+      "learning_rate": 3.7222222222222225e-06,
+      "loss": 1.0126,
+      "step": 7980
+    },
+    {
+      "epoch": 1.5660525284202274,
+      "grad_norm": 6.945578098297119,
+      "learning_rate": 3.7129629629629633e-06,
+      "loss": 1.0141,
+      "step": 7990
+    },
+    {
+      "epoch": 1.5680125441003527,
+      "grad_norm": 5.595407485961914,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 1.0077,
+      "step": 8000
+    },
+    {
+      "epoch": 1.5699725597804783,
+      "grad_norm": 6.3146867752075195,
+      "learning_rate": 3.694444444444445e-06,
+      "loss": 1.0178,
+      "step": 8010
+    },
+    {
+      "epoch": 1.5719325754606037,
+      "grad_norm": 8.867284774780273,
+      "learning_rate": 3.6851851851851854e-06,
+      "loss": 0.9974,
+      "step": 8020
+    },
+    {
+      "epoch": 1.573892591140729,
+      "grad_norm": 5.997375011444092,
+      "learning_rate": 3.675925925925926e-06,
+      "loss": 1.0169,
+      "step": 8030
+    },
+    {
+      "epoch": 1.5758526068208547,
+      "grad_norm": 6.150747776031494,
+      "learning_rate": 3.6666666666666666e-06,
+      "loss": 1.0087,
+      "step": 8040
+    },
+    {
+      "epoch": 1.5778126225009799,
+      "grad_norm": 6.344914436340332,
+      "learning_rate": 3.657407407407408e-06,
+      "loss": 1.008,
+      "step": 8050
+    },
+    {
+      "epoch": 1.5797726381811055,
+      "grad_norm": 5.457338333129883,
+      "learning_rate": 3.6481481481481486e-06,
+      "loss": 1.0169,
+      "step": 8060
+    },
+    {
+      "epoch": 1.5817326538612309,
+      "grad_norm": 6.128787994384766,
+      "learning_rate": 3.638888888888889e-06,
+      "loss": 1.0104,
+      "step": 8070
+    },
+    {
+      "epoch": 1.5836926695413562,
+      "grad_norm": 6.624429702758789,
+      "learning_rate": 3.6296296296296302e-06,
+      "loss": 0.9982,
+      "step": 8080
+    },
+    {
+      "epoch": 1.5856526852214818,
+      "grad_norm": 5.785494327545166,
+      "learning_rate": 3.6203703703703706e-06,
+      "loss": 0.9968,
+      "step": 8090
+    },
+    {
+      "epoch": 1.5876127009016072,
+      "grad_norm": 6.378090858459473,
+      "learning_rate": 3.6111111111111115e-06,
+      "loss": 0.9972,
+      "step": 8100
+    },
+    {
+      "epoch": 1.5895727165817326,
+      "grad_norm": 7.523343086242676,
+      "learning_rate": 3.601851851851852e-06,
+      "loss": 1.0077,
+      "step": 8110
+    },
+    {
+      "epoch": 1.5915327322618582,
+      "grad_norm": 5.578686714172363,
+      "learning_rate": 3.592592592592593e-06,
+      "loss": 1.0103,
+      "step": 8120
+    },
+    {
+      "epoch": 1.5934927479419836,
+      "grad_norm": 6.031825065612793,
+      "learning_rate": 3.5833333333333335e-06,
+      "loss": 1.015,
+      "step": 8130
+    },
+    {
+      "epoch": 1.595452763622109,
+      "grad_norm": 6.009352207183838,
+      "learning_rate": 3.5740740740740743e-06,
+      "loss": 0.9997,
+      "step": 8140
+    },
+    {
+      "epoch": 1.5974127793022346,
+      "grad_norm": 6.1863694190979,
+      "learning_rate": 3.5648148148148147e-06,
+      "loss": 0.9902,
+      "step": 8150
+    },
+    {
+      "epoch": 1.5993727949823597,
+      "grad_norm": 5.477410316467285,
+      "learning_rate": 3.555555555555556e-06,
+      "loss": 1.0077,
+      "step": 8160
+    },
+    {
+      "epoch": 1.6013328106624853,
+      "grad_norm": 6.446470260620117,
+      "learning_rate": 3.5462962962962967e-06,
+      "loss": 1.0123,
+      "step": 8170
+    },
+    {
+      "epoch": 1.6032928263426107,
+      "grad_norm": 5.979601860046387,
+      "learning_rate": 3.537037037037037e-06,
+      "loss": 1.0106,
+      "step": 8180
+    },
+    {
+      "epoch": 1.605252842022736,
+      "grad_norm": 5.748382091522217,
+      "learning_rate": 3.5277777777777784e-06,
+      "loss": 1.0083,
+      "step": 8190
+    },
+    {
+      "epoch": 1.6072128577028617,
+      "grad_norm": 6.122110843658447,
+      "learning_rate": 3.5185185185185187e-06,
+      "loss": 1.0124,
+      "step": 8200
+    },
+    {
+      "epoch": 1.609172873382987,
+      "grad_norm": 6.104898452758789,
+      "learning_rate": 3.5092592592592596e-06,
+      "loss": 1.0118,
+      "step": 8210
+    },
+    {
+      "epoch": 1.6111328890631125,
+      "grad_norm": 6.527801513671875,
+      "learning_rate": 3.5e-06,
+      "loss": 1.0084,
+      "step": 8220
+    },
+    {
+      "epoch": 1.613092904743238,
+      "grad_norm": 6.3016676902771,
+      "learning_rate": 3.490740740740741e-06,
+      "loss": 1.0104,
+      "step": 8230
+    },
+    {
+      "epoch": 1.6150529204233632,
+      "grad_norm": 6.44483757019043,
+      "learning_rate": 3.481481481481482e-06,
+      "loss": 1.0102,
+      "step": 8240
+    },
+    {
+      "epoch": 1.6170129361034888,
+      "grad_norm": 6.052326679229736,
+      "learning_rate": 3.4722222222222224e-06,
+      "loss": 1.0147,
+      "step": 8250
+    },
+    {
+      "epoch": 1.6189729517836142,
+      "grad_norm": 5.94256067276001,
+      "learning_rate": 3.4629629629629628e-06,
+      "loss": 1.0108,
+      "step": 8260
+    },
+    {
+      "epoch": 1.6209329674637396,
+      "grad_norm": 6.634081840515137,
+      "learning_rate": 3.453703703703704e-06,
+      "loss": 1.0139,
+      "step": 8270
+    },
+    {
+      "epoch": 1.6228929831438652,
+      "grad_norm": 6.239657878875732,
+      "learning_rate": 3.444444444444445e-06,
+      "loss": 0.9949,
+      "step": 8280
+    },
+    {
+      "epoch": 1.6248529988239906,
+      "grad_norm": 6.047982692718506,
+      "learning_rate": 3.4351851851851852e-06,
+      "loss": 1.0223,
+      "step": 8290
+    },
+    {
+      "epoch": 1.626813014504116,
+      "grad_norm": 5.618074417114258,
+      "learning_rate": 3.4259259259259265e-06,
+      "loss": 1.022,
+      "step": 8300
+    },
+    {
+      "epoch": 1.6287730301842416,
+      "grad_norm": 5.856754779815674,
+      "learning_rate": 3.416666666666667e-06,
+      "loss": 1.0053,
+      "step": 8310
+    },
+    {
+      "epoch": 1.630733045864367,
+      "grad_norm": 6.435812473297119,
+      "learning_rate": 3.4074074074074077e-06,
+      "loss": 0.9989,
+      "step": 8320
+    },
+    {
+      "epoch": 1.6326930615444923,
+      "grad_norm": 5.980440139770508,
+      "learning_rate": 3.398148148148148e-06,
+      "loss": 1.0108,
+      "step": 8330
+    },
+    {
+      "epoch": 1.634653077224618,
+      "grad_norm": 6.19628381729126,
+      "learning_rate": 3.3888888888888893e-06,
+      "loss": 0.995,
+      "step": 8340
+    },
+    {
+      "epoch": 1.636613092904743,
+      "grad_norm": 5.549736976623535,
+      "learning_rate": 3.37962962962963e-06,
+      "loss": 1.0181,
+      "step": 8350
+    },
+    {
+      "epoch": 1.6385731085848687,
+      "grad_norm": 6.485236644744873,
+      "learning_rate": 3.3703703703703705e-06,
+      "loss": 1.0092,
+      "step": 8360
+    },
+    {
+      "epoch": 1.640533124264994,
+      "grad_norm": 6.0221848487854,
+      "learning_rate": 3.3611111111111117e-06,
+      "loss": 1.0038,
+      "step": 8370
+    },
+    {
+      "epoch": 1.6424931399451195,
+      "grad_norm": 5.699035167694092,
+      "learning_rate": 3.351851851851852e-06,
+      "loss": 1.0092,
+      "step": 8380
+    },
+    {
+      "epoch": 1.644453155625245,
+      "grad_norm": 6.401322364807129,
+      "learning_rate": 3.342592592592593e-06,
+      "loss": 1.0013,
+      "step": 8390
+    },
+    {
+      "epoch": 1.6464131713053705,
+      "grad_norm": 6.202515125274658,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.0165,
+      "step": 8400
+    },
+    {
+      "epoch": 1.6464131713053705,
+      "eval_loss": 1.0264891386032104,
+      "eval_runtime": 14.05,
+      "eval_samples_per_second": 46.548,
+      "eval_steps_per_second": 5.836,
+      "step": 8400
+    },
+    {
+      "epoch": 1.6483731869854958,
+      "grad_norm": 6.184670448303223,
+      "learning_rate": 3.3240740740740746e-06,
+      "loss": 1.0055,
+      "step": 8410
+    },
+    {
+      "epoch": 1.6503332026656214,
+      "grad_norm": 6.040178298950195,
+      "learning_rate": 3.314814814814815e-06,
+      "loss": 1.007,
+      "step": 8420
+    },
+    {
+      "epoch": 1.6522932183457466,
+      "grad_norm": 6.925061225891113,
+      "learning_rate": 3.3055555555555558e-06,
+      "loss": 1.0101,
+      "step": 8430
+    },
+    {
+      "epoch": 1.6542532340258722,
+      "grad_norm": 6.050986289978027,
+      "learning_rate": 3.296296296296296e-06,
+      "loss": 0.9969,
+      "step": 8440
+    },
+    {
+      "epoch": 1.6562132497059978,
+      "grad_norm": 5.321831703186035,
+      "learning_rate": 3.2870370370370374e-06,
+      "loss": 1.0092,
+      "step": 8450
+    },
+    {
+      "epoch": 1.658173265386123,
+      "grad_norm": 5.440800666809082,
+      "learning_rate": 3.277777777777778e-06,
+      "loss": 1.0115,
+      "step": 8460
+    },
+    {
+      "epoch": 1.6601332810662486,
+      "grad_norm": 8.311331748962402,
+      "learning_rate": 3.2685185185185186e-06,
+      "loss": 1.0014,
+      "step": 8470
+    },
+    {
+      "epoch": 1.662093296746374,
+      "grad_norm": 6.014708518981934,
+      "learning_rate": 3.25925925925926e-06,
+      "loss": 1.0135,
+      "step": 8480
+    },
+    {
+      "epoch": 1.6640533124264993,
+      "grad_norm": 6.3742475509643555,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 1.0024,
+      "step": 8490
+    },
+    {
+      "epoch": 1.666013328106625,
+      "grad_norm": 6.1811041831970215,
+      "learning_rate": 3.240740740740741e-06,
+      "loss": 0.9983,
+      "step": 8500
+    },
+    {
+      "epoch": 1.6679733437867503,
+      "grad_norm": 6.055778503417969,
+      "learning_rate": 3.2314814814814814e-06,
+      "loss": 0.9986,
+      "step": 8510
+    },
+    {
+      "epoch": 1.6699333594668757,
+      "grad_norm": 6.654355525970459,
+      "learning_rate": 3.2222222222222227e-06,
+      "loss": 1.0111,
+      "step": 8520
+    },
+    {
+      "epoch": 1.6718933751470013,
+      "grad_norm": 6.156193733215332,
+      "learning_rate": 3.2129629629629635e-06,
+      "loss": 0.9879,
+      "step": 8530
+    },
+    {
+      "epoch": 1.6738533908271265,
+      "grad_norm": 6.491064548492432,
+      "learning_rate": 3.203703703703704e-06,
+      "loss": 1.0197,
+      "step": 8540
+    },
+    {
+      "epoch": 1.675813406507252,
+      "grad_norm": 5.876689434051514,
+      "learning_rate": 3.1944444444444443e-06,
+      "loss": 0.9934,
+      "step": 8550
+    },
+    {
+      "epoch": 1.6777734221873775,
+      "grad_norm": 6.113574504852295,
+      "learning_rate": 3.1851851851851855e-06,
+      "loss": 0.9992,
+      "step": 8560
+    },
+    {
+      "epoch": 1.6797334378675028,
+      "grad_norm": 6.2524518966674805,
+      "learning_rate": 3.1759259259259263e-06,
+      "loss": 1.0101,
+      "step": 8570
+    },
+    {
+      "epoch": 1.6816934535476284,
+      "grad_norm": 6.034154891967773,
+      "learning_rate": 3.1666666666666667e-06,
+      "loss": 1.0009,
+      "step": 8580
+    },
+    {
+      "epoch": 1.6836534692277538,
+      "grad_norm": 5.920796871185303,
+      "learning_rate": 3.157407407407408e-06,
+      "loss": 1.0088,
+      "step": 8590
+    },
+    {
+      "epoch": 1.6856134849078792,
+      "grad_norm": 5.578927993774414,
+      "learning_rate": 3.1481481481481483e-06,
+      "loss": 1.0004,
+      "step": 8600
+    },
+    {
+      "epoch": 1.6875735005880048,
+      "grad_norm": 6.1217803955078125,
+      "learning_rate": 3.138888888888889e-06,
+      "loss": 1.0034,
+      "step": 8610
+    },
+    {
+      "epoch": 1.6895335162681302,
+      "grad_norm": 6.209445953369141,
+      "learning_rate": 3.1296296296296295e-06,
+      "loss": 0.9912,
+      "step": 8620
+    },
+    {
+      "epoch": 1.6914935319482556,
+      "grad_norm": 6.419200897216797,
+      "learning_rate": 3.1203703703703708e-06,
+      "loss": 1.0048,
+      "step": 8630
+    },
+    {
+      "epoch": 1.6934535476283812,
+      "grad_norm": 6.344346046447754,
+      "learning_rate": 3.1111111111111116e-06,
+      "loss": 0.9986,
+      "step": 8640
+    },
+    {
+      "epoch": 1.6954135633085063,
+      "grad_norm": 6.1742963790893555,
+      "learning_rate": 3.101851851851852e-06,
+      "loss": 1.0085,
+      "step": 8650
+    },
+    {
+      "epoch": 1.697373578988632,
+      "grad_norm": 5.269043445587158,
+      "learning_rate": 3.0925925925925928e-06,
+      "loss": 1.0045,
+      "step": 8660
+    },
+    {
+      "epoch": 1.6993335946687573,
+      "grad_norm": 6.707599639892578,
+      "learning_rate": 3.0833333333333336e-06,
+      "loss": 1.0074,
+      "step": 8670
+    },
+    {
+      "epoch": 1.7012936103488827,
+      "grad_norm": 6.476319313049316,
+      "learning_rate": 3.0740740740740744e-06,
+      "loss": 1.0066,
+      "step": 8680
+    },
+    {
+      "epoch": 1.7032536260290083,
+      "grad_norm": 5.808709621429443,
+      "learning_rate": 3.064814814814815e-06,
+      "loss": 0.9969,
+      "step": 8690
+    },
+    {
+      "epoch": 1.7052136417091337,
+      "grad_norm": 10.500265121459961,
+      "learning_rate": 3.055555555555556e-06,
+      "loss": 1.0062,
+      "step": 8700
+    },
+    {
+      "epoch": 1.707173657389259,
+      "grad_norm": 5.992859840393066,
+      "learning_rate": 3.0462962962962964e-06,
+      "loss": 1.0133,
+      "step": 8710
+    },
+    {
+      "epoch": 1.7091336730693847,
+      "grad_norm": 6.0215325355529785,
+      "learning_rate": 3.0370370370370372e-06,
+      "loss": 1.0175,
+      "step": 8720
+    },
+    {
+      "epoch": 1.7110936887495098,
+      "grad_norm": 6.671189785003662,
+      "learning_rate": 3.0277777777777776e-06,
+      "loss": 1.0113,
+      "step": 8730
+    },
+    {
+      "epoch": 1.7130537044296354,
+      "grad_norm": 6.012123107910156,
+      "learning_rate": 3.018518518518519e-06,
+      "loss": 0.9931,
+      "step": 8740
+    },
+    {
+      "epoch": 1.7150137201097608,
+      "grad_norm": 5.328975200653076,
+      "learning_rate": 3.0092592592592597e-06,
+      "loss": 1.0052,
+      "step": 8750
+    },
+    {
+      "epoch": 1.7169737357898862,
+      "grad_norm": 6.184416770935059,
+      "learning_rate": 3e-06,
+      "loss": 1.0125,
+      "step": 8760
+    },
+    {
+      "epoch": 1.7189337514700118,
+      "grad_norm": 6.222989559173584,
+      "learning_rate": 2.990740740740741e-06,
+      "loss": 1.0166,
+      "step": 8770
+    },
+    {
+      "epoch": 1.7208937671501372,
+      "grad_norm": 6.706179141998291,
+      "learning_rate": 2.9814814814814817e-06,
+      "loss": 1.0095,
+      "step": 8780
+    },
+    {
+      "epoch": 1.7228537828302626,
+      "grad_norm": 6.823479652404785,
+      "learning_rate": 2.9722222222222225e-06,
+      "loss": 1.0147,
+      "step": 8790
+    },
+    {
+      "epoch": 1.7248137985103882,
+      "grad_norm": 7.233177185058594,
+      "learning_rate": 2.962962962962963e-06,
+      "loss": 0.9963,
+      "step": 8800
+    },
+    {
+      "epoch": 1.7267738141905136,
+      "grad_norm": 5.568224906921387,
+      "learning_rate": 2.953703703703704e-06,
+      "loss": 1.0052,
+      "step": 8810
+    },
+    {
+      "epoch": 1.728733829870639,
+      "grad_norm": 6.352148532867432,
+      "learning_rate": 2.944444444444445e-06,
+      "loss": 0.9951,
+      "step": 8820
+    },
+    {
+      "epoch": 1.7306938455507646,
+      "grad_norm": 6.312885284423828,
+      "learning_rate": 2.9351851851851853e-06,
+      "loss": 1.0126,
+      "step": 8830
+    },
+    {
+      "epoch": 1.7326538612308897,
+      "grad_norm": 6.648190975189209,
+      "learning_rate": 2.9259259259259257e-06,
+      "loss": 1.0095,
+      "step": 8840
+    },
+    {
+      "epoch": 1.7346138769110153,
+      "grad_norm": 7.448049545288086,
+      "learning_rate": 2.916666666666667e-06,
+      "loss": 1.0058,
+      "step": 8850
+    },
+    {
+      "epoch": 1.7365738925911407,
+      "grad_norm": 6.0499653816223145,
+      "learning_rate": 2.907407407407408e-06,
+      "loss": 1.002,
+      "step": 8860
+    },
+    {
+      "epoch": 1.738533908271266,
+      "grad_norm": 6.453138828277588,
+      "learning_rate": 2.898148148148148e-06,
+      "loss": 1.0107,
+      "step": 8870
+    },
+    {
+      "epoch": 1.7404939239513917,
+      "grad_norm": 6.5974555015563965,
+      "learning_rate": 2.888888888888889e-06,
+      "loss": 1.0119,
+      "step": 8880
+    },
+    {
+      "epoch": 1.742453939631517,
+      "grad_norm": 6.174962043762207,
+      "learning_rate": 2.87962962962963e-06,
+      "loss": 1.0067,
+      "step": 8890
+    },
+    {
+      "epoch": 1.7444139553116425,
+      "grad_norm": 6.343291282653809,
+      "learning_rate": 2.8703703703703706e-06,
+      "loss": 0.9903,
+      "step": 8900
+    },
+    {
+      "epoch": 1.746373970991768,
+      "grad_norm": 6.71571159362793,
+      "learning_rate": 2.861111111111111e-06,
+      "loss": 1.018,
+      "step": 8910
+    },
+    {
+      "epoch": 1.7483339866718932,
+      "grad_norm": 6.948620319366455,
+      "learning_rate": 2.8518518518518522e-06,
+      "loss": 1.0138,
+      "step": 8920
+    },
+    {
+      "epoch": 1.7502940023520188,
+      "grad_norm": 6.206993579864502,
+      "learning_rate": 2.842592592592593e-06,
+      "loss": 1.0024,
+      "step": 8930
+    },
+    {
+      "epoch": 1.7522540180321444,
+      "grad_norm": 6.701568603515625,
+      "learning_rate": 2.8333333333333335e-06,
+      "loss": 1.0117,
+      "step": 8940
+    },
+    {
+      "epoch": 1.7542140337122696,
+      "grad_norm": 6.476100921630859,
+      "learning_rate": 2.8240740740740743e-06,
+      "loss": 1.0,
+      "step": 8950
+    },
+    {
+      "epoch": 1.7561740493923952,
+      "grad_norm": 7.1540703773498535,
+      "learning_rate": 2.814814814814815e-06,
+      "loss": 0.9933,
+      "step": 8960
+    },
+    {
+      "epoch": 1.7581340650725206,
+      "grad_norm": 6.764838695526123,
+      "learning_rate": 2.805555555555556e-06,
+      "loss": 1.0051,
+      "step": 8970
+    },
+    {
+      "epoch": 1.760094080752646,
+      "grad_norm": 5.835758209228516,
+      "learning_rate": 2.7962962962962963e-06,
+      "loss": 1.0058,
+      "step": 8980
+    },
+    {
+      "epoch": 1.7620540964327716,
+      "grad_norm": 6.640206813812256,
+      "learning_rate": 2.7870370370370375e-06,
+      "loss": 1.0012,
+      "step": 8990
+    },
+    {
+      "epoch": 1.764014112112897,
+      "grad_norm": 7.032010555267334,
+      "learning_rate": 2.7777777777777783e-06,
+      "loss": 0.9988,
+      "step": 9000
+    },
+    {
+      "epoch": 1.7659741277930223,
+      "grad_norm": 6.398713111877441,
+      "learning_rate": 2.7685185185185187e-06,
+      "loss": 1.0179,
+      "step": 9010
+    },
+    {
+      "epoch": 1.767934143473148,
+      "grad_norm": 6.625948905944824,
+      "learning_rate": 2.759259259259259e-06,
+      "loss": 1.0089,
+      "step": 9020
+    },
+    {
+      "epoch": 1.769894159153273,
+      "grad_norm": 5.858684062957764,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 1.0057,
+      "step": 9030
+    },
+    {
+      "epoch": 1.7718541748333987,
+      "grad_norm": 7.03538703918457,
+      "learning_rate": 2.740740740740741e-06,
+      "loss": 1.0223,
+      "step": 9040
+    },
+    {
+      "epoch": 1.773814190513524,
+      "grad_norm": 6.332048416137695,
+      "learning_rate": 2.7314814814814816e-06,
+      "loss": 1.0027,
+      "step": 9050
+    },
+    {
+      "epoch": 1.7757742061936495,
+      "grad_norm": 6.304274082183838,
+      "learning_rate": 2.7222222222222224e-06,
+      "loss": 0.994,
+      "step": 9060
+    },
+    {
+      "epoch": 1.777734221873775,
+      "grad_norm": 5.588042736053467,
+      "learning_rate": 2.712962962962963e-06,
+      "loss": 1.0044,
+      "step": 9070
+    },
+    {
+      "epoch": 1.7796942375539004,
+      "grad_norm": 6.531040191650391,
+      "learning_rate": 2.703703703703704e-06,
+      "loss": 1.0059,
+      "step": 9080
+    },
+    {
+      "epoch": 1.7816542532340258,
+      "grad_norm": 6.511843681335449,
+      "learning_rate": 2.6944444444444444e-06,
+      "loss": 1.0046,
+      "step": 9090
+    },
+    {
+      "epoch": 1.7836142689141514,
+      "grad_norm": 6.718422889709473,
+      "learning_rate": 2.6851851851851856e-06,
+      "loss": 1.0084,
+      "step": 9100
+    },
+    {
+      "epoch": 1.7836142689141514,
+      "eval_loss": 1.0242658853530884,
+      "eval_runtime": 14.0168,
+      "eval_samples_per_second": 46.658,
+      "eval_steps_per_second": 5.85,
+      "step": 9100
+    },
+    {
+      "epoch": 1.7855742845942768,
+      "grad_norm": 5.511220455169678,
+      "learning_rate": 2.6759259259259264e-06,
+      "loss": 1.0028,
+      "step": 9110
+    },
+    {
+      "epoch": 1.7875343002744022,
+      "grad_norm": 6.125180244445801,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.9937,
+      "step": 9120
+    },
+    {
+      "epoch": 1.7894943159545278,
+      "grad_norm": 6.439998626708984,
+      "learning_rate": 2.6574074074074076e-06,
+      "loss": 0.9946,
+      "step": 9130
+    },
+    {
+      "epoch": 1.791454331634653,
+      "grad_norm": 6.341091156005859,
+      "learning_rate": 2.6481481481481485e-06,
+      "loss": 1.0131,
+      "step": 9140
+    },
+    {
+      "epoch": 1.7934143473147786,
+      "grad_norm": 6.909265041351318,
+      "learning_rate": 2.6388888888888893e-06,
+      "loss": 1.019,
+      "step": 9150
+    },
+    {
+      "epoch": 1.795374362994904,
+      "grad_norm": 5.798973083496094,
+      "learning_rate": 2.6296296296296297e-06,
+      "loss": 1.0058,
+      "step": 9160
+    },
+    {
+      "epoch": 1.7973343786750293,
+      "grad_norm": 6.564090728759766,
+      "learning_rate": 2.6203703703703705e-06,
+      "loss": 1.0012,
+      "step": 9170
+    },
+    {
+      "epoch": 1.799294394355155,
+      "grad_norm": 6.725678443908691,
+      "learning_rate": 2.6111111111111113e-06,
+      "loss": 0.9987,
+      "step": 9180
+    },
+    {
+      "epoch": 1.8012544100352803,
+      "grad_norm": 6.482659339904785,
+      "learning_rate": 2.601851851851852e-06,
+      "loss": 0.9986,
+      "step": 9190
+    },
+    {
+      "epoch": 1.8032144257154057,
+      "grad_norm": 10.73702621459961,
+      "learning_rate": 2.5925925925925925e-06,
+      "loss": 0.9968,
+      "step": 9200
+    },
+    {
+      "epoch": 1.8051744413955313,
+      "grad_norm": 6.893209457397461,
+      "learning_rate": 2.5833333333333337e-06,
+      "loss": 1.0111,
+      "step": 9210
+    },
+    {
+      "epoch": 1.8071344570756565,
+      "grad_norm": 6.212709426879883,
+      "learning_rate": 2.5740740740740745e-06,
+      "loss": 1.0036,
+      "step": 9220
+    },
+    {
+      "epoch": 1.809094472755782,
+      "grad_norm": 7.134426116943359,
+      "learning_rate": 2.564814814814815e-06,
+      "loss": 1.0008,
+      "step": 9230
+    },
+    {
+      "epoch": 1.8110544884359074,
+      "grad_norm": 12.403793334960938,
+      "learning_rate": 2.5555555555555557e-06,
+      "loss": 0.9933,
+      "step": 9240
+    },
+    {
+      "epoch": 1.8130145041160328,
+      "grad_norm": 6.31683349609375,
+      "learning_rate": 2.5462962962962966e-06,
+      "loss": 1.0083,
+      "step": 9250
+    },
+    {
+      "epoch": 1.8149745197961584,
+      "grad_norm": 5.857621669769287,
+      "learning_rate": 2.5370370370370374e-06,
+      "loss": 0.993,
+      "step": 9260
+    },
+    {
+      "epoch": 1.8169345354762838,
+      "grad_norm": 5.9981584548950195,
+      "learning_rate": 2.5277777777777778e-06,
+      "loss": 1.008,
+      "step": 9270
+    },
+    {
+      "epoch": 1.8188945511564092,
+      "grad_norm": 6.6797404289245605,
+      "learning_rate": 2.5185185185185186e-06,
+      "loss": 0.9915,
+      "step": 9280
+    },
+    {
+      "epoch": 1.8208545668365348,
+      "grad_norm": 6.050951957702637,
+      "learning_rate": 2.50925925925926e-06,
+      "loss": 1.0248,
+      "step": 9290
+    },
+    {
+      "epoch": 1.8228145825166602,
+      "grad_norm": 6.692429542541504,
+      "learning_rate": 2.5e-06,
+      "loss": 1.0103,
+      "step": 9300
+    },
+    {
+      "epoch": 1.8247745981967856,
+      "grad_norm": 6.396121978759766,
+      "learning_rate": 2.490740740740741e-06,
+      "loss": 1.0026,
+      "step": 9310
+    },
+    {
+      "epoch": 1.8267346138769112,
+      "grad_norm": 6.65963077545166,
+      "learning_rate": 2.481481481481482e-06,
+      "loss": 0.9912,
+      "step": 9320
+    },
+    {
+      "epoch": 1.8286946295570363,
+      "grad_norm": 7.833820343017578,
+      "learning_rate": 2.4722222222222226e-06,
+      "loss": 0.994,
+      "step": 9330
+    },
+    {
+      "epoch": 1.830654645237162,
+      "grad_norm": 5.663256645202637,
+      "learning_rate": 2.462962962962963e-06,
+      "loss": 1.0014,
+      "step": 9340
+    },
+    {
+      "epoch": 1.8326146609172873,
+      "grad_norm": 6.487533092498779,
+      "learning_rate": 2.453703703703704e-06,
+      "loss": 1.0022,
+      "step": 9350
+    },
+    {
+      "epoch": 1.8345746765974127,
+      "grad_norm": 6.143520832061768,
+      "learning_rate": 2.4444444444444447e-06,
+      "loss": 0.9852,
+      "step": 9360
+    },
+    {
+      "epoch": 1.8365346922775383,
+      "grad_norm": 6.246576309204102,
+      "learning_rate": 2.4351851851851855e-06,
+      "loss": 0.9995,
+      "step": 9370
+    },
+    {
+      "epoch": 1.8384947079576637,
+      "grad_norm": 6.709537982940674,
+      "learning_rate": 2.425925925925926e-06,
+      "loss": 1.0039,
+      "step": 9380
+    },
+    {
+      "epoch": 1.840454723637789,
+      "grad_norm": 6.031428813934326,
+      "learning_rate": 2.4166666666666667e-06,
+      "loss": 1.0103,
+      "step": 9390
+    },
+    {
+      "epoch": 1.8424147393179147,
+      "grad_norm": 6.703726291656494,
+      "learning_rate": 2.4074074074074075e-06,
+      "loss": 0.9999,
+      "step": 9400
+    },
+    {
+      "epoch": 1.8443747549980398,
+      "grad_norm": 6.8644819259643555,
+      "learning_rate": 2.3981481481481483e-06,
+      "loss": 0.9994,
+      "step": 9410
+    },
+    {
+      "epoch": 1.8463347706781654,
+      "grad_norm": 7.306818962097168,
+      "learning_rate": 2.388888888888889e-06,
+      "loss": 0.9995,
+      "step": 9420
+    },
+    {
+      "epoch": 1.848294786358291,
+      "grad_norm": 5.757288932800293,
+      "learning_rate": 2.37962962962963e-06,
+      "loss": 0.9883,
+      "step": 9430
+    },
+    {
+      "epoch": 1.8502548020384162,
+      "grad_norm": 6.339224815368652,
+      "learning_rate": 2.3703703703703707e-06,
+      "loss": 0.9953,
+      "step": 9440
+    },
+    {
+      "epoch": 1.8522148177185418,
+      "grad_norm": 8.186257362365723,
+      "learning_rate": 2.361111111111111e-06,
+      "loss": 0.9999,
+      "step": 9450
+    },
+    {
+      "epoch": 1.8541748333986672,
+      "grad_norm": 6.2635111808776855,
+      "learning_rate": 2.351851851851852e-06,
+      "loss": 1.0047,
+      "step": 9460
+    },
+    {
+      "epoch": 1.8561348490787926,
+      "grad_norm": 6.483547210693359,
+      "learning_rate": 2.3425925925925928e-06,
+      "loss": 1.0003,
+      "step": 9470
+    },
+    {
+      "epoch": 1.8580948647589182,
+      "grad_norm": 6.6517333984375,
+      "learning_rate": 2.3333333333333336e-06,
+      "loss": 1.0152,
+      "step": 9480
+    },
+    {
+      "epoch": 1.8600548804390435,
+      "grad_norm": 6.463938236236572,
+      "learning_rate": 2.324074074074074e-06,
+      "loss": 0.9917,
+      "step": 9490
+    },
+    {
+      "epoch": 1.862014896119169,
+      "grad_norm": 6.184695243835449,
+      "learning_rate": 2.314814814814815e-06,
+      "loss": 0.9889,
+      "step": 9500
+    },
+    {
+      "epoch": 1.8639749117992945,
+      "grad_norm": 6.435100078582764,
+      "learning_rate": 2.305555555555556e-06,
+      "loss": 1.002,
+      "step": 9510
+    },
+    {
+      "epoch": 1.8659349274794197,
+      "grad_norm": 5.961505889892578,
+      "learning_rate": 2.2962962962962964e-06,
+      "loss": 0.9931,
+      "step": 9520
+    },
+    {
+      "epoch": 1.8678949431595453,
+      "grad_norm": 6.590498924255371,
+      "learning_rate": 2.2870370370370372e-06,
+      "loss": 0.9967,
+      "step": 9530
+    },
+    {
+      "epoch": 1.8698549588396707,
+      "grad_norm": 6.019999027252197,
+      "learning_rate": 2.277777777777778e-06,
+      "loss": 0.9887,
+      "step": 9540
+    },
+    {
+      "epoch": 1.871814974519796,
+      "grad_norm": 6.332427978515625,
+      "learning_rate": 2.268518518518519e-06,
+      "loss": 1.0014,
+      "step": 9550
+    },
+    {
+      "epoch": 1.8737749901999217,
+      "grad_norm": 6.301807880401611,
+      "learning_rate": 2.2592592592592592e-06,
+      "loss": 0.9969,
+      "step": 9560
+    },
+    {
+      "epoch": 1.875735005880047,
+      "grad_norm": 6.279837131500244,
+      "learning_rate": 2.25e-06,
+      "loss": 0.9944,
+      "step": 9570
+    },
+    {
+      "epoch": 1.8776950215601724,
+      "grad_norm": 7.422451496124268,
+      "learning_rate": 2.240740740740741e-06,
+      "loss": 1.0007,
+      "step": 9580
+    },
+    {
+      "epoch": 1.879655037240298,
+      "grad_norm": 6.722255229949951,
+      "learning_rate": 2.2314814814814817e-06,
+      "loss": 0.9985,
+      "step": 9590
+    },
+    {
+      "epoch": 1.8816150529204234,
+      "grad_norm": 6.471696853637695,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 0.9984,
+      "step": 9600
+    },
+    {
+      "epoch": 1.8835750686005488,
+      "grad_norm": 6.347057819366455,
+      "learning_rate": 2.2129629629629633e-06,
+      "loss": 1.0092,
+      "step": 9610
+    },
+    {
+      "epoch": 1.8855350842806744,
+      "grad_norm": 6.340747833251953,
+      "learning_rate": 2.203703703703704e-06,
+      "loss": 0.9978,
+      "step": 9620
+    },
+    {
+      "epoch": 1.8874950999607996,
+      "grad_norm": 5.975046157836914,
+      "learning_rate": 2.1944444444444445e-06,
+      "loss": 1.0006,
+      "step": 9630
+    },
+    {
+      "epoch": 1.8894551156409252,
+      "grad_norm": 6.538125038146973,
+      "learning_rate": 2.1851851851851853e-06,
+      "loss": 1.006,
+      "step": 9640
+    },
+    {
+      "epoch": 1.8914151313210505,
+      "grad_norm": 6.345036506652832,
+      "learning_rate": 2.175925925925926e-06,
+      "loss": 1.0106,
+      "step": 9650
+    },
+    {
+      "epoch": 1.893375147001176,
+      "grad_norm": 6.604660987854004,
+      "learning_rate": 2.166666666666667e-06,
+      "loss": 0.9964,
+      "step": 9660
+    },
+    {
+      "epoch": 1.8953351626813015,
+      "grad_norm": 6.32635498046875,
+      "learning_rate": 2.1574074074074073e-06,
+      "loss": 1.0108,
+      "step": 9670
+    },
+    {
+      "epoch": 1.897295178361427,
+      "grad_norm": 6.804265975952148,
+      "learning_rate": 2.148148148148148e-06,
+      "loss": 1.0053,
+      "step": 9680
+    },
+    {
+      "epoch": 1.8992551940415523,
+      "grad_norm": 6.880146503448486,
+      "learning_rate": 2.138888888888889e-06,
+      "loss": 1.0073,
+      "step": 9690
+    },
+    {
+      "epoch": 1.901215209721678,
+      "grad_norm": 7.2419538497924805,
+      "learning_rate": 2.1296296296296298e-06,
+      "loss": 1.004,
+      "step": 9700
+    },
+    {
+      "epoch": 1.903175225401803,
+      "grad_norm": 6.319647789001465,
+      "learning_rate": 2.1203703703703706e-06,
+      "loss": 1.0049,
+      "step": 9710
+    },
+    {
+      "epoch": 1.9051352410819287,
+      "grad_norm": 6.6896491050720215,
+      "learning_rate": 2.1111111111111114e-06,
+      "loss": 1.0007,
+      "step": 9720
+    },
+    {
+      "epoch": 1.907095256762054,
+      "grad_norm": 5.643420696258545,
+      "learning_rate": 2.1018518518518522e-06,
+      "loss": 0.9898,
+      "step": 9730
+    },
+    {
+      "epoch": 1.9090552724421794,
+      "grad_norm": 6.660647869110107,
+      "learning_rate": 2.0925925925925926e-06,
+      "loss": 1.0039,
+      "step": 9740
+    },
+    {
+      "epoch": 1.911015288122305,
+      "grad_norm": 7.0795464515686035,
+      "learning_rate": 2.0833333333333334e-06,
+      "loss": 0.986,
+      "step": 9750
+    },
+    {
+      "epoch": 1.9129753038024304,
+      "grad_norm": 6.72257137298584,
+      "learning_rate": 2.0740740740740742e-06,
+      "loss": 1.0045,
+      "step": 9760
+    },
+    {
+      "epoch": 1.9149353194825558,
+      "grad_norm": 7.132964134216309,
+      "learning_rate": 2.064814814814815e-06,
+      "loss": 1.0043,
+      "step": 9770
+    },
+    {
+      "epoch": 1.9168953351626814,
+      "grad_norm": 6.032354354858398,
+      "learning_rate": 2.0555555555555555e-06,
+      "loss": 1.0159,
+      "step": 9780
+    },
+    {
+      "epoch": 1.9188553508428068,
+      "grad_norm": 6.651278495788574,
+      "learning_rate": 2.0462962962962967e-06,
+      "loss": 1.0009,
+      "step": 9790
+    },
+    {
+      "epoch": 1.9208153665229322,
+      "grad_norm": 6.432239532470703,
+      "learning_rate": 2.037037037037037e-06,
+      "loss": 1.0035,
+      "step": 9800
+    },
+    {
+      "epoch": 1.9208153665229322,
+      "eval_loss": 1.0201009511947632,
+      "eval_runtime": 14.0626,
+      "eval_samples_per_second": 46.506,
+      "eval_steps_per_second": 5.831,
+      "step": 9800
+    },
+    {
+      "epoch": 1.9227753822030578,
+      "grad_norm": 6.820202827453613,
+      "learning_rate": 2.027777777777778e-06,
+      "loss": 0.9879,
+      "step": 9810
+    },
+    {
+      "epoch": 1.924735397883183,
+      "grad_norm": 6.324273586273193,
+      "learning_rate": 2.0185185185185187e-06,
+      "loss": 1.0022,
+      "step": 9820
+    },
+    {
+      "epoch": 1.9266954135633085,
+      "grad_norm": 5.975357532501221,
+      "learning_rate": 2.0092592592592595e-06,
+      "loss": 1.0168,
+      "step": 9830
+    },
+    {
+      "epoch": 1.928655429243434,
+      "grad_norm": 7.312973976135254,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.0017,
+      "step": 9840
+    },
+    {
+      "epoch": 1.9306154449235593,
+      "grad_norm": 7.051765441894531,
+      "learning_rate": 1.9907407407407407e-06,
+      "loss": 0.992,
+      "step": 9850
+    },
+    {
+      "epoch": 1.932575460603685,
+      "grad_norm": 6.269126892089844,
+      "learning_rate": 1.9814814814814815e-06,
+      "loss": 1.0018,
+      "step": 9860
+    },
+    {
+      "epoch": 1.9345354762838103,
+      "grad_norm": 6.204473972320557,
+      "learning_rate": 1.9722222222222224e-06,
+      "loss": 0.999,
+      "step": 9870
+    },
+    {
+      "epoch": 1.9364954919639357,
+      "grad_norm": 6.758467197418213,
+      "learning_rate": 1.962962962962963e-06,
+      "loss": 1.0023,
+      "step": 9880
+    },
+    {
+      "epoch": 1.9384555076440613,
+      "grad_norm": 6.538867950439453,
+      "learning_rate": 1.953703703703704e-06,
+      "loss": 0.9907,
+      "step": 9890
+    },
+    {
+      "epoch": 1.9404155233241864,
+      "grad_norm": 7.003431797027588,
+      "learning_rate": 1.944444444444445e-06,
+      "loss": 1.0047,
+      "step": 9900
+    },
+    {
+      "epoch": 1.942375539004312,
+      "grad_norm": 6.028417587280273,
+      "learning_rate": 1.935185185185185e-06,
+      "loss": 1.0027,
+      "step": 9910
+    },
+    {
+      "epoch": 1.9443355546844376,
+      "grad_norm": 6.356867790222168,
+      "learning_rate": 1.925925925925926e-06,
+      "loss": 0.9902,
+      "step": 9920
+    },
+    {
+      "epoch": 1.9462955703645628,
+      "grad_norm": 7.202489376068115,
+      "learning_rate": 1.916666666666667e-06,
+      "loss": 1.0106,
+      "step": 9930
+    },
+    {
+      "epoch": 1.9482555860446884,
+      "grad_norm": 6.344156265258789,
+      "learning_rate": 1.9074074074074076e-06,
+      "loss": 0.9916,
+      "step": 9940
+    },
+    {
+      "epoch": 1.9502156017248138,
+      "grad_norm": 6.817245006561279,
+      "learning_rate": 1.8981481481481484e-06,
+      "loss": 0.9829,
+      "step": 9950
+    },
+    {
+      "epoch": 1.9521756174049392,
+      "grad_norm": 6.988001823425293,
+      "learning_rate": 1.888888888888889e-06,
+      "loss": 0.9997,
+      "step": 9960
+    },
+    {
+      "epoch": 1.9541356330850648,
+      "grad_norm": 6.65484619140625,
+      "learning_rate": 1.8796296296296299e-06,
+      "loss": 1.0026,
+      "step": 9970
+    },
+    {
+      "epoch": 1.9560956487651902,
+      "grad_norm": 6.717133045196533,
+      "learning_rate": 1.8703703703703705e-06,
+      "loss": 1.0071,
+      "step": 9980
+    },
+    {
+      "epoch": 1.9580556644453155,
+      "grad_norm": 6.037536144256592,
+      "learning_rate": 1.8611111111111113e-06,
+      "loss": 0.993,
+      "step": 9990
+    },
+    {
+      "epoch": 1.9600156801254411,
+      "grad_norm": 6.965274810791016,
+      "learning_rate": 1.8518518518518519e-06,
+      "loss": 0.9918,
+      "step": 10000
+    },
+    {
+      "epoch": 1.9619756958055663,
+      "grad_norm": 6.768052101135254,
+      "learning_rate": 1.8425925925925927e-06,
+      "loss": 0.9873,
+      "step": 10010
+    },
+    {
+      "epoch": 1.963935711485692,
+      "grad_norm": 6.111681938171387,
+      "learning_rate": 1.8333333333333333e-06,
+      "loss": 1.0111,
+      "step": 10020
+    },
+    {
+      "epoch": 1.9658957271658173,
+      "grad_norm": 6.606689453125,
+      "learning_rate": 1.8240740740740743e-06,
+      "loss": 0.9997,
+      "step": 10030
+    },
+    {
+      "epoch": 1.9678557428459427,
+      "grad_norm": 6.256832599639893,
+      "learning_rate": 1.8148148148148151e-06,
+      "loss": 0.9977,
+      "step": 10040
+    },
+    {
+      "epoch": 1.9698157585260683,
+      "grad_norm": 6.401496410369873,
+      "learning_rate": 1.8055555555555557e-06,
+      "loss": 1.0069,
+      "step": 10050
+    },
+    {
+      "epoch": 1.9717757742061937,
+      "grad_norm": 6.166873455047607,
+      "learning_rate": 1.7962962962962965e-06,
+      "loss": 0.9998,
+      "step": 10060
+    },
+    {
+      "epoch": 1.973735789886319,
+      "grad_norm": 5.77340030670166,
+      "learning_rate": 1.7870370370370371e-06,
+      "loss": 1.0008,
+      "step": 10070
+    },
+    {
+      "epoch": 1.9756958055664446,
+      "grad_norm": 7.6040120124816895,
+      "learning_rate": 1.777777777777778e-06,
+      "loss": 0.9991,
+      "step": 10080
+    },
+    {
+      "epoch": 1.97765582124657,
+      "grad_norm": 6.114837169647217,
+      "learning_rate": 1.7685185185185186e-06,
+      "loss": 0.9949,
+      "step": 10090
+    },
+    {
+      "epoch": 1.9796158369266954,
+      "grad_norm": 6.565464496612549,
+      "learning_rate": 1.7592592592592594e-06,
+      "loss": 1.0141,
+      "step": 10100
+    },
+    {
+      "epoch": 1.981575852606821,
+      "grad_norm": 6.436923980712891,
+      "learning_rate": 1.75e-06,
+      "loss": 0.9975,
+      "step": 10110
+    },
+    {
+      "epoch": 1.9835358682869462,
+      "grad_norm": 7.330633640289307,
+      "learning_rate": 1.740740740740741e-06,
+      "loss": 1.0014,
+      "step": 10120
+    },
+    {
+      "epoch": 1.9854958839670718,
+      "grad_norm": 7.172529697418213,
+      "learning_rate": 1.7314814814814814e-06,
+      "loss": 0.9811,
+      "step": 10130
+    },
+    {
+      "epoch": 1.9874558996471972,
+      "grad_norm": 6.164400577545166,
+      "learning_rate": 1.7222222222222224e-06,
+      "loss": 0.9972,
+      "step": 10140
+    },
+    {
+      "epoch": 1.9894159153273225,
+      "grad_norm": 6.392111778259277,
+      "learning_rate": 1.7129629629629632e-06,
+      "loss": 1.0001,
+      "step": 10150
+    },
+    {
+      "epoch": 1.9913759310074481,
+      "grad_norm": 8.478386878967285,
+      "learning_rate": 1.7037037037037038e-06,
+      "loss": 1.0015,
+      "step": 10160
+    },
+    {
+      "epoch": 1.9933359466875735,
+      "grad_norm": 6.89587926864624,
+      "learning_rate": 1.6944444444444446e-06,
+      "loss": 1.0047,
+      "step": 10170
+    },
+    {
+      "epoch": 1.995295962367699,
+      "grad_norm": 6.686724662780762,
+      "learning_rate": 1.6851851851851852e-06,
+      "loss": 0.9953,
+      "step": 10180
+    },
+    {
+      "epoch": 1.9972559780478245,
+      "grad_norm": 6.857761859893799,
+      "learning_rate": 1.675925925925926e-06,
+      "loss": 0.9836,
+      "step": 10190
+    },
+    {
+      "epoch": 1.9992159937279497,
+      "grad_norm": 6.607189178466797,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 0.9979,
+      "step": 10200
+    },
+    {
+      "epoch": 2.0011760094080753,
+      "grad_norm": 6.168262481689453,
+      "learning_rate": 1.6574074074074075e-06,
+      "loss": 0.9719,
+      "step": 10210
+    },
+    {
+      "epoch": 2.003136025088201,
+      "grad_norm": 6.525495529174805,
+      "learning_rate": 1.648148148148148e-06,
+      "loss": 0.9789,
+      "step": 10220
+    },
+    {
+      "epoch": 2.005096040768326,
+      "grad_norm": 5.562708854675293,
+      "learning_rate": 1.638888888888889e-06,
+      "loss": 0.98,
+      "step": 10230
+    },
+    {
+      "epoch": 2.0070560564484516,
+      "grad_norm": 7.023379802703857,
+      "learning_rate": 1.62962962962963e-06,
+      "loss": 0.985,
+      "step": 10240
+    },
+    {
+      "epoch": 2.0090160721285772,
+      "grad_norm": 7.55203104019165,
+      "learning_rate": 1.6203703703703705e-06,
+      "loss": 0.9758,
+      "step": 10250
+    },
+    {
+      "epoch": 2.0109760878087024,
+      "grad_norm": 13.453210830688477,
+      "learning_rate": 1.6111111111111113e-06,
+      "loss": 0.9656,
+      "step": 10260
+    },
+    {
+      "epoch": 2.012936103488828,
+      "grad_norm": 6.530963897705078,
+      "learning_rate": 1.601851851851852e-06,
+      "loss": 0.9615,
+      "step": 10270
+    },
+    {
+      "epoch": 2.014896119168953,
+      "grad_norm": 6.9765520095825195,
+      "learning_rate": 1.5925925925925927e-06,
+      "loss": 0.9786,
+      "step": 10280
+    },
+    {
+      "epoch": 2.0168561348490788,
+      "grad_norm": 7.124659538269043,
+      "learning_rate": 1.5833333333333333e-06,
+      "loss": 0.9795,
+      "step": 10290
+    },
+    {
+      "epoch": 2.0188161505292044,
+      "grad_norm": 7.119600772857666,
+      "learning_rate": 1.5740740740740742e-06,
+      "loss": 0.9672,
+      "step": 10300
+    },
+    {
+      "epoch": 2.0207761662093295,
+      "grad_norm": 6.711264133453369,
+      "learning_rate": 1.5648148148148148e-06,
+      "loss": 0.965,
+      "step": 10310
+    },
+    {
+      "epoch": 2.022736181889455,
+      "grad_norm": 7.027464389801025,
+      "learning_rate": 1.5555555555555558e-06,
+      "loss": 0.97,
+      "step": 10320
+    },
+    {
+      "epoch": 2.0246961975695807,
+      "grad_norm": 7.815002918243408,
+      "learning_rate": 1.5462962962962964e-06,
+      "loss": 0.9706,
+      "step": 10330
+    },
+    {
+      "epoch": 2.026656213249706,
+      "grad_norm": 6.989712238311768,
+      "learning_rate": 1.5370370370370372e-06,
+      "loss": 0.9741,
+      "step": 10340
+    },
+    {
+      "epoch": 2.0286162289298315,
+      "grad_norm": 6.903376579284668,
+      "learning_rate": 1.527777777777778e-06,
+      "loss": 0.9707,
+      "step": 10350
+    },
+    {
+      "epoch": 2.0305762446099567,
+      "grad_norm": 5.8991570472717285,
+      "learning_rate": 1.5185185185185186e-06,
+      "loss": 0.9731,
+      "step": 10360
+    },
+    {
+      "epoch": 2.0325362602900823,
+      "grad_norm": 7.208601474761963,
+      "learning_rate": 1.5092592592592594e-06,
+      "loss": 0.9845,
+      "step": 10370
+    },
+    {
+      "epoch": 2.034496275970208,
+      "grad_norm": 6.827620506286621,
+      "learning_rate": 1.5e-06,
+      "loss": 0.968,
+      "step": 10380
+    },
+    {
+      "epoch": 2.036456291650333,
+      "grad_norm": 6.6345295906066895,
+      "learning_rate": 1.4907407407407409e-06,
+      "loss": 0.9626,
+      "step": 10390
+    },
+    {
+      "epoch": 2.0384163073304586,
+      "grad_norm": 7.7344865798950195,
+      "learning_rate": 1.4814814814814815e-06,
+      "loss": 0.9717,
+      "step": 10400
+    },
+    {
+      "epoch": 2.0403763230105842,
+      "grad_norm": 7.092048168182373,
+      "learning_rate": 1.4722222222222225e-06,
+      "loss": 0.9721,
+      "step": 10410
+    },
+    {
+      "epoch": 2.0423363386907094,
+      "grad_norm": 6.746728897094727,
+      "learning_rate": 1.4629629629629629e-06,
+      "loss": 0.9755,
+      "step": 10420
+    },
+    {
+      "epoch": 2.044296354370835,
+      "grad_norm": 7.295462131500244,
+      "learning_rate": 1.453703703703704e-06,
+      "loss": 0.9633,
+      "step": 10430
+    },
+    {
+      "epoch": 2.0462563700509606,
+      "grad_norm": 6.882177829742432,
+      "learning_rate": 1.4444444444444445e-06,
+      "loss": 0.9668,
+      "step": 10440
+    },
+    {
+      "epoch": 2.0482163857310858,
+      "grad_norm": 6.1587300300598145,
+      "learning_rate": 1.4351851851851853e-06,
+      "loss": 0.9773,
+      "step": 10450
+    },
+    {
+      "epoch": 2.0501764014112114,
+      "grad_norm": 7.219092845916748,
+      "learning_rate": 1.4259259259259261e-06,
+      "loss": 0.9724,
+      "step": 10460
+    },
+    {
+      "epoch": 2.0521364170913365,
+      "grad_norm": 6.330183029174805,
+      "learning_rate": 1.4166666666666667e-06,
+      "loss": 0.965,
+      "step": 10470
+    },
+    {
+      "epoch": 2.054096432771462,
+      "grad_norm": 7.83421516418457,
+      "learning_rate": 1.4074074074074075e-06,
+      "loss": 0.9715,
+      "step": 10480
+    },
+    {
+      "epoch": 2.0560564484515877,
+      "grad_norm": 6.530287265777588,
+      "learning_rate": 1.3981481481481481e-06,
+      "loss": 0.9859,
+      "step": 10490
+    },
+    {
+      "epoch": 2.058016464131713,
+      "grad_norm": 7.4246625900268555,
+      "learning_rate": 1.3888888888888892e-06,
+      "loss": 0.9567,
+      "step": 10500
+    },
+    {
+      "epoch": 2.058016464131713,
+      "eval_loss": 1.0192580223083496,
+      "eval_runtime": 14.0156,
+      "eval_samples_per_second": 46.662,
+      "eval_steps_per_second": 5.851,
+      "step": 10500
+    },
+    {
+      "epoch": 2.0599764798118385,
+      "grad_norm": 7.052817344665527,
+      "learning_rate": 1.3796296296296296e-06,
+      "loss": 0.9657,
+      "step": 10510
+    },
+    {
+      "epoch": 2.061936495491964,
+      "grad_norm": 7.35888147354126,
+      "learning_rate": 1.3703703703703706e-06,
+      "loss": 0.9702,
+      "step": 10520
+    },
+    {
+      "epoch": 2.0638965111720893,
+      "grad_norm": 6.373018264770508,
+      "learning_rate": 1.3611111111111112e-06,
+      "loss": 0.9753,
+      "step": 10530
+    },
+    {
+      "epoch": 2.065856526852215,
+      "grad_norm": 6.720044136047363,
+      "learning_rate": 1.351851851851852e-06,
+      "loss": 0.9708,
+      "step": 10540
+    },
+    {
+      "epoch": 2.0678165425323405,
+      "grad_norm": 6.653097152709961,
+      "learning_rate": 1.3425925925925928e-06,
+      "loss": 0.9702,
+      "step": 10550
+    },
+    {
+      "epoch": 2.0697765582124656,
+      "grad_norm": 7.028474807739258,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 0.9787,
+      "step": 10560
+    },
+    {
+      "epoch": 2.0717365738925912,
+      "grad_norm": 7.457616329193115,
+      "learning_rate": 1.3240740740740742e-06,
+      "loss": 0.9585,
+      "step": 10570
+    },
+    {
+      "epoch": 2.0736965895727164,
+      "grad_norm": 6.340860843658447,
+      "learning_rate": 1.3148148148148148e-06,
+      "loss": 0.9592,
+      "step": 10580
+    },
+    {
+      "epoch": 2.075656605252842,
+      "grad_norm": 7.086592674255371,
+      "learning_rate": 1.3055555555555556e-06,
+      "loss": 0.9701,
+      "step": 10590
+    },
+    {
+      "epoch": 2.0776166209329676,
+      "grad_norm": 7.2209320068359375,
+      "learning_rate": 1.2962962962962962e-06,
+      "loss": 0.9655,
+      "step": 10600
+    },
+    {
+      "epoch": 2.0795766366130928,
+      "grad_norm": 6.678873062133789,
+      "learning_rate": 1.2870370370370373e-06,
+      "loss": 0.9674,
+      "step": 10610
+    },
+    {
+      "epoch": 2.0815366522932184,
+      "grad_norm": 7.573465347290039,
+      "learning_rate": 1.2777777777777779e-06,
+      "loss": 0.9754,
+      "step": 10620
+    },
+    {
+      "epoch": 2.083496667973344,
+      "grad_norm": 7.289187908172607,
+      "learning_rate": 1.2685185185185187e-06,
+      "loss": 0.9769,
+      "step": 10630
+    },
+    {
+      "epoch": 2.085456683653469,
+      "grad_norm": 6.930860996246338,
+      "learning_rate": 1.2592592592592593e-06,
+      "loss": 0.9697,
+      "step": 10640
+    },
+    {
+      "epoch": 2.0874166993335947,
+      "grad_norm": 6.5301618576049805,
+      "learning_rate": 1.25e-06,
+      "loss": 0.9766,
+      "step": 10650
+    },
+    {
+      "epoch": 2.08937671501372,
+      "grad_norm": 8.65379524230957,
+      "learning_rate": 1.240740740740741e-06,
+      "loss": 0.9545,
+      "step": 10660
+    },
+    {
+      "epoch": 2.0913367306938455,
+      "grad_norm": 6.923322677612305,
+      "learning_rate": 1.2314814814814815e-06,
+      "loss": 0.9643,
+      "step": 10670
+    },
+    {
+      "epoch": 2.093296746373971,
+      "grad_norm": 7.2696614265441895,
+      "learning_rate": 1.2222222222222223e-06,
+      "loss": 0.9677,
+      "step": 10680
+    },
+    {
+      "epoch": 2.0952567620540963,
+      "grad_norm": 7.1026835441589355,
+      "learning_rate": 1.212962962962963e-06,
+      "loss": 0.9791,
+      "step": 10690
+    },
+    {
+      "epoch": 2.097216777734222,
+      "grad_norm": 6.935880661010742,
+      "learning_rate": 1.2037037037037037e-06,
+      "loss": 0.9639,
+      "step": 10700
+    },
+    {
+      "epoch": 2.0991767934143475,
+      "grad_norm": 6.7199201583862305,
+      "learning_rate": 1.1944444444444446e-06,
+      "loss": 0.9795,
+      "step": 10710
+    },
+    {
+      "epoch": 2.1011368090944726,
+      "grad_norm": 7.126111030578613,
+      "learning_rate": 1.1851851851851854e-06,
+      "loss": 0.9604,
+      "step": 10720
+    },
+    {
+      "epoch": 2.1030968247745983,
+      "grad_norm": 7.274319171905518,
+      "learning_rate": 1.175925925925926e-06,
+      "loss": 0.9696,
+      "step": 10730
+    },
+    {
+      "epoch": 2.105056840454724,
+      "grad_norm": 7.07975959777832,
+      "learning_rate": 1.1666666666666668e-06,
+      "loss": 0.9708,
+      "step": 10740
+    },
+    {
+      "epoch": 2.107016856134849,
+      "grad_norm": 24.05130386352539,
+      "learning_rate": 1.1574074074074076e-06,
+      "loss": 0.9765,
+      "step": 10750
+    },
+    {
+      "epoch": 2.1089768718149746,
+      "grad_norm": 7.428647518157959,
+      "learning_rate": 1.1481481481481482e-06,
+      "loss": 0.9665,
+      "step": 10760
+    },
+    {
+      "epoch": 2.1109368874951,
+      "grad_norm": 6.800421714782715,
+      "learning_rate": 1.138888888888889e-06,
+      "loss": 0.9687,
+      "step": 10770
+    },
+    {
+      "epoch": 2.1128969031752254,
+      "grad_norm": 8.014981269836426,
+      "learning_rate": 1.1296296296296296e-06,
+      "loss": 0.9773,
+      "step": 10780
+    },
+    {
+      "epoch": 2.114856918855351,
+      "grad_norm": 7.821423053741455,
+      "learning_rate": 1.1203703703703704e-06,
+      "loss": 0.9694,
+      "step": 10790
+    },
+    {
+      "epoch": 2.116816934535476,
+      "grad_norm": 8.069396018981934,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 0.9741,
+      "step": 10800
+    },
+    {
+      "epoch": 2.1187769502156018,
+      "grad_norm": 7.268844127655029,
+      "learning_rate": 1.101851851851852e-06,
+      "loss": 0.9685,
+      "step": 10810
+    },
+    {
+      "epoch": 2.1207369658957274,
+      "grad_norm": 7.122548580169678,
+      "learning_rate": 1.0925925925925927e-06,
+      "loss": 0.9491,
+      "step": 10820
+    },
+    {
+      "epoch": 2.1226969815758525,
+      "grad_norm": 6.834280014038086,
+      "learning_rate": 1.0833333333333335e-06,
+      "loss": 0.9836,
+      "step": 10830
+    },
+    {
+      "epoch": 2.124656997255978,
+      "grad_norm": 6.231177806854248,
+      "learning_rate": 1.074074074074074e-06,
+      "loss": 0.9729,
+      "step": 10840
+    },
+    {
+      "epoch": 2.1266170129361033,
+      "grad_norm": 7.387545585632324,
+      "learning_rate": 1.0648148148148149e-06,
+      "loss": 0.9658,
+      "step": 10850
+    },
+    {
+      "epoch": 2.128577028616229,
+      "grad_norm": 8.538785934448242,
+      "learning_rate": 1.0555555555555557e-06,
+      "loss": 0.9521,
+      "step": 10860
+    },
+    {
+      "epoch": 2.1305370442963545,
+      "grad_norm": 7.329588890075684,
+      "learning_rate": 1.0462962962962963e-06,
+      "loss": 0.9804,
+      "step": 10870
+    },
+    {
+      "epoch": 2.1324970599764796,
+      "grad_norm": 6.391878128051758,
+      "learning_rate": 1.0370370370370371e-06,
+      "loss": 0.9684,
+      "step": 10880
+    },
+    {
+      "epoch": 2.1344570756566053,
+      "grad_norm": 7.090411186218262,
+      "learning_rate": 1.0277777777777777e-06,
+      "loss": 0.9622,
+      "step": 10890
+    },
+    {
+      "epoch": 2.136417091336731,
+      "grad_norm": 6.1176958084106445,
+      "learning_rate": 1.0185185185185185e-06,
+      "loss": 0.9696,
+      "step": 10900
+    },
+    {
+      "epoch": 2.138377107016856,
+      "grad_norm": 10.32075023651123,
+      "learning_rate": 1.0092592592592594e-06,
+      "loss": 0.9686,
+      "step": 10910
+    },
+    {
+      "epoch": 2.1403371226969816,
+      "grad_norm": 7.174540996551514,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.9647,
+      "step": 10920
+    },
+    {
+      "epoch": 2.1422971383771072,
+      "grad_norm": 7.739523410797119,
+      "learning_rate": 9.907407407407408e-07,
+      "loss": 0.9646,
+      "step": 10930
+    },
+    {
+      "epoch": 2.1442571540572324,
+      "grad_norm": 7.084393501281738,
+      "learning_rate": 9.814814814814816e-07,
+      "loss": 0.9695,
+      "step": 10940
+    },
+    {
+      "epoch": 2.146217169737358,
+      "grad_norm": 7.240608215332031,
+      "learning_rate": 9.722222222222224e-07,
+      "loss": 0.9655,
+      "step": 10950
+    },
+    {
+      "epoch": 2.148177185417483,
+      "grad_norm": 7.608102798461914,
+      "learning_rate": 9.62962962962963e-07,
+      "loss": 0.9598,
+      "step": 10960
+    },
+    {
+      "epoch": 2.1501372010976088,
+      "grad_norm": 7.160534858703613,
+      "learning_rate": 9.537037037037038e-07,
+      "loss": 0.9649,
+      "step": 10970
+    },
+    {
+      "epoch": 2.1520972167777344,
+      "grad_norm": 7.145265102386475,
+      "learning_rate": 9.444444444444445e-07,
+      "loss": 0.9646,
+      "step": 10980
+    },
+    {
+      "epoch": 2.1540572324578595,
+      "grad_norm": 25.257863998413086,
+      "learning_rate": 9.351851851851852e-07,
+      "loss": 0.9529,
+      "step": 10990
+    },
+    {
+      "epoch": 2.156017248137985,
+      "grad_norm": 7.226001262664795,
+      "learning_rate": 9.259259259259259e-07,
+      "loss": 0.9556,
+      "step": 11000
+    },
+    {
+      "epoch": 2.1579772638181107,
+      "grad_norm": 8.324951171875,
+      "learning_rate": 9.166666666666666e-07,
+      "loss": 0.9666,
+      "step": 11010
+    },
+    {
+      "epoch": 2.159937279498236,
+      "grad_norm": 7.676420211791992,
+      "learning_rate": 9.074074074074076e-07,
+      "loss": 0.9822,
+      "step": 11020
+    },
+    {
+      "epoch": 2.1618972951783615,
+      "grad_norm": 8.159540176391602,
+      "learning_rate": 8.981481481481483e-07,
+      "loss": 0.9608,
+      "step": 11030
+    },
+    {
+      "epoch": 2.1638573108584866,
+      "grad_norm": 7.420976161956787,
+      "learning_rate": 8.88888888888889e-07,
+      "loss": 0.974,
+      "step": 11040
+    },
+    {
+      "epoch": 2.1658173265386123,
+      "grad_norm": 7.221758842468262,
+      "learning_rate": 8.796296296296297e-07,
+      "loss": 0.9679,
+      "step": 11050
+    },
+    {
+      "epoch": 2.167777342218738,
+      "grad_norm": 7.399634838104248,
+      "learning_rate": 8.703703703703705e-07,
+      "loss": 0.9699,
+      "step": 11060
+    },
+    {
+      "epoch": 2.169737357898863,
+      "grad_norm": 6.9496026039123535,
+      "learning_rate": 8.611111111111112e-07,
+      "loss": 0.963,
+      "step": 11070
+    },
+    {
+      "epoch": 2.1716973735789886,
+      "grad_norm": 7.964886665344238,
+      "learning_rate": 8.518518518518519e-07,
+      "loss": 0.9649,
+      "step": 11080
+    },
+    {
+      "epoch": 2.1736573892591142,
+      "grad_norm": 6.945036888122559,
+      "learning_rate": 8.425925925925926e-07,
+      "loss": 0.9658,
+      "step": 11090
+    },
+    {
+      "epoch": 2.1756174049392394,
+      "grad_norm": 7.315770149230957,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 0.9715,
+      "step": 11100
+    },
+    {
+      "epoch": 2.177577420619365,
+      "grad_norm": 7.41797399520874,
+      "learning_rate": 8.24074074074074e-07,
+      "loss": 0.9579,
+      "step": 11110
+    },
+    {
+      "epoch": 2.1795374362994906,
+      "grad_norm": 6.9420366287231445,
+      "learning_rate": 8.14814814814815e-07,
+      "loss": 0.9785,
+      "step": 11120
+    },
+    {
+      "epoch": 2.1814974519796158,
+      "grad_norm": 6.257632255554199,
+      "learning_rate": 8.055555555555557e-07,
+      "loss": 0.9673,
+      "step": 11130
+    },
+    {
+      "epoch": 2.1834574676597414,
+      "grad_norm": 7.180861949920654,
+      "learning_rate": 7.962962962962964e-07,
+      "loss": 0.9691,
+      "step": 11140
+    },
+    {
+      "epoch": 2.1854174833398665,
+      "grad_norm": 7.286172389984131,
+      "learning_rate": 7.870370370370371e-07,
+      "loss": 0.9583,
+      "step": 11150
+    },
+    {
+      "epoch": 2.187377499019992,
+      "grad_norm": 7.305385589599609,
+      "learning_rate": 7.777777777777779e-07,
+      "loss": 0.9632,
+      "step": 11160
+    },
+    {
+      "epoch": 2.1893375147001177,
+      "grad_norm": 7.4768290519714355,
+      "learning_rate": 7.685185185185186e-07,
+      "loss": 0.9777,
+      "step": 11170
+    },
+    {
+      "epoch": 2.191297530380243,
+      "grad_norm": 7.44431209564209,
+      "learning_rate": 7.592592592592593e-07,
+      "loss": 0.9748,
+      "step": 11180
+    },
+    {
+      "epoch": 2.1932575460603685,
+      "grad_norm": 7.362459182739258,
+      "learning_rate": 7.5e-07,
+      "loss": 0.97,
+      "step": 11190
+    },
+    {
+      "epoch": 2.195217561740494,
+      "grad_norm": 6.118969917297363,
+      "learning_rate": 7.407407407407407e-07,
+      "loss": 0.967,
+      "step": 11200
+    },
+    {
+      "epoch": 2.195217561740494,
+      "eval_loss": 1.015982985496521,
+      "eval_runtime": 14.0402,
+      "eval_samples_per_second": 46.58,
+      "eval_steps_per_second": 5.84,
+      "step": 11200
+    },
+    {
+      "epoch": 2.1971775774206193,
+      "grad_norm": 7.092718601226807,
+      "learning_rate": 7.314814814814814e-07,
+      "loss": 0.9661,
+      "step": 11210
+    },
+    {
+      "epoch": 2.199137593100745,
+      "grad_norm": 7.384366512298584,
+      "learning_rate": 7.222222222222222e-07,
+      "loss": 0.9669,
+      "step": 11220
+    },
+    {
+      "epoch": 2.20109760878087,
+      "grad_norm": 7.769095420837402,
+      "learning_rate": 7.129629629629631e-07,
+      "loss": 0.9674,
+      "step": 11230
+    },
+    {
+      "epoch": 2.2030576244609956,
+      "grad_norm": 6.9204864501953125,
+      "learning_rate": 7.037037037037038e-07,
+      "loss": 0.9661,
+      "step": 11240
+    },
+    {
+      "epoch": 2.2050176401411212,
+      "grad_norm": 6.869389057159424,
+      "learning_rate": 6.944444444444446e-07,
+      "loss": 0.9619,
+      "step": 11250
+    },
+    {
+      "epoch": 2.2069776558212464,
+      "grad_norm": 7.39153528213501,
+      "learning_rate": 6.851851851851853e-07,
+      "loss": 0.9605,
+      "step": 11260
+    },
+    {
+      "epoch": 2.208937671501372,
+      "grad_norm": 6.61447811126709,
+      "learning_rate": 6.75925925925926e-07,
+      "loss": 0.9796,
+      "step": 11270
+    },
+    {
+      "epoch": 2.2108976871814976,
+      "grad_norm": 6.7030863761901855,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 0.96,
+      "step": 11280
+    },
+    {
+      "epoch": 2.2128577028616228,
+      "grad_norm": 7.7893829345703125,
+      "learning_rate": 6.574074074074074e-07,
+      "loss": 0.9571,
+      "step": 11290
+    },
+    {
+      "epoch": 2.2148177185417484,
+      "grad_norm": 7.597008228302002,
+      "learning_rate": 6.481481481481481e-07,
+      "loss": 0.9672,
+      "step": 11300
+    },
+    {
+      "epoch": 2.216777734221874,
+      "grad_norm": 6.373151779174805,
+      "learning_rate": 6.388888888888889e-07,
+      "loss": 0.9713,
+      "step": 11310
+    },
+    {
+      "epoch": 2.218737749901999,
+      "grad_norm": 7.9771575927734375,
+      "learning_rate": 6.296296296296296e-07,
+      "loss": 0.9674,
+      "step": 11320
+    },
+    {
+      "epoch": 2.2206977655821247,
+      "grad_norm": 7.23372745513916,
+      "learning_rate": 6.203703703703705e-07,
+      "loss": 0.9787,
+      "step": 11330
+    },
+    {
+      "epoch": 2.2226577812622503,
+      "grad_norm": 7.515969276428223,
+      "learning_rate": 6.111111111111112e-07,
+      "loss": 0.9702,
+      "step": 11340
+    },
+    {
+      "epoch": 2.2246177969423755,
+      "grad_norm": 6.468687534332275,
+      "learning_rate": 6.018518518518519e-07,
+      "loss": 0.981,
+      "step": 11350
+    },
+    {
+      "epoch": 2.226577812622501,
+      "grad_norm": 6.765408992767334,
+      "learning_rate": 5.925925925925927e-07,
+      "loss": 0.9757,
+      "step": 11360
+    },
+    {
+      "epoch": 2.2285378283026263,
+      "grad_norm": 8.031702041625977,
+      "learning_rate": 5.833333333333334e-07,
+      "loss": 0.9643,
+      "step": 11370
+    },
+    {
+      "epoch": 2.230497843982752,
+      "grad_norm": 7.857892990112305,
+      "learning_rate": 5.740740740740741e-07,
+      "loss": 0.9712,
+      "step": 11380
+    },
+    {
+      "epoch": 2.2324578596628775,
+      "grad_norm": 6.461490154266357,
+      "learning_rate": 5.648148148148148e-07,
+      "loss": 0.9664,
+      "step": 11390
+    },
+    {
+      "epoch": 2.2344178753430026,
+      "grad_norm": 7.290163993835449,
+      "learning_rate": 5.555555555555555e-07,
+      "loss": 0.9554,
+      "step": 11400
+    },
+    {
+      "epoch": 2.2363778910231282,
+      "grad_norm": 7.196146011352539,
+      "learning_rate": 5.462962962962963e-07,
+      "loss": 0.9755,
+      "step": 11410
+    },
+    {
+      "epoch": 2.238337906703254,
+      "grad_norm": 7.482301712036133,
+      "learning_rate": 5.37037037037037e-07,
+      "loss": 0.9577,
+      "step": 11420
+    },
+    {
+      "epoch": 2.240297922383379,
+      "grad_norm": 7.794521808624268,
+      "learning_rate": 5.277777777777779e-07,
+      "loss": 0.9679,
+      "step": 11430
+    },
+    {
+      "epoch": 2.2422579380635046,
+      "grad_norm": 6.857132434844971,
+      "learning_rate": 5.185185185185186e-07,
+      "loss": 0.9561,
+      "step": 11440
+    },
+    {
+      "epoch": 2.2442179537436298,
+      "grad_norm": 6.887381553649902,
+      "learning_rate": 5.092592592592593e-07,
+      "loss": 0.9712,
+      "step": 11450
+    },
+    {
+      "epoch": 2.2461779694237554,
+      "grad_norm": 7.537138938903809,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 0.9755,
+      "step": 11460
+    },
+    {
+      "epoch": 2.248137985103881,
+      "grad_norm": 7.604095935821533,
+      "learning_rate": 4.907407407407408e-07,
+      "loss": 0.968,
+      "step": 11470
+    },
+    {
+      "epoch": 2.250098000784006,
+      "grad_norm": 6.840078830718994,
+      "learning_rate": 4.814814814814815e-07,
+      "loss": 0.9677,
+      "step": 11480
+    },
+    {
+      "epoch": 2.2520580164641317,
+      "grad_norm": 7.0317301750183105,
+      "learning_rate": 4.7222222222222226e-07,
+      "loss": 0.9635,
+      "step": 11490
+    },
+    {
+      "epoch": 2.2540180321442573,
+      "grad_norm": 7.006657123565674,
+      "learning_rate": 4.6296296296296297e-07,
+      "loss": 0.96,
+      "step": 11500
+    },
+    {
+      "epoch": 2.2559780478243825,
+      "grad_norm": 7.701179504394531,
+      "learning_rate": 4.537037037037038e-07,
+      "loss": 0.9747,
+      "step": 11510
+    },
+    {
+      "epoch": 2.257938063504508,
+      "grad_norm": 7.04123592376709,
+      "learning_rate": 4.444444444444445e-07,
+      "loss": 0.9632,
+      "step": 11520
+    },
+    {
+      "epoch": 2.2598980791846337,
+      "grad_norm": 7.217432022094727,
+      "learning_rate": 4.3518518518518525e-07,
+      "loss": 0.9675,
+      "step": 11530
+    },
+    {
+      "epoch": 2.261858094864759,
+      "grad_norm": 7.661186695098877,
+      "learning_rate": 4.2592592592592596e-07,
+      "loss": 0.984,
+      "step": 11540
+    },
+    {
+      "epoch": 2.2638181105448845,
+      "grad_norm": 11.790616989135742,
+      "learning_rate": 4.1666666666666667e-07,
+      "loss": 0.9724,
+      "step": 11550
+    },
+    {
+      "epoch": 2.2657781262250096,
+      "grad_norm": 7.301620960235596,
+      "learning_rate": 4.074074074074075e-07,
+      "loss": 0.9601,
+      "step": 11560
+    },
+    {
+      "epoch": 2.2677381419051352,
+      "grad_norm": 8.462876319885254,
+      "learning_rate": 3.981481481481482e-07,
+      "loss": 0.9752,
+      "step": 11570
+    },
+    {
+      "epoch": 2.269698157585261,
+      "grad_norm": 6.7517547607421875,
+      "learning_rate": 3.8888888888888895e-07,
+      "loss": 0.9762,
+      "step": 11580
+    },
+    {
+      "epoch": 2.271658173265386,
+      "grad_norm": 7.299182891845703,
+      "learning_rate": 3.7962962962962966e-07,
+      "loss": 0.972,
+      "step": 11590
+    },
+    {
+      "epoch": 2.2736181889455116,
+      "grad_norm": 7.082560062408447,
+      "learning_rate": 3.7037037037037036e-07,
+      "loss": 0.9679,
+      "step": 11600
+    },
+    {
+      "epoch": 2.2755782046256368,
+      "grad_norm": 7.469284534454346,
+      "learning_rate": 3.611111111111111e-07,
+      "loss": 0.9546,
+      "step": 11610
+    },
+    {
+      "epoch": 2.2775382203057624,
+      "grad_norm": 7.5840535163879395,
+      "learning_rate": 3.518518518518519e-07,
+      "loss": 0.9761,
+      "step": 11620
+    },
+    {
+      "epoch": 2.279498235985888,
+      "grad_norm": 8.262914657592773,
+      "learning_rate": 3.4259259259259265e-07,
+      "loss": 0.9646,
+      "step": 11630
+    },
+    {
+      "epoch": 2.281458251666013,
+      "grad_norm": 8.672895431518555,
+      "learning_rate": 3.3333333333333335e-07,
+      "loss": 0.9732,
+      "step": 11640
+    },
+    {
+      "epoch": 2.2834182673461387,
+      "grad_norm": 7.6948347091674805,
+      "learning_rate": 3.2407407407407406e-07,
+      "loss": 0.9726,
+      "step": 11650
+    },
+    {
+      "epoch": 2.2853782830262643,
+      "grad_norm": 6.536133289337158,
+      "learning_rate": 3.148148148148148e-07,
+      "loss": 0.9694,
+      "step": 11660
+    },
+    {
+      "epoch": 2.2873382987063895,
+      "grad_norm": 8.90619945526123,
+      "learning_rate": 3.055555555555556e-07,
+      "loss": 0.9793,
+      "step": 11670
+    },
+    {
+      "epoch": 2.289298314386515,
+      "grad_norm": 7.702218055725098,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 0.9642,
+      "step": 11680
+    },
+    {
+      "epoch": 2.2912583300666407,
+      "grad_norm": 8.119073867797852,
+      "learning_rate": 2.8703703703703705e-07,
+      "loss": 0.9641,
+      "step": 11690
+    },
+    {
+      "epoch": 2.293218345746766,
+      "grad_norm": 7.159200191497803,
+      "learning_rate": 2.7777777777777776e-07,
+      "loss": 0.9636,
+      "step": 11700
+    },
+    {
+      "epoch": 2.2951783614268915,
+      "grad_norm": 6.998493671417236,
+      "learning_rate": 2.685185185185185e-07,
+      "loss": 0.9664,
+      "step": 11710
+    },
+    {
+      "epoch": 2.297138377107017,
+      "grad_norm": 7.50088357925415,
+      "learning_rate": 2.592592592592593e-07,
+      "loss": 0.9787,
+      "step": 11720
+    },
+    {
+      "epoch": 2.2990983927871422,
+      "grad_norm": 7.634678363800049,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 0.9712,
+      "step": 11730
+    },
+    {
+      "epoch": 2.301058408467268,
+      "grad_norm": 7.635111331939697,
+      "learning_rate": 2.4074074074074075e-07,
+      "loss": 0.9694,
+      "step": 11740
+    },
+    {
+      "epoch": 2.303018424147393,
+      "grad_norm": 7.645463466644287,
+      "learning_rate": 2.3148148148148148e-07,
+      "loss": 0.9717,
+      "step": 11750
+    },
+    {
+      "epoch": 2.3049784398275186,
+      "grad_norm": 6.8600568771362305,
+      "learning_rate": 2.2222222222222224e-07,
+      "loss": 0.9495,
+      "step": 11760
+    },
+    {
+      "epoch": 2.306938455507644,
+      "grad_norm": 7.019852638244629,
+      "learning_rate": 2.1296296296296298e-07,
+      "loss": 0.9527,
+      "step": 11770
+    },
+    {
+      "epoch": 2.3088984711877694,
+      "grad_norm": 8.227087020874023,
+      "learning_rate": 2.0370370370370374e-07,
+      "loss": 0.9608,
+      "step": 11780
+    },
+    {
+      "epoch": 2.310858486867895,
+      "grad_norm": 7.640336036682129,
+      "learning_rate": 1.9444444444444447e-07,
+      "loss": 0.9684,
+      "step": 11790
+    },
+    {
+      "epoch": 2.3128185025480206,
+      "grad_norm": 7.304652690887451,
+      "learning_rate": 1.8518518518518518e-07,
+      "loss": 0.9698,
+      "step": 11800
+    },
+    {
+      "epoch": 2.3147785182281457,
+      "grad_norm": 7.173428535461426,
+      "learning_rate": 1.7592592592592594e-07,
+      "loss": 0.9635,
+      "step": 11810
+    },
+    {
+      "epoch": 2.3167385339082713,
+      "grad_norm": 7.876226902008057,
+      "learning_rate": 1.6666666666666668e-07,
+      "loss": 0.9606,
+      "step": 11820
+    },
+    {
+      "epoch": 2.3186985495883965,
+      "grad_norm": 6.757920265197754,
+      "learning_rate": 1.574074074074074e-07,
+      "loss": 0.9692,
+      "step": 11830
+    },
+    {
+      "epoch": 2.320658565268522,
+      "grad_norm": 8.267070770263672,
+      "learning_rate": 1.4814814814814817e-07,
+      "loss": 0.9565,
+      "step": 11840
+    },
+    {
+      "epoch": 2.3226185809486477,
+      "grad_norm": 7.382258892059326,
+      "learning_rate": 1.3888888888888888e-07,
+      "loss": 0.9672,
+      "step": 11850
+    },
+    {
+      "epoch": 2.324578596628773,
+      "grad_norm": 7.452052116394043,
+      "learning_rate": 1.2962962962962964e-07,
+      "loss": 0.9649,
+      "step": 11860
+    },
+    {
+      "epoch": 2.3265386123088985,
+      "grad_norm": 9.273903846740723,
+      "learning_rate": 1.2037037037037037e-07,
+      "loss": 0.9652,
+      "step": 11870
+    },
+    {
+      "epoch": 2.328498627989024,
+      "grad_norm": 6.987192153930664,
+      "learning_rate": 1.1111111111111112e-07,
+      "loss": 0.9571,
+      "step": 11880
+    },
+    {
+      "epoch": 2.3304586436691492,
+      "grad_norm": 6.799464702606201,
+      "learning_rate": 1.0185185185185187e-07,
+      "loss": 0.9707,
+      "step": 11890
+    },
+    {
+      "epoch": 2.332418659349275,
+      "grad_norm": 7.16199254989624,
+      "learning_rate": 9.259259259259259e-08,
+      "loss": 0.9734,
+      "step": 11900
+    },
+    {
+      "epoch": 2.332418659349275,
+      "eval_loss": 1.0144330263137817,
+      "eval_runtime": 14.0924,
+      "eval_samples_per_second": 46.408,
+      "eval_steps_per_second": 5.819,
+      "step": 11900
+    },
+    {
+      "epoch": 2.3343786750294004,
+      "grad_norm": 8.338574409484863,
+      "learning_rate": 8.333333333333334e-08,
+      "loss": 0.9714,
+      "step": 11910
+    },
+    {
+      "epoch": 2.3363386907095256,
+      "grad_norm": 6.403331279754639,
+      "learning_rate": 7.407407407407409e-08,
+      "loss": 0.959,
+      "step": 11920
+    },
+    {
+      "epoch": 2.338298706389651,
+      "grad_norm": 8.215003967285156,
+      "learning_rate": 6.481481481481482e-08,
+      "loss": 0.964,
+      "step": 11930
+    },
+    {
+      "epoch": 2.3402587220697764,
+      "grad_norm": 7.930845737457275,
+      "learning_rate": 5.555555555555556e-08,
+      "loss": 0.9575,
+      "step": 11940
+    },
+    {
+      "epoch": 2.342218737749902,
+      "grad_norm": 7.033510684967041,
+      "learning_rate": 4.6296296296296295e-08,
+      "loss": 0.9626,
+      "step": 11950
+    },
+    {
+      "epoch": 2.3441787534300276,
+      "grad_norm": 7.8353800773620605,
+      "learning_rate": 3.703703703703704e-08,
+      "loss": 0.9673,
+      "step": 11960
+    },
+    {
+      "epoch": 2.3461387691101527,
+      "grad_norm": 7.304898738861084,
+      "learning_rate": 2.777777777777778e-08,
+      "loss": 0.9618,
+      "step": 11970
+    },
+    {
+      "epoch": 2.3480987847902783,
+      "grad_norm": 7.420264720916748,
+      "learning_rate": 1.851851851851852e-08,
+      "loss": 0.9544,
+      "step": 11980
+    },
+    {
+      "epoch": 2.350058800470404,
+      "grad_norm": 7.217324256896973,
+      "learning_rate": 9.25925925925926e-09,
+      "loss": 0.9752,
+      "step": 11990
+    },
+    {
+      "epoch": 2.352018816150529,
+      "grad_norm": 7.595900058746338,
+      "learning_rate": 0.0,
+      "loss": 0.9548,
+      "step": 12000
+    },
+    {
+      "epoch": 2.352018816150529,
+      "eval_loss": 1.0147355794906616,
+      "eval_runtime": 14.1258,
+      "eval_samples_per_second": 46.298,
+      "eval_steps_per_second": 5.805,
+      "step": 12000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 12000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 1,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.2834751593089335e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}